* [PATCH v4 0/9] Exynos DRM: rewrite IPP subsystem and userspace API [not found] <CGME20171023075448eucas1p25e25bb36d36139e3917211ab8ac999ec@eucas1p2.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski [not found] ` <CGME20171023075448eucas1p10255809be25caa7e44a3889f26561bc8@eucas1p1.samsung.com> ` (8 more replies) 0 siblings, 9 replies; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Marek Szyprowski, Inki Dae, Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon Dear all, This patchset performs complete rewrite of Exynos DRM IPP subsystem and its userspace API. Why such rewrite is needed? Exynos DRM IPP API is over-engineered in general, but not really extensible on the other side. It is also buggy, with significant design flaws: - Userspace API covers memory-2-memory picture operations together with CRTC writeback and duplicating features, which belongs to video plane. - Lack of support of the all required image formats (for example NV12 Samsung-tiled cannot be used due to lack of pixel format modifier support). - Userspace API designed only to mimic hardware behaviour, not easy to understand. - Lack of proper input validation in the core, drivers also didn't do that correctly, so it was possible to set incorrect parameters and easil trigger IOMMU fault or memory trash. - Drivers were partially disfunctional or supported only a subset of modes. Due to the above limitations and issues the Exynos DRM IPP API was not used by any of the open-source projects. I assume that it is safe to remove this broken API without any damage to open-source community. All remaining users (mainly Tizen project related) will be updated to the new version. This patchset changes Exynos DRM IPP subsystem to something useful. The userspace API is much simpler, state-less and easy to understand. Also the code of the core and driver is significantly smaller and easier to understand. Patches were tested on Exynos4412 based Odroid U3, Exynos5422 Odroid XU3 and Exynos5433 TM2 boards, on top of Linux next-20171016 kernel. A simple userspace test tool has been sent together with v1 of this patchset: https://www.spinics.net/lists/linux-samsung-soc/msg60498.html Tobias Jakobi has added support for this new API to his fork of libdrm and mpv video player: https://github.com/tobiasjakobi/libdrm/tree/ippv2 https://github.com/tobiasjakobi/mpv Best regards Marek Szyprowski Samsung R&D Institute Poland Changelog: v4: - fixed Exynos GSC limits (alignment) and operation in 90 degree rotation mode - fixed some style issues in Scaler driver (thanks to Andrzej) - fixed copy/paste typos in commit messages - improved debug messages, especially if provided parameters exceeds hardwave limits v3: https://www.spinics.net/lists/linux-samsung-soc/msg60981.html - fixed minor issues and added features pointed by other developers: * fixed missing ipp_unregister (Hoegeun) * added missing limits to FIMC and GSC driver (Hoegeun) * added more specific compatible strings to GSC driver (Hoegeun) * added Exynos5433 support to GSC driver (Hoegeun) * added autosuspend runtime PM to all IPP drivers (Tobias) - added Exynos5433 support to Scaler driver (thanks to Andrzej) - dropped Exynos5420 clk patch, which has been alredy merged v2: https://www.spinics.net/lists/dri-devel/msg153418.html - fixed minor issues pointed by other developers: * fixed possible null pointer dereferrence (Tobias) * changed limits_size to limits_count (Tobias) * renamed struct exynos_drm_ipp_format to drm_exynos_ipp_format (Andrzej) * added proper return value from exynos_drm_ipp_get_res_ioctl when no IPP driver is present (Andrzej) * properly aligned all uapi structures to be 32/64 bit safe (Emil) * properly initialize all strucutres - added new Exynos Scaler driver from Andrzej Pietrasiewicz v1: https://www.spinics.net/lists/linux-samsung-soc/msg60492.html - initial version of IPP v2 My previous works in this area: "[RFC v2 0/2] Exynos DRM: add Picture Processor extension" https://www.spinics.net/lists/dri-devel/msg140669.html - removed usage of DRM objects and properties - replaced them with simple list of parameters with predefined IDs "[RFC 0/4] Exynos DRM: add Picture Processor extension" https://www.spinics.net/lists/linux-samsung-soc/msg59323.html - moved this feature from DRM core to Exynos DRM driver - changed name from framebuffer processor to picture processor - simplified code to cover only things needed by Exynos drivers - implemented simple fifo task scheduler - cleaned up rotator driver conversion (removed IPP remainings) "[RFC 0/2] New feature: Framebuffer processors" https://www.spinics.net/lists/linux-samsung-soc/msg54810.html - generic approach implemented in DRM core, rejected Patch summary: Andrzej Pietrasiewicz (3): drm/exynos: Add driver for Exynos Scaler module ARM: dts: exynos: Add mem-2-mem Scaler devices ARM64: dts: exynos: Add mem-2-mem Scaler devices Marek Szyprowski (6): drm/exynos: ipp: Remove Exynos DRM IPP subsystem drm/exynos: ipp: Add IPP v2 framework drm/exynos: rotator: Convert driver to IPP v2 core API drm/exynos: gsc: Convert driver to IPP v2 core API drm/exynos: Add generic support for devices shared with V4L2 subsystem drm/exynos: fimc: Convert driver to IPP v2 core API .../devicetree/bindings/gpu/samsung-scaler.txt | 27 + arch/arm/boot/dts/exynos5420.dtsi | 35 + arch/arm64/boot/dts/exynos/exynos5433.dtsi | 42 + drivers/gpu/drm/exynos/Kconfig | 17 +- drivers/gpu/drm/exynos/Makefile | 1 + drivers/gpu/drm/exynos/exynos_drm_drv.c | 36 +- drivers/gpu/drm/exynos/exynos_drm_drv.h | 5 +- drivers/gpu/drm/exynos/exynos_drm_fimc.c | 930 +++----- drivers/gpu/drm/exynos/exynos_drm_fimc.h | 23 - drivers/gpu/drm/exynos/exynos_drm_gsc.c | 1001 +++------ drivers/gpu/drm/exynos/exynos_drm_gsc.h | 24 - drivers/gpu/drm/exynos/exynos_drm_ipp.c | 2253 ++++++-------------- drivers/gpu/drm/exynos/exynos_drm_ipp.h | 357 ++-- drivers/gpu/drm/exynos/exynos_drm_rotator.c | 740 ++----- drivers/gpu/drm/exynos/exynos_drm_rotator.h | 19 - drivers/gpu/drm/exynos/exynos_drm_scaler.c | 716 +++++++ drivers/gpu/drm/exynos/regs-scaler.h | 426 ++++ include/uapi/drm/exynos_drm.h | 326 +-- 18 files changed, 3078 insertions(+), 3900 deletions(-) create mode 100644 Documentation/devicetree/bindings/gpu/samsung-scaler.txt delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_fimc.h delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_gsc.h delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_rotator.h create mode 100644 drivers/gpu/drm/exynos/exynos_drm_scaler.c create mode 100644 drivers/gpu/drm/exynos/regs-scaler.h -- 2.14.2 ^ permalink raw reply [flat|nested] 20+ messages in thread
[parent not found: <CGME20171023075448eucas1p10255809be25caa7e44a3889f26561bc8@eucas1p1.samsung.com>]
* [PATCH v4 1/9] drm/exynos: ipp: Remove Exynos DRM IPP subsystem [not found] ` <CGME20171023075448eucas1p10255809be25caa7e44a3889f26561bc8@eucas1p1.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski 0 siblings, 0 replies; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Marek Szyprowski, Inki Dae, Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon Exynos IPP will be rewritten, so remove current IPP core code and mark existing drivers as BROKEN. Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> --- drivers/gpu/drm/exynos/Kconfig | 11 +- drivers/gpu/drm/exynos/Makefile | 1 - drivers/gpu/drm/exynos/exynos_drm_drv.c | 12 - drivers/gpu/drm/exynos/exynos_drm_drv.h | 2 - drivers/gpu/drm/exynos/exynos_drm_ipp.c | 1806 ------------------------------- drivers/gpu/drm/exynos/exynos_drm_ipp.h | 252 ----- include/uapi/drm/exynos_drm.h | 192 +--- 7 files changed, 4 insertions(+), 2272 deletions(-) delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.c delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.h diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig index 305dc3d4ff77..88cff0e039b6 100644 --- a/drivers/gpu/drm/exynos/Kconfig +++ b/drivers/gpu/drm/exynos/Kconfig @@ -94,26 +94,21 @@ config DRM_EXYNOS_G2D help Choose this option if you want to use Exynos G2D for DRM. -config DRM_EXYNOS_IPP - bool "Image Post Processor" - help - Choose this option if you want to use IPP feature for DRM. - config DRM_EXYNOS_FIMC bool "FIMC" - depends on DRM_EXYNOS_IPP && MFD_SYSCON + depends on BROKEN && MFD_SYSCON help Choose this option if you want to use Exynos FIMC for DRM. config DRM_EXYNOS_ROTATOR bool "Rotator" - depends on DRM_EXYNOS_IPP + depends on BROKEN help Choose this option if you want to use Exynos Rotator for DRM. config DRM_EXYNOS_GSC bool "GScaler" - depends on DRM_EXYNOS_IPP && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n + depends on BROKEN && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n help Choose this option if you want to use Exynos GSC for DRM. diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile index f663490e949d..09bb002c9555 100644 --- a/drivers/gpu/drm/exynos/Makefile +++ b/drivers/gpu/drm/exynos/Makefile @@ -17,7 +17,6 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER) += exynos_mixer.o exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI) += exynos_hdmi.o exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI) += exynos_drm_vidi.o exynosdrm-$(CONFIG_DRM_EXYNOS_G2D) += exynos_drm_g2d.o -exynosdrm-$(CONFIG_DRM_EXYNOS_IPP) += exynos_drm_ipp.o exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC) += exynos_drm_fimc.o exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR) += exynos_drm_rotator.o exynosdrm-$(CONFIG_DRM_EXYNOS_GSC) += exynos_drm_gsc.o diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c index e651a58c18cf..2fc5d3c01390 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c @@ -28,7 +28,6 @@ #include "exynos_drm_plane.h" #include "exynos_drm_vidi.h" #include "exynos_drm_g2d.h" -#include "exynos_drm_ipp.h" #include "exynos_drm_iommu.h" #define DRIVER_NAME "exynos" @@ -115,14 +114,6 @@ static const struct drm_ioctl_desc exynos_ioctls[] = { DRM_AUTH | DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl, DRM_AUTH | DRM_RENDER_ALLOW), - DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_PROPERTY, exynos_drm_ipp_get_property, - DRM_AUTH | DRM_RENDER_ALLOW), - DRM_IOCTL_DEF_DRV(EXYNOS_IPP_SET_PROPERTY, exynos_drm_ipp_set_property, - DRM_AUTH | DRM_RENDER_ALLOW), - DRM_IOCTL_DEF_DRV(EXYNOS_IPP_QUEUE_BUF, exynos_drm_ipp_queue_buf, - DRM_AUTH | DRM_RENDER_ALLOW), - DRM_IOCTL_DEF_DRV(EXYNOS_IPP_CMD_CTRL, exynos_drm_ipp_cmd_ctrl, - DRM_AUTH | DRM_RENDER_ALLOW), }; static const struct file_operations exynos_drm_driver_fops = { @@ -259,9 +250,6 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { DRV_PTR(rotator_driver, CONFIG_DRM_EXYNOS_ROTATOR), }, { DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC), - }, { - DRV_PTR(ipp_driver, CONFIG_DRM_EXYNOS_IPP), - DRM_VIRTUAL_DEVICE }, { &exynos_drm_platform_driver, DRM_VIRTUAL_DEVICE diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h b/drivers/gpu/drm/exynos/exynos_drm_drv.h index f8bae4cb4823..b47f810d64d2 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.h +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h @@ -185,7 +185,6 @@ struct exynos_drm_g2d_private { struct drm_exynos_file_private { struct exynos_drm_g2d_private *g2d_priv; - struct device *ipp_dev; }; /* @@ -293,6 +292,5 @@ extern struct platform_driver g2d_driver; extern struct platform_driver fimc_driver; extern struct platform_driver rotator_driver; extern struct platform_driver gsc_driver; -extern struct platform_driver ipp_driver; extern struct platform_driver mic_driver; #endif diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.c b/drivers/gpu/drm/exynos/exynos_drm_ipp.c deleted file mode 100644 index 3edda18cc2d2..000000000000 --- a/drivers/gpu/drm/exynos/exynos_drm_ipp.c +++ /dev/null @@ -1,1806 +0,0 @@ -/* - * Copyright (C) 2012 Samsung Electronics Co.Ltd - * Authors: - * Eunchul Kim <chulspro.kim@samsung.com> - * Jinyoung Jeon <jy0.jeon@samsung.com> - * Sangmin Lee <lsmin.lee@samsung.com> - * - * This program is free software; you can redistribute it and/or modify it - * under the terms of the GNU General Public License as published by the - * Free Software Foundation; either version 2 of the License, or (at your - * option) any later version. - * - */ -#include <linux/kernel.h> -#include <linux/platform_device.h> -#include <linux/types.h> -#include <linux/clk.h> -#include <linux/pm_runtime.h> - -#include <drm/drmP.h> -#include <drm/exynos_drm.h> -#include "exynos_drm_drv.h" -#include "exynos_drm_gem.h" -#include "exynos_drm_ipp.h" -#include "exynos_drm_iommu.h" - -/* - * IPP stands for Image Post Processing and - * supports image scaler/rotator and input/output DMA operations. - * using FIMC, GSC, Rotator, so on. - * IPP is integration device driver of same attribute h/w - */ - -/* - * TODO - * 1. expand command control id. - * 2. integrate property and config. - * 3. removed send_event id check routine. - * 4. compare send_event id if needed. - * 5. free subdrv_remove notifier callback list if needed. - * 6. need to check subdrv_open about multi-open. - * 7. need to power_on implement power and sysmmu ctrl. - */ - -#define get_ipp_context(dev) platform_get_drvdata(to_platform_device(dev)) -#define ipp_is_m2m_cmd(c) (c == IPP_CMD_M2M) - -/* - * A structure of event. - * - * @base: base of event. - * @event: ipp event. - */ -struct drm_exynos_ipp_send_event { - struct drm_pending_event base; - struct drm_exynos_ipp_event event; -}; - -/* - * A structure of memory node. - * - * @list: list head to memory queue information. - * @ops_id: id of operations. - * @prop_id: id of property. - * @buf_id: id of buffer. - * @buf_info: gem objects and dma address, size. - * @filp: a pointer to drm_file. - */ -struct drm_exynos_ipp_mem_node { - struct list_head list; - enum drm_exynos_ops_id ops_id; - u32 prop_id; - u32 buf_id; - struct drm_exynos_ipp_buf_info buf_info; -}; - -/* - * A structure of ipp context. - * - * @subdrv: prepare initialization using subdrv. - * @ipp_lock: lock for synchronization of access to ipp_idr. - * @prop_lock: lock for synchronization of access to prop_idr. - * @ipp_idr: ipp driver idr. - * @prop_idr: property idr. - * @event_workq: event work queue. - * @cmd_workq: command work queue. - */ -struct ipp_context { - struct exynos_drm_subdrv subdrv; - struct mutex ipp_lock; - struct mutex prop_lock; - struct idr ipp_idr; - struct idr prop_idr; - struct workqueue_struct *event_workq; - struct workqueue_struct *cmd_workq; -}; - -static LIST_HEAD(exynos_drm_ippdrv_list); -static DEFINE_MUTEX(exynos_drm_ippdrv_lock); -static BLOCKING_NOTIFIER_HEAD(exynos_drm_ippnb_list); - -int exynos_drm_ippdrv_register(struct exynos_drm_ippdrv *ippdrv) -{ - mutex_lock(&exynos_drm_ippdrv_lock); - list_add_tail(&ippdrv->drv_list, &exynos_drm_ippdrv_list); - mutex_unlock(&exynos_drm_ippdrv_lock); - - return 0; -} - -int exynos_drm_ippdrv_unregister(struct exynos_drm_ippdrv *ippdrv) -{ - mutex_lock(&exynos_drm_ippdrv_lock); - list_del(&ippdrv->drv_list); - mutex_unlock(&exynos_drm_ippdrv_lock); - - return 0; -} - -static int ipp_create_id(struct idr *id_idr, struct mutex *lock, void *obj) -{ - int ret; - - mutex_lock(lock); - ret = idr_alloc(id_idr, obj, 1, 0, GFP_KERNEL); - mutex_unlock(lock); - - return ret; -} - -static void ipp_remove_id(struct idr *id_idr, struct mutex *lock, u32 id) -{ - mutex_lock(lock); - idr_remove(id_idr, id); - mutex_unlock(lock); -} - -static void *ipp_find_obj(struct idr *id_idr, struct mutex *lock, u32 id) -{ - void *obj; - - mutex_lock(lock); - obj = idr_find(id_idr, id); - mutex_unlock(lock); - - return obj; -} - -static int ipp_check_driver(struct exynos_drm_ippdrv *ippdrv, - struct drm_exynos_ipp_property *property) -{ - if (ippdrv->dedicated || (!ipp_is_m2m_cmd(property->cmd) && - !pm_runtime_suspended(ippdrv->dev))) - return -EBUSY; - - if (ippdrv->check_property && - ippdrv->check_property(ippdrv->dev, property)) - return -EINVAL; - - return 0; -} - -static struct exynos_drm_ippdrv *ipp_find_driver(struct ipp_context *ctx, - struct drm_exynos_ipp_property *property) -{ - struct exynos_drm_ippdrv *ippdrv; - u32 ipp_id = property->ipp_id; - int ret; - - if (ipp_id) { - ippdrv = ipp_find_obj(&ctx->ipp_idr, &ctx->ipp_lock, ipp_id); - if (!ippdrv) { - DRM_DEBUG("ipp%d driver not found\n", ipp_id); - return ERR_PTR(-ENODEV); - } - - ret = ipp_check_driver(ippdrv, property); - if (ret < 0) { - DRM_DEBUG("ipp%d driver check error %d\n", ipp_id, ret); - return ERR_PTR(ret); - } - - return ippdrv; - } else { - list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) { - ret = ipp_check_driver(ippdrv, property); - if (ret == 0) - return ippdrv; - } - - DRM_DEBUG("cannot find driver suitable for given property.\n"); - } - - return ERR_PTR(-ENODEV); -} - -static struct exynos_drm_ippdrv *ipp_find_drv_by_handle(u32 prop_id) -{ - struct exynos_drm_ippdrv *ippdrv; - struct drm_exynos_ipp_cmd_node *c_node; - int count = 0; - - DRM_DEBUG_KMS("prop_id[%d]\n", prop_id); - - /* - * This case is search ipp driver by prop_id handle. - * sometimes, ipp subsystem find driver by prop_id. - * e.g PAUSE state, queue buf, command control. - */ - list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) { - DRM_DEBUG_KMS("count[%d]ippdrv[%pK]\n", count++, ippdrv); - - mutex_lock(&ippdrv->cmd_lock); - list_for_each_entry(c_node, &ippdrv->cmd_list, list) { - if (c_node->property.prop_id == prop_id) { - mutex_unlock(&ippdrv->cmd_lock); - return ippdrv; - } - } - mutex_unlock(&ippdrv->cmd_lock); - } - - return ERR_PTR(-ENODEV); -} - -int exynos_drm_ipp_get_property(struct drm_device *drm_dev, void *data, - struct drm_file *file) -{ - struct drm_exynos_file_private *file_priv = file->driver_priv; - struct device *dev = file_priv->ipp_dev; - struct ipp_context *ctx = get_ipp_context(dev); - struct drm_exynos_ipp_prop_list *prop_list = data; - struct exynos_drm_ippdrv *ippdrv; - int count = 0; - - if (!ctx) { - DRM_ERROR("invalid context.\n"); - return -EINVAL; - } - - if (!prop_list) { - DRM_ERROR("invalid property parameter.\n"); - return -EINVAL; - } - - DRM_DEBUG_KMS("ipp_id[%d]\n", prop_list->ipp_id); - - if (!prop_list->ipp_id) { - list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) - count++; - - /* - * Supports ippdrv list count for user application. - * First step user application getting ippdrv count. - * and second step getting ippdrv capability using ipp_id. - */ - prop_list->count = count; - } else { - /* - * Getting ippdrv capability by ipp_id. - * some device not supported wb, output interface. - * so, user application detect correct ipp driver - * using this ioctl. - */ - ippdrv = ipp_find_obj(&ctx->ipp_idr, &ctx->ipp_lock, - prop_list->ipp_id); - if (!ippdrv) { - DRM_ERROR("not found ipp%d driver.\n", - prop_list->ipp_id); - return -ENODEV; - } - - *prop_list = ippdrv->prop_list; - } - - return 0; -} - -static void ipp_print_property(struct drm_exynos_ipp_property *property, - int idx) -{ - struct drm_exynos_ipp_config *config = &property->config[idx]; - struct drm_exynos_pos *pos = &config->pos; - struct drm_exynos_sz *sz = &config->sz; - - DRM_DEBUG_KMS("prop_id[%d]ops[%s]fmt[0x%x]\n", - property->prop_id, idx ? "dst" : "src", config->fmt); - - DRM_DEBUG_KMS("pos[%d %d %d %d]sz[%d %d]f[%d]r[%d]\n", - pos->x, pos->y, pos->w, pos->h, - sz->hsize, sz->vsize, config->flip, config->degree); -} - -static struct drm_exynos_ipp_cmd_work *ipp_create_cmd_work(void) -{ - struct drm_exynos_ipp_cmd_work *cmd_work; - - cmd_work = kzalloc(sizeof(*cmd_work), GFP_KERNEL); - if (!cmd_work) - return ERR_PTR(-ENOMEM); - - INIT_WORK((struct work_struct *)cmd_work, ipp_sched_cmd); - - return cmd_work; -} - -static struct drm_exynos_ipp_event_work *ipp_create_event_work(void) -{ - struct drm_exynos_ipp_event_work *event_work; - - event_work = kzalloc(sizeof(*event_work), GFP_KERNEL); - if (!event_work) - return ERR_PTR(-ENOMEM); - - INIT_WORK(&event_work->work, ipp_sched_event); - - return event_work; -} - -int exynos_drm_ipp_set_property(struct drm_device *drm_dev, void *data, - struct drm_file *file) -{ - struct drm_exynos_file_private *file_priv = file->driver_priv; - struct device *dev = file_priv->ipp_dev; - struct ipp_context *ctx = get_ipp_context(dev); - struct drm_exynos_ipp_property *property = data; - struct exynos_drm_ippdrv *ippdrv; - struct drm_exynos_ipp_cmd_node *c_node; - u32 prop_id; - int ret, i; - - if (!ctx) { - DRM_ERROR("invalid context.\n"); - return -EINVAL; - } - - if (!property) { - DRM_ERROR("invalid property parameter.\n"); - return -EINVAL; - } - - prop_id = property->prop_id; - - /* - * This is log print for user application property. - * user application set various property. - */ - for_each_ipp_ops(i) - ipp_print_property(property, i); - - /* - * In case prop_id is not zero try to set existing property. - */ - if (prop_id) { - c_node = ipp_find_obj(&ctx->prop_idr, &ctx->prop_lock, prop_id); - - if (!c_node || c_node->filp != file) { - DRM_DEBUG_KMS("prop_id[%d] not found\n", prop_id); - return -EINVAL; - } - - if (c_node->state != IPP_STATE_STOP) { - DRM_DEBUG_KMS("prop_id[%d] not stopped\n", prop_id); - return -EINVAL; - } - - c_node->property = *property; - - return 0; - } - - /* find ipp driver using ipp id */ - ippdrv = ipp_find_driver(ctx, property); - if (IS_ERR(ippdrv)) { - DRM_ERROR("failed to get ipp driver.\n"); - return -EINVAL; - } - - /* allocate command node */ - c_node = kzalloc(sizeof(*c_node), GFP_KERNEL); - if (!c_node) - return -ENOMEM; - - ret = ipp_create_id(&ctx->prop_idr, &ctx->prop_lock, c_node); - if (ret < 0) { - DRM_ERROR("failed to create id.\n"); - goto err_clear; - } - property->prop_id = ret; - - DRM_DEBUG_KMS("created prop_id[%d]cmd[%d]ippdrv[%pK]\n", - property->prop_id, property->cmd, ippdrv); - - /* stored property information and ippdrv in private data */ - c_node->property = *property; - c_node->state = IPP_STATE_IDLE; - c_node->filp = file; - - c_node->start_work = ipp_create_cmd_work(); - if (IS_ERR(c_node->start_work)) { - DRM_ERROR("failed to create start work.\n"); - ret = PTR_ERR(c_node->start_work); - goto err_remove_id; - } - - c_node->stop_work = ipp_create_cmd_work(); - if (IS_ERR(c_node->stop_work)) { - DRM_ERROR("failed to create stop work.\n"); - ret = PTR_ERR(c_node->stop_work); - goto err_free_start; - } - - c_node->event_work = ipp_create_event_work(); - if (IS_ERR(c_node->event_work)) { - DRM_ERROR("failed to create event work.\n"); - ret = PTR_ERR(c_node->event_work); - goto err_free_stop; - } - - mutex_init(&c_node->lock); - mutex_init(&c_node->mem_lock); - mutex_init(&c_node->event_lock); - - init_completion(&c_node->start_complete); - init_completion(&c_node->stop_complete); - - for_each_ipp_ops(i) - INIT_LIST_HEAD(&c_node->mem_list[i]); - - INIT_LIST_HEAD(&c_node->event_list); - mutex_lock(&ippdrv->cmd_lock); - list_add_tail(&c_node->list, &ippdrv->cmd_list); - mutex_unlock(&ippdrv->cmd_lock); - - /* make dedicated state without m2m */ - if (!ipp_is_m2m_cmd(property->cmd)) - ippdrv->dedicated = true; - - return 0; - -err_free_stop: - kfree(c_node->stop_work); -err_free_start: - kfree(c_node->start_work); -err_remove_id: - ipp_remove_id(&ctx->prop_idr, &ctx->prop_lock, property->prop_id); -err_clear: - kfree(c_node); - return ret; -} - -static int ipp_validate_mem_node(struct drm_device *drm_dev, - struct drm_exynos_ipp_mem_node *m_node, - struct drm_exynos_ipp_cmd_node *c_node) -{ - struct drm_exynos_ipp_config *ipp_cfg; - unsigned int num_plane; - unsigned long size, buf_size = 0, plane_size, img_size = 0; - unsigned int bpp, width, height; - int i; - - ipp_cfg = &c_node->property.config[m_node->ops_id]; - num_plane = drm_format_num_planes(ipp_cfg->fmt); - - /** - * This is a rather simplified validation of a memory node. - * It basically verifies provided gem object handles - * and the buffer sizes with respect to current configuration. - * This is not the best that can be done - * but it seems more than enough - */ - for (i = 0; i < num_plane; ++i) { - width = ipp_cfg->sz.hsize; - height = ipp_cfg->sz.vsize; - bpp = drm_format_plane_cpp(ipp_cfg->fmt, i); - - /* - * The result of drm_format_plane_cpp() for chroma planes must - * be used with drm_format_xxxx_chroma_subsampling() for - * correct result. - */ - if (i > 0) { - width /= drm_format_horz_chroma_subsampling( - ipp_cfg->fmt); - height /= drm_format_vert_chroma_subsampling( - ipp_cfg->fmt); - } - plane_size = width * height * bpp; - img_size += plane_size; - - if (m_node->buf_info.handles[i]) { - size = exynos_drm_gem_get_size(drm_dev, - m_node->buf_info.handles[i], - c_node->filp); - if (plane_size > size) { - DRM_ERROR( - "buffer %d is smaller than required\n", - i); - return -EINVAL; - } - - buf_size += size; - } - } - - if (buf_size < img_size) { - DRM_ERROR("size of buffers(%lu) is smaller than image(%lu)\n", - buf_size, img_size); - return -EINVAL; - } - - return 0; -} - -static int ipp_put_mem_node(struct drm_device *drm_dev, - struct drm_exynos_ipp_cmd_node *c_node, - struct drm_exynos_ipp_mem_node *m_node) -{ - int i; - - DRM_DEBUG_KMS("node[%pK]\n", m_node); - - if (!m_node) { - DRM_ERROR("invalid dequeue node.\n"); - return -EFAULT; - } - - DRM_DEBUG_KMS("ops_id[%d]\n", m_node->ops_id); - - /* put gem buffer */ - for_each_ipp_planar(i) { - unsigned long handle = m_node->buf_info.handles[i]; - if (handle) - exynos_drm_gem_put_dma_addr(drm_dev, handle, - c_node->filp); - } - - list_del(&m_node->list); - kfree(m_node); - - return 0; -} - -static struct drm_exynos_ipp_mem_node - *ipp_get_mem_node(struct drm_device *drm_dev, - struct drm_exynos_ipp_cmd_node *c_node, - struct drm_exynos_ipp_queue_buf *qbuf) -{ - struct drm_exynos_ipp_mem_node *m_node; - struct drm_exynos_ipp_buf_info *buf_info; - int i; - - m_node = kzalloc(sizeof(*m_node), GFP_KERNEL); - if (!m_node) - return ERR_PTR(-ENOMEM); - - buf_info = &m_node->buf_info; - - /* operations, buffer id */ - m_node->ops_id = qbuf->ops_id; - m_node->prop_id = qbuf->prop_id; - m_node->buf_id = qbuf->buf_id; - INIT_LIST_HEAD(&m_node->list); - - DRM_DEBUG_KMS("m_node[%pK]ops_id[%d]\n", m_node, qbuf->ops_id); - DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]\n", qbuf->prop_id, m_node->buf_id); - - for_each_ipp_planar(i) { - DRM_DEBUG_KMS("i[%d]handle[0x%x]\n", i, qbuf->handle[i]); - - /* get dma address by handle */ - if (qbuf->handle[i]) { - dma_addr_t *addr; - - addr = exynos_drm_gem_get_dma_addr(drm_dev, - qbuf->handle[i], c_node->filp); - if (IS_ERR(addr)) { - DRM_ERROR("failed to get addr.\n"); - ipp_put_mem_node(drm_dev, c_node, m_node); - return ERR_PTR(-EFAULT); - } - - buf_info->handles[i] = qbuf->handle[i]; - buf_info->base[i] = *addr; - DRM_DEBUG_KMS("i[%d]base[%pad]hd[0x%lx]\n", i, - &buf_info->base[i], buf_info->handles[i]); - } - } - - mutex_lock(&c_node->mem_lock); - if (ipp_validate_mem_node(drm_dev, m_node, c_node)) { - ipp_put_mem_node(drm_dev, c_node, m_node); - mutex_unlock(&c_node->mem_lock); - return ERR_PTR(-EFAULT); - } - list_add_tail(&m_node->list, &c_node->mem_list[qbuf->ops_id]); - mutex_unlock(&c_node->mem_lock); - - return m_node; -} - -static void ipp_clean_mem_nodes(struct drm_device *drm_dev, - struct drm_exynos_ipp_cmd_node *c_node, int ops) -{ - struct drm_exynos_ipp_mem_node *m_node, *tm_node; - struct list_head *head = &c_node->mem_list[ops]; - - mutex_lock(&c_node->mem_lock); - - list_for_each_entry_safe(m_node, tm_node, head, list) { - int ret; - - ret = ipp_put_mem_node(drm_dev, c_node, m_node); - if (ret) - DRM_ERROR("failed to put m_node.\n"); - } - - mutex_unlock(&c_node->mem_lock); -} - -static int ipp_get_event(struct drm_device *drm_dev, - struct drm_exynos_ipp_cmd_node *c_node, - struct drm_exynos_ipp_queue_buf *qbuf) -{ - struct drm_exynos_ipp_send_event *e; - int ret; - - DRM_DEBUG_KMS("ops_id[%d]buf_id[%d]\n", qbuf->ops_id, qbuf->buf_id); - - e = kzalloc(sizeof(*e), GFP_KERNEL); - if (!e) - return -ENOMEM; - - /* make event */ - e->event.base.type = DRM_EXYNOS_IPP_EVENT; - e->event.base.length = sizeof(e->event); - e->event.user_data = qbuf->user_data; - e->event.prop_id = qbuf->prop_id; - e->event.buf_id[EXYNOS_DRM_OPS_DST] = qbuf->buf_id; - - ret = drm_event_reserve_init(drm_dev, c_node->filp, &e->base, &e->event.base); - if (ret) { - kfree(e); - return ret; - } - - mutex_lock(&c_node->event_lock); - list_add_tail(&e->base.link, &c_node->event_list); - mutex_unlock(&c_node->event_lock); - - return 0; -} - -static void ipp_put_event(struct drm_exynos_ipp_cmd_node *c_node, - struct drm_exynos_ipp_queue_buf *qbuf) -{ - struct drm_exynos_ipp_send_event *e, *te; - int count = 0; - - mutex_lock(&c_node->event_lock); - list_for_each_entry_safe(e, te, &c_node->event_list, base.link) { - DRM_DEBUG_KMS("count[%d]e[%pK]\n", count++, e); - - /* - * qbuf == NULL condition means all event deletion. - * stop operations want to delete all event list. - * another case delete only same buf id. - */ - if (!qbuf) { - /* delete list */ - list_del(&e->base.link); - kfree(e); - } - - /* compare buffer id */ - if (qbuf && (qbuf->buf_id == - e->event.buf_id[EXYNOS_DRM_OPS_DST])) { - /* delete list */ - list_del(&e->base.link); - kfree(e); - goto out_unlock; - } - } - -out_unlock: - mutex_unlock(&c_node->event_lock); - return; -} - -static void ipp_clean_cmd_node(struct ipp_context *ctx, - struct drm_exynos_ipp_cmd_node *c_node) -{ - int i; - - /* cancel works */ - cancel_work_sync(&c_node->start_work->work); - cancel_work_sync(&c_node->stop_work->work); - cancel_work_sync(&c_node->event_work->work); - - /* put event */ - ipp_put_event(c_node, NULL); - - for_each_ipp_ops(i) - ipp_clean_mem_nodes(ctx->subdrv.drm_dev, c_node, i); - - /* delete list */ - list_del(&c_node->list); - - ipp_remove_id(&ctx->prop_idr, &ctx->prop_lock, - c_node->property.prop_id); - - /* destroy mutex */ - mutex_destroy(&c_node->lock); - mutex_destroy(&c_node->mem_lock); - mutex_destroy(&c_node->event_lock); - - /* free command node */ - kfree(c_node->start_work); - kfree(c_node->stop_work); - kfree(c_node->event_work); - kfree(c_node); -} - -static bool ipp_check_mem_list(struct drm_exynos_ipp_cmd_node *c_node) -{ - switch (c_node->property.cmd) { - case IPP_CMD_WB: - return !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_DST]); - case IPP_CMD_OUTPUT: - return !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_SRC]); - case IPP_CMD_M2M: - default: - return !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_SRC]) && - !list_empty(&c_node->mem_list[EXYNOS_DRM_OPS_DST]); - } -} - -static struct drm_exynos_ipp_mem_node - *ipp_find_mem_node(struct drm_exynos_ipp_cmd_node *c_node, - struct drm_exynos_ipp_queue_buf *qbuf) -{ - struct drm_exynos_ipp_mem_node *m_node; - struct list_head *head; - int count = 0; - - DRM_DEBUG_KMS("buf_id[%d]\n", qbuf->buf_id); - - /* source/destination memory list */ - head = &c_node->mem_list[qbuf->ops_id]; - - /* find memory node from memory list */ - list_for_each_entry(m_node, head, list) { - DRM_DEBUG_KMS("count[%d]m_node[%pK]\n", count++, m_node); - - /* compare buffer id */ - if (m_node->buf_id == qbuf->buf_id) - return m_node; - } - - return NULL; -} - -static int ipp_set_mem_node(struct exynos_drm_ippdrv *ippdrv, - struct drm_exynos_ipp_cmd_node *c_node, - struct drm_exynos_ipp_mem_node *m_node) -{ - struct exynos_drm_ipp_ops *ops = NULL; - int ret = 0; - - DRM_DEBUG_KMS("node[%pK]\n", m_node); - - if (!m_node) { - DRM_ERROR("invalid queue node.\n"); - return -EFAULT; - } - - DRM_DEBUG_KMS("ops_id[%d]\n", m_node->ops_id); - - /* get operations callback */ - ops = ippdrv->ops[m_node->ops_id]; - if (!ops) { - DRM_ERROR("not support ops.\n"); - return -EFAULT; - } - - /* set address and enable irq */ - if (ops->set_addr) { - ret = ops->set_addr(ippdrv->dev, &m_node->buf_info, - m_node->buf_id, IPP_BUF_ENQUEUE); - if (ret) { - DRM_ERROR("failed to set addr.\n"); - return ret; - } - } - - return ret; -} - -static void ipp_handle_cmd_work(struct device *dev, - struct exynos_drm_ippdrv *ippdrv, - struct drm_exynos_ipp_cmd_work *cmd_work, - struct drm_exynos_ipp_cmd_node *c_node) -{ - struct ipp_context *ctx = get_ipp_context(dev); - - cmd_work->ippdrv = ippdrv; - cmd_work->c_node = c_node; - queue_work(ctx->cmd_workq, &cmd_work->work); -} - -static int ipp_queue_buf_with_run(struct device *dev, - struct drm_exynos_ipp_cmd_node *c_node, - struct drm_exynos_ipp_mem_node *m_node, - struct drm_exynos_ipp_queue_buf *qbuf) -{ - struct exynos_drm_ippdrv *ippdrv; - struct drm_exynos_ipp_property *property; - struct exynos_drm_ipp_ops *ops; - int ret; - - ippdrv = ipp_find_drv_by_handle(qbuf->prop_id); - if (IS_ERR(ippdrv)) { - DRM_ERROR("failed to get ipp driver.\n"); - return -EFAULT; - } - - ops = ippdrv->ops[qbuf->ops_id]; - if (!ops) { - DRM_ERROR("failed to get ops.\n"); - return -EFAULT; - } - - property = &c_node->property; - - if (c_node->state != IPP_STATE_START) { - DRM_DEBUG_KMS("bypass for invalid state.\n"); - return 0; - } - - mutex_lock(&c_node->mem_lock); - if (!ipp_check_mem_list(c_node)) { - mutex_unlock(&c_node->mem_lock); - DRM_DEBUG_KMS("empty memory.\n"); - return 0; - } - - /* - * If set destination buffer and enabled clock, - * then m2m operations need start operations at queue_buf - */ - if (ipp_is_m2m_cmd(property->cmd)) { - struct drm_exynos_ipp_cmd_work *cmd_work = c_node->start_work; - - cmd_work->ctrl = IPP_CTRL_PLAY; - ipp_handle_cmd_work(dev, ippdrv, cmd_work, c_node); - } else { - ret = ipp_set_mem_node(ippdrv, c_node, m_node); - if (ret) { - mutex_unlock(&c_node->mem_lock); - DRM_ERROR("failed to set m node.\n"); - return ret; - } - } - mutex_unlock(&c_node->mem_lock); - - return 0; -} - -static void ipp_clean_queue_buf(struct drm_device *drm_dev, - struct drm_exynos_ipp_cmd_node *c_node, - struct drm_exynos_ipp_queue_buf *qbuf) -{ - struct drm_exynos_ipp_mem_node *m_node, *tm_node; - - /* delete list */ - mutex_lock(&c_node->mem_lock); - list_for_each_entry_safe(m_node, tm_node, - &c_node->mem_list[qbuf->ops_id], list) { - if (m_node->buf_id == qbuf->buf_id && - m_node->ops_id == qbuf->ops_id) - ipp_put_mem_node(drm_dev, c_node, m_node); - } - mutex_unlock(&c_node->mem_lock); -} - -int exynos_drm_ipp_queue_buf(struct drm_device *drm_dev, void *data, - struct drm_file *file) -{ - struct drm_exynos_file_private *file_priv = file->driver_priv; - struct device *dev = file_priv->ipp_dev; - struct ipp_context *ctx = get_ipp_context(dev); - struct drm_exynos_ipp_queue_buf *qbuf = data; - struct drm_exynos_ipp_cmd_node *c_node; - struct drm_exynos_ipp_mem_node *m_node; - int ret; - - if (!qbuf) { - DRM_ERROR("invalid buf parameter.\n"); - return -EINVAL; - } - - if (qbuf->ops_id >= EXYNOS_DRM_OPS_MAX) { - DRM_ERROR("invalid ops parameter.\n"); - return -EINVAL; - } - - DRM_DEBUG_KMS("prop_id[%d]ops_id[%s]buf_id[%d]buf_type[%d]\n", - qbuf->prop_id, qbuf->ops_id ? "dst" : "src", - qbuf->buf_id, qbuf->buf_type); - - /* find command node */ - c_node = ipp_find_obj(&ctx->prop_idr, &ctx->prop_lock, - qbuf->prop_id); - if (!c_node || c_node->filp != file) { - DRM_ERROR("failed to get command node.\n"); - return -ENODEV; - } - - /* buffer control */ - switch (qbuf->buf_type) { - case IPP_BUF_ENQUEUE: - /* get memory node */ - m_node = ipp_get_mem_node(drm_dev, c_node, qbuf); - if (IS_ERR(m_node)) { - DRM_ERROR("failed to get m_node.\n"); - return PTR_ERR(m_node); - } - - /* - * first step get event for destination buffer. - * and second step when M2M case run with destination buffer - * if needed. - */ - if (qbuf->ops_id == EXYNOS_DRM_OPS_DST) { - /* get event for destination buffer */ - ret = ipp_get_event(drm_dev, c_node, qbuf); - if (ret) { - DRM_ERROR("failed to get event.\n"); - goto err_clean_node; - } - - /* - * M2M case run play control for streaming feature. - * other case set address and waiting. - */ - ret = ipp_queue_buf_with_run(dev, c_node, m_node, qbuf); - if (ret) { - DRM_ERROR("failed to run command.\n"); - goto err_clean_node; - } - } - break; - case IPP_BUF_DEQUEUE: - mutex_lock(&c_node->lock); - - /* put event for destination buffer */ - if (qbuf->ops_id == EXYNOS_DRM_OPS_DST) - ipp_put_event(c_node, qbuf); - - ipp_clean_queue_buf(drm_dev, c_node, qbuf); - - mutex_unlock(&c_node->lock); - break; - default: - DRM_ERROR("invalid buffer control.\n"); - return -EINVAL; - } - - return 0; - -err_clean_node: - DRM_ERROR("clean memory nodes.\n"); - - ipp_clean_queue_buf(drm_dev, c_node, qbuf); - return ret; -} - -static bool exynos_drm_ipp_check_valid(struct device *dev, - enum drm_exynos_ipp_ctrl ctrl, enum drm_exynos_ipp_state state) -{ - if (ctrl != IPP_CTRL_PLAY) { - if (pm_runtime_suspended(dev)) { - DRM_ERROR("pm:runtime_suspended.\n"); - goto err_status; - } - } - - switch (ctrl) { - case IPP_CTRL_PLAY: - if (state != IPP_STATE_IDLE) - goto err_status; - break; - case IPP_CTRL_STOP: - if (state == IPP_STATE_STOP) - goto err_status; - break; - case IPP_CTRL_PAUSE: - if (state != IPP_STATE_START) - goto err_status; - break; - case IPP_CTRL_RESUME: - if (state != IPP_STATE_STOP) - goto err_status; - break; - default: - DRM_ERROR("invalid state.\n"); - goto err_status; - } - - return true; - -err_status: - DRM_ERROR("invalid status:ctrl[%d]state[%d]\n", ctrl, state); - return false; -} - -int exynos_drm_ipp_cmd_ctrl(struct drm_device *drm_dev, void *data, - struct drm_file *file) -{ - struct drm_exynos_file_private *file_priv = file->driver_priv; - struct exynos_drm_ippdrv *ippdrv = NULL; - struct device *dev = file_priv->ipp_dev; - struct ipp_context *ctx = get_ipp_context(dev); - struct drm_exynos_ipp_cmd_ctrl *cmd_ctrl = data; - struct drm_exynos_ipp_cmd_work *cmd_work; - struct drm_exynos_ipp_cmd_node *c_node; - - if (!ctx) { - DRM_ERROR("invalid context.\n"); - return -EINVAL; - } - - if (!cmd_ctrl) { - DRM_ERROR("invalid control parameter.\n"); - return -EINVAL; - } - - DRM_DEBUG_KMS("ctrl[%d]prop_id[%d]\n", - cmd_ctrl->ctrl, cmd_ctrl->prop_id); - - ippdrv = ipp_find_drv_by_handle(cmd_ctrl->prop_id); - if (IS_ERR(ippdrv)) { - DRM_ERROR("failed to get ipp driver.\n"); - return PTR_ERR(ippdrv); - } - - c_node = ipp_find_obj(&ctx->prop_idr, &ctx->prop_lock, - cmd_ctrl->prop_id); - if (!c_node || c_node->filp != file) { - DRM_ERROR("invalid command node list.\n"); - return -ENODEV; - } - - if (!exynos_drm_ipp_check_valid(ippdrv->dev, cmd_ctrl->ctrl, - c_node->state)) { - DRM_ERROR("invalid state.\n"); - return -EINVAL; - } - - switch (cmd_ctrl->ctrl) { - case IPP_CTRL_PLAY: - if (pm_runtime_suspended(ippdrv->dev)) - pm_runtime_get_sync(ippdrv->dev); - - c_node->state = IPP_STATE_START; - - cmd_work = c_node->start_work; - cmd_work->ctrl = cmd_ctrl->ctrl; - ipp_handle_cmd_work(dev, ippdrv, cmd_work, c_node); - break; - case IPP_CTRL_STOP: - cmd_work = c_node->stop_work; - cmd_work->ctrl = cmd_ctrl->ctrl; - ipp_handle_cmd_work(dev, ippdrv, cmd_work, c_node); - - if (!wait_for_completion_timeout(&c_node->stop_complete, - msecs_to_jiffies(300))) { - DRM_ERROR("timeout stop:prop_id[%d]\n", - c_node->property.prop_id); - } - - c_node->state = IPP_STATE_STOP; - ippdrv->dedicated = false; - mutex_lock(&ippdrv->cmd_lock); - ipp_clean_cmd_node(ctx, c_node); - - if (list_empty(&ippdrv->cmd_list)) - pm_runtime_put_sync(ippdrv->dev); - mutex_unlock(&ippdrv->cmd_lock); - break; - case IPP_CTRL_PAUSE: - cmd_work = c_node->stop_work; - cmd_work->ctrl = cmd_ctrl->ctrl; - ipp_handle_cmd_work(dev, ippdrv, cmd_work, c_node); - - if (!wait_for_completion_timeout(&c_node->stop_complete, - msecs_to_jiffies(200))) { - DRM_ERROR("timeout stop:prop_id[%d]\n", - c_node->property.prop_id); - } - - c_node->state = IPP_STATE_STOP; - break; - case IPP_CTRL_RESUME: - c_node->state = IPP_STATE_START; - cmd_work = c_node->start_work; - cmd_work->ctrl = cmd_ctrl->ctrl; - ipp_handle_cmd_work(dev, ippdrv, cmd_work, c_node); - break; - default: - DRM_ERROR("could not support this state currently.\n"); - return -EINVAL; - } - - DRM_DEBUG_KMS("done ctrl[%d]prop_id[%d]\n", - cmd_ctrl->ctrl, cmd_ctrl->prop_id); - - return 0; -} - -int exynos_drm_ippnb_register(struct notifier_block *nb) -{ - return blocking_notifier_chain_register( - &exynos_drm_ippnb_list, nb); -} - -int exynos_drm_ippnb_unregister(struct notifier_block *nb) -{ - return blocking_notifier_chain_unregister( - &exynos_drm_ippnb_list, nb); -} - -int exynos_drm_ippnb_send_event(unsigned long val, void *v) -{ - return blocking_notifier_call_chain( - &exynos_drm_ippnb_list, val, v); -} - -static int ipp_set_property(struct exynos_drm_ippdrv *ippdrv, - struct drm_exynos_ipp_property *property) -{ - struct exynos_drm_ipp_ops *ops = NULL; - bool swap = false; - int ret, i; - - if (!property) { - DRM_ERROR("invalid property parameter.\n"); - return -EINVAL; - } - - DRM_DEBUG_KMS("prop_id[%d]\n", property->prop_id); - - /* reset h/w block */ - if (ippdrv->reset && - ippdrv->reset(ippdrv->dev)) { - return -EINVAL; - } - - /* set source,destination operations */ - for_each_ipp_ops(i) { - struct drm_exynos_ipp_config *config = - &property->config[i]; - - ops = ippdrv->ops[i]; - if (!ops || !config) { - DRM_ERROR("not support ops and config.\n"); - return -EINVAL; - } - - /* set format */ - if (ops->set_fmt) { - ret = ops->set_fmt(ippdrv->dev, config->fmt); - if (ret) - return ret; - } - - /* set transform for rotation, flip */ - if (ops->set_transf) { - ret = ops->set_transf(ippdrv->dev, config->degree, - config->flip, &swap); - if (ret) - return ret; - } - - /* set size */ - if (ops->set_size) { - ret = ops->set_size(ippdrv->dev, swap, &config->pos, - &config->sz); - if (ret) - return ret; - } - } - - return 0; -} - -static int ipp_start_property(struct exynos_drm_ippdrv *ippdrv, - struct drm_exynos_ipp_cmd_node *c_node) -{ - struct drm_exynos_ipp_mem_node *m_node; - struct drm_exynos_ipp_property *property = &c_node->property; - struct list_head *head; - int ret, i; - - DRM_DEBUG_KMS("prop_id[%d]\n", property->prop_id); - - /* store command info in ippdrv */ - ippdrv->c_node = c_node; - - mutex_lock(&c_node->mem_lock); - if (!ipp_check_mem_list(c_node)) { - DRM_DEBUG_KMS("empty memory.\n"); - ret = -ENOMEM; - goto err_unlock; - } - - /* set current property in ippdrv */ - ret = ipp_set_property(ippdrv, property); - if (ret) { - DRM_ERROR("failed to set property.\n"); - ippdrv->c_node = NULL; - goto err_unlock; - } - - /* check command */ - switch (property->cmd) { - case IPP_CMD_M2M: - for_each_ipp_ops(i) { - /* source/destination memory list */ - head = &c_node->mem_list[i]; - - m_node = list_first_entry(head, - struct drm_exynos_ipp_mem_node, list); - - DRM_DEBUG_KMS("m_node[%pK]\n", m_node); - - ret = ipp_set_mem_node(ippdrv, c_node, m_node); - if (ret) { - DRM_ERROR("failed to set m node.\n"); - goto err_unlock; - } - } - break; - case IPP_CMD_WB: - /* destination memory list */ - head = &c_node->mem_list[EXYNOS_DRM_OPS_DST]; - - list_for_each_entry(m_node, head, list) { - ret = ipp_set_mem_node(ippdrv, c_node, m_node); - if (ret) { - DRM_ERROR("failed to set m node.\n"); - goto err_unlock; - } - } - break; - case IPP_CMD_OUTPUT: - /* source memory list */ - head = &c_node->mem_list[EXYNOS_DRM_OPS_SRC]; - - list_for_each_entry(m_node, head, list) { - ret = ipp_set_mem_node(ippdrv, c_node, m_node); - if (ret) { - DRM_ERROR("failed to set m node.\n"); - goto err_unlock; - } - } - break; - default: - DRM_ERROR("invalid operations.\n"); - ret = -EINVAL; - goto err_unlock; - } - mutex_unlock(&c_node->mem_lock); - - DRM_DEBUG_KMS("cmd[%d]\n", property->cmd); - - /* start operations */ - if (ippdrv->start) { - ret = ippdrv->start(ippdrv->dev, property->cmd); - if (ret) { - DRM_ERROR("failed to start ops.\n"); - ippdrv->c_node = NULL; - return ret; - } - } - - return 0; - -err_unlock: - mutex_unlock(&c_node->mem_lock); - ippdrv->c_node = NULL; - return ret; -} - -static int ipp_stop_property(struct drm_device *drm_dev, - struct exynos_drm_ippdrv *ippdrv, - struct drm_exynos_ipp_cmd_node *c_node) -{ - struct drm_exynos_ipp_property *property = &c_node->property; - int i; - - DRM_DEBUG_KMS("prop_id[%d]\n", property->prop_id); - - /* stop operations */ - if (ippdrv->stop) - ippdrv->stop(ippdrv->dev, property->cmd); - - /* check command */ - switch (property->cmd) { - case IPP_CMD_M2M: - for_each_ipp_ops(i) - ipp_clean_mem_nodes(drm_dev, c_node, i); - break; - case IPP_CMD_WB: - ipp_clean_mem_nodes(drm_dev, c_node, EXYNOS_DRM_OPS_DST); - break; - case IPP_CMD_OUTPUT: - ipp_clean_mem_nodes(drm_dev, c_node, EXYNOS_DRM_OPS_SRC); - break; - default: - DRM_ERROR("invalid operations.\n"); - return -EINVAL; - } - - return 0; -} - -void ipp_sched_cmd(struct work_struct *work) -{ - struct drm_exynos_ipp_cmd_work *cmd_work = - container_of(work, struct drm_exynos_ipp_cmd_work, work); - struct exynos_drm_ippdrv *ippdrv; - struct drm_exynos_ipp_cmd_node *c_node; - struct drm_exynos_ipp_property *property; - int ret; - - ippdrv = cmd_work->ippdrv; - if (!ippdrv) { - DRM_ERROR("invalid ippdrv list.\n"); - return; - } - - c_node = cmd_work->c_node; - if (!c_node) { - DRM_ERROR("invalid command node list.\n"); - return; - } - - mutex_lock(&c_node->lock); - - property = &c_node->property; - - switch (cmd_work->ctrl) { - case IPP_CTRL_PLAY: - case IPP_CTRL_RESUME: - ret = ipp_start_property(ippdrv, c_node); - if (ret) { - DRM_ERROR("failed to start property:prop_id[%d]\n", - c_node->property.prop_id); - goto err_unlock; - } - - /* - * M2M case supports wait_completion of transfer. - * because M2M case supports single unit operation - * with multiple queue. - * M2M need to wait completion of data transfer. - */ - if (ipp_is_m2m_cmd(property->cmd)) { - if (!wait_for_completion_timeout - (&c_node->start_complete, msecs_to_jiffies(200))) { - DRM_ERROR("timeout event:prop_id[%d]\n", - c_node->property.prop_id); - goto err_unlock; - } - } - break; - case IPP_CTRL_STOP: - case IPP_CTRL_PAUSE: - ret = ipp_stop_property(ippdrv->drm_dev, ippdrv, - c_node); - if (ret) { - DRM_ERROR("failed to stop property.\n"); - goto err_unlock; - } - - complete(&c_node->stop_complete); - break; - default: - DRM_ERROR("unknown control type\n"); - break; - } - - DRM_DEBUG_KMS("ctrl[%d] done.\n", cmd_work->ctrl); - -err_unlock: - mutex_unlock(&c_node->lock); -} - -static int ipp_send_event(struct exynos_drm_ippdrv *ippdrv, - struct drm_exynos_ipp_cmd_node *c_node, int *buf_id) -{ - struct drm_device *drm_dev = ippdrv->drm_dev; - struct drm_exynos_ipp_property *property = &c_node->property; - struct drm_exynos_ipp_mem_node *m_node; - struct drm_exynos_ipp_queue_buf qbuf; - struct drm_exynos_ipp_send_event *e; - struct list_head *head; - struct timeval now; - u32 tbuf_id[EXYNOS_DRM_OPS_MAX] = {0, }; - int ret, i; - - for_each_ipp_ops(i) - DRM_DEBUG_KMS("%s buf_id[%d]\n", i ? "dst" : "src", buf_id[i]); - - if (!drm_dev) { - DRM_ERROR("failed to get drm_dev.\n"); - return -EINVAL; - } - - if (!property) { - DRM_ERROR("failed to get property.\n"); - return -EINVAL; - } - - mutex_lock(&c_node->event_lock); - if (list_empty(&c_node->event_list)) { - DRM_DEBUG_KMS("event list is empty.\n"); - ret = 0; - goto err_event_unlock; - } - - mutex_lock(&c_node->mem_lock); - if (!ipp_check_mem_list(c_node)) { - DRM_DEBUG_KMS("empty memory.\n"); - ret = 0; - goto err_mem_unlock; - } - - /* check command */ - switch (property->cmd) { - case IPP_CMD_M2M: - for_each_ipp_ops(i) { - /* source/destination memory list */ - head = &c_node->mem_list[i]; - - m_node = list_first_entry(head, - struct drm_exynos_ipp_mem_node, list); - - tbuf_id[i] = m_node->buf_id; - DRM_DEBUG_KMS("%s buf_id[%d]\n", - i ? "dst" : "src", tbuf_id[i]); - - ret = ipp_put_mem_node(drm_dev, c_node, m_node); - if (ret) - DRM_ERROR("failed to put m_node.\n"); - } - break; - case IPP_CMD_WB: - /* clear buf for finding */ - memset(&qbuf, 0x0, sizeof(qbuf)); - qbuf.ops_id = EXYNOS_DRM_OPS_DST; - qbuf.buf_id = buf_id[EXYNOS_DRM_OPS_DST]; - - /* get memory node entry */ - m_node = ipp_find_mem_node(c_node, &qbuf); - if (!m_node) { - DRM_ERROR("empty memory node.\n"); - ret = -ENOMEM; - goto err_mem_unlock; - } - - tbuf_id[EXYNOS_DRM_OPS_DST] = m_node->buf_id; - - ret = ipp_put_mem_node(drm_dev, c_node, m_node); - if (ret) - DRM_ERROR("failed to put m_node.\n"); - break; - case IPP_CMD_OUTPUT: - /* source memory list */ - head = &c_node->mem_list[EXYNOS_DRM_OPS_SRC]; - - m_node = list_first_entry(head, - struct drm_exynos_ipp_mem_node, list); - - tbuf_id[EXYNOS_DRM_OPS_SRC] = m_node->buf_id; - - ret = ipp_put_mem_node(drm_dev, c_node, m_node); - if (ret) - DRM_ERROR("failed to put m_node.\n"); - break; - default: - DRM_ERROR("invalid operations.\n"); - ret = -EINVAL; - goto err_mem_unlock; - } - mutex_unlock(&c_node->mem_lock); - - if (tbuf_id[EXYNOS_DRM_OPS_DST] != buf_id[EXYNOS_DRM_OPS_DST]) - DRM_ERROR("failed to match buf_id[%d %d]prop_id[%d]\n", - tbuf_id[1], buf_id[1], property->prop_id); - - /* - * command node have event list of destination buffer - * If destination buffer enqueue to mem list, - * then we make event and link to event list tail. - * so, we get first event for first enqueued buffer. - */ - e = list_first_entry(&c_node->event_list, - struct drm_exynos_ipp_send_event, base.link); - - do_gettimeofday(&now); - DRM_DEBUG_KMS("tv_sec[%ld]tv_usec[%ld]\n", now.tv_sec, now.tv_usec); - e->event.tv_sec = now.tv_sec; - e->event.tv_usec = now.tv_usec; - e->event.prop_id = property->prop_id; - - /* set buffer id about source destination */ - for_each_ipp_ops(i) - e->event.buf_id[i] = tbuf_id[i]; - - drm_send_event(drm_dev, &e->base); - mutex_unlock(&c_node->event_lock); - - DRM_DEBUG_KMS("done cmd[%d]prop_id[%d]buf_id[%d]\n", - property->cmd, property->prop_id, tbuf_id[EXYNOS_DRM_OPS_DST]); - - return 0; - -err_mem_unlock: - mutex_unlock(&c_node->mem_lock); -err_event_unlock: - mutex_unlock(&c_node->event_lock); - return ret; -} - -void ipp_sched_event(struct work_struct *work) -{ - struct drm_exynos_ipp_event_work *event_work = - container_of(work, struct drm_exynos_ipp_event_work, work); - struct exynos_drm_ippdrv *ippdrv; - struct drm_exynos_ipp_cmd_node *c_node; - int ret; - - if (!event_work) { - DRM_ERROR("failed to get event_work.\n"); - return; - } - - DRM_DEBUG_KMS("buf_id[%d]\n", event_work->buf_id[EXYNOS_DRM_OPS_DST]); - - ippdrv = event_work->ippdrv; - if (!ippdrv) { - DRM_ERROR("failed to get ipp driver.\n"); - return; - } - - c_node = ippdrv->c_node; - if (!c_node) { - DRM_ERROR("failed to get command node.\n"); - return; - } - - /* - * IPP supports command thread, event thread synchronization. - * If IPP close immediately from user land, then IPP make - * synchronization with command thread, so make complete event. - * or going out operations. - */ - if (c_node->state != IPP_STATE_START) { - DRM_DEBUG_KMS("bypass state[%d]prop_id[%d]\n", - c_node->state, c_node->property.prop_id); - goto err_completion; - } - - ret = ipp_send_event(ippdrv, c_node, event_work->buf_id); - if (ret) { - DRM_ERROR("failed to send event.\n"); - goto err_completion; - } - -err_completion: - if (ipp_is_m2m_cmd(c_node->property.cmd)) - complete(&c_node->start_complete); -} - -static int ipp_subdrv_probe(struct drm_device *drm_dev, struct device *dev) -{ - struct ipp_context *ctx = get_ipp_context(dev); - struct exynos_drm_ippdrv *ippdrv; - int ret, count = 0; - - /* get ipp driver entry */ - list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) { - ippdrv->drm_dev = drm_dev; - - ret = ipp_create_id(&ctx->ipp_idr, &ctx->ipp_lock, ippdrv); - if (ret < 0) { - DRM_ERROR("failed to create id.\n"); - goto err; - } - ippdrv->prop_list.ipp_id = ret; - - DRM_DEBUG_KMS("count[%d]ippdrv[%pK]ipp_id[%d]\n", - count++, ippdrv, ret); - - /* store parent device for node */ - ippdrv->parent_dev = dev; - - /* store event work queue and handler */ - ippdrv->event_workq = ctx->event_workq; - ippdrv->sched_event = ipp_sched_event; - INIT_LIST_HEAD(&ippdrv->cmd_list); - mutex_init(&ippdrv->cmd_lock); - - ret = drm_iommu_attach_device(drm_dev, ippdrv->dev); - if (ret) { - DRM_ERROR("failed to activate iommu\n"); - goto err; - } - } - - return 0; - -err: - /* get ipp driver entry */ - list_for_each_entry_continue_reverse(ippdrv, &exynos_drm_ippdrv_list, - drv_list) { - drm_iommu_detach_device(drm_dev, ippdrv->dev); - - ipp_remove_id(&ctx->ipp_idr, &ctx->ipp_lock, - ippdrv->prop_list.ipp_id); - } - - return ret; -} - -static void ipp_subdrv_remove(struct drm_device *drm_dev, struct device *dev) -{ - struct exynos_drm_ippdrv *ippdrv, *t; - struct ipp_context *ctx = get_ipp_context(dev); - - /* get ipp driver entry */ - list_for_each_entry_safe(ippdrv, t, &exynos_drm_ippdrv_list, drv_list) { - drm_iommu_detach_device(drm_dev, ippdrv->dev); - - ipp_remove_id(&ctx->ipp_idr, &ctx->ipp_lock, - ippdrv->prop_list.ipp_id); - - ippdrv->drm_dev = NULL; - exynos_drm_ippdrv_unregister(ippdrv); - } -} - -static int ipp_subdrv_open(struct drm_device *drm_dev, struct device *dev, - struct drm_file *file) -{ - struct drm_exynos_file_private *file_priv = file->driver_priv; - - file_priv->ipp_dev = dev; - - DRM_DEBUG_KMS("done priv[%pK]\n", dev); - - return 0; -} - -static void ipp_subdrv_close(struct drm_device *drm_dev, struct device *dev, - struct drm_file *file) -{ - struct exynos_drm_ippdrv *ippdrv = NULL; - struct ipp_context *ctx = get_ipp_context(dev); - struct drm_exynos_ipp_cmd_node *c_node, *tc_node; - int count = 0; - - list_for_each_entry(ippdrv, &exynos_drm_ippdrv_list, drv_list) { - mutex_lock(&ippdrv->cmd_lock); - list_for_each_entry_safe(c_node, tc_node, - &ippdrv->cmd_list, list) { - DRM_DEBUG_KMS("count[%d]ippdrv[%pK]\n", - count++, ippdrv); - - if (c_node->filp == file) { - /* - * userland goto unnormal state. process killed. - * and close the file. - * so, IPP didn't called stop cmd ctrl. - * so, we are make stop operation in this state. - */ - if (c_node->state == IPP_STATE_START) { - ipp_stop_property(drm_dev, ippdrv, - c_node); - c_node->state = IPP_STATE_STOP; - } - - ippdrv->dedicated = false; - ipp_clean_cmd_node(ctx, c_node); - if (list_empty(&ippdrv->cmd_list)) - pm_runtime_put_sync(ippdrv->dev); - } - } - mutex_unlock(&ippdrv->cmd_lock); - } - - return; -} - -static int ipp_probe(struct platform_device *pdev) -{ - struct device *dev = &pdev->dev; - struct ipp_context *ctx; - struct exynos_drm_subdrv *subdrv; - int ret; - - ctx = devm_kzalloc(dev, sizeof(*ctx), GFP_KERNEL); - if (!ctx) - return -ENOMEM; - - mutex_init(&ctx->ipp_lock); - mutex_init(&ctx->prop_lock); - - idr_init(&ctx->ipp_idr); - idr_init(&ctx->prop_idr); - - /* - * create single thread for ipp event - * IPP supports event thread for IPP drivers. - * IPP driver send event_work to this thread. - * and IPP event thread send event to user process. - */ - ctx->event_workq = create_singlethread_workqueue("ipp_event"); - if (!ctx->event_workq) { - dev_err(dev, "failed to create event workqueue\n"); - return -EINVAL; - } - - /* - * create single thread for ipp command - * IPP supports command thread for user process. - * user process make command node using set property ioctl. - * and make start_work and send this work to command thread. - * and then this command thread start property. - */ - ctx->cmd_workq = create_singlethread_workqueue("ipp_cmd"); - if (!ctx->cmd_workq) { - dev_err(dev, "failed to create cmd workqueue\n"); - ret = -EINVAL; - goto err_event_workq; - } - - /* set sub driver informations */ - subdrv = &ctx->subdrv; - subdrv->dev = dev; - subdrv->probe = ipp_subdrv_probe; - subdrv->remove = ipp_subdrv_remove; - subdrv->open = ipp_subdrv_open; - subdrv->close = ipp_subdrv_close; - - platform_set_drvdata(pdev, ctx); - - ret = exynos_drm_subdrv_register(subdrv); - if (ret < 0) { - DRM_ERROR("failed to register drm ipp device.\n"); - goto err_cmd_workq; - } - - dev_info(dev, "drm ipp registered successfully.\n"); - - return 0; - -err_cmd_workq: - destroy_workqueue(ctx->cmd_workq); -err_event_workq: - destroy_workqueue(ctx->event_workq); - return ret; -} - -static int ipp_remove(struct platform_device *pdev) -{ - struct ipp_context *ctx = platform_get_drvdata(pdev); - - /* unregister sub driver */ - exynos_drm_subdrv_unregister(&ctx->subdrv); - - /* remove,destroy ipp idr */ - idr_destroy(&ctx->ipp_idr); - idr_destroy(&ctx->prop_idr); - - mutex_destroy(&ctx->ipp_lock); - mutex_destroy(&ctx->prop_lock); - - /* destroy command, event work queue */ - destroy_workqueue(ctx->cmd_workq); - destroy_workqueue(ctx->event_workq); - - return 0; -} - -struct platform_driver ipp_driver = { - .probe = ipp_probe, - .remove = ipp_remove, - .driver = { - .name = "exynos-drm-ipp", - .owner = THIS_MODULE, - }, -}; - diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.h b/drivers/gpu/drm/exynos/exynos_drm_ipp.h deleted file mode 100644 index 2a61547a39d0..000000000000 --- a/drivers/gpu/drm/exynos/exynos_drm_ipp.h +++ /dev/null @@ -1,252 +0,0 @@ -/* - * Copyright (c) 2012 Samsung Electronics Co., Ltd. - * - * Authors: - * Eunchul Kim <chulspro.kim@samsung.com> - * Jinyoung Jeon <jy0.jeon@samsung.com> - * Sangmin Lee <lsmin.lee@samsung.com> - * - * This program is free software; you can redistribute it and/or modify it - * under the terms of the GNU General Public License as published by the - * Free Software Foundation; either version 2 of the License, or (at your - * option) any later version. - */ - -#ifndef _EXYNOS_DRM_IPP_H_ -#define _EXYNOS_DRM_IPP_H_ - -#define for_each_ipp_ops(pos) \ - for (pos = 0; pos < EXYNOS_DRM_OPS_MAX; pos++) -#define for_each_ipp_planar(pos) \ - for (pos = 0; pos < EXYNOS_DRM_PLANAR_MAX; pos++) - -#define IPP_GET_LCD_WIDTH _IOR('F', 302, int) -#define IPP_GET_LCD_HEIGHT _IOR('F', 303, int) -#define IPP_SET_WRITEBACK _IOW('F', 304, u32) - -/* definition of state */ -enum drm_exynos_ipp_state { - IPP_STATE_IDLE, - IPP_STATE_START, - IPP_STATE_STOP, -}; - -/* - * A structure of command work information. - * @work: work structure. - * @ippdrv: current work ippdrv. - * @c_node: command node information. - * @ctrl: command control. - */ -struct drm_exynos_ipp_cmd_work { - struct work_struct work; - struct exynos_drm_ippdrv *ippdrv; - struct drm_exynos_ipp_cmd_node *c_node; - enum drm_exynos_ipp_ctrl ctrl; -}; - -/* - * A structure of command node. - * - * @list: list head to command queue information. - * @event_list: list head of event. - * @mem_list: list head to source,destination memory queue information. - * @lock: lock for synchronization of access to ioctl. - * @mem_lock: lock for synchronization of access to memory nodes. - * @event_lock: lock for synchronization of access to scheduled event. - * @start_complete: completion of start of command. - * @stop_complete: completion of stop of command. - * @property: property information. - * @start_work: start command work structure. - * @stop_work: stop command work structure. - * @event_work: event work structure. - * @state: state of command node. - * @filp: associated file pointer. - */ -struct drm_exynos_ipp_cmd_node { - struct list_head list; - struct list_head event_list; - struct list_head mem_list[EXYNOS_DRM_OPS_MAX]; - struct mutex lock; - struct mutex mem_lock; - struct mutex event_lock; - struct completion start_complete; - struct completion stop_complete; - struct drm_exynos_ipp_property property; - struct drm_exynos_ipp_cmd_work *start_work; - struct drm_exynos_ipp_cmd_work *stop_work; - struct drm_exynos_ipp_event_work *event_work; - enum drm_exynos_ipp_state state; - struct drm_file *filp; -}; - -/* - * A structure of buffer information. - * - * @handles: Y, Cb, Cr each gem object handle. - * @base: Y, Cb, Cr each planar address. - */ -struct drm_exynos_ipp_buf_info { - unsigned long handles[EXYNOS_DRM_PLANAR_MAX]; - dma_addr_t base[EXYNOS_DRM_PLANAR_MAX]; -}; - -/* - * A structure of wb setting information. - * - * @enable: enable flag for wb. - * @refresh: HZ of the refresh rate. - */ -struct drm_exynos_ipp_set_wb { - __u32 enable; - __u32 refresh; -}; - -/* - * A structure of event work information. - * - * @work: work structure. - * @ippdrv: current work ippdrv. - * @buf_id: id of src, dst buffer. - */ -struct drm_exynos_ipp_event_work { - struct work_struct work; - struct exynos_drm_ippdrv *ippdrv; - u32 buf_id[EXYNOS_DRM_OPS_MAX]; -}; - -/* - * A structure of source,destination operations. - * - * @set_fmt: set format of image. - * @set_transf: set transform(rotations, flip). - * @set_size: set size of region. - * @set_addr: set address for dma. - */ -struct exynos_drm_ipp_ops { - int (*set_fmt)(struct device *dev, u32 fmt); - int (*set_transf)(struct device *dev, - enum drm_exynos_degree degree, - enum drm_exynos_flip flip, bool *swap); - int (*set_size)(struct device *dev, int swap, - struct drm_exynos_pos *pos, struct drm_exynos_sz *sz); - int (*set_addr)(struct device *dev, - struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id, - enum drm_exynos_ipp_buf_type buf_type); -}; - -/* - * A structure of ipp driver. - * - * @drv_list: list head for registed sub driver information. - * @parent_dev: parent device information. - * @dev: platform device. - * @drm_dev: drm device. - * @dedicated: dedicated ipp device. - * @ops: source, destination operations. - * @event_workq: event work queue. - * @c_node: current command information. - * @cmd_list: list head for command information. - * @cmd_lock: lock for synchronization of access to cmd_list. - * @prop_list: property informations of current ipp driver. - * @check_property: check property about format, size, buffer. - * @reset: reset ipp block. - * @start: ipp each device start. - * @stop: ipp each device stop. - * @sched_event: work schedule handler. - */ -struct exynos_drm_ippdrv { - struct list_head drv_list; - struct device *parent_dev; - struct device *dev; - struct drm_device *drm_dev; - bool dedicated; - struct exynos_drm_ipp_ops *ops[EXYNOS_DRM_OPS_MAX]; - struct workqueue_struct *event_workq; - struct drm_exynos_ipp_cmd_node *c_node; - struct list_head cmd_list; - struct mutex cmd_lock; - struct drm_exynos_ipp_prop_list prop_list; - - int (*check_property)(struct device *dev, - struct drm_exynos_ipp_property *property); - int (*reset)(struct device *dev); - int (*start)(struct device *dev, enum drm_exynos_ipp_cmd cmd); - void (*stop)(struct device *dev, enum drm_exynos_ipp_cmd cmd); - void (*sched_event)(struct work_struct *work); -}; - -#ifdef CONFIG_DRM_EXYNOS_IPP -extern int exynos_drm_ippdrv_register(struct exynos_drm_ippdrv *ippdrv); -extern int exynos_drm_ippdrv_unregister(struct exynos_drm_ippdrv *ippdrv); -extern int exynos_drm_ipp_get_property(struct drm_device *drm_dev, void *data, - struct drm_file *file); -extern int exynos_drm_ipp_set_property(struct drm_device *drm_dev, void *data, - struct drm_file *file); -extern int exynos_drm_ipp_queue_buf(struct drm_device *drm_dev, void *data, - struct drm_file *file); -extern int exynos_drm_ipp_cmd_ctrl(struct drm_device *drm_dev, void *data, - struct drm_file *file); -extern int exynos_drm_ippnb_register(struct notifier_block *nb); -extern int exynos_drm_ippnb_unregister(struct notifier_block *nb); -extern int exynos_drm_ippnb_send_event(unsigned long val, void *v); -extern void ipp_sched_cmd(struct work_struct *work); -extern void ipp_sched_event(struct work_struct *work); - -#else -static inline int exynos_drm_ippdrv_register(struct exynos_drm_ippdrv *ippdrv) -{ - return -ENODEV; -} - -static inline int exynos_drm_ippdrv_unregister(struct exynos_drm_ippdrv *ippdrv) -{ - return -ENODEV; -} - -static inline int exynos_drm_ipp_get_property(struct drm_device *drm_dev, - void *data, - struct drm_file *file_priv) -{ - return -ENOTTY; -} - -static inline int exynos_drm_ipp_set_property(struct drm_device *drm_dev, - void *data, - struct drm_file *file_priv) -{ - return -ENOTTY; -} - -static inline int exynos_drm_ipp_queue_buf(struct drm_device *drm_dev, - void *data, - struct drm_file *file) -{ - return -ENOTTY; -} - -static inline int exynos_drm_ipp_cmd_ctrl(struct drm_device *drm_dev, - void *data, - struct drm_file *file) -{ - return -ENOTTY; -} - -static inline int exynos_drm_ippnb_register(struct notifier_block *nb) -{ - return -ENODEV; -} - -static inline int exynos_drm_ippnb_unregister(struct notifier_block *nb) -{ - return -ENODEV; -} - -static inline int exynos_drm_ippnb_send_event(unsigned long val, void *v) -{ - return -ENOTTY; -} -#endif - -#endif /* _EXYNOS_DRM_IPP_H_ */ - diff --git a/include/uapi/drm/exynos_drm.h b/include/uapi/drm/exynos_drm.h index cb3e9f9d029f..5d9a08d69ba7 100644 --- a/include/uapi/drm/exynos_drm.h +++ b/include/uapi/drm/exynos_drm.h @@ -134,172 +134,6 @@ struct drm_exynos_g2d_exec { __u64 async; }; -enum drm_exynos_ops_id { - EXYNOS_DRM_OPS_SRC, - EXYNOS_DRM_OPS_DST, - EXYNOS_DRM_OPS_MAX, -}; - -struct drm_exynos_sz { - __u32 hsize; - __u32 vsize; -}; - -struct drm_exynos_pos { - __u32 x; - __u32 y; - __u32 w; - __u32 h; -}; - -enum drm_exynos_flip { - EXYNOS_DRM_FLIP_NONE = (0 << 0), - EXYNOS_DRM_FLIP_VERTICAL = (1 << 0), - EXYNOS_DRM_FLIP_HORIZONTAL = (1 << 1), - EXYNOS_DRM_FLIP_BOTH = EXYNOS_DRM_FLIP_VERTICAL | - EXYNOS_DRM_FLIP_HORIZONTAL, -}; - -enum drm_exynos_degree { - EXYNOS_DRM_DEGREE_0, - EXYNOS_DRM_DEGREE_90, - EXYNOS_DRM_DEGREE_180, - EXYNOS_DRM_DEGREE_270, -}; - -enum drm_exynos_planer { - EXYNOS_DRM_PLANAR_Y, - EXYNOS_DRM_PLANAR_CB, - EXYNOS_DRM_PLANAR_CR, - EXYNOS_DRM_PLANAR_MAX, -}; - -/** - * A structure for ipp supported property list. - * - * @version: version of this structure. - * @ipp_id: id of ipp driver. - * @count: count of ipp driver. - * @writeback: flag of writeback supporting. - * @flip: flag of flip supporting. - * @degree: flag of degree information. - * @csc: flag of csc supporting. - * @crop: flag of crop supporting. - * @scale: flag of scale supporting. - * @refresh_min: min hz of refresh. - * @refresh_max: max hz of refresh. - * @crop_min: crop min resolution. - * @crop_max: crop max resolution. - * @scale_min: scale min resolution. - * @scale_max: scale max resolution. - */ -struct drm_exynos_ipp_prop_list { - __u32 version; - __u32 ipp_id; - __u32 count; - __u32 writeback; - __u32 flip; - __u32 degree; - __u32 csc; - __u32 crop; - __u32 scale; - __u32 refresh_min; - __u32 refresh_max; - __u32 reserved; - struct drm_exynos_sz crop_min; - struct drm_exynos_sz crop_max; - struct drm_exynos_sz scale_min; - struct drm_exynos_sz scale_max; -}; - -/** - * A structure for ipp config. - * - * @ops_id: property of operation directions. - * @flip: property of mirror, flip. - * @degree: property of rotation degree. - * @fmt: property of image format. - * @sz: property of image size. - * @pos: property of image position(src-cropped,dst-scaler). - */ -struct drm_exynos_ipp_config { - __u32 ops_id; - __u32 flip; - __u32 degree; - __u32 fmt; - struct drm_exynos_sz sz; - struct drm_exynos_pos pos; -}; - -enum drm_exynos_ipp_cmd { - IPP_CMD_NONE, - IPP_CMD_M2M, - IPP_CMD_WB, - IPP_CMD_OUTPUT, - IPP_CMD_MAX, -}; - -/** - * A structure for ipp property. - * - * @config: source, destination config. - * @cmd: definition of command. - * @ipp_id: id of ipp driver. - * @prop_id: id of property. - * @refresh_rate: refresh rate. - */ -struct drm_exynos_ipp_property { - struct drm_exynos_ipp_config config[EXYNOS_DRM_OPS_MAX]; - __u32 cmd; - __u32 ipp_id; - __u32 prop_id; - __u32 refresh_rate; -}; - -enum drm_exynos_ipp_buf_type { - IPP_BUF_ENQUEUE, - IPP_BUF_DEQUEUE, -}; - -/** - * A structure for ipp buffer operations. - * - * @ops_id: operation directions. - * @buf_type: definition of buffer. - * @prop_id: id of property. - * @buf_id: id of buffer. - * @handle: Y, Cb, Cr each planar handle. - * @user_data: user data. - */ -struct drm_exynos_ipp_queue_buf { - __u32 ops_id; - __u32 buf_type; - __u32 prop_id; - __u32 buf_id; - __u32 handle[EXYNOS_DRM_PLANAR_MAX]; - __u32 reserved; - __u64 user_data; -}; - -enum drm_exynos_ipp_ctrl { - IPP_CTRL_PLAY, - IPP_CTRL_STOP, - IPP_CTRL_PAUSE, - IPP_CTRL_RESUME, - IPP_CTRL_MAX, -}; - -/** - * A structure for ipp start/stop operations. - * - * @prop_id: id of property. - * @ctrl: definition of control. - */ -struct drm_exynos_ipp_cmd_ctrl { - __u32 prop_id; - __u32 ctrl; -}; - #define DRM_EXYNOS_GEM_CREATE 0x00 #define DRM_EXYNOS_GEM_MAP 0x01 /* Reserved 0x03 ~ 0x05 for exynos specific gem ioctl */ @@ -311,11 +145,7 @@ struct drm_exynos_ipp_cmd_ctrl { #define DRM_EXYNOS_G2D_SET_CMDLIST 0x21 #define DRM_EXYNOS_G2D_EXEC 0x22 -/* IPP - Image Post Processing */ -#define DRM_EXYNOS_IPP_GET_PROPERTY 0x30 -#define DRM_EXYNOS_IPP_SET_PROPERTY 0x31 -#define DRM_EXYNOS_IPP_QUEUE_BUF 0x32 -#define DRM_EXYNOS_IPP_CMD_CTRL 0x33 +/* Reserved 0x30 ~ 0x33 for obsolete Exynos IPP ioctls */ #define DRM_IOCTL_EXYNOS_GEM_CREATE DRM_IOWR(DRM_COMMAND_BASE + \ DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create) @@ -334,18 +164,8 @@ struct drm_exynos_ipp_cmd_ctrl { #define DRM_IOCTL_EXYNOS_G2D_EXEC DRM_IOWR(DRM_COMMAND_BASE + \ DRM_EXYNOS_G2D_EXEC, struct drm_exynos_g2d_exec) -#define DRM_IOCTL_EXYNOS_IPP_GET_PROPERTY DRM_IOWR(DRM_COMMAND_BASE + \ - DRM_EXYNOS_IPP_GET_PROPERTY, struct drm_exynos_ipp_prop_list) -#define DRM_IOCTL_EXYNOS_IPP_SET_PROPERTY DRM_IOWR(DRM_COMMAND_BASE + \ - DRM_EXYNOS_IPP_SET_PROPERTY, struct drm_exynos_ipp_property) -#define DRM_IOCTL_EXYNOS_IPP_QUEUE_BUF DRM_IOWR(DRM_COMMAND_BASE + \ - DRM_EXYNOS_IPP_QUEUE_BUF, struct drm_exynos_ipp_queue_buf) -#define DRM_IOCTL_EXYNOS_IPP_CMD_CTRL DRM_IOWR(DRM_COMMAND_BASE + \ - DRM_EXYNOS_IPP_CMD_CTRL, struct drm_exynos_ipp_cmd_ctrl) - /* EXYNOS specific events */ #define DRM_EXYNOS_G2D_EVENT 0x80000000 -#define DRM_EXYNOS_IPP_EVENT 0x80000001 struct drm_exynos_g2d_event { struct drm_event base; @@ -356,16 +176,6 @@ struct drm_exynos_g2d_event { __u32 reserved; }; -struct drm_exynos_ipp_event { - struct drm_event base; - __u64 user_data; - __u32 tv_sec; - __u32 tv_usec; - __u32 prop_id; - __u32 reserved; - __u32 buf_id[EXYNOS_DRM_OPS_MAX]; -}; - #if defined(__cplusplus) } #endif -- 2.14.2 ^ permalink raw reply related [flat|nested] 20+ messages in thread
[parent not found: <CGME20171023075449eucas1p18211ed64e964860d201aaeb53a87a8be@eucas1p1.samsung.com>]
* [PATCH v4 2/9] drm/exynos: ipp: Add IPP v2 framework [not found] ` <CGME20171023075449eucas1p18211ed64e964860d201aaeb53a87a8be@eucas1p1.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski 2017-11-01 1:13 ` Inki Dae 2017-11-01 3:47 ` Dave Airlie 0 siblings, 2 replies; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Marek Szyprowski, Inki Dae, Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon This patch adds Exynos IPP v2 subsystem and userspace API. New userspace API is focused ONLY on memory-to-memory image processing. The two remainging IPP operation modes (framebuffer writeback and local-path output with image processing) can be implemented using standard DRM features: writeback connectors and additional DRM planes with scaling features. V2 IPP userspace API is not compatible with old IPP ioctls. This is a significant change, but the old IPP subsystem in mainline Linux kernel was partially disfunctional anyway and not used in any open-source project. V2 IPP userspace API is based on stateless approach, which much better fits to memory-to-memory image processing model. It also provides support for all image formats, which are both already defined in DRM API and supported by the existing IPP hardware modules. The API consists of the following ioctls: - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image processing modules, - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image formats of given IPP module, - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for selected image format of given IPP module, - DRM_IOCTL_EXYNOS_IPP_COMMIT: to perform operation described by the provided structures (source and destination buffers, operation rectangle, transformation, etc). The proposed userspace API is extensible. In the future more advanced image processing operations can be defined to support for example blending. Userspace API is fully functional also on DRM render nodes, so it is not limited to the root/privileged client. Internal driver API also has been completely rewritten. New IPP core performs all possible input validation, checks and object life-time control. The drivers can focus only on writing configuration to hardware registers. Stateless nature of DRM_IOCTL_EXYNOS_IPP_COMMIT ioctl simplifies the driver API. Minimal driver needs to provide a single callback for starting processing and an array with supported image formats. Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Hoegeun Kwon <hoegeun.kwon@samsung.com> --- drivers/gpu/drm/exynos/Kconfig | 3 + drivers/gpu/drm/exynos/Makefile | 1 + drivers/gpu/drm/exynos/exynos_drm_drv.c | 9 + drivers/gpu/drm/exynos/exynos_drm_ipp.c | 911 ++++++++++++++++++++++++++++++++ drivers/gpu/drm/exynos/exynos_drm_ipp.h | 195 +++++++ include/uapi/drm/exynos_drm.h | 238 +++++++++ 6 files changed, 1357 insertions(+) create mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.c create mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.h diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig index 88cff0e039b6..2e097a81df12 100644 --- a/drivers/gpu/drm/exynos/Kconfig +++ b/drivers/gpu/drm/exynos/Kconfig @@ -94,6 +94,9 @@ config DRM_EXYNOS_G2D help Choose this option if you want to use Exynos G2D for DRM. +config DRM_EXYNOS_IPP + bool + config DRM_EXYNOS_FIMC bool "FIMC" depends on BROKEN && MFD_SYSCON diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile index 09bb002c9555..f663490e949d 100644 --- a/drivers/gpu/drm/exynos/Makefile +++ b/drivers/gpu/drm/exynos/Makefile @@ -17,6 +17,7 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER) += exynos_mixer.o exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI) += exynos_hdmi.o exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI) += exynos_drm_vidi.o exynosdrm-$(CONFIG_DRM_EXYNOS_G2D) += exynos_drm_g2d.o +exynosdrm-$(CONFIG_DRM_EXYNOS_IPP) += exynos_drm_ipp.o exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC) += exynos_drm_fimc.o exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR) += exynos_drm_rotator.o exynosdrm-$(CONFIG_DRM_EXYNOS_GSC) += exynos_drm_gsc.o diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c index 2fc5d3c01390..de4cce258a5b 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c @@ -26,6 +26,7 @@ #include "exynos_drm_fb.h" #include "exynos_drm_gem.h" #include "exynos_drm_plane.h" +#include "exynos_drm_ipp.h" #include "exynos_drm_vidi.h" #include "exynos_drm_g2d.h" #include "exynos_drm_iommu.h" @@ -114,6 +115,14 @@ static const struct drm_ioctl_desc exynos_ioctls[] = { DRM_AUTH | DRM_RENDER_ALLOW), DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl, DRM_AUTH | DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_RESOURCES, exynos_drm_ipp_get_res_ioctl, + DRM_AUTH | DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_CAPS, exynos_drm_ipp_get_caps_ioctl, + DRM_AUTH | DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_LIMITS, exynos_drm_ipp_get_limits_ioctl, + DRM_AUTH | DRM_RENDER_ALLOW), + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_COMMIT, exynos_drm_ipp_commit_ioctl, + DRM_AUTH | DRM_RENDER_ALLOW), }; static const struct file_operations exynos_drm_driver_fops = { diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.c b/drivers/gpu/drm/exynos/exynos_drm_ipp.c new file mode 100644 index 000000000000..d54d23c6f2b0 --- /dev/null +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.c @@ -0,0 +1,911 @@ +/* + * Copyright (C) 2017 Samsung Electronics Co.Ltd + * Authors: + * Marek Szyprowski <m.szyprowski@samsung.com> + * + * Exynos DRM Image Post Processing (IPP) related functions + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + */ + + +#include <drm/drmP.h> +#include <drm/drm_mode.h> +#include <uapi/drm/exynos_drm.h> + +#include "exynos_drm_drv.h" +#include "exynos_drm_gem.h" +#include "exynos_drm_ipp.h" + +static int num_ipp; +static LIST_HEAD(ipp_list); + +struct drm_pending_exynos_ipp_event { + struct drm_pending_event base; + struct drm_exynos_ipp_event event; +}; + +/** + * exynos_drm_ipp_register - Register a new picture processor hardware module + * @dev: DRM device + * @ipp: ipp module to init + * @funcs: callbacks for the new ipp object + * @caps: bitmask of ipp capabilities (%DRM_EXYNOS_IPP_CAP_*) + * @formats: array of supported formats (%DRM_FORMAT_*) + * @name: name (for debugging purposes) + * + * Initializes a ipp module. + * + * Returns: + * Zero on success, error code on failure. + */ +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp, + const struct exynos_drm_ipp_funcs *funcs, unsigned int caps, + const struct exynos_drm_ipp_formats *formats, const char *name) +{ + WARN_ON(!ipp); + WARN_ON(!funcs); + WARN_ON(!formats); + + spin_lock_init(&ipp->lock); + INIT_LIST_HEAD(&ipp->todo_list); + init_waitqueue_head(&ipp->done_wq); + ipp->dev = dev; + ipp->funcs = funcs; + ipp->capabilities = caps; + ipp->name = name; + ipp->num_formats = 0; + ipp->formats = formats; + while (formats++->fourcc) + ipp->num_formats++; + + list_add_tail(&ipp->head, &ipp_list); + ipp->id = num_ipp++; + + DRM_DEBUG_DRIVER("Registered ipp %d\n", ipp->id); + + return 0; +} + +/** + * exynos_drm_ipp_unregister - Unregister the picture processor module + * @dev: DRM device + * @ipp: ipp module + */ +void exynos_drm_ipp_unregister(struct drm_device *dev, + struct exynos_drm_ipp *ipp) +{ + BUG_ON(ipp->task); + BUG_ON(!list_empty(&ipp->todo_list)); + list_del(&ipp->head); +} + +/** + * exynos_drm_ipp_ioctl_get_res_ioctl - enumerate all ipp modules + * @dev: DRM device + * @data: ioctl data + * @file_priv: DRM file info + * + * Construct a list of ipp ids. + * + * Called by the user via ioctl. + * + * Returns: + * Zero on success, negative errno on failure. + */ +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_exynos_ioctl_ipp_get_res *resp = data; + struct exynos_drm_ipp *ipp; + uint32_t __user *ipp_ptr = (uint32_t __user *) + (unsigned long)resp->ipp_id_ptr; + unsigned int count = num_ipp, copied = 0; + + /* + * This ioctl is called twice, once to determine how much space is + * needed, and the 2nd time to fill it. + */ + if (count && resp->count_ipps >= count) { + list_for_each_entry(ipp, &ipp_list, head) { + if (put_user(ipp->id, ipp_ptr + copied)) + return -EFAULT; + copied++; + } + } + resp->count_ipps = count; + + return 0; +} + +static inline struct exynos_drm_ipp *__ipp_get(uint32_t id) +{ + struct exynos_drm_ipp *ipp; + + list_for_each_entry(ipp, &ipp_list, head) + if (ipp->id == id) + return ipp; + return NULL; +} + +/** + * exynos_drm_ipp_ioctl_get_caps - get ipp module capabilities and formats + * @dev: DRM device + * @data: ioctl data + * @file_priv: DRM file info + * + * Construct a structure describing ipp module capabilities. + * + * Called by the user via ioctl. + * + * Returns: + * Zero on success, negative errno on failure. + */ +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_exynos_ioctl_ipp_get_caps *resp = data; + void __user *ptr = (void __user *)(unsigned long)resp->formats_ptr; + struct exynos_drm_ipp *ipp; + int i; + + ipp = __ipp_get(resp->ipp_id); + if (!ipp) + return -ENOENT; + + resp->ipp_id = ipp->id; + resp->capabilities = ipp->capabilities; + + /* + * This ioctl is called twice, once to determine how much space is + * needed, and the 2nd time to fill it. + */ + if (resp->formats_count >= ipp->num_formats) { + for (i = 0; i < ipp->num_formats; i++) { + struct drm_exynos_ipp_format tmp = { + .fourcc = ipp->formats[i].fourcc, + .type = ipp->formats[i].type, + .modifier = ipp->formats[i].modifier, + }; + + if (copy_to_user(ptr, &tmp, sizeof(tmp))) + return -EFAULT; + ptr += sizeof(tmp); + } + } + resp->formats_count = ipp->num_formats; + + return 0; +} + +static inline const struct exynos_drm_ipp_formats *__ipp_format_get( + uint32_t fourcc, uint64_t mod, unsigned long type, + const struct exynos_drm_ipp_formats *f) +{ + for (; f->fourcc; f++) + if ((f->type & type) && f->fourcc == fourcc && + f->modifier == mod) + return f; + return NULL; +} + +/** + * exynos_drm_ipp_get_limits_ioctl - get ipp module limits + * @dev: DRM device + * @data: ioctl data + * @file_priv: DRM file info + * + * Construct a structure describing ipp module limitations for provided + * picture format. + * + * Called by the user via ioctl. + * + * Returns: + * Zero on success, negative errno on failure. + */ +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_exynos_ioctl_ipp_get_limits *resp = data; + void __user *ptr = (void __user *)(unsigned long)resp->limits_ptr; + const struct exynos_drm_ipp_formats *format; + const struct drm_exynos_ipp_limit *limits, *l; + struct exynos_drm_ipp *ipp; + int count = 0; + + if (resp->type != DRM_EXYNOS_IPP_FORMAT_SOURCE && + resp->type != DRM_EXYNOS_IPP_FORMAT_DESTINATION) + return -EINVAL; + + ipp = __ipp_get(resp->ipp_id); + if (!ipp) + return -ENOENT; + + format = __ipp_format_get(resp->fourcc, resp->modifier, resp->type, + ipp->formats); + if (!format) + return -EINVAL; + + limits = format->limits; + if (limits) + for (l = limits; l->type; l++) + count++; + + /* + * This ioctl is called twice, once to determine how much space is + * needed, and the 2nd time to fill it. + */ + if (count && resp->limits_count >= count) { + for (l = limits; l->type; l++, ptr += sizeof(*l)) + if (copy_to_user((void __user *)ptr, l, sizeof(*l))) + return -EFAULT; + } + resp->limits_count = count; + + return 0; +} + +static inline struct exynos_drm_ipp_task * + exynos_drm_ipp_task_alloc(struct exynos_drm_ipp *ipp) +{ + struct exynos_drm_ipp_task *task; + + task = kzalloc(sizeof(*task), GFP_KERNEL); + if (!task) + return NULL; + + task->dev = ipp->dev; + task->ipp = ipp; + + /* some defaults */ + task->src.rect.w = task->dst.rect.w = UINT_MAX; + task->src.rect.h = task->dst.rect.h = UINT_MAX; + task->transform.rotation = DRM_MODE_ROTATE_0; + + DRM_DEBUG_DRIVER("Allocated task %pK\n", task); + + return task; +} + +struct exynos_drm_param_map { + unsigned int id; + unsigned int size; + unsigned int offset; +} exynos_drm_ipp_params_maps[] = { + { + DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE, + sizeof(struct drm_exynos_ipp_task_buffer), + offsetof(struct exynos_drm_ipp_task, src.buf), + }, { + DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION, + sizeof(struct drm_exynos_ipp_task_buffer), + offsetof(struct exynos_drm_ipp_task, dst.buf), + }, { + DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE, + sizeof(struct drm_exynos_ipp_task_rect), + offsetof(struct exynos_drm_ipp_task, src.rect), + }, { + DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION, + sizeof(struct drm_exynos_ipp_task_rect), + offsetof(struct exynos_drm_ipp_task, dst.rect), + }, { + DRM_EXYNOS_IPP_TASK_TRANSFORM, + sizeof(struct drm_exynos_ipp_task_transform), + offsetof(struct exynos_drm_ipp_task, transform), + }, { + DRM_EXYNOS_IPP_TASK_ALPHA, + sizeof(struct drm_exynos_ipp_task_alpha), + offsetof(struct exynos_drm_ipp_task, alpha), + }, +}; + +static int exynos_drm_ipp_task_set(struct exynos_drm_ipp_task *task, + struct drm_exynos_ioctl_ipp_commit *arg) +{ + unsigned int size = arg->params_size; + void *p, *params; + int ret = 0; + + if (size > PAGE_SIZE) + return -ERANGE; + + p = kmalloc(size, GFP_KERNEL); + if (!p) + return -ENOMEM; + + if (copy_from_user(p, (void __user *)(unsigned long)arg->params_ptr, + size)) { + ret = -EFAULT; + DRM_DEBUG_DRIVER("Failed to copy configuration from userspace\n"); + goto free; + } + + params = p; + while (size) { + struct exynos_drm_param_map *map = exynos_drm_ipp_params_maps; + u32 *id = params; + int i; + + for (i = 0; i < ARRAY_SIZE(exynos_drm_ipp_params_maps); i++) + if (map[i].id == *id) + break; + + if (i < ARRAY_SIZE(exynos_drm_ipp_params_maps) && + map[i].size <= size) { + memcpy((void *)task + map[i].offset, params, + map[i].size); + params += map[i].size; + size -= map[i].size; + } else { + ret = -EINVAL; + goto free; + } + } + + DRM_DEBUG_DRIVER("Got task %pK configuration from userspace\n", task); +free: + kfree(p); + return ret; +} + +static int exynos_drm_ipp_task_setup_buffer(struct exynos_drm_ipp_buffer *buf, + struct drm_file *filp) +{ + int ret = 0; + int i; + + /* basic checks */ + if (buf->buf.width == 0 || buf->buf.height == 0) + return -EINVAL; + buf->format = drm_format_info(buf->buf.fourcc); + for (i = 0; i < buf->format->num_planes; i++) { + unsigned int width = (i == 0) ? buf->buf.width : + DIV_ROUND_UP(buf->buf.width, buf->format->hsub); + + if (buf->buf.pitch[i] == 0) + buf->buf.pitch[i] = width * buf->format->cpp[i]; + if (buf->buf.pitch[i] < width * buf->format->cpp[i]) + return -EINVAL; + if (!buf->buf.gem_id[i]) + return -ENOENT; + } + + /* get GEM buffers and check their size */ + for (i = 0; i < buf->format->num_planes; i++) { + unsigned int height = (i == 0) ? buf->buf.height : + DIV_ROUND_UP(buf->buf.height, buf->format->vsub); + unsigned long size = height * buf->buf.pitch[i] + + buf->buf.offset[i]; + struct drm_gem_object *obj = drm_gem_object_lookup(filp, + buf->buf.gem_id[i]); + if (!obj) { + ret = -ENOENT; + goto gem_free; + } + buf->exynos_gem[i] = to_exynos_gem(obj); + + if (size > buf->exynos_gem[i]->size) { + i++; + ret = -EINVAL; + goto gem_free; + } + buf->dma_addr[i] = buf->exynos_gem[i]->dma_addr + + buf->buf.offset[i]; + } + + return 0; +gem_free: + while (i--) { + drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base); + buf->exynos_gem[i] = NULL; + } + return ret; +} + +static void exynos_drm_ipp_task_release_buffer(struct exynos_drm_ipp_buffer *buf) +{ + int i; + + if (!buf->exynos_gem[0]) + return; + for (i = 0; i < buf->format->num_planes; i++) + drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base); +} + +static void exynos_drm_ipp_task_free(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) +{ + DRM_DEBUG_DRIVER("Freeing task %pK\n", task); + + exynos_drm_ipp_task_release_buffer(&task->src); + exynos_drm_ipp_task_release_buffer(&task->dst); + if (task->event) + drm_event_cancel_free(ipp->dev, &task->event->base); + kfree(task); +} + +struct drm_ipp_limit { + struct drm_exynos_ipp_limit_val h; + struct drm_exynos_ipp_limit_val v; +}; + +enum drm_ipp_size_id { + IPP_LIMIT_BUFFER, IPP_LIMIT_AREA, IPP_LIMIT_ROTATED, IPP_LIMIT_MAX +}; + +static const enum drm_ipp_size_id limit_id_fallback[IPP_LIMIT_MAX][4] = { + [IPP_LIMIT_BUFFER] = { DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER }, + [IPP_LIMIT_AREA] = { DRM_EXYNOS_IPP_LIMIT_SIZE_AREA, + DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER }, + [IPP_LIMIT_ROTATED] = { DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED, + DRM_EXYNOS_IPP_LIMIT_SIZE_AREA, + DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER }, +}; + +static inline void __limit_set_val(unsigned int *ptr, unsigned int val) +{ + if (!*ptr) + *ptr = val; +} + +static void __get_size_limit(const struct drm_exynos_ipp_limit *limits, + enum drm_ipp_size_id id, struct drm_ipp_limit *res) +{ + const struct drm_exynos_ipp_limit *l = limits; + int i = 0; + + memset(res, 0, sizeof(*res)); + for (i = 0; limit_id_fallback[id][i]; i++) + for (l = limits; l->type; l++) { + if (((l->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) != + DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE) || + ((l->type & DRM_EXYNOS_IPP_LIMIT_SIZE_MASK) != + limit_id_fallback[id][i])) + continue; + __limit_set_val(&res->h.min, l->h.min); + __limit_set_val(&res->h.max, l->h.max); + __limit_set_val(&res->h.align, l->h.align); + __limit_set_val(&res->v.min, l->v.min); + __limit_set_val(&res->v.max, l->v.max); + __limit_set_val(&res->v.align, l->v.align); + } +} + +static inline bool __align_check(unsigned int val, unsigned int align) +{ + if (align && (val & (align - 1))) { + DRM_DEBUG_DRIVER("Value %d exceeds hw limits (align %d)\n", + val, align); + return false; + } + return true; +} + +static inline bool __size_limit_check(unsigned int val, + struct drm_exynos_ipp_limit_val *l) +{ + if ((l->min && val < l->min) || (l->max && val > l->max)) { + DRM_DEBUG_DRIVER("Value %d exceeds hw limits (min %d, max %d)\n", + val, l->min, l->max); + return false; + } + return __align_check(val, l->align); +} + +static int exynos_drm_ipp_check_size_limits(struct exynos_drm_ipp_buffer *buf, + const struct drm_exynos_ipp_limit *limits, bool rotate, bool swap) +{ + enum drm_ipp_size_id id = rotate ? IPP_LIMIT_ROTATED : IPP_LIMIT_AREA; + struct drm_ipp_limit l; + struct drm_exynos_ipp_limit_val *lh = &l.h, *lv = &l.v; + + if (!limits) + return 0; + + __get_size_limit(limits, IPP_LIMIT_BUFFER, &l); + if (!__size_limit_check(buf->buf.width, &l.h) || + !__size_limit_check(buf->buf.height, &l.v)) + return -EINVAL; + + if (swap) { + lv = &l.h; + lh = &l.v; + } + __get_size_limit(limits, id, &l); + if (!__size_limit_check(buf->rect.w, lh) || + !__align_check(buf->rect.x, lh->align) || + !__size_limit_check(buf->rect.h, lv) || + !__align_check(buf->rect.y, lv->align)) + return -EINVAL; + + return 0; +} + +static inline bool __scale_limit_check(unsigned int src, unsigned int dst, + unsigned int min, unsigned int max) +{ + if ((max && (dst << 16) > src * max) || + (min && (dst << 16) < src * min)) { + DRM_DEBUG_DRIVER("Scale from %d to %d exceeds hw limits (ratio min %d.%05d, max %d.%05d)\n", + src, dst, + min >> 16, 100000 * (min & 0xffff) / (1 << 16), + max >> 16, 100000 * (max & 0xffff) / (1 << 16)); + return false; + } + return true; +} + +static int exynos_drm_ipp_check_scale_limits(struct drm_exynos_ipp_task_rect *src, + struct drm_exynos_ipp_task_rect *dst, + const struct drm_exynos_ipp_limit *limits, bool swap) +{ + const struct drm_exynos_ipp_limit_val *lh, *lv; + + if (!limits) + return 0; + + for (; limits->type; limits++) + if ((limits->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) == + DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE) + break; + if (!limits->type) + return 0; + + lh = (!swap) ? &limits->h : &limits->v; + lv = (!swap) ? &limits->v : &limits->h; + + if (!__scale_limit_check(src->w, dst->w, lh->min, lh->max) || + !__scale_limit_check(src->h, dst->h, lv->min, lv->max)) + return -EINVAL; + + return 0; +} + +static int exynos_drm_ipp_task_check(struct exynos_drm_ipp_task *task, + struct drm_file *filp) +{ + struct exynos_drm_ipp *ipp = task->ipp; + const struct exynos_drm_ipp_formats *src_format, *dst_format; + struct exynos_drm_ipp_buffer *src = &task->src, *dst = &task->dst; + unsigned int rotation = task->transform.rotation; + int ret = 0; + bool scale = false, swap = false; + + DRM_DEBUG_DRIVER("Checking task %pK\n", task); + + if (src->rect.w == UINT_MAX) + src->rect.w = src->buf.width; + if (src->rect.h == UINT_MAX) + src->rect.h = src->buf.height; + if (dst->rect.w == UINT_MAX) + dst->rect.w = dst->buf.width; + if (dst->rect.h == UINT_MAX) + dst->rect.h = dst->buf.height; + + if (src->rect.x + src->rect.w > (src->buf.width) || + src->rect.y + src->rect.h > (src->buf.height) || + dst->rect.x + dst->rect.w > (dst->buf.width) || + dst->rect.y + dst->rect.h > (dst->buf.height)) { + DRM_DEBUG_DRIVER("Task %pK: defined area is outside provided buffers\n", + task); + return -EINVAL; + } + + if (drm_rotation_90_or_270(rotation)) + swap = true; + + if ((!swap && (src->rect.w != dst->rect.w || src->rect.h != dst->rect.h)) || + (swap && (src->rect.w != dst->rect.h || src->rect.h != dst->rect.w))) + scale = true; + + if ((!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CROP) && + (src->rect.x || src->rect.y || dst->rect.x || dst->rect.y)) || + (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_ROTATE) && + rotation != DRM_MODE_ROTATE_0) || + (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_SCALE) && scale) || + (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CONVERT) && + src->buf.fourcc != dst->buf.fourcc)) { + DRM_DEBUG_DRIVER("Task %pK: hw capabilities exceeded\n", task); + return -EINVAL; + } + + src_format = __ipp_format_get(src->buf.fourcc, src->buf.modifier, + DRM_EXYNOS_IPP_FORMAT_SOURCE, ipp->formats); + if (!src_format) { + DRM_DEBUG_DRIVER("Task %pK: src format not supported\n", task); + return -EINVAL; + } + ret = exynos_drm_ipp_check_size_limits(src, src_format->limits, + rotation != DRM_MODE_ROTATE_0, false); + if (ret) + return ret; + ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect, + src_format->limits, swap); + if (ret) + return ret; + + dst_format = __ipp_format_get(dst->buf.fourcc, dst->buf.modifier, + DRM_EXYNOS_IPP_FORMAT_DESTINATION, ipp->formats); + if (!dst_format) { + DRM_DEBUG_DRIVER("Task %pK: dst format not supported\n", task); + return -EINVAL; + } + ret = exynos_drm_ipp_check_size_limits(dst, dst_format->limits, + rotation != DRM_MODE_ROTATE_0, swap); + if (ret) + return ret; + ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect, + dst_format->limits, swap); + if (ret) + return ret; + + ret = exynos_drm_ipp_task_setup_buffer(src, filp); + if (ret) { + DRM_DEBUG_DRIVER("Task %pK: src buffer setup failed\n", task); + return ret; + } + ret = exynos_drm_ipp_task_setup_buffer(dst, filp); + if (ret) { + DRM_DEBUG_DRIVER("Task %pK: dst buffer setup failed\n", task); + return ret; + } + + if (ipp->funcs->check) { + ret = ipp->funcs->check(ipp, task); + if (ret) { + DRM_DEBUG_DRIVER("Task %pK: driver specific check failed\n", + task); + return ret; + } + } + + DRM_DEBUG_DRIVER("Task %pK: all checks done.\n", task); + + return ret; +} + +static int exynos_drm_ipp_event_create(struct exynos_drm_ipp_task *task, + struct drm_file *file_priv, uint64_t user_data) +{ + struct drm_pending_exynos_ipp_event *e = NULL; + int ret; + + e = kzalloc(sizeof(*e), GFP_KERNEL); + if (!e) + return -ENOMEM; + + e->event.base.type = DRM_EXYNOS_IPP_EVENT; + e->event.base.length = sizeof(e->event); + e->event.user_data = user_data; + + ret = drm_event_reserve_init(task->dev, file_priv, &e->base, + &e->event.base); + if (ret) + goto free; + + task->event = e; + return 0; +free: + kfree(e); + return ret; +} + +static void exynos_drm_ipp_event_send(struct exynos_drm_ipp_task *task) +{ + struct timeval now = ktime_to_timeval(ktime_get()); + + task->event->event.tv_sec = now.tv_sec; + task->event->event.tv_usec = now.tv_usec; + task->event->event.sequence = atomic_inc_return(&task->ipp->sequence); + + drm_send_event(task->dev, &task->event->base); +} + +static int exynos_drm_ipp_task_cleanup(struct exynos_drm_ipp_task *task) +{ + int ret = task->ret; + + if (ret == 0 && task->event) { + exynos_drm_ipp_event_send(task); + /* ensure event won't be canceled on task free */ + task->event = NULL; + } + + exynos_drm_ipp_task_free(task->ipp, task); + return ret; +} + +static void exynos_drm_ipp_cleanup_work(struct work_struct *work) +{ + struct exynos_drm_ipp_task *task = container_of(work, + struct exynos_drm_ipp_task, cleanup_work); + + exynos_drm_ipp_task_cleanup(task); +} + +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp); + +/** + * exynos_drm_ipp_task_done - finish given task and set return code + * @task: ipp task to finish + * @ret: error code or 0 if operation has been performed successfully + */ +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret) +{ + struct exynos_drm_ipp *ipp = task->ipp; + unsigned long flags; + + DRM_DEBUG_DRIVER("ipp: %d, task %pK done\n", ipp->id, task); + + spin_lock_irqsave(&ipp->lock, flags); + if (ipp->task == task) + ipp->task = NULL; + task->flags |= DRM_EXYNOS_IPP_TASK_DONE; + task->ret = ret; + spin_unlock_irqrestore(&ipp->lock, flags); + + exynos_drm_ipp_next_task(ipp); + wake_up(&ipp->done_wq); + + if (task->flags & DRM_EXYNOS_IPP_TASK_ASYNC) { + INIT_WORK(&task->cleanup_work, exynos_drm_ipp_cleanup_work); + schedule_work(&task->cleanup_work); + } +} + +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp) +{ + struct exynos_drm_ipp_task *task; + unsigned long flags; + int ret; + + DRM_DEBUG_DRIVER("ipp: %d, try to run new task\n", ipp->id); + + spin_lock_irqsave(&ipp->lock, flags); + + if (ipp->task || list_empty(&ipp->todo_list)) { + spin_unlock_irqrestore(&ipp->lock, flags); + return; + } + + task = list_first_entry(&ipp->todo_list, struct exynos_drm_ipp_task, + head); + list_del_init(&task->head); + ipp->task = task; + + spin_unlock_irqrestore(&ipp->lock, flags); + + DRM_DEBUG_DRIVER("ipp: %d, selected task %pK to run\n", ipp->id, task); + + ret = ipp->funcs->commit(ipp, task); + if (ret) + exynos_drm_ipp_task_done(task, ret); +} + +static void exynos_drm_ipp_schedule_task(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) +{ + unsigned long flags; + + spin_lock_irqsave(&ipp->lock, flags); + list_add(&task->head, &ipp->todo_list); + spin_unlock_irqrestore(&ipp->lock, flags); + + exynos_drm_ipp_next_task(ipp); +} + +static void exynos_drm_ipp_task_abort(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) +{ + unsigned long flags; + + spin_lock_irqsave(&ipp->lock, flags); + if (task->flags & DRM_EXYNOS_IPP_TASK_DONE) { + /* already completed task */ + exynos_drm_ipp_task_cleanup(task); + } else if (ipp->task != task) { + /* task has not been scheduled for execution yet */ + list_del_init(&task->head); + exynos_drm_ipp_task_cleanup(task); + } else { + /* + * currently processed task, call abort() and perform + * cleanup with async worker + */ + task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC; + spin_unlock_irqrestore(&ipp->lock, flags); + if (ipp->funcs->abort) + ipp->funcs->abort(ipp, task); + return; + } + spin_unlock_irqrestore(&ipp->lock, flags); +} + +/** + * exynos_drm_ipp_commit_ioctl - perform image processing operation + * @dev: DRM device + * @data: ioctl data + * @file_priv: DRM file info + * + * Construct a ipp task from the set of properties provided from the user + * and try to schedule it to framebuffer processor hardware. + * + * Called by the user via ioctl. + * + * Returns: + * Zero on success, negative errno on failure. + */ +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv) +{ + struct drm_exynos_ioctl_ipp_commit *arg = data; + struct exynos_drm_ipp *ipp; + struct exynos_drm_ipp_task *task; + int ret = 0; + + if ((arg->flags & ~DRM_EXYNOS_IPP_FLAGS) || arg->reserved) + return -EINVAL; + + /* can't test and expect an event at the same time */ + if ((arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY) && + (arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT)) + return -EINVAL; + + ipp = __ipp_get(arg->ipp_id); + if (!ipp) + return -ENOENT; + + task = exynos_drm_ipp_task_alloc(ipp); + if (!task) + return -ENOMEM; + + ret = exynos_drm_ipp_task_set(task, arg); + if (ret) + goto free; + + ret = exynos_drm_ipp_task_check(task, file_priv); + if (ret || arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY) + goto free; + + if (arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT) { + ret = exynos_drm_ipp_event_create(task, file_priv, + arg->user_data); + if (ret) + goto free; + } + + /* + * Queue task for processing on the hardware. task object will be + * then freed after exynos_drm_ipp_task_done() + */ + if (arg->flags & DRM_EXYNOS_IPP_FLAG_NONBLOCK) { + DRM_DEBUG_DRIVER("ipp: %d, nonblocking processing task %pK\n", + ipp->id, task); + + task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC; + exynos_drm_ipp_schedule_task(task->ipp, task); + ret = 0; + } else { + DRM_DEBUG_DRIVER("ipp: %d, processing task %pK\n", ipp->id, task); + exynos_drm_ipp_schedule_task(ipp, task); + ret = wait_event_interruptible(ipp->done_wq, + task->flags & DRM_EXYNOS_IPP_TASK_DONE); + if (ret) + exynos_drm_ipp_task_abort(ipp, task); + else + ret = exynos_drm_ipp_task_cleanup(task); + } + return ret; +free: + exynos_drm_ipp_task_free(ipp, task); + + return ret; +} diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.h b/drivers/gpu/drm/exynos/exynos_drm_ipp.h new file mode 100644 index 000000000000..199eb2006410 --- /dev/null +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.h @@ -0,0 +1,195 @@ +/* + * Copyright (c) 2017 Samsung Electronics Co., Ltd. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2 of the License, or (at your + * option) any later version. + */ + +#ifndef _EXYNOS_DRM_IPP_H_ +#define _EXYNOS_DRM_IPP_H_ + +#include <drm/drmP.h> + +struct exynos_drm_ipp; +struct exynos_drm_ipp_task; + +/** + * struct exynos_drm_ipp_funcs - exynos_drm_ipp control functions + */ +struct exynos_drm_ipp_funcs { + /** + * @check: + * + * This is the optional hook to validate an ipp task. This function + * must reject any task which the hardware or driver doesn't support. + * This includes but is of course not limited to: + * + * - Checking that the buffers, scaling and placement requirements + * and so on are within the limits of the hardware not specified by + * limits array. + * + * - The driver does not need to repeat basic input validation like + * done in the exynos_drm_ipp_task_check() function. The core does + * that before calling this hook. + * + * RETURNS: + * + * 0 on success or one of the below negative error codes: + * + * - -EINVAL, if any of the above constraints are violated. + */ + int (*check)(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task); + + /** + * @commit: + * + * This is the main entry point to start framebuffer processing + * in the hardware. The exynos_drm_ipp_task has been already validated. + * This function must not wait until the device finishes processing. + * When the driver finishes processing, it has to call + * exynos_exynos_drm_ipp_task_done() function. + * + * RETURNS: + * + * 0 on success or negative error codes in case of failure. + */ + int (*commit)(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task); + + /** + * @abort: + * + * Informs the driver that it has to abort the currently running + * task as soon as possible (i.e. as soon as it can stop the device + * safely), even if the task would not have been finished by then. + * After the driver performs the necessary steps, it has to call + * exynos_drm_ipp_task_done() (as if the task ended normally). + * This function does not have to (and will usually not) wait + * until the device enters a state when it can be stopped. + */ + void (*abort)(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task); +}; + +/** + * struct exynos_drm_ipp - central picture processor module structure + */ +struct exynos_drm_ipp { + struct drm_device *dev; + struct list_head head; + unsigned int id; + + const char *name; + const struct exynos_drm_ipp_funcs *funcs; + unsigned int capabilities; + const struct exynos_drm_ipp_formats *formats; + unsigned int num_formats; + atomic_t sequence; + + spinlock_t lock; + struct exynos_drm_ipp_task *task; + struct list_head todo_list; + wait_queue_head_t done_wq; +}; + +struct exynos_drm_ipp_buffer { + struct drm_exynos_ipp_task_buffer buf; + struct drm_exynos_ipp_task_rect rect; + + struct exynos_drm_gem *exynos_gem[MAX_FB_BUFFER]; + const struct drm_format_info *format; + dma_addr_t dma_addr[MAX_FB_BUFFER]; +}; + +/** + * struct exynos_drm_ipp_task - a structure describing transformation that + * has to be performed by the picture processor hardware module + */ +struct exynos_drm_ipp_task { + struct drm_device *dev; + struct exynos_drm_ipp *ipp; + struct list_head head; + + struct exynos_drm_ipp_buffer src; + struct exynos_drm_ipp_buffer dst; + + struct drm_exynos_ipp_task_transform transform; + struct drm_exynos_ipp_task_alpha alpha; + + struct work_struct cleanup_work; + unsigned int flags; + int ret; + + struct drm_pending_exynos_ipp_event *event; +}; + +#define DRM_EXYNOS_IPP_TASK_DONE (1 << 0) +#define DRM_EXYNOS_IPP_TASK_ASYNC (1 << 1) + +struct exynos_drm_ipp_formats { + u32 fourcc; + u32 type; + u64 modifier; + const struct drm_exynos_ipp_limit *limits; +}; + +/* helper macros to set exynos_drm_ipp_formats structure and limits*/ +#define IPP_SRCDST_MFORMAT(f, m, l) \ + .fourcc = DRM_FORMAT_##f, .modifier = m, .limits = l, \ + .type = (DRM_EXYNOS_IPP_FORMAT_SOURCE | DRM_EXYNOS_IPP_FORMAT_DESTINATION) + +#define IPP_SRCDST_FORMAT(f, l) IPP_SRCDST_MFORMAT(f, 0, l) + +#define IPP_SIZE_LIMIT(l, val...) \ + .type = (DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE | DRM_EXYNOS_IPP_LIMIT_SIZE_##l), \ + val + +#define IPP_SCALE_LIMIT(val...) \ + .type = (DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE),\ + val + +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp, + const struct exynos_drm_ipp_funcs *funcs, unsigned int caps, + const struct exynos_drm_ipp_formats *formats, const char *name); +void exynos_drm_ipp_unregister(struct drm_device *dev, struct exynos_drm_ipp *ipp); + +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret); + +#ifdef CONFIG_DRM_EXYNOS_IPP +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data, + struct drm_file *file_priv); +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv); +#else +static inline int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + struct drm_exynos_ioctl_ipp_get_res *resp = data; + + resp->count_ipps = 0; + return 0; +} +static inline int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + return -ENODEV; +} +static inline int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + return -ENODEV; +} +static inline int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, + void *data, struct drm_file *file_priv) +{ + return -ENODEV; +} +#endif +#endif diff --git a/include/uapi/drm/exynos_drm.h b/include/uapi/drm/exynos_drm.h index 5d9a08d69ba7..974cbb579f90 100644 --- a/include/uapi/drm/exynos_drm.h +++ b/include/uapi/drm/exynos_drm.h @@ -134,6 +134,219 @@ struct drm_exynos_g2d_exec { __u64 async; }; +/* Exynos DRM IPP v2 API */ + +/** + * Enumerate available IPP hardware modules. + * + * @count_ipps: size of ipp_id array / number of ipp modules (set by driver) + * @reserved: padding + * @ipp_id_ptr: pointer to ipp_id array or NULL + */ +struct drm_exynos_ioctl_ipp_get_res { + __u32 count_ipps; + __u32 reserved; + __u64 ipp_id_ptr; +}; + +enum drm_exynos_ipp_format_type { + DRM_EXYNOS_IPP_FORMAT_SOURCE = 0x01, + DRM_EXYNOS_IPP_FORMAT_DESTINATION = 0x02, +}; + +struct drm_exynos_ipp_format { + __u32 fourcc; + __u32 type; + __u64 modifier; +}; + +enum drm_exynos_ipp_capability { + DRM_EXYNOS_IPP_CAP_CROP = 0x01, + DRM_EXYNOS_IPP_CAP_ROTATE = 0x02, + DRM_EXYNOS_IPP_CAP_SCALE = 0x04, + DRM_EXYNOS_IPP_CAP_CONVERT = 0x08, +}; + +/** + * Get IPP hardware capabilities and supported image formats. + * + * @ipp_id: id of IPP module to query + * @capabilities: bitmask of drm_exynos_ipp_capability (set by driver) + * @reserved: padding + * @formats_count: size of formats array (in entries) / number of filled + * formats (set by driver) + * @formats_ptr: pointer to formats array or NULL + */ +struct drm_exynos_ioctl_ipp_get_caps { + __u32 ipp_id; + __u32 capabilities; + __u32 reserved; + __u32 formats_count; + __u64 formats_ptr; +}; + +enum drm_exynos_ipp_limit_type { + /* size (horizontal/vertial) limits, in pixels (min, max, alignment) */ + DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE = 0x0001, + /* scale ratio (horizonta/vertial), 16.16 fixed point (min, max) */ + DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE = 0x0002, + + /* image buffer area */ + DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER = 0x0001 << 16, + /* src/dst rectangle area */ + DRM_EXYNOS_IPP_LIMIT_SIZE_AREA = 0x0002 << 16, + /* src/dst rectangle area when rotation enabled */ + DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED = 0x0003 << 16, + + DRM_EXYNOS_IPP_LIMIT_TYPE_MASK = 0x000f, + DRM_EXYNOS_IPP_LIMIT_SIZE_MASK = 0x000f << 16, +}; + +struct drm_exynos_ipp_limit_val { + __u32 min; + __u32 max; + __u32 align; + __u32 reserved; +}; + +/** + * IPP module limitation. + * + * @type: limit type (see drm_exynos_ipp_limit_type enum) + * @reserved: padding + * @h: horizontal limits + * @v: vertical limits + */ +struct drm_exynos_ipp_limit { + __u32 type; + __u32 reserved; + struct drm_exynos_ipp_limit_val h; + struct drm_exynos_ipp_limit_val v; +}; + +/** + * Get IPP limits for given image format. + * + * @ipp_id: id of IPP module to query + * @fourcc: image format code (see DRM_FORMAT_* in drm_fourcc.h) + * @modifier: image format modifier (see DRM_FORMAT_MOD_* in drm_fourcc.h) + * @type: source/destination identifier (drm_exynos_ipp_format_flag enum) + * @limits_count: size of limits array (in entries) / number of filled entries + * (set by driver) + * @limits_ptr: pointer to limits array or NULL + */ +struct drm_exynos_ioctl_ipp_get_limits { + __u32 ipp_id; + __u32 fourcc; + __u64 modifier; + __u32 type; + __u32 limits_count; + __u64 limits_ptr; +}; + +enum drm_exynos_ipp_task_id { + /* buffer described by struct drm_exynos_ipp_task_buffer */ + DRM_EXYNOS_IPP_TASK_BUFFER = 0x0001, + /* rectangle described by struct drm_exynos_ipp_task_rect */ + DRM_EXYNOS_IPP_TASK_RECTANGLE = 0x0002, + /* transformation described by struct drm_exynos_ipp_task_transform */ + DRM_EXYNOS_IPP_TASK_TRANSFORM = 0x0003, + /* alpha configuration described by struct drm_exynos_ipp_task_alpha */ + DRM_EXYNOS_IPP_TASK_ALPHA = 0x0004, + + /* source image data (for buffer and rectangle chunks) */ + DRM_EXYNOS_IPP_TASK_TYPE_SOURCE = 0x0001 << 16, + /* destination image data (for buffer and rectangle chunks) */ + DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION = 0x0002 << 16, +}; + +/** + * Memory buffer with image data. + * + * @id: must be DRM_EXYNOS_IPP_TASK_BUFFER + * other parameters are same as for AddFB2 generic DRM ioctl + */ +struct drm_exynos_ipp_task_buffer { + __u32 id; + __u32 fourcc; + __u32 width, height; + __u32 gem_id[4]; + __u32 offset[4]; + __u32 pitch[4]; + __u64 modifier; +}; + +/** + * Rectangle for processing. + * + * @id: must be DRM_EXYNOS_IPP_TASK_RECTANGLE + * @reserved: padding + * @x,@y: left corner in pixels + * @w,@h: width/height in pixels + */ +struct drm_exynos_ipp_task_rect { + __u32 id; + __u32 reserved; + __u32 x; + __u32 y; + __u32 w; + __u32 h; +}; + +/** + * Image tranformation description. + * + * @id: must be DRM_EXYNOS_IPP_TASK_TRANSFORM + * @rotation: DRM_MODE_ROTATE_* and DRM_MODE_REFLECT_* values + */ +struct drm_exynos_ipp_task_transform { + __u32 id; + __u32 rotation; +}; + +/** + * Image global alpha configuration for formats without alpha values. + * + * @id: must be DRM_EXYNOS_IPP_TASK_ALPHA + * @value: global alpha value (0-255) + */ +struct drm_exynos_ipp_task_alpha { + __u32 id; + __u32 value; +}; + +enum drm_exynos_ipp_flag { + /* generate DRM event after processing */ + DRM_EXYNOS_IPP_FLAG_EVENT = 0x01, + /* dry run, only check task parameters */ + DRM_EXYNOS_IPP_FLAG_TEST_ONLY = 0x02, + /* non-blocking processing */ + DRM_EXYNOS_IPP_FLAG_NONBLOCK = 0x04, +}; + +#define DRM_EXYNOS_IPP_FLAGS (DRM_EXYNOS_IPP_FLAG_EVENT |\ + DRM_EXYNOS_IPP_FLAG_TEST_ONLY | DRM_EXYNOS_IPP_FLAG_NONBLOCK) + +/** + * Perform image processing described by array of drm_exynos_ipp_task_* + * structures (parameters array). + * + * @ipp_id: id of IPP module to run the task + * @flags: bitmask of drm_exynos_ipp_flag values + * @reserved: padding + * @params_size: size of parameters array (in bytes) + * @params_ptr: pointer to parameters array or NULL + * @user_data: (optional) data for drm event + */ +struct drm_exynos_ioctl_ipp_commit { + __u32 ipp_id; + __u32 flags; + __u32 reserved; + __u32 params_size; + __u64 params_ptr; + __u64 user_data; +}; + #define DRM_EXYNOS_GEM_CREATE 0x00 #define DRM_EXYNOS_GEM_MAP 0x01 /* Reserved 0x03 ~ 0x05 for exynos specific gem ioctl */ @@ -146,6 +359,11 @@ struct drm_exynos_g2d_exec { #define DRM_EXYNOS_G2D_EXEC 0x22 /* Reserved 0x30 ~ 0x33 for obsolete Exynos IPP ioctls */ +/* IPP - Image Post Processing */ +#define DRM_EXYNOS_IPP_GET_RESOURCES 0x40 +#define DRM_EXYNOS_IPP_GET_CAPS 0x41 +#define DRM_EXYNOS_IPP_GET_LIMITS 0x42 +#define DRM_EXYNOS_IPP_COMMIT 0x43 #define DRM_IOCTL_EXYNOS_GEM_CREATE DRM_IOWR(DRM_COMMAND_BASE + \ DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create) @@ -164,8 +382,18 @@ struct drm_exynos_g2d_exec { #define DRM_IOCTL_EXYNOS_G2D_EXEC DRM_IOWR(DRM_COMMAND_BASE + \ DRM_EXYNOS_G2D_EXEC, struct drm_exynos_g2d_exec) +#define DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES DRM_IOWR(DRM_COMMAND_BASE + \ + DRM_EXYNOS_IPP_GET_RESOURCES, struct drm_exynos_ioctl_ipp_get_res) +#define DRM_IOCTL_EXYNOS_IPP_GET_CAPS DRM_IOWR(DRM_COMMAND_BASE + \ + DRM_EXYNOS_IPP_GET_CAPS, struct drm_exynos_ioctl_ipp_get_caps) +#define DRM_IOCTL_EXYNOS_IPP_GET_LIMITS DRM_IOWR(DRM_COMMAND_BASE + \ + DRM_EXYNOS_IPP_GET_LIMITS, struct drm_exynos_ioctl_ipp_get_limits) +#define DRM_IOCTL_EXYNOS_IPP_COMMIT DRM_IOWR(DRM_COMMAND_BASE + \ + DRM_EXYNOS_IPP_COMMIT, struct drm_exynos_ioctl_ipp_commit) + /* EXYNOS specific events */ #define DRM_EXYNOS_G2D_EVENT 0x80000000 +#define DRM_EXYNOS_IPP_EVENT 0x80000002 struct drm_exynos_g2d_event { struct drm_event base; @@ -176,6 +404,16 @@ struct drm_exynos_g2d_event { __u32 reserved; }; +struct drm_exynos_ipp_event { + struct drm_event base; + __u64 user_data; + __u32 tv_sec; + __u32 tv_usec; + __u32 ipp_id; + __u32 sequence; + __u64 reserved; +}; + #if defined(__cplusplus) } #endif -- 2.14.2 ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH v4 2/9] drm/exynos: ipp: Add IPP v2 framework 2017-10-23 7:54 ` [PATCH v4 2/9] drm/exynos: ipp: Add IPP v2 framework Marek Szyprowski @ 2017-11-01 1:13 ` Inki Dae 2017-11-03 11:20 ` Marek Szyprowski 2017-11-01 3:47 ` Dave Airlie 1 sibling, 1 reply; 20+ messages in thread From: Inki Dae @ 2017-11-01 1:13 UTC (permalink / raw) To: Marek Szyprowski, dri-devel, linux-samsung-soc Cc: Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon Hi Marek, Thanks for your effort and sorry for late. Below are trivial my comments, 2017년 10월 23일 16:54에 Marek Szyprowski 이(가) 쓴 글: > This patch adds Exynos IPP v2 subsystem and userspace API. > > New userspace API is focused ONLY on memory-to-memory image processing. > The two remainging IPP operation modes (framebuffer writeback and > local-path output with image processing) can be implemented using > standard DRM features: writeback connectors and additional DRM planes with > scaling features. > > V2 IPP userspace API is not compatible with old IPP ioctls. This is a > significant change, but the old IPP subsystem in mainline Linux kernel was > partially disfunctional anyway and not used in any open-source project. > > V2 IPP userspace API is based on stateless approach, which much better fits > to memory-to-memory image processing model. It also provides support for > all image formats, which are both already defined in DRM API and supported > by the existing IPP hardware modules. > > The API consists of the following ioctls: > - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image > processing modules, > - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image > formats of given IPP module, > - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for > selected image format of given IPP module, > - DRM_IOCTL_EXYNOS_IPP_COMMIT: to perform operation described by the > provided structures (source and destination buffers, operation rectangle, > transformation, etc). > > The proposed userspace API is extensible. In the future more advanced image > processing operations can be defined to support for example blending. > > Userspace API is fully functional also on DRM render nodes, so it is not > limited to the root/privileged client. > > Internal driver API also has been completely rewritten. New IPP core > performs all possible input validation, checks and object life-time > control. The drivers can focus only on writing configuration to hardware > registers. Stateless nature of DRM_IOCTL_EXYNOS_IPP_COMMIT ioctl simplifies > the driver API. Minimal driver needs to provide a single callback for > starting processing and an array with supported image formats. > > Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> > Tested-by: Hoegeun Kwon <hoegeun.kwon@samsung.com> > --- > drivers/gpu/drm/exynos/Kconfig | 3 + > drivers/gpu/drm/exynos/Makefile | 1 + > drivers/gpu/drm/exynos/exynos_drm_drv.c | 9 + > drivers/gpu/drm/exynos/exynos_drm_ipp.c | 911 ++++++++++++++++++++++++++++++++ > drivers/gpu/drm/exynos/exynos_drm_ipp.h | 195 +++++++ > include/uapi/drm/exynos_drm.h | 238 +++++++++ > 6 files changed, 1357 insertions(+) > create mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.c > create mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.h > > diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig > index 88cff0e039b6..2e097a81df12 100644 > --- a/drivers/gpu/drm/exynos/Kconfig > +++ b/drivers/gpu/drm/exynos/Kconfig > @@ -94,6 +94,9 @@ config DRM_EXYNOS_G2D > help > Choose this option if you want to use Exynos G2D for DRM. > > +config DRM_EXYNOS_IPP > + bool > + > config DRM_EXYNOS_FIMC > bool "FIMC" > depends on BROKEN && MFD_SYSCON > diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile > index 09bb002c9555..f663490e949d 100644 > --- a/drivers/gpu/drm/exynos/Makefile > +++ b/drivers/gpu/drm/exynos/Makefile > @@ -17,6 +17,7 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER) += exynos_mixer.o > exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI) += exynos_hdmi.o > exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI) += exynos_drm_vidi.o > exynosdrm-$(CONFIG_DRM_EXYNOS_G2D) += exynos_drm_g2d.o > +exynosdrm-$(CONFIG_DRM_EXYNOS_IPP) += exynos_drm_ipp.o > exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC) += exynos_drm_fimc.o > exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR) += exynos_drm_rotator.o > exynosdrm-$(CONFIG_DRM_EXYNOS_GSC) += exynos_drm_gsc.o > diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c > index 2fc5d3c01390..de4cce258a5b 100644 > --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c > +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c > @@ -26,6 +26,7 @@ > #include "exynos_drm_fb.h" > #include "exynos_drm_gem.h" > #include "exynos_drm_plane.h" > +#include "exynos_drm_ipp.h" > #include "exynos_drm_vidi.h" > #include "exynos_drm_g2d.h" > #include "exynos_drm_iommu.h" > @@ -114,6 +115,14 @@ static const struct drm_ioctl_desc exynos_ioctls[] = { > DRM_AUTH | DRM_RENDER_ALLOW), > DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl, > DRM_AUTH | DRM_RENDER_ALLOW), > + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_RESOURCES, exynos_drm_ipp_get_res_ioctl, > + DRM_AUTH | DRM_RENDER_ALLOW), > + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_CAPS, exynos_drm_ipp_get_caps_ioctl, > + DRM_AUTH | DRM_RENDER_ALLOW), > + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_LIMITS, exynos_drm_ipp_get_limits_ioctl, > + DRM_AUTH | DRM_RENDER_ALLOW), > + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_COMMIT, exynos_drm_ipp_commit_ioctl, > + DRM_AUTH | DRM_RENDER_ALLOW), > }; > > static const struct file_operations exynos_drm_driver_fops = { > diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.c b/drivers/gpu/drm/exynos/exynos_drm_ipp.c > new file mode 100644 > index 000000000000..d54d23c6f2b0 > --- /dev/null > +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.c > @@ -0,0 +1,911 @@ > +/* > + * Copyright (C) 2017 Samsung Electronics Co.Ltd > + * Authors: > + * Marek Szyprowski <m.szyprowski@samsung.com> > + * > + * Exynos DRM Image Post Processing (IPP) related functions > + * > + * Permission is hereby granted, free of charge, to any person obtaining a > + * copy of this software and associated documentation files (the "Software"), > + * to deal in the Software without restriction, including without limitation > + * the rights to use, copy, modify, merge, publish, distribute, sublicense, > + * and/or sell copies of the Software, and to permit persons to whom the > + * Software is furnished to do so, subject to the following conditions: > + * > + * The above copyright notice and this permission notice shall be included in > + * all copies or substantial portions of the Software. > + */ > + > + > +#include <drm/drmP.h> > +#include <drm/drm_mode.h> > +#include <uapi/drm/exynos_drm.h> > + > +#include "exynos_drm_drv.h" > +#include "exynos_drm_gem.h" > +#include "exynos_drm_ipp.h" > + > +static int num_ipp; > +static LIST_HEAD(ipp_list); > + > +struct drm_pending_exynos_ipp_event { > + struct drm_pending_event base; > + struct drm_exynos_ipp_event event; > +}; > + > +/** > + * exynos_drm_ipp_register - Register a new picture processor hardware module > + * @dev: DRM device > + * @ipp: ipp module to init > + * @funcs: callbacks for the new ipp object > + * @caps: bitmask of ipp capabilities (%DRM_EXYNOS_IPP_CAP_*) > + * @formats: array of supported formats (%DRM_FORMAT_*) > + * @name: name (for debugging purposes) > + * > + * Initializes a ipp module. > + * > + * Returns: > + * Zero on success, error code on failure. > + */ > +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp, > + const struct exynos_drm_ipp_funcs *funcs, unsigned int caps, > + const struct exynos_drm_ipp_formats *formats, const char *name) > +{ > + WARN_ON(!ipp); > + WARN_ON(!funcs); > + WARN_ON(!formats); > + > + spin_lock_init(&ipp->lock); > + INIT_LIST_HEAD(&ipp->todo_list); > + init_waitqueue_head(&ipp->done_wq); > + ipp->dev = dev; > + ipp->funcs = funcs; > + ipp->capabilities = caps; > + ipp->name = name; > + ipp->num_formats = 0; > + ipp->formats = formats; > + while (formats++->fourcc) > + ipp->num_formats++; > + > + list_add_tail(&ipp->head, &ipp_list); mutex lock/unlock would be required as component framework did. > + ipp->id = num_ipp++; > + > + DRM_DEBUG_DRIVER("Registered ipp %d\n", ipp->id); > + > + return 0; > +} > + > +/** > + * exynos_drm_ipp_unregister - Unregister the picture processor module > + * @dev: DRM device > + * @ipp: ipp module > + */ > +void exynos_drm_ipp_unregister(struct drm_device *dev, > + struct exynos_drm_ipp *ipp) > +{ > + BUG_ON(ipp->task); > + BUG_ON(!list_empty(&ipp->todo_list)); > + list_del(&ipp->head); ditto. > +} > + > +/** > + * exynos_drm_ipp_ioctl_get_res_ioctl - enumerate all ipp modules > + * @dev: DRM device > + * @data: ioctl data > + * @file_priv: DRM file info > + * > + * Construct a list of ipp ids. > + * > + * Called by the user via ioctl. > + * > + * Returns: > + * Zero on success, negative errno on failure. > + */ > +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data, > + struct drm_file *file_priv) > +{ > + struct drm_exynos_ioctl_ipp_get_res *resp = data; > + struct exynos_drm_ipp *ipp; > + uint32_t __user *ipp_ptr = (uint32_t __user *) > + (unsigned long)resp->ipp_id_ptr; > + unsigned int count = num_ipp, copied = 0; > + > + /* > + * This ioctl is called twice, once to determine how much space is > + * needed, and the 2nd time to fill it. This is a same way as drm_mode_getresources function did. I'm not sure whether calling this function twice is best or not but no problem. > + */ > + if (count && resp->count_ipps >= count) { > + list_for_each_entry(ipp, &ipp_list, head) { > + if (put_user(ipp->id, ipp_ptr + copied)) > + return -EFAULT; > + copied++; > + } > + } > + resp->count_ipps = count; > + > + return 0; > +} > + > +static inline struct exynos_drm_ipp *__ipp_get(uint32_t id) > +{ > + struct exynos_drm_ipp *ipp; > + > + list_for_each_entry(ipp, &ipp_list, head) mutex lock here. ipp could be deleted by exynos_drm_ipp_unregister function. > + if (ipp->id == id) > + return ipp; > + return NULL; > +} > + > +/** > + * exynos_drm_ipp_ioctl_get_caps - get ipp module capabilities and formats > + * @dev: DRM device > + * @data: ioctl data > + * @file_priv: DRM file info > + * > + * Construct a structure describing ipp module capabilities. > + * > + * Called by the user via ioctl. > + * > + * Returns: > + * Zero on success, negative errno on failure. > + */ > +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data, > + struct drm_file *file_priv) > +{ > + struct drm_exynos_ioctl_ipp_get_caps *resp = data; > + void __user *ptr = (void __user *)(unsigned long)resp->formats_ptr; > + struct exynos_drm_ipp *ipp; > + int i; > + > + ipp = __ipp_get(resp->ipp_id); > + if (!ipp) > + return -ENOENT; > + > + resp->ipp_id = ipp->id; > + resp->capabilities = ipp->capabilities; > + > + /* > + * This ioctl is called twice, once to determine how much space is > + * needed, and the 2nd time to fill it. > + */ > + if (resp->formats_count >= ipp->num_formats) { > + for (i = 0; i < ipp->num_formats; i++) { > + struct drm_exynos_ipp_format tmp = { > + .fourcc = ipp->formats[i].fourcc, > + .type = ipp->formats[i].type, > + .modifier = ipp->formats[i].modifier, > + }; > + > + if (copy_to_user(ptr, &tmp, sizeof(tmp))) > + return -EFAULT; > + ptr += sizeof(tmp); > + } > + } > + resp->formats_count = ipp->num_formats; > + > + return 0; > +} > + > +static inline const struct exynos_drm_ipp_formats *__ipp_format_get( > + uint32_t fourcc, uint64_t mod, unsigned long type, Should argument, 'type', be 'unsigned long' data type? not 'unsigned int'? > + const struct exynos_drm_ipp_formats *f) > +{ > + for (; f->fourcc; f++) > + if ((f->type & type) && f->fourcc == fourcc && Here 'f->type' has 'unsigned int' data type. > + f->modifier == mod) > + return f; > + return NULL; > +} > + > +/** > + * exynos_drm_ipp_get_limits_ioctl - get ipp module limits > + * @dev: DRM device > + * @data: ioctl data > + * @file_priv: DRM file info > + * > + * Construct a structure describing ipp module limitations for provided > + * picture format. > + * > + * Called by the user via ioctl. > + * > + * Returns: > + * Zero on success, negative errno on failure. > + */ > +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data, > + struct drm_file *file_priv) > +{ > + struct drm_exynos_ioctl_ipp_get_limits *resp = data; > + void __user *ptr = (void __user *)(unsigned long)resp->limits_ptr; > + const struct exynos_drm_ipp_formats *format; > + const struct drm_exynos_ipp_limit *limits, *l; > + struct exynos_drm_ipp *ipp; > + int count = 0; > + > + if (resp->type != DRM_EXYNOS_IPP_FORMAT_SOURCE && > + resp->type != DRM_EXYNOS_IPP_FORMAT_DESTINATION) > + return -EINVAL; > + > + ipp = __ipp_get(resp->ipp_id); > + if (!ipp) > + return -ENOENT; > + > + format = __ipp_format_get(resp->fourcc, resp->modifier, resp->type, > + ipp->formats); > + if (!format) > + return -EINVAL; > + > + limits = format->limits; > + if (limits) > + for (l = limits; l->type; l++) > + count++; It seems no need to use loop to count. How about this? at device driver, static struct gsc_driverdata gsc_exynosxxxx_drvdata = { .clk_name... .. .limits = gsc_xxxx_limits, .limit_cnt = ARRAY_SIZE(gsc_xxx_limits), }; > + > + /* > + * This ioctl is called twice, once to determine how much space is > + * needed, and the 2nd time to fill it. > + */ > + if (count && resp->limits_count >= count) { > + for (l = limits; l->type; l++, ptr += sizeof(*l)) > + if (copy_to_user((void __user *)ptr, l, sizeof(*l))) > + return -EFAULT; > + } > + resp->limits_count = count; > + > + return 0; > +} > + > +static inline struct exynos_drm_ipp_task * > + exynos_drm_ipp_task_alloc(struct exynos_drm_ipp *ipp) > +{ > + struct exynos_drm_ipp_task *task; > + > + task = kzalloc(sizeof(*task), GFP_KERNEL); > + if (!task) > + return NULL; > + > + task->dev = ipp->dev; > + task->ipp = ipp; > + > + /* some defaults */ > + task->src.rect.w = task->dst.rect.w = UINT_MAX; > + task->src.rect.h = task->dst.rect.h = UINT_MAX; > + task->transform.rotation = DRM_MODE_ROTATE_0; > + > + DRM_DEBUG_DRIVER("Allocated task %pK\n", task); > + > + return task; > +} > + > +struct exynos_drm_param_map { > + unsigned int id; > + unsigned int size; > + unsigned int offset; > +} exynos_drm_ipp_params_maps[] = { > + { > + DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE, > + sizeof(struct drm_exynos_ipp_task_buffer), > + offsetof(struct exynos_drm_ipp_task, src.buf), > + }, { > + DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION, > + sizeof(struct drm_exynos_ipp_task_buffer), > + offsetof(struct exynos_drm_ipp_task, dst.buf), > + }, { > + DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE, > + sizeof(struct drm_exynos_ipp_task_rect), > + offsetof(struct exynos_drm_ipp_task, src.rect), > + }, { > + DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION, > + sizeof(struct drm_exynos_ipp_task_rect), > + offsetof(struct exynos_drm_ipp_task, dst.rect), > + }, { > + DRM_EXYNOS_IPP_TASK_TRANSFORM, > + sizeof(struct drm_exynos_ipp_task_transform), > + offsetof(struct exynos_drm_ipp_task, transform), > + }, { > + DRM_EXYNOS_IPP_TASK_ALPHA, > + sizeof(struct drm_exynos_ipp_task_alpha), > + offsetof(struct exynos_drm_ipp_task, alpha), > + }, > +}; > + > +static int exynos_drm_ipp_task_set(struct exynos_drm_ipp_task *task, > + struct drm_exynos_ioctl_ipp_commit *arg) > +{ > + unsigned int size = arg->params_size; > + void *p, *params; > + int ret = 0; > + > + if (size > PAGE_SIZE) > + return -ERANGE; I think you would need to check whether commands - which are passed by user space and arg->params_ptr points to - are valid or not here. With this verification, checking if size > PAGE_SIZE wouldn't be needed because valid commands include only below data structures, struct drm_exynos_ipp_task_buffer, struct drm_exynos_ipp_task_rect, struct drm_exynos_ipp_task_transform, and struct drm_exynos_ipp_task_alpha. This would be done by exynos_drm_ipp_task_check function which is called at exynos_drm_ipp_commit_ioctl function after exynos_drm_ipp_task_set funcion call. However, if task checking could be done prior to exynos_drm_ipp_set function and the commands passed by user-space have wrong data then you could skip unnecessary operation - copying commands from user-space to kernel. > + > + p = kmalloc(size, GFP_KERNEL); > + if (!p) > + return -ENOMEM; > + > + if (copy_from_user(p, (void __user *)(unsigned long)arg->params_ptr, > + size)) { > + ret = -EFAULT; > + DRM_DEBUG_DRIVER("Failed to copy configuration from userspace\n"); > + goto free; > + } > + > + params = p; > + while (size) { > + struct exynos_drm_param_map *map = exynos_drm_ipp_params_maps; > + u32 *id = params; > + int i; > + > + for (i = 0; i < ARRAY_SIZE(exynos_drm_ipp_params_maps); i++) > + if (map[i].id == *id) > + break; > + > + if (i < ARRAY_SIZE(exynos_drm_ipp_params_maps) && > + map[i].size <= size) { > + memcpy((void *)task + map[i].offset, params, > + map[i].size); > + params += map[i].size; > + size -= map[i].size; > + } else { > + ret = -EINVAL; > + goto free; > + } > + } > + > + DRM_DEBUG_DRIVER("Got task %pK configuration from userspace\n", task); > +free: > + kfree(p); > + return ret; > +} > + > +static int exynos_drm_ipp_task_setup_buffer(struct exynos_drm_ipp_buffer *buf, > + struct drm_file *filp) > +{ > + int ret = 0; > + int i; > + > + /* basic checks */ > + if (buf->buf.width == 0 || buf->buf.height == 0) > + return -EINVAL; > + buf->format = drm_format_info(buf->buf.fourcc); > + for (i = 0; i < buf->format->num_planes; i++) { > + unsigned int width = (i == 0) ? buf->buf.width : > + DIV_ROUND_UP(buf->buf.width, buf->format->hsub); > + > + if (buf->buf.pitch[i] == 0) > + buf->buf.pitch[i] = width * buf->format->cpp[i]; > + if (buf->buf.pitch[i] < width * buf->format->cpp[i]) > + return -EINVAL; > + if (!buf->buf.gem_id[i]) > + return -ENOENT; > + } > + > + /* get GEM buffers and check their size */ > + for (i = 0; i < buf->format->num_planes; i++) { > + unsigned int height = (i == 0) ? buf->buf.height : > + DIV_ROUND_UP(buf->buf.height, buf->format->vsub); > + unsigned long size = height * buf->buf.pitch[i] + > + buf->buf.offset[i]; > + struct drm_gem_object *obj = drm_gem_object_lookup(filp, > + buf->buf.gem_id[i]); > + if (!obj) { > + ret = -ENOENT; > + goto gem_free; > + } > + buf->exynos_gem[i] = to_exynos_gem(obj); > + > + if (size > buf->exynos_gem[i]->size) { > + i++; > + ret = -EINVAL; > + goto gem_free; > + } > + buf->dma_addr[i] = buf->exynos_gem[i]->dma_addr + > + buf->buf.offset[i]; > + } > + > + return 0; > +gem_free: > + while (i--) { > + drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base); > + buf->exynos_gem[i] = NULL; > + } > + return ret; > +} > + > +static void exynos_drm_ipp_task_release_buffer(struct exynos_drm_ipp_buffer *buf) > +{ > + int i; > + > + if (!buf->exynos_gem[0]) > + return; > + for (i = 0; i < buf->format->num_planes; i++) > + drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base); > +} > + > +static void exynos_drm_ipp_task_free(struct exynos_drm_ipp *ipp, > + struct exynos_drm_ipp_task *task) > +{ > + DRM_DEBUG_DRIVER("Freeing task %pK\n", task); > + > + exynos_drm_ipp_task_release_buffer(&task->src); > + exynos_drm_ipp_task_release_buffer(&task->dst); > + if (task->event) > + drm_event_cancel_free(ipp->dev, &task->event->base); > + kfree(task); > +} > + > +struct drm_ipp_limit { > + struct drm_exynos_ipp_limit_val h; > + struct drm_exynos_ipp_limit_val v; > +}; > + > +enum drm_ipp_size_id { > + IPP_LIMIT_BUFFER, IPP_LIMIT_AREA, IPP_LIMIT_ROTATED, IPP_LIMIT_MAX > +}; > + > +static const enum drm_ipp_size_id limit_id_fallback[IPP_LIMIT_MAX][4] = { > + [IPP_LIMIT_BUFFER] = { DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER }, > + [IPP_LIMIT_AREA] = { DRM_EXYNOS_IPP_LIMIT_SIZE_AREA, > + DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER }, > + [IPP_LIMIT_ROTATED] = { DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED, > + DRM_EXYNOS_IPP_LIMIT_SIZE_AREA, > + DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER }, > +}; > + > +static inline void __limit_set_val(unsigned int *ptr, unsigned int val) > +{ > + if (!*ptr) > + *ptr = val; > +} > + > +static void __get_size_limit(const struct drm_exynos_ipp_limit *limits, > + enum drm_ipp_size_id id, struct drm_ipp_limit *res) > +{ > + const struct drm_exynos_ipp_limit *l = limits; > + int i = 0; > + > + memset(res, 0, sizeof(*res)); > + for (i = 0; limit_id_fallback[id][i]; i++) > + for (l = limits; l->type; l++) { > + if (((l->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) != > + DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE) || > + ((l->type & DRM_EXYNOS_IPP_LIMIT_SIZE_MASK) != > + limit_id_fallback[id][i])) > + continue; > + __limit_set_val(&res->h.min, l->h.min); > + __limit_set_val(&res->h.max, l->h.max); > + __limit_set_val(&res->h.align, l->h.align); > + __limit_set_val(&res->v.min, l->v.min); > + __limit_set_val(&res->v.max, l->v.max); > + __limit_set_val(&res->v.align, l->v.align); > + } > +} > + > +static inline bool __align_check(unsigned int val, unsigned int align) > +{ > + if (align && (val & (align - 1))) { > + DRM_DEBUG_DRIVER("Value %d exceeds hw limits (align %d)\n", > + val, align); > + return false; > + } > + return true; > +} > + > +static inline bool __size_limit_check(unsigned int val, > + struct drm_exynos_ipp_limit_val *l) > +{ > + if ((l->min && val < l->min) || (l->max && val > l->max)) { > + DRM_DEBUG_DRIVER("Value %d exceeds hw limits (min %d, max %d)\n", > + val, l->min, l->max); > + return false; > + } > + return __align_check(val, l->align); > +} > + > +static int exynos_drm_ipp_check_size_limits(struct exynos_drm_ipp_buffer *buf, > + const struct drm_exynos_ipp_limit *limits, bool rotate, bool swap) > +{ > + enum drm_ipp_size_id id = rotate ? IPP_LIMIT_ROTATED : IPP_LIMIT_AREA; > + struct drm_ipp_limit l; > + struct drm_exynos_ipp_limit_val *lh = &l.h, *lv = &l.v; > + > + if (!limits) > + return 0; > + > + __get_size_limit(limits, IPP_LIMIT_BUFFER, &l); > + if (!__size_limit_check(buf->buf.width, &l.h) || > + !__size_limit_check(buf->buf.height, &l.v)) > + return -EINVAL; Printing out warning message here would be better for developers as a hint. > + > + if (swap) { > + lv = &l.h; > + lh = &l.v; > + } > + __get_size_limit(limits, id, &l); > + if (!__size_limit_check(buf->rect.w, lh) || > + !__align_check(buf->rect.x, lh->align) || > + !__size_limit_check(buf->rect.h, lv) || > + !__align_check(buf->rect.y, lv->align)) > + return -EINVAL; ditto. > + > + return 0; > +} > + > +static inline bool __scale_limit_check(unsigned int src, unsigned int dst, > + unsigned int min, unsigned int max) > +{ > + if ((max && (dst << 16) > src * max) || > + (min && (dst << 16) < src * min)) { > + DRM_DEBUG_DRIVER("Scale from %d to %d exceeds hw limits (ratio min %d.%05d, max %d.%05d)\n", > + src, dst, > + min >> 16, 100000 * (min & 0xffff) / (1 << 16), > + max >> 16, 100000 * (max & 0xffff) / (1 << 16)); > + return false; > + } > + return true; > +} > + > +static int exynos_drm_ipp_check_scale_limits(struct drm_exynos_ipp_task_rect *src, > + struct drm_exynos_ipp_task_rect *dst, > + const struct drm_exynos_ipp_limit *limits, bool swap) > +{ > + const struct drm_exynos_ipp_limit_val *lh, *lv; > + > + if (!limits) > + return 0; > + > + for (; limits->type; limits++) > + if ((limits->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) == > + DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE) > + break; > + if (!limits->type) > + return 0; > + > + lh = (!swap) ? &limits->h : &limits->v; > + lv = (!swap) ? &limits->v : &limits->h; > + > + if (!__scale_limit_check(src->w, dst->w, lh->min, lh->max) || > + !__scale_limit_check(src->h, dst->h, lv->min, lv->max)) > + return -EINVAL; ditto, hint as a warning. > + > + return 0; > +} > + > +static int exynos_drm_ipp_task_check(struct exynos_drm_ipp_task *task, > + struct drm_file *filp) > +{ > + struct exynos_drm_ipp *ipp = task->ipp; > + const struct exynos_drm_ipp_formats *src_format, *dst_format; > + struct exynos_drm_ipp_buffer *src = &task->src, *dst = &task->dst; > + unsigned int rotation = task->transform.rotation; > + int ret = 0; > + bool scale = false, swap = false; > + > + DRM_DEBUG_DRIVER("Checking task %pK\n", task); > + > + if (src->rect.w == UINT_MAX) > + src->rect.w = src->buf.width; > + if (src->rect.h == UINT_MAX) > + src->rect.h = src->buf.height; > + if (dst->rect.w == UINT_MAX) > + dst->rect.w = dst->buf.width; > + if (dst->rect.h == UINT_MAX) > + dst->rect.h = dst->buf.height; > + > + if (src->rect.x + src->rect.w > (src->buf.width) || > + src->rect.y + src->rect.h > (src->buf.height) || > + dst->rect.x + dst->rect.w > (dst->buf.width) || > + dst->rect.y + dst->rect.h > (dst->buf.height)) { > + DRM_DEBUG_DRIVER("Task %pK: defined area is outside provided buffers\n", > + task); > + return -EINVAL; > + } > + > + if (drm_rotation_90_or_270(rotation)) > + swap = true; > + > + if ((!swap && (src->rect.w != dst->rect.w || src->rect.h != dst->rect.h)) || > + (swap && (src->rect.w != dst->rect.h || src->rect.h != dst->rect.w))) > + scale = true; > + > + if ((!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CROP) && > + (src->rect.x || src->rect.y || dst->rect.x || dst->rect.y)) || > + (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_ROTATE) && > + rotation != DRM_MODE_ROTATE_0) || > + (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_SCALE) && scale) || > + (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CONVERT) && > + src->buf.fourcc != dst->buf.fourcc)) { > + DRM_DEBUG_DRIVER("Task %pK: hw capabilities exceeded\n", task); > + return -EINVAL; > + } > + > + src_format = __ipp_format_get(src->buf.fourcc, src->buf.modifier, > + DRM_EXYNOS_IPP_FORMAT_SOURCE, ipp->formats); > + if (!src_format) { > + DRM_DEBUG_DRIVER("Task %pK: src format not supported\n", task); > + return -EINVAL; > + } > + ret = exynos_drm_ipp_check_size_limits(src, src_format->limits, > + rotation != DRM_MODE_ROTATE_0, false); > + if (ret) > + return ret; > + ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect, > + src_format->limits, swap); > + if (ret) > + return ret; > + > + dst_format = __ipp_format_get(dst->buf.fourcc, dst->buf.modifier, > + DRM_EXYNOS_IPP_FORMAT_DESTINATION, ipp->formats); > + if (!dst_format) { > + DRM_DEBUG_DRIVER("Task %pK: dst format not supported\n", task); > + return -EINVAL; > + } > + ret = exynos_drm_ipp_check_size_limits(dst, dst_format->limits, > + rotation != DRM_MODE_ROTATE_0, swap); > + if (ret) > + return ret; > + ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect, > + dst_format->limits, swap); > + if (ret) > + return ret; > + > + ret = exynos_drm_ipp_task_setup_buffer(src, filp); > + if (ret) { > + DRM_DEBUG_DRIVER("Task %pK: src buffer setup failed\n", task); > + return ret; > + } > + ret = exynos_drm_ipp_task_setup_buffer(dst, filp); > + if (ret) { > + DRM_DEBUG_DRIVER("Task %pK: dst buffer setup failed\n", task); > + return ret; > + } > + > + if (ipp->funcs->check) { > + ret = ipp->funcs->check(ipp, task); It would be better to check HW or driver specific things prior to buffer setup. > + if (ret) { > + DRM_DEBUG_DRIVER("Task %pK: driver specific check failed\n", > + task); > + return ret; > + } > + } > + > + DRM_DEBUG_DRIVER("Task %pK: all checks done.\n", task); > + > + return ret; > +} > + > +static int exynos_drm_ipp_event_create(struct exynos_drm_ipp_task *task, > + struct drm_file *file_priv, uint64_t user_data) > +{ > + struct drm_pending_exynos_ipp_event *e = NULL; > + int ret; > + > + e = kzalloc(sizeof(*e), GFP_KERNEL); > + if (!e) > + return -ENOMEM; > + > + e->event.base.type = DRM_EXYNOS_IPP_EVENT; > + e->event.base.length = sizeof(e->event); > + e->event.user_data = user_data; > + > + ret = drm_event_reserve_init(task->dev, file_priv, &e->base, > + &e->event.base); > + if (ret) > + goto free; > + > + task->event = e; > + return 0; > +free: > + kfree(e); > + return ret; > +} > + > +static void exynos_drm_ipp_event_send(struct exynos_drm_ipp_task *task) > +{ > + struct timeval now = ktime_to_timeval(ktime_get()); > + > + task->event->event.tv_sec = now.tv_sec; > + task->event->event.tv_usec = now.tv_usec; > + task->event->event.sequence = atomic_inc_return(&task->ipp->sequence); > + > + drm_send_event(task->dev, &task->event->base); > +} > + > +static int exynos_drm_ipp_task_cleanup(struct exynos_drm_ipp_task *task) > +{ > + int ret = task->ret; > + > + if (ret == 0 && task->event) { > + exynos_drm_ipp_event_send(task); > + /* ensure event won't be canceled on task free */ > + task->event = NULL; > + } > + > + exynos_drm_ipp_task_free(task->ipp, task); > + return ret; > +} > + > +static void exynos_drm_ipp_cleanup_work(struct work_struct *work) > +{ > + struct exynos_drm_ipp_task *task = container_of(work, > + struct exynos_drm_ipp_task, cleanup_work); > + > + exynos_drm_ipp_task_cleanup(task); > +} > + > +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp); It's not good. How about just moving implementaion of this function above side? Thanks, Inki Dae > + > +/** > + * exynos_drm_ipp_task_done - finish given task and set return code > + * @task: ipp task to finish > + * @ret: error code or 0 if operation has been performed successfully > + */ > +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret) > +{ > + struct exynos_drm_ipp *ipp = task->ipp; > + unsigned long flags; > + > + DRM_DEBUG_DRIVER("ipp: %d, task %pK done\n", ipp->id, task); > + > + spin_lock_irqsave(&ipp->lock, flags); > + if (ipp->task == task) > + ipp->task = NULL; > + task->flags |= DRM_EXYNOS_IPP_TASK_DONE; > + task->ret = ret; > + spin_unlock_irqrestore(&ipp->lock, flags); > + > + exynos_drm_ipp_next_task(ipp); > + wake_up(&ipp->done_wq); > + > + if (task->flags & DRM_EXYNOS_IPP_TASK_ASYNC) { > + INIT_WORK(&task->cleanup_work, exynos_drm_ipp_cleanup_work); > + schedule_work(&task->cleanup_work); > + } > +} > + > +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp) > +{ > + struct exynos_drm_ipp_task *task; > + unsigned long flags; > + int ret; > + > + DRM_DEBUG_DRIVER("ipp: %d, try to run new task\n", ipp->id); > + > + spin_lock_irqsave(&ipp->lock, flags); > + > + if (ipp->task || list_empty(&ipp->todo_list)) { > + spin_unlock_irqrestore(&ipp->lock, flags); > + return; > + } > + > + task = list_first_entry(&ipp->todo_list, struct exynos_drm_ipp_task, > + head); > + list_del_init(&task->head); > + ipp->task = task; > + > + spin_unlock_irqrestore(&ipp->lock, flags); > + > + DRM_DEBUG_DRIVER("ipp: %d, selected task %pK to run\n", ipp->id, task); > + > + ret = ipp->funcs->commit(ipp, task); > + if (ret) > + exynos_drm_ipp_task_done(task, ret); > +} > + > +static void exynos_drm_ipp_schedule_task(struct exynos_drm_ipp *ipp, > + struct exynos_drm_ipp_task *task) > +{ > + unsigned long flags; > + > + spin_lock_irqsave(&ipp->lock, flags); > + list_add(&task->head, &ipp->todo_list); > + spin_unlock_irqrestore(&ipp->lock, flags); > + > + exynos_drm_ipp_next_task(ipp); > +} > + > +static void exynos_drm_ipp_task_abort(struct exynos_drm_ipp *ipp, > + struct exynos_drm_ipp_task *task) > +{ > + unsigned long flags; > + > + spin_lock_irqsave(&ipp->lock, flags); > + if (task->flags & DRM_EXYNOS_IPP_TASK_DONE) { > + /* already completed task */ > + exynos_drm_ipp_task_cleanup(task); > + } else if (ipp->task != task) { > + /* task has not been scheduled for execution yet */ > + list_del_init(&task->head); > + exynos_drm_ipp_task_cleanup(task); > + } else { > + /* > + * currently processed task, call abort() and perform > + * cleanup with async worker > + */ > + task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC; > + spin_unlock_irqrestore(&ipp->lock, flags); > + if (ipp->funcs->abort) > + ipp->funcs->abort(ipp, task); > + return; > + } > + spin_unlock_irqrestore(&ipp->lock, flags); > +} > + > +/** > + * exynos_drm_ipp_commit_ioctl - perform image processing operation > + * @dev: DRM device > + * @data: ioctl data > + * @file_priv: DRM file info > + * > + * Construct a ipp task from the set of properties provided from the user > + * and try to schedule it to framebuffer processor hardware. > + * > + * Called by the user via ioctl. > + * > + * Returns: > + * Zero on success, negative errno on failure. > + */ > +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, void *data, > + struct drm_file *file_priv) > +{ > + struct drm_exynos_ioctl_ipp_commit *arg = data; > + struct exynos_drm_ipp *ipp; > + struct exynos_drm_ipp_task *task; > + int ret = 0; > + > + if ((arg->flags & ~DRM_EXYNOS_IPP_FLAGS) || arg->reserved) > + return -EINVAL; > + > + /* can't test and expect an event at the same time */ > + if ((arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY) && > + (arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT)) > + return -EINVAL; > + > + ipp = __ipp_get(arg->ipp_id); > + if (!ipp) > + return -ENOENT; > + > + task = exynos_drm_ipp_task_alloc(ipp); > + if (!task) > + return -ENOMEM; > + > + ret = exynos_drm_ipp_task_set(task, arg); > + if (ret) > + goto free; > + > + ret = exynos_drm_ipp_task_check(task, file_priv); > + if (ret || arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY) > + goto free; > + > + if (arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT) { > + ret = exynos_drm_ipp_event_create(task, file_priv, > + arg->user_data); > + if (ret) > + goto free; > + } > + > + /* > + * Queue task for processing on the hardware. task object will be > + * then freed after exynos_drm_ipp_task_done() > + */ > + if (arg->flags & DRM_EXYNOS_IPP_FLAG_NONBLOCK) { > + DRM_DEBUG_DRIVER("ipp: %d, nonblocking processing task %pK\n", > + ipp->id, task); > + > + task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC; > + exynos_drm_ipp_schedule_task(task->ipp, task); > + ret = 0; > + } else { > + DRM_DEBUG_DRIVER("ipp: %d, processing task %pK\n", ipp->id, task); > + exynos_drm_ipp_schedule_task(ipp, task); > + ret = wait_event_interruptible(ipp->done_wq, > + task->flags & DRM_EXYNOS_IPP_TASK_DONE); > + if (ret) > + exynos_drm_ipp_task_abort(ipp, task); > + else > + ret = exynos_drm_ipp_task_cleanup(task); > + } > + return ret; > +free: > + exynos_drm_ipp_task_free(ipp, task); > + > + return ret; > +} > diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.h b/drivers/gpu/drm/exynos/exynos_drm_ipp.h > new file mode 100644 > index 000000000000..199eb2006410 > --- /dev/null > +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.h > @@ -0,0 +1,195 @@ > +/* > + * Copyright (c) 2017 Samsung Electronics Co., Ltd. > + * > + * This program is free software; you can redistribute it and/or modify it > + * under the terms of the GNU General Public License as published by the > + * Free Software Foundation; either version 2 of the License, or (at your > + * option) any later version. > + */ > + > +#ifndef _EXYNOS_DRM_IPP_H_ > +#define _EXYNOS_DRM_IPP_H_ > + > +#include <drm/drmP.h> > + > +struct exynos_drm_ipp; > +struct exynos_drm_ipp_task; > + > +/** > + * struct exynos_drm_ipp_funcs - exynos_drm_ipp control functions > + */ > +struct exynos_drm_ipp_funcs { > + /** > + * @check: > + * > + * This is the optional hook to validate an ipp task. This function > + * must reject any task which the hardware or driver doesn't support. > + * This includes but is of course not limited to: > + * > + * - Checking that the buffers, scaling and placement requirements > + * and so on are within the limits of the hardware not specified by > + * limits array. > + * > + * - The driver does not need to repeat basic input validation like > + * done in the exynos_drm_ipp_task_check() function. The core does > + * that before calling this hook. > + * > + * RETURNS: > + * > + * 0 on success or one of the below negative error codes: > + * > + * - -EINVAL, if any of the above constraints are violated. > + */ > + int (*check)(struct exynos_drm_ipp *ipp, > + struct exynos_drm_ipp_task *task); > + > + /** > + * @commit: > + * > + * This is the main entry point to start framebuffer processing > + * in the hardware. The exynos_drm_ipp_task has been already validated. > + * This function must not wait until the device finishes processing. > + * When the driver finishes processing, it has to call > + * exynos_exynos_drm_ipp_task_done() function. > + * > + * RETURNS: > + * > + * 0 on success or negative error codes in case of failure. > + */ > + int (*commit)(struct exynos_drm_ipp *ipp, > + struct exynos_drm_ipp_task *task); > + > + /** > + * @abort: > + * > + * Informs the driver that it has to abort the currently running > + * task as soon as possible (i.e. as soon as it can stop the device > + * safely), even if the task would not have been finished by then. > + * After the driver performs the necessary steps, it has to call > + * exynos_drm_ipp_task_done() (as if the task ended normally). > + * This function does not have to (and will usually not) wait > + * until the device enters a state when it can be stopped. > + */ > + void (*abort)(struct exynos_drm_ipp *ipp, > + struct exynos_drm_ipp_task *task); > +}; > + > +/** > + * struct exynos_drm_ipp - central picture processor module structure > + */ > +struct exynos_drm_ipp { > + struct drm_device *dev; > + struct list_head head; > + unsigned int id; > + > + const char *name; > + const struct exynos_drm_ipp_funcs *funcs; > + unsigned int capabilities; > + const struct exynos_drm_ipp_formats *formats; > + unsigned int num_formats; > + atomic_t sequence; > + > + spinlock_t lock; > + struct exynos_drm_ipp_task *task; > + struct list_head todo_list; > + wait_queue_head_t done_wq; > +}; > + > +struct exynos_drm_ipp_buffer { > + struct drm_exynos_ipp_task_buffer buf; > + struct drm_exynos_ipp_task_rect rect; > + > + struct exynos_drm_gem *exynos_gem[MAX_FB_BUFFER]; > + const struct drm_format_info *format; > + dma_addr_t dma_addr[MAX_FB_BUFFER]; > +}; > + > +/** > + * struct exynos_drm_ipp_task - a structure describing transformation that > + * has to be performed by the picture processor hardware module > + */ > +struct exynos_drm_ipp_task { > + struct drm_device *dev; > + struct exynos_drm_ipp *ipp; > + struct list_head head; > + > + struct exynos_drm_ipp_buffer src; > + struct exynos_drm_ipp_buffer dst; > + > + struct drm_exynos_ipp_task_transform transform; > + struct drm_exynos_ipp_task_alpha alpha; > + > + struct work_struct cleanup_work; > + unsigned int flags; > + int ret; > + > + struct drm_pending_exynos_ipp_event *event; > +}; > + > +#define DRM_EXYNOS_IPP_TASK_DONE (1 << 0) > +#define DRM_EXYNOS_IPP_TASK_ASYNC (1 << 1) > + > +struct exynos_drm_ipp_formats { > + u32 fourcc; > + u32 type; > + u64 modifier; > + const struct drm_exynos_ipp_limit *limits; > +}; > + > +/* helper macros to set exynos_drm_ipp_formats structure and limits*/ > +#define IPP_SRCDST_MFORMAT(f, m, l) \ > + .fourcc = DRM_FORMAT_##f, .modifier = m, .limits = l, \ > + .type = (DRM_EXYNOS_IPP_FORMAT_SOURCE | DRM_EXYNOS_IPP_FORMAT_DESTINATION) > + > +#define IPP_SRCDST_FORMAT(f, l) IPP_SRCDST_MFORMAT(f, 0, l) > + > +#define IPP_SIZE_LIMIT(l, val...) \ > + .type = (DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE | DRM_EXYNOS_IPP_LIMIT_SIZE_##l), \ > + val > + > +#define IPP_SCALE_LIMIT(val...) \ > + .type = (DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE),\ > + val > + > +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp, > + const struct exynos_drm_ipp_funcs *funcs, unsigned int caps, > + const struct exynos_drm_ipp_formats *formats, const char *name); > +void exynos_drm_ipp_unregister(struct drm_device *dev, struct exynos_drm_ipp *ipp); > + > +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret); > + > +#ifdef CONFIG_DRM_EXYNOS_IPP > +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data, > + struct drm_file *file_priv); > +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data, > + struct drm_file *file_priv); > +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data, > + struct drm_file *file_priv); > +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, > + void *data, struct drm_file *file_priv); > +#else > +static inline int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, > + void *data, struct drm_file *file_priv) > +{ > + struct drm_exynos_ioctl_ipp_get_res *resp = data; > + > + resp->count_ipps = 0; > + return 0; > +} > +static inline int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, > + void *data, struct drm_file *file_priv) > +{ > + return -ENODEV; > +} > +static inline int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, > + void *data, struct drm_file *file_priv) > +{ > + return -ENODEV; > +} > +static inline int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, > + void *data, struct drm_file *file_priv) > +{ > + return -ENODEV; > +} > +#endif > +#endif > diff --git a/include/uapi/drm/exynos_drm.h b/include/uapi/drm/exynos_drm.h > index 5d9a08d69ba7..974cbb579f90 100644 > --- a/include/uapi/drm/exynos_drm.h > +++ b/include/uapi/drm/exynos_drm.h > @@ -134,6 +134,219 @@ struct drm_exynos_g2d_exec { > __u64 async; > }; > > +/* Exynos DRM IPP v2 API */ > + > +/** > + * Enumerate available IPP hardware modules. > + * > + * @count_ipps: size of ipp_id array / number of ipp modules (set by driver) > + * @reserved: padding > + * @ipp_id_ptr: pointer to ipp_id array or NULL > + */ > +struct drm_exynos_ioctl_ipp_get_res { > + __u32 count_ipps; > + __u32 reserved; > + __u64 ipp_id_ptr; > +}; > + > +enum drm_exynos_ipp_format_type { > + DRM_EXYNOS_IPP_FORMAT_SOURCE = 0x01, > + DRM_EXYNOS_IPP_FORMAT_DESTINATION = 0x02, > +}; > + > +struct drm_exynos_ipp_format { > + __u32 fourcc; > + __u32 type; > + __u64 modifier; > +}; > + > +enum drm_exynos_ipp_capability { > + DRM_EXYNOS_IPP_CAP_CROP = 0x01, > + DRM_EXYNOS_IPP_CAP_ROTATE = 0x02, > + DRM_EXYNOS_IPP_CAP_SCALE = 0x04, > + DRM_EXYNOS_IPP_CAP_CONVERT = 0x08, > +}; > + > +/** > + * Get IPP hardware capabilities and supported image formats. > + * > + * @ipp_id: id of IPP module to query > + * @capabilities: bitmask of drm_exynos_ipp_capability (set by driver) > + * @reserved: padding > + * @formats_count: size of formats array (in entries) / number of filled > + * formats (set by driver) > + * @formats_ptr: pointer to formats array or NULL > + */ > +struct drm_exynos_ioctl_ipp_get_caps { > + __u32 ipp_id; > + __u32 capabilities; > + __u32 reserved; > + __u32 formats_count; > + __u64 formats_ptr; > +}; > + > +enum drm_exynos_ipp_limit_type { > + /* size (horizontal/vertial) limits, in pixels (min, max, alignment) */ > + DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE = 0x0001, > + /* scale ratio (horizonta/vertial), 16.16 fixed point (min, max) */ > + DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE = 0x0002, > + > + /* image buffer area */ > + DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER = 0x0001 << 16, > + /* src/dst rectangle area */ > + DRM_EXYNOS_IPP_LIMIT_SIZE_AREA = 0x0002 << 16, > + /* src/dst rectangle area when rotation enabled */ > + DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED = 0x0003 << 16, > + > + DRM_EXYNOS_IPP_LIMIT_TYPE_MASK = 0x000f, > + DRM_EXYNOS_IPP_LIMIT_SIZE_MASK = 0x000f << 16, > +}; > + > +struct drm_exynos_ipp_limit_val { > + __u32 min; > + __u32 max; > + __u32 align; > + __u32 reserved; > +}; > + > +/** > + * IPP module limitation. > + * > + * @type: limit type (see drm_exynos_ipp_limit_type enum) > + * @reserved: padding > + * @h: horizontal limits > + * @v: vertical limits > + */ > +struct drm_exynos_ipp_limit { > + __u32 type; > + __u32 reserved; > + struct drm_exynos_ipp_limit_val h; > + struct drm_exynos_ipp_limit_val v; > +}; > + > +/** > + * Get IPP limits for given image format. > + * > + * @ipp_id: id of IPP module to query > + * @fourcc: image format code (see DRM_FORMAT_* in drm_fourcc.h) > + * @modifier: image format modifier (see DRM_FORMAT_MOD_* in drm_fourcc.h) > + * @type: source/destination identifier (drm_exynos_ipp_format_flag enum) > + * @limits_count: size of limits array (in entries) / number of filled entries > + * (set by driver) > + * @limits_ptr: pointer to limits array or NULL > + */ > +struct drm_exynos_ioctl_ipp_get_limits { > + __u32 ipp_id; > + __u32 fourcc; > + __u64 modifier; > + __u32 type; > + __u32 limits_count; > + __u64 limits_ptr; > +}; > + > +enum drm_exynos_ipp_task_id { > + /* buffer described by struct drm_exynos_ipp_task_buffer */ > + DRM_EXYNOS_IPP_TASK_BUFFER = 0x0001, > + /* rectangle described by struct drm_exynos_ipp_task_rect */ > + DRM_EXYNOS_IPP_TASK_RECTANGLE = 0x0002, > + /* transformation described by struct drm_exynos_ipp_task_transform */ > + DRM_EXYNOS_IPP_TASK_TRANSFORM = 0x0003, > + /* alpha configuration described by struct drm_exynos_ipp_task_alpha */ > + DRM_EXYNOS_IPP_TASK_ALPHA = 0x0004, > + > + /* source image data (for buffer and rectangle chunks) */ > + DRM_EXYNOS_IPP_TASK_TYPE_SOURCE = 0x0001 << 16, > + /* destination image data (for buffer and rectangle chunks) */ > + DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION = 0x0002 << 16, > +}; > + > +/** > + * Memory buffer with image data. > + * > + * @id: must be DRM_EXYNOS_IPP_TASK_BUFFER > + * other parameters are same as for AddFB2 generic DRM ioctl > + */ > +struct drm_exynos_ipp_task_buffer { > + __u32 id; > + __u32 fourcc; > + __u32 width, height; > + __u32 gem_id[4]; > + __u32 offset[4]; > + __u32 pitch[4]; > + __u64 modifier; > +}; > + > +/** > + * Rectangle for processing. > + * > + * @id: must be DRM_EXYNOS_IPP_TASK_RECTANGLE > + * @reserved: padding > + * @x,@y: left corner in pixels > + * @w,@h: width/height in pixels > + */ > +struct drm_exynos_ipp_task_rect { > + __u32 id; > + __u32 reserved; > + __u32 x; > + __u32 y; > + __u32 w; > + __u32 h; > +}; > + > +/** > + * Image tranformation description. > + * > + * @id: must be DRM_EXYNOS_IPP_TASK_TRANSFORM > + * @rotation: DRM_MODE_ROTATE_* and DRM_MODE_REFLECT_* values > + */ > +struct drm_exynos_ipp_task_transform { > + __u32 id; > + __u32 rotation; > +}; > + > +/** > + * Image global alpha configuration for formats without alpha values. > + * > + * @id: must be DRM_EXYNOS_IPP_TASK_ALPHA > + * @value: global alpha value (0-255) > + */ > +struct drm_exynos_ipp_task_alpha { > + __u32 id; > + __u32 value; > +}; > + > +enum drm_exynos_ipp_flag { > + /* generate DRM event after processing */ > + DRM_EXYNOS_IPP_FLAG_EVENT = 0x01, > + /* dry run, only check task parameters */ > + DRM_EXYNOS_IPP_FLAG_TEST_ONLY = 0x02, > + /* non-blocking processing */ > + DRM_EXYNOS_IPP_FLAG_NONBLOCK = 0x04, > +}; > + > +#define DRM_EXYNOS_IPP_FLAGS (DRM_EXYNOS_IPP_FLAG_EVENT |\ > + DRM_EXYNOS_IPP_FLAG_TEST_ONLY | DRM_EXYNOS_IPP_FLAG_NONBLOCK) > + > +/** > + * Perform image processing described by array of drm_exynos_ipp_task_* > + * structures (parameters array). > + * > + * @ipp_id: id of IPP module to run the task > + * @flags: bitmask of drm_exynos_ipp_flag values > + * @reserved: padding > + * @params_size: size of parameters array (in bytes) > + * @params_ptr: pointer to parameters array or NULL > + * @user_data: (optional) data for drm event > + */ > +struct drm_exynos_ioctl_ipp_commit { > + __u32 ipp_id; > + __u32 flags; > + __u32 reserved; > + __u32 params_size; > + __u64 params_ptr; > + __u64 user_data; > +}; > + > #define DRM_EXYNOS_GEM_CREATE 0x00 > #define DRM_EXYNOS_GEM_MAP 0x01 > /* Reserved 0x03 ~ 0x05 for exynos specific gem ioctl */ > @@ -146,6 +359,11 @@ struct drm_exynos_g2d_exec { > #define DRM_EXYNOS_G2D_EXEC 0x22 > > /* Reserved 0x30 ~ 0x33 for obsolete Exynos IPP ioctls */ > +/* IPP - Image Post Processing */ > +#define DRM_EXYNOS_IPP_GET_RESOURCES 0x40 > +#define DRM_EXYNOS_IPP_GET_CAPS 0x41 > +#define DRM_EXYNOS_IPP_GET_LIMITS 0x42 > +#define DRM_EXYNOS_IPP_COMMIT 0x43 > > #define DRM_IOCTL_EXYNOS_GEM_CREATE DRM_IOWR(DRM_COMMAND_BASE + \ > DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create) > @@ -164,8 +382,18 @@ struct drm_exynos_g2d_exec { > #define DRM_IOCTL_EXYNOS_G2D_EXEC DRM_IOWR(DRM_COMMAND_BASE + \ > DRM_EXYNOS_G2D_EXEC, struct drm_exynos_g2d_exec) > > +#define DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES DRM_IOWR(DRM_COMMAND_BASE + \ > + DRM_EXYNOS_IPP_GET_RESOURCES, struct drm_exynos_ioctl_ipp_get_res) > +#define DRM_IOCTL_EXYNOS_IPP_GET_CAPS DRM_IOWR(DRM_COMMAND_BASE + \ > + DRM_EXYNOS_IPP_GET_CAPS, struct drm_exynos_ioctl_ipp_get_caps) > +#define DRM_IOCTL_EXYNOS_IPP_GET_LIMITS DRM_IOWR(DRM_COMMAND_BASE + \ > + DRM_EXYNOS_IPP_GET_LIMITS, struct drm_exynos_ioctl_ipp_get_limits) > +#define DRM_IOCTL_EXYNOS_IPP_COMMIT DRM_IOWR(DRM_COMMAND_BASE + \ > + DRM_EXYNOS_IPP_COMMIT, struct drm_exynos_ioctl_ipp_commit) > + > /* EXYNOS specific events */ > #define DRM_EXYNOS_G2D_EVENT 0x80000000 > +#define DRM_EXYNOS_IPP_EVENT 0x80000002 > > struct drm_exynos_g2d_event { > struct drm_event base; > @@ -176,6 +404,16 @@ struct drm_exynos_g2d_event { > __u32 reserved; > }; > > +struct drm_exynos_ipp_event { > + struct drm_event base; > + __u64 user_data; > + __u32 tv_sec; > + __u32 tv_usec; > + __u32 ipp_id; > + __u32 sequence; > + __u64 reserved; > +}; > + > #if defined(__cplusplus) > } > #endif > ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH v4 2/9] drm/exynos: ipp: Add IPP v2 framework 2017-11-01 1:13 ` Inki Dae @ 2017-11-03 11:20 ` Marek Szyprowski 0 siblings, 0 replies; 20+ messages in thread From: Marek Szyprowski @ 2017-11-03 11:20 UTC (permalink / raw) To: Inki Dae, dri-devel, linux-samsung-soc Cc: Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon On 2017-11-01 02:13, Inki Dae wrote: > Hi Marek, > > Thanks for your effort and sorry for late. Below are trivial my comments, > > 2017년 10월 23일 16:54에 Marek Szyprowski 이(가) 쓴 글: >> This patch adds Exynos IPP v2 subsystem and userspace API. >> >> New userspace API is focused ONLY on memory-to-memory image processing. >> The two remainging IPP operation modes (framebuffer writeback and >> local-path output with image processing) can be implemented using >> standard DRM features: writeback connectors and additional DRM planes with >> scaling features. >> >> V2 IPP userspace API is not compatible with old IPP ioctls. This is a >> significant change, but the old IPP subsystem in mainline Linux kernel was >> partially disfunctional anyway and not used in any open-source project. >> >> V2 IPP userspace API is based on stateless approach, which much better fits >> to memory-to-memory image processing model. It also provides support for >> all image formats, which are both already defined in DRM API and supported >> by the existing IPP hardware modules. >> >> The API consists of the following ioctls: >> - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image >> processing modules, >> - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image >> formats of given IPP module, >> - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for >> selected image format of given IPP module, >> - DRM_IOCTL_EXYNOS_IPP_COMMIT: to perform operation described by the >> provided structures (source and destination buffers, operation rectangle, >> transformation, etc). >> >> The proposed userspace API is extensible. In the future more advanced image >> processing operations can be defined to support for example blending. >> >> Userspace API is fully functional also on DRM render nodes, so it is not >> limited to the root/privileged client. >> >> Internal driver API also has been completely rewritten. New IPP core >> performs all possible input validation, checks and object life-time >> control. The drivers can focus only on writing configuration to hardware >> registers. Stateless nature of DRM_IOCTL_EXYNOS_IPP_COMMIT ioctl simplifies >> the driver API. Minimal driver needs to provide a single callback for >> starting processing and an array with supported image formats. >> >> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> >> Tested-by: Hoegeun Kwon <hoegeun.kwon@samsung.com> >> --- >> drivers/gpu/drm/exynos/Kconfig | 3 + >> drivers/gpu/drm/exynos/Makefile | 1 + >> drivers/gpu/drm/exynos/exynos_drm_drv.c | 9 + >> drivers/gpu/drm/exynos/exynos_drm_ipp.c | 911 ++++++++++++++++++++++++++++++++ >> drivers/gpu/drm/exynos/exynos_drm_ipp.h | 195 +++++++ >> include/uapi/drm/exynos_drm.h | 238 +++++++++ >> 6 files changed, 1357 insertions(+) >> create mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.c >> create mode 100644 drivers/gpu/drm/exynos/exynos_drm_ipp.h >> >> diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig >> index 88cff0e039b6..2e097a81df12 100644 >> --- a/drivers/gpu/drm/exynos/Kconfig >> +++ b/drivers/gpu/drm/exynos/Kconfig >> @@ -94,6 +94,9 @@ config DRM_EXYNOS_G2D >> help >> Choose this option if you want to use Exynos G2D for DRM. >> >> +config DRM_EXYNOS_IPP >> + bool >> + >> config DRM_EXYNOS_FIMC >> bool "FIMC" >> depends on BROKEN && MFD_SYSCON >> diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile >> index 09bb002c9555..f663490e949d 100644 >> --- a/drivers/gpu/drm/exynos/Makefile >> +++ b/drivers/gpu/drm/exynos/Makefile >> @@ -17,6 +17,7 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_MIXER) += exynos_mixer.o >> exynosdrm-$(CONFIG_DRM_EXYNOS_HDMI) += exynos_hdmi.o >> exynosdrm-$(CONFIG_DRM_EXYNOS_VIDI) += exynos_drm_vidi.o >> exynosdrm-$(CONFIG_DRM_EXYNOS_G2D) += exynos_drm_g2d.o >> +exynosdrm-$(CONFIG_DRM_EXYNOS_IPP) += exynos_drm_ipp.o >> exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC) += exynos_drm_fimc.o >> exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR) += exynos_drm_rotator.o >> exynosdrm-$(CONFIG_DRM_EXYNOS_GSC) += exynos_drm_gsc.o >> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c >> index 2fc5d3c01390..de4cce258a5b 100644 >> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c >> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c >> @@ -26,6 +26,7 @@ >> #include "exynos_drm_fb.h" >> #include "exynos_drm_gem.h" >> #include "exynos_drm_plane.h" >> +#include "exynos_drm_ipp.h" >> #include "exynos_drm_vidi.h" >> #include "exynos_drm_g2d.h" >> #include "exynos_drm_iommu.h" >> @@ -114,6 +115,14 @@ static const struct drm_ioctl_desc exynos_ioctls[] = { >> DRM_AUTH | DRM_RENDER_ALLOW), >> DRM_IOCTL_DEF_DRV(EXYNOS_G2D_EXEC, exynos_g2d_exec_ioctl, >> DRM_AUTH | DRM_RENDER_ALLOW), >> + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_RESOURCES, exynos_drm_ipp_get_res_ioctl, >> + DRM_AUTH | DRM_RENDER_ALLOW), >> + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_CAPS, exynos_drm_ipp_get_caps_ioctl, >> + DRM_AUTH | DRM_RENDER_ALLOW), >> + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_GET_LIMITS, exynos_drm_ipp_get_limits_ioctl, >> + DRM_AUTH | DRM_RENDER_ALLOW), >> + DRM_IOCTL_DEF_DRV(EXYNOS_IPP_COMMIT, exynos_drm_ipp_commit_ioctl, >> + DRM_AUTH | DRM_RENDER_ALLOW), >> }; >> >> static const struct file_operations exynos_drm_driver_fops = { >> diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.c b/drivers/gpu/drm/exynos/exynos_drm_ipp.c >> new file mode 100644 >> index 000000000000..d54d23c6f2b0 >> --- /dev/null >> +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.c >> @@ -0,0 +1,911 @@ >> +/* >> + * Copyright (C) 2017 Samsung Electronics Co.Ltd >> + * Authors: >> + * Marek Szyprowski <m.szyprowski@samsung.com> >> + * >> + * Exynos DRM Image Post Processing (IPP) related functions >> + * >> + * Permission is hereby granted, free of charge, to any person obtaining a >> + * copy of this software and associated documentation files (the "Software"), >> + * to deal in the Software without restriction, including without limitation >> + * the rights to use, copy, modify, merge, publish, distribute, sublicense, >> + * and/or sell copies of the Software, and to permit persons to whom the >> + * Software is furnished to do so, subject to the following conditions: >> + * >> + * The above copyright notice and this permission notice shall be included in >> + * all copies or substantial portions of the Software. >> + */ >> + >> + >> +#include <drm/drmP.h> >> +#include <drm/drm_mode.h> >> +#include <uapi/drm/exynos_drm.h> >> + >> +#include "exynos_drm_drv.h" >> +#include "exynos_drm_gem.h" >> +#include "exynos_drm_ipp.h" >> + >> +static int num_ipp; >> +static LIST_HEAD(ipp_list); >> + >> +struct drm_pending_exynos_ipp_event { >> + struct drm_pending_event base; >> + struct drm_exynos_ipp_event event; >> +}; >> + >> +/** >> + * exynos_drm_ipp_register - Register a new picture processor hardware module >> + * @dev: DRM device >> + * @ipp: ipp module to init >> + * @funcs: callbacks for the new ipp object >> + * @caps: bitmask of ipp capabilities (%DRM_EXYNOS_IPP_CAP_*) >> + * @formats: array of supported formats (%DRM_FORMAT_*) >> + * @name: name (for debugging purposes) >> + * >> + * Initializes a ipp module. >> + * >> + * Returns: >> + * Zero on success, error code on failure. >> + */ >> +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp, >> + const struct exynos_drm_ipp_funcs *funcs, unsigned int caps, >> + const struct exynos_drm_ipp_formats *formats, const char *name) >> +{ >> + WARN_ON(!ipp); >> + WARN_ON(!funcs); >> + WARN_ON(!formats); >> + >> + spin_lock_init(&ipp->lock); >> + INIT_LIST_HEAD(&ipp->todo_list); >> + init_waitqueue_head(&ipp->done_wq); >> + ipp->dev = dev; >> + ipp->funcs = funcs; >> + ipp->capabilities = caps; >> + ipp->name = name; >> + ipp->num_formats = 0; >> + ipp->formats = formats; >> + while (formats++->fourcc) >> + ipp->num_formats++; >> + >> + list_add_tail(&ipp->head, &ipp_list); > mutex lock/unlock would be required as component framework did. Frankly, add IPP drivers are now registered using component framework, which serializes the both registration and unregistration calls. The other place where this list is used is userspace ioctls, which in turn can be called only when DRM device is registered. This means that there is no need for any additional mutexes protecting the ipp_list. I will add a comment explaining it, as this is not obvious on the first sight. >> + ipp->id = num_ipp++; >> + >> + DRM_DEBUG_DRIVER("Registered ipp %d\n", ipp->id); >> + >> + return 0; >> +} >> + >> +/** >> + * exynos_drm_ipp_unregister - Unregister the picture processor module >> + * @dev: DRM device >> + * @ipp: ipp module >> + */ >> +void exynos_drm_ipp_unregister(struct drm_device *dev, >> + struct exynos_drm_ipp *ipp) >> +{ >> + BUG_ON(ipp->task); >> + BUG_ON(!list_empty(&ipp->todo_list)); >> + list_del(&ipp->head); > ditto. > >> +} >> + >> +/** >> + * exynos_drm_ipp_ioctl_get_res_ioctl - enumerate all ipp modules >> + * @dev: DRM device >> + * @data: ioctl data >> + * @file_priv: DRM file info >> + * >> + * Construct a list of ipp ids. >> + * >> + * Called by the user via ioctl. >> + * >> + * Returns: >> + * Zero on success, negative errno on failure. >> + */ >> +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data, >> + struct drm_file *file_priv) >> +{ >> + struct drm_exynos_ioctl_ipp_get_res *resp = data; >> + struct exynos_drm_ipp *ipp; >> + uint32_t __user *ipp_ptr = (uint32_t __user *) >> + (unsigned long)resp->ipp_id_ptr; >> + unsigned int count = num_ipp, copied = 0; >> + >> + /* >> + * This ioctl is called twice, once to determine how much space is >> + * needed, and the 2nd time to fill it. > This is a same way as drm_mode_getresources function did. I'm not sure whether calling this function twice is best or not but no problem. > >> + */ >> + if (count && resp->count_ipps >= count) { >> + list_for_each_entry(ipp, &ipp_list, head) { >> + if (put_user(ipp->id, ipp_ptr + copied)) >> + return -EFAULT; >> + copied++; >> + } >> + } >> + resp->count_ipps = count; >> + >> + return 0; >> +} >> + >> +static inline struct exynos_drm_ipp *__ipp_get(uint32_t id) >> +{ >> + struct exynos_drm_ipp *ipp; >> + >> + list_for_each_entry(ipp, &ipp_list, head) > mutex lock here. ipp could be deleted by exynos_drm_ipp_unregister function. > >> + if (ipp->id == id) >> + return ipp; >> + return NULL; >> +} >> + >> +/** >> + * exynos_drm_ipp_ioctl_get_caps - get ipp module capabilities and formats >> + * @dev: DRM device >> + * @data: ioctl data >> + * @file_priv: DRM file info >> + * >> + * Construct a structure describing ipp module capabilities. >> + * >> + * Called by the user via ioctl. >> + * >> + * Returns: >> + * Zero on success, negative errno on failure. >> + */ >> +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data, >> + struct drm_file *file_priv) >> +{ >> + struct drm_exynos_ioctl_ipp_get_caps *resp = data; >> + void __user *ptr = (void __user *)(unsigned long)resp->formats_ptr; >> + struct exynos_drm_ipp *ipp; >> + int i; >> + >> + ipp = __ipp_get(resp->ipp_id); >> + if (!ipp) >> + return -ENOENT; >> + >> + resp->ipp_id = ipp->id; >> + resp->capabilities = ipp->capabilities; >> + >> + /* >> + * This ioctl is called twice, once to determine how much space is >> + * needed, and the 2nd time to fill it. >> + */ >> + if (resp->formats_count >= ipp->num_formats) { >> + for (i = 0; i < ipp->num_formats; i++) { >> + struct drm_exynos_ipp_format tmp = { >> + .fourcc = ipp->formats[i].fourcc, >> + .type = ipp->formats[i].type, >> + .modifier = ipp->formats[i].modifier, >> + }; >> + >> + if (copy_to_user(ptr, &tmp, sizeof(tmp))) >> + return -EFAULT; >> + ptr += sizeof(tmp); >> + } >> + } >> + resp->formats_count = ipp->num_formats; >> + >> + return 0; >> +} >> + >> +static inline const struct exynos_drm_ipp_formats *__ipp_format_get( >> + uint32_t fourcc, uint64_t mod, unsigned long type, > Should argument, 'type', be 'unsigned long' data type? not 'unsigned int'? Yes, it looks I missed this during refactoring. >> + const struct exynos_drm_ipp_formats *f) >> +{ >> + for (; f->fourcc; f++) >> + if ((f->type & type) && f->fourcc == fourcc && > Here 'f->type' has 'unsigned int' data type. > >> + f->modifier == mod) >> + return f; >> + return NULL; >> +} >> + >> +/** >> + * exynos_drm_ipp_get_limits_ioctl - get ipp module limits >> + * @dev: DRM device >> + * @data: ioctl data >> + * @file_priv: DRM file info >> + * >> + * Construct a structure describing ipp module limitations for provided >> + * picture format. >> + * >> + * Called by the user via ioctl. >> + * >> + * Returns: >> + * Zero on success, negative errno on failure. >> + */ >> +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data, >> + struct drm_file *file_priv) >> +{ >> + struct drm_exynos_ioctl_ipp_get_limits *resp = data; >> + void __user *ptr = (void __user *)(unsigned long)resp->limits_ptr; >> + const struct exynos_drm_ipp_formats *format; >> + const struct drm_exynos_ipp_limit *limits, *l; >> + struct exynos_drm_ipp *ipp; >> + int count = 0; >> + >> + if (resp->type != DRM_EXYNOS_IPP_FORMAT_SOURCE && >> + resp->type != DRM_EXYNOS_IPP_FORMAT_DESTINATION) >> + return -EINVAL; >> + >> + ipp = __ipp_get(resp->ipp_id); >> + if (!ipp) >> + return -ENOENT; >> + >> + format = __ipp_format_get(resp->fourcc, resp->modifier, resp->type, >> + ipp->formats); >> + if (!format) >> + return -EINVAL; >> + >> + limits = format->limits; >> + if (limits) >> + for (l = limits; l->type; l++) >> + count++; > It seems no need to use loop to count. How about this? > at device driver, > static struct gsc_driverdata gsc_exynosxxxx_drvdata = { > .clk_name... > .. > .limits = gsc_xxxx_limits, > .limit_cnt = ARRAY_SIZE(gsc_xxx_limits), > }; If you prefer to have explicit number of limits provided by the driver, I can change the code. >> + >> + /* >> + * This ioctl is called twice, once to determine how much space is >> + * needed, and the 2nd time to fill it. >> + */ >> + if (count && resp->limits_count >= count) { >> + for (l = limits; l->type; l++, ptr += sizeof(*l)) >> + if (copy_to_user((void __user *)ptr, l, sizeof(*l))) >> + return -EFAULT; >> + } >> + resp->limits_count = count; >> + >> + return 0; >> +} >> + >> +static inline struct exynos_drm_ipp_task * >> + exynos_drm_ipp_task_alloc(struct exynos_drm_ipp *ipp) >> +{ >> + struct exynos_drm_ipp_task *task; >> + >> + task = kzalloc(sizeof(*task), GFP_KERNEL); >> + if (!task) >> + return NULL; >> + >> + task->dev = ipp->dev; >> + task->ipp = ipp; >> + >> + /* some defaults */ >> + task->src.rect.w = task->dst.rect.w = UINT_MAX; >> + task->src.rect.h = task->dst.rect.h = UINT_MAX; >> + task->transform.rotation = DRM_MODE_ROTATE_0; >> + >> + DRM_DEBUG_DRIVER("Allocated task %pK\n", task); >> + >> + return task; >> +} >> + >> +struct exynos_drm_param_map { >> + unsigned int id; >> + unsigned int size; >> + unsigned int offset; >> +} exynos_drm_ipp_params_maps[] = { >> + { >> + DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE, >> + sizeof(struct drm_exynos_ipp_task_buffer), >> + offsetof(struct exynos_drm_ipp_task, src.buf), >> + }, { >> + DRM_EXYNOS_IPP_TASK_BUFFER | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION, >> + sizeof(struct drm_exynos_ipp_task_buffer), >> + offsetof(struct exynos_drm_ipp_task, dst.buf), >> + }, { >> + DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_SOURCE, >> + sizeof(struct drm_exynos_ipp_task_rect), >> + offsetof(struct exynos_drm_ipp_task, src.rect), >> + }, { >> + DRM_EXYNOS_IPP_TASK_RECTANGLE | DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION, >> + sizeof(struct drm_exynos_ipp_task_rect), >> + offsetof(struct exynos_drm_ipp_task, dst.rect), >> + }, { >> + DRM_EXYNOS_IPP_TASK_TRANSFORM, >> + sizeof(struct drm_exynos_ipp_task_transform), >> + offsetof(struct exynos_drm_ipp_task, transform), >> + }, { >> + DRM_EXYNOS_IPP_TASK_ALPHA, >> + sizeof(struct drm_exynos_ipp_task_alpha), >> + offsetof(struct exynos_drm_ipp_task, alpha), >> + }, >> +}; >> + >> +static int exynos_drm_ipp_task_set(struct exynos_drm_ipp_task *task, >> + struct drm_exynos_ioctl_ipp_commit *arg) >> +{ >> + unsigned int size = arg->params_size; >> + void *p, *params; >> + int ret = 0; >> + >> + if (size > PAGE_SIZE) >> + return -ERANGE; > I think you would need to check whether commands - which are passed by user space and arg->params_ptr points to - are valid or not here. > > With this verification, checking if size > PAGE_SIZE wouldn't be needed because valid commands include only below data structures, > struct drm_exynos_ipp_task_buffer, struct drm_exynos_ipp_task_rect, struct drm_exynos_ipp_task_transform, and struct drm_exynos_ipp_task_alpha. > > This would be done by exynos_drm_ipp_task_check function which is called at exynos_drm_ipp_commit_ioctl function after exynos_drm_ipp_task_set funcion call. > However, if task checking could be done prior to exynos_drm_ipp_set function and the commands passed by user-space have wrong data then you could skip unnecessary operation - copying commands from user-space to kernel. > >> + >> + p = kmalloc(size, GFP_KERNEL); >> + if (!p) >> + return -ENOMEM; >> + >> + if (copy_from_user(p, (void __user *)(unsigned long)arg->params_ptr, >> + size)) { >> + ret = -EFAULT; >> + DRM_DEBUG_DRIVER("Failed to copy configuration from userspace\n"); >> + goto free; >> + } >> + >> + params = p; >> + while (size) { >> + struct exynos_drm_param_map *map = exynos_drm_ipp_params_maps; >> + u32 *id = params; >> + int i; >> + >> + for (i = 0; i < ARRAY_SIZE(exynos_drm_ipp_params_maps); i++) >> + if (map[i].id == *id) >> + break; >> + >> + if (i < ARRAY_SIZE(exynos_drm_ipp_params_maps) && >> + map[i].size <= size) { >> + memcpy((void *)task + map[i].offset, params, >> + map[i].size); >> + params += map[i].size; >> + size -= map[i].size; >> + } else { >> + ret = -EINVAL; >> + goto free; >> + } >> + } >> + >> + DRM_DEBUG_DRIVER("Got task %pK configuration from userspace\n", task); >> +free: >> + kfree(p); >> + return ret; >> +} >> + >> +static int exynos_drm_ipp_task_setup_buffer(struct exynos_drm_ipp_buffer *buf, >> + struct drm_file *filp) >> +{ >> + int ret = 0; >> + int i; >> + >> + /* basic checks */ >> + if (buf->buf.width == 0 || buf->buf.height == 0) >> + return -EINVAL; >> + buf->format = drm_format_info(buf->buf.fourcc); >> + for (i = 0; i < buf->format->num_planes; i++) { >> + unsigned int width = (i == 0) ? buf->buf.width : >> + DIV_ROUND_UP(buf->buf.width, buf->format->hsub); >> + >> + if (buf->buf.pitch[i] == 0) >> + buf->buf.pitch[i] = width * buf->format->cpp[i]; >> + if (buf->buf.pitch[i] < width * buf->format->cpp[i]) >> + return -EINVAL; >> + if (!buf->buf.gem_id[i]) >> + return -ENOENT; >> + } >> + >> + /* get GEM buffers and check their size */ >> + for (i = 0; i < buf->format->num_planes; i++) { >> + unsigned int height = (i == 0) ? buf->buf.height : >> + DIV_ROUND_UP(buf->buf.height, buf->format->vsub); >> + unsigned long size = height * buf->buf.pitch[i] + >> + buf->buf.offset[i]; >> + struct drm_gem_object *obj = drm_gem_object_lookup(filp, >> + buf->buf.gem_id[i]); >> + if (!obj) { >> + ret = -ENOENT; >> + goto gem_free; >> + } >> + buf->exynos_gem[i] = to_exynos_gem(obj); >> + >> + if (size > buf->exynos_gem[i]->size) { >> + i++; >> + ret = -EINVAL; >> + goto gem_free; >> + } >> + buf->dma_addr[i] = buf->exynos_gem[i]->dma_addr + >> + buf->buf.offset[i]; >> + } >> + >> + return 0; >> +gem_free: >> + while (i--) { >> + drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base); >> + buf->exynos_gem[i] = NULL; >> + } >> + return ret; >> +} >> + >> +static void exynos_drm_ipp_task_release_buffer(struct exynos_drm_ipp_buffer *buf) >> +{ >> + int i; >> + >> + if (!buf->exynos_gem[0]) >> + return; >> + for (i = 0; i < buf->format->num_planes; i++) >> + drm_gem_object_put_unlocked(&buf->exynos_gem[i]->base); >> +} >> + >> +static void exynos_drm_ipp_task_free(struct exynos_drm_ipp *ipp, >> + struct exynos_drm_ipp_task *task) >> +{ >> + DRM_DEBUG_DRIVER("Freeing task %pK\n", task); >> + >> + exynos_drm_ipp_task_release_buffer(&task->src); >> + exynos_drm_ipp_task_release_buffer(&task->dst); >> + if (task->event) >> + drm_event_cancel_free(ipp->dev, &task->event->base); >> + kfree(task); >> +} >> + >> +struct drm_ipp_limit { >> + struct drm_exynos_ipp_limit_val h; >> + struct drm_exynos_ipp_limit_val v; >> +}; >> + >> +enum drm_ipp_size_id { >> + IPP_LIMIT_BUFFER, IPP_LIMIT_AREA, IPP_LIMIT_ROTATED, IPP_LIMIT_MAX >> +}; >> + >> +static const enum drm_ipp_size_id limit_id_fallback[IPP_LIMIT_MAX][4] = { >> + [IPP_LIMIT_BUFFER] = { DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER }, >> + [IPP_LIMIT_AREA] = { DRM_EXYNOS_IPP_LIMIT_SIZE_AREA, >> + DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER }, >> + [IPP_LIMIT_ROTATED] = { DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED, >> + DRM_EXYNOS_IPP_LIMIT_SIZE_AREA, >> + DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER }, >> +}; >> + >> +static inline void __limit_set_val(unsigned int *ptr, unsigned int val) >> +{ >> + if (!*ptr) >> + *ptr = val; >> +} >> + >> +static void __get_size_limit(const struct drm_exynos_ipp_limit *limits, >> + enum drm_ipp_size_id id, struct drm_ipp_limit *res) >> +{ >> + const struct drm_exynos_ipp_limit *l = limits; >> + int i = 0; >> + >> + memset(res, 0, sizeof(*res)); >> + for (i = 0; limit_id_fallback[id][i]; i++) >> + for (l = limits; l->type; l++) { >> + if (((l->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) != >> + DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE) || >> + ((l->type & DRM_EXYNOS_IPP_LIMIT_SIZE_MASK) != >> + limit_id_fallback[id][i])) >> + continue; >> + __limit_set_val(&res->h.min, l->h.min); >> + __limit_set_val(&res->h.max, l->h.max); >> + __limit_set_val(&res->h.align, l->h.align); >> + __limit_set_val(&res->v.min, l->v.min); >> + __limit_set_val(&res->v.max, l->v.max); >> + __limit_set_val(&res->v.align, l->v.align); >> + } >> +} >> + >> +static inline bool __align_check(unsigned int val, unsigned int align) >> +{ >> + if (align && (val & (align - 1))) { >> + DRM_DEBUG_DRIVER("Value %d exceeds hw limits (align %d)\n", >> + val, align); >> + return false; >> + } >> + return true; >> +} >> + >> +static inline bool __size_limit_check(unsigned int val, >> + struct drm_exynos_ipp_limit_val *l) >> +{ >> + if ((l->min && val < l->min) || (l->max && val > l->max)) { >> + DRM_DEBUG_DRIVER("Value %d exceeds hw limits (min %d, max %d)\n", >> + val, l->min, l->max); >> + return false; >> + } >> + return __align_check(val, l->align); >> +} >> + >> +static int exynos_drm_ipp_check_size_limits(struct exynos_drm_ipp_buffer *buf, >> + const struct drm_exynos_ipp_limit *limits, bool rotate, bool swap) >> +{ >> + enum drm_ipp_size_id id = rotate ? IPP_LIMIT_ROTATED : IPP_LIMIT_AREA; >> + struct drm_ipp_limit l; >> + struct drm_exynos_ipp_limit_val *lh = &l.h, *lv = &l.v; >> + >> + if (!limits) >> + return 0; >> + >> + __get_size_limit(limits, IPP_LIMIT_BUFFER, &l); >> + if (!__size_limit_check(buf->buf.width, &l.h) || >> + !__size_limit_check(buf->buf.height, &l.v)) >> + return -EINVAL; > Printing out warning message here would be better for developers as a hint. __size_limit_check() already prints a warning with exact values, imho there is no need to duplicate it here. > >> + >> + if (swap) { >> + lv = &l.h; >> + lh = &l.v; >> + } >> + __get_size_limit(limits, id, &l); >> + if (!__size_limit_check(buf->rect.w, lh) || >> + !__align_check(buf->rect.x, lh->align) || >> + !__size_limit_check(buf->rect.h, lv) || >> + !__align_check(buf->rect.y, lv->align)) >> + return -EINVAL; > ditto. > >> + >> + return 0; >> +} >> + >> +static inline bool __scale_limit_check(unsigned int src, unsigned int dst, >> + unsigned int min, unsigned int max) >> +{ >> + if ((max && (dst << 16) > src * max) || >> + (min && (dst << 16) < src * min)) { >> + DRM_DEBUG_DRIVER("Scale from %d to %d exceeds hw limits (ratio min %d.%05d, max %d.%05d)\n", >> + src, dst, >> + min >> 16, 100000 * (min & 0xffff) / (1 << 16), >> + max >> 16, 100000 * (max & 0xffff) / (1 << 16)); >> + return false; >> + } >> + return true; >> +} >> + >> +static int exynos_drm_ipp_check_scale_limits(struct drm_exynos_ipp_task_rect *src, >> + struct drm_exynos_ipp_task_rect *dst, >> + const struct drm_exynos_ipp_limit *limits, bool swap) >> +{ >> + const struct drm_exynos_ipp_limit_val *lh, *lv; >> + >> + if (!limits) >> + return 0; >> + >> + for (; limits->type; limits++) >> + if ((limits->type & DRM_EXYNOS_IPP_LIMIT_TYPE_MASK) == >> + DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE) >> + break; >> + if (!limits->type) >> + return 0; >> + >> + lh = (!swap) ? &limits->h : &limits->v; >> + lv = (!swap) ? &limits->v : &limits->h; >> + >> + if (!__scale_limit_check(src->w, dst->w, lh->min, lh->max) || >> + !__scale_limit_check(src->h, dst->h, lv->min, lv->max)) >> + return -EINVAL; > ditto, hint as a warning. __scale_limit_check() also print warning. > >> + >> + return 0; >> +} >> + >> +static int exynos_drm_ipp_task_check(struct exynos_drm_ipp_task *task, >> + struct drm_file *filp) >> +{ >> + struct exynos_drm_ipp *ipp = task->ipp; >> + const struct exynos_drm_ipp_formats *src_format, *dst_format; >> + struct exynos_drm_ipp_buffer *src = &task->src, *dst = &task->dst; >> + unsigned int rotation = task->transform.rotation; >> + int ret = 0; >> + bool scale = false, swap = false; >> + >> + DRM_DEBUG_DRIVER("Checking task %pK\n", task); >> + >> + if (src->rect.w == UINT_MAX) >> + src->rect.w = src->buf.width; >> + if (src->rect.h == UINT_MAX) >> + src->rect.h = src->buf.height; >> + if (dst->rect.w == UINT_MAX) >> + dst->rect.w = dst->buf.width; >> + if (dst->rect.h == UINT_MAX) >> + dst->rect.h = dst->buf.height; >> + >> + if (src->rect.x + src->rect.w > (src->buf.width) || >> + src->rect.y + src->rect.h > (src->buf.height) || >> + dst->rect.x + dst->rect.w > (dst->buf.width) || >> + dst->rect.y + dst->rect.h > (dst->buf.height)) { >> + DRM_DEBUG_DRIVER("Task %pK: defined area is outside provided buffers\n", >> + task); >> + return -EINVAL; >> + } >> + >> + if (drm_rotation_90_or_270(rotation)) >> + swap = true; >> + >> + if ((!swap && (src->rect.w != dst->rect.w || src->rect.h != dst->rect.h)) || >> + (swap && (src->rect.w != dst->rect.h || src->rect.h != dst->rect.w))) >> + scale = true; >> + >> + if ((!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CROP) && >> + (src->rect.x || src->rect.y || dst->rect.x || dst->rect.y)) || >> + (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_ROTATE) && >> + rotation != DRM_MODE_ROTATE_0) || >> + (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_SCALE) && scale) || >> + (!(ipp->capabilities & DRM_EXYNOS_IPP_CAP_CONVERT) && >> + src->buf.fourcc != dst->buf.fourcc)) { >> + DRM_DEBUG_DRIVER("Task %pK: hw capabilities exceeded\n", task); >> + return -EINVAL; >> + } >> + >> + src_format = __ipp_format_get(src->buf.fourcc, src->buf.modifier, >> + DRM_EXYNOS_IPP_FORMAT_SOURCE, ipp->formats); >> + if (!src_format) { >> + DRM_DEBUG_DRIVER("Task %pK: src format not supported\n", task); >> + return -EINVAL; >> + } >> + ret = exynos_drm_ipp_check_size_limits(src, src_format->limits, >> + rotation != DRM_MODE_ROTATE_0, false); >> + if (ret) >> + return ret; >> + ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect, >> + src_format->limits, swap); >> + if (ret) >> + return ret; >> + >> + dst_format = __ipp_format_get(dst->buf.fourcc, dst->buf.modifier, >> + DRM_EXYNOS_IPP_FORMAT_DESTINATION, ipp->formats); >> + if (!dst_format) { >> + DRM_DEBUG_DRIVER("Task %pK: dst format not supported\n", task); >> + return -EINVAL; >> + } >> + ret = exynos_drm_ipp_check_size_limits(dst, dst_format->limits, >> + rotation != DRM_MODE_ROTATE_0, swap); >> + if (ret) >> + return ret; >> + ret = exynos_drm_ipp_check_scale_limits(&src->rect, &dst->rect, >> + dst_format->limits, swap); >> + if (ret) >> + return ret; >> + >> + ret = exynos_drm_ipp_task_setup_buffer(src, filp); >> + if (ret) { >> + DRM_DEBUG_DRIVER("Task %pK: src buffer setup failed\n", task); >> + return ret; >> + } >> + ret = exynos_drm_ipp_task_setup_buffer(dst, filp); >> + if (ret) { >> + DRM_DEBUG_DRIVER("Task %pK: dst buffer setup failed\n", task); >> + return ret; >> + } >> + >> + if (ipp->funcs->check) { >> + ret = ipp->funcs->check(ipp, task); > It would be better to check HW or driver specific things prior to buffer setup. Frankly, I would probably remove driver specific check for now. The code in Exynos Scaler's driver should be moved to the core, as it is valid for all Exynos drivers. >> + if (ret) { >> + DRM_DEBUG_DRIVER("Task %pK: driver specific check failed\n", >> + task); >> + return ret; >> + } >> + } >> + >> + DRM_DEBUG_DRIVER("Task %pK: all checks done.\n", task); >> + >> + return ret; >> +} >> + >> +static int exynos_drm_ipp_event_create(struct exynos_drm_ipp_task *task, >> + struct drm_file *file_priv, uint64_t user_data) >> +{ >> + struct drm_pending_exynos_ipp_event *e = NULL; >> + int ret; >> + >> + e = kzalloc(sizeof(*e), GFP_KERNEL); >> + if (!e) >> + return -ENOMEM; >> + >> + e->event.base.type = DRM_EXYNOS_IPP_EVENT; >> + e->event.base.length = sizeof(e->event); >> + e->event.user_data = user_data; >> + >> + ret = drm_event_reserve_init(task->dev, file_priv, &e->base, >> + &e->event.base); >> + if (ret) >> + goto free; >> + >> + task->event = e; >> + return 0; >> +free: >> + kfree(e); >> + return ret; >> +} >> + >> +static void exynos_drm_ipp_event_send(struct exynos_drm_ipp_task *task) >> +{ >> + struct timeval now = ktime_to_timeval(ktime_get()); >> + >> + task->event->event.tv_sec = now.tv_sec; >> + task->event->event.tv_usec = now.tv_usec; >> + task->event->event.sequence = atomic_inc_return(&task->ipp->sequence); >> + >> + drm_send_event(task->dev, &task->event->base); >> +} >> + >> +static int exynos_drm_ipp_task_cleanup(struct exynos_drm_ipp_task *task) >> +{ >> + int ret = task->ret; >> + >> + if (ret == 0 && task->event) { >> + exynos_drm_ipp_event_send(task); >> + /* ensure event won't be canceled on task free */ >> + task->event = NULL; >> + } >> + >> + exynos_drm_ipp_task_free(task->ipp, task); >> + return ret; >> +} >> + >> +static void exynos_drm_ipp_cleanup_work(struct work_struct *work) >> +{ >> + struct exynos_drm_ipp_task *task = container_of(work, >> + struct exynos_drm_ipp_task, cleanup_work); >> + >> + exynos_drm_ipp_task_cleanup(task); >> +} >> + >> +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp); > It's not good. How about just moving implementaion of this function above side? exynos_drm_ipp_next_task() calls exynos_drm_ipp_task_done(), which might call exynos_drm_ipp_next_task() again. A forward declaration is required in such cycle. > > Thanks, > Inki Dae > >> + >> +/** >> + * exynos_drm_ipp_task_done - finish given task and set return code >> + * @task: ipp task to finish >> + * @ret: error code or 0 if operation has been performed successfully >> + */ >> +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret) >> +{ >> + struct exynos_drm_ipp *ipp = task->ipp; >> + unsigned long flags; >> + >> + DRM_DEBUG_DRIVER("ipp: %d, task %pK done\n", ipp->id, task); >> + >> + spin_lock_irqsave(&ipp->lock, flags); >> + if (ipp->task == task) >> + ipp->task = NULL; >> + task->flags |= DRM_EXYNOS_IPP_TASK_DONE; >> + task->ret = ret; >> + spin_unlock_irqrestore(&ipp->lock, flags); >> + >> + exynos_drm_ipp_next_task(ipp); >> + wake_up(&ipp->done_wq); >> + >> + if (task->flags & DRM_EXYNOS_IPP_TASK_ASYNC) { >> + INIT_WORK(&task->cleanup_work, exynos_drm_ipp_cleanup_work); >> + schedule_work(&task->cleanup_work); >> + } >> +} >> + >> +static void exynos_drm_ipp_next_task(struct exynos_drm_ipp *ipp) >> +{ >> + struct exynos_drm_ipp_task *task; >> + unsigned long flags; >> + int ret; >> + >> + DRM_DEBUG_DRIVER("ipp: %d, try to run new task\n", ipp->id); >> + >> + spin_lock_irqsave(&ipp->lock, flags); >> + >> + if (ipp->task || list_empty(&ipp->todo_list)) { >> + spin_unlock_irqrestore(&ipp->lock, flags); >> + return; >> + } >> + >> + task = list_first_entry(&ipp->todo_list, struct exynos_drm_ipp_task, >> + head); >> + list_del_init(&task->head); >> + ipp->task = task; >> + >> + spin_unlock_irqrestore(&ipp->lock, flags); >> + >> + DRM_DEBUG_DRIVER("ipp: %d, selected task %pK to run\n", ipp->id, task); >> + >> + ret = ipp->funcs->commit(ipp, task); >> + if (ret) >> + exynos_drm_ipp_task_done(task, ret); >> +} >> + >> +static void exynos_drm_ipp_schedule_task(struct exynos_drm_ipp *ipp, >> + struct exynos_drm_ipp_task *task) >> +{ >> + unsigned long flags; >> + >> + spin_lock_irqsave(&ipp->lock, flags); >> + list_add(&task->head, &ipp->todo_list); >> + spin_unlock_irqrestore(&ipp->lock, flags); >> + >> + exynos_drm_ipp_next_task(ipp); >> +} >> + >> +static void exynos_drm_ipp_task_abort(struct exynos_drm_ipp *ipp, >> + struct exynos_drm_ipp_task *task) >> +{ >> + unsigned long flags; >> + >> + spin_lock_irqsave(&ipp->lock, flags); >> + if (task->flags & DRM_EXYNOS_IPP_TASK_DONE) { >> + /* already completed task */ >> + exynos_drm_ipp_task_cleanup(task); >> + } else if (ipp->task != task) { >> + /* task has not been scheduled for execution yet */ >> + list_del_init(&task->head); >> + exynos_drm_ipp_task_cleanup(task); >> + } else { >> + /* >> + * currently processed task, call abort() and perform >> + * cleanup with async worker >> + */ >> + task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC; >> + spin_unlock_irqrestore(&ipp->lock, flags); >> + if (ipp->funcs->abort) >> + ipp->funcs->abort(ipp, task); >> + return; >> + } >> + spin_unlock_irqrestore(&ipp->lock, flags); >> +} >> + >> +/** >> + * exynos_drm_ipp_commit_ioctl - perform image processing operation >> + * @dev: DRM device >> + * @data: ioctl data >> + * @file_priv: DRM file info >> + * >> + * Construct a ipp task from the set of properties provided from the user >> + * and try to schedule it to framebuffer processor hardware. >> + * >> + * Called by the user via ioctl. >> + * >> + * Returns: >> + * Zero on success, negative errno on failure. >> + */ >> +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, void *data, >> + struct drm_file *file_priv) >> +{ >> + struct drm_exynos_ioctl_ipp_commit *arg = data; >> + struct exynos_drm_ipp *ipp; >> + struct exynos_drm_ipp_task *task; >> + int ret = 0; >> + >> + if ((arg->flags & ~DRM_EXYNOS_IPP_FLAGS) || arg->reserved) >> + return -EINVAL; >> + >> + /* can't test and expect an event at the same time */ >> + if ((arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY) && >> + (arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT)) >> + return -EINVAL; >> + >> + ipp = __ipp_get(arg->ipp_id); >> + if (!ipp) >> + return -ENOENT; >> + >> + task = exynos_drm_ipp_task_alloc(ipp); >> + if (!task) >> + return -ENOMEM; >> + >> + ret = exynos_drm_ipp_task_set(task, arg); >> + if (ret) >> + goto free; >> + >> + ret = exynos_drm_ipp_task_check(task, file_priv); >> + if (ret || arg->flags & DRM_EXYNOS_IPP_FLAG_TEST_ONLY) >> + goto free; >> + >> + if (arg->flags & DRM_EXYNOS_IPP_FLAG_EVENT) { >> + ret = exynos_drm_ipp_event_create(task, file_priv, >> + arg->user_data); >> + if (ret) >> + goto free; >> + } >> + >> + /* >> + * Queue task for processing on the hardware. task object will be >> + * then freed after exynos_drm_ipp_task_done() >> + */ >> + if (arg->flags & DRM_EXYNOS_IPP_FLAG_NONBLOCK) { >> + DRM_DEBUG_DRIVER("ipp: %d, nonblocking processing task %pK\n", >> + ipp->id, task); >> + >> + task->flags |= DRM_EXYNOS_IPP_TASK_ASYNC; >> + exynos_drm_ipp_schedule_task(task->ipp, task); >> + ret = 0; >> + } else { >> + DRM_DEBUG_DRIVER("ipp: %d, processing task %pK\n", ipp->id, task); >> + exynos_drm_ipp_schedule_task(ipp, task); >> + ret = wait_event_interruptible(ipp->done_wq, >> + task->flags & DRM_EXYNOS_IPP_TASK_DONE); >> + if (ret) >> + exynos_drm_ipp_task_abort(ipp, task); >> + else >> + ret = exynos_drm_ipp_task_cleanup(task); >> + } >> + return ret; >> +free: >> + exynos_drm_ipp_task_free(ipp, task); >> + >> + return ret; >> +} >> diff --git a/drivers/gpu/drm/exynos/exynos_drm_ipp.h b/drivers/gpu/drm/exynos/exynos_drm_ipp.h >> new file mode 100644 >> index 000000000000..199eb2006410 >> --- /dev/null >> +++ b/drivers/gpu/drm/exynos/exynos_drm_ipp.h >> @@ -0,0 +1,195 @@ >> +/* >> + * Copyright (c) 2017 Samsung Electronics Co., Ltd. >> + * >> + * This program is free software; you can redistribute it and/or modify it >> + * under the terms of the GNU General Public License as published by the >> + * Free Software Foundation; either version 2 of the License, or (at your >> + * option) any later version. >> + */ >> + >> +#ifndef _EXYNOS_DRM_IPP_H_ >> +#define _EXYNOS_DRM_IPP_H_ >> + >> +#include <drm/drmP.h> >> + >> +struct exynos_drm_ipp; >> +struct exynos_drm_ipp_task; >> + >> +/** >> + * struct exynos_drm_ipp_funcs - exynos_drm_ipp control functions >> + */ >> +struct exynos_drm_ipp_funcs { >> + /** >> + * @check: >> + * >> + * This is the optional hook to validate an ipp task. This function >> + * must reject any task which the hardware or driver doesn't support. >> + * This includes but is of course not limited to: >> + * >> + * - Checking that the buffers, scaling and placement requirements >> + * and so on are within the limits of the hardware not specified by >> + * limits array. >> + * >> + * - The driver does not need to repeat basic input validation like >> + * done in the exynos_drm_ipp_task_check() function. The core does >> + * that before calling this hook. >> + * >> + * RETURNS: >> + * >> + * 0 on success or one of the below negative error codes: >> + * >> + * - -EINVAL, if any of the above constraints are violated. >> + */ >> + int (*check)(struct exynos_drm_ipp *ipp, >> + struct exynos_drm_ipp_task *task); >> + >> + /** >> + * @commit: >> + * >> + * This is the main entry point to start framebuffer processing >> + * in the hardware. The exynos_drm_ipp_task has been already validated. >> + * This function must not wait until the device finishes processing. >> + * When the driver finishes processing, it has to call >> + * exynos_exynos_drm_ipp_task_done() function. >> + * >> + * RETURNS: >> + * >> + * 0 on success or negative error codes in case of failure. >> + */ >> + int (*commit)(struct exynos_drm_ipp *ipp, >> + struct exynos_drm_ipp_task *task); >> + >> + /** >> + * @abort: >> + * >> + * Informs the driver that it has to abort the currently running >> + * task as soon as possible (i.e. as soon as it can stop the device >> + * safely), even if the task would not have been finished by then. >> + * After the driver performs the necessary steps, it has to call >> + * exynos_drm_ipp_task_done() (as if the task ended normally). >> + * This function does not have to (and will usually not) wait >> + * until the device enters a state when it can be stopped. >> + */ >> + void (*abort)(struct exynos_drm_ipp *ipp, >> + struct exynos_drm_ipp_task *task); >> +}; >> + >> +/** >> + * struct exynos_drm_ipp - central picture processor module structure >> + */ >> +struct exynos_drm_ipp { >> + struct drm_device *dev; >> + struct list_head head; >> + unsigned int id; >> + >> + const char *name; >> + const struct exynos_drm_ipp_funcs *funcs; >> + unsigned int capabilities; >> + const struct exynos_drm_ipp_formats *formats; >> + unsigned int num_formats; >> + atomic_t sequence; >> + >> + spinlock_t lock; >> + struct exynos_drm_ipp_task *task; >> + struct list_head todo_list; >> + wait_queue_head_t done_wq; >> +}; >> + >> +struct exynos_drm_ipp_buffer { >> + struct drm_exynos_ipp_task_buffer buf; >> + struct drm_exynos_ipp_task_rect rect; >> + >> + struct exynos_drm_gem *exynos_gem[MAX_FB_BUFFER]; >> + const struct drm_format_info *format; >> + dma_addr_t dma_addr[MAX_FB_BUFFER]; >> +}; >> + >> +/** >> + * struct exynos_drm_ipp_task - a structure describing transformation that >> + * has to be performed by the picture processor hardware module >> + */ >> +struct exynos_drm_ipp_task { >> + struct drm_device *dev; >> + struct exynos_drm_ipp *ipp; >> + struct list_head head; >> + >> + struct exynos_drm_ipp_buffer src; >> + struct exynos_drm_ipp_buffer dst; >> + >> + struct drm_exynos_ipp_task_transform transform; >> + struct drm_exynos_ipp_task_alpha alpha; >> + >> + struct work_struct cleanup_work; >> + unsigned int flags; >> + int ret; >> + >> + struct drm_pending_exynos_ipp_event *event; >> +}; >> + >> +#define DRM_EXYNOS_IPP_TASK_DONE (1 << 0) >> +#define DRM_EXYNOS_IPP_TASK_ASYNC (1 << 1) >> + >> +struct exynos_drm_ipp_formats { >> + u32 fourcc; >> + u32 type; >> + u64 modifier; >> + const struct drm_exynos_ipp_limit *limits; >> +}; >> + >> +/* helper macros to set exynos_drm_ipp_formats structure and limits*/ >> +#define IPP_SRCDST_MFORMAT(f, m, l) \ >> + .fourcc = DRM_FORMAT_##f, .modifier = m, .limits = l, \ >> + .type = (DRM_EXYNOS_IPP_FORMAT_SOURCE | DRM_EXYNOS_IPP_FORMAT_DESTINATION) >> + >> +#define IPP_SRCDST_FORMAT(f, l) IPP_SRCDST_MFORMAT(f, 0, l) >> + >> +#define IPP_SIZE_LIMIT(l, val...) \ >> + .type = (DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE | DRM_EXYNOS_IPP_LIMIT_SIZE_##l), \ >> + val >> + >> +#define IPP_SCALE_LIMIT(val...) \ >> + .type = (DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE),\ >> + val >> + >> +int exynos_drm_ipp_register(struct drm_device *dev, struct exynos_drm_ipp *ipp, >> + const struct exynos_drm_ipp_funcs *funcs, unsigned int caps, >> + const struct exynos_drm_ipp_formats *formats, const char *name); >> +void exynos_drm_ipp_unregister(struct drm_device *dev, struct exynos_drm_ipp *ipp); >> + >> +void exynos_drm_ipp_task_done(struct exynos_drm_ipp_task *task, int ret); >> + >> +#ifdef CONFIG_DRM_EXYNOS_IPP >> +int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, void *data, >> + struct drm_file *file_priv); >> +int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, void *data, >> + struct drm_file *file_priv); >> +int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, void *data, >> + struct drm_file *file_priv); >> +int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, >> + void *data, struct drm_file *file_priv); >> +#else >> +static inline int exynos_drm_ipp_get_res_ioctl(struct drm_device *dev, >> + void *data, struct drm_file *file_priv) >> +{ >> + struct drm_exynos_ioctl_ipp_get_res *resp = data; >> + >> + resp->count_ipps = 0; >> + return 0; >> +} >> +static inline int exynos_drm_ipp_get_caps_ioctl(struct drm_device *dev, >> + void *data, struct drm_file *file_priv) >> +{ >> + return -ENODEV; >> +} >> +static inline int exynos_drm_ipp_get_limits_ioctl(struct drm_device *dev, >> + void *data, struct drm_file *file_priv) >> +{ >> + return -ENODEV; >> +} >> +static inline int exynos_drm_ipp_commit_ioctl(struct drm_device *dev, >> + void *data, struct drm_file *file_priv) >> +{ >> + return -ENODEV; >> +} >> +#endif >> +#endif >> diff --git a/include/uapi/drm/exynos_drm.h b/include/uapi/drm/exynos_drm.h >> index 5d9a08d69ba7..974cbb579f90 100644 >> --- a/include/uapi/drm/exynos_drm.h >> +++ b/include/uapi/drm/exynos_drm.h >> @@ -134,6 +134,219 @@ struct drm_exynos_g2d_exec { >> __u64 async; >> }; >> >> +/* Exynos DRM IPP v2 API */ >> + >> +/** >> + * Enumerate available IPP hardware modules. >> + * >> + * @count_ipps: size of ipp_id array / number of ipp modules (set by driver) >> + * @reserved: padding >> + * @ipp_id_ptr: pointer to ipp_id array or NULL >> + */ >> +struct drm_exynos_ioctl_ipp_get_res { >> + __u32 count_ipps; >> + __u32 reserved; >> + __u64 ipp_id_ptr; >> +}; >> + >> +enum drm_exynos_ipp_format_type { >> + DRM_EXYNOS_IPP_FORMAT_SOURCE = 0x01, >> + DRM_EXYNOS_IPP_FORMAT_DESTINATION = 0x02, >> +}; >> + >> +struct drm_exynos_ipp_format { >> + __u32 fourcc; >> + __u32 type; >> + __u64 modifier; >> +}; >> + >> +enum drm_exynos_ipp_capability { >> + DRM_EXYNOS_IPP_CAP_CROP = 0x01, >> + DRM_EXYNOS_IPP_CAP_ROTATE = 0x02, >> + DRM_EXYNOS_IPP_CAP_SCALE = 0x04, >> + DRM_EXYNOS_IPP_CAP_CONVERT = 0x08, >> +}; >> + >> +/** >> + * Get IPP hardware capabilities and supported image formats. >> + * >> + * @ipp_id: id of IPP module to query >> + * @capabilities: bitmask of drm_exynos_ipp_capability (set by driver) >> + * @reserved: padding >> + * @formats_count: size of formats array (in entries) / number of filled >> + * formats (set by driver) >> + * @formats_ptr: pointer to formats array or NULL >> + */ >> +struct drm_exynos_ioctl_ipp_get_caps { >> + __u32 ipp_id; >> + __u32 capabilities; >> + __u32 reserved; >> + __u32 formats_count; >> + __u64 formats_ptr; >> +}; >> + >> +enum drm_exynos_ipp_limit_type { >> + /* size (horizontal/vertial) limits, in pixels (min, max, alignment) */ >> + DRM_EXYNOS_IPP_LIMIT_TYPE_SIZE = 0x0001, >> + /* scale ratio (horizonta/vertial), 16.16 fixed point (min, max) */ >> + DRM_EXYNOS_IPP_LIMIT_TYPE_SCALE = 0x0002, >> + >> + /* image buffer area */ >> + DRM_EXYNOS_IPP_LIMIT_SIZE_BUFFER = 0x0001 << 16, >> + /* src/dst rectangle area */ >> + DRM_EXYNOS_IPP_LIMIT_SIZE_AREA = 0x0002 << 16, >> + /* src/dst rectangle area when rotation enabled */ >> + DRM_EXYNOS_IPP_LIMIT_SIZE_ROTATED = 0x0003 << 16, >> + >> + DRM_EXYNOS_IPP_LIMIT_TYPE_MASK = 0x000f, >> + DRM_EXYNOS_IPP_LIMIT_SIZE_MASK = 0x000f << 16, >> +}; >> + >> +struct drm_exynos_ipp_limit_val { >> + __u32 min; >> + __u32 max; >> + __u32 align; >> + __u32 reserved; >> +}; >> + >> +/** >> + * IPP module limitation. >> + * >> + * @type: limit type (see drm_exynos_ipp_limit_type enum) >> + * @reserved: padding >> + * @h: horizontal limits >> + * @v: vertical limits >> + */ >> +struct drm_exynos_ipp_limit { >> + __u32 type; >> + __u32 reserved; >> + struct drm_exynos_ipp_limit_val h; >> + struct drm_exynos_ipp_limit_val v; >> +}; >> + >> +/** >> + * Get IPP limits for given image format. >> + * >> + * @ipp_id: id of IPP module to query >> + * @fourcc: image format code (see DRM_FORMAT_* in drm_fourcc.h) >> + * @modifier: image format modifier (see DRM_FORMAT_MOD_* in drm_fourcc.h) >> + * @type: source/destination identifier (drm_exynos_ipp_format_flag enum) >> + * @limits_count: size of limits array (in entries) / number of filled entries >> + * (set by driver) >> + * @limits_ptr: pointer to limits array or NULL >> + */ >> +struct drm_exynos_ioctl_ipp_get_limits { >> + __u32 ipp_id; >> + __u32 fourcc; >> + __u64 modifier; >> + __u32 type; >> + __u32 limits_count; >> + __u64 limits_ptr; >> +}; >> + >> +enum drm_exynos_ipp_task_id { >> + /* buffer described by struct drm_exynos_ipp_task_buffer */ >> + DRM_EXYNOS_IPP_TASK_BUFFER = 0x0001, >> + /* rectangle described by struct drm_exynos_ipp_task_rect */ >> + DRM_EXYNOS_IPP_TASK_RECTANGLE = 0x0002, >> + /* transformation described by struct drm_exynos_ipp_task_transform */ >> + DRM_EXYNOS_IPP_TASK_TRANSFORM = 0x0003, >> + /* alpha configuration described by struct drm_exynos_ipp_task_alpha */ >> + DRM_EXYNOS_IPP_TASK_ALPHA = 0x0004, >> + >> + /* source image data (for buffer and rectangle chunks) */ >> + DRM_EXYNOS_IPP_TASK_TYPE_SOURCE = 0x0001 << 16, >> + /* destination image data (for buffer and rectangle chunks) */ >> + DRM_EXYNOS_IPP_TASK_TYPE_DESTINATION = 0x0002 << 16, >> +}; >> + >> +/** >> + * Memory buffer with image data. >> + * >> + * @id: must be DRM_EXYNOS_IPP_TASK_BUFFER >> + * other parameters are same as for AddFB2 generic DRM ioctl >> + */ >> +struct drm_exynos_ipp_task_buffer { >> + __u32 id; >> + __u32 fourcc; >> + __u32 width, height; >> + __u32 gem_id[4]; >> + __u32 offset[4]; >> + __u32 pitch[4]; >> + __u64 modifier; >> +}; >> + >> +/** >> + * Rectangle for processing. >> + * >> + * @id: must be DRM_EXYNOS_IPP_TASK_RECTANGLE >> + * @reserved: padding >> + * @x,@y: left corner in pixels >> + * @w,@h: width/height in pixels >> + */ >> +struct drm_exynos_ipp_task_rect { >> + __u32 id; >> + __u32 reserved; >> + __u32 x; >> + __u32 y; >> + __u32 w; >> + __u32 h; >> +}; >> + >> +/** >> + * Image tranformation description. >> + * >> + * @id: must be DRM_EXYNOS_IPP_TASK_TRANSFORM >> + * @rotation: DRM_MODE_ROTATE_* and DRM_MODE_REFLECT_* values >> + */ >> +struct drm_exynos_ipp_task_transform { >> + __u32 id; >> + __u32 rotation; >> +}; >> + >> +/** >> + * Image global alpha configuration for formats without alpha values. >> + * >> + * @id: must be DRM_EXYNOS_IPP_TASK_ALPHA >> + * @value: global alpha value (0-255) >> + */ >> +struct drm_exynos_ipp_task_alpha { >> + __u32 id; >> + __u32 value; >> +}; >> + >> +enum drm_exynos_ipp_flag { >> + /* generate DRM event after processing */ >> + DRM_EXYNOS_IPP_FLAG_EVENT = 0x01, >> + /* dry run, only check task parameters */ >> + DRM_EXYNOS_IPP_FLAG_TEST_ONLY = 0x02, >> + /* non-blocking processing */ >> + DRM_EXYNOS_IPP_FLAG_NONBLOCK = 0x04, >> +}; >> + >> +#define DRM_EXYNOS_IPP_FLAGS (DRM_EXYNOS_IPP_FLAG_EVENT |\ >> + DRM_EXYNOS_IPP_FLAG_TEST_ONLY | DRM_EXYNOS_IPP_FLAG_NONBLOCK) >> + >> +/** >> + * Perform image processing described by array of drm_exynos_ipp_task_* >> + * structures (parameters array). >> + * >> + * @ipp_id: id of IPP module to run the task >> + * @flags: bitmask of drm_exynos_ipp_flag values >> + * @reserved: padding >> + * @params_size: size of parameters array (in bytes) >> + * @params_ptr: pointer to parameters array or NULL >> + * @user_data: (optional) data for drm event >> + */ >> +struct drm_exynos_ioctl_ipp_commit { >> + __u32 ipp_id; >> + __u32 flags; >> + __u32 reserved; >> + __u32 params_size; >> + __u64 params_ptr; >> + __u64 user_data; >> +}; >> + >> #define DRM_EXYNOS_GEM_CREATE 0x00 >> #define DRM_EXYNOS_GEM_MAP 0x01 >> /* Reserved 0x03 ~ 0x05 for exynos specific gem ioctl */ >> @@ -146,6 +359,11 @@ struct drm_exynos_g2d_exec { >> #define DRM_EXYNOS_G2D_EXEC 0x22 >> >> /* Reserved 0x30 ~ 0x33 for obsolete Exynos IPP ioctls */ >> +/* IPP - Image Post Processing */ >> +#define DRM_EXYNOS_IPP_GET_RESOURCES 0x40 >> +#define DRM_EXYNOS_IPP_GET_CAPS 0x41 >> +#define DRM_EXYNOS_IPP_GET_LIMITS 0x42 >> +#define DRM_EXYNOS_IPP_COMMIT 0x43 >> >> #define DRM_IOCTL_EXYNOS_GEM_CREATE DRM_IOWR(DRM_COMMAND_BASE + \ >> DRM_EXYNOS_GEM_CREATE, struct drm_exynos_gem_create) >> @@ -164,8 +382,18 @@ struct drm_exynos_g2d_exec { >> #define DRM_IOCTL_EXYNOS_G2D_EXEC DRM_IOWR(DRM_COMMAND_BASE + \ >> DRM_EXYNOS_G2D_EXEC, struct drm_exynos_g2d_exec) >> >> +#define DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES DRM_IOWR(DRM_COMMAND_BASE + \ >> + DRM_EXYNOS_IPP_GET_RESOURCES, struct drm_exynos_ioctl_ipp_get_res) >> +#define DRM_IOCTL_EXYNOS_IPP_GET_CAPS DRM_IOWR(DRM_COMMAND_BASE + \ >> + DRM_EXYNOS_IPP_GET_CAPS, struct drm_exynos_ioctl_ipp_get_caps) >> +#define DRM_IOCTL_EXYNOS_IPP_GET_LIMITS DRM_IOWR(DRM_COMMAND_BASE + \ >> + DRM_EXYNOS_IPP_GET_LIMITS, struct drm_exynos_ioctl_ipp_get_limits) >> +#define DRM_IOCTL_EXYNOS_IPP_COMMIT DRM_IOWR(DRM_COMMAND_BASE + \ >> + DRM_EXYNOS_IPP_COMMIT, struct drm_exynos_ioctl_ipp_commit) >> + >> /* EXYNOS specific events */ >> #define DRM_EXYNOS_G2D_EVENT 0x80000000 >> +#define DRM_EXYNOS_IPP_EVENT 0x80000002 >> >> struct drm_exynos_g2d_event { >> struct drm_event base; >> @@ -176,6 +404,16 @@ struct drm_exynos_g2d_event { >> __u32 reserved; >> }; >> >> +struct drm_exynos_ipp_event { >> + struct drm_event base; >> + __u64 user_data; >> + __u32 tv_sec; >> + __u32 tv_usec; >> + __u32 ipp_id; >> + __u32 sequence; >> + __u64 reserved; >> +}; >> + >> #if defined(__cplusplus) >> } >> #endif >> > Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH v4 2/9] drm/exynos: ipp: Add IPP v2 framework 2017-10-23 7:54 ` [PATCH v4 2/9] drm/exynos: ipp: Add IPP v2 framework Marek Szyprowski 2017-11-01 1:13 ` Inki Dae @ 2017-11-01 3:47 ` Dave Airlie 2017-11-01 14:18 ` Tobias Jakobi 1 sibling, 1 reply; 20+ messages in thread From: Dave Airlie @ 2017-11-01 3:47 UTC (permalink / raw) To: Marek Szyprowski Cc: dri-devel, linux-samsung-soc, Bartlomiej Zolnierkiewicz, Seung-Woo Kim, Krzysztof Kozlowski, Tobias Jakobi, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon On 23 October 2017 at 17:54, Marek Szyprowski <m.szyprowski@samsung.com> wrote: > This patch adds Exynos IPP v2 subsystem and userspace API. > > New userspace API is focused ONLY on memory-to-memory image processing. > The two remainging IPP operation modes (framebuffer writeback and > local-path output with image processing) can be implemented using > standard DRM features: writeback connectors and additional DRM planes with > scaling features. > > V2 IPP userspace API is not compatible with old IPP ioctls. This is a > significant change, but the old IPP subsystem in mainline Linux kernel was > partially disfunctional anyway and not used in any open-source project. > > V2 IPP userspace API is based on stateless approach, which much better fits > to memory-to-memory image processing model. It also provides support for > all image formats, which are both already defined in DRM API and supported > by the existing IPP hardware modules. > > The API consists of the following ioctls: > - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image > processing modules, > - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image > formats of given IPP module, > - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for > selected image format of given IPP module, > - DRM_IOCTL_EXYNOS_IPP_COMMIT: to perform operation described by the > provided structures (source and destination buffers, operation rectangle, > transformation, etc). > > The proposed userspace API is extensible. In the future more advanced image > processing operations can be defined to support for example blending. > > Userspace API is fully functional also on DRM render nodes, so it is not > limited to the root/privileged client. > > Internal driver API also has been completely rewritten. New IPP core > performs all possible input validation, checks and object life-time > control. The drivers can focus only on writing configuration to hardware > registers. Stateless nature of DRM_IOCTL_EXYNOS_IPP_COMMIT ioctl simplifies > the driver API. Minimal driver needs to provide a single callback for > starting processing and an array with supported image formats. Is there userspace user for this? This is pretty significant driver-specific API addition. The commit should contain pointers to a userspace project using this, or patches to one. Dave. ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH v4 2/9] drm/exynos: ipp: Add IPP v2 framework 2017-11-01 3:47 ` Dave Airlie @ 2017-11-01 14:18 ` Tobias Jakobi 0 siblings, 0 replies; 20+ messages in thread From: Tobias Jakobi @ 2017-11-01 14:18 UTC (permalink / raw) To: Dave Airlie, Marek Szyprowski Cc: dri-devel, linux-samsung-soc, Bartlomiej Zolnierkiewicz, Seung-Woo Kim, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon Dave Airlie wrote: > On 23 October 2017 at 17:54, Marek Szyprowski <m.szyprowski@samsung.com> wrote: >> This patch adds Exynos IPP v2 subsystem and userspace API. >> >> New userspace API is focused ONLY on memory-to-memory image processing. >> The two remainging IPP operation modes (framebuffer writeback and >> local-path output with image processing) can be implemented using >> standard DRM features: writeback connectors and additional DRM planes with >> scaling features. >> >> V2 IPP userspace API is not compatible with old IPP ioctls. This is a >> significant change, but the old IPP subsystem in mainline Linux kernel was >> partially disfunctional anyway and not used in any open-source project. >> >> V2 IPP userspace API is based on stateless approach, which much better fits >> to memory-to-memory image processing model. It also provides support for >> all image formats, which are both already defined in DRM API and supported >> by the existing IPP hardware modules. >> >> The API consists of the following ioctls: >> - DRM_IOCTL_EXYNOS_IPP_GET_RESOURCES: to enumerate all available image >> processing modules, >> - DRM_IOCTL_EXYNOS_IPP_GET_CAPS: to query capabilities and supported image >> formats of given IPP module, >> - DRM_IOCTL_EXYNOS_IPP_GET_LIMITS: to query hardware limitiations for >> selected image format of given IPP module, >> - DRM_IOCTL_EXYNOS_IPP_COMMIT: to perform operation described by the >> provided structures (source and destination buffers, operation rectangle, >> transformation, etc). >> >> The proposed userspace API is extensible. In the future more advanced image >> processing operations can be defined to support for example blending. >> >> Userspace API is fully functional also on DRM render nodes, so it is not >> limited to the root/privileged client. >> >> Internal driver API also has been completely rewritten. New IPP core >> performs all possible input validation, checks and object life-time >> control. The drivers can focus only on writing configuration to hardware >> registers. Stateless nature of DRM_IOCTL_EXYNOS_IPP_COMMIT ioctl simplifies >> the driver API. Minimal driver needs to provide a single callback for >> starting processing and an array with supported image formats. > > Is there userspace user for this? This is pretty significant > driver-specific API addition. > > The commit should contain pointers to a userspace project using this, > or patches to one. The pointers are in the cover letter of the series. - Tobias > Dave. > ^ permalink raw reply [flat|nested] 20+ messages in thread
[parent not found: <CGME20171023075449eucas1p14adcde4cea94ca82a13a822ee59d9446@eucas1p1.samsung.com>]
* [PATCH v4 3/9] drm/exynos: rotator: Convert driver to IPP v2 core API [not found] ` <CGME20171023075449eucas1p14adcde4cea94ca82a13a822ee59d9446@eucas1p1.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski 2017-11-01 5:40 ` Inki Dae 0 siblings, 1 reply; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Bartlomiej Zolnierkiewicz, Seung-Woo Kim, Krzysztof Kozlowski, Tobias Jakobi, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon, Marek Szyprowski This patch adapts Exynos DRM rotator driver to new IPP v2 core API. The side effect of this conversion is a switch to driver component API to register properly in the Exynos DRM core. Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> --- drivers/gpu/drm/exynos/Kconfig | 2 +- drivers/gpu/drm/exynos/exynos_drm_drv.c | 1 + drivers/gpu/drm/exynos/exynos_drm_rotator.c | 740 +++++++--------------------- drivers/gpu/drm/exynos/exynos_drm_rotator.h | 19 - 4 files changed, 178 insertions(+), 584 deletions(-) delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_rotator.h diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig index 2e097a81df12..c10c9ca0d8b4 100644 --- a/drivers/gpu/drm/exynos/Kconfig +++ b/drivers/gpu/drm/exynos/Kconfig @@ -105,7 +105,7 @@ config DRM_EXYNOS_FIMC config DRM_EXYNOS_ROTATOR bool "Rotator" - depends on BROKEN + select DRM_EXYNOS_IPP help Choose this option if you want to use Exynos Rotator for DRM. diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c index de4cce258a5b..3ade2b0ad15d 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c @@ -257,6 +257,7 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { DRV_PTR(fimc_driver, CONFIG_DRM_EXYNOS_FIMC), }, { DRV_PTR(rotator_driver, CONFIG_DRM_EXYNOS_ROTATOR), + DRM_COMPONENT_DRIVER }, { DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC), }, { diff --git a/drivers/gpu/drm/exynos/exynos_drm_rotator.c b/drivers/gpu/drm/exynos/exynos_drm_rotator.c index 79282a820ecc..ffbc49b84562 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_rotator.c +++ b/drivers/gpu/drm/exynos/exynos_drm_rotator.c @@ -10,6 +10,7 @@ */ #include <linux/kernel.h> +#include <linux/component.h> #include <linux/err.h> #include <linux/interrupt.h> #include <linux/io.h> @@ -22,29 +23,18 @@ #include <drm/exynos_drm.h> #include "regs-rotator.h" #include "exynos_drm_drv.h" +#include "exynos_drm_iommu.h" #include "exynos_drm_ipp.h" /* * Rotator supports image crop/rotator and input/output DMA operations. * input DMA reads image data from the memory. * output DMA writes image data to memory. - * - * M2M operation : supports crop/scale/rotation/csc so on. - * Memory ----> Rotator H/W ----> Memory. */ -/* - * TODO - * 1. check suspend/resume api if needed. - * 2. need to check use case platform_device_id. - * 3. check src/dst size with, height. - * 4. need to add supported list in prop_list. - */ +#define ROTATOR_AUTOSUSPEND_DELAY 2000 -#define get_rot_context(dev) platform_get_drvdata(to_platform_device(dev)) -#define get_ctx_from_ippdrv(ippdrv) container_of(ippdrv,\ - struct rot_context, ippdrv); -#define rot_read(offset) readl(rot->regs + (offset)) +#define rot_read(offset) readl(rot->regs + (offset)) #define rot_write(cfg, offset) writel(cfg, rot->regs + (offset)) enum rot_irq_status { @@ -52,54 +42,22 @@ enum rot_irq_status { ROT_IRQ_STATUS_ILLEGAL = 9, }; -/* - * A structure of limitation. - * - * @min_w: minimum width. - * @min_h: minimum height. - * @max_w: maximum width. - * @max_h: maximum height. - * @align: align size. - */ -struct rot_limit { - u32 min_w; - u32 min_h; - u32 max_w; - u32 max_h; - u32 align; -}; - -/* - * A structure of limitation table. - * - * @ycbcr420_2p: case of YUV. - * @rgb888: case of RGB. - */ -struct rot_limit_table { - struct rot_limit ycbcr420_2p; - struct rot_limit rgb888; -}; - /* * A structure of rotator context. * @ippdrv: prepare initialization using ippdrv. - * @regs_res: register resources. * @regs: memory mapped io registers. * @clock: rotator gate clock. * @limit_tbl: limitation of rotator. * @irq: irq number. - * @cur_buf_id: current operation buffer id. - * @suspended: suspended state. */ struct rot_context { - struct exynos_drm_ippdrv ippdrv; - struct resource *regs_res; + struct exynos_drm_ipp ipp; + struct drm_device *drm_dev; + struct device *dev; void __iomem *regs; struct clk *clock; - struct rot_limit_table *limit_tbl; - int irq; - int cur_buf_id[EXYNOS_DRM_OPS_MAX]; - bool suspended; + const struct exynos_drm_ipp_formats *formats; + struct exynos_drm_ipp_task *task; }; static void rotator_reg_set_irq(struct rot_context *rot, bool enable) @@ -114,15 +72,6 @@ static void rotator_reg_set_irq(struct rot_context *rot, bool enable) rot_write(val, ROT_CONFIG); } -static u32 rotator_reg_get_fmt(struct rot_context *rot) -{ - u32 val = rot_read(ROT_CONTROL); - - val &= ROT_CONTROL_FMT_MASK; - - return val; -} - static enum rot_irq_status rotator_reg_get_irq_status(struct rot_context *rot) { u32 val = rot_read(ROT_STATUS); @@ -138,9 +87,6 @@ static enum rot_irq_status rotator_reg_get_irq_status(struct rot_context *rot) static irqreturn_t rotator_irq_handler(int irq, void *arg) { struct rot_context *rot = arg; - struct exynos_drm_ippdrv *ippdrv = &rot->ippdrv; - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; - struct drm_exynos_ipp_event_work *event_work = c_node->event_work; enum rot_irq_status irq_status; u32 val; @@ -152,56 +98,21 @@ static irqreturn_t rotator_irq_handler(int irq, void *arg) val |= ROT_STATUS_IRQ_PENDING((u32)irq_status); rot_write(val, ROT_STATUS); - if (irq_status == ROT_IRQ_STATUS_COMPLETE) { - event_work->ippdrv = ippdrv; - event_work->buf_id[EXYNOS_DRM_OPS_DST] = - rot->cur_buf_id[EXYNOS_DRM_OPS_DST]; - queue_work(ippdrv->event_workq, &event_work->work); - } else { - DRM_ERROR("the SFR is set illegally\n"); + if (rot->task) { + struct exynos_drm_ipp_task *task = rot->task; + + rot->task = NULL; + pm_runtime_mark_last_busy(rot->dev); + pm_runtime_put_autosuspend(rot->dev); + exynos_drm_ipp_task_done(task, + irq_status == ROT_IRQ_STATUS_COMPLETE ? 0 : -EINVAL); } return IRQ_HANDLED; } -static void rotator_align_size(struct rot_context *rot, u32 fmt, u32 *hsize, - u32 *vsize) -{ - struct rot_limit_table *limit_tbl = rot->limit_tbl; - struct rot_limit *limit; - u32 mask, val; - - /* Get size limit */ - if (fmt == ROT_CONTROL_FMT_RGB888) - limit = &limit_tbl->rgb888; - else - limit = &limit_tbl->ycbcr420_2p; - - /* Get mask for rounding to nearest aligned val */ - mask = ~((1 << limit->align) - 1); - - /* Set aligned width */ - val = ROT_ALIGN(*hsize, limit->align, mask); - if (val < limit->min_w) - *hsize = ROT_MIN(limit->min_w, mask); - else if (val > limit->max_w) - *hsize = ROT_MAX(limit->max_w, mask); - else - *hsize = val; - - /* Set aligned height */ - val = ROT_ALIGN(*vsize, limit->align, mask); - if (val < limit->min_h) - *vsize = ROT_MIN(limit->min_h, mask); - else if (val > limit->max_h) - *vsize = ROT_MAX(limit->max_h, mask); - else - *vsize = val; -} - -static int rotator_src_set_fmt(struct device *dev, u32 fmt) +static int rotator_src_set_fmt(struct rot_context *rot, u32 fmt) { - struct rot_context *rot = dev_get_drvdata(dev); u32 val; val = rot_read(ROT_CONTROL); @@ -214,9 +125,6 @@ static int rotator_src_set_fmt(struct device *dev, u32 fmt) case DRM_FORMAT_XRGB8888: val |= ROT_CONTROL_FMT_RGB888; break; - default: - DRM_ERROR("invalid image format\n"); - return -EINVAL; } rot_write(val, ROT_CONTROL); @@ -224,505 +132,176 @@ static int rotator_src_set_fmt(struct device *dev, u32 fmt) return 0; } -static inline bool rotator_check_reg_fmt(u32 fmt) +static int rotator_src_set_buf(struct rot_context *rot, + struct exynos_drm_ipp_buffer *buf) { - if ((fmt == ROT_CONTROL_FMT_YCBCR420_2P) || - (fmt == ROT_CONTROL_FMT_RGB888)) - return true; - - return false; -} - -static int rotator_src_set_size(struct device *dev, int swap, - struct drm_exynos_pos *pos, - struct drm_exynos_sz *sz) -{ - struct rot_context *rot = dev_get_drvdata(dev); - u32 fmt, hsize, vsize; u32 val; - /* Get format */ - fmt = rotator_reg_get_fmt(rot); - if (!rotator_check_reg_fmt(fmt)) { - DRM_ERROR("invalid format.\n"); - return -EINVAL; - } - - /* Align buffer size */ - hsize = sz->hsize; - vsize = sz->vsize; - rotator_align_size(rot, fmt, &hsize, &vsize); - /* Set buffer size configuration */ - val = ROT_SET_BUF_SIZE_H(vsize) | ROT_SET_BUF_SIZE_W(hsize); + val = ROT_SET_BUF_SIZE_H(buf->buf.height) | + ROT_SET_BUF_SIZE_W(buf->buf.pitch[0] / buf->format->cpp[0]); rot_write(val, ROT_SRC_BUF_SIZE); /* Set crop image position configuration */ - val = ROT_CROP_POS_Y(pos->y) | ROT_CROP_POS_X(pos->x); + val = ROT_CROP_POS_Y(buf->rect.y) | ROT_CROP_POS_X(buf->rect.x); rot_write(val, ROT_SRC_CROP_POS); - val = ROT_SRC_CROP_SIZE_H(pos->h) | ROT_SRC_CROP_SIZE_W(pos->w); + val = ROT_SRC_CROP_SIZE_H(buf->rect.h) | + ROT_SRC_CROP_SIZE_W(buf->rect.w); rot_write(val, ROT_SRC_CROP_SIZE); - return 0; -} - -static int rotator_src_set_addr(struct device *dev, - struct drm_exynos_ipp_buf_info *buf_info, - u32 buf_id, enum drm_exynos_ipp_buf_type buf_type) -{ - struct rot_context *rot = dev_get_drvdata(dev); - dma_addr_t addr[EXYNOS_DRM_PLANAR_MAX]; - u32 val, fmt, hsize, vsize; - int i; - - /* Set current buf_id */ - rot->cur_buf_id[EXYNOS_DRM_OPS_SRC] = buf_id; - - switch (buf_type) { - case IPP_BUF_ENQUEUE: - /* Set address configuration */ - for_each_ipp_planar(i) - addr[i] = buf_info->base[i]; - - /* Get format */ - fmt = rotator_reg_get_fmt(rot); - if (!rotator_check_reg_fmt(fmt)) { - DRM_ERROR("invalid format.\n"); - return -EINVAL; - } - - /* Re-set cb planar for NV12 format */ - if ((fmt == ROT_CONTROL_FMT_YCBCR420_2P) && - !addr[EXYNOS_DRM_PLANAR_CB]) { - - val = rot_read(ROT_SRC_BUF_SIZE); - hsize = ROT_GET_BUF_SIZE_W(val); - vsize = ROT_GET_BUF_SIZE_H(val); - - /* Set cb planar */ - addr[EXYNOS_DRM_PLANAR_CB] = - addr[EXYNOS_DRM_PLANAR_Y] + hsize * vsize; - } - - for_each_ipp_planar(i) - rot_write(addr[i], ROT_SRC_BUF_ADDR(i)); - break; - case IPP_BUF_DEQUEUE: - for_each_ipp_planar(i) - rot_write(0x0, ROT_SRC_BUF_ADDR(i)); - break; - default: - /* Nothing to do */ - break; - } + /* Set buffer DMA address */ + rot_write(buf->dma_addr[0], ROT_SRC_BUF_ADDR(0)); + rot_write(buf->dma_addr[1], ROT_SRC_BUF_ADDR(1)); return 0; } -static int rotator_dst_set_transf(struct device *dev, - enum drm_exynos_degree degree, - enum drm_exynos_flip flip, bool *swap) +static int rotator_dst_set_transf(struct rot_context *rot, + unsigned int rotation) { - struct rot_context *rot = dev_get_drvdata(dev); u32 val; /* Set transform configuration */ val = rot_read(ROT_CONTROL); val &= ~ROT_CONTROL_FLIP_MASK; - switch (flip) { - case EXYNOS_DRM_FLIP_VERTICAL: - val |= ROT_CONTROL_FLIP_VERTICAL; - break; - case EXYNOS_DRM_FLIP_HORIZONTAL: + if (rotation & DRM_MODE_REFLECT_X) val |= ROT_CONTROL_FLIP_HORIZONTAL; - break; - default: - /* Flip None */ - break; - } + if (rotation & DRM_MODE_REFLECT_Y) + val |= ROT_CONTROL_FLIP_VERTICAL; val &= ~ROT_CONTROL_ROT_MASK; - switch (degree) { - case EXYNOS_DRM_DEGREE_90: + if (rotation & DRM_MODE_ROTATE_90) val |= ROT_CONTROL_ROT_90; - break; - case EXYNOS_DRM_DEGREE_180: + else if (rotation & DRM_MODE_ROTATE_180) val |= ROT_CONTROL_ROT_180; - break; - case EXYNOS_DRM_DEGREE_270: + else if (rotation & DRM_MODE_ROTATE_270) val |= ROT_CONTROL_ROT_270; - break; - default: - /* Rotation 0 Degree */ - break; - } rot_write(val, ROT_CONTROL); - /* Check degree for setting buffer size swap */ - if ((degree == EXYNOS_DRM_DEGREE_90) || - (degree == EXYNOS_DRM_DEGREE_270)) - *swap = true; - else - *swap = false; - return 0; } -static int rotator_dst_set_size(struct device *dev, int swap, - struct drm_exynos_pos *pos, - struct drm_exynos_sz *sz) +static int rotator_dst_set_buf(struct rot_context *rot, + struct exynos_drm_ipp_buffer *buf) { - struct rot_context *rot = dev_get_drvdata(dev); - u32 val, fmt, hsize, vsize; - - /* Get format */ - fmt = rotator_reg_get_fmt(rot); - if (!rotator_check_reg_fmt(fmt)) { - DRM_ERROR("invalid format.\n"); - return -EINVAL; - } - - /* Align buffer size */ - hsize = sz->hsize; - vsize = sz->vsize; - rotator_align_size(rot, fmt, &hsize, &vsize); + u32 val; /* Set buffer size configuration */ - val = ROT_SET_BUF_SIZE_H(vsize) | ROT_SET_BUF_SIZE_W(hsize); + val = ROT_SET_BUF_SIZE_H(buf->buf.height) | + ROT_SET_BUF_SIZE_W(buf->buf.pitch[0] / buf->format->cpp[0]); rot_write(val, ROT_DST_BUF_SIZE); /* Set crop image position configuration */ - val = ROT_CROP_POS_Y(pos->y) | ROT_CROP_POS_X(pos->x); + val = ROT_CROP_POS_Y(buf->rect.y) | ROT_CROP_POS_X(buf->rect.x); rot_write(val, ROT_DST_CROP_POS); - return 0; -} - -static int rotator_dst_set_addr(struct device *dev, - struct drm_exynos_ipp_buf_info *buf_info, - u32 buf_id, enum drm_exynos_ipp_buf_type buf_type) -{ - struct rot_context *rot = dev_get_drvdata(dev); - dma_addr_t addr[EXYNOS_DRM_PLANAR_MAX]; - u32 val, fmt, hsize, vsize; - int i; - - /* Set current buf_id */ - rot->cur_buf_id[EXYNOS_DRM_OPS_DST] = buf_id; - - switch (buf_type) { - case IPP_BUF_ENQUEUE: - /* Set address configuration */ - for_each_ipp_planar(i) - addr[i] = buf_info->base[i]; - - /* Get format */ - fmt = rotator_reg_get_fmt(rot); - if (!rotator_check_reg_fmt(fmt)) { - DRM_ERROR("invalid format.\n"); - return -EINVAL; - } - - /* Re-set cb planar for NV12 format */ - if ((fmt == ROT_CONTROL_FMT_YCBCR420_2P) && - !addr[EXYNOS_DRM_PLANAR_CB]) { - /* Get buf size */ - val = rot_read(ROT_DST_BUF_SIZE); - - hsize = ROT_GET_BUF_SIZE_W(val); - vsize = ROT_GET_BUF_SIZE_H(val); - - /* Set cb planar */ - addr[EXYNOS_DRM_PLANAR_CB] = - addr[EXYNOS_DRM_PLANAR_Y] + hsize * vsize; - } - - for_each_ipp_planar(i) - rot_write(addr[i], ROT_DST_BUF_ADDR(i)); - break; - case IPP_BUF_DEQUEUE: - for_each_ipp_planar(i) - rot_write(0x0, ROT_DST_BUF_ADDR(i)); - break; - default: - /* Nothing to do */ - break; - } + /* Set buffer DMA address */ + rot_write(buf->dma_addr[0], ROT_DST_BUF_ADDR(0)); + rot_write(buf->dma_addr[1], ROT_DST_BUF_ADDR(1)); return 0; } -static struct exynos_drm_ipp_ops rot_src_ops = { - .set_fmt = rotator_src_set_fmt, - .set_size = rotator_src_set_size, - .set_addr = rotator_src_set_addr, -}; +static int rotator_start(struct rot_context *rot) +{ + u32 val; -static struct exynos_drm_ipp_ops rot_dst_ops = { - .set_transf = rotator_dst_set_transf, - .set_size = rotator_dst_set_size, - .set_addr = rotator_dst_set_addr, -}; + /* Set interrupt enable */ + rotator_reg_set_irq(rot, true); -static int rotator_init_prop_list(struct exynos_drm_ippdrv *ippdrv) -{ - struct drm_exynos_ipp_prop_list *prop_list = &ippdrv->prop_list; - - prop_list->version = 1; - prop_list->flip = (1 << EXYNOS_DRM_FLIP_VERTICAL) | - (1 << EXYNOS_DRM_FLIP_HORIZONTAL); - prop_list->degree = (1 << EXYNOS_DRM_DEGREE_0) | - (1 << EXYNOS_DRM_DEGREE_90) | - (1 << EXYNOS_DRM_DEGREE_180) | - (1 << EXYNOS_DRM_DEGREE_270); - prop_list->csc = 0; - prop_list->crop = 0; - prop_list->scale = 0; + val = rot_read(ROT_CONTROL); + val |= ROT_CONTROL_START; + rot_write(val, ROT_CONTROL); return 0; } -static inline bool rotator_check_drm_fmt(u32 fmt) +static int rotator_commit(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) { - switch (fmt) { - case DRM_FORMAT_XRGB8888: - case DRM_FORMAT_NV12: - return true; - default: - DRM_DEBUG_KMS("not support format\n"); - return false; - } -} + struct rot_context *rot = + container_of(ipp, struct rot_context, ipp); -static inline bool rotator_check_drm_flip(enum drm_exynos_flip flip) -{ - switch (flip) { - case EXYNOS_DRM_FLIP_NONE: - case EXYNOS_DRM_FLIP_VERTICAL: - case EXYNOS_DRM_FLIP_HORIZONTAL: - case EXYNOS_DRM_FLIP_BOTH: - return true; - default: - DRM_DEBUG_KMS("invalid flip\n"); - return false; - } -} + pm_runtime_get_sync(rot->dev); + rot->task = task; -static int rotator_ippdrv_check_property(struct device *dev, - struct drm_exynos_ipp_property *property) -{ - struct drm_exynos_ipp_config *src_config = - &property->config[EXYNOS_DRM_OPS_SRC]; - struct drm_exynos_ipp_config *dst_config = - &property->config[EXYNOS_DRM_OPS_DST]; - struct drm_exynos_pos *src_pos = &src_config->pos; - struct drm_exynos_pos *dst_pos = &dst_config->pos; - struct drm_exynos_sz *src_sz = &src_config->sz; - struct drm_exynos_sz *dst_sz = &dst_config->sz; - bool swap = false; - - /* Check format configuration */ - if (src_config->fmt != dst_config->fmt) { - DRM_DEBUG_KMS("not support csc feature\n"); - return -EINVAL; - } - - if (!rotator_check_drm_fmt(dst_config->fmt)) { - DRM_DEBUG_KMS("invalid format\n"); - return -EINVAL; - } - - /* Check transform configuration */ - if (src_config->degree != EXYNOS_DRM_DEGREE_0) { - DRM_DEBUG_KMS("not support source-side rotation\n"); - return -EINVAL; - } - - switch (dst_config->degree) { - case EXYNOS_DRM_DEGREE_90: - case EXYNOS_DRM_DEGREE_270: - swap = true; - case EXYNOS_DRM_DEGREE_0: - case EXYNOS_DRM_DEGREE_180: - /* No problem */ - break; - default: - DRM_DEBUG_KMS("invalid degree\n"); - return -EINVAL; - } - - if (src_config->flip != EXYNOS_DRM_FLIP_NONE) { - DRM_DEBUG_KMS("not support source-side flip\n"); - return -EINVAL; - } - - if (!rotator_check_drm_flip(dst_config->flip)) { - DRM_DEBUG_KMS("invalid flip\n"); - return -EINVAL; - } - - /* Check size configuration */ - if ((src_pos->x + src_pos->w > src_sz->hsize) || - (src_pos->y + src_pos->h > src_sz->vsize)) { - DRM_DEBUG_KMS("out of source buffer bound\n"); - return -EINVAL; - } - - if (swap) { - if ((dst_pos->x + dst_pos->h > dst_sz->vsize) || - (dst_pos->y + dst_pos->w > dst_sz->hsize)) { - DRM_DEBUG_KMS("out of destination buffer bound\n"); - return -EINVAL; - } - - if ((src_pos->w != dst_pos->h) || (src_pos->h != dst_pos->w)) { - DRM_DEBUG_KMS("not support scale feature\n"); - return -EINVAL; - } - } else { - if ((dst_pos->x + dst_pos->w > dst_sz->hsize) || - (dst_pos->y + dst_pos->h > dst_sz->vsize)) { - DRM_DEBUG_KMS("out of destination buffer bound\n"); - return -EINVAL; - } - - if ((src_pos->w != dst_pos->w) || (src_pos->h != dst_pos->h)) { - DRM_DEBUG_KMS("not support scale feature\n"); - return -EINVAL; - } - } + rotator_src_set_fmt(rot, task->src.buf.fourcc); + rotator_src_set_buf(rot, &task->src); + rotator_dst_set_transf(rot, task->transform.rotation); + rotator_dst_set_buf(rot, &task->dst); + rotator_start(rot); return 0; } -static int rotator_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd) +static const struct exynos_drm_ipp_funcs ipp_funcs = { + .commit = rotator_commit, +}; + +static int rotator_bind(struct device *dev, struct device *master, void *data) { struct rot_context *rot = dev_get_drvdata(dev); - u32 val; - - if (rot->suspended) { - DRM_ERROR("suspended state\n"); - return -EPERM; - } - - if (cmd != IPP_CMD_M2M) { - DRM_ERROR("not support cmd: %d\n", cmd); - return -EINVAL; - } + struct drm_device *drm_dev = data; + struct exynos_drm_ipp *ipp = &rot->ipp; - /* Set interrupt enable */ - rotator_reg_set_irq(rot, true); + rot->drm_dev = drm_dev; + drm_iommu_attach_device(drm_dev, dev); - val = rot_read(ROT_CONTROL); - val |= ROT_CONTROL_START; + exynos_drm_ipp_register(drm_dev, ipp, &ipp_funcs, + DRM_EXYNOS_IPP_CAP_CROP | DRM_EXYNOS_IPP_CAP_ROTATE, + rot->formats, "rotator"); - rot_write(val, ROT_CONTROL); + dev_info(dev, "The exynos rotator has been probed successfully\n"); return 0; } -static struct rot_limit_table rot_limit_tbl_4210 = { - .ycbcr420_2p = { - .min_w = 32, - .min_h = 32, - .max_w = SZ_64K, - .max_h = SZ_64K, - .align = 3, - }, - .rgb888 = { - .min_w = 8, - .min_h = 8, - .max_w = SZ_16K, - .max_h = SZ_16K, - .align = 2, - }, -}; +static void rotator_unbind(struct device *dev, struct device *master, + void *data) +{ + struct rot_context *rot = dev_get_drvdata(dev); + struct drm_device *drm_dev = data; + struct exynos_drm_ipp *ipp = &rot->ipp; -static struct rot_limit_table rot_limit_tbl_4x12 = { - .ycbcr420_2p = { - .min_w = 32, - .min_h = 32, - .max_w = SZ_32K, - .max_h = SZ_32K, - .align = 3, - }, - .rgb888 = { - .min_w = 8, - .min_h = 8, - .max_w = SZ_8K, - .max_h = SZ_8K, - .align = 2, - }, -}; + exynos_drm_ipp_unregister(drm_dev, ipp); + drm_iommu_detach_device(rot->drm_dev, rot->dev); +} -static struct rot_limit_table rot_limit_tbl_5250 = { - .ycbcr420_2p = { - .min_w = 32, - .min_h = 32, - .max_w = SZ_32K, - .max_h = SZ_32K, - .align = 3, - }, - .rgb888 = { - .min_w = 8, - .min_h = 8, - .max_w = SZ_8K, - .max_h = SZ_8K, - .align = 1, - }, +static const struct component_ops rotator_component_ops = { + .bind = rotator_bind, + .unbind = rotator_unbind, }; -static const struct of_device_id exynos_rotator_match[] = { - { - .compatible = "samsung,exynos4210-rotator", - .data = &rot_limit_tbl_4210, - }, - { - .compatible = "samsung,exynos4212-rotator", - .data = &rot_limit_tbl_4x12, - }, - { - .compatible = "samsung,exynos5250-rotator", - .data = &rot_limit_tbl_5250, - }, - {}, -}; -MODULE_DEVICE_TABLE(of, exynos_rotator_match); - static int rotator_probe(struct platform_device *pdev) { struct device *dev = &pdev->dev; + struct resource *regs_res; struct rot_context *rot; - struct exynos_drm_ippdrv *ippdrv; + int irq; int ret; - if (!dev->of_node) { - dev_err(dev, "cannot find of_node.\n"); - return -ENODEV; - } - rot = devm_kzalloc(dev, sizeof(*rot), GFP_KERNEL); if (!rot) return -ENOMEM; - rot->limit_tbl = (struct rot_limit_table *) - of_device_get_match_data(dev); - rot->regs_res = platform_get_resource(pdev, IORESOURCE_MEM, 0); - rot->regs = devm_ioremap_resource(dev, rot->regs_res); + rot->formats = of_device_get_match_data(dev); + rot->dev = dev; + regs_res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + rot->regs = devm_ioremap_resource(dev, regs_res); if (IS_ERR(rot->regs)) return PTR_ERR(rot->regs); - rot->irq = platform_get_irq(pdev, 0); - if (rot->irq < 0) { + irq = platform_get_irq(pdev, 0); + if (irq < 0) { dev_err(dev, "failed to get irq\n"); - return rot->irq; + return irq; } - ret = devm_request_threaded_irq(dev, rot->irq, NULL, - rotator_irq_handler, IRQF_ONESHOT, "drm_rotator", rot); + ret = devm_request_irq(dev, irq, rotator_irq_handler, 0, dev_name(dev), + rot); if (ret < 0) { dev_err(dev, "failed to request irq\n"); return ret; @@ -734,35 +313,19 @@ static int rotator_probe(struct platform_device *pdev) return PTR_ERR(rot->clock); } + pm_runtime_use_autosuspend(dev); + pm_runtime_set_autosuspend_delay(dev, ROTATOR_AUTOSUSPEND_DELAY); pm_runtime_enable(dev); - - ippdrv = &rot->ippdrv; - ippdrv->dev = dev; - ippdrv->ops[EXYNOS_DRM_OPS_SRC] = &rot_src_ops; - ippdrv->ops[EXYNOS_DRM_OPS_DST] = &rot_dst_ops; - ippdrv->check_property = rotator_ippdrv_check_property; - ippdrv->start = rotator_ippdrv_start; - ret = rotator_init_prop_list(ippdrv); - if (ret < 0) { - dev_err(dev, "failed to init property list.\n"); - goto err_ippdrv_register; - } - - DRM_DEBUG_KMS("ippdrv[%pK]\n", ippdrv); - platform_set_drvdata(pdev, rot); - ret = exynos_drm_ippdrv_register(ippdrv); - if (ret < 0) { - dev_err(dev, "failed to register drm rotator device\n"); - goto err_ippdrv_register; - } - - dev_info(dev, "The exynos rotator is probed successfully\n"); + ret = component_add(dev, &rotator_component_ops); + if (ret) + goto err_component; return 0; -err_ippdrv_register: +err_component: + pm_runtime_dont_use_autosuspend(dev); pm_runtime_disable(dev); return ret; } @@ -770,45 +333,94 @@ static int rotator_probe(struct platform_device *pdev) static int rotator_remove(struct platform_device *pdev) { struct device *dev = &pdev->dev; - struct rot_context *rot = dev_get_drvdata(dev); - struct exynos_drm_ippdrv *ippdrv = &rot->ippdrv; - - exynos_drm_ippdrv_unregister(ippdrv); + component_del(dev, &rotator_component_ops); + pm_runtime_dont_use_autosuspend(dev); pm_runtime_disable(dev); return 0; } #ifdef CONFIG_PM -static int rotator_clk_crtl(struct rot_context *rot, bool enable) -{ - if (enable) { - clk_prepare_enable(rot->clock); - rot->suspended = false; - } else { - clk_disable_unprepare(rot->clock); - rot->suspended = true; - } - - return 0; -} - static int rotator_runtime_suspend(struct device *dev) { struct rot_context *rot = dev_get_drvdata(dev); - return rotator_clk_crtl(rot, false); + clk_disable_unprepare(rot->clock); + return 0; } static int rotator_runtime_resume(struct device *dev) { struct rot_context *rot = dev_get_drvdata(dev); - return rotator_clk_crtl(rot, true); + return clk_prepare_enable(rot->clock); } #endif +static const struct drm_exynos_ipp_limit rotator_4210_rbg888_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {8, SZ_16K}, .v = {8, SZ_16K}) }, + { IPP_SIZE_LIMIT(AREA, .h.align = 4, .v.align = 4) }, + { } +}; + +static const struct drm_exynos_ipp_limit rotator_4412_rbg888_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {8, SZ_8K}, .v = {8, SZ_8K}) }, + { IPP_SIZE_LIMIT(AREA, .h.align = 4, .v.align = 4) }, + { } +}; + +static const struct drm_exynos_ipp_limit rotator_5250_rbg888_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {8, SZ_8K}, .v = {8, SZ_8K}) }, + { IPP_SIZE_LIMIT(AREA, .h.align = 2, .v.align = 2) }, + { } +}; + +static const struct drm_exynos_ipp_limit rotator_4210_yuv_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {32, SZ_64K}, .v = {32, SZ_64K}) }, + { IPP_SIZE_LIMIT(AREA, .h.align = 8, .v.align = 8) }, + { } +}; + +static const struct drm_exynos_ipp_limit rotator_4412_yuv_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {32, SZ_32K}, .v = {32, SZ_32K}) }, + { IPP_SIZE_LIMIT(AREA, .h.align = 8, .v.align = 8) }, + { } +}; + +static const struct exynos_drm_ipp_formats rotator_4210_formats[] = { + { IPP_SRCDST_FORMAT(XRGB8888, rotator_4210_rbg888_limits) }, + { IPP_SRCDST_FORMAT(NV12, rotator_4210_yuv_limits) }, + { } +}; + +static const struct exynos_drm_ipp_formats rotator_4412_formats[] = { + { IPP_SRCDST_FORMAT(XRGB8888, rotator_4412_rbg888_limits) }, + { IPP_SRCDST_FORMAT(NV12, rotator_4412_yuv_limits) }, + { } +}; + +static const struct exynos_drm_ipp_formats rotator_5250_formats[] = { + { IPP_SRCDST_FORMAT(XRGB8888, rotator_5250_rbg888_limits) }, + { IPP_SRCDST_FORMAT(NV12, rotator_4412_yuv_limits) }, + { } +}; + +static const struct of_device_id exynos_rotator_match[] = { + { + .compatible = "samsung,exynos4210-rotator", + .data = rotator_4210_formats, + }, { + .compatible = "samsung,exynos4212-rotator", + .data = rotator_4412_formats, + }, { + .compatible = "samsung,exynos5250-rotator", + .data = rotator_5250_formats, + }, { + }, +}; +MODULE_DEVICE_TABLE(of, exynos_rotator_match); + static const struct dev_pm_ops rotator_pm_ops = { SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend, pm_runtime_force_resume) @@ -820,7 +432,7 @@ struct platform_driver rotator_driver = { .probe = rotator_probe, .remove = rotator_remove, .driver = { - .name = "exynos-rot", + .name = "exynos-rotator", .owner = THIS_MODULE, .pm = &rotator_pm_ops, .of_match_table = exynos_rotator_match, diff --git a/drivers/gpu/drm/exynos/exynos_drm_rotator.h b/drivers/gpu/drm/exynos/exynos_drm_rotator.h deleted file mode 100644 index 71a0b4c0c1e8..000000000000 --- a/drivers/gpu/drm/exynos/exynos_drm_rotator.h +++ /dev/null @@ -1,19 +0,0 @@ -/* - * Copyright (c) 2012 Samsung Electronics Co., Ltd. - * - * Authors: - * YoungJun Cho <yj44.cho@samsung.com> - * Eunchul Kim <chulspro.kim@samsung.com> - * - * This program is free software; you can redistribute it and/or modify it - * under the terms of the GNU General Public License as published by the - * Free Software Foundation; either version 2 of the License, or (at your - * option) any later version. - */ - -#ifndef _EXYNOS_DRM_ROTATOR_H_ -#define _EXYNOS_DRM_ROTATOR_H_ - -/* TODO */ - -#endif -- 2.14.2 _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH v4 3/9] drm/exynos: rotator: Convert driver to IPP v2 core API 2017-10-23 7:54 ` [PATCH v4 3/9] drm/exynos: rotator: Convert driver to IPP v2 core API Marek Szyprowski @ 2017-11-01 5:40 ` Inki Dae 0 siblings, 0 replies; 20+ messages in thread From: Inki Dae @ 2017-11-01 5:40 UTC (permalink / raw) To: Marek Szyprowski, dri-devel, linux-samsung-soc Cc: Tobias Jakobi, Bartlomiej Zolnierkiewicz, Seung-Woo Kim, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon 2017년 10월 23일 16:54에 Marek Szyprowski 이(가) 쓴 글: > This patch adapts Exynos DRM rotator driver to new IPP v2 core API. > The side effect of this conversion is a switch to driver component API > to register properly in the Exynos DRM core. > > Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> > --- > drivers/gpu/drm/exynos/Kconfig | 2 +- > drivers/gpu/drm/exynos/exynos_drm_drv.c | 1 + > drivers/gpu/drm/exynos/exynos_drm_rotator.c | 740 +++++++--------------------- > drivers/gpu/drm/exynos/exynos_drm_rotator.h | 19 - > 4 files changed, 178 insertions(+), 584 deletions(-) > delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_rotator.h > > diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig > index 2e097a81df12..c10c9ca0d8b4 100644 > --- a/drivers/gpu/drm/exynos/Kconfig > +++ b/drivers/gpu/drm/exynos/Kconfig > @@ -105,7 +105,7 @@ config DRM_EXYNOS_FIMC > > config DRM_EXYNOS_ROTATOR > bool "Rotator" > - depends on BROKEN > + select DRM_EXYNOS_IPP > help > Choose this option if you want to use Exynos Rotator for DRM. > > diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c > index de4cce258a5b..3ade2b0ad15d 100644 > --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c > +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c > @@ -257,6 +257,7 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { > DRV_PTR(fimc_driver, CONFIG_DRM_EXYNOS_FIMC), > }, { > DRV_PTR(rotator_driver, CONFIG_DRM_EXYNOS_ROTATOR), > + DRM_COMPONENT_DRIVER > }, { > DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC), > }, { > diff --git a/drivers/gpu/drm/exynos/exynos_drm_rotator.c b/drivers/gpu/drm/exynos/exynos_drm_rotator.c > index 79282a820ecc..ffbc49b84562 100644 > --- a/drivers/gpu/drm/exynos/exynos_drm_rotator.c > +++ b/drivers/gpu/drm/exynos/exynos_drm_rotator.c > @@ -10,6 +10,7 @@ > */ > > #include <linux/kernel.h> > +#include <linux/component.h> > #include <linux/err.h> > #include <linux/interrupt.h> > #include <linux/io.h> > @@ -22,29 +23,18 @@ > #include <drm/exynos_drm.h> > #include "regs-rotator.h" > #include "exynos_drm_drv.h" > +#include "exynos_drm_iommu.h" > #include "exynos_drm_ipp.h" > > /* > * Rotator supports image crop/rotator and input/output DMA operations. > * input DMA reads image data from the memory. > * output DMA writes image data to memory. > - * > - * M2M operation : supports crop/scale/rotation/csc so on. > - * Memory ----> Rotator H/W ----> Memory. > */ > > -/* > - * TODO > - * 1. check suspend/resume api if needed. > - * 2. need to check use case platform_device_id. > - * 3. check src/dst size with, height. > - * 4. need to add supported list in prop_list. > - */ > +#define ROTATOR_AUTOSUSPEND_DELAY 2000 > > -#define get_rot_context(dev) platform_get_drvdata(to_platform_device(dev)) > -#define get_ctx_from_ippdrv(ippdrv) container_of(ippdrv,\ > - struct rot_context, ippdrv); > -#define rot_read(offset) readl(rot->regs + (offset)) > +#define rot_read(offset) readl(rot->regs + (offset)) > #define rot_write(cfg, offset) writel(cfg, rot->regs + (offset)) > > enum rot_irq_status { > @@ -52,54 +42,22 @@ enum rot_irq_status { > ROT_IRQ_STATUS_ILLEGAL = 9, > }; > > -/* > - * A structure of limitation. > - * > - * @min_w: minimum width. > - * @min_h: minimum height. > - * @max_w: maximum width. > - * @max_h: maximum height. > - * @align: align size. > - */ > -struct rot_limit { > - u32 min_w; > - u32 min_h; > - u32 max_w; > - u32 max_h; > - u32 align; > -}; > - > -/* > - * A structure of limitation table. > - * > - * @ycbcr420_2p: case of YUV. > - * @rgb888: case of RGB. > - */ > -struct rot_limit_table { > - struct rot_limit ycbcr420_2p; > - struct rot_limit rgb888; > -}; > - > /* > * A structure of rotator context. > * @ippdrv: prepare initialization using ippdrv. > - * @regs_res: register resources. > * @regs: memory mapped io registers. > * @clock: rotator gate clock. > * @limit_tbl: limitation of rotator. > * @irq: irq number. > - * @cur_buf_id: current operation buffer id. > - * @suspended: suspended state. > */ > struct rot_context { > - struct exynos_drm_ippdrv ippdrv; > - struct resource *regs_res; > + struct exynos_drm_ipp ipp; > + struct drm_device *drm_dev; > + struct device *dev; > void __iomem *regs; > struct clk *clock; > - struct rot_limit_table *limit_tbl; > - int irq; > - int cur_buf_id[EXYNOS_DRM_OPS_MAX]; > - bool suspended; > + const struct exynos_drm_ipp_formats *formats; > + struct exynos_drm_ipp_task *task; > }; > > static void rotator_reg_set_irq(struct rot_context *rot, bool enable) > @@ -114,15 +72,6 @@ static void rotator_reg_set_irq(struct rot_context *rot, bool enable) > rot_write(val, ROT_CONFIG); > } > > -static u32 rotator_reg_get_fmt(struct rot_context *rot) > -{ > - u32 val = rot_read(ROT_CONTROL); > - > - val &= ROT_CONTROL_FMT_MASK; > - > - return val; > -} > - > static enum rot_irq_status rotator_reg_get_irq_status(struct rot_context *rot) > { > u32 val = rot_read(ROT_STATUS); > @@ -138,9 +87,6 @@ static enum rot_irq_status rotator_reg_get_irq_status(struct rot_context *rot) > static irqreturn_t rotator_irq_handler(int irq, void *arg) > { > struct rot_context *rot = arg; > - struct exynos_drm_ippdrv *ippdrv = &rot->ippdrv; > - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; > - struct drm_exynos_ipp_event_work *event_work = c_node->event_work; > enum rot_irq_status irq_status; > u32 val; > > @@ -152,56 +98,21 @@ static irqreturn_t rotator_irq_handler(int irq, void *arg) > val |= ROT_STATUS_IRQ_PENDING((u32)irq_status); > rot_write(val, ROT_STATUS); > > - if (irq_status == ROT_IRQ_STATUS_COMPLETE) { > - event_work->ippdrv = ippdrv; > - event_work->buf_id[EXYNOS_DRM_OPS_DST] = > - rot->cur_buf_id[EXYNOS_DRM_OPS_DST]; > - queue_work(ippdrv->event_workq, &event_work->work); > - } else { > - DRM_ERROR("the SFR is set illegally\n"); > + if (rot->task) { > + struct exynos_drm_ipp_task *task = rot->task; > + > + rot->task = NULL; > + pm_runtime_mark_last_busy(rot->dev); > + pm_runtime_put_autosuspend(rot->dev); > + exynos_drm_ipp_task_done(task, > + irq_status == ROT_IRQ_STATUS_COMPLETE ? 0 : -EINVAL); > } > > return IRQ_HANDLED; > } > > -static void rotator_align_size(struct rot_context *rot, u32 fmt, u32 *hsize, > - u32 *vsize) > -{ > - struct rot_limit_table *limit_tbl = rot->limit_tbl; > - struct rot_limit *limit; > - u32 mask, val; > - > - /* Get size limit */ > - if (fmt == ROT_CONTROL_FMT_RGB888) > - limit = &limit_tbl->rgb888; > - else > - limit = &limit_tbl->ycbcr420_2p; > - > - /* Get mask for rounding to nearest aligned val */ > - mask = ~((1 << limit->align) - 1); > - > - /* Set aligned width */ > - val = ROT_ALIGN(*hsize, limit->align, mask); > - if (val < limit->min_w) > - *hsize = ROT_MIN(limit->min_w, mask); > - else if (val > limit->max_w) > - *hsize = ROT_MAX(limit->max_w, mask); > - else > - *hsize = val; > - > - /* Set aligned height */ > - val = ROT_ALIGN(*vsize, limit->align, mask); > - if (val < limit->min_h) > - *vsize = ROT_MIN(limit->min_h, mask); > - else if (val > limit->max_h) > - *vsize = ROT_MAX(limit->max_h, mask); > - else > - *vsize = val; > -} > - > -static int rotator_src_set_fmt(struct device *dev, u32 fmt) > +static int rotator_src_set_fmt(struct rot_context *rot, u32 fmt) Never return any error so this function doesn't need return type. > { > - struct rot_context *rot = dev_get_drvdata(dev); > u32 val; > > val = rot_read(ROT_CONTROL); > @@ -214,9 +125,6 @@ static int rotator_src_set_fmt(struct device *dev, u32 fmt) > case DRM_FORMAT_XRGB8888: > val |= ROT_CONTROL_FMT_RGB888; > break; > - default: > - DRM_ERROR("invalid image format\n"); > - return -EINVAL; > } > > rot_write(val, ROT_CONTROL); > @@ -224,505 +132,176 @@ static int rotator_src_set_fmt(struct device *dev, u32 fmt) > return 0; > } > > -static inline bool rotator_check_reg_fmt(u32 fmt) > +static int rotator_src_set_buf(struct rot_context *rot, > + struct exynos_drm_ipp_buffer *buf) ditto. > { > - if ((fmt == ROT_CONTROL_FMT_YCBCR420_2P) || > - (fmt == ROT_CONTROL_FMT_RGB888)) > - return true; > - > - return false; > -} > - > -static int rotator_src_set_size(struct device *dev, int swap, > - struct drm_exynos_pos *pos, > - struct drm_exynos_sz *sz) > -{ > - struct rot_context *rot = dev_get_drvdata(dev); > - u32 fmt, hsize, vsize; > u32 val; > > - /* Get format */ > - fmt = rotator_reg_get_fmt(rot); > - if (!rotator_check_reg_fmt(fmt)) { > - DRM_ERROR("invalid format.\n"); > - return -EINVAL; > - } > - > - /* Align buffer size */ > - hsize = sz->hsize; > - vsize = sz->vsize; > - rotator_align_size(rot, fmt, &hsize, &vsize); > - > /* Set buffer size configuration */ > - val = ROT_SET_BUF_SIZE_H(vsize) | ROT_SET_BUF_SIZE_W(hsize); > + val = ROT_SET_BUF_SIZE_H(buf->buf.height) | > + ROT_SET_BUF_SIZE_W(buf->buf.pitch[0] / buf->format->cpp[0]); > rot_write(val, ROT_SRC_BUF_SIZE); > > /* Set crop image position configuration */ > - val = ROT_CROP_POS_Y(pos->y) | ROT_CROP_POS_X(pos->x); > + val = ROT_CROP_POS_Y(buf->rect.y) | ROT_CROP_POS_X(buf->rect.x); > rot_write(val, ROT_SRC_CROP_POS); > - val = ROT_SRC_CROP_SIZE_H(pos->h) | ROT_SRC_CROP_SIZE_W(pos->w); > + val = ROT_SRC_CROP_SIZE_H(buf->rect.h) | > + ROT_SRC_CROP_SIZE_W(buf->rect.w); > rot_write(val, ROT_SRC_CROP_SIZE); > > - return 0; > -} > - > -static int rotator_src_set_addr(struct device *dev, > - struct drm_exynos_ipp_buf_info *buf_info, > - u32 buf_id, enum drm_exynos_ipp_buf_type buf_type) > -{ > - struct rot_context *rot = dev_get_drvdata(dev); > - dma_addr_t addr[EXYNOS_DRM_PLANAR_MAX]; > - u32 val, fmt, hsize, vsize; > - int i; > - > - /* Set current buf_id */ > - rot->cur_buf_id[EXYNOS_DRM_OPS_SRC] = buf_id; > - > - switch (buf_type) { > - case IPP_BUF_ENQUEUE: > - /* Set address configuration */ > - for_each_ipp_planar(i) > - addr[i] = buf_info->base[i]; > - > - /* Get format */ > - fmt = rotator_reg_get_fmt(rot); > - if (!rotator_check_reg_fmt(fmt)) { > - DRM_ERROR("invalid format.\n"); > - return -EINVAL; > - } > - > - /* Re-set cb planar for NV12 format */ > - if ((fmt == ROT_CONTROL_FMT_YCBCR420_2P) && > - !addr[EXYNOS_DRM_PLANAR_CB]) { > - > - val = rot_read(ROT_SRC_BUF_SIZE); > - hsize = ROT_GET_BUF_SIZE_W(val); > - vsize = ROT_GET_BUF_SIZE_H(val); > - > - /* Set cb planar */ > - addr[EXYNOS_DRM_PLANAR_CB] = > - addr[EXYNOS_DRM_PLANAR_Y] + hsize * vsize; > - } > - > - for_each_ipp_planar(i) > - rot_write(addr[i], ROT_SRC_BUF_ADDR(i)); > - break; > - case IPP_BUF_DEQUEUE: > - for_each_ipp_planar(i) > - rot_write(0x0, ROT_SRC_BUF_ADDR(i)); > - break; > - default: > - /* Nothing to do */ > - break; > - } > + /* Set buffer DMA address */ > + rot_write(buf->dma_addr[0], ROT_SRC_BUF_ADDR(0)); > + rot_write(buf->dma_addr[1], ROT_SRC_BUF_ADDR(1)); > > return 0; > } > > -static int rotator_dst_set_transf(struct device *dev, > - enum drm_exynos_degree degree, > - enum drm_exynos_flip flip, bool *swap) > +static int rotator_dst_set_transf(struct rot_context *rot, > + unsigned int rotation) ditto. > { > - struct rot_context *rot = dev_get_drvdata(dev); > u32 val; > > /* Set transform configuration */ > val = rot_read(ROT_CONTROL); > val &= ~ROT_CONTROL_FLIP_MASK; > > - switch (flip) { > - case EXYNOS_DRM_FLIP_VERTICAL: > - val |= ROT_CONTROL_FLIP_VERTICAL; > - break; > - case EXYNOS_DRM_FLIP_HORIZONTAL: > + if (rotation & DRM_MODE_REFLECT_X) > val |= ROT_CONTROL_FLIP_HORIZONTAL; > - break; > - default: > - /* Flip None */ > - break; > - } > + if (rotation & DRM_MODE_REFLECT_Y) > + val |= ROT_CONTROL_FLIP_VERTICAL; > > val &= ~ROT_CONTROL_ROT_MASK; > > - switch (degree) { > - case EXYNOS_DRM_DEGREE_90: > + if (rotation & DRM_MODE_ROTATE_90) > val |= ROT_CONTROL_ROT_90; > - break; > - case EXYNOS_DRM_DEGREE_180: > + else if (rotation & DRM_MODE_ROTATE_180) > val |= ROT_CONTROL_ROT_180; > - break; > - case EXYNOS_DRM_DEGREE_270: > + else if (rotation & DRM_MODE_ROTATE_270) > val |= ROT_CONTROL_ROT_270; > - break; > - default: > - /* Rotation 0 Degree */ > - break; > - } > > rot_write(val, ROT_CONTROL); > > - /* Check degree for setting buffer size swap */ > - if ((degree == EXYNOS_DRM_DEGREE_90) || > - (degree == EXYNOS_DRM_DEGREE_270)) > - *swap = true; > - else > - *swap = false; > - > return 0; > } > > -static int rotator_dst_set_size(struct device *dev, int swap, > - struct drm_exynos_pos *pos, > - struct drm_exynos_sz *sz) > +static int rotator_dst_set_buf(struct rot_context *rot, > + struct exynos_drm_ipp_buffer *buf) ditto. > { > - struct rot_context *rot = dev_get_drvdata(dev); > - u32 val, fmt, hsize, vsize; > - > - /* Get format */ > - fmt = rotator_reg_get_fmt(rot); > - if (!rotator_check_reg_fmt(fmt)) { > - DRM_ERROR("invalid format.\n"); > - return -EINVAL; > - } > - > - /* Align buffer size */ > - hsize = sz->hsize; > - vsize = sz->vsize; > - rotator_align_size(rot, fmt, &hsize, &vsize); > + u32 val; > > /* Set buffer size configuration */ > - val = ROT_SET_BUF_SIZE_H(vsize) | ROT_SET_BUF_SIZE_W(hsize); > + val = ROT_SET_BUF_SIZE_H(buf->buf.height) | > + ROT_SET_BUF_SIZE_W(buf->buf.pitch[0] / buf->format->cpp[0]); > rot_write(val, ROT_DST_BUF_SIZE); > > /* Set crop image position configuration */ > - val = ROT_CROP_POS_Y(pos->y) | ROT_CROP_POS_X(pos->x); > + val = ROT_CROP_POS_Y(buf->rect.y) | ROT_CROP_POS_X(buf->rect.x); > rot_write(val, ROT_DST_CROP_POS); > > - return 0; > -} > - > -static int rotator_dst_set_addr(struct device *dev, > - struct drm_exynos_ipp_buf_info *buf_info, > - u32 buf_id, enum drm_exynos_ipp_buf_type buf_type) > -{ > - struct rot_context *rot = dev_get_drvdata(dev); > - dma_addr_t addr[EXYNOS_DRM_PLANAR_MAX]; > - u32 val, fmt, hsize, vsize; > - int i; > - > - /* Set current buf_id */ > - rot->cur_buf_id[EXYNOS_DRM_OPS_DST] = buf_id; > - > - switch (buf_type) { > - case IPP_BUF_ENQUEUE: > - /* Set address configuration */ > - for_each_ipp_planar(i) > - addr[i] = buf_info->base[i]; > - > - /* Get format */ > - fmt = rotator_reg_get_fmt(rot); > - if (!rotator_check_reg_fmt(fmt)) { > - DRM_ERROR("invalid format.\n"); > - return -EINVAL; > - } > - > - /* Re-set cb planar for NV12 format */ > - if ((fmt == ROT_CONTROL_FMT_YCBCR420_2P) && > - !addr[EXYNOS_DRM_PLANAR_CB]) { > - /* Get buf size */ > - val = rot_read(ROT_DST_BUF_SIZE); > - > - hsize = ROT_GET_BUF_SIZE_W(val); > - vsize = ROT_GET_BUF_SIZE_H(val); > - > - /* Set cb planar */ > - addr[EXYNOS_DRM_PLANAR_CB] = > - addr[EXYNOS_DRM_PLANAR_Y] + hsize * vsize; > - } > - > - for_each_ipp_planar(i) > - rot_write(addr[i], ROT_DST_BUF_ADDR(i)); > - break; > - case IPP_BUF_DEQUEUE: > - for_each_ipp_planar(i) > - rot_write(0x0, ROT_DST_BUF_ADDR(i)); > - break; > - default: > - /* Nothing to do */ > - break; > - } > + /* Set buffer DMA address */ > + rot_write(buf->dma_addr[0], ROT_DST_BUF_ADDR(0)); > + rot_write(buf->dma_addr[1], ROT_DST_BUF_ADDR(1)); > > return 0; > } > > -static struct exynos_drm_ipp_ops rot_src_ops = { > - .set_fmt = rotator_src_set_fmt, > - .set_size = rotator_src_set_size, > - .set_addr = rotator_src_set_addr, > -}; > +static int rotator_start(struct rot_context *rot) ditto. > +{ > + u32 val; > > -static struct exynos_drm_ipp_ops rot_dst_ops = { > - .set_transf = rotator_dst_set_transf, > - .set_size = rotator_dst_set_size, > - .set_addr = rotator_dst_set_addr, > -}; > + /* Set interrupt enable */ > + rotator_reg_set_irq(rot, true); > > -static int rotator_init_prop_list(struct exynos_drm_ippdrv *ippdrv) > -{ > - struct drm_exynos_ipp_prop_list *prop_list = &ippdrv->prop_list; > - > - prop_list->version = 1; > - prop_list->flip = (1 << EXYNOS_DRM_FLIP_VERTICAL) | > - (1 << EXYNOS_DRM_FLIP_HORIZONTAL); > - prop_list->degree = (1 << EXYNOS_DRM_DEGREE_0) | > - (1 << EXYNOS_DRM_DEGREE_90) | > - (1 << EXYNOS_DRM_DEGREE_180) | > - (1 << EXYNOS_DRM_DEGREE_270); > - prop_list->csc = 0; > - prop_list->crop = 0; > - prop_list->scale = 0; > + val = rot_read(ROT_CONTROL); > + val |= ROT_CONTROL_START; > + rot_write(val, ROT_CONTROL); > > return 0; > } > > -static inline bool rotator_check_drm_fmt(u32 fmt) > +static int rotator_commit(struct exynos_drm_ipp *ipp, > + struct exynos_drm_ipp_task *task) > { > - switch (fmt) { > - case DRM_FORMAT_XRGB8888: > - case DRM_FORMAT_NV12: > - return true; > - default: > - DRM_DEBUG_KMS("not support format\n"); > - return false; > - } > -} > + struct rot_context *rot = > + container_of(ipp, struct rot_context, ipp); > > -static inline bool rotator_check_drm_flip(enum drm_exynos_flip flip) > -{ > - switch (flip) { > - case EXYNOS_DRM_FLIP_NONE: > - case EXYNOS_DRM_FLIP_VERTICAL: > - case EXYNOS_DRM_FLIP_HORIZONTAL: > - case EXYNOS_DRM_FLIP_BOTH: > - return true; > - default: > - DRM_DEBUG_KMS("invalid flip\n"); > - return false; > - } > -} > + pm_runtime_get_sync(rot->dev); > + rot->task = task; > > -static int rotator_ippdrv_check_property(struct device *dev, > - struct drm_exynos_ipp_property *property) > -{ > - struct drm_exynos_ipp_config *src_config = > - &property->config[EXYNOS_DRM_OPS_SRC]; > - struct drm_exynos_ipp_config *dst_config = > - &property->config[EXYNOS_DRM_OPS_DST]; > - struct drm_exynos_pos *src_pos = &src_config->pos; > - struct drm_exynos_pos *dst_pos = &dst_config->pos; > - struct drm_exynos_sz *src_sz = &src_config->sz; > - struct drm_exynos_sz *dst_sz = &dst_config->sz; > - bool swap = false; > - > - /* Check format configuration */ > - if (src_config->fmt != dst_config->fmt) { > - DRM_DEBUG_KMS("not support csc feature\n"); > - return -EINVAL; > - } > - > - if (!rotator_check_drm_fmt(dst_config->fmt)) { > - DRM_DEBUG_KMS("invalid format\n"); > - return -EINVAL; > - } > - > - /* Check transform configuration */ > - if (src_config->degree != EXYNOS_DRM_DEGREE_0) { > - DRM_DEBUG_KMS("not support source-side rotation\n"); > - return -EINVAL; > - } > - > - switch (dst_config->degree) { > - case EXYNOS_DRM_DEGREE_90: > - case EXYNOS_DRM_DEGREE_270: > - swap = true; > - case EXYNOS_DRM_DEGREE_0: > - case EXYNOS_DRM_DEGREE_180: > - /* No problem */ > - break; > - default: > - DRM_DEBUG_KMS("invalid degree\n"); > - return -EINVAL; > - } > - > - if (src_config->flip != EXYNOS_DRM_FLIP_NONE) { > - DRM_DEBUG_KMS("not support source-side flip\n"); > - return -EINVAL; > - } > - > - if (!rotator_check_drm_flip(dst_config->flip)) { > - DRM_DEBUG_KMS("invalid flip\n"); > - return -EINVAL; > - } > - > - /* Check size configuration */ > - if ((src_pos->x + src_pos->w > src_sz->hsize) || > - (src_pos->y + src_pos->h > src_sz->vsize)) { > - DRM_DEBUG_KMS("out of source buffer bound\n"); > - return -EINVAL; > - } > - > - if (swap) { > - if ((dst_pos->x + dst_pos->h > dst_sz->vsize) || > - (dst_pos->y + dst_pos->w > dst_sz->hsize)) { > - DRM_DEBUG_KMS("out of destination buffer bound\n"); > - return -EINVAL; > - } > - > - if ((src_pos->w != dst_pos->h) || (src_pos->h != dst_pos->w)) { > - DRM_DEBUG_KMS("not support scale feature\n"); > - return -EINVAL; > - } > - } else { > - if ((dst_pos->x + dst_pos->w > dst_sz->hsize) || > - (dst_pos->y + dst_pos->h > dst_sz->vsize)) { > - DRM_DEBUG_KMS("out of destination buffer bound\n"); > - return -EINVAL; > - } > - > - if ((src_pos->w != dst_pos->w) || (src_pos->h != dst_pos->h)) { > - DRM_DEBUG_KMS("not support scale feature\n"); > - return -EINVAL; > - } > - } > + rotator_src_set_fmt(rot, task->src.buf.fourcc); > + rotator_src_set_buf(rot, &task->src); > + rotator_dst_set_transf(rot, task->transform.rotation); > + rotator_dst_set_buf(rot, &task->dst); > + rotator_start(rot); > > return 0; > } > > -static int rotator_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd) > +static const struct exynos_drm_ipp_funcs ipp_funcs = { > + .commit = rotator_commit, > +}; > + > +static int rotator_bind(struct device *dev, struct device *master, void *data) > { > struct rot_context *rot = dev_get_drvdata(dev); > - u32 val; > - > - if (rot->suspended) { > - DRM_ERROR("suspended state\n"); > - return -EPERM; > - } > - > - if (cmd != IPP_CMD_M2M) { > - DRM_ERROR("not support cmd: %d\n", cmd); > - return -EINVAL; > - } > + struct drm_device *drm_dev = data; > + struct exynos_drm_ipp *ipp = &rot->ipp; > > - /* Set interrupt enable */ > - rotator_reg_set_irq(rot, true); > + rot->drm_dev = drm_dev; > + drm_iommu_attach_device(drm_dev, dev); > > - val = rot_read(ROT_CONTROL); > - val |= ROT_CONTROL_START; > + exynos_drm_ipp_register(drm_dev, ipp, &ipp_funcs, > + DRM_EXYNOS_IPP_CAP_CROP | DRM_EXYNOS_IPP_CAP_ROTATE, > + rot->formats, "rotator"); > > - rot_write(val, ROT_CONTROL); > + dev_info(dev, "The exynos rotator has been probed successfully\n"); > > return 0; > } > > -static struct rot_limit_table rot_limit_tbl_4210 = { > - .ycbcr420_2p = { > - .min_w = 32, > - .min_h = 32, > - .max_w = SZ_64K, > - .max_h = SZ_64K, > - .align = 3, > - }, > - .rgb888 = { > - .min_w = 8, > - .min_h = 8, > - .max_w = SZ_16K, > - .max_h = SZ_16K, > - .align = 2, > - }, > -}; > +static void rotator_unbind(struct device *dev, struct device *master, > + void *data) > +{ > + struct rot_context *rot = dev_get_drvdata(dev); > + struct drm_device *drm_dev = data; > + struct exynos_drm_ipp *ipp = &rot->ipp; > > -static struct rot_limit_table rot_limit_tbl_4x12 = { > - .ycbcr420_2p = { > - .min_w = 32, > - .min_h = 32, > - .max_w = SZ_32K, > - .max_h = SZ_32K, > - .align = 3, > - }, > - .rgb888 = { > - .min_w = 8, > - .min_h = 8, > - .max_w = SZ_8K, > - .max_h = SZ_8K, > - .align = 2, > - }, > -}; > + exynos_drm_ipp_unregister(drm_dev, ipp); > + drm_iommu_detach_device(rot->drm_dev, rot->dev); > +} > > -static struct rot_limit_table rot_limit_tbl_5250 = { > - .ycbcr420_2p = { > - .min_w = 32, > - .min_h = 32, > - .max_w = SZ_32K, > - .max_h = SZ_32K, > - .align = 3, > - }, > - .rgb888 = { > - .min_w = 8, > - .min_h = 8, > - .max_w = SZ_8K, > - .max_h = SZ_8K, > - .align = 1, > - }, > +static const struct component_ops rotator_component_ops = { > + .bind = rotator_bind, > + .unbind = rotator_unbind, > }; > > -static const struct of_device_id exynos_rotator_match[] = { > - { > - .compatible = "samsung,exynos4210-rotator", > - .data = &rot_limit_tbl_4210, > - }, > - { > - .compatible = "samsung,exynos4212-rotator", > - .data = &rot_limit_tbl_4x12, > - }, > - { > - .compatible = "samsung,exynos5250-rotator", > - .data = &rot_limit_tbl_5250, > - }, > - {}, > -}; > -MODULE_DEVICE_TABLE(of, exynos_rotator_match); > - > static int rotator_probe(struct platform_device *pdev) > { > struct device *dev = &pdev->dev; > + struct resource *regs_res; > struct rot_context *rot; > - struct exynos_drm_ippdrv *ippdrv; > + int irq; > int ret; > > - if (!dev->of_node) { > - dev_err(dev, "cannot find of_node.\n"); > - return -ENODEV; > - } > - > rot = devm_kzalloc(dev, sizeof(*rot), GFP_KERNEL); > if (!rot) > return -ENOMEM; > > - rot->limit_tbl = (struct rot_limit_table *) > - of_device_get_match_data(dev); > - rot->regs_res = platform_get_resource(pdev, IORESOURCE_MEM, 0); > - rot->regs = devm_ioremap_resource(dev, rot->regs_res); > + rot->formats = of_device_get_match_data(dev); > + rot->dev = dev; > + regs_res = platform_get_resource(pdev, IORESOURCE_MEM, 0); > + rot->regs = devm_ioremap_resource(dev, regs_res); > if (IS_ERR(rot->regs)) > return PTR_ERR(rot->regs); > > - rot->irq = platform_get_irq(pdev, 0); > - if (rot->irq < 0) { > + irq = platform_get_irq(pdev, 0); > + if (irq < 0) { > dev_err(dev, "failed to get irq\n"); > - return rot->irq; > + return irq; > } > > - ret = devm_request_threaded_irq(dev, rot->irq, NULL, > - rotator_irq_handler, IRQF_ONESHOT, "drm_rotator", rot); > + ret = devm_request_irq(dev, irq, rotator_irq_handler, 0, dev_name(dev), > + rot); > if (ret < 0) { > dev_err(dev, "failed to request irq\n"); > return ret; > @@ -734,35 +313,19 @@ static int rotator_probe(struct platform_device *pdev) > return PTR_ERR(rot->clock); > } > > + pm_runtime_use_autosuspend(dev); > + pm_runtime_set_autosuspend_delay(dev, ROTATOR_AUTOSUSPEND_DELAY); > pm_runtime_enable(dev); > - > - ippdrv = &rot->ippdrv; > - ippdrv->dev = dev; > - ippdrv->ops[EXYNOS_DRM_OPS_SRC] = &rot_src_ops; > - ippdrv->ops[EXYNOS_DRM_OPS_DST] = &rot_dst_ops; > - ippdrv->check_property = rotator_ippdrv_check_property; > - ippdrv->start = rotator_ippdrv_start; > - ret = rotator_init_prop_list(ippdrv); > - if (ret < 0) { > - dev_err(dev, "failed to init property list.\n"); > - goto err_ippdrv_register; > - } > - > - DRM_DEBUG_KMS("ippdrv[%pK]\n", ippdrv); > - > platform_set_drvdata(pdev, rot); > > - ret = exynos_drm_ippdrv_register(ippdrv); > - if (ret < 0) { > - dev_err(dev, "failed to register drm rotator device\n"); > - goto err_ippdrv_register; > - } > - > - dev_info(dev, "The exynos rotator is probed successfully\n"); > + ret = component_add(dev, &rotator_component_ops); > + if (ret) > + goto err_component; > > return 0; > > -err_ippdrv_register: > +err_component: > + pm_runtime_dont_use_autosuspend(dev); > pm_runtime_disable(dev); > return ret; > } > @@ -770,45 +333,94 @@ static int rotator_probe(struct platform_device *pdev) > static int rotator_remove(struct platform_device *pdev) > { > struct device *dev = &pdev->dev; > - struct rot_context *rot = dev_get_drvdata(dev); > - struct exynos_drm_ippdrv *ippdrv = &rot->ippdrv; > - > - exynos_drm_ippdrv_unregister(ippdrv); > > + component_del(dev, &rotator_component_ops); > + pm_runtime_dont_use_autosuspend(dev); > pm_runtime_disable(dev); > > return 0; > } > > #ifdef CONFIG_PM > -static int rotator_clk_crtl(struct rot_context *rot, bool enable) > -{ > - if (enable) { > - clk_prepare_enable(rot->clock); > - rot->suspended = false; > - } else { > - clk_disable_unprepare(rot->clock); > - rot->suspended = true; > - } > - > - return 0; > -} > - > static int rotator_runtime_suspend(struct device *dev) > { > struct rot_context *rot = dev_get_drvdata(dev); > > - return rotator_clk_crtl(rot, false); > + clk_disable_unprepare(rot->clock); > + return 0; > } > > static int rotator_runtime_resume(struct device *dev) > { > struct rot_context *rot = dev_get_drvdata(dev); > > - return rotator_clk_crtl(rot, true); > + return clk_prepare_enable(rot->clock); > } > #endif > > +static const struct drm_exynos_ipp_limit rotator_4210_rbg888_limits[] = { > + { IPP_SIZE_LIMIT(BUFFER, .h = {8, SZ_16K}, .v = {8, SZ_16K}) }, Leave a space brace and constant value, and below declarations also. Thanks, Inki Dae > + { IPP_SIZE_LIMIT(AREA, .h.align = 4, .v.align = 4) }, > + { } > +}; > + > +static const struct drm_exynos_ipp_limit rotator_4412_rbg888_limits[] = { > + { IPP_SIZE_LIMIT(BUFFER, .h = {8, SZ_8K}, .v = {8, SZ_8K}) }, > + { IPP_SIZE_LIMIT(AREA, .h.align = 4, .v.align = 4) }, > + { } > +}; > + > +static const struct drm_exynos_ipp_limit rotator_5250_rbg888_limits[] = { > + { IPP_SIZE_LIMIT(BUFFER, .h = {8, SZ_8K}, .v = {8, SZ_8K}) }, > + { IPP_SIZE_LIMIT(AREA, .h.align = 2, .v.align = 2) }, > + { } > +}; > + > +static const struct drm_exynos_ipp_limit rotator_4210_yuv_limits[] = { > + { IPP_SIZE_LIMIT(BUFFER, .h = {32, SZ_64K}, .v = {32, SZ_64K}) }, > + { IPP_SIZE_LIMIT(AREA, .h.align = 8, .v.align = 8) }, > + { } > +}; > + > +static const struct drm_exynos_ipp_limit rotator_4412_yuv_limits[] = { > + { IPP_SIZE_LIMIT(BUFFER, .h = {32, SZ_32K}, .v = {32, SZ_32K}) }, > + { IPP_SIZE_LIMIT(AREA, .h.align = 8, .v.align = 8) }, > + { } > +}; > + > +static const struct exynos_drm_ipp_formats rotator_4210_formats[] = { > + { IPP_SRCDST_FORMAT(XRGB8888, rotator_4210_rbg888_limits) }, > + { IPP_SRCDST_FORMAT(NV12, rotator_4210_yuv_limits) }, > + { } > +}; > + > +static const struct exynos_drm_ipp_formats rotator_4412_formats[] = { > + { IPP_SRCDST_FORMAT(XRGB8888, rotator_4412_rbg888_limits) }, > + { IPP_SRCDST_FORMAT(NV12, rotator_4412_yuv_limits) }, > + { } > +}; > + > +static const struct exynos_drm_ipp_formats rotator_5250_formats[] = { > + { IPP_SRCDST_FORMAT(XRGB8888, rotator_5250_rbg888_limits) }, > + { IPP_SRCDST_FORMAT(NV12, rotator_4412_yuv_limits) }, > + { } > +}; > + > +static const struct of_device_id exynos_rotator_match[] = { > + { > + .compatible = "samsung,exynos4210-rotator", > + .data = rotator_4210_formats, > + }, { > + .compatible = "samsung,exynos4212-rotator", > + .data = rotator_4412_formats, > + }, { > + .compatible = "samsung,exynos5250-rotator", > + .data = rotator_5250_formats, > + }, { > + }, > +}; > +MODULE_DEVICE_TABLE(of, exynos_rotator_match); > + > static const struct dev_pm_ops rotator_pm_ops = { > SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend, > pm_runtime_force_resume) > @@ -820,7 +432,7 @@ struct platform_driver rotator_driver = { > .probe = rotator_probe, > .remove = rotator_remove, > .driver = { > - .name = "exynos-rot", > + .name = "exynos-rotator", > .owner = THIS_MODULE, > .pm = &rotator_pm_ops, > .of_match_table = exynos_rotator_match, > diff --git a/drivers/gpu/drm/exynos/exynos_drm_rotator.h b/drivers/gpu/drm/exynos/exynos_drm_rotator.h > deleted file mode 100644 > index 71a0b4c0c1e8..000000000000 > --- a/drivers/gpu/drm/exynos/exynos_drm_rotator.h > +++ /dev/null > @@ -1,19 +0,0 @@ > -/* > - * Copyright (c) 2012 Samsung Electronics Co., Ltd. > - * > - * Authors: > - * YoungJun Cho <yj44.cho@samsung.com> > - * Eunchul Kim <chulspro.kim@samsung.com> > - * > - * This program is free software; you can redistribute it and/or modify it > - * under the terms of the GNU General Public License as published by the > - * Free Software Foundation; either version 2 of the License, or (at your > - * option) any later version. > - */ > - > -#ifndef _EXYNOS_DRM_ROTATOR_H_ > -#define _EXYNOS_DRM_ROTATOR_H_ > - > -/* TODO */ > - > -#endif > _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel ^ permalink raw reply [flat|nested] 20+ messages in thread
[parent not found: <CGME20171023075450eucas1p155afda476e368ae57ba7fe1a347a5f4c@eucas1p1.samsung.com>]
* [PATCH v4 4/9] drm/exynos: gsc: Convert driver to IPP v2 core API [not found] ` <CGME20171023075450eucas1p155afda476e368ae57ba7fe1a347a5f4c@eucas1p1.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski 2017-11-01 6:02 ` Inki Dae 0 siblings, 1 reply; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Marek Szyprowski, Inki Dae, Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon This patch adapts Exynos DRM GScaler driver to new IPP v2 core API. The side effect of this conversion is a switch to driver component API to register properly in the Exynos DRM core. During the conversion driver has been adapted to support more specific compatible strings to distinguish between Exynos5250 and Exynos5420 (different hardware limits). Support for Exynos5433 variant has been added too (different limits table, removed dependency on ARCH_EXYNOS5). Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Hoegeun Kwon <hoegeun.kwon@samsung.com> --- drivers/gpu/drm/exynos/Kconfig | 3 +- drivers/gpu/drm/exynos/exynos_drm_drv.c | 1 + drivers/gpu/drm/exynos/exynos_drm_gsc.c | 1001 ++++++++++--------------------- drivers/gpu/drm/exynos/exynos_drm_gsc.h | 24 - 4 files changed, 321 insertions(+), 708 deletions(-) delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_gsc.h diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig index c10c9ca0d8b4..9b66a9838dca 100644 --- a/drivers/gpu/drm/exynos/Kconfig +++ b/drivers/gpu/drm/exynos/Kconfig @@ -111,7 +111,8 @@ config DRM_EXYNOS_ROTATOR config DRM_EXYNOS_GSC bool "GScaler" - depends on BROKEN && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n + depends on VIDEO_SAMSUNG_EXYNOS_GSC=n + select DRM_EXYNOS_IPP help Choose this option if you want to use Exynos GSC for DRM. diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c index 3ade2b0ad15d..cac0d84385d3 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c @@ -260,6 +260,7 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { DRM_COMPONENT_DRIVER }, { DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC), + DRM_COMPONENT_DRIVER }, { &exynos_drm_platform_driver, DRM_VIRTUAL_DEVICE diff --git a/drivers/gpu/drm/exynos/exynos_drm_gsc.c b/drivers/gpu/drm/exynos/exynos_drm_gsc.c index 0506b2b17ac1..4671ac41469d 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_gsc.c +++ b/drivers/gpu/drm/exynos/exynos_drm_gsc.c @@ -12,18 +12,20 @@ * */ #include <linux/kernel.h> +#include <linux/component.h> #include <linux/platform_device.h> #include <linux/clk.h> #include <linux/pm_runtime.h> #include <linux/mfd/syscon.h> +#include <linux/of_device.h> #include <linux/regmap.h> #include <drm/drmP.h> #include <drm/exynos_drm.h> #include "regs-gsc.h" #include "exynos_drm_drv.h" +#include "exynos_drm_iommu.h" #include "exynos_drm_ipp.h" -#include "exynos_drm_gsc.h" /* * GSC stands for General SCaler and @@ -31,26 +33,10 @@ * input DMA reads image data from the memory. * output DMA writes image data to memory. * GSC supports image rotation and image effect functions. - * - * M2M operation : supports crop/scale/rotation/csc so on. - * Memory ----> GSC H/W ----> Memory. - * Writeback operation : supports cloned screen with FIMD. - * FIMD ----> GSC H/W ----> Memory. - * Output operation : supports direct display using local path. - * Memory ----> GSC H/W ----> FIMD, Mixer. */ -/* - * TODO - * 1. check suspend/resume api if needed. - * 2. need to check use case platform_device_id. - * 3. check src/dst size with, height. - * 4. added check_prepare api for right register. - * 5. need to add supported list in prop_list. - * 6. check prescaler/scaler optimization. - */ -#define GSC_MAX_DEVS 4 +#define GSC_MAX_CLOCKS 8 #define GSC_MAX_SRC 4 #define GSC_MAX_DST 16 #define GSC_RESET_TIMEOUT 50 @@ -65,8 +51,6 @@ #define GSC_SC_DOWN_RATIO_4_8 131072 #define GSC_SC_DOWN_RATIO_3_8 174762 #define GSC_SC_DOWN_RATIO_2_8 262144 -#define GSC_REFRESH_MIN 12 -#define GSC_REFRESH_MAX 60 #define GSC_CROP_MAX 8192 #define GSC_CROP_MIN 32 #define GSC_SCALE_MAX 4224 @@ -77,10 +61,9 @@ #define GSC_COEF_H_8T 8 #define GSC_COEF_V_4T 4 #define GSC_COEF_DEPTH 3 +#define GSC_AUTOSUSPEND_DELAY 2000 #define get_gsc_context(dev) platform_get_drvdata(to_platform_device(dev)) -#define get_ctx_from_ippdrv(ippdrv) container_of(ippdrv,\ - struct gsc_context, ippdrv); #define gsc_read(offset) readl(ctx->regs + (offset)) #define gsc_write(cfg, offset) writel(cfg, ctx->regs + (offset)) @@ -124,7 +107,6 @@ struct gsc_capability { /* * A structure of gsc context. * - * @ippdrv: prepare initialization using ippdrv. * @regs_res: register resources. * @regs: memory mapped io registers. * @sysreg: handle to SYSREG block regmap. @@ -137,12 +119,18 @@ struct gsc_capability { * @suspended: qos operations. */ struct gsc_context { - struct exynos_drm_ippdrv ippdrv; + struct exynos_drm_ipp ipp; + struct drm_device *drm_dev; + struct device *dev; + struct exynos_drm_ipp_task *task; + struct exynos_drm_ipp_formats *formats; + struct resource *regs_res; void __iomem *regs; struct regmap *sysreg; - struct mutex lock; - struct clk *gsc_clk; + const char **clk_names; + struct clk *clocks[GSC_MAX_CLOCKS]; + int num_clocks; struct gsc_scaler sc; int id; int irq; @@ -150,6 +138,19 @@ struct gsc_context { bool suspended; }; +/** + * struct gsc_driverdata - per device type driver data for init time. + * + * @limits: picture size limits array + * @clk_names: names of clocks needed by this variant + * @num_clocks: the number of clocks needed by this variant + */ +struct gsc_driverdata { + const struct drm_exynos_ipp_limit *limits; + const char *clk_names[GSC_MAX_CLOCKS]; + int num_clocks; +}; + /* 8-tap Filter Coefficient */ static const int h_coef_8t[GSC_COEF_RATIO][GSC_COEF_ATTR][GSC_COEF_H_8T] = { { /* Ratio <= 65536 (~8:8) */ @@ -438,25 +439,6 @@ static int gsc_sw_reset(struct gsc_context *ctx) return 0; } -static void gsc_set_gscblk_fimd_wb(struct gsc_context *ctx, bool enable) -{ - unsigned int gscblk_cfg; - - if (!ctx->sysreg) - return; - - regmap_read(ctx->sysreg, SYSREG_GSCBLK_CFG1, &gscblk_cfg); - - if (enable) - gscblk_cfg |= GSC_BLK_DISP1WB_DEST(ctx->id) | - GSC_BLK_GSCL_WB_IN_SRC_SEL(ctx->id) | - GSC_BLK_SW_RESET_WB_DEST(ctx->id); - else - gscblk_cfg |= GSC_BLK_PXLASYNC_LO_MASK_WB(ctx->id); - - regmap_write(ctx->sysreg, SYSREG_GSCBLK_CFG1, gscblk_cfg); -} - static void gsc_handle_irq(struct gsc_context *ctx, bool enable, bool overflow, bool done) { @@ -487,10 +469,8 @@ static void gsc_handle_irq(struct gsc_context *ctx, bool enable, } -static int gsc_src_set_fmt(struct device *dev, u32 fmt) +static int gsc_src_set_fmt(struct gsc_context *ctx, u32 fmt) { - struct gsc_context *ctx = get_gsc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; u32 cfg; DRM_DEBUG_KMS("fmt[0x%x]\n", fmt); @@ -549,7 +529,7 @@ static int gsc_src_set_fmt(struct device *dev, u32 fmt) GSC_IN_YUV420_2P); break; default: - dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt); + dev_err(ctx->dev, "invalid target yuv order 0x%x.\n", fmt); return -EINVAL; } @@ -558,105 +538,81 @@ static int gsc_src_set_fmt(struct device *dev, u32 fmt) return 0; } -static int gsc_src_set_transf(struct device *dev, - enum drm_exynos_degree degree, - enum drm_exynos_flip flip, bool *swap) +static int gsc_src_set_transf(struct gsc_context *ctx, unsigned int rotation) { - struct gsc_context *ctx = get_gsc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; + unsigned int degree = rotation & DRM_MODE_ROTATE_MASK; u32 cfg; - DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip); - cfg = gsc_read(GSC_IN_CON); cfg &= ~GSC_IN_ROT_MASK; switch (degree) { - case EXYNOS_DRM_DEGREE_0: - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + case DRM_MODE_ROTATE_0: + if (rotation & DRM_MODE_REFLECT_Y) cfg |= GSC_IN_ROT_XFLIP; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_X) cfg |= GSC_IN_ROT_YFLIP; break; - case EXYNOS_DRM_DEGREE_90: - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + case DRM_MODE_ROTATE_90: + if (rotation & DRM_MODE_REFLECT_Y) cfg |= GSC_IN_ROT_90_XFLIP; - else if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + else if (rotation & DRM_MODE_REFLECT_X) cfg |= GSC_IN_ROT_90_YFLIP; else cfg |= GSC_IN_ROT_90; break; - case EXYNOS_DRM_DEGREE_180: + case DRM_MODE_ROTATE_180: cfg |= GSC_IN_ROT_180; - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg &= ~GSC_IN_ROT_XFLIP; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_X) cfg &= ~GSC_IN_ROT_YFLIP; break; - case EXYNOS_DRM_DEGREE_270: + case DRM_MODE_ROTATE_270: cfg |= GSC_IN_ROT_270; - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg &= ~GSC_IN_ROT_XFLIP; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_X) cfg &= ~GSC_IN_ROT_YFLIP; break; - default: - dev_err(ippdrv->dev, "invalid degree value %d.\n", degree); - return -EINVAL; } gsc_write(cfg, GSC_IN_CON); ctx->rotation = (cfg & GSC_IN_ROT_90) ? 1 : 0; - *swap = ctx->rotation; - return 0; } -static int gsc_src_set_size(struct device *dev, int swap, - struct drm_exynos_pos *pos, struct drm_exynos_sz *sz) +static int gsc_src_set_size(struct gsc_context *ctx, + struct exynos_drm_ipp_buffer *buf) { - struct gsc_context *ctx = get_gsc_context(dev); - struct drm_exynos_pos img_pos = *pos; struct gsc_scaler *sc = &ctx->sc; u32 cfg; - DRM_DEBUG_KMS("swap[%d]x[%d]y[%d]w[%d]h[%d]\n", - swap, pos->x, pos->y, pos->w, pos->h); - - if (swap) { - img_pos.w = pos->h; - img_pos.h = pos->w; - } - /* pixel offset */ - cfg = (GSC_SRCIMG_OFFSET_X(img_pos.x) | - GSC_SRCIMG_OFFSET_Y(img_pos.y)); + cfg = (GSC_SRCIMG_OFFSET_X(buf->rect.x) | + GSC_SRCIMG_OFFSET_Y(buf->rect.y)); gsc_write(cfg, GSC_SRCIMG_OFFSET); /* cropped size */ - cfg = (GSC_CROPPED_WIDTH(img_pos.w) | - GSC_CROPPED_HEIGHT(img_pos.h)); + cfg = (GSC_CROPPED_WIDTH(buf->rect.w) | + GSC_CROPPED_HEIGHT(buf->rect.h)); gsc_write(cfg, GSC_CROPPED_SIZE); - DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", sz->hsize, sz->vsize); - /* original size */ cfg = gsc_read(GSC_SRCIMG_SIZE); cfg &= ~(GSC_SRCIMG_HEIGHT_MASK | GSC_SRCIMG_WIDTH_MASK); - cfg |= (GSC_SRCIMG_WIDTH(sz->hsize) | - GSC_SRCIMG_HEIGHT(sz->vsize)); + cfg |= (GSC_SRCIMG_WIDTH(buf->buf.width) | + GSC_SRCIMG_HEIGHT(buf->buf.height)); gsc_write(cfg, GSC_SRCIMG_SIZE); cfg = gsc_read(GSC_IN_CON); cfg &= ~GSC_IN_RGB_TYPE_MASK; - DRM_DEBUG_KMS("width[%d]range[%d]\n", pos->w, sc->range); - - if (pos->w >= GSC_WIDTH_ITU_709) + if (buf->rect.w >= GSC_WIDTH_ITU_709) if (sc->range) cfg |= GSC_IN_RGB_HD_WIDE; else @@ -673,30 +629,15 @@ static int gsc_src_set_size(struct device *dev, int swap, } static int gsc_src_set_buf_seq(struct gsc_context *ctx, u32 buf_id, - enum drm_exynos_ipp_buf_type buf_type) + bool enqueue) { - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - bool masked; + bool masked = !enqueue; u32 cfg; u32 mask = 0x00000001 << buf_id; - DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type); - /* mask register set */ cfg = gsc_read(GSC_IN_BASE_ADDR_Y_MASK); - switch (buf_type) { - case IPP_BUF_ENQUEUE: - masked = false; - break; - case IPP_BUF_DEQUEUE: - masked = true; - break; - default: - dev_err(ippdrv->dev, "invalid buf ctrl parameter.\n"); - return -EINVAL; - } - /* sequence id */ cfg &= ~mask; cfg |= masked << buf_id; @@ -707,64 +648,19 @@ static int gsc_src_set_buf_seq(struct gsc_context *ctx, u32 buf_id, return 0; } -static int gsc_src_set_addr(struct device *dev, - struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id, - enum drm_exynos_ipp_buf_type buf_type) +static int gsc_src_set_addr(struct gsc_context *ctx, u32 buf_id, + struct exynos_drm_ipp_buffer *buf) { - struct gsc_context *ctx = get_gsc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; - struct drm_exynos_ipp_property *property; - - if (!c_node) { - DRM_ERROR("failed to get c_node.\n"); - return -EFAULT; - } - - property = &c_node->property; - - DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n", - property->prop_id, buf_id, buf_type); - - if (buf_id > GSC_MAX_SRC) { - dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id); - return -EINVAL; - } - /* address register set */ - switch (buf_type) { - case IPP_BUF_ENQUEUE: - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_Y], - GSC_IN_BASE_ADDR_Y(buf_id)); - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CB], - GSC_IN_BASE_ADDR_CB(buf_id)); - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CR], - GSC_IN_BASE_ADDR_CR(buf_id)); - break; - case IPP_BUF_DEQUEUE: - gsc_write(0x0, GSC_IN_BASE_ADDR_Y(buf_id)); - gsc_write(0x0, GSC_IN_BASE_ADDR_CB(buf_id)); - gsc_write(0x0, GSC_IN_BASE_ADDR_CR(buf_id)); - break; - default: - /* bypass */ - break; - } + gsc_write(buf->dma_addr[0], GSC_IN_BASE_ADDR_Y(buf_id)); + gsc_write(buf->dma_addr[1], GSC_IN_BASE_ADDR_CB(buf_id)); + gsc_write(buf->dma_addr[2], GSC_IN_BASE_ADDR_CR(buf_id)); - return gsc_src_set_buf_seq(ctx, buf_id, buf_type); + return gsc_src_set_buf_seq(ctx, buf_id, true); } -static struct exynos_drm_ipp_ops gsc_src_ops = { - .set_fmt = gsc_src_set_fmt, - .set_transf = gsc_src_set_transf, - .set_size = gsc_src_set_size, - .set_addr = gsc_src_set_addr, -}; - -static int gsc_dst_set_fmt(struct device *dev, u32 fmt) +static int gsc_dst_set_fmt(struct gsc_context *ctx, u32 fmt) { - struct gsc_context *ctx = get_gsc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; u32 cfg; DRM_DEBUG_KMS("fmt[0x%x]\n", fmt); @@ -819,9 +715,6 @@ static int gsc_dst_set_fmt(struct device *dev, u32 fmt) cfg |= (GSC_OUT_CHROMA_ORDER_CBCR | GSC_OUT_YUV420_2P); break; - default: - dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt); - return -EINVAL; } gsc_write(cfg, GSC_OUT_CON); @@ -829,61 +722,6 @@ static int gsc_dst_set_fmt(struct device *dev, u32 fmt) return 0; } -static int gsc_dst_set_transf(struct device *dev, - enum drm_exynos_degree degree, - enum drm_exynos_flip flip, bool *swap) -{ - struct gsc_context *ctx = get_gsc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - u32 cfg; - - DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip); - - cfg = gsc_read(GSC_IN_CON); - cfg &= ~GSC_IN_ROT_MASK; - - switch (degree) { - case EXYNOS_DRM_DEGREE_0: - if (flip & EXYNOS_DRM_FLIP_VERTICAL) - cfg |= GSC_IN_ROT_XFLIP; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) - cfg |= GSC_IN_ROT_YFLIP; - break; - case EXYNOS_DRM_DEGREE_90: - if (flip & EXYNOS_DRM_FLIP_VERTICAL) - cfg |= GSC_IN_ROT_90_XFLIP; - else if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) - cfg |= GSC_IN_ROT_90_YFLIP; - else - cfg |= GSC_IN_ROT_90; - break; - case EXYNOS_DRM_DEGREE_180: - cfg |= GSC_IN_ROT_180; - if (flip & EXYNOS_DRM_FLIP_VERTICAL) - cfg &= ~GSC_IN_ROT_XFLIP; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) - cfg &= ~GSC_IN_ROT_YFLIP; - break; - case EXYNOS_DRM_DEGREE_270: - cfg |= GSC_IN_ROT_270; - if (flip & EXYNOS_DRM_FLIP_VERTICAL) - cfg &= ~GSC_IN_ROT_XFLIP; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) - cfg &= ~GSC_IN_ROT_YFLIP; - break; - default: - dev_err(ippdrv->dev, "invalid degree value %d.\n", degree); - return -EINVAL; - } - - gsc_write(cfg, GSC_IN_CON); - - ctx->rotation = (cfg & GSC_IN_ROT_90) ? 1 : 0; - *swap = ctx->rotation; - - return 0; -} - static int gsc_get_ratio_shift(u32 src, u32 dst, u32 *ratio) { DRM_DEBUG_KMS("src[%d]dst[%d]\n", src, dst); @@ -919,9 +757,8 @@ static void gsc_get_prescaler_shfactor(u32 hratio, u32 vratio, u32 *shfactor) } static int gsc_set_prescaler(struct gsc_context *ctx, struct gsc_scaler *sc, - struct drm_exynos_pos *src, struct drm_exynos_pos *dst) + struct drm_exynos_ipp_task_rect *src, struct drm_exynos_ipp_task_rect *dst) { - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; u32 cfg; u32 src_w, src_h, dst_w, dst_h; int ret = 0; @@ -939,13 +776,13 @@ static int gsc_set_prescaler(struct gsc_context *ctx, struct gsc_scaler *sc, ret = gsc_get_ratio_shift(src_w, dst_w, &sc->pre_hratio); if (ret) { - dev_err(ippdrv->dev, "failed to get ratio horizontal.\n"); + dev_err(ctx->dev, "failed to get ratio horizontal.\n"); return ret; } ret = gsc_get_ratio_shift(src_h, dst_h, &sc->pre_vratio); if (ret) { - dev_err(ippdrv->dev, "failed to get ratio vertical.\n"); + dev_err(ctx->dev, "failed to get ratio vertical.\n"); return ret; } @@ -1039,47 +876,37 @@ static void gsc_set_scaler(struct gsc_context *ctx, struct gsc_scaler *sc) gsc_write(cfg, GSC_MAIN_V_RATIO); } -static int gsc_dst_set_size(struct device *dev, int swap, - struct drm_exynos_pos *pos, struct drm_exynos_sz *sz) +static int gsc_dst_set_size(struct gsc_context *ctx, + struct exynos_drm_ipp_buffer *buf) { - struct gsc_context *ctx = get_gsc_context(dev); - struct drm_exynos_pos img_pos = *pos; struct gsc_scaler *sc = &ctx->sc; u32 cfg; - DRM_DEBUG_KMS("swap[%d]x[%d]y[%d]w[%d]h[%d]\n", - swap, pos->x, pos->y, pos->w, pos->h); - - if (swap) { - img_pos.w = pos->h; - img_pos.h = pos->w; - } - /* pixel offset */ - cfg = (GSC_DSTIMG_OFFSET_X(pos->x) | - GSC_DSTIMG_OFFSET_Y(pos->y)); + cfg = (GSC_DSTIMG_OFFSET_X(buf->rect.x) | + GSC_DSTIMG_OFFSET_Y(buf->rect.y)); gsc_write(cfg, GSC_DSTIMG_OFFSET); /* scaled size */ - cfg = (GSC_SCALED_WIDTH(img_pos.w) | GSC_SCALED_HEIGHT(img_pos.h)); + if (ctx->rotation) + cfg = (GSC_SCALED_WIDTH(buf->rect.h) | + GSC_SCALED_HEIGHT(buf->rect.w)); + else + cfg = (GSC_SCALED_WIDTH(buf->rect.w) | + GSC_SCALED_HEIGHT(buf->rect.h)); gsc_write(cfg, GSC_SCALED_SIZE); - DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", sz->hsize, sz->vsize); - /* original size */ cfg = gsc_read(GSC_DSTIMG_SIZE); - cfg &= ~(GSC_DSTIMG_HEIGHT_MASK | - GSC_DSTIMG_WIDTH_MASK); - cfg |= (GSC_DSTIMG_WIDTH(sz->hsize) | - GSC_DSTIMG_HEIGHT(sz->vsize)); + cfg &= ~(GSC_DSTIMG_HEIGHT_MASK | GSC_DSTIMG_WIDTH_MASK); + cfg |= GSC_DSTIMG_WIDTH(buf->buf.width) | + GSC_DSTIMG_HEIGHT(buf->buf.height); gsc_write(cfg, GSC_DSTIMG_SIZE); cfg = gsc_read(GSC_OUT_CON); cfg &= ~GSC_OUT_RGB_TYPE_MASK; - DRM_DEBUG_KMS("width[%d]range[%d]\n", pos->w, sc->range); - - if (pos->w >= GSC_WIDTH_ITU_709) + if (buf->rect.w >= GSC_WIDTH_ITU_709) if (sc->range) cfg |= GSC_OUT_RGB_HD_WIDE; else @@ -1112,34 +939,16 @@ static int gsc_dst_get_buf_seq(struct gsc_context *ctx) } static int gsc_dst_set_buf_seq(struct gsc_context *ctx, u32 buf_id, - enum drm_exynos_ipp_buf_type buf_type) + bool enqueue) { - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - bool masked; + bool masked = !enqueue; u32 cfg; u32 mask = 0x00000001 << buf_id; int ret = 0; - DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type); - - mutex_lock(&ctx->lock); - /* mask register set */ cfg = gsc_read(GSC_OUT_BASE_ADDR_Y_MASK); - switch (buf_type) { - case IPP_BUF_ENQUEUE: - masked = false; - break; - case IPP_BUF_DEQUEUE: - masked = true; - break; - default: - dev_err(ippdrv->dev, "invalid buf ctrl parameter.\n"); - ret = -EINVAL; - goto err_unlock; - } - /* sequence id */ cfg &= ~mask; cfg |= masked << buf_id; @@ -1148,87 +957,25 @@ static int gsc_dst_set_buf_seq(struct gsc_context *ctx, u32 buf_id, gsc_write(cfg, GSC_OUT_BASE_ADDR_CR_MASK); /* interrupt enable */ - if (buf_type == IPP_BUF_ENQUEUE && - gsc_dst_get_buf_seq(ctx) >= GSC_BUF_START) + if (enqueue && gsc_dst_get_buf_seq(ctx) >= GSC_BUF_START) gsc_handle_irq(ctx, true, false, true); /* interrupt disable */ - if (buf_type == IPP_BUF_DEQUEUE && - gsc_dst_get_buf_seq(ctx) <= GSC_BUF_STOP) + if (!enqueue && gsc_dst_get_buf_seq(ctx) <= GSC_BUF_STOP) gsc_handle_irq(ctx, false, false, true); -err_unlock: - mutex_unlock(&ctx->lock); return ret; } -static int gsc_dst_set_addr(struct device *dev, - struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id, - enum drm_exynos_ipp_buf_type buf_type) +static int gsc_dst_set_addr(struct gsc_context *ctx, + u32 buf_id, struct exynos_drm_ipp_buffer *buf) { - struct gsc_context *ctx = get_gsc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; - struct drm_exynos_ipp_property *property; - - if (!c_node) { - DRM_ERROR("failed to get c_node.\n"); - return -EFAULT; - } - - property = &c_node->property; - - DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n", - property->prop_id, buf_id, buf_type); - - if (buf_id > GSC_MAX_DST) { - dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id); - return -EINVAL; - } - /* address register set */ - switch (buf_type) { - case IPP_BUF_ENQUEUE: - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_Y], - GSC_OUT_BASE_ADDR_Y(buf_id)); - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CB], - GSC_OUT_BASE_ADDR_CB(buf_id)); - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CR], - GSC_OUT_BASE_ADDR_CR(buf_id)); - break; - case IPP_BUF_DEQUEUE: - gsc_write(0x0, GSC_OUT_BASE_ADDR_Y(buf_id)); - gsc_write(0x0, GSC_OUT_BASE_ADDR_CB(buf_id)); - gsc_write(0x0, GSC_OUT_BASE_ADDR_CR(buf_id)); - break; - default: - /* bypass */ - break; - } - - return gsc_dst_set_buf_seq(ctx, buf_id, buf_type); -} - -static struct exynos_drm_ipp_ops gsc_dst_ops = { - .set_fmt = gsc_dst_set_fmt, - .set_transf = gsc_dst_set_transf, - .set_size = gsc_dst_set_size, - .set_addr = gsc_dst_set_addr, -}; - -static int gsc_clk_ctrl(struct gsc_context *ctx, bool enable) -{ - DRM_DEBUG_KMS("enable[%d]\n", enable); - - if (enable) { - clk_prepare_enable(ctx->gsc_clk); - ctx->suspended = false; - } else { - clk_disable_unprepare(ctx->gsc_clk); - ctx->suspended = true; - } + gsc_write(buf->dma_addr[0], GSC_OUT_BASE_ADDR_Y(buf_id)); + gsc_write(buf->dma_addr[1], GSC_OUT_BASE_ADDR_CB(buf_id)); + gsc_write(buf->dma_addr[2], GSC_OUT_BASE_ADDR_CR(buf_id)); - return 0; + return gsc_dst_set_buf_seq(ctx, buf_id, true); } static int gsc_get_src_buf_index(struct gsc_context *ctx) @@ -1249,20 +996,20 @@ static int gsc_get_src_buf_index(struct gsc_context *ctx) } } + DRM_DEBUG_KMS("cfg[0x%x]curr_index[%d]buf_id[%d]\n", cfg, + curr_index, buf_id); + if (buf_id == GSC_MAX_SRC) { DRM_ERROR("failed to get in buffer index.\n"); return -EINVAL; } - ret = gsc_src_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE); + ret = gsc_src_set_buf_seq(ctx, buf_id, false); if (ret < 0) { DRM_ERROR("failed to dequeue.\n"); return ret; } - DRM_DEBUG_KMS("cfg[0x%x]curr_index[%d]buf_id[%d]\n", cfg, - curr_index, buf_id); - return buf_id; } @@ -1289,7 +1036,7 @@ static int gsc_get_dst_buf_index(struct gsc_context *ctx) return -EINVAL; } - ret = gsc_dst_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE); + ret = gsc_dst_set_buf_seq(ctx, buf_id, false); if (ret < 0) { DRM_ERROR("failed to dequeue.\n"); return ret; @@ -1304,215 +1051,55 @@ static int gsc_get_dst_buf_index(struct gsc_context *ctx) static irqreturn_t gsc_irq_handler(int irq, void *dev_id) { struct gsc_context *ctx = dev_id; - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; - struct drm_exynos_ipp_event_work *event_work = - c_node->event_work; u32 status; - int buf_id[EXYNOS_DRM_OPS_MAX]; + int err = 0; DRM_DEBUG_KMS("gsc id[%d]\n", ctx->id); status = gsc_read(GSC_IRQ); if (status & GSC_IRQ_STATUS_OR_IRQ) { - dev_err(ippdrv->dev, "occurred overflow at %d, status 0x%x.\n", + dev_err(ctx->dev, "occurred overflow at %d, status 0x%x.\n", ctx->id, status); - return IRQ_NONE; + err = -EINVAL; } if (status & GSC_IRQ_STATUS_OR_FRM_DONE) { - dev_dbg(ippdrv->dev, "occurred frame done at %d, status 0x%x.\n", - ctx->id, status); - - buf_id[EXYNOS_DRM_OPS_SRC] = gsc_get_src_buf_index(ctx); - if (buf_id[EXYNOS_DRM_OPS_SRC] < 0) - return IRQ_HANDLED; - - buf_id[EXYNOS_DRM_OPS_DST] = gsc_get_dst_buf_index(ctx); - if (buf_id[EXYNOS_DRM_OPS_DST] < 0) - return IRQ_HANDLED; - - DRM_DEBUG_KMS("buf_id_src[%d]buf_id_dst[%d]\n", - buf_id[EXYNOS_DRM_OPS_SRC], buf_id[EXYNOS_DRM_OPS_DST]); - - event_work->ippdrv = ippdrv; - event_work->buf_id[EXYNOS_DRM_OPS_SRC] = - buf_id[EXYNOS_DRM_OPS_SRC]; - event_work->buf_id[EXYNOS_DRM_OPS_DST] = - buf_id[EXYNOS_DRM_OPS_DST]; - queue_work(ippdrv->event_workq, &event_work->work); - } - - return IRQ_HANDLED; -} - -static int gsc_init_prop_list(struct exynos_drm_ippdrv *ippdrv) -{ - struct drm_exynos_ipp_prop_list *prop_list = &ippdrv->prop_list; - - prop_list->version = 1; - prop_list->writeback = 1; - prop_list->refresh_min = GSC_REFRESH_MIN; - prop_list->refresh_max = GSC_REFRESH_MAX; - prop_list->flip = (1 << EXYNOS_DRM_FLIP_VERTICAL) | - (1 << EXYNOS_DRM_FLIP_HORIZONTAL); - prop_list->degree = (1 << EXYNOS_DRM_DEGREE_0) | - (1 << EXYNOS_DRM_DEGREE_90) | - (1 << EXYNOS_DRM_DEGREE_180) | - (1 << EXYNOS_DRM_DEGREE_270); - prop_list->csc = 1; - prop_list->crop = 1; - prop_list->crop_max.hsize = GSC_CROP_MAX; - prop_list->crop_max.vsize = GSC_CROP_MAX; - prop_list->crop_min.hsize = GSC_CROP_MIN; - prop_list->crop_min.vsize = GSC_CROP_MIN; - prop_list->scale = 1; - prop_list->scale_max.hsize = GSC_SCALE_MAX; - prop_list->scale_max.vsize = GSC_SCALE_MAX; - prop_list->scale_min.hsize = GSC_SCALE_MIN; - prop_list->scale_min.vsize = GSC_SCALE_MIN; - - return 0; -} + int src_buf_id, dst_buf_id; -static inline bool gsc_check_drm_flip(enum drm_exynos_flip flip) -{ - switch (flip) { - case EXYNOS_DRM_FLIP_NONE: - case EXYNOS_DRM_FLIP_VERTICAL: - case EXYNOS_DRM_FLIP_HORIZONTAL: - case EXYNOS_DRM_FLIP_BOTH: - return true; - default: - DRM_DEBUG_KMS("invalid flip\n"); - return false; - } -} - -static int gsc_ippdrv_check_property(struct device *dev, - struct drm_exynos_ipp_property *property) -{ - struct gsc_context *ctx = get_gsc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_prop_list *pp = &ippdrv->prop_list; - struct drm_exynos_ipp_config *config; - struct drm_exynos_pos *pos; - struct drm_exynos_sz *sz; - bool swap; - int i; - - for_each_ipp_ops(i) { - if ((i == EXYNOS_DRM_OPS_SRC) && - (property->cmd == IPP_CMD_WB)) - continue; - - config = &property->config[i]; - pos = &config->pos; - sz = &config->sz; - - /* check for flip */ - if (!gsc_check_drm_flip(config->flip)) { - DRM_ERROR("invalid flip.\n"); - goto err_property; - } - - /* check for degree */ - switch (config->degree) { - case EXYNOS_DRM_DEGREE_90: - case EXYNOS_DRM_DEGREE_270: - swap = true; - break; - case EXYNOS_DRM_DEGREE_0: - case EXYNOS_DRM_DEGREE_180: - swap = false; - break; - default: - DRM_ERROR("invalid degree.\n"); - goto err_property; - } + dev_dbg(ctx->dev, "occurred frame done at %d, status 0x%x.\n", + ctx->id, status); - /* check for buffer bound */ - if ((pos->x + pos->w > sz->hsize) || - (pos->y + pos->h > sz->vsize)) { - DRM_ERROR("out of buf bound.\n"); - goto err_property; - } + src_buf_id = gsc_get_src_buf_index(ctx); + dst_buf_id = gsc_get_dst_buf_index(ctx); - /* check for crop */ - if ((i == EXYNOS_DRM_OPS_SRC) && (pp->crop)) { - if (swap) { - if ((pos->h < pp->crop_min.hsize) || - (sz->vsize > pp->crop_max.hsize) || - (pos->w < pp->crop_min.vsize) || - (sz->hsize > pp->crop_max.vsize)) { - DRM_ERROR("out of crop size.\n"); - goto err_property; - } - } else { - if ((pos->w < pp->crop_min.hsize) || - (sz->hsize > pp->crop_max.hsize) || - (pos->h < pp->crop_min.vsize) || - (sz->vsize > pp->crop_max.vsize)) { - DRM_ERROR("out of crop size.\n"); - goto err_property; - } - } - } + DRM_DEBUG_KMS("buf_id_src[%d]buf_id_dst[%d]\n", src_buf_id, + dst_buf_id); - /* check for scale */ - if ((i == EXYNOS_DRM_OPS_DST) && (pp->scale)) { - if (swap) { - if ((pos->h < pp->scale_min.hsize) || - (sz->vsize > pp->scale_max.hsize) || - (pos->w < pp->scale_min.vsize) || - (sz->hsize > pp->scale_max.vsize)) { - DRM_ERROR("out of scale size.\n"); - goto err_property; - } - } else { - if ((pos->w < pp->scale_min.hsize) || - (sz->hsize > pp->scale_max.hsize) || - (pos->h < pp->scale_min.vsize) || - (sz->vsize > pp->scale_max.vsize)) { - DRM_ERROR("out of scale size.\n"); - goto err_property; - } - } - } + if (src_buf_id < 0 || dst_buf_id < 0) + err = -EINVAL; } - return 0; + if (ctx->task) { + struct exynos_drm_ipp_task *task = ctx->task; -err_property: - for_each_ipp_ops(i) { - if ((i == EXYNOS_DRM_OPS_SRC) && - (property->cmd == IPP_CMD_WB)) - continue; - - config = &property->config[i]; - pos = &config->pos; - sz = &config->sz; - - DRM_ERROR("[%s]f[%d]r[%d]pos[%d %d %d %d]sz[%d %d]\n", - i ? "dst" : "src", config->flip, config->degree, - pos->x, pos->y, pos->w, pos->h, - sz->hsize, sz->vsize); + ctx->task = NULL; + pm_runtime_mark_last_busy(ctx->dev); + pm_runtime_put_autosuspend(ctx->dev); + exynos_drm_ipp_task_done(task, err); } - return -EINVAL; + return IRQ_HANDLED; } - -static int gsc_ippdrv_reset(struct device *dev) +static int gsc_reset(struct gsc_context *ctx) { - struct gsc_context *ctx = get_gsc_context(dev); struct gsc_scaler *sc = &ctx->sc; int ret; /* reset h/w block */ ret = gsc_sw_reset(ctx); if (ret < 0) { - dev_err(dev, "failed to reset hardware.\n"); + dev_err(ctx->dev, "failed to reset hardware.\n"); return ret; } @@ -1523,96 +1110,29 @@ static int gsc_ippdrv_reset(struct device *dev) return 0; } -static int gsc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd) +static int gsc_start(struct gsc_context *ctx) { - struct gsc_context *ctx = get_gsc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; - struct drm_exynos_ipp_property *property; - struct drm_exynos_ipp_config *config; - struct drm_exynos_pos img_pos[EXYNOS_DRM_OPS_MAX]; - struct drm_exynos_ipp_set_wb set_wb; u32 cfg; - int ret, i; - - DRM_DEBUG_KMS("cmd[%d]\n", cmd); - - if (!c_node) { - DRM_ERROR("failed to get c_node.\n"); - return -EINVAL; - } - - property = &c_node->property; gsc_handle_irq(ctx, true, false, true); - for_each_ipp_ops(i) { - config = &property->config[i]; - img_pos[i] = config->pos; - } + /* enable one shot */ + cfg = gsc_read(GSC_ENABLE); + cfg &= ~(GSC_ENABLE_ON_CLEAR_MASK | + GSC_ENABLE_CLK_GATE_MODE_MASK); + cfg |= GSC_ENABLE_ON_CLEAR_ONESHOT; + gsc_write(cfg, GSC_ENABLE); - switch (cmd) { - case IPP_CMD_M2M: - /* enable one shot */ - cfg = gsc_read(GSC_ENABLE); - cfg &= ~(GSC_ENABLE_ON_CLEAR_MASK | - GSC_ENABLE_CLK_GATE_MODE_MASK); - cfg |= GSC_ENABLE_ON_CLEAR_ONESHOT; - gsc_write(cfg, GSC_ENABLE); - - /* src dma memory */ - cfg = gsc_read(GSC_IN_CON); - cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK); - cfg |= GSC_IN_PATH_MEMORY; - gsc_write(cfg, GSC_IN_CON); - - /* dst dma memory */ - cfg = gsc_read(GSC_OUT_CON); - cfg |= GSC_OUT_PATH_MEMORY; - gsc_write(cfg, GSC_OUT_CON); - break; - case IPP_CMD_WB: - set_wb.enable = 1; - set_wb.refresh = property->refresh_rate; - gsc_set_gscblk_fimd_wb(ctx, set_wb.enable); - exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb); - - /* src local path */ - cfg = gsc_read(GSC_IN_CON); - cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK); - cfg |= (GSC_IN_PATH_LOCAL | GSC_IN_LOCAL_FIMD_WB); - gsc_write(cfg, GSC_IN_CON); - - /* dst dma memory */ - cfg = gsc_read(GSC_OUT_CON); - cfg |= GSC_OUT_PATH_MEMORY; - gsc_write(cfg, GSC_OUT_CON); - break; - case IPP_CMD_OUTPUT: - /* src dma memory */ - cfg = gsc_read(GSC_IN_CON); - cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK); - cfg |= GSC_IN_PATH_MEMORY; - gsc_write(cfg, GSC_IN_CON); - - /* dst local path */ - cfg = gsc_read(GSC_OUT_CON); - cfg |= GSC_OUT_PATH_MEMORY; - gsc_write(cfg, GSC_OUT_CON); - break; - default: - ret = -EINVAL; - dev_err(dev, "invalid operations.\n"); - return ret; - } + /* src dma memory */ + cfg = gsc_read(GSC_IN_CON); + cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK); + cfg |= GSC_IN_PATH_MEMORY; + gsc_write(cfg, GSC_IN_CON); - ret = gsc_set_prescaler(ctx, &ctx->sc, - &img_pos[EXYNOS_DRM_OPS_SRC], - &img_pos[EXYNOS_DRM_OPS_DST]); - if (ret) { - dev_err(dev, "failed to set prescaler.\n"); - return ret; - } + /* dst dma memory */ + cfg = gsc_read(GSC_OUT_CON); + cfg |= GSC_OUT_PATH_MEMORY; + gsc_write(cfg, GSC_OUT_CON); gsc_set_scaler(ctx, &ctx->sc); @@ -1623,66 +1143,132 @@ static int gsc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd) return 0; } -static void gsc_ippdrv_stop(struct device *dev, enum drm_exynos_ipp_cmd cmd) +static int gsc_commit(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) { - struct gsc_context *ctx = get_gsc_context(dev); - struct drm_exynos_ipp_set_wb set_wb = {0, 0}; - u32 cfg; + struct gsc_context *ctx = container_of(ipp, struct gsc_context, ipp); + + pm_runtime_get_sync(ctx->dev); + ctx->task = task; + + gsc_reset(ctx); + gsc_src_set_fmt(ctx, task->src.buf.fourcc); + gsc_src_set_transf(ctx, task->transform.rotation); + gsc_src_set_size(ctx, &task->src); + gsc_src_set_addr(ctx, 0, &task->src); + gsc_dst_set_fmt(ctx, task->dst.buf.fourcc); + gsc_dst_set_size(ctx, &task->dst); + gsc_dst_set_addr(ctx, 0, &task->dst); + gsc_set_prescaler(ctx, &ctx->sc, &task->src.rect, &task->dst.rect); + gsc_start(ctx); - DRM_DEBUG_KMS("cmd[%d]\n", cmd); + return 0; +} - switch (cmd) { - case IPP_CMD_M2M: - /* bypass */ - break; - case IPP_CMD_WB: - gsc_set_gscblk_fimd_wb(ctx, set_wb.enable); - exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb); - break; - case IPP_CMD_OUTPUT: - default: - dev_err(dev, "invalid operations.\n"); - break; +static void gsc_abort(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) +{ + struct gsc_context *ctx = + container_of(ipp, struct gsc_context, ipp); + + gsc_reset(ctx); + if (ctx->task) { + struct exynos_drm_ipp_task *task = ctx->task; + + ctx->task = NULL; + pm_runtime_mark_last_busy(ctx->dev); + pm_runtime_put_autosuspend(ctx->dev); + exynos_drm_ipp_task_done(task, -EIO); } +} - gsc_handle_irq(ctx, false, false, true); +static struct exynos_drm_ipp_funcs ipp_funcs = { + .commit = gsc_commit, + .abort = gsc_abort, +}; - /* reset sequence */ - gsc_write(0xff, GSC_OUT_BASE_ADDR_Y_MASK); - gsc_write(0xff, GSC_OUT_BASE_ADDR_CB_MASK); - gsc_write(0xff, GSC_OUT_BASE_ADDR_CR_MASK); +static int gsc_bind(struct device *dev, struct device *master, void *data) +{ + struct gsc_context *ctx = dev_get_drvdata(dev); + struct drm_device *drm_dev = data; + struct exynos_drm_ipp *ipp = &ctx->ipp; - cfg = gsc_read(GSC_ENABLE); - cfg &= ~GSC_ENABLE_ON; - gsc_write(cfg, GSC_ENABLE); + ctx->drm_dev = drm_dev; + drm_iommu_attach_device(drm_dev, dev); + + exynos_drm_ipp_register(drm_dev, ipp, &ipp_funcs, + DRM_EXYNOS_IPP_CAP_CROP | DRM_EXYNOS_IPP_CAP_ROTATE | + DRM_EXYNOS_IPP_CAP_SCALE | DRM_EXYNOS_IPP_CAP_CONVERT, + ctx->formats, "gsc"); + + dev_info(dev, "The exynos gscaler is probed successfully\n"); + + return 0; +} + +static void gsc_unbind(struct device *dev, struct device *master, + void *data) +{ + struct gsc_context *ctx = dev_get_drvdata(dev); + struct drm_device *drm_dev = data; + struct exynos_drm_ipp *ipp = &ctx->ipp; + + exynos_drm_ipp_unregister(drm_dev, ipp); + drm_iommu_detach_device(drm_dev, dev); } +static const struct component_ops gsc_component_ops = { + .bind = gsc_bind, + .unbind = gsc_unbind, +}; + +static const unsigned int gsc_formats[] = { + DRM_FORMAT_XRGB8888, DRM_FORMAT_RGB565, DRM_FORMAT_BGRX8888, + DRM_FORMAT_NV12, DRM_FORMAT_NV16, DRM_FORMAT_NV21, DRM_FORMAT_NV61, + DRM_FORMAT_UYVY, DRM_FORMAT_VYUY, DRM_FORMAT_YUYV, DRM_FORMAT_YVYU, + DRM_FORMAT_YUV420, DRM_FORMAT_YVU420, DRM_FORMAT_YUV422, +}; + static int gsc_probe(struct platform_device *pdev) { struct device *dev = &pdev->dev; + struct gsc_driverdata *driver_data; + struct exynos_drm_ipp_formats *formats; struct gsc_context *ctx; struct resource *res; - struct exynos_drm_ippdrv *ippdrv; - int ret; + int ret, i; ctx = devm_kzalloc(dev, sizeof(*ctx), GFP_KERNEL); if (!ctx) return -ENOMEM; - if (dev->of_node) { - ctx->sysreg = syscon_regmap_lookup_by_phandle(dev->of_node, - "samsung,sysreg"); - if (IS_ERR(ctx->sysreg)) { - dev_warn(dev, "failed to get system register.\n"); - ctx->sysreg = NULL; - } + formats = devm_kzalloc(dev, sizeof(*formats) * + (ARRAY_SIZE(gsc_formats) + 1), GFP_KERNEL); + if (!formats) + return -ENOMEM; + + + driver_data = (struct gsc_driverdata *)of_device_get_match_data(dev); + ctx->dev = dev; + ctx->num_clocks = driver_data->num_clocks; + ctx->clk_names = driver_data->clk_names; + + for (i = 0; i < ARRAY_SIZE(gsc_formats); i++) { + formats[i].fourcc = gsc_formats[i]; + formats[i].type = DRM_EXYNOS_IPP_FORMAT_SOURCE | + DRM_EXYNOS_IPP_FORMAT_DESTINATION; + formats[i].limits = driver_data->limits; } + ctx->formats = formats; /* clock control */ - ctx->gsc_clk = devm_clk_get(dev, "gscl"); - if (IS_ERR(ctx->gsc_clk)) { - dev_err(dev, "failed to get gsc clock.\n"); - return PTR_ERR(ctx->gsc_clk); + for (i = 0; i < ctx->num_clocks; i++) { + ctx->clocks[i] = devm_clk_get(dev, ctx->clk_names[i]); + if (IS_ERR(ctx->clocks[i])) { + dev_err(dev, "failed to get clock: %s\n", + ctx->clk_names[i]); + return PTR_ERR(ctx->clocks[i]); + } } /* resource memory */ @@ -1699,8 +1285,8 @@ static int gsc_probe(struct platform_device *pdev) } ctx->irq = res->start; - ret = devm_request_threaded_irq(dev, ctx->irq, NULL, gsc_irq_handler, - IRQF_ONESHOT, "drm_gsc", ctx); + ret = devm_request_irq(dev, ctx->irq, gsc_irq_handler, 0, + dev_name(dev), ctx); if (ret < 0) { dev_err(dev, "failed to request irq.\n"); return ret; @@ -1709,38 +1295,22 @@ static int gsc_probe(struct platform_device *pdev) /* context initailization */ ctx->id = pdev->id; - ippdrv = &ctx->ippdrv; - ippdrv->dev = dev; - ippdrv->ops[EXYNOS_DRM_OPS_SRC] = &gsc_src_ops; - ippdrv->ops[EXYNOS_DRM_OPS_DST] = &gsc_dst_ops; - ippdrv->check_property = gsc_ippdrv_check_property; - ippdrv->reset = gsc_ippdrv_reset; - ippdrv->start = gsc_ippdrv_start; - ippdrv->stop = gsc_ippdrv_stop; - ret = gsc_init_prop_list(ippdrv); - if (ret < 0) { - dev_err(dev, "failed to init property list.\n"); - return ret; - } - - DRM_DEBUG_KMS("id[%d]ippdrv[%pK]\n", ctx->id, ippdrv); - - mutex_init(&ctx->lock); platform_set_drvdata(pdev, ctx); + pm_runtime_use_autosuspend(dev); + pm_runtime_set_autosuspend_delay(dev, GSC_AUTOSUSPEND_DELAY); pm_runtime_enable(dev); - ret = exynos_drm_ippdrv_register(ippdrv); - if (ret < 0) { - dev_err(dev, "failed to register drm gsc device.\n"); - goto err_ippdrv_register; - } + ret = component_add(dev, &gsc_component_ops); + if (ret) + goto err_pm_dis; dev_info(dev, "drm gsc registered successfully.\n"); return 0; -err_ippdrv_register: +err_pm_dis: + pm_runtime_dont_use_autosuspend(dev); pm_runtime_disable(dev); return ret; } @@ -1748,13 +1318,8 @@ static int gsc_probe(struct platform_device *pdev) static int gsc_remove(struct platform_device *pdev) { struct device *dev = &pdev->dev; - struct gsc_context *ctx = get_gsc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - - exynos_drm_ippdrv_unregister(ippdrv); - mutex_destroy(&ctx->lock); - pm_runtime_set_suspended(dev); + pm_runtime_dont_use_autosuspend(dev); pm_runtime_disable(dev); return 0; @@ -1763,19 +1328,32 @@ static int gsc_remove(struct platform_device *pdev) static int __maybe_unused gsc_runtime_suspend(struct device *dev) { struct gsc_context *ctx = get_gsc_context(dev); + int i; DRM_DEBUG_KMS("id[%d]\n", ctx->id); - return gsc_clk_ctrl(ctx, false); + for (i = ctx->num_clocks - 1; i >= 0; i--) + clk_disable_unprepare(ctx->clocks[i]); + + return 0; } static int __maybe_unused gsc_runtime_resume(struct device *dev) { struct gsc_context *ctx = get_gsc_context(dev); + int i; DRM_DEBUG_KMS("id[%d]\n", ctx->id); - return gsc_clk_ctrl(ctx, true); + for (i = 0; i < ctx->num_clocks; i++) { + int ret = clk_prepare_enable(ctx->clocks[i]); + if (ret) { + while (--i > 0) + clk_disable_unprepare(ctx->clocks[i]); + return ret; + } + } + return 0; } static const struct dev_pm_ops gsc_pm_ops = { @@ -1784,9 +1362,66 @@ static const struct dev_pm_ops gsc_pm_ops = { SET_RUNTIME_PM_OPS(gsc_runtime_suspend, gsc_runtime_resume, NULL) }; +static const struct drm_exynos_ipp_limit gsc_5250_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {32, 4800, 8}, .v = {16, 3344, 8}) }, + { IPP_SIZE_LIMIT(AREA, .h = {16, 4800, 2}, .v = {8, 3344, 2}) }, + { IPP_SIZE_LIMIT(ROTATED, .h = {32, 2048}, .v = {16, 2048}) }, + { IPP_SCALE_LIMIT(.h = { (1 << 16) / 16, (1 << 16) * 8}, + .v = { (1 << 16) / 16, (1 << 16) * 8}) }, + { } +}; + +static const struct drm_exynos_ipp_limit gsc_5420_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {32, 4800, 8}, .v = {16, 3344, 8}) }, + { IPP_SIZE_LIMIT(AREA, .h = {16, 4800, 2}, .v = {8, 3344, 2}) }, + { IPP_SIZE_LIMIT(ROTATED, .h = {16, 2016}, .v = {8, 2016}) }, + { IPP_SCALE_LIMIT(.h = { (1 << 16) / 16, (1 << 16) * 8}, + .v = { (1 << 16) / 16, (1 << 16) * 8}) }, + { } +}; + +static const struct drm_exynos_ipp_limit gsc_5433_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {32, 8191, 8}, .v = {16, 8191, 8}) }, + { IPP_SIZE_LIMIT(AREA, .h = {16, 4800, 2}, .v = {8, 3344, 2}) }, + { IPP_SIZE_LIMIT(ROTATED, .h = {16, 2047}, .v = {8, 2047}) }, + { IPP_SCALE_LIMIT(.h = { (1 << 16) / 16, (1 << 16) * 8}, + .v = { (1 << 16) / 16, (1 << 16) * 8}) }, + { } +}; + +static struct gsc_driverdata gsc_exynos5250_drvdata = { + .clk_names = {"gscl"}, + .num_clocks = 1, + .limits = gsc_5250_limits, +}; + +static struct gsc_driverdata gsc_exynos5420_drvdata = { + .clk_names = {"gscl"}, + .num_clocks = 1, + .limits = gsc_5420_limits, +}; + +static struct gsc_driverdata gsc_exynos5433_drvdata = { + .clk_names = {"pclk", "aclk", "aclk_xiu", "aclk_gsclbend"}, + .num_clocks = 4, + .limits = gsc_5433_limits, +}; + static const struct of_device_id exynos_drm_gsc_of_match[] = { - { .compatible = "samsung,exynos5-gsc" }, - { }, + { + .compatible = "samsung,exynos5-gsc", + .data = &gsc_exynos5250_drvdata, + }, { + .compatible = "samsung,exynos5250-gsc", + .data = &gsc_exynos5250_drvdata, + }, { + .compatible = "samsung,exynos5420-gsc", + .data = &gsc_exynos5420_drvdata, + }, { + .compatible = "samsung,exynos5433-gsc", + .data = &gsc_exynos5433_drvdata, + }, { + }, }; MODULE_DEVICE_TABLE(of, exynos_drm_gsc_of_match); diff --git a/drivers/gpu/drm/exynos/exynos_drm_gsc.h b/drivers/gpu/drm/exynos/exynos_drm_gsc.h deleted file mode 100644 index 29ec1c5efcf2..000000000000 --- a/drivers/gpu/drm/exynos/exynos_drm_gsc.h +++ /dev/null @@ -1,24 +0,0 @@ -/* - * Copyright (c) 2012 Samsung Electronics Co., Ltd. - * - * Authors: - * Eunchul Kim <chulspro.kim@samsung.com> - * Jinyoung Jeon <jy0.jeon@samsung.com> - * Sangmin Lee <lsmin.lee@samsung.com> - * - * This program is free software; you can redistribute it and/or modify it - * under the terms of the GNU General Public License as published by the - * Free Software Foundation; either version 2 of the License, or (at your - * option) any later version. - */ - -#ifndef _EXYNOS_DRM_GSC_H_ -#define _EXYNOS_DRM_GSC_H_ - -/* - * TODO - * FIMD output interface notifier callback. - * Mixer output interface notifier callback. - */ - -#endif /* _EXYNOS_DRM_GSC_H_ */ -- 2.14.2 ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH v4 4/9] drm/exynos: gsc: Convert driver to IPP v2 core API 2017-10-23 7:54 ` [PATCH v4 4/9] drm/exynos: gsc: " Marek Szyprowski @ 2017-11-01 6:02 ` Inki Dae 0 siblings, 0 replies; 20+ messages in thread From: Inki Dae @ 2017-11-01 6:02 UTC (permalink / raw) To: Marek Szyprowski, dri-devel, linux-samsung-soc Cc: Tobias Jakobi, Bartlomiej Zolnierkiewicz, Seung-Woo Kim, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon 2017년 10월 23일 16:54에 Marek Szyprowski 이(가) 쓴 글: > This patch adapts Exynos DRM GScaler driver to new IPP v2 core API. > The side effect of this conversion is a switch to driver component API > to register properly in the Exynos DRM core. During the conversion > driver has been adapted to support more specific compatible strings > to distinguish between Exynos5250 and Exynos5420 (different hardware > limits). Support for Exynos5433 variant has been added too > (different limits table, removed dependency on ARCH_EXYNOS5). > > Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> > Tested-by: Hoegeun Kwon <hoegeun.kwon@samsung.com> > --- > drivers/gpu/drm/exynos/Kconfig | 3 +- > drivers/gpu/drm/exynos/exynos_drm_drv.c | 1 + > drivers/gpu/drm/exynos/exynos_drm_gsc.c | 1001 ++++++++++--------------------- > drivers/gpu/drm/exynos/exynos_drm_gsc.h | 24 - > 4 files changed, 321 insertions(+), 708 deletions(-) > delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_gsc.h > > diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig > index c10c9ca0d8b4..9b66a9838dca 100644 > --- a/drivers/gpu/drm/exynos/Kconfig > +++ b/drivers/gpu/drm/exynos/Kconfig > @@ -111,7 +111,8 @@ config DRM_EXYNOS_ROTATOR > > config DRM_EXYNOS_GSC > bool "GScaler" > - depends on BROKEN && ARCH_EXYNOS5 && VIDEO_SAMSUNG_EXYNOS_GSC=n > + depends on VIDEO_SAMSUNG_EXYNOS_GSC=n > + select DRM_EXYNOS_IPP > help > Choose this option if you want to use Exynos GSC for DRM. > > diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c > index 3ade2b0ad15d..cac0d84385d3 100644 > --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c > +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c > @@ -260,6 +260,7 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { > DRM_COMPONENT_DRIVER > }, { > DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC), > + DRM_COMPONENT_DRIVER > }, { > &exynos_drm_platform_driver, > DRM_VIRTUAL_DEVICE > diff --git a/drivers/gpu/drm/exynos/exynos_drm_gsc.c b/drivers/gpu/drm/exynos/exynos_drm_gsc.c > index 0506b2b17ac1..4671ac41469d 100644 > --- a/drivers/gpu/drm/exynos/exynos_drm_gsc.c > +++ b/drivers/gpu/drm/exynos/exynos_drm_gsc.c > @@ -12,18 +12,20 @@ > * > */ > #include <linux/kernel.h> > +#include <linux/component.h> > #include <linux/platform_device.h> > #include <linux/clk.h> > #include <linux/pm_runtime.h> > #include <linux/mfd/syscon.h> > +#include <linux/of_device.h> > #include <linux/regmap.h> > > #include <drm/drmP.h> > #include <drm/exynos_drm.h> > #include "regs-gsc.h" > #include "exynos_drm_drv.h" > +#include "exynos_drm_iommu.h" > #include "exynos_drm_ipp.h" > -#include "exynos_drm_gsc.h" > > /* > * GSC stands for General SCaler and > @@ -31,26 +33,10 @@ > * input DMA reads image data from the memory. > * output DMA writes image data to memory. > * GSC supports image rotation and image effect functions. > - * > - * M2M operation : supports crop/scale/rotation/csc so on. > - * Memory ----> GSC H/W ----> Memory. > - * Writeback operation : supports cloned screen with FIMD. > - * FIMD ----> GSC H/W ----> Memory. > - * Output operation : supports direct display using local path. > - * Memory ----> GSC H/W ----> FIMD, Mixer. > */ > > -/* > - * TODO > - * 1. check suspend/resume api if needed. > - * 2. need to check use case platform_device_id. > - * 3. check src/dst size with, height. > - * 4. added check_prepare api for right register. > - * 5. need to add supported list in prop_list. > - * 6. check prescaler/scaler optimization. > - */ > > -#define GSC_MAX_DEVS 4 > +#define GSC_MAX_CLOCKS 8 > #define GSC_MAX_SRC 4 > #define GSC_MAX_DST 16 > #define GSC_RESET_TIMEOUT 50 > @@ -65,8 +51,6 @@ > #define GSC_SC_DOWN_RATIO_4_8 131072 > #define GSC_SC_DOWN_RATIO_3_8 174762 > #define GSC_SC_DOWN_RATIO_2_8 262144 > -#define GSC_REFRESH_MIN 12 > -#define GSC_REFRESH_MAX 60 > #define GSC_CROP_MAX 8192 > #define GSC_CROP_MIN 32 > #define GSC_SCALE_MAX 4224 > @@ -77,10 +61,9 @@ > #define GSC_COEF_H_8T 8 > #define GSC_COEF_V_4T 4 > #define GSC_COEF_DEPTH 3 > +#define GSC_AUTOSUSPEND_DELAY 2000 > > #define get_gsc_context(dev) platform_get_drvdata(to_platform_device(dev)) > -#define get_ctx_from_ippdrv(ippdrv) container_of(ippdrv,\ > - struct gsc_context, ippdrv); > #define gsc_read(offset) readl(ctx->regs + (offset)) > #define gsc_write(cfg, offset) writel(cfg, ctx->regs + (offset)) > > @@ -124,7 +107,6 @@ struct gsc_capability { > /* > * A structure of gsc context. > * > - * @ippdrv: prepare initialization using ippdrv. > * @regs_res: register resources. > * @regs: memory mapped io registers. > * @sysreg: handle to SYSREG block regmap. > @@ -137,12 +119,18 @@ struct gsc_capability { > * @suspended: qos operations. > */ > struct gsc_context { > - struct exynos_drm_ippdrv ippdrv; > + struct exynos_drm_ipp ipp; > + struct drm_device *drm_dev; > + struct device *dev; > + struct exynos_drm_ipp_task *task; > + struct exynos_drm_ipp_formats *formats; > + > struct resource *regs_res; > void __iomem *regs; > struct regmap *sysreg; > - struct mutex lock; > - struct clk *gsc_clk; > + const char **clk_names; > + struct clk *clocks[GSC_MAX_CLOCKS]; > + int num_clocks; > struct gsc_scaler sc; > int id; > int irq; > @@ -150,6 +138,19 @@ struct gsc_context { > bool suspended; > }; > > +/** > + * struct gsc_driverdata - per device type driver data for init time. > + * > + * @limits: picture size limits array > + * @clk_names: names of clocks needed by this variant > + * @num_clocks: the number of clocks needed by this variant > + */ > +struct gsc_driverdata { > + const struct drm_exynos_ipp_limit *limits; > + const char *clk_names[GSC_MAX_CLOCKS]; > + int num_clocks; > +}; > + > /* 8-tap Filter Coefficient */ > static const int h_coef_8t[GSC_COEF_RATIO][GSC_COEF_ATTR][GSC_COEF_H_8T] = { > { /* Ratio <= 65536 (~8:8) */ > @@ -438,25 +439,6 @@ static int gsc_sw_reset(struct gsc_context *ctx) > return 0; > } > > -static void gsc_set_gscblk_fimd_wb(struct gsc_context *ctx, bool enable) > -{ > - unsigned int gscblk_cfg; > - > - if (!ctx->sysreg) > - return; > - > - regmap_read(ctx->sysreg, SYSREG_GSCBLK_CFG1, &gscblk_cfg); > - > - if (enable) > - gscblk_cfg |= GSC_BLK_DISP1WB_DEST(ctx->id) | > - GSC_BLK_GSCL_WB_IN_SRC_SEL(ctx->id) | > - GSC_BLK_SW_RESET_WB_DEST(ctx->id); > - else > - gscblk_cfg |= GSC_BLK_PXLASYNC_LO_MASK_WB(ctx->id); > - > - regmap_write(ctx->sysreg, SYSREG_GSCBLK_CFG1, gscblk_cfg); > -} > - > static void gsc_handle_irq(struct gsc_context *ctx, bool enable, > bool overflow, bool done) > { > @@ -487,10 +469,8 @@ static void gsc_handle_irq(struct gsc_context *ctx, bool enable, > } > > > -static int gsc_src_set_fmt(struct device *dev, u32 fmt) > +static int gsc_src_set_fmt(struct gsc_context *ctx, u32 fmt) > { > - struct gsc_context *ctx = get_gsc_context(dev); > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > u32 cfg; > > DRM_DEBUG_KMS("fmt[0x%x]\n", fmt); > @@ -549,7 +529,7 @@ static int gsc_src_set_fmt(struct device *dev, u32 fmt) > GSC_IN_YUV420_2P); > break; > default: > - dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt); > + dev_err(ctx->dev, "invalid target yuv order 0x%x.\n", fmt); > return -EINVAL; > } > > @@ -558,105 +538,81 @@ static int gsc_src_set_fmt(struct device *dev, u32 fmt) > return 0; > } > > -static int gsc_src_set_transf(struct device *dev, > - enum drm_exynos_degree degree, > - enum drm_exynos_flip flip, bool *swap) > +static int gsc_src_set_transf(struct gsc_context *ctx, unsigned int rotation) This function never return a error. > { > - struct gsc_context *ctx = get_gsc_context(dev); > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > + unsigned int degree = rotation & DRM_MODE_ROTATE_MASK; > u32 cfg; > > - DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip); > - > cfg = gsc_read(GSC_IN_CON); > cfg &= ~GSC_IN_ROT_MASK; > > switch (degree) { > - case EXYNOS_DRM_DEGREE_0: > - if (flip & EXYNOS_DRM_FLIP_VERTICAL) > + case DRM_MODE_ROTATE_0: > + if (rotation & DRM_MODE_REFLECT_Y) > cfg |= GSC_IN_ROT_XFLIP; > - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) > + if (rotation & DRM_MODE_REFLECT_X) > cfg |= GSC_IN_ROT_YFLIP; > break; > - case EXYNOS_DRM_DEGREE_90: > - if (flip & EXYNOS_DRM_FLIP_VERTICAL) > + case DRM_MODE_ROTATE_90: > + if (rotation & DRM_MODE_REFLECT_Y) > cfg |= GSC_IN_ROT_90_XFLIP; > - else if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) > + else if (rotation & DRM_MODE_REFLECT_X) > cfg |= GSC_IN_ROT_90_YFLIP; > else > cfg |= GSC_IN_ROT_90; > break; > - case EXYNOS_DRM_DEGREE_180: > + case DRM_MODE_ROTATE_180: > cfg |= GSC_IN_ROT_180; > - if (flip & EXYNOS_DRM_FLIP_VERTICAL) > + if (rotation & DRM_MODE_REFLECT_Y) > cfg &= ~GSC_IN_ROT_XFLIP; > - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) > + if (rotation & DRM_MODE_REFLECT_X) > cfg &= ~GSC_IN_ROT_YFLIP; > break; > - case EXYNOS_DRM_DEGREE_270: > + case DRM_MODE_ROTATE_270: > cfg |= GSC_IN_ROT_270; > - if (flip & EXYNOS_DRM_FLIP_VERTICAL) > + if (rotation & DRM_MODE_REFLECT_Y) > cfg &= ~GSC_IN_ROT_XFLIP; > - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) > + if (rotation & DRM_MODE_REFLECT_X) > cfg &= ~GSC_IN_ROT_YFLIP; > break; > - default: > - dev_err(ippdrv->dev, "invalid degree value %d.\n", degree); > - return -EINVAL; > } > > gsc_write(cfg, GSC_IN_CON); > > ctx->rotation = (cfg & GSC_IN_ROT_90) ? 1 : 0; > - *swap = ctx->rotation; > - > return 0; > } > > -static int gsc_src_set_size(struct device *dev, int swap, > - struct drm_exynos_pos *pos, struct drm_exynos_sz *sz) > +static int gsc_src_set_size(struct gsc_context *ctx, > + struct exynos_drm_ipp_buffer *buf) ditto. > { > - struct gsc_context *ctx = get_gsc_context(dev); > - struct drm_exynos_pos img_pos = *pos; > struct gsc_scaler *sc = &ctx->sc; > u32 cfg; > > - DRM_DEBUG_KMS("swap[%d]x[%d]y[%d]w[%d]h[%d]\n", > - swap, pos->x, pos->y, pos->w, pos->h); > - > - if (swap) { > - img_pos.w = pos->h; > - img_pos.h = pos->w; > - } > - > /* pixel offset */ > - cfg = (GSC_SRCIMG_OFFSET_X(img_pos.x) | > - GSC_SRCIMG_OFFSET_Y(img_pos.y)); > + cfg = (GSC_SRCIMG_OFFSET_X(buf->rect.x) | > + GSC_SRCIMG_OFFSET_Y(buf->rect.y)); > gsc_write(cfg, GSC_SRCIMG_OFFSET); > > /* cropped size */ > - cfg = (GSC_CROPPED_WIDTH(img_pos.w) | > - GSC_CROPPED_HEIGHT(img_pos.h)); > + cfg = (GSC_CROPPED_WIDTH(buf->rect.w) | > + GSC_CROPPED_HEIGHT(buf->rect.h)); > gsc_write(cfg, GSC_CROPPED_SIZE); > > - DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", sz->hsize, sz->vsize); > - > /* original size */ > cfg = gsc_read(GSC_SRCIMG_SIZE); > cfg &= ~(GSC_SRCIMG_HEIGHT_MASK | > GSC_SRCIMG_WIDTH_MASK); > > - cfg |= (GSC_SRCIMG_WIDTH(sz->hsize) | > - GSC_SRCIMG_HEIGHT(sz->vsize)); > + cfg |= (GSC_SRCIMG_WIDTH(buf->buf.width) | > + GSC_SRCIMG_HEIGHT(buf->buf.height)); > > gsc_write(cfg, GSC_SRCIMG_SIZE); > > cfg = gsc_read(GSC_IN_CON); > cfg &= ~GSC_IN_RGB_TYPE_MASK; > > - DRM_DEBUG_KMS("width[%d]range[%d]\n", pos->w, sc->range); > - > - if (pos->w >= GSC_WIDTH_ITU_709) > + if (buf->rect.w >= GSC_WIDTH_ITU_709) > if (sc->range) > cfg |= GSC_IN_RGB_HD_WIDE; > else > @@ -673,30 +629,15 @@ static int gsc_src_set_size(struct device *dev, int swap, > } > > static int gsc_src_set_buf_seq(struct gsc_context *ctx, u32 buf_id, > - enum drm_exynos_ipp_buf_type buf_type) > + bool enqueue) ditto. > { > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > - bool masked; > + bool masked = !enqueue; > u32 cfg; > u32 mask = 0x00000001 << buf_id; > > - DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type); > - > /* mask register set */ > cfg = gsc_read(GSC_IN_BASE_ADDR_Y_MASK); > > - switch (buf_type) { > - case IPP_BUF_ENQUEUE: > - masked = false; > - break; > - case IPP_BUF_DEQUEUE: > - masked = true; > - break; > - default: > - dev_err(ippdrv->dev, "invalid buf ctrl parameter.\n"); > - return -EINVAL; > - } > - > /* sequence id */ > cfg &= ~mask; > cfg |= masked << buf_id; > @@ -707,64 +648,19 @@ static int gsc_src_set_buf_seq(struct gsc_context *ctx, u32 buf_id, > return 0; > } > > -static int gsc_src_set_addr(struct device *dev, > - struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id, > - enum drm_exynos_ipp_buf_type buf_type) > +static int gsc_src_set_addr(struct gsc_context *ctx, u32 buf_id, > + struct exynos_drm_ipp_buffer *buf) > { > - struct gsc_context *ctx = get_gsc_context(dev); > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; > - struct drm_exynos_ipp_property *property; > - > - if (!c_node) { > - DRM_ERROR("failed to get c_node.\n"); > - return -EFAULT; > - } > - > - property = &c_node->property; > - > - DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n", > - property->prop_id, buf_id, buf_type); > - > - if (buf_id > GSC_MAX_SRC) { > - dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id); > - return -EINVAL; > - } > - > /* address register set */ > - switch (buf_type) { > - case IPP_BUF_ENQUEUE: > - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_Y], > - GSC_IN_BASE_ADDR_Y(buf_id)); > - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CB], > - GSC_IN_BASE_ADDR_CB(buf_id)); > - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CR], > - GSC_IN_BASE_ADDR_CR(buf_id)); > - break; > - case IPP_BUF_DEQUEUE: > - gsc_write(0x0, GSC_IN_BASE_ADDR_Y(buf_id)); > - gsc_write(0x0, GSC_IN_BASE_ADDR_CB(buf_id)); > - gsc_write(0x0, GSC_IN_BASE_ADDR_CR(buf_id)); > - break; > - default: > - /* bypass */ > - break; > - } > + gsc_write(buf->dma_addr[0], GSC_IN_BASE_ADDR_Y(buf_id)); > + gsc_write(buf->dma_addr[1], GSC_IN_BASE_ADDR_CB(buf_id)); > + gsc_write(buf->dma_addr[2], GSC_IN_BASE_ADDR_CR(buf_id)); > > - return gsc_src_set_buf_seq(ctx, buf_id, buf_type); > + return gsc_src_set_buf_seq(ctx, buf_id, true); > } > > -static struct exynos_drm_ipp_ops gsc_src_ops = { > - .set_fmt = gsc_src_set_fmt, > - .set_transf = gsc_src_set_transf, > - .set_size = gsc_src_set_size, > - .set_addr = gsc_src_set_addr, > -}; > - > -static int gsc_dst_set_fmt(struct device *dev, u32 fmt) > +static int gsc_dst_set_fmt(struct gsc_context *ctx, u32 fmt) ditto. > { > - struct gsc_context *ctx = get_gsc_context(dev); > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > u32 cfg; > > DRM_DEBUG_KMS("fmt[0x%x]\n", fmt); > @@ -819,9 +715,6 @@ static int gsc_dst_set_fmt(struct device *dev, u32 fmt) > cfg |= (GSC_OUT_CHROMA_ORDER_CBCR | > GSC_OUT_YUV420_2P); > break; > - default: > - dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt); > - return -EINVAL; > } > > gsc_write(cfg, GSC_OUT_CON); > @@ -829,61 +722,6 @@ static int gsc_dst_set_fmt(struct device *dev, u32 fmt) > return 0; > } > > -static int gsc_dst_set_transf(struct device *dev, > - enum drm_exynos_degree degree, > - enum drm_exynos_flip flip, bool *swap) > -{ > - struct gsc_context *ctx = get_gsc_context(dev); > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > - u32 cfg; > - > - DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip); > - > - cfg = gsc_read(GSC_IN_CON); > - cfg &= ~GSC_IN_ROT_MASK; > - > - switch (degree) { > - case EXYNOS_DRM_DEGREE_0: > - if (flip & EXYNOS_DRM_FLIP_VERTICAL) > - cfg |= GSC_IN_ROT_XFLIP; > - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) > - cfg |= GSC_IN_ROT_YFLIP; > - break; > - case EXYNOS_DRM_DEGREE_90: > - if (flip & EXYNOS_DRM_FLIP_VERTICAL) > - cfg |= GSC_IN_ROT_90_XFLIP; > - else if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) > - cfg |= GSC_IN_ROT_90_YFLIP; > - else > - cfg |= GSC_IN_ROT_90; > - break; > - case EXYNOS_DRM_DEGREE_180: > - cfg |= GSC_IN_ROT_180; > - if (flip & EXYNOS_DRM_FLIP_VERTICAL) > - cfg &= ~GSC_IN_ROT_XFLIP; > - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) > - cfg &= ~GSC_IN_ROT_YFLIP; > - break; > - case EXYNOS_DRM_DEGREE_270: > - cfg |= GSC_IN_ROT_270; > - if (flip & EXYNOS_DRM_FLIP_VERTICAL) > - cfg &= ~GSC_IN_ROT_XFLIP; > - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) > - cfg &= ~GSC_IN_ROT_YFLIP; > - break; > - default: > - dev_err(ippdrv->dev, "invalid degree value %d.\n", degree); > - return -EINVAL; > - } > - > - gsc_write(cfg, GSC_IN_CON); > - > - ctx->rotation = (cfg & GSC_IN_ROT_90) ? 1 : 0; > - *swap = ctx->rotation; > - > - return 0; > -} > - > static int gsc_get_ratio_shift(u32 src, u32 dst, u32 *ratio) > { > DRM_DEBUG_KMS("src[%d]dst[%d]\n", src, dst); > @@ -919,9 +757,8 @@ static void gsc_get_prescaler_shfactor(u32 hratio, u32 vratio, u32 *shfactor) > } > > static int gsc_set_prescaler(struct gsc_context *ctx, struct gsc_scaler *sc, > - struct drm_exynos_pos *src, struct drm_exynos_pos *dst) > + struct drm_exynos_ipp_task_rect *src, struct drm_exynos_ipp_task_rect *dst) > { > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > u32 cfg; > u32 src_w, src_h, dst_w, dst_h; > int ret = 0; > @@ -939,13 +776,13 @@ static int gsc_set_prescaler(struct gsc_context *ctx, struct gsc_scaler *sc, > > ret = gsc_get_ratio_shift(src_w, dst_w, &sc->pre_hratio); > if (ret) { > - dev_err(ippdrv->dev, "failed to get ratio horizontal.\n"); > + dev_err(ctx->dev, "failed to get ratio horizontal.\n"); > return ret; > } > > ret = gsc_get_ratio_shift(src_h, dst_h, &sc->pre_vratio); > if (ret) { > - dev_err(ippdrv->dev, "failed to get ratio vertical.\n"); > + dev_err(ctx->dev, "failed to get ratio vertical.\n"); > return ret; > } > > @@ -1039,47 +876,37 @@ static void gsc_set_scaler(struct gsc_context *ctx, struct gsc_scaler *sc) > gsc_write(cfg, GSC_MAIN_V_RATIO); > } > > -static int gsc_dst_set_size(struct device *dev, int swap, > - struct drm_exynos_pos *pos, struct drm_exynos_sz *sz) > +static int gsc_dst_set_size(struct gsc_context *ctx, > + struct exynos_drm_ipp_buffer *buf) ditto. > { > - struct gsc_context *ctx = get_gsc_context(dev); > - struct drm_exynos_pos img_pos = *pos; > struct gsc_scaler *sc = &ctx->sc; > u32 cfg; > > - DRM_DEBUG_KMS("swap[%d]x[%d]y[%d]w[%d]h[%d]\n", > - swap, pos->x, pos->y, pos->w, pos->h); > - > - if (swap) { > - img_pos.w = pos->h; > - img_pos.h = pos->w; > - } > - > /* pixel offset */ > - cfg = (GSC_DSTIMG_OFFSET_X(pos->x) | > - GSC_DSTIMG_OFFSET_Y(pos->y)); > + cfg = (GSC_DSTIMG_OFFSET_X(buf->rect.x) | > + GSC_DSTIMG_OFFSET_Y(buf->rect.y)); > gsc_write(cfg, GSC_DSTIMG_OFFSET); > > /* scaled size */ > - cfg = (GSC_SCALED_WIDTH(img_pos.w) | GSC_SCALED_HEIGHT(img_pos.h)); > + if (ctx->rotation) > + cfg = (GSC_SCALED_WIDTH(buf->rect.h) | > + GSC_SCALED_HEIGHT(buf->rect.w)); > + else > + cfg = (GSC_SCALED_WIDTH(buf->rect.w) | > + GSC_SCALED_HEIGHT(buf->rect.h)); > gsc_write(cfg, GSC_SCALED_SIZE); > > - DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", sz->hsize, sz->vsize); > - > /* original size */ > cfg = gsc_read(GSC_DSTIMG_SIZE); > - cfg &= ~(GSC_DSTIMG_HEIGHT_MASK | > - GSC_DSTIMG_WIDTH_MASK); > - cfg |= (GSC_DSTIMG_WIDTH(sz->hsize) | > - GSC_DSTIMG_HEIGHT(sz->vsize)); > + cfg &= ~(GSC_DSTIMG_HEIGHT_MASK | GSC_DSTIMG_WIDTH_MASK); > + cfg |= GSC_DSTIMG_WIDTH(buf->buf.width) | > + GSC_DSTIMG_HEIGHT(buf->buf.height); > gsc_write(cfg, GSC_DSTIMG_SIZE); > > cfg = gsc_read(GSC_OUT_CON); > cfg &= ~GSC_OUT_RGB_TYPE_MASK; > > - DRM_DEBUG_KMS("width[%d]range[%d]\n", pos->w, sc->range); > - > - if (pos->w >= GSC_WIDTH_ITU_709) > + if (buf->rect.w >= GSC_WIDTH_ITU_709) > if (sc->range) > cfg |= GSC_OUT_RGB_HD_WIDE; > else > @@ -1112,34 +939,16 @@ static int gsc_dst_get_buf_seq(struct gsc_context *ctx) > } > > static int gsc_dst_set_buf_seq(struct gsc_context *ctx, u32 buf_id, > - enum drm_exynos_ipp_buf_type buf_type) > + bool enqueue) > { > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > - bool masked; > + bool masked = !enqueue; > u32 cfg; > u32 mask = 0x00000001 << buf_id; > int ret = 0; 'ret' variable always returns 0 so seems unnecessary, and this function doesn't need 'int' return data type. > > - DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type); > - > - mutex_lock(&ctx->lock); > - > /* mask register set */ > cfg = gsc_read(GSC_OUT_BASE_ADDR_Y_MASK); > > - switch (buf_type) { > - case IPP_BUF_ENQUEUE: > - masked = false; > - break; > - case IPP_BUF_DEQUEUE: > - masked = true; > - break; > - default: > - dev_err(ippdrv->dev, "invalid buf ctrl parameter.\n"); > - ret = -EINVAL; > - goto err_unlock; > - } > - > /* sequence id */ > cfg &= ~mask; > cfg |= masked << buf_id; > @@ -1148,87 +957,25 @@ static int gsc_dst_set_buf_seq(struct gsc_context *ctx, u32 buf_id, > gsc_write(cfg, GSC_OUT_BASE_ADDR_CR_MASK); > > /* interrupt enable */ > - if (buf_type == IPP_BUF_ENQUEUE && > - gsc_dst_get_buf_seq(ctx) >= GSC_BUF_START) > + if (enqueue && gsc_dst_get_buf_seq(ctx) >= GSC_BUF_START) > gsc_handle_irq(ctx, true, false, true); > > /* interrupt disable */ > - if (buf_type == IPP_BUF_DEQUEUE && > - gsc_dst_get_buf_seq(ctx) <= GSC_BUF_STOP) > + if (!enqueue && gsc_dst_get_buf_seq(ctx) <= GSC_BUF_STOP) > gsc_handle_irq(ctx, false, false, true); > > -err_unlock: > - mutex_unlock(&ctx->lock); > return ret; > } > > -static int gsc_dst_set_addr(struct device *dev, > - struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id, > - enum drm_exynos_ipp_buf_type buf_type) > +static int gsc_dst_set_addr(struct gsc_context *ctx, > + u32 buf_id, struct exynos_drm_ipp_buffer *buf) > { No need return data type because the return value of gsc_dst_set_buf_seq function isn't needed. > - struct gsc_context *ctx = get_gsc_context(dev); > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; > - struct drm_exynos_ipp_property *property; > - > - if (!c_node) { > - DRM_ERROR("failed to get c_node.\n"); > - return -EFAULT; > - } > - > - property = &c_node->property; > - > - DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n", > - property->prop_id, buf_id, buf_type); > - > - if (buf_id > GSC_MAX_DST) { > - dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id); > - return -EINVAL; > - } > - > /* address register set */ > - switch (buf_type) { > - case IPP_BUF_ENQUEUE: > - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_Y], > - GSC_OUT_BASE_ADDR_Y(buf_id)); > - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CB], > - GSC_OUT_BASE_ADDR_CB(buf_id)); > - gsc_write(buf_info->base[EXYNOS_DRM_PLANAR_CR], > - GSC_OUT_BASE_ADDR_CR(buf_id)); > - break; > - case IPP_BUF_DEQUEUE: > - gsc_write(0x0, GSC_OUT_BASE_ADDR_Y(buf_id)); > - gsc_write(0x0, GSC_OUT_BASE_ADDR_CB(buf_id)); > - gsc_write(0x0, GSC_OUT_BASE_ADDR_CR(buf_id)); > - break; > - default: > - /* bypass */ > - break; > - } > - > - return gsc_dst_set_buf_seq(ctx, buf_id, buf_type); > -} > - > -static struct exynos_drm_ipp_ops gsc_dst_ops = { > - .set_fmt = gsc_dst_set_fmt, > - .set_transf = gsc_dst_set_transf, > - .set_size = gsc_dst_set_size, > - .set_addr = gsc_dst_set_addr, > -}; > - > -static int gsc_clk_ctrl(struct gsc_context *ctx, bool enable) > -{ > - DRM_DEBUG_KMS("enable[%d]\n", enable); > - > - if (enable) { > - clk_prepare_enable(ctx->gsc_clk); > - ctx->suspended = false; > - } else { > - clk_disable_unprepare(ctx->gsc_clk); > - ctx->suspended = true; > - } > + gsc_write(buf->dma_addr[0], GSC_OUT_BASE_ADDR_Y(buf_id)); > + gsc_write(buf->dma_addr[1], GSC_OUT_BASE_ADDR_CB(buf_id)); > + gsc_write(buf->dma_addr[2], GSC_OUT_BASE_ADDR_CR(buf_id)); > > - return 0; > + return gsc_dst_set_buf_seq(ctx, buf_id, true); So just call gsc_dst_set_buf_seq(ctx, buf_id, true); > } > > static int gsc_get_src_buf_index(struct gsc_context *ctx) > @@ -1249,20 +996,20 @@ static int gsc_get_src_buf_index(struct gsc_context *ctx) > } > } > > + DRM_DEBUG_KMS("cfg[0x%x]curr_index[%d]buf_id[%d]\n", cfg, > + curr_index, buf_id); > + > if (buf_id == GSC_MAX_SRC) { > DRM_ERROR("failed to get in buffer index.\n"); > return -EINVAL; > } > > - ret = gsc_src_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE); > + ret = gsc_src_set_buf_seq(ctx, buf_id, false); So unnecessary error check. > if (ret < 0) { > DRM_ERROR("failed to dequeue.\n"); > return ret; > } > > - DRM_DEBUG_KMS("cfg[0x%x]curr_index[%d]buf_id[%d]\n", cfg, > - curr_index, buf_id); > - > return buf_id; > } > > @@ -1289,7 +1036,7 @@ static int gsc_get_dst_buf_index(struct gsc_context *ctx) > return -EINVAL; > } > > - ret = gsc_dst_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE); > + ret = gsc_dst_set_buf_seq(ctx, buf_id, false); ditto. > if (ret < 0) { > DRM_ERROR("failed to dequeue.\n"); > return ret; > @@ -1304,215 +1051,55 @@ static int gsc_get_dst_buf_index(struct gsc_context *ctx) > static irqreturn_t gsc_irq_handler(int irq, void *dev_id) > { > struct gsc_context *ctx = dev_id; > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; > - struct drm_exynos_ipp_event_work *event_work = > - c_node->event_work; > u32 status; > - int buf_id[EXYNOS_DRM_OPS_MAX]; > + int err = 0; > > DRM_DEBUG_KMS("gsc id[%d]\n", ctx->id); > > status = gsc_read(GSC_IRQ); > if (status & GSC_IRQ_STATUS_OR_IRQ) { > - dev_err(ippdrv->dev, "occurred overflow at %d, status 0x%x.\n", > + dev_err(ctx->dev, "occurred overflow at %d, status 0x%x.\n", > ctx->id, status); > - return IRQ_NONE; > + err = -EINVAL; > } > > if (status & GSC_IRQ_STATUS_OR_FRM_DONE) { > - dev_dbg(ippdrv->dev, "occurred frame done at %d, status 0x%x.\n", > - ctx->id, status); > - > - buf_id[EXYNOS_DRM_OPS_SRC] = gsc_get_src_buf_index(ctx); > - if (buf_id[EXYNOS_DRM_OPS_SRC] < 0) > - return IRQ_HANDLED; > - > - buf_id[EXYNOS_DRM_OPS_DST] = gsc_get_dst_buf_index(ctx); > - if (buf_id[EXYNOS_DRM_OPS_DST] < 0) > - return IRQ_HANDLED; > - > - DRM_DEBUG_KMS("buf_id_src[%d]buf_id_dst[%d]\n", > - buf_id[EXYNOS_DRM_OPS_SRC], buf_id[EXYNOS_DRM_OPS_DST]); > - > - event_work->ippdrv = ippdrv; > - event_work->buf_id[EXYNOS_DRM_OPS_SRC] = > - buf_id[EXYNOS_DRM_OPS_SRC]; > - event_work->buf_id[EXYNOS_DRM_OPS_DST] = > - buf_id[EXYNOS_DRM_OPS_DST]; > - queue_work(ippdrv->event_workq, &event_work->work); > - } > - > - return IRQ_HANDLED; > -} > - > -static int gsc_init_prop_list(struct exynos_drm_ippdrv *ippdrv) > -{ > - struct drm_exynos_ipp_prop_list *prop_list = &ippdrv->prop_list; > - > - prop_list->version = 1; > - prop_list->writeback = 1; > - prop_list->refresh_min = GSC_REFRESH_MIN; > - prop_list->refresh_max = GSC_REFRESH_MAX; > - prop_list->flip = (1 << EXYNOS_DRM_FLIP_VERTICAL) | > - (1 << EXYNOS_DRM_FLIP_HORIZONTAL); > - prop_list->degree = (1 << EXYNOS_DRM_DEGREE_0) | > - (1 << EXYNOS_DRM_DEGREE_90) | > - (1 << EXYNOS_DRM_DEGREE_180) | > - (1 << EXYNOS_DRM_DEGREE_270); > - prop_list->csc = 1; > - prop_list->crop = 1; > - prop_list->crop_max.hsize = GSC_CROP_MAX; > - prop_list->crop_max.vsize = GSC_CROP_MAX; > - prop_list->crop_min.hsize = GSC_CROP_MIN; > - prop_list->crop_min.vsize = GSC_CROP_MIN; > - prop_list->scale = 1; > - prop_list->scale_max.hsize = GSC_SCALE_MAX; > - prop_list->scale_max.vsize = GSC_SCALE_MAX; > - prop_list->scale_min.hsize = GSC_SCALE_MIN; > - prop_list->scale_min.vsize = GSC_SCALE_MIN; > - > - return 0; > -} > + int src_buf_id, dst_buf_id; > > -static inline bool gsc_check_drm_flip(enum drm_exynos_flip flip) > -{ > - switch (flip) { > - case EXYNOS_DRM_FLIP_NONE: > - case EXYNOS_DRM_FLIP_VERTICAL: > - case EXYNOS_DRM_FLIP_HORIZONTAL: > - case EXYNOS_DRM_FLIP_BOTH: > - return true; > - default: > - DRM_DEBUG_KMS("invalid flip\n"); > - return false; > - } > -} > - > -static int gsc_ippdrv_check_property(struct device *dev, > - struct drm_exynos_ipp_property *property) > -{ > - struct gsc_context *ctx = get_gsc_context(dev); > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > - struct drm_exynos_ipp_prop_list *pp = &ippdrv->prop_list; > - struct drm_exynos_ipp_config *config; > - struct drm_exynos_pos *pos; > - struct drm_exynos_sz *sz; > - bool swap; > - int i; > - > - for_each_ipp_ops(i) { > - if ((i == EXYNOS_DRM_OPS_SRC) && > - (property->cmd == IPP_CMD_WB)) > - continue; > - > - config = &property->config[i]; > - pos = &config->pos; > - sz = &config->sz; > - > - /* check for flip */ > - if (!gsc_check_drm_flip(config->flip)) { > - DRM_ERROR("invalid flip.\n"); > - goto err_property; > - } > - > - /* check for degree */ > - switch (config->degree) { > - case EXYNOS_DRM_DEGREE_90: > - case EXYNOS_DRM_DEGREE_270: > - swap = true; > - break; > - case EXYNOS_DRM_DEGREE_0: > - case EXYNOS_DRM_DEGREE_180: > - swap = false; > - break; > - default: > - DRM_ERROR("invalid degree.\n"); > - goto err_property; > - } > + dev_dbg(ctx->dev, "occurred frame done at %d, status 0x%x.\n", > + ctx->id, status); > > - /* check for buffer bound */ > - if ((pos->x + pos->w > sz->hsize) || > - (pos->y + pos->h > sz->vsize)) { > - DRM_ERROR("out of buf bound.\n"); > - goto err_property; > - } > + src_buf_id = gsc_get_src_buf_index(ctx); > + dst_buf_id = gsc_get_dst_buf_index(ctx); > > - /* check for crop */ > - if ((i == EXYNOS_DRM_OPS_SRC) && (pp->crop)) { > - if (swap) { > - if ((pos->h < pp->crop_min.hsize) || > - (sz->vsize > pp->crop_max.hsize) || > - (pos->w < pp->crop_min.vsize) || > - (sz->hsize > pp->crop_max.vsize)) { > - DRM_ERROR("out of crop size.\n"); > - goto err_property; > - } > - } else { > - if ((pos->w < pp->crop_min.hsize) || > - (sz->hsize > pp->crop_max.hsize) || > - (pos->h < pp->crop_min.vsize) || > - (sz->vsize > pp->crop_max.vsize)) { > - DRM_ERROR("out of crop size.\n"); > - goto err_property; > - } > - } > - } > + DRM_DEBUG_KMS("buf_id_src[%d]buf_id_dst[%d]\n", src_buf_id, > + dst_buf_id); > > - /* check for scale */ > - if ((i == EXYNOS_DRM_OPS_DST) && (pp->scale)) { > - if (swap) { > - if ((pos->h < pp->scale_min.hsize) || > - (sz->vsize > pp->scale_max.hsize) || > - (pos->w < pp->scale_min.vsize) || > - (sz->hsize > pp->scale_max.vsize)) { > - DRM_ERROR("out of scale size.\n"); > - goto err_property; > - } > - } else { > - if ((pos->w < pp->scale_min.hsize) || > - (sz->hsize > pp->scale_max.hsize) || > - (pos->h < pp->scale_min.vsize) || > - (sz->vsize > pp->scale_max.vsize)) { > - DRM_ERROR("out of scale size.\n"); > - goto err_property; > - } > - } > - } > + if (src_buf_id < 0 || dst_buf_id < 0) > + err = -EINVAL; > } > > - return 0; > + if (ctx->task) { > + struct exynos_drm_ipp_task *task = ctx->task; > > -err_property: > - for_each_ipp_ops(i) { > - if ((i == EXYNOS_DRM_OPS_SRC) && > - (property->cmd == IPP_CMD_WB)) > - continue; > - > - config = &property->config[i]; > - pos = &config->pos; > - sz = &config->sz; > - > - DRM_ERROR("[%s]f[%d]r[%d]pos[%d %d %d %d]sz[%d %d]\n", > - i ? "dst" : "src", config->flip, config->degree, > - pos->x, pos->y, pos->w, pos->h, > - sz->hsize, sz->vsize); > + ctx->task = NULL; > + pm_runtime_mark_last_busy(ctx->dev); > + pm_runtime_put_autosuspend(ctx->dev); > + exynos_drm_ipp_task_done(task, err); > } > > - return -EINVAL; > + return IRQ_HANDLED; > } > > - > -static int gsc_ippdrv_reset(struct device *dev) > +static int gsc_reset(struct gsc_context *ctx) > { > - struct gsc_context *ctx = get_gsc_context(dev); > struct gsc_scaler *sc = &ctx->sc; > int ret; > > /* reset h/w block */ > ret = gsc_sw_reset(ctx); > if (ret < 0) { > - dev_err(dev, "failed to reset hardware.\n"); > + dev_err(ctx->dev, "failed to reset hardware.\n"); > return ret; > } > > @@ -1523,96 +1110,29 @@ static int gsc_ippdrv_reset(struct device *dev) > return 0; > } > > -static int gsc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd) > +static int gsc_start(struct gsc_context *ctx) No need 'int' return data type. > { > - struct gsc_context *ctx = get_gsc_context(dev); > - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; > - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; > - struct drm_exynos_ipp_property *property; > - struct drm_exynos_ipp_config *config; > - struct drm_exynos_pos img_pos[EXYNOS_DRM_OPS_MAX]; > - struct drm_exynos_ipp_set_wb set_wb; > u32 cfg; > - int ret, i; > - > - DRM_DEBUG_KMS("cmd[%d]\n", cmd); > - > - if (!c_node) { > - DRM_ERROR("failed to get c_node.\n"); > - return -EINVAL; > - } > - > - property = &c_node->property; > > gsc_handle_irq(ctx, true, false, true); > > - for_each_ipp_ops(i) { > - config = &property->config[i]; > - img_pos[i] = config->pos; > - } > + /* enable one shot */ > + cfg = gsc_read(GSC_ENABLE); > + cfg &= ~(GSC_ENABLE_ON_CLEAR_MASK | > + GSC_ENABLE_CLK_GATE_MODE_MASK); > + cfg |= GSC_ENABLE_ON_CLEAR_ONESHOT; > + gsc_write(cfg, GSC_ENABLE); > > - switch (cmd) { > - case IPP_CMD_M2M: > - /* enable one shot */ > - cfg = gsc_read(GSC_ENABLE); > - cfg &= ~(GSC_ENABLE_ON_CLEAR_MASK | > - GSC_ENABLE_CLK_GATE_MODE_MASK); > - cfg |= GSC_ENABLE_ON_CLEAR_ONESHOT; > - gsc_write(cfg, GSC_ENABLE); > - > - /* src dma memory */ > - cfg = gsc_read(GSC_IN_CON); > - cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK); > - cfg |= GSC_IN_PATH_MEMORY; > - gsc_write(cfg, GSC_IN_CON); > - > - /* dst dma memory */ > - cfg = gsc_read(GSC_OUT_CON); > - cfg |= GSC_OUT_PATH_MEMORY; > - gsc_write(cfg, GSC_OUT_CON); > - break; > - case IPP_CMD_WB: > - set_wb.enable = 1; > - set_wb.refresh = property->refresh_rate; > - gsc_set_gscblk_fimd_wb(ctx, set_wb.enable); > - exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb); > - > - /* src local path */ > - cfg = gsc_read(GSC_IN_CON); > - cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK); > - cfg |= (GSC_IN_PATH_LOCAL | GSC_IN_LOCAL_FIMD_WB); > - gsc_write(cfg, GSC_IN_CON); > - > - /* dst dma memory */ > - cfg = gsc_read(GSC_OUT_CON); > - cfg |= GSC_OUT_PATH_MEMORY; > - gsc_write(cfg, GSC_OUT_CON); > - break; > - case IPP_CMD_OUTPUT: > - /* src dma memory */ > - cfg = gsc_read(GSC_IN_CON); > - cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK); > - cfg |= GSC_IN_PATH_MEMORY; > - gsc_write(cfg, GSC_IN_CON); > - > - /* dst local path */ > - cfg = gsc_read(GSC_OUT_CON); > - cfg |= GSC_OUT_PATH_MEMORY; > - gsc_write(cfg, GSC_OUT_CON); > - break; > - default: > - ret = -EINVAL; > - dev_err(dev, "invalid operations.\n"); > - return ret; > - } > + /* src dma memory */ > + cfg = gsc_read(GSC_IN_CON); > + cfg &= ~(GSC_IN_PATH_MASK | GSC_IN_LOCAL_SEL_MASK); > + cfg |= GSC_IN_PATH_MEMORY; > + gsc_write(cfg, GSC_IN_CON); > > - ret = gsc_set_prescaler(ctx, &ctx->sc, > - &img_pos[EXYNOS_DRM_OPS_SRC], > - &img_pos[EXYNOS_DRM_OPS_DST]); > - if (ret) { > - dev_err(dev, "failed to set prescaler.\n"); > - return ret; > - } > + /* dst dma memory */ > + cfg = gsc_read(GSC_OUT_CON); > + cfg |= GSC_OUT_PATH_MEMORY; > + gsc_write(cfg, GSC_OUT_CON); > > gsc_set_scaler(ctx, &ctx->sc); > > @@ -1623,66 +1143,132 @@ static int gsc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd) > return 0; > } > > -static void gsc_ippdrv_stop(struct device *dev, enum drm_exynos_ipp_cmd cmd) > +static int gsc_commit(struct exynos_drm_ipp *ipp, > + struct exynos_drm_ipp_task *task) > { > - struct gsc_context *ctx = get_gsc_context(dev); > - struct drm_exynos_ipp_set_wb set_wb = {0, 0}; > - u32 cfg; > + struct gsc_context *ctx = container_of(ipp, struct gsc_context, ipp); > + > + pm_runtime_get_sync(ctx->dev); > + ctx->task = task; > + > + gsc_reset(ctx); > + gsc_src_set_fmt(ctx, task->src.buf.fourcc); This function can return a error so please check a error here. Thanks, Inki Dae > + gsc_src_set_transf(ctx, task->transform.rotation); > + gsc_src_set_size(ctx, &task->src); > + gsc_src_set_addr(ctx, 0, &task->src); > + gsc_dst_set_fmt(ctx, task->dst.buf.fourcc); > + gsc_dst_set_size(ctx, &task->dst); > + gsc_dst_set_addr(ctx, 0, &task->dst); > + gsc_set_prescaler(ctx, &ctx->sc, &task->src.rect, &task->dst.rect); > + gsc_start(ctx); > > - DRM_DEBUG_KMS("cmd[%d]\n", cmd); > + return 0; > +} _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel ^ permalink raw reply [flat|nested] 20+ messages in thread
[parent not found: <CGME20171023075450eucas1p24923c2a0793ac24364b97770dadd56e0@eucas1p2.samsung.com>]
* [PATCH v4 5/9] drm/exynos: Add generic support for devices shared with V4L2 subsystem [not found] ` <CGME20171023075450eucas1p24923c2a0793ac24364b97770dadd56e0@eucas1p2.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski 2017-11-01 6:26 ` Inki Dae 0 siblings, 1 reply; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Marek Szyprowski, Inki Dae, Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon Some hardware modules, like FIMC in Exynos4 series are shared between V4L2 (camera support) and DRM (memory-to-memory processing) subsystems. This patch provides a simple check to let such drivers to be used in the driver components framework. Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> --- drivers/gpu/drm/exynos/exynos_drm_drv.c | 17 ++++++++++++++++- drivers/gpu/drm/exynos/exynos_drm_drv.h | 2 ++ 2 files changed, 18 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c index cac0d84385d3..60ae6ae06eb2 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c @@ -216,6 +216,7 @@ struct exynos_drm_driver_info { #define DRM_COMPONENT_DRIVER BIT(0) /* supports component framework */ #define DRM_VIRTUAL_DEVICE BIT(1) /* create virtual platform device */ #define DRM_DMA_DEVICE BIT(2) /* can be used for dma allocations */ +#define DRM_SHARED_DEVICE BIT(3) /* devices shared with V4L2 subsystem */ #define DRV_PTR(drv, cond) (IS_ENABLED(cond) ? &drv : NULL) @@ -267,6 +268,17 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { } }; +int exynos_drm_check_shared_device(struct device *dev) +{ + /* + * Exynos DRM drivers handle only devices that support + * the LCD Writeback data path, rest is handled by V4L2 driver + */ + if (!of_property_read_bool(dev->of_node, "samsung,lcd-wb")) + return -ENODEV; + return 0; +} + static int compare_dev(struct device *dev, void *data) { return dev == (struct device *)data; @@ -288,7 +300,10 @@ static struct component_match *exynos_drm_match_add(struct device *dev) &info->driver->driver, (void *)platform_bus_type.match))) { put_device(p); - component_match_add(dev, &match, compare_dev, d); + + if (!(info->flags & DRM_SHARED_DEVICE) || + exynos_drm_check_shared_device(d) == 0) + component_match_add(dev, &match, compare_dev, d); p = d; } put_device(p); diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h b/drivers/gpu/drm/exynos/exynos_drm_drv.h index b47f810d64d2..8b3b31d35168 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.h +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h @@ -275,6 +275,8 @@ static inline int exynos_dpi_bind(struct drm_device *dev, } #endif +int exynos_drm_check_shared_device(struct device *dev); + int exynos_atomic_commit(struct drm_device *dev, struct drm_atomic_state *state, bool nonblock); int exynos_atomic_check(struct drm_device *dev, struct drm_atomic_state *state); -- 2.14.2 ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH v4 5/9] drm/exynos: Add generic support for devices shared with V4L2 subsystem 2017-10-23 7:54 ` [PATCH v4 5/9] drm/exynos: Add generic support for devices shared with V4L2 subsystem Marek Szyprowski @ 2017-11-01 6:26 ` Inki Dae 2017-11-03 12:28 ` Marek Szyprowski 0 siblings, 1 reply; 20+ messages in thread From: Inki Dae @ 2017-11-01 6:26 UTC (permalink / raw) To: Marek Szyprowski, dri-devel, linux-samsung-soc Cc: Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon 2017년 10월 23일 16:54에 Marek Szyprowski 이(가) 쓴 글: > Some hardware modules, like FIMC in Exynos4 series are shared between > V4L2 (camera support) and DRM (memory-to-memory processing) subsystems. > This patch provides a simple check to let such drivers to be used in the > driver components framework. > > Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> > --- > drivers/gpu/drm/exynos/exynos_drm_drv.c | 17 ++++++++++++++++- > drivers/gpu/drm/exynos/exynos_drm_drv.h | 2 ++ > 2 files changed, 18 insertions(+), 1 deletion(-) > > diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c > index cac0d84385d3..60ae6ae06eb2 100644 > --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c > +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c > @@ -216,6 +216,7 @@ struct exynos_drm_driver_info { > #define DRM_COMPONENT_DRIVER BIT(0) /* supports component framework */ > #define DRM_VIRTUAL_DEVICE BIT(1) /* create virtual platform device */ > #define DRM_DMA_DEVICE BIT(2) /* can be used for dma allocations */ > +#define DRM_SHARED_DEVICE BIT(3) /* devices shared with V4L2 subsystem */ > > #define DRV_PTR(drv, cond) (IS_ENABLED(cond) ? &drv : NULL) > > @@ -267,6 +268,17 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { > } > }; > > +int exynos_drm_check_shared_device(struct device *dev) > +{ > + /* > + * Exynos DRM drivers handle only devices that support > + * the LCD Writeback data path, rest is handled by V4L2 driver > + */ > + if (!of_property_read_bool(dev->of_node, "samsung,lcd-wb")) > + return -ENODEV; > + return 0; > +} > + > static int compare_dev(struct device *dev, void *data) > { > return dev == (struct device *)data; > @@ -288,7 +300,10 @@ static struct component_match *exynos_drm_match_add(struct device *dev) > &info->driver->driver, > (void *)platform_bus_type.match))) { > put_device(p); > - component_match_add(dev, &match, compare_dev, d); > + > + if (!(info->flags & DRM_SHARED_DEVICE) || > + exynos_drm_check_shared_device(d) == 0) > + component_match_add(dev, &match, compare_dev, d); Seems above line make fimc device driver to be bound only when fimc device node has "samsung,lcd-wb" property. However, Exynos DRM FIMC driver is able to various transfomations such as color space conversion, scaling up/down and rotation. And this driver is used as mem to mem device driver. However, 'writeback' feature means 'dma to memory', which delivers one blended image in display controller into system memory. Thanks, Inki Dae > p = d; > } > put_device(p); > diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h b/drivers/gpu/drm/exynos/exynos_drm_drv.h > index b47f810d64d2..8b3b31d35168 100644 > --- a/drivers/gpu/drm/exynos/exynos_drm_drv.h > +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h > @@ -275,6 +275,8 @@ static inline int exynos_dpi_bind(struct drm_device *dev, > } > #endif > > +int exynos_drm_check_shared_device(struct device *dev); > + > int exynos_atomic_commit(struct drm_device *dev, struct drm_atomic_state *state, > bool nonblock); > int exynos_atomic_check(struct drm_device *dev, struct drm_atomic_state *state); > ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH v4 5/9] drm/exynos: Add generic support for devices shared with V4L2 subsystem 2017-11-01 6:26 ` Inki Dae @ 2017-11-03 12:28 ` Marek Szyprowski 2017-11-06 1:20 ` Inki Dae 0 siblings, 1 reply; 20+ messages in thread From: Marek Szyprowski @ 2017-11-03 12:28 UTC (permalink / raw) To: Inki Dae, dri-devel, linux-samsung-soc Cc: Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon Hi Inki, On 2017-11-01 07:26, Inki Dae wrote: > 2017년 10월 23일 16:54에 Marek Szyprowski 이(가) 쓴 글: >> Some hardware modules, like FIMC in Exynos4 series are shared between >> V4L2 (camera support) and DRM (memory-to-memory processing) subsystems. >> This patch provides a simple check to let such drivers to be used in the >> driver components framework. >> >> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> >> --- >> drivers/gpu/drm/exynos/exynos_drm_drv.c | 17 ++++++++++++++++- >> drivers/gpu/drm/exynos/exynos_drm_drv.h | 2 ++ >> 2 files changed, 18 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c >> index cac0d84385d3..60ae6ae06eb2 100644 >> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c >> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c >> @@ -216,6 +216,7 @@ struct exynos_drm_driver_info { >> #define DRM_COMPONENT_DRIVER BIT(0) /* supports component framework */ >> #define DRM_VIRTUAL_DEVICE BIT(1) /* create virtual platform device */ >> #define DRM_DMA_DEVICE BIT(2) /* can be used for dma allocations */ >> +#define DRM_SHARED_DEVICE BIT(3) /* devices shared with V4L2 subsystem */ >> >> #define DRV_PTR(drv, cond) (IS_ENABLED(cond) ? &drv : NULL) >> >> @@ -267,6 +268,17 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { >> } >> }; >> >> +int exynos_drm_check_shared_device(struct device *dev) >> +{ >> + /* >> + * Exynos DRM drivers handle only devices that support >> + * the LCD Writeback data path, rest is handled by V4L2 driver >> + */ >> + if (!of_property_read_bool(dev->of_node, "samsung,lcd-wb")) >> + return -ENODEV; >> + return 0; >> +} >> + >> static int compare_dev(struct device *dev, void *data) >> { >> return dev == (struct device *)data; >> @@ -288,7 +300,10 @@ static struct component_match *exynos_drm_match_add(struct device *dev) >> &info->driver->driver, >> (void *)platform_bus_type.match))) { >> put_device(p); >> - component_match_add(dev, &match, compare_dev, d); >> + >> + if (!(info->flags & DRM_SHARED_DEVICE) || >> + exynos_drm_check_shared_device(d) == 0) >> + component_match_add(dev, &match, compare_dev, d); > Seems above line make fimc device driver to be bound only when fimc device node has "samsung,lcd-wb" property. However, Exynos DRM FIMC driver is able to various transfomations such as color space conversion, scaling up/down and rotation. > And this driver is used as mem to mem device driver. However, 'writeback' feature means 'dma to memory', which delivers one blended image in display controller into system memory. This patch is only to bind fimc.2 and fimc.3 devices to DRM and fimc.0 and fimc.1 to V4L2. Exactly the same checks are in V4l2 and old DRM FIMC drivers. > > Thanks, > Inki Dae > >> p = d; >> } >> put_device(p); >> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h b/drivers/gpu/drm/exynos/exynos_drm_drv.h >> index b47f810d64d2..8b3b31d35168 100644 >> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.h >> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h >> @@ -275,6 +275,8 @@ static inline int exynos_dpi_bind(struct drm_device *dev, >> } >> #endif >> >> +int exynos_drm_check_shared_device(struct device *dev); >> + >> int exynos_atomic_commit(struct drm_device *dev, struct drm_atomic_state *state, >> bool nonblock); >> int exynos_atomic_check(struct drm_device *dev, struct drm_atomic_state *state); >> > Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH v4 5/9] drm/exynos: Add generic support for devices shared with V4L2 subsystem 2017-11-03 12:28 ` Marek Szyprowski @ 2017-11-06 1:20 ` Inki Dae 2017-11-07 9:07 ` Marek Szyprowski 0 siblings, 1 reply; 20+ messages in thread From: Inki Dae @ 2017-11-06 1:20 UTC (permalink / raw) To: Marek Szyprowski, dri-devel, linux-samsung-soc Cc: Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon Hi Marek, 2017년 11월 03일 21:28에 Marek Szyprowski 이(가) 쓴 글: > Hi Inki, > > On 2017-11-01 07:26, Inki Dae wrote: >> 2017년 10월 23일 16:54에 Marek Szyprowski 이(가) 쓴 글: >>> Some hardware modules, like FIMC in Exynos4 series are shared between >>> V4L2 (camera support) and DRM (memory-to-memory processing) subsystems. >>> This patch provides a simple check to let such drivers to be used in the >>> driver components framework. >>> >>> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> >>> --- >>> drivers/gpu/drm/exynos/exynos_drm_drv.c | 17 ++++++++++++++++- >>> drivers/gpu/drm/exynos/exynos_drm_drv.h | 2 ++ >>> 2 files changed, 18 insertions(+), 1 deletion(-) >>> >>> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c >>> index cac0d84385d3..60ae6ae06eb2 100644 >>> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c >>> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c >>> @@ -216,6 +216,7 @@ struct exynos_drm_driver_info { >>> #define DRM_COMPONENT_DRIVER BIT(0) /* supports component framework */ >>> #define DRM_VIRTUAL_DEVICE BIT(1) /* create virtual platform device */ >>> #define DRM_DMA_DEVICE BIT(2) /* can be used for dma allocations */ >>> +#define DRM_SHARED_DEVICE BIT(3) /* devices shared with V4L2 subsystem */ >>> #define DRV_PTR(drv, cond) (IS_ENABLED(cond) ? &drv : NULL) >>> @@ -267,6 +268,17 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { >>> } >>> }; >>> +int exynos_drm_check_shared_device(struct device *dev) >>> +{ >>> + /* >>> + * Exynos DRM drivers handle only devices that support >>> + * the LCD Writeback data path, rest is handled by V4L2 driver >>> + */ >>> + if (!of_property_read_bool(dev->of_node, "samsung,lcd-wb")) >>> + return -ENODEV; >>> + return 0; >>> +} >>> + >>> static int compare_dev(struct device *dev, void *data) >>> { >>> return dev == (struct device *)data; >>> @@ -288,7 +300,10 @@ static struct component_match *exynos_drm_match_add(struct device *dev) >>> &info->driver->driver, >>> (void *)platform_bus_type.match))) { >>> put_device(p); >>> - component_match_add(dev, &match, compare_dev, d); >>> + >>> + if (!(info->flags & DRM_SHARED_DEVICE) || >>> + exynos_drm_check_shared_device(d) == 0) >>> + component_match_add(dev, &match, compare_dev, d); >> Seems above line make fimc device driver to be bound only when fimc device node has "samsung,lcd-wb" property. However, Exynos DRM FIMC driver is able to various transfomations such as color space conversion, scaling up/down and rotation. >> And this driver is used as mem to mem device driver. However, 'writeback' feature means 'dma to memory', which delivers one blended image in display controller into system memory. > > This patch is only to bind fimc.2 and fimc.3 devices to DRM and fimc.0 > and fimc.1 to V4L2. Exactly the same checks are in V4l2 and old DRM FIMC > drivers. Indeed. No change for behaivor. When Sylwester posted old version[1] of DRM fimc device tree support, seems he and even me missed the behaiver of DRM FIMC driver. According to below patch description, it says, "The DRM driver uses this interface for setting up the FIFO data link between FIMD and FIMC IP blocks" We thought we could use 'samsung,lcd-wb' property to distinguish FIMC devices for V4L2 and ones for DRM - only fimc 2 and fimc 3 - have 'samsung'lcd-wb' property'. but it's not true. DRM FIMC driver is used as a general post processing hardware such as color space conversion, scaling up/down and rotation operations. In FIMC device's perspective, 'samsung,lcd-wb' means 'dma to memory' - as I mentioned before, which delivers one blended image in display controller into system memory and it's just one of several features FIMC has - so it doesn't make sense. For example, User - some device tree - can remove 'samsung,lcd-wb' property from fimc device node because this property is optional. In this case, binding of DRM FIMC driver will fail even though FIMC driver can provide other functions. We shouldn't make FIMC device to be limited with just LCD write back feature and we need to find a good way to distinguish FIMC devices for V4L2 and DRM. Anyway, we are trying to merge new version of IPP driver which is really a big change so how about making sure the behaiver of this driver, not depending on 'old version'? I think we need to take care of this carefully because ABI interface could be changed according to our decision. To Sylwester, Could you give us your option? According to Exynos4412 datasheet, 'isp-wb' can go to input of FIMC 0, 1 and 2, and 'lcd-wb' can go to input of FIMC 2 and 3 if Figure 43-1 in the datasheet is right. So it says, 1. FIMC 0 and 1 could be used for isp-wb. 2. FIMC 2 could be used for isp-wb and lcd-wb. 3. FIMC 3 could be used for lcd-wb. But you updated binding document as like below, "Optional properties: ... - samsung,isp-wb: this property must be present if the IP block has the ISP writeback input. - samsung,lcd-wb: this property must be present if the IP block has the LCD writeback input." This would mean that all FIMC devices - 0 to 3 - could be used for isp-wb or lcd-wb, and it wouldn't make sense. My opinion is, 1. we could dedicate FIMC 0 and 1 for isp-wb, and FIMC 3 for lcd-wb, and binding document may be changed like below, - For FIMC 0 and 1, isp-wb property should be a required property, not optional. - For FIMC 3, lcd-wb propery should be a required property, not optionl. 2. we could share FIMC 2 for isp-wb and lcd-wb. - For FIMC 2, isp-wb and lcd-wb should be required properties but only one driver - V4L2 or DRM driver - can be bound. [1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=5186fc5e8eda184935467aa84295b2897166fecd Thanks, Inki Dae > >> >> Thanks, >> Inki Dae >> >>> p = d; >>> } >>> put_device(p); >>> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h b/drivers/gpu/drm/exynos/exynos_drm_drv.h >>> index b47f810d64d2..8b3b31d35168 100644 >>> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.h >>> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h >>> @@ -275,6 +275,8 @@ static inline int exynos_dpi_bind(struct drm_device *dev, >>> } >>> #endif >>> +int exynos_drm_check_shared_device(struct device *dev); >>> + >>> int exynos_atomic_commit(struct drm_device *dev, struct drm_atomic_state *state, >>> bool nonblock); >>> int exynos_atomic_check(struct drm_device *dev, struct drm_atomic_state *state); >>> >> > > Best regards ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH v4 5/9] drm/exynos: Add generic support for devices shared with V4L2 subsystem 2017-11-06 1:20 ` Inki Dae @ 2017-11-07 9:07 ` Marek Szyprowski 0 siblings, 0 replies; 20+ messages in thread From: Marek Szyprowski @ 2017-11-07 9:07 UTC (permalink / raw) To: Inki Dae, dri-devel, linux-samsung-soc Cc: Tobias Jakobi, Bartlomiej Zolnierkiewicz, Seung-Woo Kim, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon Hi Inki, On 2017-11-06 02:20, Inki Dae wrote: > 2017년 11월 03일 21:28에 Marek Szyprowski 이(가) 쓴 글: >> On 2017-11-01 07:26, Inki Dae wrote: >>> 2017년 10월 23일 16:54에 Marek Szyprowski 이(가) 쓴 글: >>>> Some hardware modules, like FIMC in Exynos4 series are shared between >>>> V4L2 (camera support) and DRM (memory-to-memory processing) subsystems. >>>> This patch provides a simple check to let such drivers to be used in the >>>> driver components framework. >>>> >>>> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> >>>> --- >>>> drivers/gpu/drm/exynos/exynos_drm_drv.c | 17 ++++++++++++++++- >>>> drivers/gpu/drm/exynos/exynos_drm_drv.h | 2 ++ >>>> 2 files changed, 18 insertions(+), 1 deletion(-) >>>> >>>> diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c >>>> index cac0d84385d3..60ae6ae06eb2 100644 >>>> --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c >>>> +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c >>>> @@ -216,6 +216,7 @@ struct exynos_drm_driver_info { >>>> #define DRM_COMPONENT_DRIVER BIT(0) /* supports component framework */ >>>> #define DRM_VIRTUAL_DEVICE BIT(1) /* create virtual platform device */ >>>> #define DRM_DMA_DEVICE BIT(2) /* can be used for dma allocations */ >>>> +#define DRM_SHARED_DEVICE BIT(3) /* devices shared with V4L2 subsystem */ >>>> #define DRV_PTR(drv, cond) (IS_ENABLED(cond) ? &drv : NULL) >>>> @@ -267,6 +268,17 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { >>>> } >>>> }; >>>> +int exynos_drm_check_shared_device(struct device *dev) >>>> +{ >>>> + /* >>>> + * Exynos DRM drivers handle only devices that support >>>> + * the LCD Writeback data path, rest is handled by V4L2 driver >>>> + */ >>>> + if (!of_property_read_bool(dev->of_node, "samsung,lcd-wb")) >>>> + return -ENODEV; >>>> + return 0; >>>> +} >>>> + >>>> static int compare_dev(struct device *dev, void *data) >>>> { >>>> return dev == (struct device *)data; >>>> @@ -288,7 +300,10 @@ static struct component_match *exynos_drm_match_add(struct device *dev) >>>> &info->driver->driver, >>>> (void *)platform_bus_type.match))) { >>>> put_device(p); >>>> - component_match_add(dev, &match, compare_dev, d); >>>> + >>>> + if (!(info->flags & DRM_SHARED_DEVICE) || >>>> + exynos_drm_check_shared_device(d) == 0) >>>> + component_match_add(dev, &match, compare_dev, d); >>> Seems above line make fimc device driver to be bound only when fimc device node has "samsung,lcd-wb" property. However, Exynos DRM FIMC driver is able to various transfomations such as color space conversion, scaling up/down and rotation. >>> And this driver is used as mem to mem device driver. However, 'writeback' feature means 'dma to memory', which delivers one blended image in display controller into system memory. >> This patch is only to bind fimc.2 and fimc.3 devices to DRM and fimc.0 >> and fimc.1 to V4L2. Exactly the same checks are in V4l2 and old DRM FIMC >> drivers. > Indeed. No change for behaivor. > > When Sylwester posted old version[1] of DRM fimc device tree support, seems he and even me missed the behaiver of DRM FIMC driver. > According to below patch description, it says, > "The DRM driver uses this interface for setting up the FIFO data link between FIMD and FIMC IP blocks" > > We thought we could use 'samsung,lcd-wb' property to distinguish FIMC devices for V4L2 and ones for DRM - only fimc 2 and fimc 3 - have 'samsung'lcd-wb' property'. > > but it's not true. DRM FIMC driver is used as a general post processing hardware such as color space conversion, scaling up/down and rotation operations. > In FIMC device's perspective, 'samsung,lcd-wb' means 'dma to memory' - as I mentioned before, which delivers one blended image in display controller into system memory and it's just one of several features FIMC has - so it doesn't make sense. > > For example, > User - some device tree - can remove 'samsung,lcd-wb' property from fimc device node because this property is optional. In this case, binding of DRM FIMC driver will fail even though FIMC driver can provide other functions. > We shouldn't make FIMC device to be limited with just LCD write back feature and we need to find a good way to distinguish FIMC devices for V4L2 and DRM. > > Anyway, we are trying to merge new version of IPP driver which is really a big change so how about making sure the behaiver of this driver, not depending on 'old version'? > I think we need to take care of this carefully because ABI interface could be changed according to our decision. > > To Sylwester, > Could you give us your option? > > According to Exynos4412 datasheet, > 'isp-wb' can go to input of FIMC 0, 1 and 2, and 'lcd-wb' can go to input of FIMC 2 and 3 if Figure 43-1 in the datasheet is right. > > So it says, > 1. FIMC 0 and 1 could be used for isp-wb. > 2. FIMC 2 could be used for isp-wb and lcd-wb. > 3. FIMC 3 could be used for lcd-wb. > > But you updated binding document as like below, > "Optional properties: > ... > - samsung,isp-wb: this property must be present if the IP block has the ISP > writeback input. > - samsung,lcd-wb: this property must be present if the IP block has the LCD > writeback input." > > This would mean that all FIMC devices - 0 to 3 - could be used for isp-wb or lcd-wb, and it wouldn't make sense. > > My opinion is, > 1. we could dedicate FIMC 0 and 1 for isp-wb, and FIMC 3 for lcd-wb, and binding document may be changed like below, > - For FIMC 0 and 1, isp-wb property should be a required property, not optional. > - For FIMC 3, lcd-wb propery should be a required property, not optionl. > 2. we could share FIMC 2 for isp-wb and lcd-wb. > - For FIMC 2, isp-wb and lcd-wb should be required properties but only one driver - V4L2 or DRM driver - can be bound. It is a matter of use-cases which FIMC device is used for camera/v4l2 and which for m2m processing. I think this should be simply configurable by some kernel/module option(s), which defaults to current setup (fimc.0 & fimc.1 for v4l2, fimc.2 & fimc.3 for drm). Best regards -- Marek Szyprowski, PhD Samsung R&D Institute Poland _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/dri-devel ^ permalink raw reply [flat|nested] 20+ messages in thread
[parent not found: <CGME20171023075451eucas1p1a833d9c805f6b7dc1b984e59afce23f3@eucas1p1.samsung.com>]
* [PATCH v4 6/9] drm/exynos: fimc: Convert driver to IPP v2 core API [not found] ` <CGME20171023075451eucas1p1a833d9c805f6b7dc1b984e59afce23f3@eucas1p1.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski 0 siblings, 0 replies; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Marek Szyprowski, Inki Dae, Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon This patch adapts Exynos DRM FIMC driver to new IPP v2 core API. The side effect of this conversion is a switch to driver component API to register properly in the Exynos DRM core. Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> --- drivers/gpu/drm/exynos/Kconfig | 2 +- drivers/gpu/drm/exynos/exynos_drm_drv.c | 1 + drivers/gpu/drm/exynos/exynos_drm_fimc.c | 930 ++++++++++--------------------- drivers/gpu/drm/exynos/exynos_drm_fimc.h | 23 - 4 files changed, 282 insertions(+), 674 deletions(-) delete mode 100644 drivers/gpu/drm/exynos/exynos_drm_fimc.h diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig index 9b66a9838dca..73f06a3a8bec 100644 --- a/drivers/gpu/drm/exynos/Kconfig +++ b/drivers/gpu/drm/exynos/Kconfig @@ -99,7 +99,7 @@ config DRM_EXYNOS_IPP config DRM_EXYNOS_FIMC bool "FIMC" - depends on BROKEN && MFD_SYSCON + select DRM_EXYNOS_IPP help Choose this option if you want to use Exynos FIMC for DRM. diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c index 60ae6ae06eb2..4afeffa024f3 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c @@ -256,6 +256,7 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { DRV_PTR(g2d_driver, CONFIG_DRM_EXYNOS_G2D), }, { DRV_PTR(fimc_driver, CONFIG_DRM_EXYNOS_FIMC), + DRM_COMPONENT_DRIVER | DRM_SHARED_DEVICE, }, { DRV_PTR(rotator_driver, CONFIG_DRM_EXYNOS_ROTATOR), DRM_COMPONENT_DRIVER diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimc.c b/drivers/gpu/drm/exynos/exynos_drm_fimc.c index 5b18b5c5fdf2..870f14b775b5 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_fimc.c +++ b/drivers/gpu/drm/exynos/exynos_drm_fimc.c @@ -12,6 +12,7 @@ * */ #include <linux/kernel.h> +#include <linux/component.h> #include <linux/platform_device.h> #include <linux/mfd/syscon.h> #include <linux/regmap.h> @@ -24,8 +25,9 @@ #include <drm/exynos_drm.h> #include "regs-fimc.h" #include "exynos_drm_drv.h" +#include "exynos_drm_iommu.h" #include "exynos_drm_ipp.h" -#include "exynos_drm_fimc.h" + /* * FIMC stands for Fully Interactive Mobile Camera and @@ -33,23 +35,6 @@ * input DMA reads image data from the memory. * output DMA writes image data to memory. * FIMC supports image rotation and image effect functions. - * - * M2M operation : supports crop/scale/rotation/csc so on. - * Memory ----> FIMC H/W ----> Memory. - * Writeback operation : supports cloned screen with FIMD. - * FIMD ----> FIMC H/W ----> Memory. - * Output operation : supports direct display using local path. - * Memory ----> FIMC H/W ----> FIMD. - */ - -/* - * TODO - * 1. check suspend/resume api if needed. - * 2. need to check use case platform_device_id. - * 3. check src/dst size with, height. - * 4. added check_prepare api for right register. - * 5. need to add supported list in prop_list. - * 6. check prescaler/scaler optimization. */ #define FIMC_MAX_DEVS 4 @@ -59,29 +44,15 @@ #define FIMC_BUF_STOP 1 #define FIMC_BUF_START 2 #define FIMC_WIDTH_ITU_709 1280 -#define FIMC_REFRESH_MAX 60 -#define FIMC_REFRESH_MIN 12 -#define FIMC_CROP_MAX 8192 -#define FIMC_CROP_MIN 32 -#define FIMC_SCALE_MAX 4224 -#define FIMC_SCALE_MIN 32 +#define FIMC_AUTOSUSPEND_DELAY 2000 #define get_fimc_context(dev) platform_get_drvdata(to_platform_device(dev)) -#define get_ctx_from_ippdrv(ippdrv) container_of(ippdrv,\ - struct fimc_context, ippdrv); -enum fimc_wb { - FIMC_WB_NONE, - FIMC_WB_A, - FIMC_WB_B, -}; enum { FIMC_CLK_LCLK, FIMC_CLK_GATE, FIMC_CLK_WB_A, FIMC_CLK_WB_B, - FIMC_CLK_MUX, - FIMC_CLK_PARENT, FIMC_CLKS_MAX }; @@ -90,8 +61,6 @@ static const char * const fimc_clock_names[] = { [FIMC_CLK_GATE] = "fimc", [FIMC_CLK_WB_A] = "pxl_async0", [FIMC_CLK_WB_B] = "pxl_async1", - [FIMC_CLK_MUX] = "mux", - [FIMC_CLK_PARENT] = "parent", }; #define FIMC_DEFAULT_LCLK_FREQUENCY 133000000UL @@ -141,31 +110,29 @@ struct fimc_capability { /* * A structure of fimc context. * - * @ippdrv: prepare initialization using ippdrv. * @regs_res: register resources. * @regs: memory mapped io registers. * @lock: locking of operations. * @clocks: fimc clocks. - * @clk_frequency: LCLK clock frequency. - * @sysreg: handle to SYSREG block regmap. * @sc: scaler infomations. * @pol: porarity of writeback. * @id: fimc id. * @irq: irq number. - * @suspended: qos operations. */ struct fimc_context { - struct exynos_drm_ippdrv ippdrv; + struct exynos_drm_ipp ipp; + struct drm_device *drm_dev; + struct device *dev; + struct exynos_drm_ipp_task *task; + struct exynos_drm_ipp_formats *formats; + struct resource *regs_res; void __iomem *regs; spinlock_t lock; struct clk *clocks[FIMC_CLKS_MAX]; - u32 clk_frequency; - struct regmap *sysreg; struct fimc_scaler sc; int id; int irq; - bool suspended; }; static u32 fimc_read(struct fimc_context *ctx, u32 reg) @@ -217,19 +184,10 @@ static void fimc_sw_reset(struct fimc_context *ctx) fimc_write(ctx, 0x0, EXYNOS_CIFCNTSEQ); } -static int fimc_set_camblk_fimd0_wb(struct fimc_context *ctx) -{ - return regmap_update_bits(ctx->sysreg, SYSREG_CAMERA_BLK, - SYSREG_FIMD0WB_DEST_MASK, - ctx->id << SYSREG_FIMD0WB_DEST_SHIFT); -} - -static void fimc_set_type_ctrl(struct fimc_context *ctx, enum fimc_wb wb) +static void fimc_set_type_ctrl(struct fimc_context *ctx) { u32 cfg; - DRM_DEBUG_KMS("wb[%d]\n", wb); - cfg = fimc_read(ctx, EXYNOS_CIGCTRL); cfg &= ~(EXYNOS_CIGCTRL_TESTPATTERN_MASK | EXYNOS_CIGCTRL_SELCAM_ITU_MASK | @@ -238,23 +196,10 @@ static void fimc_set_type_ctrl(struct fimc_context *ctx, enum fimc_wb wb) EXYNOS_CIGCTRL_SELWB_CAMIF_MASK | EXYNOS_CIGCTRL_SELWRITEBACK_MASK); - switch (wb) { - case FIMC_WB_A: - cfg |= (EXYNOS_CIGCTRL_SELWRITEBACK_A | - EXYNOS_CIGCTRL_SELWB_CAMIF_WRITEBACK); - break; - case FIMC_WB_B: - cfg |= (EXYNOS_CIGCTRL_SELWRITEBACK_B | - EXYNOS_CIGCTRL_SELWB_CAMIF_WRITEBACK); - break; - case FIMC_WB_NONE: - default: - cfg |= (EXYNOS_CIGCTRL_SELCAM_ITU_A | - EXYNOS_CIGCTRL_SELWRITEBACK_A | - EXYNOS_CIGCTRL_SELCAM_MIPI_A | - EXYNOS_CIGCTRL_SELCAM_FIMC_ITU); - break; - } + cfg |= (EXYNOS_CIGCTRL_SELCAM_ITU_A | + EXYNOS_CIGCTRL_SELWRITEBACK_A | + EXYNOS_CIGCTRL_SELCAM_MIPI_A | + EXYNOS_CIGCTRL_SELCAM_FIMC_ITU); fimc_write(ctx, cfg, EXYNOS_CIGCTRL); } @@ -296,7 +241,6 @@ static void fimc_clear_irq(struct fimc_context *ctx) static bool fimc_check_ovf(struct fimc_context *ctx) { - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; u32 status, flag; status = fimc_read(ctx, EXYNOS_CISTATUS); @@ -310,7 +254,7 @@ static bool fimc_check_ovf(struct fimc_context *ctx) EXYNOS_CIWDOFST_CLROVFIY | EXYNOS_CIWDOFST_CLROVFICB | EXYNOS_CIWDOFST_CLROVFICR); - dev_err(ippdrv->dev, "occurred overflow at %d, status 0x%x.\n", + dev_err(ctx->dev, "occurred overflow at %d, status 0x%x.\n", ctx->id, status); return true; } @@ -379,7 +323,6 @@ static void fimc_handle_lastend(struct fimc_context *ctx, bool enable) static int fimc_src_set_fmt_order(struct fimc_context *ctx, u32 fmt) { - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; u32 cfg; DRM_DEBUG_KMS("fmt[0x%x]\n", fmt); @@ -439,7 +382,7 @@ static int fimc_src_set_fmt_order(struct fimc_context *ctx, u32 fmt) EXYNOS_MSCTRL_C_INT_IN_2PLANE); break; default: - dev_err(ippdrv->dev, "invalid source yuv order 0x%x.\n", fmt); + dev_err(ctx->dev, "invalid source yuv order 0x%x.\n", fmt); return -EINVAL; } @@ -448,10 +391,8 @@ static int fimc_src_set_fmt_order(struct fimc_context *ctx, u32 fmt) return 0; } -static int fimc_src_set_fmt(struct device *dev, u32 fmt) +static int fimc_src_set_fmt(struct fimc_context *ctx, u32 fmt) { - struct fimc_context *ctx = get_fimc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; u32 cfg; DRM_DEBUG_KMS("fmt[0x%x]\n", fmt); @@ -486,7 +427,7 @@ static int fimc_src_set_fmt(struct device *dev, u32 fmt) cfg |= EXYNOS_MSCTRL_INFORMAT_YCBCR420; break; default: - dev_err(ippdrv->dev, "invalid source format 0x%x.\n", fmt); + dev_err(ctx->dev, "invalid source format 0x%x.\n", fmt); return -EINVAL; } @@ -502,15 +443,12 @@ static int fimc_src_set_fmt(struct device *dev, u32 fmt) return fimc_src_set_fmt_order(ctx, fmt); } -static int fimc_src_set_transf(struct device *dev, - enum drm_exynos_degree degree, - enum drm_exynos_flip flip, bool *swap) +static int fimc_src_set_transf(struct fimc_context *ctx, unsigned int rotation) { - struct fimc_context *ctx = get_fimc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; + unsigned int degree = rotation & DRM_MODE_ROTATE_MASK; u32 cfg1, cfg2; - DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip); + DRM_DEBUG_KMS("rotation[%x]\n", rotation); cfg1 = fimc_read(ctx, EXYNOS_MSCTRL); cfg1 &= ~(EXYNOS_MSCTRL_FLIP_X_MIRROR | @@ -520,61 +458,58 @@ static int fimc_src_set_transf(struct device *dev, cfg2 &= ~EXYNOS_CITRGFMT_INROT90_CLOCKWISE; switch (degree) { - case EXYNOS_DRM_DEGREE_0: - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + case DRM_MODE_ROTATE_0: + if (rotation & DRM_MODE_REFLECT_X) cfg1 |= EXYNOS_MSCTRL_FLIP_X_MIRROR; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg1 |= EXYNOS_MSCTRL_FLIP_Y_MIRROR; break; - case EXYNOS_DRM_DEGREE_90: + case DRM_MODE_ROTATE_90: cfg2 |= EXYNOS_CITRGFMT_INROT90_CLOCKWISE; - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + if (rotation & DRM_MODE_REFLECT_X) cfg1 |= EXYNOS_MSCTRL_FLIP_X_MIRROR; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg1 |= EXYNOS_MSCTRL_FLIP_Y_MIRROR; break; - case EXYNOS_DRM_DEGREE_180: + case DRM_MODE_ROTATE_180: cfg1 |= (EXYNOS_MSCTRL_FLIP_X_MIRROR | EXYNOS_MSCTRL_FLIP_Y_MIRROR); - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + if (rotation & DRM_MODE_REFLECT_X) cfg1 &= ~EXYNOS_MSCTRL_FLIP_X_MIRROR; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg1 &= ~EXYNOS_MSCTRL_FLIP_Y_MIRROR; break; - case EXYNOS_DRM_DEGREE_270: + case DRM_MODE_ROTATE_270: cfg1 |= (EXYNOS_MSCTRL_FLIP_X_MIRROR | EXYNOS_MSCTRL_FLIP_Y_MIRROR); cfg2 |= EXYNOS_CITRGFMT_INROT90_CLOCKWISE; - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + if (rotation & DRM_MODE_REFLECT_X) cfg1 &= ~EXYNOS_MSCTRL_FLIP_X_MIRROR; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg1 &= ~EXYNOS_MSCTRL_FLIP_Y_MIRROR; break; - default: - dev_err(ippdrv->dev, "invalid degree value %d.\n", degree); - return -EINVAL; } fimc_write(ctx, cfg1, EXYNOS_MSCTRL); fimc_write(ctx, cfg2, EXYNOS_CITRGFMT); - *swap = (cfg2 & EXYNOS_CITRGFMT_INROT90_CLOCKWISE) ? 1 : 0; return 0; } static int fimc_set_window(struct fimc_context *ctx, - struct drm_exynos_pos *pos, struct drm_exynos_sz *sz) + struct exynos_drm_ipp_buffer *buf) { u32 cfg, h1, h2, v1, v2; /* cropped image */ - h1 = pos->x; - h2 = sz->hsize - pos->w - pos->x; - v1 = pos->y; - v2 = sz->vsize - pos->h - pos->y; + h1 = buf->rect.x; + h2 = buf->buf.width - buf->rect.w - buf->rect.x; + v1 = buf->rect.y; + v2 = buf->buf.height - buf->rect.h - buf->rect.y; DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]hsize[%d]vsize[%d]\n", - pos->x, pos->y, pos->w, pos->h, sz->hsize, sz->vsize); + buf->rect.x, buf->rect.y, buf->rect.w, buf->rect.h, + buf->buf.width, buf->buf.height); DRM_DEBUG_KMS("h1[%d]h2[%d]v1[%d]v2[%d]\n", h1, h2, v1, v2); /* @@ -596,38 +531,28 @@ static int fimc_set_window(struct fimc_context *ctx, return 0; } -static int fimc_src_set_size(struct device *dev, int swap, - struct drm_exynos_pos *pos, struct drm_exynos_sz *sz) +static int fimc_src_set_size(struct fimc_context *ctx, + struct exynos_drm_ipp_buffer *buf) { - struct fimc_context *ctx = get_fimc_context(dev); - struct drm_exynos_pos img_pos = *pos; - struct drm_exynos_sz img_sz = *sz; u32 cfg; - DRM_DEBUG_KMS("swap[%d]hsize[%d]vsize[%d]\n", - swap, sz->hsize, sz->vsize); + DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", buf->buf.width, buf->buf.height); /* original size */ - cfg = (EXYNOS_ORGISIZE_HORIZONTAL(img_sz.hsize) | - EXYNOS_ORGISIZE_VERTICAL(img_sz.vsize)); + cfg = (EXYNOS_ORGISIZE_HORIZONTAL(buf->buf.width) | + EXYNOS_ORGISIZE_VERTICAL(buf->buf.height)); fimc_write(ctx, cfg, EXYNOS_ORGISIZE); - DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]\n", pos->x, pos->y, pos->w, pos->h); - - if (swap) { - img_pos.w = pos->h; - img_pos.h = pos->w; - img_sz.hsize = sz->vsize; - img_sz.vsize = sz->hsize; - } + DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]\n", buf->rect.x, buf->rect.y, + buf->rect.w, buf->rect.h); /* set input DMA image size */ cfg = fimc_read(ctx, EXYNOS_CIREAL_ISIZE); cfg &= ~(EXYNOS_CIREAL_ISIZE_HEIGHT_MASK | EXYNOS_CIREAL_ISIZE_WIDTH_MASK); - cfg |= (EXYNOS_CIREAL_ISIZE_WIDTH(img_pos.w) | - EXYNOS_CIREAL_ISIZE_HEIGHT(img_pos.h)); + cfg |= (EXYNOS_CIREAL_ISIZE_WIDTH(buf->rect.w) | + EXYNOS_CIREAL_ISIZE_HEIGHT(buf->rect.h)); fimc_write(ctx, cfg, EXYNOS_CIREAL_ISIZE); /* @@ -635,91 +560,36 @@ static int fimc_src_set_size(struct device *dev, int swap, * for now, we support only ITU601 8 bit mode */ cfg = (EXYNOS_CISRCFMT_ITU601_8BIT | - EXYNOS_CISRCFMT_SOURCEHSIZE(img_sz.hsize) | - EXYNOS_CISRCFMT_SOURCEVSIZE(img_sz.vsize)); + EXYNOS_CISRCFMT_SOURCEHSIZE(buf->buf.width) | + EXYNOS_CISRCFMT_SOURCEVSIZE(buf->buf.height)); fimc_write(ctx, cfg, EXYNOS_CISRCFMT); /* offset Y(RGB), Cb, Cr */ - cfg = (EXYNOS_CIIYOFF_HORIZONTAL(img_pos.x) | - EXYNOS_CIIYOFF_VERTICAL(img_pos.y)); + cfg = (EXYNOS_CIIYOFF_HORIZONTAL(buf->rect.x) | + EXYNOS_CIIYOFF_VERTICAL(buf->rect.y)); fimc_write(ctx, cfg, EXYNOS_CIIYOFF); - cfg = (EXYNOS_CIICBOFF_HORIZONTAL(img_pos.x) | - EXYNOS_CIICBOFF_VERTICAL(img_pos.y)); + cfg = (EXYNOS_CIICBOFF_HORIZONTAL(buf->rect.x) | + EXYNOS_CIICBOFF_VERTICAL(buf->rect.y)); fimc_write(ctx, cfg, EXYNOS_CIICBOFF); - cfg = (EXYNOS_CIICROFF_HORIZONTAL(img_pos.x) | - EXYNOS_CIICROFF_VERTICAL(img_pos.y)); + cfg = (EXYNOS_CIICROFF_HORIZONTAL(buf->rect.x) | + EXYNOS_CIICROFF_VERTICAL(buf->rect.y)); fimc_write(ctx, cfg, EXYNOS_CIICROFF); - return fimc_set_window(ctx, &img_pos, &img_sz); + return fimc_set_window(ctx, buf); } -static int fimc_src_set_addr(struct device *dev, - struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id, - enum drm_exynos_ipp_buf_type buf_type) +static int fimc_src_set_addr(struct fimc_context *ctx, + struct exynos_drm_ipp_buffer *buf) { - struct fimc_context *ctx = get_fimc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; - struct drm_exynos_ipp_property *property; - struct drm_exynos_ipp_config *config; - - if (!c_node) { - DRM_ERROR("failed to get c_node.\n"); - return -EINVAL; - } - - property = &c_node->property; - - DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n", - property->prop_id, buf_id, buf_type); - - if (buf_id > FIMC_MAX_SRC) { - dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id); - return -ENOMEM; - } - - /* address register set */ - switch (buf_type) { - case IPP_BUF_ENQUEUE: - config = &property->config[EXYNOS_DRM_OPS_SRC]; - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_Y], - EXYNOS_CIIYSA0); - - if (config->fmt == DRM_FORMAT_YVU420) { - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CR], - EXYNOS_CIICBSA0); - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CB], - EXYNOS_CIICRSA0); - } else { - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CB], - EXYNOS_CIICBSA0); - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CR], - EXYNOS_CIICRSA0); - } - break; - case IPP_BUF_DEQUEUE: - fimc_write(ctx, 0x0, EXYNOS_CIIYSA0); - fimc_write(ctx, 0x0, EXYNOS_CIICBSA0); - fimc_write(ctx, 0x0, EXYNOS_CIICRSA0); - break; - default: - /* bypass */ - break; - } + fimc_write(ctx, buf->dma_addr[0], EXYNOS_CIIYSA(0)); + fimc_write(ctx, buf->dma_addr[1], EXYNOS_CIICBSA(0)); + fimc_write(ctx, buf->dma_addr[2], EXYNOS_CIICRSA(0)); return 0; } -static struct exynos_drm_ipp_ops fimc_src_ops = { - .set_fmt = fimc_src_set_fmt, - .set_transf = fimc_src_set_transf, - .set_size = fimc_src_set_size, - .set_addr = fimc_src_set_addr, -}; - static int fimc_dst_set_fmt_order(struct fimc_context *ctx, u32 fmt) { - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; u32 cfg; DRM_DEBUG_KMS("fmt[0x%x]\n", fmt); @@ -785,7 +655,7 @@ static int fimc_dst_set_fmt_order(struct fimc_context *ctx, u32 fmt) cfg |= EXYNOS_CIOCTRL_YCBCR_2PLANE; break; default: - dev_err(ippdrv->dev, "invalid target yuv order 0x%x.\n", fmt); + dev_err(ctx->dev, "invalid target yuv order 0x%x.\n", fmt); return -EINVAL; } @@ -794,10 +664,8 @@ static int fimc_dst_set_fmt_order(struct fimc_context *ctx, u32 fmt) return 0; } -static int fimc_dst_set_fmt(struct device *dev, u32 fmt) +static int fimc_dst_set_fmt(struct fimc_context *ctx, u32 fmt) { - struct fimc_context *ctx = get_fimc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; u32 cfg; DRM_DEBUG_KMS("fmt[0x%x]\n", fmt); @@ -838,7 +706,7 @@ static int fimc_dst_set_fmt(struct device *dev, u32 fmt) cfg |= EXYNOS_CITRGFMT_OUTFORMAT_YCBCR420; break; default: - dev_err(ippdrv->dev, "invalid target format 0x%x.\n", + dev_err(ctx->dev, "invalid target format 0x%x.\n", fmt); return -EINVAL; } @@ -856,66 +724,58 @@ static int fimc_dst_set_fmt(struct device *dev, u32 fmt) return fimc_dst_set_fmt_order(ctx, fmt); } -static int fimc_dst_set_transf(struct device *dev, - enum drm_exynos_degree degree, - enum drm_exynos_flip flip, bool *swap) +static int fimc_dst_set_transf(struct fimc_context *ctx, unsigned int rotation) { - struct fimc_context *ctx = get_fimc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; + unsigned int degree = rotation & DRM_MODE_ROTATE_MASK; u32 cfg; - DRM_DEBUG_KMS("degree[%d]flip[0x%x]\n", degree, flip); + DRM_DEBUG_KMS("rotation[0x%x]\n", rotation); cfg = fimc_read(ctx, EXYNOS_CITRGFMT); cfg &= ~EXYNOS_CITRGFMT_FLIP_MASK; cfg &= ~EXYNOS_CITRGFMT_OUTROT90_CLOCKWISE; switch (degree) { - case EXYNOS_DRM_DEGREE_0: - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + case DRM_MODE_ROTATE_0: + if (rotation & DRM_MODE_REFLECT_X) cfg |= EXYNOS_CITRGFMT_FLIP_X_MIRROR; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg |= EXYNOS_CITRGFMT_FLIP_Y_MIRROR; break; - case EXYNOS_DRM_DEGREE_90: + case DRM_MODE_ROTATE_90: cfg |= EXYNOS_CITRGFMT_OUTROT90_CLOCKWISE; - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + if (rotation & DRM_MODE_REFLECT_X) cfg |= EXYNOS_CITRGFMT_FLIP_X_MIRROR; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg |= EXYNOS_CITRGFMT_FLIP_Y_MIRROR; break; - case EXYNOS_DRM_DEGREE_180: + case DRM_MODE_ROTATE_180: cfg |= (EXYNOS_CITRGFMT_FLIP_X_MIRROR | EXYNOS_CITRGFMT_FLIP_Y_MIRROR); - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + if (rotation & DRM_MODE_REFLECT_X) cfg &= ~EXYNOS_CITRGFMT_FLIP_X_MIRROR; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg &= ~EXYNOS_CITRGFMT_FLIP_Y_MIRROR; break; - case EXYNOS_DRM_DEGREE_270: + case DRM_MODE_ROTATE_270: cfg |= (EXYNOS_CITRGFMT_OUTROT90_CLOCKWISE | EXYNOS_CITRGFMT_FLIP_X_MIRROR | EXYNOS_CITRGFMT_FLIP_Y_MIRROR); - if (flip & EXYNOS_DRM_FLIP_VERTICAL) + if (rotation & DRM_MODE_REFLECT_X) cfg &= ~EXYNOS_CITRGFMT_FLIP_X_MIRROR; - if (flip & EXYNOS_DRM_FLIP_HORIZONTAL) + if (rotation & DRM_MODE_REFLECT_Y) cfg &= ~EXYNOS_CITRGFMT_FLIP_Y_MIRROR; break; - default: - dev_err(ippdrv->dev, "invalid degree value %d.\n", degree); - return -EINVAL; } fimc_write(ctx, cfg, EXYNOS_CITRGFMT); - *swap = (cfg & EXYNOS_CITRGFMT_OUTROT90_CLOCKWISE) ? 1 : 0; return 0; } static int fimc_set_prescaler(struct fimc_context *ctx, struct fimc_scaler *sc, - struct drm_exynos_pos *src, struct drm_exynos_pos *dst) + struct drm_exynos_ipp_task_rect *src, struct drm_exynos_ipp_task_rect *dst) { - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; u32 cfg, cfg_ext, shfactor; u32 pre_dst_width, pre_dst_height; u32 hfactor, vfactor; @@ -942,13 +802,13 @@ static int fimc_set_prescaler(struct fimc_context *ctx, struct fimc_scaler *sc, /* fimc_ippdrv_check_property assures that dividers are not null */ hfactor = fls(src_w / dst_w / 2); if (hfactor > FIMC_SHFACTOR / 2) { - dev_err(ippdrv->dev, "failed to get ratio horizontal.\n"); + dev_err(ctx->dev, "failed to get ratio horizontal.\n"); return -EINVAL; } vfactor = fls(src_h / dst_h / 2); if (vfactor > FIMC_SHFACTOR / 2) { - dev_err(ippdrv->dev, "failed to get ratio vertical.\n"); + dev_err(ctx->dev, "failed to get ratio vertical.\n"); return -EINVAL; } @@ -1019,83 +879,73 @@ static void fimc_set_scaler(struct fimc_context *ctx, struct fimc_scaler *sc) fimc_write(ctx, cfg_ext, EXYNOS_CIEXTEN); } -static int fimc_dst_set_size(struct device *dev, int swap, - struct drm_exynos_pos *pos, struct drm_exynos_sz *sz) +static int fimc_dst_set_size(struct fimc_context *ctx, + struct exynos_drm_ipp_buffer *buf) { - struct fimc_context *ctx = get_fimc_context(dev); - struct drm_exynos_pos img_pos = *pos; - struct drm_exynos_sz img_sz = *sz; u32 cfg; - DRM_DEBUG_KMS("swap[%d]hsize[%d]vsize[%d]\n", - swap, sz->hsize, sz->vsize); + DRM_DEBUG_KMS("hsize[%d]vsize[%d]\n", buf->buf.width, buf->buf.height); /* original size */ - cfg = (EXYNOS_ORGOSIZE_HORIZONTAL(img_sz.hsize) | - EXYNOS_ORGOSIZE_VERTICAL(img_sz.vsize)); + cfg = (EXYNOS_ORGOSIZE_HORIZONTAL(buf->buf.width) | + EXYNOS_ORGOSIZE_VERTICAL(buf->buf.height)); fimc_write(ctx, cfg, EXYNOS_ORGOSIZE); - DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]\n", pos->x, pos->y, pos->w, pos->h); + DRM_DEBUG_KMS("x[%d]y[%d]w[%d]h[%d]\n", buf->rect.x, buf->rect.y, + buf->rect.w, buf->rect.h); /* CSC ITU */ cfg = fimc_read(ctx, EXYNOS_CIGCTRL); cfg &= ~EXYNOS_CIGCTRL_CSC_MASK; - if (sz->hsize >= FIMC_WIDTH_ITU_709) + if (buf->buf.width >= FIMC_WIDTH_ITU_709) cfg |= EXYNOS_CIGCTRL_CSC_ITU709; else cfg |= EXYNOS_CIGCTRL_CSC_ITU601; fimc_write(ctx, cfg, EXYNOS_CIGCTRL); - if (swap) { - img_pos.w = pos->h; - img_pos.h = pos->w; - img_sz.hsize = sz->vsize; - img_sz.vsize = sz->hsize; - } - /* target image size */ cfg = fimc_read(ctx, EXYNOS_CITRGFMT); cfg &= ~(EXYNOS_CITRGFMT_TARGETH_MASK | EXYNOS_CITRGFMT_TARGETV_MASK); - cfg |= (EXYNOS_CITRGFMT_TARGETHSIZE(img_pos.w) | - EXYNOS_CITRGFMT_TARGETVSIZE(img_pos.h)); + cfg |= (EXYNOS_CITRGFMT_TARGETHSIZE(buf->rect.w) | + EXYNOS_CITRGFMT_TARGETVSIZE(buf->rect.h)); fimc_write(ctx, cfg, EXYNOS_CITRGFMT); /* target area */ - cfg = EXYNOS_CITAREA_TARGET_AREA(img_pos.w * img_pos.h); + cfg = EXYNOS_CITAREA_TARGET_AREA(buf->rect.w * buf->rect.h); fimc_write(ctx, cfg, EXYNOS_CITAREA); /* offset Y(RGB), Cb, Cr */ - cfg = (EXYNOS_CIOYOFF_HORIZONTAL(img_pos.x) | - EXYNOS_CIOYOFF_VERTICAL(img_pos.y)); + cfg = (EXYNOS_CIOYOFF_HORIZONTAL(buf->rect.x) | + EXYNOS_CIOYOFF_VERTICAL(buf->rect.y)); fimc_write(ctx, cfg, EXYNOS_CIOYOFF); - cfg = (EXYNOS_CIOCBOFF_HORIZONTAL(img_pos.x) | - EXYNOS_CIOCBOFF_VERTICAL(img_pos.y)); + cfg = (EXYNOS_CIOCBOFF_HORIZONTAL(buf->rect.x) | + EXYNOS_CIOCBOFF_VERTICAL(buf->rect.y)); fimc_write(ctx, cfg, EXYNOS_CIOCBOFF); - cfg = (EXYNOS_CIOCROFF_HORIZONTAL(img_pos.x) | - EXYNOS_CIOCROFF_VERTICAL(img_pos.y)); + cfg = (EXYNOS_CIOCROFF_HORIZONTAL(buf->rect.x) | + EXYNOS_CIOCROFF_VERTICAL(buf->rect.y)); fimc_write(ctx, cfg, EXYNOS_CIOCROFF); return 0; } static void fimc_dst_set_buf_seq(struct fimc_context *ctx, u32 buf_id, - enum drm_exynos_ipp_buf_type buf_type) + bool enqueue) { unsigned long flags; u32 buf_num; u32 cfg; - DRM_DEBUG_KMS("buf_id[%d]buf_type[%d]\n", buf_id, buf_type); + DRM_DEBUG_KMS("buf_id[%d]enqueu[%d]\n", buf_id, enqueue); spin_lock_irqsave(&ctx->lock, flags); cfg = fimc_read(ctx, EXYNOS_CIFCNTSEQ); - if (buf_type == IPP_BUF_ENQUEUE) + if (enqueue) cfg |= (1 << buf_id); else cfg &= ~(1 << buf_id); @@ -1104,88 +954,31 @@ static void fimc_dst_set_buf_seq(struct fimc_context *ctx, u32 buf_id, buf_num = hweight32(cfg); - if (buf_type == IPP_BUF_ENQUEUE && buf_num >= FIMC_BUF_START) + if (enqueue && buf_num >= FIMC_BUF_START) fimc_mask_irq(ctx, true); - else if (buf_type == IPP_BUF_DEQUEUE && buf_num <= FIMC_BUF_STOP) + else if (!enqueue && buf_num <= FIMC_BUF_STOP) fimc_mask_irq(ctx, false); spin_unlock_irqrestore(&ctx->lock, flags); } -static int fimc_dst_set_addr(struct device *dev, - struct drm_exynos_ipp_buf_info *buf_info, u32 buf_id, - enum drm_exynos_ipp_buf_type buf_type) +static int fimc_dst_set_addr(struct fimc_context *ctx, + struct exynos_drm_ipp_buffer *buf) { - struct fimc_context *ctx = get_fimc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; - struct drm_exynos_ipp_property *property; - struct drm_exynos_ipp_config *config; + fimc_write(ctx, buf->dma_addr[0], EXYNOS_CIOYSA(0)); + fimc_write(ctx, buf->dma_addr[1], EXYNOS_CIOCBSA(0)); + fimc_write(ctx, buf->dma_addr[2], EXYNOS_CIOCRSA(0)); - if (!c_node) { - DRM_ERROR("failed to get c_node.\n"); - return -EINVAL; - } - - property = &c_node->property; - - DRM_DEBUG_KMS("prop_id[%d]buf_id[%d]buf_type[%d]\n", - property->prop_id, buf_id, buf_type); - - if (buf_id > FIMC_MAX_DST) { - dev_info(ippdrv->dev, "invalid buf_id %d.\n", buf_id); - return -ENOMEM; - } - - /* address register set */ - switch (buf_type) { - case IPP_BUF_ENQUEUE: - config = &property->config[EXYNOS_DRM_OPS_DST]; - - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_Y], - EXYNOS_CIOYSA(buf_id)); - - if (config->fmt == DRM_FORMAT_YVU420) { - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CR], - EXYNOS_CIOCBSA(buf_id)); - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CB], - EXYNOS_CIOCRSA(buf_id)); - } else { - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CB], - EXYNOS_CIOCBSA(buf_id)); - fimc_write(ctx, buf_info->base[EXYNOS_DRM_PLANAR_CR], - EXYNOS_CIOCRSA(buf_id)); - } - break; - case IPP_BUF_DEQUEUE: - fimc_write(ctx, 0x0, EXYNOS_CIOYSA(buf_id)); - fimc_write(ctx, 0x0, EXYNOS_CIOCBSA(buf_id)); - fimc_write(ctx, 0x0, EXYNOS_CIOCRSA(buf_id)); - break; - default: - /* bypass */ - break; - } - - fimc_dst_set_buf_seq(ctx, buf_id, buf_type); + fimc_dst_set_buf_seq(ctx, 0, true); return 0; } -static struct exynos_drm_ipp_ops fimc_dst_ops = { - .set_fmt = fimc_dst_set_fmt, - .set_transf = fimc_dst_set_transf, - .set_size = fimc_dst_set_size, - .set_addr = fimc_dst_set_addr, -}; +static void fimc_stop(struct fimc_context *ctx); static irqreturn_t fimc_irq_handler(int irq, void *dev_id) { struct fimc_context *ctx = dev_id; - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; - struct drm_exynos_ipp_event_work *event_work = - c_node->event_work; int buf_id; DRM_DEBUG_KMS("fimc id[%d]\n", ctx->id); @@ -1203,170 +996,19 @@ static irqreturn_t fimc_irq_handler(int irq, void *dev_id) DRM_DEBUG_KMS("buf_id[%d]\n", buf_id); - fimc_dst_set_buf_seq(ctx, buf_id, IPP_BUF_DEQUEUE); + if (ctx->task) { + struct exynos_drm_ipp_task *task = ctx->task; - event_work->ippdrv = ippdrv; - event_work->buf_id[EXYNOS_DRM_OPS_DST] = buf_id; - queue_work(ippdrv->event_workq, &event_work->work); - - return IRQ_HANDLED; -} - -static int fimc_init_prop_list(struct exynos_drm_ippdrv *ippdrv) -{ - struct drm_exynos_ipp_prop_list *prop_list = &ippdrv->prop_list; - - prop_list->version = 1; - prop_list->writeback = 1; - prop_list->refresh_min = FIMC_REFRESH_MIN; - prop_list->refresh_max = FIMC_REFRESH_MAX; - prop_list->flip = (1 << EXYNOS_DRM_FLIP_NONE) | - (1 << EXYNOS_DRM_FLIP_VERTICAL) | - (1 << EXYNOS_DRM_FLIP_HORIZONTAL); - prop_list->degree = (1 << EXYNOS_DRM_DEGREE_0) | - (1 << EXYNOS_DRM_DEGREE_90) | - (1 << EXYNOS_DRM_DEGREE_180) | - (1 << EXYNOS_DRM_DEGREE_270); - prop_list->csc = 1; - prop_list->crop = 1; - prop_list->crop_max.hsize = FIMC_CROP_MAX; - prop_list->crop_max.vsize = FIMC_CROP_MAX; - prop_list->crop_min.hsize = FIMC_CROP_MIN; - prop_list->crop_min.vsize = FIMC_CROP_MIN; - prop_list->scale = 1; - prop_list->scale_max.hsize = FIMC_SCALE_MAX; - prop_list->scale_max.vsize = FIMC_SCALE_MAX; - prop_list->scale_min.hsize = FIMC_SCALE_MIN; - prop_list->scale_min.vsize = FIMC_SCALE_MIN; - - return 0; -} - -static inline bool fimc_check_drm_flip(enum drm_exynos_flip flip) -{ - switch (flip) { - case EXYNOS_DRM_FLIP_NONE: - case EXYNOS_DRM_FLIP_VERTICAL: - case EXYNOS_DRM_FLIP_HORIZONTAL: - case EXYNOS_DRM_FLIP_BOTH: - return true; - default: - DRM_DEBUG_KMS("invalid flip\n"); - return false; + ctx->task = NULL; + pm_runtime_mark_last_busy(ctx->dev); + pm_runtime_put_autosuspend(ctx->dev); + exynos_drm_ipp_task_done(task, 0); } -} - -static int fimc_ippdrv_check_property(struct device *dev, - struct drm_exynos_ipp_property *property) -{ - struct fimc_context *ctx = get_fimc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_prop_list *pp = &ippdrv->prop_list; - struct drm_exynos_ipp_config *config; - struct drm_exynos_pos *pos; - struct drm_exynos_sz *sz; - bool swap; - int i; - for_each_ipp_ops(i) { - if ((i == EXYNOS_DRM_OPS_SRC) && - (property->cmd == IPP_CMD_WB)) - continue; - - config = &property->config[i]; - pos = &config->pos; - sz = &config->sz; - - /* check for flip */ - if (!fimc_check_drm_flip(config->flip)) { - DRM_ERROR("invalid flip.\n"); - goto err_property; - } - - /* check for degree */ - switch (config->degree) { - case EXYNOS_DRM_DEGREE_90: - case EXYNOS_DRM_DEGREE_270: - swap = true; - break; - case EXYNOS_DRM_DEGREE_0: - case EXYNOS_DRM_DEGREE_180: - swap = false; - break; - default: - DRM_ERROR("invalid degree.\n"); - goto err_property; - } - - /* check for buffer bound */ - if ((pos->x + pos->w > sz->hsize) || - (pos->y + pos->h > sz->vsize)) { - DRM_ERROR("out of buf bound.\n"); - goto err_property; - } - - /* check for crop */ - if ((i == EXYNOS_DRM_OPS_SRC) && (pp->crop)) { - if (swap) { - if ((pos->h < pp->crop_min.hsize) || - (sz->vsize > pp->crop_max.hsize) || - (pos->w < pp->crop_min.vsize) || - (sz->hsize > pp->crop_max.vsize)) { - DRM_ERROR("out of crop size.\n"); - goto err_property; - } - } else { - if ((pos->w < pp->crop_min.hsize) || - (sz->hsize > pp->crop_max.hsize) || - (pos->h < pp->crop_min.vsize) || - (sz->vsize > pp->crop_max.vsize)) { - DRM_ERROR("out of crop size.\n"); - goto err_property; - } - } - } - - /* check for scale */ - if ((i == EXYNOS_DRM_OPS_DST) && (pp->scale)) { - if (swap) { - if ((pos->h < pp->scale_min.hsize) || - (sz->vsize > pp->scale_max.hsize) || - (pos->w < pp->scale_min.vsize) || - (sz->hsize > pp->scale_max.vsize)) { - DRM_ERROR("out of scale size.\n"); - goto err_property; - } - } else { - if ((pos->w < pp->scale_min.hsize) || - (sz->hsize > pp->scale_max.hsize) || - (pos->h < pp->scale_min.vsize) || - (sz->vsize > pp->scale_max.vsize)) { - DRM_ERROR("out of scale size.\n"); - goto err_property; - } - } - } - } - - return 0; + fimc_dst_set_buf_seq(ctx, buf_id, false); + fimc_stop(ctx); -err_property: - for_each_ipp_ops(i) { - if ((i == EXYNOS_DRM_OPS_SRC) && - (property->cmd == IPP_CMD_WB)) - continue; - - config = &property->config[i]; - pos = &config->pos; - sz = &config->sz; - - DRM_ERROR("[%s]f[%d]r[%d]pos[%d %d %d %d]sz[%d %d]\n", - i ? "dst" : "src", config->flip, config->degree, - pos->x, pos->y, pos->w, pos->h, - sz->hsize, sz->vsize); - } - - return -EINVAL; + return IRQ_HANDLED; } static void fimc_clear_addr(struct fimc_context *ctx) @@ -1386,10 +1028,8 @@ static void fimc_clear_addr(struct fimc_context *ctx) } } -static int fimc_ippdrv_reset(struct device *dev) +static int fimc_reset(struct fimc_context *ctx) { - struct fimc_context *ctx = get_fimc_context(dev); - /* reset h/w block */ fimc_sw_reset(ctx); @@ -1401,78 +1041,24 @@ static int fimc_ippdrv_reset(struct device *dev) return 0; } -static int fimc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd) +static int fimc_start(struct fimc_context *ctx) { - struct fimc_context *ctx = get_fimc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - struct drm_exynos_ipp_cmd_node *c_node = ippdrv->c_node; - struct drm_exynos_ipp_property *property; - struct drm_exynos_ipp_config *config; - struct drm_exynos_pos img_pos[EXYNOS_DRM_OPS_MAX]; - struct drm_exynos_ipp_set_wb set_wb; - int ret, i; u32 cfg0, cfg1; - DRM_DEBUG_KMS("cmd[%d]\n", cmd); - - if (!c_node) { - DRM_ERROR("failed to get c_node.\n"); - return -EINVAL; - } - - property = &c_node->property; - fimc_mask_irq(ctx, true); - for_each_ipp_ops(i) { - config = &property->config[i]; - img_pos[i] = config->pos; - } - - ret = fimc_set_prescaler(ctx, &ctx->sc, - &img_pos[EXYNOS_DRM_OPS_SRC], - &img_pos[EXYNOS_DRM_OPS_DST]); - if (ret) { - dev_err(dev, "failed to set prescaler.\n"); - return ret; - } - /* If set ture, we can save jpeg about screen */ fimc_handle_jpeg(ctx, false); fimc_set_scaler(ctx, &ctx->sc); - switch (cmd) { - case IPP_CMD_M2M: - fimc_set_type_ctrl(ctx, FIMC_WB_NONE); - fimc_handle_lastend(ctx, false); + fimc_set_type_ctrl(ctx); + fimc_handle_lastend(ctx, false); - /* setup dma */ - cfg0 = fimc_read(ctx, EXYNOS_MSCTRL); - cfg0 &= ~EXYNOS_MSCTRL_INPUT_MASK; - cfg0 |= EXYNOS_MSCTRL_INPUT_MEMORY; - fimc_write(ctx, cfg0, EXYNOS_MSCTRL); - break; - case IPP_CMD_WB: - fimc_set_type_ctrl(ctx, FIMC_WB_A); - fimc_handle_lastend(ctx, true); - - /* setup FIMD */ - ret = fimc_set_camblk_fimd0_wb(ctx); - if (ret < 0) { - dev_err(dev, "camblk setup failed.\n"); - return ret; - } - - set_wb.enable = 1; - set_wb.refresh = property->refresh_rate; - exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb); - break; - case IPP_CMD_OUTPUT: - default: - ret = -EINVAL; - dev_err(dev, "invalid operations.\n"); - return ret; - } + /* setup dma */ + cfg0 = fimc_read(ctx, EXYNOS_MSCTRL); + cfg0 &= ~EXYNOS_MSCTRL_INPUT_MASK; + cfg0 |= EXYNOS_MSCTRL_INPUT_MEMORY; + fimc_write(ctx, cfg0, EXYNOS_MSCTRL); /* Reset status */ fimc_write(ctx, 0x0, EXYNOS_CISTATUS); @@ -1498,36 +1084,20 @@ static int fimc_ippdrv_start(struct device *dev, enum drm_exynos_ipp_cmd cmd) fimc_clear_bits(ctx, EXYNOS_CIOCTRL, EXYNOS_CIOCTRL_WEAVE_MASK); - if (cmd == IPP_CMD_M2M) - fimc_set_bits(ctx, EXYNOS_MSCTRL, EXYNOS_MSCTRL_ENVID); + fimc_set_bits(ctx, EXYNOS_MSCTRL, EXYNOS_MSCTRL_ENVID); return 0; } -static void fimc_ippdrv_stop(struct device *dev, enum drm_exynos_ipp_cmd cmd) +static void fimc_stop(struct fimc_context *ctx) { - struct fimc_context *ctx = get_fimc_context(dev); - struct drm_exynos_ipp_set_wb set_wb = {0, 0}; u32 cfg; - DRM_DEBUG_KMS("cmd[%d]\n", cmd); - - switch (cmd) { - case IPP_CMD_M2M: - /* Source clear */ - cfg = fimc_read(ctx, EXYNOS_MSCTRL); - cfg &= ~EXYNOS_MSCTRL_INPUT_MASK; - cfg &= ~EXYNOS_MSCTRL_ENVID; - fimc_write(ctx, cfg, EXYNOS_MSCTRL); - break; - case IPP_CMD_WB: - exynos_drm_ippnb_send_event(IPP_SET_WRITEBACK, (void *)&set_wb); - break; - case IPP_CMD_OUTPUT: - default: - dev_err(dev, "invalid operations.\n"); - break; - } + /* Source clear */ + cfg = fimc_read(ctx, EXYNOS_MSCTRL); + cfg &= ~EXYNOS_MSCTRL_INPUT_MASK; + cfg &= ~EXYNOS_MSCTRL_ENVID; + fimc_write(ctx, cfg, EXYNOS_MSCTRL); fimc_mask_irq(ctx, false); @@ -1545,6 +1115,87 @@ static void fimc_ippdrv_stop(struct device *dev, enum drm_exynos_ipp_cmd cmd) fimc_set_bits(ctx, EXYNOS_CIGCTRL, EXYNOS_CIGCTRL_IRQ_END_DISABLE); } +static int fimc_commit(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) +{ + struct fimc_context *ctx = + container_of(ipp, struct fimc_context, ipp); + + pm_runtime_get_sync(ctx->dev); + ctx->task = task; + + fimc_src_set_fmt(ctx, task->src.buf.fourcc); + fimc_src_set_size(ctx, &task->src); + fimc_src_set_transf(ctx, DRM_MODE_ROTATE_0); + fimc_src_set_addr(ctx, &task->src); + fimc_dst_set_fmt(ctx, task->dst.buf.fourcc); + fimc_dst_set_transf(ctx, task->transform.rotation); + fimc_dst_set_size(ctx, &task->dst); + fimc_dst_set_addr(ctx, &task->dst); + fimc_set_prescaler(ctx, &ctx->sc, &task->src.rect, &task->dst.rect); + fimc_start(ctx); + + return 0; +} + +static void fimc_abort(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) +{ + struct fimc_context *ctx = + container_of(ipp, struct fimc_context, ipp); + + fimc_reset(ctx); + + if (ctx->task) { + struct exynos_drm_ipp_task *task = ctx->task; + + ctx->task = NULL; + pm_runtime_mark_last_busy(ctx->dev); + pm_runtime_put_autosuspend(ctx->dev); + exynos_drm_ipp_task_done(task, -EIO); + } +} + +static struct exynos_drm_ipp_funcs ipp_funcs = { + .commit = fimc_commit, + .abort = fimc_abort, +}; + +static int fimc_bind(struct device *dev, struct device *master, void *data) +{ + struct fimc_context *ctx = dev_get_drvdata(dev); + struct drm_device *drm_dev = data; + struct exynos_drm_ipp *ipp = &ctx->ipp; + + ctx->drm_dev = drm_dev; + drm_iommu_attach_device(drm_dev, dev); + + exynos_drm_ipp_register(drm_dev, ipp, &ipp_funcs, + DRM_EXYNOS_IPP_CAP_CROP | DRM_EXYNOS_IPP_CAP_ROTATE | + DRM_EXYNOS_IPP_CAP_SCALE | DRM_EXYNOS_IPP_CAP_CONVERT, + ctx->formats, "fimc"); + + dev_info(dev, "The exynos fimc is probed successfully\n"); + + return 0; +} + +static void fimc_unbind(struct device *dev, struct device *master, + void *data) +{ + struct fimc_context *ctx = dev_get_drvdata(dev); + struct drm_device *drm_dev = data; + struct exynos_drm_ipp *ipp = &ctx->ipp; + + exynos_drm_ipp_unregister(drm_dev, ipp); + drm_iommu_detach_device(drm_dev, dev); +} + +static const struct component_ops fimc_component_ops = { + .bind = fimc_bind, + .unbind = fimc_unbind, +}; + static void fimc_put_clocks(struct fimc_context *ctx) { int i; @@ -1559,7 +1210,7 @@ static void fimc_put_clocks(struct fimc_context *ctx) static int fimc_setup_clocks(struct fimc_context *ctx) { - struct device *fimc_dev = ctx->ippdrv.dev; + struct device *fimc_dev = ctx->dev; struct device *dev; int ret, i; @@ -1574,8 +1225,6 @@ static int fimc_setup_clocks(struct fimc_context *ctx) ctx->clocks[i] = clk_get(dev, fimc_clock_names[i]); if (IS_ERR(ctx->clocks[i])) { - if (i >= FIMC_CLK_MUX) - break; ret = PTR_ERR(ctx->clocks[i]); dev_err(fimc_dev, "failed to get clock: %s\n", fimc_clock_names[i]); @@ -1583,20 +1232,6 @@ static int fimc_setup_clocks(struct fimc_context *ctx) } } - /* Optional FIMC LCLK parent clock setting */ - if (!IS_ERR(ctx->clocks[FIMC_CLK_PARENT])) { - ret = clk_set_parent(ctx->clocks[FIMC_CLK_MUX], - ctx->clocks[FIMC_CLK_PARENT]); - if (ret < 0) { - dev_err(fimc_dev, "failed to set parent.\n"); - goto e_clk_free; - } - } - - ret = clk_set_rate(ctx->clocks[FIMC_CLK_LCLK], ctx->clk_frequency); - if (ret < 0) - goto e_clk_free; - ret = clk_prepare_enable(ctx->clocks[FIMC_CLK_LCLK]); if (!ret) return ret; @@ -1605,22 +1240,42 @@ static int fimc_setup_clocks(struct fimc_context *ctx) return ret; } +static const unsigned int fimc_formats[] = { + DRM_FORMAT_XRGB8888, DRM_FORMAT_RGB565, + DRM_FORMAT_NV12, DRM_FORMAT_NV16, DRM_FORMAT_NV21, DRM_FORMAT_NV61, + DRM_FORMAT_UYVY, DRM_FORMAT_VYUY, DRM_FORMAT_YUYV, DRM_FORMAT_YVYU, + DRM_FORMAT_YUV420, DRM_FORMAT_YVU420, DRM_FORMAT_YUV422, + DRM_FORMAT_YUV444, +}; + +static const struct drm_exynos_ipp_limit fimc_4210_limits_v1[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {32, 8192, 8}, .v = {32, 8192, 2}) }, + { IPP_SIZE_LIMIT(AREA, .h = {32, 4224, 2}, .v = {32, 0, 2}) }, + { IPP_SIZE_LIMIT(ROTATED, .h = {128, 1920}, .v = {128, 0}) }, + { IPP_SCALE_LIMIT(.h = { (1 << 16) / 64, (1 << 16) * 64}, + .v = { (1 << 16) / 64, (1 << 16) * 64}) }, + { } +}; + +static const struct drm_exynos_ipp_limit fimc_4210_limits_v2[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = {32, 8192, 8}, .v = {32, 8192, 2}) }, + { IPP_SIZE_LIMIT(AREA, .h = {32, 1920, 2}, .v = {32, 0, 2}) }, + { IPP_SIZE_LIMIT(ROTATED, .h = {128, 1366}, .v = {128, 0}) }, + { IPP_SCALE_LIMIT(.h = { (1 << 16) / 64, (1 << 16) * 64}, + .v = { (1 << 16) / 64, (1 << 16) * 64}) }, + { } +}; + static int fimc_parse_dt(struct fimc_context *ctx) { - struct device_node *node = ctx->ippdrv.dev->of_node; + struct device_node *node = ctx->dev->of_node; - /* Handle only devices that support the LCD Writeback data path */ - if (!of_property_read_bool(node, "samsung,lcd-wb")) + if (exynos_drm_check_shared_device(ctx->dev) != 0) return -ENODEV; - if (of_property_read_u32(node, "clock-frequency", - &ctx->clk_frequency)) - ctx->clk_frequency = FIMC_DEFAULT_LCLK_FREQUENCY; - ctx->id = of_alias_get_id(node, "fimc"); - if (ctx->id < 0) { - dev_err(ctx->ippdrv.dev, "failed to get node alias id.\n"); + dev_err(ctx->dev, "failed to get node alias id.\n"); return -EINVAL; } @@ -1629,11 +1284,13 @@ static int fimc_parse_dt(struct fimc_context *ctx) static int fimc_probe(struct platform_device *pdev) { + const struct drm_exynos_ipp_limit *limits; + struct exynos_drm_ipp_formats *formats; struct device *dev = &pdev->dev; struct fimc_context *ctx; struct resource *res; - struct exynos_drm_ippdrv *ippdrv; int ret; + int i; if (!dev->of_node) { dev_err(dev, "device tree node not found.\n"); @@ -1644,18 +1301,26 @@ static int fimc_probe(struct platform_device *pdev) if (!ctx) return -ENOMEM; - ctx->ippdrv.dev = dev; + ctx->dev = dev; ret = fimc_parse_dt(ctx); if (ret < 0) return ret; - ctx->sysreg = syscon_regmap_lookup_by_phandle(dev->of_node, - "samsung,sysreg"); - if (IS_ERR(ctx->sysreg)) { - dev_err(dev, "syscon regmap lookup failed.\n"); - return PTR_ERR(ctx->sysreg); + /* construct formats/limits array */ + formats = devm_kzalloc(dev, sizeof(*formats) * + (ARRAY_SIZE(fimc_formats) + 1), GFP_KERNEL); + if (!formats) + return -ENOMEM; + + limits = (ctx->id < 3) ? fimc_4210_limits_v1 : fimc_4210_limits_v2; + for (i = 0; i < ARRAY_SIZE(fimc_formats); i++) { + formats[i].fourcc = fimc_formats[i]; + formats[i].type = DRM_EXYNOS_IPP_FORMAT_SOURCE | + DRM_EXYNOS_IPP_FORMAT_DESTINATION; + formats[i].limits = limits; } + ctx->formats = formats; /* resource memory */ ctx->regs_res = platform_get_resource(pdev, IORESOURCE_MEM, 0); @@ -1670,9 +1335,8 @@ static int fimc_probe(struct platform_device *pdev) return -ENOENT; } - ctx->irq = res->start; - ret = devm_request_threaded_irq(dev, ctx->irq, NULL, fimc_irq_handler, - IRQF_ONESHOT, "drm_fimc", ctx); + ret = devm_request_irq(dev, res->start, fimc_irq_handler, + 0, dev_name(dev), ctx); if (ret < 0) { dev_err(dev, "failed to request irq.\n"); return ret; @@ -1682,39 +1346,24 @@ static int fimc_probe(struct platform_device *pdev) if (ret < 0) return ret; - ippdrv = &ctx->ippdrv; - ippdrv->ops[EXYNOS_DRM_OPS_SRC] = &fimc_src_ops; - ippdrv->ops[EXYNOS_DRM_OPS_DST] = &fimc_dst_ops; - ippdrv->check_property = fimc_ippdrv_check_property; - ippdrv->reset = fimc_ippdrv_reset; - ippdrv->start = fimc_ippdrv_start; - ippdrv->stop = fimc_ippdrv_stop; - ret = fimc_init_prop_list(ippdrv); - if (ret < 0) { - dev_err(dev, "failed to init property list.\n"); - goto err_put_clk; - } - - DRM_DEBUG_KMS("id[%d]ippdrv[%pK]\n", ctx->id, ippdrv); - spin_lock_init(&ctx->lock); platform_set_drvdata(pdev, ctx); + pm_runtime_use_autosuspend(dev); + pm_runtime_set_autosuspend_delay(dev, FIMC_AUTOSUSPEND_DELAY); pm_runtime_enable(dev); - ret = exynos_drm_ippdrv_register(ippdrv); - if (ret < 0) { - dev_err(dev, "failed to register drm fimc device.\n"); + ret = component_add(dev, &fimc_component_ops); + if (ret) goto err_pm_dis; - } dev_info(dev, "drm fimc registered successfully.\n"); return 0; err_pm_dis: + pm_runtime_dont_use_autosuspend(dev); pm_runtime_disable(dev); -err_put_clk: fimc_put_clocks(ctx); return ret; @@ -1724,42 +1373,24 @@ static int fimc_remove(struct platform_device *pdev) { struct device *dev = &pdev->dev; struct fimc_context *ctx = get_fimc_context(dev); - struct exynos_drm_ippdrv *ippdrv = &ctx->ippdrv; - exynos_drm_ippdrv_unregister(ippdrv); + component_del(dev, &fimc_component_ops); + pm_runtime_dont_use_autosuspend(dev); + pm_runtime_disable(dev); fimc_put_clocks(ctx); - pm_runtime_set_suspended(dev); - pm_runtime_disable(dev); return 0; } #ifdef CONFIG_PM -static int fimc_clk_ctrl(struct fimc_context *ctx, bool enable) -{ - DRM_DEBUG_KMS("enable[%d]\n", enable); - - if (enable) { - clk_prepare_enable(ctx->clocks[FIMC_CLK_GATE]); - clk_prepare_enable(ctx->clocks[FIMC_CLK_WB_A]); - ctx->suspended = false; - } else { - clk_disable_unprepare(ctx->clocks[FIMC_CLK_GATE]); - clk_disable_unprepare(ctx->clocks[FIMC_CLK_WB_A]); - ctx->suspended = true; - } - - return 0; -} - static int fimc_runtime_suspend(struct device *dev) { struct fimc_context *ctx = get_fimc_context(dev); DRM_DEBUG_KMS("id[%d]\n", ctx->id); - - return fimc_clk_ctrl(ctx, false); + clk_disable_unprepare(ctx->clocks[FIMC_CLK_GATE]); + return 0; } static int fimc_runtime_resume(struct device *dev) @@ -1767,8 +1398,7 @@ static int fimc_runtime_resume(struct device *dev) struct fimc_context *ctx = get_fimc_context(dev); DRM_DEBUG_KMS("id[%d]\n", ctx->id); - - return fimc_clk_ctrl(ctx, true); + return clk_prepare_enable(ctx->clocks[FIMC_CLK_GATE]); } #endif diff --git a/drivers/gpu/drm/exynos/exynos_drm_fimc.h b/drivers/gpu/drm/exynos/exynos_drm_fimc.h deleted file mode 100644 index 127a424c5fdf..000000000000 --- a/drivers/gpu/drm/exynos/exynos_drm_fimc.h +++ /dev/null @@ -1,23 +0,0 @@ -/* - * Copyright (c) 2012 Samsung Electronics Co., Ltd. - * - * Authors: - * Eunchul Kim <chulspro.kim@samsung.com> - * Jinyoung Jeon <jy0.jeon@samsung.com> - * Sangmin Lee <lsmin.lee@samsung.com> - * - * This program is free software; you can redistribute it and/or modify it - * under the terms of the GNU General Public License as published by the - * Free Software Foundation; either version 2 of the License, or (at your - * option) any later version. - */ - -#ifndef _EXYNOS_DRM_FIMC_H_ -#define _EXYNOS_DRM_FIMC_H_ - -/* - * TODO - * FIMD output interface notifier callback. - */ - -#endif /* _EXYNOS_DRM_FIMC_H_ */ -- 2.14.2 ^ permalink raw reply related [flat|nested] 20+ messages in thread
[parent not found: <CGME20171023075452eucas1p1dada05e443a800977d9c888744877f38@eucas1p1.samsung.com>]
* [PATCH v4 7/9] drm/exynos: Add driver for Exynos Scaler module [not found] ` <CGME20171023075452eucas1p1dada05e443a800977d9c888744877f38@eucas1p1.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski 0 siblings, 0 replies; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Marek Szyprowski, Inki Dae, Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon From: Andrzej Pietrasiewicz <andrzej.p@samsung.com> Exynos Scaler is a hardware module, which processes graphic data fetched from memory and transfers the resultant dato another memory buffer. Graphics data can be up/down-scaled, rotated, flipped and converted color space. Scaler hardware modules are a part of Exynos5420 and newer Exynos SoCs. Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> --- .../devicetree/bindings/gpu/samsung-scaler.txt | 27 + drivers/gpu/drm/exynos/Kconfig | 6 + drivers/gpu/drm/exynos/Makefile | 1 + drivers/gpu/drm/exynos/exynos_drm_drv.c | 3 + drivers/gpu/drm/exynos/exynos_drm_drv.h | 1 + drivers/gpu/drm/exynos/exynos_drm_scaler.c | 716 +++++++++++++++++++++ drivers/gpu/drm/exynos/regs-scaler.h | 426 ++++++++++++ 7 files changed, 1180 insertions(+) create mode 100644 Documentation/devicetree/bindings/gpu/samsung-scaler.txt create mode 100644 drivers/gpu/drm/exynos/exynos_drm_scaler.c create mode 100644 drivers/gpu/drm/exynos/regs-scaler.h diff --git a/Documentation/devicetree/bindings/gpu/samsung-scaler.txt b/Documentation/devicetree/bindings/gpu/samsung-scaler.txt new file mode 100644 index 000000000000..9c3d98105dfd --- /dev/null +++ b/Documentation/devicetree/bindings/gpu/samsung-scaler.txt @@ -0,0 +1,27 @@ +* Samsung Exynos Image Scaler + +Required properties: + - compatible : value should be one of the following: + (a) "samsung,exynos5420-scaler" for Scaler IP in Exynos5420 + (b) "samsung,exynos5433-scaler" for Scaler IP in Exynos5433 + + - reg : Physical base address of the IP registers and length of memory + mapped region. + + - interrupts : Interrupt specifier for scaler interrupt, according to format + specific to interrupt parent. + + - clocks : Clock specifier for scaler clock, according to generic clock + bindings. (See Documentation/devicetree/bindings/clock/exynos*.txt) + + - clock-names : Names of clocks. For exynos scaler, it should be "mscl" + on 5420 and "pclk", "aclk" and "aclk_xiu" on 5433. + +Example: + scaler@12800000 { + compatible = "samsung,exynos5420-scaler"; + reg = <0x12800000 0x1294>; + interrupts = <0 220 IRQ_TYPE_LEVEL_HIGH>; + clocks = <&clock CLK_MSCL0>; + clock-names = "mscl"; + }; diff --git a/drivers/gpu/drm/exynos/Kconfig b/drivers/gpu/drm/exynos/Kconfig index 73f06a3a8bec..bb6fad57e18a 100644 --- a/drivers/gpu/drm/exynos/Kconfig +++ b/drivers/gpu/drm/exynos/Kconfig @@ -109,6 +109,12 @@ config DRM_EXYNOS_ROTATOR help Choose this option if you want to use Exynos Rotator for DRM. +config DRM_EXYNOS_SCALER + bool "Scaler" + select DRM_EXYNOS_IPP + help + Choose this option if you want to use Exynos Scaler for DRM. + config DRM_EXYNOS_GSC bool "GScaler" depends on VIDEO_SAMSUNG_EXYNOS_GSC=n diff --git a/drivers/gpu/drm/exynos/Makefile b/drivers/gpu/drm/exynos/Makefile index f663490e949d..aa84472ac093 100644 --- a/drivers/gpu/drm/exynos/Makefile +++ b/drivers/gpu/drm/exynos/Makefile @@ -20,6 +20,7 @@ exynosdrm-$(CONFIG_DRM_EXYNOS_G2D) += exynos_drm_g2d.o exynosdrm-$(CONFIG_DRM_EXYNOS_IPP) += exynos_drm_ipp.o exynosdrm-$(CONFIG_DRM_EXYNOS_FIMC) += exynos_drm_fimc.o exynosdrm-$(CONFIG_DRM_EXYNOS_ROTATOR) += exynos_drm_rotator.o +exynosdrm-$(CONFIG_DRM_EXYNOS_SCALER) += exynos_drm_scaler.o exynosdrm-$(CONFIG_DRM_EXYNOS_GSC) += exynos_drm_gsc.o exynosdrm-$(CONFIG_DRM_EXYNOS_MIC) += exynos_drm_mic.o diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.c b/drivers/gpu/drm/exynos/exynos_drm_drv.c index 4afeffa024f3..b2582ae993ff 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.c +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.c @@ -260,6 +260,9 @@ static struct exynos_drm_driver_info exynos_drm_drivers[] = { }, { DRV_PTR(rotator_driver, CONFIG_DRM_EXYNOS_ROTATOR), DRM_COMPONENT_DRIVER + }, { + DRV_PTR(scaler_driver, CONFIG_DRM_EXYNOS_SCALER), + DRM_COMPONENT_DRIVER }, { DRV_PTR(gsc_driver, CONFIG_DRM_EXYNOS_GSC), DRM_COMPONENT_DRIVER diff --git a/drivers/gpu/drm/exynos/exynos_drm_drv.h b/drivers/gpu/drm/exynos/exynos_drm_drv.h index 8b3b31d35168..8dc8537f4fd1 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_drv.h +++ b/drivers/gpu/drm/exynos/exynos_drm_drv.h @@ -293,6 +293,7 @@ extern struct platform_driver vidi_driver; extern struct platform_driver g2d_driver; extern struct platform_driver fimc_driver; extern struct platform_driver rotator_driver; +extern struct platform_driver scaler_driver; extern struct platform_driver gsc_driver; extern struct platform_driver mic_driver; #endif diff --git a/drivers/gpu/drm/exynos/exynos_drm_scaler.c b/drivers/gpu/drm/exynos/exynos_drm_scaler.c new file mode 100644 index 000000000000..d82b474d8078 --- /dev/null +++ b/drivers/gpu/drm/exynos/exynos_drm_scaler.c @@ -0,0 +1,716 @@ +/* + * Copyright (C) 2017 Samsung Electronics Co.Ltd + * Author: + * Andrzej Pietrasiewicz <andrzej.p@samsung.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundationr + */ + +#include <linux/kernel.h> +#include <linux/component.h> +#include <linux/err.h> +#include <linux/interrupt.h> +#include <linux/io.h> +#include <linux/platform_device.h> +#include <linux/clk.h> +#include <linux/of_device.h> +#include <linux/pm_runtime.h> + +#include <drm/drmP.h> +#include <drm/exynos_drm.h> +#include "regs-scaler.h" +#include "exynos_drm_fb.h" +#include "exynos_drm_drv.h" +#include "exynos_drm_iommu.h" +#include "exynos_drm_ipp.h" + +#define scaler_read(offset) readl(scaler->regs + (offset)) +#define scaler_write(cfg, offset) writel(cfg, scaler->regs + (offset)) +#define SCALER_MAX_CLK 4 +#define SCALER_AUTOSUSPEND_DELAY 2000 + +struct scaler_data { + const char *clk_name[SCALER_MAX_CLK]; + unsigned num_clk; + const struct exynos_drm_ipp_formats *formats; +}; + +struct scaler_context { + struct exynos_drm_ipp ipp; + struct drm_device *drm_dev; + struct device *dev; + void __iomem *regs; + struct clk *clock[SCALER_MAX_CLK]; + struct exynos_drm_ipp_task *task; + const struct scaler_data *scaler_data; +}; + +static u32 scaler_get_format(u32 drm_fmt) +{ + switch (drm_fmt) { + case DRM_FORMAT_NV21: + return SCALER_YUV420_2P_UV; + case DRM_FORMAT_NV12: + return SCALER_YUV420_2P_VU; + case DRM_FORMAT_YUV420: + return SCALER_YUV420_3P; + case DRM_FORMAT_YUYV: + return SCALER_YUV422_1P_YUYV; + case DRM_FORMAT_UYVY: + return SCALER_YUV422_1P_UYVY; + case DRM_FORMAT_YVYU: + return SCALER_YUV422_1P_YVYU; + case DRM_FORMAT_NV61: + return SCALER_YUV422_2P_UV; + case DRM_FORMAT_NV16: + return SCALER_YUV422_2P_VU; + case DRM_FORMAT_YUV422: + return SCALER_YUV422_3P; + case DRM_FORMAT_NV42: + return SCALER_YUV444_2P_UV; + case DRM_FORMAT_NV24: + return SCALER_YUV444_2P_VU; + case DRM_FORMAT_YUV444: + return SCALER_YUV444_3P; + case DRM_FORMAT_RGB565: + return SCALER_RGB_565; + case DRM_FORMAT_XRGB1555: + return SCALER_ARGB1555; + case DRM_FORMAT_ARGB1555: + return SCALER_ARGB1555; + case DRM_FORMAT_XRGB4444: + return SCALER_ARGB4444; + case DRM_FORMAT_ARGB4444: + return SCALER_ARGB4444; + case DRM_FORMAT_XRGB8888: + return SCALER_ARGB8888; + case DRM_FORMAT_ARGB8888: + return SCALER_ARGB8888; + case DRM_FORMAT_RGBX8888: + return SCALER_RGBA8888; + case DRM_FORMAT_RGBA8888: + return SCALER_RGBA8888; + default: + break; + } + + return 0; +} + +static int scaler_check(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) +{ + struct exynos_drm_ipp_buffer *src_buf = &task->src; + struct exynos_drm_ipp_buffer *dst_buf = &task->dst; + + if (src_buf->format->num_planes > 2 && + src_buf->buf.pitch[2] != src_buf->buf.pitch[1]) + return -EINVAL; + + if (dst_buf->format->num_planes > 2 && + dst_buf->buf.pitch[2] != dst_buf->buf.pitch[1]) + return -EINVAL; + + return 0; +} + +static inline void scaler_enable_int(struct scaler_context *scaler) +{ + u32 val; + + val = SCALER_INT_EN_TIMEOUT | + SCALER_INT_EN_ILLEGAL_BLEND | + SCALER_INT_EN_ILLEGAL_RATIO | + SCALER_INT_EN_ILLEGAL_DST_HEIGHT | + SCALER_INT_EN_ILLEGAL_DST_WIDTH | + SCALER_INT_EN_ILLEGAL_DST_V_POS | + SCALER_INT_EN_ILLEGAL_DST_H_POS | + SCALER_INT_EN_ILLEGAL_DST_C_SPAN | + SCALER_INT_EN_ILLEGAL_DST_Y_SPAN | + SCALER_INT_EN_ILLEGAL_DST_CR_BASE | + SCALER_INT_EN_ILLEGAL_DST_CB_BASE | + SCALER_INT_EN_ILLEGAL_DST_Y_BASE | + SCALER_INT_EN_ILLEGAL_DST_COLOR | + SCALER_INT_EN_ILLEGAL_SRC_HEIGHT | + SCALER_INT_EN_ILLEGAL_SRC_WIDTH | + SCALER_INT_EN_ILLEGAL_SRC_CV_POS | + SCALER_INT_EN_ILLEGAL_SRC_CH_POS | + SCALER_INT_EN_ILLEGAL_SRC_YV_POS | + SCALER_INT_EN_ILLEGAL_SRC_YH_POS | + SCALER_INT_EN_ILLEGAL_DST_SPAN | + SCALER_INT_EN_ILLEGAL_SRC_Y_SPAN | + SCALER_INT_EN_ILLEGAL_SRC_CR_BASE | + SCALER_INT_EN_ILLEGAL_SRC_CB_BASE | + SCALER_INT_EN_ILLEGAL_SRC_Y_BASE | + SCALER_INT_EN_ILLEGAL_SRC_COLOR | + SCALER_INT_EN_FRAME_END; + scaler_write(val, SCALER_INT_EN); +} + +static inline void scaler_set_src_fmt(struct scaler_context *scaler, + u32 src_fmt) +{ + u32 val; + + val = SCALER_SRC_CFG_SET_COLOR_FORMAT(src_fmt); + scaler_write(val, SCALER_SRC_CFG); +} + +static inline void scaler_set_src_base(struct scaler_context *scaler, + struct exynos_drm_ipp_buffer *src_buf) +{ + static unsigned int bases[] = { + SCALER_SRC_Y_BASE, + SCALER_SRC_CB_BASE, + SCALER_SRC_CR_BASE, + }; + int i; + + for (i = 0; i < src_buf->format->num_planes; ++i) + scaler_write(src_buf->dma_addr[i], bases[i]); +} + +static inline void scaler_set_src_span(struct scaler_context *scaler, + struct exynos_drm_ipp_buffer *src_buf) +{ + u32 val; + + val = SCALER_SRC_SPAN_SET_Y_SPAN(src_buf->buf.pitch[0] / + src_buf->format->cpp[0]); + + if (src_buf->format->num_planes > 1) + val |= SCALER_SRC_SPAN_SET_C_SPAN(src_buf->buf.pitch[1]); + + scaler_write(val, SCALER_SRC_SPAN); +} + +static inline void scaler_set_src_luma_pos(struct scaler_context *scaler, + struct drm_exynos_ipp_task_rect *src_pos) +{ + u32 val; + + val = SCALER_SRC_Y_POS_SET_YH_POS(src_pos->x << 2); + val |= SCALER_SRC_Y_POS_SET_YV_POS(src_pos->y << 2); + scaler_write(val, SCALER_SRC_Y_POS); + scaler_write(val, SCALER_SRC_C_POS); /* ATTENTION! */ +} + +static inline void scaler_set_src_wh(struct scaler_context *scaler, + struct drm_exynos_ipp_task_rect *src_pos) +{ + u32 val; + + val = SCALER_SRC_WH_SET_WIDTH(src_pos->w); + val |= SCALER_SRC_WH_SET_HEIGHT(src_pos->h); + scaler_write(val, SCALER_SRC_WH); +} + +static inline void scaler_set_dst_fmt(struct scaler_context *scaler, + u32 dst_fmt) +{ + u32 val; + + val = SCALER_DST_CFG_SET_COLOR_FORMAT(dst_fmt); + scaler_write(val, SCALER_DST_CFG); +} + +static inline void scaler_set_dst_base(struct scaler_context *scaler, + struct exynos_drm_ipp_buffer *dst_buf) +{ + static unsigned int bases[] = { + SCALER_DST_Y_BASE, + SCALER_DST_CB_BASE, + SCALER_DST_CR_BASE, + }; + int i; + + for (i = 0; i < dst_buf->format->num_planes; ++i) + scaler_write(dst_buf->dma_addr[i], bases[i]); +} + +static inline void scaler_set_dst_span(struct scaler_context *scaler, + struct exynos_drm_ipp_buffer *dst_buf) +{ + u32 val; + + val = SCALER_DST_SPAN_SET_Y_SPAN(dst_buf->buf.pitch[0] / + dst_buf->format->cpp[0]); + + if (dst_buf->format->num_planes > 1) + val |= SCALER_DST_SPAN_SET_C_SPAN(dst_buf->buf.pitch[1]); + + scaler_write(val, SCALER_DST_SPAN); +} + +static inline void scaler_set_dst_luma_pos(struct scaler_context *scaler, + struct drm_exynos_ipp_task_rect *dst_pos) +{ + u32 val; + + val = SCALER_DST_WH_SET_WIDTH(dst_pos->w); + val |= SCALER_DST_WH_SET_HEIGHT(dst_pos->h); + scaler_write(val, SCALER_DST_WH); +} + +static inline void scaler_set_dst_wh(struct scaler_context *scaler, + struct drm_exynos_ipp_task_rect *dst_pos) +{ + u32 val; + + val = SCALER_DST_POS_SET_H_POS(dst_pos->x); + val |= SCALER_DST_POS_SET_V_POS(dst_pos->y); + scaler_write(val, SCALER_DST_POS); +} + +static inline void scaler_set_hv_ratio(struct scaler_context *scaler, + unsigned int rotation, + struct drm_exynos_ipp_task_rect *src_pos, + struct drm_exynos_ipp_task_rect *dst_pos) +{ + u32 val, h_ratio, v_ratio; + + if (drm_rotation_90_or_270(rotation)) { + h_ratio = (src_pos->h << 16) / dst_pos->w; + v_ratio = (src_pos->w << 16) / dst_pos->h; + } else { + h_ratio = (src_pos->w << 16) / dst_pos->w; + v_ratio = (src_pos->h << 16) / dst_pos->h; + } + + val = SCALER_H_RATIO_SET(h_ratio); + scaler_write(val, SCALER_H_RATIO); + + val = SCALER_V_RATIO_SET(v_ratio); + scaler_write(val, SCALER_V_RATIO); +} + +static inline void scaler_set_rotation(struct scaler_context *scaler, + unsigned int rotation) +{ + u32 val = 0; + + if (rotation & DRM_MODE_ROTATE_90) + val |= SCALER_ROT_CFG_SET_ROTMODE(SCALER_ROT_MODE_90); + else if (rotation & DRM_MODE_ROTATE_180) + val |= SCALER_ROT_CFG_SET_ROTMODE(SCALER_ROT_MODE_180); + else if (rotation & DRM_MODE_ROTATE_270) + val |= SCALER_ROT_CFG_SET_ROTMODE(SCALER_ROT_MODE_270); + if (rotation & DRM_MODE_REFLECT_X) + val |= SCALER_ROT_CFG_FLIP_X_EN; + if (rotation & DRM_MODE_REFLECT_Y) + val |= SCALER_ROT_CFG_FLIP_Y_EN; + scaler_write(val, SCALER_ROT_CFG); +} + +static inline void scaler_set_csc(struct scaler_context *scaler, + const struct drm_format_info *fmt) +{ + static const u32 csc_mtx[2][3][3] = { + { /* YCbCr to RGB */ + {0x254, 0x000, 0x331}, + {0x254, 0xf38, 0xe60}, + {0x254, 0x409, 0x000}, + }, + { /* RGB to YCbCr */ + {0x084, 0x102, 0x032}, + {0xfb4, 0xf6b, 0x0e1}, + {0x0e1, 0xf44, 0xfdc}, + }, + }; + int i, j, dir; + + switch (fmt->format) { + case DRM_FORMAT_RGB565: + case DRM_FORMAT_XRGB1555: + case DRM_FORMAT_ARGB1555: + case DRM_FORMAT_XRGB4444: + case DRM_FORMAT_ARGB4444: + case DRM_FORMAT_XRGB8888: + case DRM_FORMAT_ARGB8888: + case DRM_FORMAT_RGBX8888: + case DRM_FORMAT_RGBA8888: + dir = 1; + break; + default: + dir = 0; + } + + for (i = 0; i < 3; i++) + for (j = 0; j < 3; j++) + scaler_write(csc_mtx[dir][i][j], SCALER_CSC_COEF(j, i)); +} + +static inline void scaler_set_timer(struct scaler_context *scaler, + unsigned int timer, unsigned int divider) +{ + u32 val; + + val = SCALER_TIMEOUT_CTRL_TIMER_ENABLE; + val |= SCALER_TIMEOUT_CTRL_SET_TIMER_VALUE(timer); + val |= SCALER_TIMEOUT_CTRL_SET_TIMER_DIV(divider); + scaler_write(val, SCALER_TIMEOUT_CTRL); +} + +static inline void scaler_start_hw(struct scaler_context *scaler) +{ + scaler_write(SCALER_CFG_START_CMD, SCALER_CFG); +} + +static int scaler_commit(struct exynos_drm_ipp *ipp, + struct exynos_drm_ipp_task *task) +{ + struct scaler_context *scaler = + container_of(ipp, struct scaler_context, ipp); + + u32 src_fmt = scaler_get_format(task->src.buf.fourcc); + struct drm_exynos_ipp_task_rect *src_pos = &task->src.rect; + + u32 dst_fmt = scaler_get_format(task->dst.buf.fourcc); + struct drm_exynos_ipp_task_rect *dst_pos = &task->dst.rect; + + scaler->task = task; + + pm_runtime_get_sync(scaler->dev); + + scaler_set_src_fmt(scaler, src_fmt); + scaler_set_src_base(scaler, &task->src); + scaler_set_src_span(scaler, &task->src); + scaler_set_src_luma_pos(scaler, src_pos); + scaler_set_src_wh(scaler, src_pos); + + scaler_set_dst_fmt(scaler, dst_fmt); + scaler_set_dst_base(scaler, &task->dst); + scaler_set_dst_span(scaler, &task->dst); + scaler_set_dst_luma_pos(scaler, dst_pos); + scaler_set_dst_wh(scaler, dst_pos); + + scaler_set_hv_ratio(scaler, task->transform.rotation, src_pos, dst_pos); + scaler_set_rotation(scaler, task->transform.rotation); + + scaler_set_csc(scaler, task->src.format); + + scaler_set_timer(scaler, 0xffff, 0xf); + + scaler_enable_int(scaler); + scaler_start_hw(scaler); + + return 0; +} + +static struct exynos_drm_ipp_funcs ipp_funcs = { + .check = scaler_check, + .commit = scaler_commit, +}; + +static inline void scaler_disable_int(struct scaler_context *scaler) +{ + scaler_write(0, SCALER_INT_EN); +} + +static inline u32 scaler_get_int_status(struct scaler_context *scaler) +{ + return scaler_read(SCALER_INT_STATUS); +} + +static inline bool scaler_task_done(u32 val) +{ + return val & SCALER_INT_STATUS_FRAME_END ? 0 : -EINVAL; +} + +static irqreturn_t scaler_irq_handler(int irq, void *arg) +{ + struct scaler_context *scaler = arg; + + u32 val = scaler_get_int_status(scaler); + + scaler_disable_int(scaler); + + if (scaler->task) { + struct exynos_drm_ipp_task *task = scaler->task; + + scaler->task = NULL; + pm_runtime_mark_last_busy(scaler->dev); + pm_runtime_put_autosuspend(scaler->dev); + exynos_drm_ipp_task_done(task, scaler_task_done(val)); + } + + return IRQ_HANDLED; +} + +static int scaler_bind(struct device *dev, struct device *master, void *data) +{ + struct scaler_context *scaler = dev_get_drvdata(dev); + struct drm_device *drm_dev = data; + struct exynos_drm_ipp *ipp = &scaler->ipp; + + scaler->drm_dev = drm_dev; + drm_iommu_attach_device(drm_dev, dev); + + exynos_drm_ipp_register(drm_dev, ipp, &ipp_funcs, + DRM_EXYNOS_IPP_CAP_CROP | + DRM_EXYNOS_IPP_CAP_ROTATE | + DRM_EXYNOS_IPP_CAP_SCALE | + DRM_EXYNOS_IPP_CAP_CONVERT, + scaler->scaler_data->formats, "scaler"); + + dev_info(dev, "The exynos scaler has been probed successfully\n"); + + return 0; +} + +static void scaler_unbind(struct device *dev, struct device *master, + void *data) +{ + struct scaler_context *scaler = dev_get_drvdata(dev); + struct drm_device *drm_dev = data; + struct exynos_drm_ipp *ipp = &scaler->ipp; + + exynos_drm_ipp_unregister(drm_dev, ipp); + drm_iommu_detach_device(scaler->drm_dev, scaler->dev); +} + +static const struct component_ops scaler_component_ops = { + .bind = scaler_bind, + .unbind = scaler_unbind, +}; + +static int scaler_probe(struct platform_device *pdev) +{ + struct device *dev = &pdev->dev; + struct resource *regs_res; + struct scaler_context *scaler; + int irq; + int ret, i; + + scaler = devm_kzalloc(dev, sizeof(*scaler), GFP_KERNEL); + if (!scaler) + return -ENOMEM; + + scaler->scaler_data = + (struct scaler_data *)of_device_get_match_data(dev); + + scaler->dev = dev; + regs_res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + scaler->regs = devm_ioremap_resource(dev, regs_res); + if (IS_ERR(scaler->regs)) + return PTR_ERR(scaler->regs); + + irq = platform_get_irq(pdev, 0); + if (irq < 0) { + dev_err(dev, "failed to get irq\n"); + return irq; + } + + ret = devm_request_threaded_irq(dev, irq, NULL, scaler_irq_handler, + IRQF_ONESHOT, "drm_scaler", scaler); + if (ret < 0) { + dev_err(dev, "failed to request irq\n"); + return ret; + } + + for (i = 0; i < scaler->scaler_data->num_clk; ++i) { + scaler->clock[i] = devm_clk_get(dev, scaler->scaler_data->clk_name[i]); + if (IS_ERR(scaler->clock[i])) { + dev_err(dev, "failed to get clock\n"); + return PTR_ERR(scaler->clock[i]); + } + } + + pm_runtime_use_autosuspend(dev); + pm_runtime_set_autosuspend_delay(dev, SCALER_AUTOSUSPEND_DELAY); + pm_runtime_enable(dev); + platform_set_drvdata(pdev, scaler); + + ret = component_add(dev, &scaler_component_ops); + if (ret) + goto err_ippdrv_register; + + return 0; + +err_ippdrv_register: + pm_runtime_dont_use_autosuspend(dev); + pm_runtime_disable(dev); + return ret; +} + +static int scaler_remove(struct platform_device *pdev) +{ + struct device *dev = &pdev->dev; + + component_del(dev, &scaler_component_ops); + pm_runtime_dont_use_autosuspend(dev); + pm_runtime_disable(dev); + + return 0; +} + +#ifdef CONFIG_PM + +static int clk_disable_unprepare_wrapper(struct clk *clk) +{ + clk_disable_unprepare(clk); + + return 0; +} + +static int scaler_clk_ctrl(struct scaler_context *scaler, bool enable) +{ + int (*clk_fun)(struct clk *clk), i; + + clk_fun = enable ? clk_prepare_enable : clk_disable_unprepare_wrapper; + + for (i = 0; i < scaler->scaler_data->num_clk; ++i) + clk_fun(scaler->clock[i]); + + return 0; +} + +static int scaler_runtime_suspend(struct device *dev) +{ + struct scaler_context *scaler = dev_get_drvdata(dev); + + return scaler_clk_ctrl(scaler, false); +} + +static int scaler_runtime_resume(struct device *dev) +{ + struct scaler_context *scaler = dev_get_drvdata(dev); + + return scaler_clk_ctrl(scaler, true); +} +#endif + +static const struct dev_pm_ops scaler_pm_ops = { + SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend, + pm_runtime_force_resume) + SET_RUNTIME_PM_OPS(scaler_runtime_suspend, scaler_runtime_resume, + NULL) +}; + +static const struct drm_exynos_ipp_limit scaler_5420_two_pixel_hv_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = { 16, SZ_8K }, .v = { 16, SZ_8K }) }, + { IPP_SIZE_LIMIT(AREA, .h.align = 2, .v.align = 2) }, + { IPP_SCALE_LIMIT(.h = { 65536 * 1 / 4, 65536 * 16}, + .v = { 65536 * 1 / 4, 65536 * 16}) }, + { } +}; + +static const struct drm_exynos_ipp_limit scaler_5420_two_pixel_h_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = { 16, SZ_8K }, .v = { 16, SZ_8K }) }, + { IPP_SIZE_LIMIT(AREA, .h.align = 2, .v.align = 1) }, + { IPP_SCALE_LIMIT(.h = { 65536 * 1 / 4, 65536 * 16}, + .v = { 65536 * 1 / 4, 65536 * 16}) }, + { } +}; + +static const struct drm_exynos_ipp_limit scaler_5420_one_pixel_limits[] = { + { IPP_SIZE_LIMIT(BUFFER, .h = { 16, SZ_8K }, .v = { 16, SZ_8K }) }, + { IPP_SCALE_LIMIT(.h = { 65536 * 1 / 4, 65536 * 16}, + .v = { 65536 * 1 / 4, 65536 * 16}) }, + { } +}; + +static const struct exynos_drm_ipp_formats exynos5420_formats[] = { + /* SCALER_YUV420_2P_UV */ + { IPP_SRCDST_FORMAT(NV21, scaler_5420_two_pixel_hv_limits) }, + + /* SCALER_YUV420_2P_VU */ + { IPP_SRCDST_FORMAT(NV12, scaler_5420_two_pixel_hv_limits) }, + + /* SCALER_YUV420_3P */ + { IPP_SRCDST_FORMAT(YUV420, scaler_5420_two_pixel_hv_limits) }, + + /* SCALER_YUV422_1P_YUYV */ + { IPP_SRCDST_FORMAT(YUYV, scaler_5420_two_pixel_h_limits) }, + + /* SCALER_YUV422_1P_UYVY */ + { IPP_SRCDST_FORMAT(UYVY, scaler_5420_two_pixel_h_limits) }, + + /* SCALER_YUV422_1P_YVYU */ + { IPP_SRCDST_FORMAT(YVYU, scaler_5420_two_pixel_h_limits) }, + + /* SCALER_YUV422_2P_UV */ + { IPP_SRCDST_FORMAT(NV61, scaler_5420_two_pixel_h_limits) }, + + /* SCALER_YUV422_2P_VU */ + { IPP_SRCDST_FORMAT(NV16, scaler_5420_two_pixel_h_limits) }, + + /* SCALER_YUV422_3P */ + { IPP_SRCDST_FORMAT(YUV422, scaler_5420_two_pixel_h_limits) }, + + /* SCALER_YUV444_2P_UV */ + { IPP_SRCDST_FORMAT(NV42, scaler_5420_one_pixel_limits) }, + + /* SCALER_YUV444_2P_VU */ + { IPP_SRCDST_FORMAT(NV24, scaler_5420_one_pixel_limits) }, + + /* SCALER_YUV444_3P */ + { IPP_SRCDST_FORMAT(YUV444, scaler_5420_one_pixel_limits) }, + + /* SCALER_RGB_565 */ + { IPP_SRCDST_FORMAT(RGB565, scaler_5420_one_pixel_limits) }, + + /* SCALER_ARGB1555 */ + { IPP_SRCDST_FORMAT(XRGB1555, scaler_5420_one_pixel_limits) }, + + /* SCALER_ARGB1555 */ + { IPP_SRCDST_FORMAT(ARGB1555, scaler_5420_one_pixel_limits) }, + + /* SCALER_ARGB4444 */ + { IPP_SRCDST_FORMAT(XRGB4444, scaler_5420_one_pixel_limits) }, + + /* SCALER_ARGB4444 */ + { IPP_SRCDST_FORMAT(ARGB4444, scaler_5420_one_pixel_limits) }, + + /* SCALER_ARGB8888 */ + { IPP_SRCDST_FORMAT(XRGB8888, scaler_5420_one_pixel_limits) }, + + /* SCALER_ARGB8888 */ + { IPP_SRCDST_FORMAT(ARGB8888, scaler_5420_one_pixel_limits) }, + + /* SCALER_RGBA8888 */ + { IPP_SRCDST_FORMAT(RGBX8888, scaler_5420_one_pixel_limits) }, + + /* SCALER_RGBA8888 */ + { IPP_SRCDST_FORMAT(RGBA8888, scaler_5420_one_pixel_limits) }, + + { }, +}; + +static const struct scaler_data exynos5420_data = { + .clk_name = {"mscl"}, + .num_clk = 1, + .formats = exynos5420_formats, +}; + +static const struct scaler_data exynos5433_data = { + .clk_name = {"pclk", "aclk", "aclk_xiu"}, + .num_clk = 3, + .formats = exynos5420_formats, /* intentional */ +}; + +static const struct of_device_id exynos_scaler_match[] = { + { + .compatible = "samsung,exynos5420-scaler", + .data = &exynos5420_data, + }, + { + .compatible = "samsung,exynos5433-scaler", + .data = &exynos5433_data, + }, + {}, +}; +MODULE_DEVICE_TABLE(of, exynos_scaler_match); + +struct platform_driver scaler_driver = { + .probe = scaler_probe, + .remove = scaler_remove, + .driver = { + .name = "exynos-scaler", + .owner = THIS_MODULE, + .pm = &scaler_pm_ops, + .of_match_table = exynos_scaler_match, + }, +}; diff --git a/drivers/gpu/drm/exynos/regs-scaler.h b/drivers/gpu/drm/exynos/regs-scaler.h new file mode 100644 index 000000000000..fc7ccad75e74 --- /dev/null +++ b/drivers/gpu/drm/exynos/regs-scaler.h @@ -0,0 +1,426 @@ +/* drivers/gpu/drm/exynos/regs-scaler.h + * + * Copyright (c) 2017 Samsung Electronics Co., Ltd. + * http://www.samsung.com/ + * Author: Andrzej Pietrasiewicz <andrzej.p@samsung.com> + * + * Register definition file for Samsung scaler driver + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#ifndef EXYNOS_REGS_SCALER_H +#define EXYNOS_REGS_SCALER_H + +/* Register part */ + +/* Global setting */ +#define SCALER_STATUS 0x0 /* no shadow */ +#define SCALER_CFG 0x4 + +/* Interrupt */ +#define SCALER_INT_EN 0x8 /* no shadow */ +#define SCALER_INT_STATUS 0xc /* no shadow */ + +/* SRC */ +#define SCALER_SRC_CFG 0x10 +#define SCALER_SRC_Y_BASE 0x14 +#define SCALER_SRC_CB_BASE 0x18 +#define SCALER_SRC_CR_BASE 0x294 +#define SCALER_SRC_SPAN 0x1c +#define SCALER_SRC_Y_POS 0x20 +#define SCALER_SRC_WH 0x24 +#define SCALER_SRC_C_POS 0x28 + +/* DST */ +#define SCALER_DST_CFG 0x30 +#define SCALER_DST_Y_BASE 0x34 +#define SCALER_DST_CB_BASE 0x38 +#define SCALER_DST_CR_BASE 0x298 +#define SCALER_DST_SPAN 0x3c +#define SCALER_DST_WH 0x40 +#define SCALER_DST_POS 0x44 + +/* Ratio */ +#define SCALER_H_RATIO 0x50 +#define SCALER_V_RATIO 0x54 + +/* Rotation */ +#define SCALER_ROT_CFG 0x58 + +/* Coefficient */ +/* + * YHCOEF_{x}{A|B|C|D} CHCOEF_{x}{A|B|C|D} + * + * A B C D A B C D + * 0 60 64 68 6c 140 144 148 14c + * 1 70 74 78 7c 150 154 158 15c + * 2 80 84 88 8c 160 164 168 16c + * 3 90 94 98 9c 170 174 178 17c + * 4 a0 a4 a8 ac 180 184 188 18c + * 5 b0 b4 b8 bc 190 194 198 19c + * 6 c0 c4 c8 cc 1a0 1a4 1a8 1ac + * 7 d0 d4 d8 dc 1b0 1b4 1b8 1bc + * 8 e0 e4 e8 ec 1c0 1c4 1c8 1cc + * + * + * YVCOEF_{x}{A|B} CVCOEF_{x}{A|B} + * + * A B A B + * 0 f0 f4 1d0 1d4 + * 1 f8 fc 1d8 1dc + * 2 100 104 1e0 1e4 + * 3 108 10c 1e8 1ec + * 4 110 114 1f0 1f4 + * 5 118 11c 1f8 1fc + * 6 120 124 200 204 + * 7 128 12c 208 20c + * 8 130 134 210 214 + */ +#define _SCALER_HCOEF_DELTA(r, c) ((r) * 0x10 + (c) * 0x4) +#define _SCALER_VCOEF_DELTA(r, c) ((r) * 0x8 + (c) * 0x4) + +#define SCALER_YHCOEF(r, c) (0x60 + _SCALER_HCOEF_DELTA((r), (c))) +#define SCALER_YVCOEF(r, c) (0xf0 + _SCALER_VCOEF_DELTA((r), (c))) +#define SCALER_CHCOEF(r, c) (0x140 + _SCALER_HCOEF_DELTA((r), (c))) +#define SCALER_CVCOEF(r, c) (0x1d0 + _SCALER_VCOEF_DELTA((r), (c))) + + +/* Color Space Conversion */ +#define SCALER_CSC_COEF(x, y) (0x220 + (y) * 0xc + (x) * 0x4) + +/* Dithering */ +#define SCALER_DITH_CFG 0x250 + +/* Version Number */ +#define SCALER_VER 0x260 /* no shadow */ + +/* Cycle count and Timeout */ +#define SCALER_CYCLE_COUNT 0x278 /* no shadow */ +#define SCALER_TIMEOUT_CTRL 0x2c0 /* no shadow */ +#define SCALER_TIMEOUT_CNT 0x2c4 /* no shadow */ + +/* Blending */ +#define SCALER_SRC_BLEND_COLOR 0x280 +#define SCALER_SRC_BLEND_ALPHA 0x284 +#define SCALER_DST_BLEND_COLOR 0x288 +#define SCALER_DST_BLEND_ALPHA 0x28c + +/* Color Fill */ +#define SCALER_FILL_COLOR 0x290 + +/* Multiple Command Queue */ +#define SCALER_ADDR_Q_CONFIG 0x2a0 /* no shadow */ +#define SCALER_SRC_ADDR_Q_STATUS 0x2a4 /* no shadow */ +#define SCALER_SRC_ADDR_Q 0x2a8 /* no shadow */ + +/* CRC */ +#define SCALER_CRC_COLOR00_10 0x2b0 /* no shadow */ +#define SCALER_CRC_COLOR20_30 0x2b4 /* no shadow */ +#define SCALER_CRC_COLOR01_11 0x2b8 /* no shadow */ +#define SCALER_CRC_COLOR21_31 0x2bc /* no shadow */ + +/* Shadow Registers */ +#define SCALER_SHADOW_OFFSET 0x1000 + + +/* Bit definition part */ +#define SCALER_MASK(hi_b, lo_b) ((1 << ((hi_b) - (lo_b) + 1)) - 1) +#define SCALER_GET(reg, hi_b, lo_b) \ + (((reg) >> (lo_b)) & SCALER_MASK(hi_b, lo_b)) +#define SCALER_SET(val, hi_b, lo_b) \ + (((val) & SCALER_MASK(hi_b, lo_b)) << lo_b) + +/* SCALER_STATUS */ +#define SCALER_STATUS_SCALER_RUNNING (1 << 1) +#define SCALER_STATUS_SCALER_READY_CLK_DOWN (1 << 0) + +/* SCALER_CFG */ +#define SCALER_CFG_FILL_EN (1 << 24) +#define SCALER_CFG_BLEND_COLOR_DIVIDE_ALPHA_EN (1 << 17) +#define SCALER_CFG_BLEND_EN (1 << 16) +#define SCALER_CFG_CSC_Y_OFFSET_SRC_EN (1 << 10) +#define SCALER_CFG_CSC_Y_OFFSET_DST_EN (1 << 9) +#define SCALER_CFG_16_BURST_MODE (1 << 8) +#define SCALER_CFG_SOFT_RESET (1 << 1) +#define SCALER_CFG_START_CMD (1 << 0) + +/* SCALER_INT_EN */ +#define SCALER_INT_EN_TIMEOUT (1 << 31) +#define SCALER_INT_EN_ILLEGAL_BLEND (1 << 24) +#define SCALER_INT_EN_ILLEGAL_RATIO (1 << 23) +#define SCALER_INT_EN_ILLEGAL_DST_HEIGHT (1 << 22) +#define SCALER_INT_EN_ILLEGAL_DST_WIDTH (1 << 21) +#define SCALER_INT_EN_ILLEGAL_DST_V_POS (1 << 20) +#define SCALER_INT_EN_ILLEGAL_DST_H_POS (1 << 19) +#define SCALER_INT_EN_ILLEGAL_DST_C_SPAN (1 << 18) +#define SCALER_INT_EN_ILLEGAL_DST_Y_SPAN (1 << 17) +#define SCALER_INT_EN_ILLEGAL_DST_CR_BASE (1 << 16) +#define SCALER_INT_EN_ILLEGAL_DST_CB_BASE (1 << 15) +#define SCALER_INT_EN_ILLEGAL_DST_Y_BASE (1 << 14) +#define SCALER_INT_EN_ILLEGAL_DST_COLOR (1 << 13) +#define SCALER_INT_EN_ILLEGAL_SRC_HEIGHT (1 << 12) +#define SCALER_INT_EN_ILLEGAL_SRC_WIDTH (1 << 11) +#define SCALER_INT_EN_ILLEGAL_SRC_CV_POS (1 << 10) +#define SCALER_INT_EN_ILLEGAL_SRC_CH_POS (1 << 9) +#define SCALER_INT_EN_ILLEGAL_SRC_YV_POS (1 << 8) +#define SCALER_INT_EN_ILLEGAL_SRC_YH_POS (1 << 7) +#define SCALER_INT_EN_ILLEGAL_DST_SPAN (1 << 6) +#define SCALER_INT_EN_ILLEGAL_SRC_Y_SPAN (1 << 5) +#define SCALER_INT_EN_ILLEGAL_SRC_CR_BASE (1 << 4) +#define SCALER_INT_EN_ILLEGAL_SRC_CB_BASE (1 << 3) +#define SCALER_INT_EN_ILLEGAL_SRC_Y_BASE (1 << 2) +#define SCALER_INT_EN_ILLEGAL_SRC_COLOR (1 << 1) +#define SCALER_INT_EN_FRAME_END (1 << 0) + +/* SCALER_INT_STATUS */ +#define SCALER_INT_STATUS_TIMEOUT (1 << 31) +#define SCALER_INT_STATUS_ILLEGAL_BLEND (1 << 24) +#define SCALER_INT_STATUS_ILLEGAL_RATIO (1 << 23) +#define SCALER_INT_STATUS_ILLEGAL_DST_HEIGHT (1 << 22) +#define SCALER_INT_STATUS_ILLEGAL_DST_WIDTH (1 << 21) +#define SCALER_INT_STATUS_ILLEGAL_DST_V_POS (1 << 20) +#define SCALER_INT_STATUS_ILLEGAL_DST_H_POS (1 << 19) +#define SCALER_INT_STATUS_ILLEGAL_DST_C_SPAN (1 << 18) +#define SCALER_INT_STATUS_ILLEGAL_DST_Y_SPAN (1 << 17) +#define SCALER_INT_STATUS_ILLEGAL_DST_CR_BASE (1 << 16) +#define SCALER_INT_STATUS_ILLEGAL_DST_CB_BASE (1 << 15) +#define SCALER_INT_STATUS_ILLEGAL_DST_Y_BASE (1 << 14) +#define SCALER_INT_STATUS_ILLEGAL_DST_COLOR (1 << 13) +#define SCALER_INT_STATUS_ILLEGAL_SRC_HEIGHT (1 << 12) +#define SCALER_INT_STATUS_ILLEGAL_SRC_WIDTH (1 << 11) +#define SCALER_INT_STATUS_ILLEGAL_SRC_CV_POS (1 << 10) +#define SCALER_INT_STATUS_ILLEGAL_SRC_CH_POS (1 << 9) +#define SCALER_INT_STATUS_ILLEGAL_SRC_YV_POS (1 << 8) +#define SCALER_INT_STATUS_ILLEGAL_SRC_YH_POS (1 << 7) +#define SCALER_INT_STATUS_ILLEGAL_DST_SPAN (1 << 6) +#define SCALER_INT_STATUS_ILLEGAL_SRC_Y_SPAN (1 << 5) +#define SCALER_INT_STATUS_ILLEGAL_SRC_CR_BASE (1 << 4) +#define SCALER_INT_STATUS_ILLEGAL_SRC_CB_BASE (1 << 3) +#define SCALER_INT_STATUS_ILLEGAL_SRC_Y_BASE (1 << 2) +#define SCALER_INT_STATUS_ILLEGAL_SRC_COLOR (1 << 1) +#define SCALER_INT_STATUS_FRAME_END (1 << 0) + +/* SCALER_SRC_CFG */ +#define SCALER_SRC_CFG_TILE_EN (1 << 10) +#define SCALER_SRC_CFG_GET_BYTE_SWAP(r) SCALER_GET(r, 6, 5) +#define SCALER_SRC_CFG_SET_BYTE_SWAP(v) SCALER_SET(v, 6, 5) +#define SCALER_SRC_CFG_GET_COLOR_FORMAT(r) SCALER_GET(r, 4, 0) +#define SCALER_SRC_CFG_SET_COLOR_FORMAT(v) SCALER_SET(v, 4, 0) +#define SCALER_YUV420_2P_UV 0 +#define SCALER_YUV422_2P_UV 2 +#define SCALER_YUV444_2P_UV 3 +#define SCALER_RGB_565 4 +#define SCALER_ARGB1555 5 +#define SCALER_ARGB8888 6 +#define SCALER_ARGB8888_PRE 7 +#define SCALER_YUV422_1P_YVYU 9 +#define SCALER_YUV422_1P_YUYV 10 +#define SCALER_YUV422_1P_UYVY 11 +#define SCALER_ARGB4444 12 +#define SCALER_L8A8 13 +#define SCALER_RGBA8888 14 +#define SCALER_L8 15 +#define SCALER_YUV420_2P_VU 16 +#define SCALER_YUV422_2P_VU 18 +#define SCALER_YUV444_2P_VU 19 +#define SCALER_YUV420_3P 20 +#define SCALER_YUV422_3P 22 +#define SCALER_YUV444_3P 23 + +/* SCALER_SRC_SPAN */ +#define SCALER_SRC_SPAN_GET_C_SPAN(r) SCALER_GET(r, 29, 16) +#define SCALER_SRC_SPAN_SET_C_SPAN(v) SCALER_SET(v, 29, 16) +#define SCALER_SRC_SPAN_GET_Y_SPAN(r) SCALER_GET(r, 13, 0) +#define SCALER_SRC_SPAN_SET_Y_SPAN(v) SCALER_SET(v, 13, 0) + +/* SCALER_SRC_Y_POS */ +#define SCALER_SRC_Y_POS_GET_YH_POS(r) SCALER_GET(r, 31, 16) +#define SCALER_SRC_Y_POS_SET_YH_POS(v) SCALER_SET(v, 31, 16) +#define SCALER_SRC_Y_POS_GET_YV_POS(r) SCALER_GET(r, 15, 0) +#define SCALER_SRC_Y_POS_SET_YV_POS(v) SCALER_SET(v, 15, 0) + +/* SCALER_SRC_WH */ +#define SCALER_SRC_WH_GET_WIDTH(r) SCALER_GET(r, 29, 16) +#define SCALER_SRC_WH_SET_WIDTH(v) SCALER_SET(v, 29, 16) +#define SCALER_SRC_WH_GET_HEIGHT(r) SCALER_GET(r, 13, 0) +#define SCALER_SRC_WH_SET_HEIGHT(v) SCALER_SET(v, 13, 0) + +/* SCALER_SRC_C_POS */ +#define SCALER_SRC_C_POS_GET_CH_POS(r) SCALER_GET(r, 31, 16) +#define SCALER_SRC_C_POS_SET_CH_POS(v) SCALER_SET(v, 31, 16) +#define SCALER_SRC_C_POS_GET_CV_POS(r) SCALER_GET(r, 15, 0) +#define SCALER_SRC_C_POS_SET_CV_POS(v) SCALER_SET(v, 15, 0) + +/* SCALER_DST_CFG */ +#define SCALER_DST_CFG_GET_BYTE_SWAP(r) SCALER_GET(r, 6, 5) +#define SCALER_DST_CFG_SET_BYTE_SWAP(v) SCALER_SET(v, 6, 5) +#define SCALER_DST_CFG_GET_COLOR_FORMAT(r) SCALER_GET(r, 4, 0) +#define SCALER_DST_CFG_SET_COLOR_FORMAT(v) SCALER_SET(v, 4, 0) + +/* SCALER_DST_SPAN */ +#define SCALER_DST_SPAN_GET_C_SPAN(r) SCALER_GET(r, 29, 16) +#define SCALER_DST_SPAN_SET_C_SPAN(v) SCALER_SET(v, 29, 16) +#define SCALER_DST_SPAN_GET_Y_SPAN(r) SCALER_GET(r, 13, 0) +#define SCALER_DST_SPAN_SET_Y_SPAN(v) SCALER_SET(v, 13, 0) + +/* SCALER_DST_WH */ +#define SCALER_DST_WH_GET_WIDTH(r) SCALER_GET(r, 29, 16) +#define SCALER_DST_WH_SET_WIDTH(v) SCALER_SET(v, 29, 16) +#define SCALER_DST_WH_GET_HEIGHT(r) SCALER_GET(r, 13, 0) +#define SCALER_DST_WH_SET_HEIGHT(v) SCALER_SET(v, 13, 0) + +/* SCALER_DST_POS */ +#define SCALER_DST_POS_GET_H_POS(r) SCALER_GET(r, 29, 16) +#define SCALER_DST_POS_SET_H_POS(v) SCALER_SET(v, 29, 16) +#define SCALER_DST_POS_GET_V_POS(r) SCALER_GET(r, 13, 0) +#define SCALER_DST_POS_SET_V_POS(v) SCALER_SET(v, 13, 0) + +/* SCALER_H_RATIO */ +#define SCALER_H_RATIO_GET(r) SCALER_GET(r, 18, 0) +#define SCALER_H_RATIO_SET(v) SCALER_SET(v, 18, 0) + +/* SCALER_V_RATIO */ +#define SCALER_V_RATIO_GET(r) SCALER_GET(r, 18, 0) +#define SCALER_V_RATIO_SET(v) SCALER_SET(v, 18, 0) + +/* SCALER_ROT_CFG */ +#define SCALER_ROT_CFG_FLIP_X_EN (1 << 3) +#define SCALER_ROT_CFG_FLIP_Y_EN (1 << 2) +#define SCALER_ROT_CFG_GET_ROTMODE(r) SCALER_GET(r, 1, 0) +#define SCALER_ROT_CFG_SET_ROTMODE(v) SCALER_SET(v, 1, 0) +#define SCALER_ROT_MODE_90 1 +#define SCALER_ROT_MODE_180 2 +#define SCALER_ROT_MODE_270 3 + +/* SCALER_HCOEF, SCALER_VCOEF */ +#define SCALER_COEF_SHIFT(i) (16 * (1 - (i) % 2)) +#define SCALER_COEF_GET(r, i) \ + (((r) >> SCALER_COEF_SHIFT(i)) & 0x1ff) +#define SCALER_COEF_SET(v, i) \ + (((v) & 0x1ff) << SCALER_COEF_SHIFT(i)) + +/* SCALER_CSC_COEFxy */ +#define SCALER_CSC_COEF_GET(r) SCALER_GET(r, 11, 0) +#define SCALER_CSC_COEF_SET(v) SCALER_SET(v, 11, 0) + +/* SCALER_DITH_CFG */ +#define SCALER_DITH_CFG_GET_R_TYPE(r) SCALER_GET(r, 8, 6) +#define SCALER_DITH_CFG_SET_R_TYPE(v) SCALER_SET(v, 8, 6) +#define SCALER_DITH_CFG_GET_G_TYPE(r) SCALER_GET(r, 5, 3) +#define SCALER_DITH_CFG_SET_G_TYPE(v) SCALER_SET(v, 5, 3) +#define SCALER_DITH_CFG_GET_B_TYPE(r) SCALER_GET(r, 2, 0) +#define SCALER_DITH_CFG_SET_B_TYPE(v) SCALER_SET(v, 2, 0) + +/* SCALER_TIMEOUT_CTRL */ +#define SCALER_TIMEOUT_CTRL_GET_TIMER_VALUE(r) SCALER_GET(r, 31, 16) +#define SCALER_TIMEOUT_CTRL_SET_TIMER_VALUE(v) SCALER_SET(v, 31, 16) +#define SCALER_TIMEOUT_CTRL_GET_TIMER_DIV(r) SCALER_GET(r, 7, 4) +#define SCALER_TIMEOUT_CTRL_SET_TIMER_DIV(v) SCALER_SET(v, 7, 4) +#define SCALER_TIMEOUT_CTRL_TIMER_ENABLE (1 << 0) + +/* SCALER_TIMEOUT_CNT */ +#define SCALER_TIMEOUT_CTRL_GET_TIMER_COUNT(r) SCALER_GET(r, 31, 16) + +/* SCALER_SRC_BLEND_COLOR */ +#define SCALER_SRC_BLEND_COLOR_SEL_INV (1 << 31) +#define SCALER_SRC_BLEND_COLOR_GET_SEL(r) SCALER_GET(r, 30, 29) +#define SCALER_SRC_BLEND_COLOR_SET_SEL(v) SCALER_SET(v, 30, 29) +#define SCALER_SRC_BLEND_COLOR_OP_SEL_INV (1 << 28) +#define SCALER_SRC_BLEND_COLOR_GET_OP_SEL(r) SCALER_GET(r, 27, 24) +#define SCALER_SRC_BLEND_COLOR_SET_OP_SEL(v) SCALER_SET(v, 27, 24) +#define SCALER_SRC_BLEND_COLOR_GET_COLOR0(r) SCALER_GET(r, 23, 16) +#define SCALER_SRC_BLEND_COLOR_SET_COLOR0(v) SCALER_SET(v, 23, 16) +#define SCALER_SRC_BLEND_COLOR_GET_COLOR1(r) SCALER_GET(r, 15, 8) +#define SCALER_SRC_BLEND_COLOR_SET_COLOR1(v) SCALER_SET(v, 15, 8) +#define SCALER_SRC_BLEND_COLOR_GET_COLOR2(r) SCALER_GET(r, 7, 0) +#define SCALER_SRC_BLEND_COLOR_SET_COLOR2(v) SCALER_SET(v, 7, 0) + +/* SCALER_SRC_BLEND_ALPHA */ +#define SCALER_SRC_BLEND_ALPHA_SEL_INV (1 << 31) +#define SCALER_SRC_BLEND_ALPHA_GET_SEL(r) SCALER_GET(r, 30, 29) +#define SCALER_SRC_BLEND_ALPHA_SET_SEL(v) SCALER_SET(v, 30, 29) +#define SCALER_SRC_BLEND_ALPHA_OP_SEL_INV (1 << 28) +#define SCALER_SRC_BLEND_ALPHA_GET_OP_SEL(r) SCALER_GET(r, 27, 24) +#define SCALER_SRC_BLEND_ALPHA_SET_OP_SEL(v) SCALER_SET(v, 27, 24) +#define SCALER_SRC_BLEND_ALPHA_GET_ALPHA(r) SCALER_GET(r, 7, 0) +#define SCALER_SRC_BLEND_ALPHA_SET_ALPHA(v) SCALER_SET(v, 7, 0) + +/* SCALER_DST_BLEND_COLOR */ +#define SCALER_DST_BLEND_COLOR_SEL_INV (1 << 31) +#define SCALER_DST_BLEND_COLOR_GET_SEL(r) SCALER_GET(r, 30, 29) +#define SCALER_DST_BLEND_COLOR_SET_SEL(v) SCALER_SET(v, 30, 29) +#define SCALER_DST_BLEND_COLOR_OP_SEL_INV (1 << 28) +#define SCALER_DST_BLEND_COLOR_GET_OP_SEL(r) SCALER_GET(r, 27, 24) +#define SCALER_DST_BLEND_COLOR_SET_OP_SEL(v) SCALER_SET(v, 27, 24) +#define SCALER_DST_BLEND_COLOR_GET_COLOR0(r) SCALER_GET(r, 23, 16) +#define SCALER_DST_BLEND_COLOR_SET_COLOR0(v) SCALER_SET(v, 23, 16) +#define SCALER_DST_BLEND_COLOR_GET_COLOR1(r) SCALER_GET(r, 15, 8) +#define SCALER_DST_BLEND_COLOR_SET_COLOR1(v) SCALER_SET(v, 15, 8) +#define SCALER_DST_BLEND_COLOR_GET_COLOR2(r) SCALER_GET(r, 7, 0) +#define SCALER_DST_BLEND_COLOR_SET_COLOR2(v) SCALER_SET(v, 7, 0) + +/* SCALER_DST_BLEND_ALPHA */ +#define SCALER_DST_BLEND_ALPHA_SEL_INV (1 << 31) +#define SCALER_DST_BLEND_ALPHA_GET_SEL(r) SCALER_GET(r, 30, 29) +#define SCALER_DST_BLEND_ALPHA_SET_SEL(v) SCALER_SET(v, 30, 29) +#define SCALER_DST_BLEND_ALPHA_OP_SEL_INV (1 << 28) +#define SCALER_DST_BLEND_ALPHA_GET_OP_SEL(r) SCALER_GET(r, 27, 24) +#define SCALER_DST_BLEND_ALPHA_SET_OP_SEL(v) SCALER_SET(v, 27, 24) +#define SCALER_DST_BLEND_ALPHA_GET_ALPHA(r) SCALER_GET(r, 7, 0) +#define SCALER_DST_BLEND_ALPHA_SET_ALPHA(v) SCALER_SET(v, 7, 0) + +/* SCALER_FILL_COLOR */ +#define SCALER_FILL_COLOR_GET_ALPHA(r) SCALER_GET(r, 31, 24) +#define SCALER_FILL_COLOR_SET_ALPHA(v) SCALER_SET(v, 31, 24) +#define SCALER_FILL_COLOR_GET_FILL_COLOR0(r) SCALER_GET(r, 23, 16) +#define SCALER_FILL_COLOR_SET_FILL_COLOR0(v) SCALER_SET(v, 23, 16) +#define SCALER_FILL_COLOR_GET_FILL_COLOR1(r) SCALER_GET(r, 15, 8) +#define SCALER_FILL_COLOR_SET_FILL_COLOR1(v) SCALER_SET(v, 15, 8) +#define SCALER_FILL_COLOR_GET_FILL_COLOR2(r) SCALER_GET(r, 7, 0) +#define SCALER_FILL_COLOR_SET_FILL_COLOR2(v) SCALER_SET(v, 7, 0) + +/* SCALER_ADDR_Q_CONFIG */ +#define SCALER_ADDR_Q_CONFIG_RST (1 << 0) + +/* SCALER_SRC_ADDR_Q_STATUS */ +#define SCALER_SRC_ADDR_Q_STATUS_Y_FULL (1 << 23) +#define SCALER_SRC_ADDR_Q_STATUS_Y_EMPTY (1 << 22) +#define SCALER_SRC_ADDR_Q_STATUS_GET_Y_WR_IDX(r) SCALER_GET(r, 21, 16) +#define SCALER_SRC_ADDR_Q_STATUS_CB_FULL (1 << 15) +#define SCALER_SRC_ADDR_Q_STATUS_CB_EMPTY (1 << 14) +#define SCALER_SRC_ADDR_Q_STATUS_GET_CB_WR_IDX(r) SCALER_GET(r, 13, 8) +#define SCALER_SRC_ADDR_Q_STATUS_CR_FULL (1 << 7) +#define SCALER_SRC_ADDR_Q_STATUS_CR_EMPTY (1 << 6) +#define SCALER_SRC_ADDR_Q_STATUS_GET_CR_WR_IDX(r) SCALER_GET(r, 5, 0) + +/* SCALER_DST_ADDR_Q_STATUS */ +#define SCALER_DST_ADDR_Q_STATUS_Y_FULL (1 << 23) +#define SCALER_DST_ADDR_Q_STATUS_Y_EMPTY (1 << 22) +#define SCALER_DST_ADDR_Q_STATUS_GET_Y_WR_IDX(r) SCALER_GET(r, 21, 16) +#define SCALER_DST_ADDR_Q_STATUS_CB_FULL (1 << 15) +#define SCALER_DST_ADDR_Q_STATUS_CB_EMPTY (1 << 14) +#define SCALER_DST_ADDR_Q_STATUS_GET_CB_WR_IDX(r) SCALER_GET(r, 13, 8) +#define SCALER_DST_ADDR_Q_STATUS_CR_FULL (1 << 7) +#define SCALER_DST_ADDR_Q_STATUS_CR_EMPTY (1 << 6) +#define SCALER_DST_ADDR_Q_STATUS_GET_CR_WR_IDX(r) SCALER_GET(r, 5, 0) + +/* SCALER_CRC_COLOR00_10 */ +#define SCALER_CRC_COLOR00_10_GET_00(r) SCALER_GET(r, 31, 16) +#define SCALER_CRC_COLOR00_10_GET_10(r) SCALER_GET(r, 15, 0) + +/* SCALER_CRC_COLOR20_30 */ +#define SCALER_CRC_COLOR20_30_GET_20(r) SCALER_GET(r, 31, 16) +#define SCALER_CRC_COLOR20_30_GET_30(r) SCALER_GET(r, 15, 0) + +/* SCALER_CRC_COLOR01_11 */ +#define SCALER_CRC_COLOR01_11_GET_01(r) SCALER_GET(r, 31, 16) +#define SCALER_CRC_COLOR01_11_GET_11(r) SCALER_GET(r, 15, 0) + +/* SCALER_CRC_COLOR21_31 */ +#define SCALER_CRC_COLOR21_31_GET_21(r) SCALER_GET(r, 31, 16) +#define SCALER_CRC_COLOR21_31_GET_31(r) SCALER_GET(r, 15, 0) + +#endif /* EXYNOS_REGS_SCALER_H */ -- 2.14.2 ^ permalink raw reply related [flat|nested] 20+ messages in thread
[parent not found: <CGME20171023075452eucas1p2881393094c507e4409005a498658f28d@eucas1p2.samsung.com>]
* [PATCH v4 8/9] ARM: dts: exynos: Add mem-2-mem Scaler devices [not found] ` <CGME20171023075452eucas1p2881393094c507e4409005a498658f28d@eucas1p2.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski 0 siblings, 0 replies; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Marek Szyprowski, Inki Dae, Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon From: Andrzej Pietrasiewicz <andrzej.p@samsung.com> There are 3 scaler devices in Exynos5420 SoCs, all are a part of MSCL power domain. MSCL power domain and SYSMMU controllers (two per each scaler device) have been already added to exynos5420.dtsi earlier, so bind them to newly added devices. Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> --- arch/arm/boot/dts/exynos5420.dtsi | 35 +++++++++++++++++++++++++++++++++++ 1 file changed, 35 insertions(+) diff --git a/arch/arm/boot/dts/exynos5420.dtsi b/arch/arm/boot/dts/exynos5420.dtsi index 88e5d6d3f901..7894045bd91b 100644 --- a/arch/arm/boot/dts/exynos5420.dtsi +++ b/arch/arm/boot/dts/exynos5420.dtsi @@ -678,6 +678,35 @@ iommus = <&sysmmu_gscl1>; }; + scaler_0: scaler@12800000 { + compatible = "samsung,exynos5420-scaler"; + reg = <0x12800000 0x1294>; + interrupts = <0 220 IRQ_TYPE_LEVEL_HIGH>; + clocks = <&clock CLK_MSCL0>; + clock-names = "mscl"; + power-domains = <&msc_pd>; + iommus = <&sysmmu_scaler0r>, <&sysmmu_scaler0w>; + }; + + scaler_1: scaler@12810000 { + compatible = "samsung,exynos5420-scaler"; + reg = <0x12810000 0x1294>; + interrupts = <0 221 IRQ_TYPE_LEVEL_HIGH>; + clocks = <&clock CLK_MSCL1>; + clock-names = "mscl"; + power-domains = <&msc_pd>; + iommus = <&sysmmu_scaler1r>, <&sysmmu_scaler1w>; + }; + scaler_2: scaler@12820000 { + compatible = "samsung,exynos5420-scaler"; + reg = <0x12820000 0x1294>; + interrupts = <0 222 IRQ_TYPE_LEVEL_HIGH>; + clocks = <&clock CLK_MSCL2>; + clock-names = "mscl"; + power-domains = <&msc_pd>; + iommus = <&sysmmu_scaler2r>, <&sysmmu_scaler2w>; + }; + jpeg_0: jpeg@11F50000 { compatible = "samsung,exynos5420-jpeg"; reg = <0x11F50000 0x1000>; @@ -812,6 +841,7 @@ interrupts = <22 4>; clock-names = "sysmmu", "master"; clocks = <&clock CLK_SMMU_MSCL0>, <&clock CLK_MSCL0>; + power-domains = <&msc_pd>; #iommu-cells = <0>; }; @@ -821,6 +851,7 @@ interrupts = <GIC_SPI 186 IRQ_TYPE_LEVEL_HIGH>; clock-names = "sysmmu", "master"; clocks = <&clock CLK_SMMU_MSCL1>, <&clock CLK_MSCL1>; + power-domains = <&msc_pd>; #iommu-cells = <0>; }; @@ -830,6 +861,7 @@ interrupts = <GIC_SPI 188 IRQ_TYPE_LEVEL_HIGH>; clock-names = "sysmmu", "master"; clocks = <&clock CLK_SMMU_MSCL2>, <&clock CLK_MSCL2>; + power-domains = <&msc_pd>; #iommu-cells = <0>; }; @@ -840,6 +872,7 @@ interrupts = <27 2>; clock-names = "sysmmu", "master"; clocks = <&clock CLK_SMMU_MSCL0>, <&clock CLK_MSCL0>; + power-domains = <&msc_pd>; #iommu-cells = <0>; }; @@ -850,6 +883,7 @@ interrupts = <22 6>; clock-names = "sysmmu", "master"; clocks = <&clock CLK_SMMU_MSCL1>, <&clock CLK_MSCL1>; + power-domains = <&msc_pd>; #iommu-cells = <0>; }; @@ -860,6 +894,7 @@ interrupts = <19 6>; clock-names = "sysmmu", "master"; clocks = <&clock CLK_SMMU_MSCL2>, <&clock CLK_MSCL2>; + power-domains = <&msc_pd>; #iommu-cells = <0>; }; -- 2.14.2 ^ permalink raw reply related [flat|nested] 20+ messages in thread
[parent not found: <CGME20171023075453eucas1p18c5a22b967399885c14bff39b1e218f2@eucas1p1.samsung.com>]
* [PATCH v4 9/9] ARM64: dts: exynos: Add mem-2-mem Scaler devices [not found] ` <CGME20171023075453eucas1p18c5a22b967399885c14bff39b1e218f2@eucas1p1.samsung.com> @ 2017-10-23 7:54 ` Marek Szyprowski 0 siblings, 0 replies; 20+ messages in thread From: Marek Szyprowski @ 2017-10-23 7:54 UTC (permalink / raw) To: dri-devel, linux-samsung-soc Cc: Marek Szyprowski, Inki Dae, Seung-Woo Kim, Andrzej Hajda, Bartlomiej Zolnierkiewicz, Tobias Jakobi, Krzysztof Kozlowski, Sylwester Nawrocki, Andrzej Pietrasiewicz, Hoegeun Kwon From: Andrzej Pietrasiewicz <andrzej.p@samsung.com> There are two Scaler devices in Exynos5433 SoCs. Add nodes for them and their SYSMMU controllers. Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@samsung.com> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> --- arch/arm64/boot/dts/exynos/exynos5433.dtsi | 42 ++++++++++++++++++++++++++++++ 1 file changed, 42 insertions(+) diff --git a/arch/arm64/boot/dts/exynos/exynos5433.dtsi b/arch/arm64/boot/dts/exynos/exynos5433.dtsi index 7fe994b750da..97866114767f 100644 --- a/arch/arm64/boot/dts/exynos/exynos5433.dtsi +++ b/arch/arm64/boot/dts/exynos/exynos5433.dtsi @@ -920,6 +920,28 @@ iommus = <&sysmmu_gscl2>; }; + scaler_0: scaler@15000000 { + compatible = "samsung,exynos5433-scaler"; + reg = <0x15000000 0x1294>; + interrupts = <0 402 IRQ_TYPE_LEVEL_HIGH>; + clock-names = "pclk", "aclk", "aclk_xiu"; + clocks = <&cmu_mscl CLK_PCLK_M2MSCALER0>, + <&cmu_mscl CLK_ACLK_M2MSCALER0>, + <&cmu_mscl CLK_ACLK_XIU_MSCLX>; + iommus = <&sysmmu_scaler_0>; + }; + + scaler_1: scaler@15010000 { + compatible = "samsung,exynos5433-scaler"; + reg = <0x15010000 0x1294>; + interrupts = <0 403 IRQ_TYPE_LEVEL_HIGH>; + clock-names = "pclk", "aclk", "aclk_xiu"; + clocks = <&cmu_mscl CLK_PCLK_M2MSCALER1>, + <&cmu_mscl CLK_ACLK_M2MSCALER1>, + <&cmu_mscl CLK_ACLK_XIU_MSCLX>; + iommus = <&sysmmu_scaler_1>; + }; + jpeg: codec@15020000 { compatible = "samsung,exynos5433-jpeg"; reg = <0x15020000 0x10000>; @@ -1014,6 +1036,26 @@ #iommu-cells = <0>; }; + sysmmu_scaler_0: sysmmu@0x15040000 { + compatible = "samsung,exynos-sysmmu"; + reg = <0x15040000 0x1000>; + interrupts = <GIC_SPI 404 IRQ_TYPE_LEVEL_HIGH>; + clock-names = "pclk", "aclk"; + clocks = <&cmu_mscl CLK_PCLK_SMMU_M2MSCALER0>, + <&cmu_mscl CLK_ACLK_SMMU_M2MSCALER0>; + #iommu-cells = <0>; + }; + + sysmmu_scaler_1: sysmmu@0x15050000 { + compatible = "samsung,exynos-sysmmu"; + reg = <0x15050000 0x1000>; + interrupts = <GIC_SPI 406 IRQ_TYPE_LEVEL_HIGH>; + clock-names = "pclk", "aclk"; + clocks = <&cmu_mscl CLK_PCLK_SMMU_M2MSCALER1>, + <&cmu_mscl CLK_ACLK_SMMU_M2MSCALER1>; + #iommu-cells = <0>; + }; + sysmmu_jpeg: sysmmu@15060000 { compatible = "samsung,exynos-sysmmu"; reg = <0x15060000 0x1000>; -- 2.14.2 ^ permalink raw reply related [flat|nested] 20+ messages in thread
end of thread, other threads:[~2017-11-07 9:07 UTC | newest]
Thread overview: 20+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <CGME20171023075448eucas1p25e25bb36d36139e3917211ab8ac999ec@eucas1p2.samsung.com>
2017-10-23 7:54 ` [PATCH v4 0/9] Exynos DRM: rewrite IPP subsystem and userspace API Marek Szyprowski
[not found] ` <CGME20171023075448eucas1p10255809be25caa7e44a3889f26561bc8@eucas1p1.samsung.com>
2017-10-23 7:54 ` [PATCH v4 1/9] drm/exynos: ipp: Remove Exynos DRM IPP subsystem Marek Szyprowski
[not found] ` <CGME20171023075449eucas1p18211ed64e964860d201aaeb53a87a8be@eucas1p1.samsung.com>
2017-10-23 7:54 ` [PATCH v4 2/9] drm/exynos: ipp: Add IPP v2 framework Marek Szyprowski
2017-11-01 1:13 ` Inki Dae
2017-11-03 11:20 ` Marek Szyprowski
2017-11-01 3:47 ` Dave Airlie
2017-11-01 14:18 ` Tobias Jakobi
[not found] ` <CGME20171023075449eucas1p14adcde4cea94ca82a13a822ee59d9446@eucas1p1.samsung.com>
2017-10-23 7:54 ` [PATCH v4 3/9] drm/exynos: rotator: Convert driver to IPP v2 core API Marek Szyprowski
2017-11-01 5:40 ` Inki Dae
[not found] ` <CGME20171023075450eucas1p155afda476e368ae57ba7fe1a347a5f4c@eucas1p1.samsung.com>
2017-10-23 7:54 ` [PATCH v4 4/9] drm/exynos: gsc: " Marek Szyprowski
2017-11-01 6:02 ` Inki Dae
[not found] ` <CGME20171023075450eucas1p24923c2a0793ac24364b97770dadd56e0@eucas1p2.samsung.com>
2017-10-23 7:54 ` [PATCH v4 5/9] drm/exynos: Add generic support for devices shared with V4L2 subsystem Marek Szyprowski
2017-11-01 6:26 ` Inki Dae
2017-11-03 12:28 ` Marek Szyprowski
2017-11-06 1:20 ` Inki Dae
2017-11-07 9:07 ` Marek Szyprowski
[not found] ` <CGME20171023075451eucas1p1a833d9c805f6b7dc1b984e59afce23f3@eucas1p1.samsung.com>
2017-10-23 7:54 ` [PATCH v4 6/9] drm/exynos: fimc: Convert driver to IPP v2 core API Marek Szyprowski
[not found] ` <CGME20171023075452eucas1p1dada05e443a800977d9c888744877f38@eucas1p1.samsung.com>
2017-10-23 7:54 ` [PATCH v4 7/9] drm/exynos: Add driver for Exynos Scaler module Marek Szyprowski
[not found] ` <CGME20171023075452eucas1p2881393094c507e4409005a498658f28d@eucas1p2.samsung.com>
2017-10-23 7:54 ` [PATCH v4 8/9] ARM: dts: exynos: Add mem-2-mem Scaler devices Marek Szyprowski
[not found] ` <CGME20171023075453eucas1p18c5a22b967399885c14bff39b1e218f2@eucas1p1.samsung.com>
2017-10-23 7:54 ` [PATCH v4 9/9] ARM64: " Marek Szyprowski
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.