* [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver
@ 2022-02-07 14:19 Dmitry Osipenko
2022-02-07 14:19 ` [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags Dmitry Osipenko
` (4 more replies)
0 siblings, 5 replies; 10+ messages in thread
From: Dmitry Osipenko @ 2022-02-07 14:19 UTC (permalink / raw)
To: Thierry Reding, Jonathan Hunter, Mauro Carvalho Chehab,
Hans Verkuil, Nicolas Dufresne, Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
Support V4L stateless video decoder API by NVIDIA Tegra decoder driver.
Tested using GStreamer [1] and libvdpau-tegra [2][8].
[1] https://github.com/grate-driver/gstreamer/commit/b8509bdbb69b534e61419ea1798f32f9ad2f3597
[2] https://github.com/grate-driver/libvdpau-tegra/commit/f822e95911e5e0c39f8ba19f843ddc1e0138d5ce
[8] https://github.com/grate-driver/libvdpau-tegra/commit/80db4d02369f2a984ce3173d6bc305f32e9fdb97
Changelog:
v3: - Added new decode_params flags [7] instead of V4L2_BUF_FLAG_*FRAME flags,
as was suggested by Nicolas Dufresne.
[7] https://github.com/grate-driver/gstreamer/commit/c5cd847f9c26b7669720ae58f9058de2515f51a2
- Added new patch that removes legacy UAPI.
v2: - Made V4L2_BUF_FLAG_*FRAME flags mandatory [3] and dropped reading
of raw bitstream from the driver code, as was suggested by
Nicolas Dufresne.
[3] https://github.com/grate-driver/gstreamer/commit/aee292f0f2e84b7654a314dd7e63f916888ffaa5
- Ran v4l2-compliance [4] and fluster [5][6] tests, like was suggested by
Nicolas Dufresne. Fixed minor v4l2-compliance errors that were related
to a partial initialization of the coded format and were harmless in
practice, but made compliance checker unhappy.
[4] https://gist.github.com/digetx/5d6bcdab633488f1dcc7c141ab90d30e
[5] https://gist.github.com/digetx/b06c5d779e9d25afa41d9f46946fe399
[6] https://gist.github.com/digetx/ac4198bc340e5065aa8ec3288bb21356
Dmitry Osipenko (4):
media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags
media: staging: tegra-vde: Factor out H.264 code
media: staging: tegra-vde: Support V4L stateless video decoder API
media: staging: tegra-vde: Remove legacy UAPI support
.../media/v4l/ext-ctrls-codec-stateless.rst | 6 +
drivers/staging/media/tegra-vde/Kconfig | 7 +
drivers/staging/media/tegra-vde/Makefile | 2 +-
.../staging/media/tegra-vde/dmabuf-cache.c | 2 +-
drivers/staging/media/tegra-vde/h264.c | 946 +++++++++++++++
drivers/staging/media/tegra-vde/iommu.c | 2 +-
drivers/staging/media/tegra-vde/uapi.h | 73 --
drivers/staging/media/tegra-vde/v4l2.c | 1018 +++++++++++++++++
drivers/staging/media/tegra-vde/vde.c | 945 ++-------------
drivers/staging/media/tegra-vde/vde.h | 121 +-
include/uapi/linux/v4l2-controls.h | 2 +
11 files changed, 2170 insertions(+), 954 deletions(-)
create mode 100644 drivers/staging/media/tegra-vde/h264.c
delete mode 100644 drivers/staging/media/tegra-vde/uapi.h
create mode 100644 drivers/staging/media/tegra-vde/v4l2.c
--
2.34.1
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags
2022-02-07 14:19 [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Dmitry Osipenko
@ 2022-02-07 14:19 ` Dmitry Osipenko
2022-02-11 13:29 ` Nicolas Dufresne
2022-02-18 14:53 ` Nicolas Dufresne
2022-02-07 14:19 ` [PATCH v3 2/4] media: staging: tegra-vde: Factor out H.264 code Dmitry Osipenko
` (3 subsequent siblings)
4 siblings, 2 replies; 10+ messages in thread
From: Dmitry Osipenko @ 2022-02-07 14:19 UTC (permalink / raw)
To: Thierry Reding, Jonathan Hunter, Mauro Carvalho Chehab,
Hans Verkuil, Nicolas Dufresne, Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags that are needed by
NVIDIA Tegra video decoder. Userspace will have to set these flags in
accordance to the type of a decoded frame.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
.../userspace-api/media/v4l/ext-ctrls-codec-stateless.rst | 6 ++++++
include/uapi/linux/v4l2-controls.h | 2 ++
2 files changed, 8 insertions(+)
diff --git a/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst b/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst
index cc080c4257d0..f87584ad90ba 100644
--- a/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst
+++ b/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst
@@ -616,6 +616,12 @@ Stateless Codec Control ID
* - ``V4L2_H264_DECODE_PARAM_FLAG_BOTTOM_FIELD``
- 0x00000004
-
+ * - ``V4L2_H264_DECODE_PARAM_FLAG_PFRAME``
+ - 0x00000008
+ -
+ * - ``V4L2_H264_DECODE_PARAM_FLAG_BFRAME``
+ - 0x00000010
+ -
.. raw:: latex
diff --git a/include/uapi/linux/v4l2-controls.h b/include/uapi/linux/v4l2-controls.h
index c8e0f84d204d..e3d48d571062 100644
--- a/include/uapi/linux/v4l2-controls.h
+++ b/include/uapi/linux/v4l2-controls.h
@@ -1563,6 +1563,8 @@ struct v4l2_h264_dpb_entry {
#define V4L2_H264_DECODE_PARAM_FLAG_IDR_PIC 0x01
#define V4L2_H264_DECODE_PARAM_FLAG_FIELD_PIC 0x02
#define V4L2_H264_DECODE_PARAM_FLAG_BOTTOM_FIELD 0x04
+#define V4L2_H264_DECODE_PARAM_FLAG_PFRAME 0x08
+#define V4L2_H264_DECODE_PARAM_FLAG_BFRAME 0x10
#define V4L2_CID_STATELESS_H264_DECODE_PARAMS (V4L2_CID_CODEC_STATELESS_BASE + 7)
/**
--
2.34.1
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH v3 2/4] media: staging: tegra-vde: Factor out H.264 code
2022-02-07 14:19 [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Dmitry Osipenko
2022-02-07 14:19 ` [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags Dmitry Osipenko
@ 2022-02-07 14:19 ` Dmitry Osipenko
2022-02-07 14:19 ` [PATCH v3 3/4] media: staging: tegra-vde: Support V4L stateless video decoder API Dmitry Osipenko
` (2 subsequent siblings)
4 siblings, 0 replies; 10+ messages in thread
From: Dmitry Osipenko @ 2022-02-07 14:19 UTC (permalink / raw)
To: Thierry Reding, Jonathan Hunter, Mauro Carvalho Chehab,
Hans Verkuil, Nicolas Dufresne, Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
Factor out H.264 hardware programming code into separate source file in a
preparation to support V4L API by the Tegra video decoder driver.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
drivers/staging/media/tegra-vde/Makefile | 2 +-
drivers/staging/media/tegra-vde/h264.c | 647 +++++++++++++++++++++++
drivers/staging/media/tegra-vde/vde.c | 626 +---------------------
drivers/staging/media/tegra-vde/vde.h | 42 ++
4 files changed, 706 insertions(+), 611 deletions(-)
create mode 100644 drivers/staging/media/tegra-vde/h264.c
diff --git a/drivers/staging/media/tegra-vde/Makefile b/drivers/staging/media/tegra-vde/Makefile
index 2827f7601de8..43525b08b3b0 100644
--- a/drivers/staging/media/tegra-vde/Makefile
+++ b/drivers/staging/media/tegra-vde/Makefile
@@ -1,3 +1,3 @@
# SPDX-License-Identifier: GPL-2.0
-tegra-vde-y := vde.o iommu.o dmabuf-cache.o
+tegra-vde-y := vde.o iommu.o dmabuf-cache.o h264.o
obj-$(CONFIG_TEGRA_VDE) += tegra-vde.o
diff --git a/drivers/staging/media/tegra-vde/h264.c b/drivers/staging/media/tegra-vde/h264.c
new file mode 100644
index 000000000000..03faa705bf71
--- /dev/null
+++ b/drivers/staging/media/tegra-vde/h264.c
@@ -0,0 +1,647 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NVIDIA Tegra Video decoder driver
+ *
+ * Copyright (C) 2016-2022 Dmitry Osipenko <digetx@gmail.com>
+ *
+ */
+
+#include <linux/iopoll.h>
+#include <linux/pm_runtime.h>
+#include <linux/reset.h>
+#include <linux/slab.h>
+
+#include "trace.h"
+#include "uapi.h"
+#include "vde.h"
+
+static int tegra_vde_wait_mbe(struct tegra_vde *vde)
+{
+ u32 tmp;
+
+ return readl_relaxed_poll_timeout(vde->mbe + 0x8C, tmp,
+ tmp >= 0x10, 1, 100);
+}
+
+static int tegra_vde_setup_mbe_frame_idx(struct tegra_vde *vde,
+ unsigned int refs_nb,
+ bool setup_refs)
+{
+ u32 value, frame_idx_enb_mask = 0;
+ unsigned int frame_idx;
+ unsigned int idx;
+ int err;
+
+ tegra_vde_writel(vde, 0xD0000000 | (0 << 23), vde->mbe, 0x80);
+ tegra_vde_writel(vde, 0xD0200000 | (0 << 23), vde->mbe, 0x80);
+
+ err = tegra_vde_wait_mbe(vde);
+ if (err)
+ return err;
+
+ if (!setup_refs)
+ return 0;
+
+ for (idx = 0, frame_idx = 1; idx < refs_nb; idx++, frame_idx++) {
+ tegra_vde_writel(vde, 0xD0000000 | (frame_idx << 23),
+ vde->mbe, 0x80);
+ tegra_vde_writel(vde, 0xD0200000 | (frame_idx << 23),
+ vde->mbe, 0x80);
+
+ frame_idx_enb_mask |= frame_idx << (6 * (idx % 4));
+
+ if (idx % 4 == 3 || idx == refs_nb - 1) {
+ value = 0xC0000000;
+ value |= (idx >> 2) << 24;
+ value |= frame_idx_enb_mask;
+
+ tegra_vde_writel(vde, value, vde->mbe, 0x80);
+
+ err = tegra_vde_wait_mbe(vde);
+ if (err)
+ return err;
+
+ frame_idx_enb_mask = 0;
+ }
+ }
+
+ return 0;
+}
+
+static void tegra_vde_mbe_set_0xa_reg(struct tegra_vde *vde, int reg, u32 val)
+{
+ tegra_vde_writel(vde, 0xA0000000 | (reg << 24) | (val & 0xFFFF),
+ vde->mbe, 0x80);
+ tegra_vde_writel(vde, 0xA0000000 | ((reg + 1) << 24) | (val >> 16),
+ vde->mbe, 0x80);
+}
+
+static int tegra_vde_wait_bsev(struct tegra_vde *vde, bool wait_dma)
+{
+ struct device *dev = vde->dev;
+ u32 value;
+ int err;
+
+ err = readl_relaxed_poll_timeout(vde->bsev + INTR_STATUS, value,
+ !(value & BIT(2)), 1, 100);
+ if (err) {
+ dev_err(dev, "BSEV unknown bit timeout\n");
+ return err;
+ }
+
+ err = readl_relaxed_poll_timeout(vde->bsev + INTR_STATUS, value,
+ (value & BSE_ICMDQUE_EMPTY), 1, 100);
+ if (err) {
+ dev_err(dev, "BSEV ICMDQUE flush timeout\n");
+ return err;
+ }
+
+ if (!wait_dma)
+ return 0;
+
+ err = readl_relaxed_poll_timeout(vde->bsev + INTR_STATUS, value,
+ !(value & BSE_DMA_BUSY), 1, 100);
+ if (err) {
+ dev_err(dev, "BSEV DMA timeout\n");
+ return err;
+ }
+
+ return 0;
+}
+
+static int tegra_vde_push_to_bsev_icmdqueue(struct tegra_vde *vde,
+ u32 value, bool wait_dma)
+{
+ tegra_vde_writel(vde, value, vde->bsev, ICMDQUE_WR);
+
+ return tegra_vde_wait_bsev(vde, wait_dma);
+}
+
+static void tegra_vde_setup_frameid(struct tegra_vde *vde,
+ struct tegra_video_frame *frame,
+ unsigned int frameid,
+ u32 mbs_width, u32 mbs_height)
+{
+ u32 y_addr = frame ? frame->y_addr : 0x6CDEAD00;
+ u32 cb_addr = frame ? frame->cb_addr : 0x6CDEAD00;
+ u32 cr_addr = frame ? frame->cr_addr : 0x6CDEAD00;
+ u32 value1 = frame ? ((mbs_width << 16) | mbs_height) : 0;
+ u32 value2 = frame ? ((((mbs_width + 1) >> 1) << 6) | 1) : 0;
+
+ tegra_vde_writel(vde, y_addr >> 8, vde->frameid, 0x000 + frameid * 4);
+ tegra_vde_writel(vde, cb_addr >> 8, vde->frameid, 0x100 + frameid * 4);
+ tegra_vde_writel(vde, cr_addr >> 8, vde->frameid, 0x180 + frameid * 4);
+ tegra_vde_writel(vde, value1, vde->frameid, 0x080 + frameid * 4);
+ tegra_vde_writel(vde, value2, vde->frameid, 0x280 + frameid * 4);
+}
+
+static void tegra_setup_frameidx(struct tegra_vde *vde,
+ struct tegra_video_frame *frames,
+ unsigned int frames_nb,
+ u32 mbs_width, u32 mbs_height)
+{
+ unsigned int idx;
+
+ for (idx = 0; idx < frames_nb; idx++)
+ tegra_vde_setup_frameid(vde, &frames[idx], idx,
+ mbs_width, mbs_height);
+
+ for (; idx < 17; idx++)
+ tegra_vde_setup_frameid(vde, NULL, idx, 0, 0);
+}
+
+static void tegra_vde_setup_iram_entry(struct tegra_vde *vde,
+ unsigned int table,
+ unsigned int row,
+ u32 value1, u32 value2)
+{
+ u32 *iram_tables = vde->iram;
+
+ trace_vde_setup_iram_entry(table, row, value1, value2);
+
+ iram_tables[0x20 * table + row * 2 + 0] = value1;
+ iram_tables[0x20 * table + row * 2 + 1] = value2;
+}
+
+static void tegra_vde_setup_iram_tables(struct tegra_vde *vde,
+ struct tegra_video_frame *dpb_frames,
+ unsigned int ref_frames_nb,
+ unsigned int with_earlier_poc_nb)
+{
+ struct tegra_video_frame *frame;
+ int with_later_poc_nb;
+ u32 value, aux_addr;
+ unsigned int i, k;
+
+ trace_vde_ref_l0(dpb_frames[0].frame_num);
+
+ for (i = 0; i < 16; i++) {
+ if (i < ref_frames_nb) {
+ frame = &dpb_frames[i + 1];
+
+ aux_addr = frame->aux_addr;
+
+ value = (i + 1) << 26;
+ value |= !(frame->flags & FLAG_B_FRAME) << 25;
+ value |= 1 << 24;
+ value |= frame->frame_num;
+ } else {
+ aux_addr = 0x6ADEAD00;
+ value = 0x3f;
+ }
+
+ tegra_vde_setup_iram_entry(vde, 0, i, value, aux_addr);
+ tegra_vde_setup_iram_entry(vde, 1, i, value, aux_addr);
+ tegra_vde_setup_iram_entry(vde, 2, i, value, aux_addr);
+ tegra_vde_setup_iram_entry(vde, 3, i, value, aux_addr);
+ }
+
+ if (!(dpb_frames[0].flags & FLAG_B_FRAME))
+ return;
+
+ if (with_earlier_poc_nb >= ref_frames_nb)
+ return;
+
+ with_later_poc_nb = ref_frames_nb - with_earlier_poc_nb;
+
+ trace_vde_ref_l1(with_later_poc_nb, with_earlier_poc_nb);
+
+ for (i = 0, k = with_earlier_poc_nb; i < with_later_poc_nb; i++, k++) {
+ frame = &dpb_frames[k + 1];
+
+ aux_addr = frame->aux_addr;
+
+ value = (k + 1) << 26;
+ value |= !(frame->flags & FLAG_B_FRAME) << 25;
+ value |= 1 << 24;
+ value |= frame->frame_num;
+
+ tegra_vde_setup_iram_entry(vde, 2, i, value, aux_addr);
+ }
+
+ for (k = 0; i < ref_frames_nb; i++, k++) {
+ frame = &dpb_frames[k + 1];
+
+ aux_addr = frame->aux_addr;
+
+ value = (k + 1) << 26;
+ value |= !(frame->flags & FLAG_B_FRAME) << 25;
+ value |= 1 << 24;
+ value |= frame->frame_num;
+
+ tegra_vde_setup_iram_entry(vde, 2, i, value, aux_addr);
+ }
+}
+
+static int tegra_vde_setup_hw_context(struct tegra_vde *vde,
+ struct tegra_vde_h264_decoder_ctx *ctx,
+ struct tegra_video_frame *dpb_frames,
+ dma_addr_t bitstream_data_addr,
+ size_t bitstream_data_size,
+ unsigned int macroblocks_nb)
+{
+ struct device *dev = vde->dev;
+ u32 value;
+ int err;
+
+ tegra_vde_set_bits(vde, 0x000A, vde->sxe, 0xF0);
+ tegra_vde_set_bits(vde, 0x000B, vde->bsev, CMDQUE_CONTROL);
+ tegra_vde_set_bits(vde, 0x8002, vde->mbe, 0x50);
+ tegra_vde_set_bits(vde, 0x000A, vde->mbe, 0xA0);
+ tegra_vde_set_bits(vde, 0x000A, vde->ppe, 0x14);
+ tegra_vde_set_bits(vde, 0x000A, vde->ppe, 0x28);
+ tegra_vde_set_bits(vde, 0x0A00, vde->mce, 0x08);
+ tegra_vde_set_bits(vde, 0x000A, vde->tfe, 0x00);
+ tegra_vde_set_bits(vde, 0x0005, vde->vdma, 0x04);
+
+ tegra_vde_writel(vde, 0x00000000, vde->vdma, 0x1C);
+ tegra_vde_writel(vde, 0x00000000, vde->vdma, 0x00);
+ tegra_vde_writel(vde, 0x00000007, vde->vdma, 0x04);
+ tegra_vde_writel(vde, 0x00000007, vde->frameid, 0x200);
+ tegra_vde_writel(vde, 0x00000005, vde->tfe, 0x04);
+ tegra_vde_writel(vde, 0x00000000, vde->mbe, 0x84);
+ tegra_vde_writel(vde, 0x00000010, vde->sxe, 0x08);
+ tegra_vde_writel(vde, 0x00000150, vde->sxe, 0x54);
+ tegra_vde_writel(vde, 0x0000054C, vde->sxe, 0x58);
+ tegra_vde_writel(vde, 0x00000E34, vde->sxe, 0x5C);
+ tegra_vde_writel(vde, 0x063C063C, vde->mce, 0x10);
+ tegra_vde_writel(vde, 0x0003FC00, vde->bsev, INTR_STATUS);
+ tegra_vde_writel(vde, 0x0000150D, vde->bsev, BSE_CONFIG);
+ tegra_vde_writel(vde, 0x00000100, vde->bsev, BSE_INT_ENB);
+ tegra_vde_writel(vde, 0x00000000, vde->bsev, 0x98);
+ tegra_vde_writel(vde, 0x00000060, vde->bsev, 0x9C);
+
+ memset(vde->iram + 128, 0, macroblocks_nb / 2);
+
+ tegra_setup_frameidx(vde, dpb_frames, ctx->dpb_frames_nb,
+ ctx->pic_width_in_mbs, ctx->pic_height_in_mbs);
+
+ tegra_vde_setup_iram_tables(vde, dpb_frames,
+ ctx->dpb_frames_nb - 1,
+ ctx->dpb_ref_frames_with_earlier_poc_nb);
+
+ /*
+ * The IRAM mapping is write-combine, ensure that CPU buffers have
+ * been flushed at this point.
+ */
+ wmb();
+
+ tegra_vde_writel(vde, 0x00000000, vde->bsev, 0x8C);
+ tegra_vde_writel(vde, bitstream_data_addr + bitstream_data_size,
+ vde->bsev, 0x54);
+
+ vde->bitstream_data_addr = bitstream_data_addr;
+
+ value = ctx->pic_width_in_mbs << 11 | ctx->pic_height_in_mbs << 3;
+
+ tegra_vde_writel(vde, value, vde->bsev, 0x88);
+
+ err = tegra_vde_wait_bsev(vde, false);
+ if (err)
+ return err;
+
+ err = tegra_vde_push_to_bsev_icmdqueue(vde, 0x800003FC, false);
+ if (err)
+ return err;
+
+ value = 0x01500000;
+ value |= ((vde->iram_lists_addr + 512) >> 2) & 0xFFFF;
+
+ err = tegra_vde_push_to_bsev_icmdqueue(vde, value, true);
+ if (err)
+ return err;
+
+ err = tegra_vde_push_to_bsev_icmdqueue(vde, 0x840F054C, false);
+ if (err)
+ return err;
+
+ err = tegra_vde_push_to_bsev_icmdqueue(vde, 0x80000080, false);
+ if (err)
+ return err;
+
+ value = 0x0E340000 | ((vde->iram_lists_addr >> 2) & 0xFFFF);
+
+ err = tegra_vde_push_to_bsev_icmdqueue(vde, value, true);
+ if (err)
+ return err;
+
+ value = 0x00800005;
+ value |= ctx->pic_width_in_mbs << 11;
+ value |= ctx->pic_height_in_mbs << 3;
+
+ tegra_vde_writel(vde, value, vde->sxe, 0x10);
+
+ value = !ctx->baseline_profile << 17;
+ value |= ctx->level_idc << 13;
+ value |= ctx->log2_max_pic_order_cnt_lsb << 7;
+ value |= ctx->pic_order_cnt_type << 5;
+ value |= ctx->log2_max_frame_num;
+
+ tegra_vde_writel(vde, value, vde->sxe, 0x40);
+
+ value = ctx->pic_init_qp << 25;
+ value |= !!(ctx->deblocking_filter_control_present_flag) << 2;
+ value |= !!ctx->pic_order_present_flag;
+
+ tegra_vde_writel(vde, value, vde->sxe, 0x44);
+
+ value = ctx->chroma_qp_index_offset;
+ value |= ctx->num_ref_idx_l0_active_minus1 << 5;
+ value |= ctx->num_ref_idx_l1_active_minus1 << 10;
+ value |= !!ctx->constrained_intra_pred_flag << 15;
+
+ tegra_vde_writel(vde, value, vde->sxe, 0x48);
+
+ value = 0x0C000000;
+ value |= !!(dpb_frames[0].flags & FLAG_B_FRAME) << 24;
+
+ tegra_vde_writel(vde, value, vde->sxe, 0x4C);
+
+ value = 0x03800000;
+ value |= bitstream_data_size & GENMASK(19, 15);
+
+ tegra_vde_writel(vde, value, vde->sxe, 0x68);
+
+ tegra_vde_writel(vde, bitstream_data_addr, vde->sxe, 0x6C);
+
+ if (vde->soc->supports_ref_pic_marking)
+ tegra_vde_writel(vde, vde->secure_bo->dma_addr, vde->sxe, 0x7c);
+
+ value = 0x10000005;
+ value |= ctx->pic_width_in_mbs << 11;
+ value |= ctx->pic_height_in_mbs << 3;
+
+ tegra_vde_writel(vde, value, vde->mbe, 0x80);
+
+ value = 0x26800000;
+ value |= ctx->level_idc << 4;
+ value |= !ctx->baseline_profile << 1;
+ value |= !!ctx->direct_8x8_inference_flag;
+
+ tegra_vde_writel(vde, value, vde->mbe, 0x80);
+
+ tegra_vde_writel(vde, 0xF4000001, vde->mbe, 0x80);
+ tegra_vde_writel(vde, 0x20000000, vde->mbe, 0x80);
+ tegra_vde_writel(vde, 0xF4000101, vde->mbe, 0x80);
+
+ value = 0x20000000;
+ value |= ctx->chroma_qp_index_offset << 8;
+
+ tegra_vde_writel(vde, value, vde->mbe, 0x80);
+
+ err = tegra_vde_setup_mbe_frame_idx(vde,
+ ctx->dpb_frames_nb - 1,
+ ctx->pic_order_cnt_type == 0);
+ if (err) {
+ dev_err(dev, "MBE frames setup failed %d\n", err);
+ return err;
+ }
+
+ tegra_vde_mbe_set_0xa_reg(vde, 0, 0x000009FC);
+ tegra_vde_mbe_set_0xa_reg(vde, 2, 0x61DEAD00);
+ tegra_vde_mbe_set_0xa_reg(vde, 4, 0x62DEAD00);
+ tegra_vde_mbe_set_0xa_reg(vde, 6, 0x63DEAD00);
+ tegra_vde_mbe_set_0xa_reg(vde, 8, dpb_frames[0].aux_addr);
+
+ value = 0xFC000000;
+ value |= !!(dpb_frames[0].flags & FLAG_B_FRAME) << 2;
+
+ if (!ctx->baseline_profile)
+ value |= !!(dpb_frames[0].flags & FLAG_REFERENCE) << 1;
+
+ tegra_vde_writel(vde, value, vde->mbe, 0x80);
+
+ err = tegra_vde_wait_mbe(vde);
+ if (err) {
+ dev_err(dev, "MBE programming failed %d\n", err);
+ return err;
+ }
+
+ return 0;
+}
+
+static void tegra_vde_decode_frame(struct tegra_vde *vde,
+ unsigned int macroblocks_nb)
+{
+ reinit_completion(&vde->decode_completion);
+
+ tegra_vde_writel(vde, 0x00000001, vde->bsev, 0x8C);
+ tegra_vde_writel(vde, 0x20000000 | (macroblocks_nb - 1),
+ vde->sxe, 0x00);
+}
+
+int tegra_vde_validate_h264_frame(struct device *dev,
+ struct tegra_vde_h264_frame *frame)
+{
+ if (frame->frame_num > 0x7FFFFF) {
+ dev_err(dev, "Bad frame_num %u\n", frame->frame_num);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+int tegra_vde_validate_h264_ctx(struct device *dev,
+ struct tegra_vde_h264_decoder_ctx *ctx)
+{
+ if (ctx->dpb_frames_nb == 0 || ctx->dpb_frames_nb > 17) {
+ dev_err(dev, "Bad DPB size %u\n", ctx->dpb_frames_nb);
+ return -EINVAL;
+ }
+
+ if (ctx->level_idc > 15) {
+ dev_err(dev, "Bad level value %u\n", ctx->level_idc);
+ return -EINVAL;
+ }
+
+ if (ctx->pic_init_qp > 52) {
+ dev_err(dev, "Bad pic_init_qp value %u\n", ctx->pic_init_qp);
+ return -EINVAL;
+ }
+
+ if (ctx->log2_max_pic_order_cnt_lsb > 16) {
+ dev_err(dev, "Bad log2_max_pic_order_cnt_lsb value %u\n",
+ ctx->log2_max_pic_order_cnt_lsb);
+ return -EINVAL;
+ }
+
+ if (ctx->log2_max_frame_num > 16) {
+ dev_err(dev, "Bad log2_max_frame_num value %u\n",
+ ctx->log2_max_frame_num);
+ return -EINVAL;
+ }
+
+ if (ctx->chroma_qp_index_offset > 31) {
+ dev_err(dev, "Bad chroma_qp_index_offset value %u\n",
+ ctx->chroma_qp_index_offset);
+ return -EINVAL;
+ }
+
+ if (ctx->pic_order_cnt_type > 2) {
+ dev_err(dev, "Bad pic_order_cnt_type value %u\n",
+ ctx->pic_order_cnt_type);
+ return -EINVAL;
+ }
+
+ if (ctx->num_ref_idx_l0_active_minus1 > 15) {
+ dev_err(dev, "Bad num_ref_idx_l0_active_minus1 value %u\n",
+ ctx->num_ref_idx_l0_active_minus1);
+ return -EINVAL;
+ }
+
+ if (ctx->num_ref_idx_l1_active_minus1 > 15) {
+ dev_err(dev, "Bad num_ref_idx_l1_active_minus1 value %u\n",
+ ctx->num_ref_idx_l1_active_minus1);
+ return -EINVAL;
+ }
+
+ if (!ctx->pic_width_in_mbs || ctx->pic_width_in_mbs > 127) {
+ dev_err(dev, "Bad pic_width_in_mbs value %u\n",
+ ctx->pic_width_in_mbs);
+ return -EINVAL;
+ }
+
+ if (!ctx->pic_height_in_mbs || ctx->pic_height_in_mbs > 127) {
+ dev_err(dev, "Bad pic_height_in_mbs value %u\n",
+ ctx->pic_height_in_mbs);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int tegra_vde_decode_begin(struct tegra_vde *vde,
+ struct tegra_vde_h264_decoder_ctx *ctx,
+ struct tegra_video_frame *dpb_frames,
+ dma_addr_t bitstream_data_addr,
+ size_t bitstream_data_size)
+{
+ struct device *dev = vde->dev;
+ unsigned int macroblocks_nb;
+ int err;
+
+ err = mutex_lock_interruptible(&vde->lock);
+ if (err)
+ return err;
+
+ err = pm_runtime_resume_and_get(dev);
+ if (err < 0)
+ goto unlock;
+
+ /*
+ * We rely on the VDE registers reset value, otherwise VDE
+ * causes bus lockup.
+ */
+ err = reset_control_assert(vde->rst_mc);
+ if (err) {
+ dev_err(dev, "DEC start: Failed to assert MC reset: %d\n",
+ err);
+ goto put_runtime_pm;
+ }
+
+ err = reset_control_reset(vde->rst);
+ if (err) {
+ dev_err(dev, "DEC start: Failed to reset HW: %d\n", err);
+ goto put_runtime_pm;
+ }
+
+ err = reset_control_deassert(vde->rst_mc);
+ if (err) {
+ dev_err(dev, "DEC start: Failed to deassert MC reset: %d\n",
+ err);
+ goto put_runtime_pm;
+ }
+
+ macroblocks_nb = ctx->pic_width_in_mbs * ctx->pic_height_in_mbs;
+
+ err = tegra_vde_setup_hw_context(vde, ctx, dpb_frames,
+ bitstream_data_addr,
+ bitstream_data_size,
+ macroblocks_nb);
+ if (err)
+ goto put_runtime_pm;
+
+ tegra_vde_decode_frame(vde, macroblocks_nb);
+
+ return 0;
+
+put_runtime_pm:
+ pm_runtime_mark_last_busy(dev);
+ pm_runtime_put_autosuspend(dev);
+
+unlock:
+ mutex_unlock(&vde->lock);
+
+ return err;
+}
+
+static void tegra_vde_decode_abort(struct tegra_vde *vde)
+{
+ struct device *dev = vde->dev;
+ int err;
+
+ /*
+ * At first reset memory client to avoid resetting VDE HW in the
+ * middle of DMA which could result into memory corruption or hang
+ * the whole system.
+ */
+ err = reset_control_assert(vde->rst_mc);
+ if (err)
+ dev_err(dev, "DEC end: Failed to assert MC reset: %d\n", err);
+
+ err = reset_control_assert(vde->rst);
+ if (err)
+ dev_err(dev, "DEC end: Failed to assert HW reset: %d\n", err);
+
+ pm_runtime_mark_last_busy(dev);
+ pm_runtime_put_autosuspend(dev);
+
+ mutex_unlock(&vde->lock);
+}
+
+static int tegra_vde_decode_end(struct tegra_vde *vde)
+{
+ unsigned int read_bytes, macroblocks_nb;
+ struct device *dev = vde->dev;
+ dma_addr_t bsev_ptr;
+ long timeout;
+ int ret;
+
+ timeout = wait_for_completion_interruptible_timeout(
+ &vde->decode_completion, msecs_to_jiffies(1000));
+ if (timeout == 0) {
+ bsev_ptr = tegra_vde_readl(vde, vde->bsev, 0x10);
+ macroblocks_nb = tegra_vde_readl(vde, vde->sxe, 0xC8) & 0x1FFF;
+ read_bytes = bsev_ptr ? bsev_ptr - vde->bitstream_data_addr : 0;
+
+ dev_err(dev, "Decoding failed: read 0x%X bytes, %u macroblocks parsed\n",
+ read_bytes, macroblocks_nb);
+
+ ret = -EIO;
+ } else if (timeout < 0) {
+ ret = timeout;
+ } else {
+ ret = 0;
+ }
+
+ tegra_vde_decode_abort(vde);
+
+ return ret;
+}
+
+int tegra_vde_decode_h264(struct tegra_vde *vde,
+ struct tegra_vde_h264_decoder_ctx *ctx,
+ struct tegra_video_frame *dpb_frames,
+ dma_addr_t bitstream_data_addr,
+ size_t bitstream_data_size)
+{
+ int err;
+
+ err = tegra_vde_decode_begin(vde, ctx, dpb_frames,
+ bitstream_data_addr,
+ bitstream_data_size);
+ if (err)
+ return err;
+
+ return tegra_vde_decode_end(vde);
+}
diff --git a/drivers/staging/media/tegra-vde/vde.c b/drivers/staging/media/tegra-vde/vde.c
index a8f1a024c343..36f5595c0fd8 100644
--- a/drivers/staging/media/tegra-vde/vde.c
+++ b/drivers/staging/media/tegra-vde/vde.c
@@ -10,7 +10,6 @@
#include <linux/dma-buf.h>
#include <linux/genalloc.h>
#include <linux/interrupt.h>
-#include <linux/iopoll.h>
#include <linux/list.h>
#include <linux/miscdevice.h>
#include <linux/module.h>
@@ -29,38 +28,15 @@
#define CREATE_TRACE_POINTS
#include "trace.h"
-#define ICMDQUE_WR 0x00
-#define CMDQUE_CONTROL 0x08
-#define INTR_STATUS 0x18
-#define BSE_INT_ENB 0x40
-#define BSE_CONFIG 0x44
-
-#define BSE_ICMDQUE_EMPTY BIT(3)
-#define BSE_DMA_BUSY BIT(23)
-
-struct video_frame {
- struct dma_buf_attachment *y_dmabuf_attachment;
- struct dma_buf_attachment *cb_dmabuf_attachment;
- struct dma_buf_attachment *cr_dmabuf_attachment;
- struct dma_buf_attachment *aux_dmabuf_attachment;
- dma_addr_t y_addr;
- dma_addr_t cb_addr;
- dma_addr_t cr_addr;
- dma_addr_t aux_addr;
- u32 frame_num;
- u32 flags;
-};
-
-static void tegra_vde_writel(struct tegra_vde *vde,
- u32 value, void __iomem *base, u32 offset)
+void tegra_vde_writel(struct tegra_vde *vde, u32 value,
+ void __iomem *base, u32 offset)
{
trace_vde_writel(vde, base, offset, value);
writel_relaxed(value, base + offset);
}
-static u32 tegra_vde_readl(struct tegra_vde *vde,
- void __iomem *base, u32 offset)
+u32 tegra_vde_readl(struct tegra_vde *vde, void __iomem *base, u32 offset)
{
u32 value = readl_relaxed(base + offset);
@@ -69,22 +45,14 @@ static u32 tegra_vde_readl(struct tegra_vde *vde,
return value;
}
-static void tegra_vde_set_bits(struct tegra_vde *vde,
- u32 mask, void __iomem *base, u32 offset)
+void tegra_vde_set_bits(struct tegra_vde *vde, u32 mask,
+ void __iomem *base, u32 offset)
{
u32 value = tegra_vde_readl(vde, base, offset);
tegra_vde_writel(vde, value | mask, base, offset);
}
-static int tegra_vde_wait_mbe(struct tegra_vde *vde)
-{
- u32 tmp;
-
- return readl_relaxed_poll_timeout(vde->mbe + 0x8C, tmp,
- (tmp >= 0x10), 1, 100);
-}
-
static int tegra_vde_alloc_bo(struct tegra_vde *vde,
struct tegra_vde_bo **ret_bo,
enum dma_data_direction dma_dir,
@@ -175,412 +143,6 @@ static void tegra_vde_free_bo(struct tegra_vde_bo *bo)
kfree(bo);
}
-static int tegra_vde_setup_mbe_frame_idx(struct tegra_vde *vde,
- unsigned int refs_nb,
- bool setup_refs)
-{
- u32 frame_idx_enb_mask = 0;
- u32 value;
- unsigned int frame_idx;
- unsigned int idx;
- int err;
-
- tegra_vde_writel(vde, 0xD0000000 | (0 << 23), vde->mbe, 0x80);
- tegra_vde_writel(vde, 0xD0200000 | (0 << 23), vde->mbe, 0x80);
-
- err = tegra_vde_wait_mbe(vde);
- if (err)
- return err;
-
- if (!setup_refs)
- return 0;
-
- for (idx = 0, frame_idx = 1; idx < refs_nb; idx++, frame_idx++) {
- tegra_vde_writel(vde, 0xD0000000 | (frame_idx << 23),
- vde->mbe, 0x80);
- tegra_vde_writel(vde, 0xD0200000 | (frame_idx << 23),
- vde->mbe, 0x80);
-
- frame_idx_enb_mask |= frame_idx << (6 * (idx % 4));
-
- if (idx % 4 == 3 || idx == refs_nb - 1) {
- value = 0xC0000000;
- value |= (idx >> 2) << 24;
- value |= frame_idx_enb_mask;
-
- tegra_vde_writel(vde, value, vde->mbe, 0x80);
-
- err = tegra_vde_wait_mbe(vde);
- if (err)
- return err;
-
- frame_idx_enb_mask = 0;
- }
- }
-
- return 0;
-}
-
-static void tegra_vde_mbe_set_0xa_reg(struct tegra_vde *vde, int reg, u32 val)
-{
- tegra_vde_writel(vde, 0xA0000000 | (reg << 24) | (val & 0xFFFF),
- vde->mbe, 0x80);
- tegra_vde_writel(vde, 0xA0000000 | ((reg + 1) << 24) | (val >> 16),
- vde->mbe, 0x80);
-}
-
-static int tegra_vde_wait_bsev(struct tegra_vde *vde, bool wait_dma)
-{
- struct device *dev = vde->miscdev.parent;
- u32 value;
- int err;
-
- err = readl_relaxed_poll_timeout(vde->bsev + INTR_STATUS, value,
- !(value & BIT(2)), 1, 100);
- if (err) {
- dev_err(dev, "BSEV unknown bit timeout\n");
- return err;
- }
-
- err = readl_relaxed_poll_timeout(vde->bsev + INTR_STATUS, value,
- (value & BSE_ICMDQUE_EMPTY), 1, 100);
- if (err) {
- dev_err(dev, "BSEV ICMDQUE flush timeout\n");
- return err;
- }
-
- if (!wait_dma)
- return 0;
-
- err = readl_relaxed_poll_timeout(vde->bsev + INTR_STATUS, value,
- !(value & BSE_DMA_BUSY), 1, 100);
- if (err) {
- dev_err(dev, "BSEV DMA timeout\n");
- return err;
- }
-
- return 0;
-}
-
-static int tegra_vde_push_to_bsev_icmdqueue(struct tegra_vde *vde,
- u32 value, bool wait_dma)
-{
- tegra_vde_writel(vde, value, vde->bsev, ICMDQUE_WR);
-
- return tegra_vde_wait_bsev(vde, wait_dma);
-}
-
-static void tegra_vde_setup_frameid(struct tegra_vde *vde,
- struct video_frame *frame,
- unsigned int frameid,
- u32 mbs_width, u32 mbs_height)
-{
- u32 y_addr = frame ? frame->y_addr : 0x6CDEAD00;
- u32 cb_addr = frame ? frame->cb_addr : 0x6CDEAD00;
- u32 cr_addr = frame ? frame->cr_addr : 0x6CDEAD00;
- u32 value1 = frame ? ((mbs_width << 16) | mbs_height) : 0;
- u32 value2 = frame ? ((((mbs_width + 1) >> 1) << 6) | 1) : 0;
-
- tegra_vde_writel(vde, y_addr >> 8, vde->frameid, 0x000 + frameid * 4);
- tegra_vde_writel(vde, cb_addr >> 8, vde->frameid, 0x100 + frameid * 4);
- tegra_vde_writel(vde, cr_addr >> 8, vde->frameid, 0x180 + frameid * 4);
- tegra_vde_writel(vde, value1, vde->frameid, 0x080 + frameid * 4);
- tegra_vde_writel(vde, value2, vde->frameid, 0x280 + frameid * 4);
-}
-
-static void tegra_setup_frameidx(struct tegra_vde *vde,
- struct video_frame *frames,
- unsigned int frames_nb,
- u32 mbs_width, u32 mbs_height)
-{
- unsigned int idx;
-
- for (idx = 0; idx < frames_nb; idx++)
- tegra_vde_setup_frameid(vde, &frames[idx], idx,
- mbs_width, mbs_height);
-
- for (; idx < 17; idx++)
- tegra_vde_setup_frameid(vde, NULL, idx, 0, 0);
-}
-
-static void tegra_vde_setup_iram_entry(struct tegra_vde *vde,
- unsigned int table,
- unsigned int row,
- u32 value1, u32 value2)
-{
- u32 *iram_tables = vde->iram;
-
- trace_vde_setup_iram_entry(table, row, value1, value2);
-
- iram_tables[0x20 * table + row * 2] = value1;
- iram_tables[0x20 * table + row * 2 + 1] = value2;
-}
-
-static void tegra_vde_setup_iram_tables(struct tegra_vde *vde,
- struct video_frame *dpb_frames,
- unsigned int ref_frames_nb,
- unsigned int with_earlier_poc_nb)
-{
- struct video_frame *frame;
- u32 value, aux_addr;
- int with_later_poc_nb;
- unsigned int i, k;
-
- trace_vde_ref_l0(dpb_frames[0].frame_num);
-
- for (i = 0; i < 16; i++) {
- if (i < ref_frames_nb) {
- frame = &dpb_frames[i + 1];
-
- aux_addr = frame->aux_addr;
-
- value = (i + 1) << 26;
- value |= !(frame->flags & FLAG_B_FRAME) << 25;
- value |= 1 << 24;
- value |= frame->frame_num;
- } else {
- aux_addr = 0x6ADEAD00;
- value = 0x3f;
- }
-
- tegra_vde_setup_iram_entry(vde, 0, i, value, aux_addr);
- tegra_vde_setup_iram_entry(vde, 1, i, value, aux_addr);
- tegra_vde_setup_iram_entry(vde, 2, i, value, aux_addr);
- tegra_vde_setup_iram_entry(vde, 3, i, value, aux_addr);
- }
-
- if (!(dpb_frames[0].flags & FLAG_B_FRAME))
- return;
-
- if (with_earlier_poc_nb >= ref_frames_nb)
- return;
-
- with_later_poc_nb = ref_frames_nb - with_earlier_poc_nb;
-
- trace_vde_ref_l1(with_later_poc_nb, with_earlier_poc_nb);
-
- for (i = 0, k = with_earlier_poc_nb; i < with_later_poc_nb; i++, k++) {
- frame = &dpb_frames[k + 1];
-
- aux_addr = frame->aux_addr;
-
- value = (k + 1) << 26;
- value |= !(frame->flags & FLAG_B_FRAME) << 25;
- value |= 1 << 24;
- value |= frame->frame_num;
-
- tegra_vde_setup_iram_entry(vde, 2, i, value, aux_addr);
- }
-
- for (k = 0; i < ref_frames_nb; i++, k++) {
- frame = &dpb_frames[k + 1];
-
- aux_addr = frame->aux_addr;
-
- value = (k + 1) << 26;
- value |= !(frame->flags & FLAG_B_FRAME) << 25;
- value |= 1 << 24;
- value |= frame->frame_num;
-
- tegra_vde_setup_iram_entry(vde, 2, i, value, aux_addr);
- }
-}
-
-static int tegra_vde_setup_hw_context(struct tegra_vde *vde,
- struct tegra_vde_h264_decoder_ctx *ctx,
- struct video_frame *dpb_frames,
- dma_addr_t bitstream_data_addr,
- size_t bitstream_data_size,
- unsigned int macroblocks_nb)
-{
- struct device *dev = vde->miscdev.parent;
- u32 value;
- int err;
-
- tegra_vde_set_bits(vde, 0x000A, vde->sxe, 0xF0);
- tegra_vde_set_bits(vde, 0x000B, vde->bsev, CMDQUE_CONTROL);
- tegra_vde_set_bits(vde, 0x8002, vde->mbe, 0x50);
- tegra_vde_set_bits(vde, 0x000A, vde->mbe, 0xA0);
- tegra_vde_set_bits(vde, 0x000A, vde->ppe, 0x14);
- tegra_vde_set_bits(vde, 0x000A, vde->ppe, 0x28);
- tegra_vde_set_bits(vde, 0x0A00, vde->mce, 0x08);
- tegra_vde_set_bits(vde, 0x000A, vde->tfe, 0x00);
- tegra_vde_set_bits(vde, 0x0005, vde->vdma, 0x04);
-
- tegra_vde_writel(vde, 0x00000000, vde->vdma, 0x1C);
- tegra_vde_writel(vde, 0x00000000, vde->vdma, 0x00);
- tegra_vde_writel(vde, 0x00000007, vde->vdma, 0x04);
- tegra_vde_writel(vde, 0x00000007, vde->frameid, 0x200);
- tegra_vde_writel(vde, 0x00000005, vde->tfe, 0x04);
- tegra_vde_writel(vde, 0x00000000, vde->mbe, 0x84);
- tegra_vde_writel(vde, 0x00000010, vde->sxe, 0x08);
- tegra_vde_writel(vde, 0x00000150, vde->sxe, 0x54);
- tegra_vde_writel(vde, 0x0000054C, vde->sxe, 0x58);
- tegra_vde_writel(vde, 0x00000E34, vde->sxe, 0x5C);
- tegra_vde_writel(vde, 0x063C063C, vde->mce, 0x10);
- tegra_vde_writel(vde, 0x0003FC00, vde->bsev, INTR_STATUS);
- tegra_vde_writel(vde, 0x0000150D, vde->bsev, BSE_CONFIG);
- tegra_vde_writel(vde, 0x00000100, vde->bsev, BSE_INT_ENB);
- tegra_vde_writel(vde, 0x00000000, vde->bsev, 0x98);
- tegra_vde_writel(vde, 0x00000060, vde->bsev, 0x9C);
-
- memset(vde->iram + 128, 0, macroblocks_nb / 2);
-
- tegra_setup_frameidx(vde, dpb_frames, ctx->dpb_frames_nb,
- ctx->pic_width_in_mbs, ctx->pic_height_in_mbs);
-
- tegra_vde_setup_iram_tables(vde, dpb_frames,
- ctx->dpb_frames_nb - 1,
- ctx->dpb_ref_frames_with_earlier_poc_nb);
-
- /*
- * The IRAM mapping is write-combine, ensure that CPU buffers have
- * been flushed at this point.
- */
- wmb();
-
- tegra_vde_writel(vde, 0x00000000, vde->bsev, 0x8C);
- tegra_vde_writel(vde, bitstream_data_addr + bitstream_data_size,
- vde->bsev, 0x54);
-
- value = ctx->pic_width_in_mbs << 11 | ctx->pic_height_in_mbs << 3;
-
- tegra_vde_writel(vde, value, vde->bsev, 0x88);
-
- err = tegra_vde_wait_bsev(vde, false);
- if (err)
- return err;
-
- err = tegra_vde_push_to_bsev_icmdqueue(vde, 0x800003FC, false);
- if (err)
- return err;
-
- value = 0x01500000;
- value |= ((vde->iram_lists_addr + 512) >> 2) & 0xFFFF;
-
- err = tegra_vde_push_to_bsev_icmdqueue(vde, value, true);
- if (err)
- return err;
-
- err = tegra_vde_push_to_bsev_icmdqueue(vde, 0x840F054C, false);
- if (err)
- return err;
-
- err = tegra_vde_push_to_bsev_icmdqueue(vde, 0x80000080, false);
- if (err)
- return err;
-
- value = 0x0E340000 | ((vde->iram_lists_addr >> 2) & 0xFFFF);
-
- err = tegra_vde_push_to_bsev_icmdqueue(vde, value, true);
- if (err)
- return err;
-
- value = 0x00800005;
- value |= ctx->pic_width_in_mbs << 11;
- value |= ctx->pic_height_in_mbs << 3;
-
- tegra_vde_writel(vde, value, vde->sxe, 0x10);
-
- value = !ctx->baseline_profile << 17;
- value |= ctx->level_idc << 13;
- value |= ctx->log2_max_pic_order_cnt_lsb << 7;
- value |= ctx->pic_order_cnt_type << 5;
- value |= ctx->log2_max_frame_num;
-
- tegra_vde_writel(vde, value, vde->sxe, 0x40);
-
- value = ctx->pic_init_qp << 25;
- value |= !!(ctx->deblocking_filter_control_present_flag) << 2;
- value |= !!ctx->pic_order_present_flag;
-
- tegra_vde_writel(vde, value, vde->sxe, 0x44);
-
- value = ctx->chroma_qp_index_offset;
- value |= ctx->num_ref_idx_l0_active_minus1 << 5;
- value |= ctx->num_ref_idx_l1_active_minus1 << 10;
- value |= !!ctx->constrained_intra_pred_flag << 15;
-
- tegra_vde_writel(vde, value, vde->sxe, 0x48);
-
- value = 0x0C000000;
- value |= !!(dpb_frames[0].flags & FLAG_B_FRAME) << 24;
-
- tegra_vde_writel(vde, value, vde->sxe, 0x4C);
-
- value = 0x03800000;
- value |= bitstream_data_size & GENMASK(19, 15);
-
- tegra_vde_writel(vde, value, vde->sxe, 0x68);
-
- tegra_vde_writel(vde, bitstream_data_addr, vde->sxe, 0x6C);
-
- if (vde->soc->supports_ref_pic_marking)
- tegra_vde_writel(vde, vde->secure_bo->dma_addr, vde->sxe, 0x7c);
-
- value = 0x10000005;
- value |= ctx->pic_width_in_mbs << 11;
- value |= ctx->pic_height_in_mbs << 3;
-
- tegra_vde_writel(vde, value, vde->mbe, 0x80);
-
- value = 0x26800000;
- value |= ctx->level_idc << 4;
- value |= !ctx->baseline_profile << 1;
- value |= !!ctx->direct_8x8_inference_flag;
-
- tegra_vde_writel(vde, value, vde->mbe, 0x80);
-
- tegra_vde_writel(vde, 0xF4000001, vde->mbe, 0x80);
- tegra_vde_writel(vde, 0x20000000, vde->mbe, 0x80);
- tegra_vde_writel(vde, 0xF4000101, vde->mbe, 0x80);
-
- value = 0x20000000;
- value |= ctx->chroma_qp_index_offset << 8;
-
- tegra_vde_writel(vde, value, vde->mbe, 0x80);
-
- err = tegra_vde_setup_mbe_frame_idx(vde,
- ctx->dpb_frames_nb - 1,
- ctx->pic_order_cnt_type == 0);
- if (err) {
- dev_err(dev, "MBE frames setup failed %d\n", err);
- return err;
- }
-
- tegra_vde_mbe_set_0xa_reg(vde, 0, 0x000009FC);
- tegra_vde_mbe_set_0xa_reg(vde, 2, 0x61DEAD00);
- tegra_vde_mbe_set_0xa_reg(vde, 4, 0x62DEAD00);
- tegra_vde_mbe_set_0xa_reg(vde, 6, 0x63DEAD00);
- tegra_vde_mbe_set_0xa_reg(vde, 8, dpb_frames[0].aux_addr);
-
- value = 0xFC000000;
- value |= !!(dpb_frames[0].flags & FLAG_B_FRAME) << 2;
-
- if (!ctx->baseline_profile)
- value |= !!(dpb_frames[0].flags & FLAG_REFERENCE) << 1;
-
- tegra_vde_writel(vde, value, vde->mbe, 0x80);
-
- err = tegra_vde_wait_mbe(vde);
- if (err) {
- dev_err(dev, "MBE programming failed %d\n", err);
- return err;
- }
-
- return 0;
-}
-
-static void tegra_vde_decode_frame(struct tegra_vde *vde,
- unsigned int macroblocks_nb)
-{
- reinit_completion(&vde->decode_completion);
-
- tegra_vde_writel(vde, 0x00000001, vde->bsev, 0x8C);
- tegra_vde_writel(vde, 0x20000000 | (macroblocks_nb - 1),
- vde->sxe, 0x00);
-}
-
static int tegra_vde_attach_dmabuf(struct tegra_vde *vde,
int fd,
unsigned long offset,
@@ -631,7 +193,7 @@ static int tegra_vde_attach_dmabuf(struct tegra_vde *vde,
}
static int tegra_vde_attach_dmabufs_to_frame(struct tegra_vde *vde,
- struct video_frame *frame,
+ struct tegra_video_frame *frame,
struct tegra_vde_h264_frame *src,
enum dma_data_direction dma_dir,
bool baseline_profile,
@@ -689,7 +251,7 @@ static int tegra_vde_attach_dmabufs_to_frame(struct tegra_vde *vde,
}
static void tegra_vde_release_frame_dmabufs(struct tegra_vde *vde,
- struct video_frame *frame,
+ struct tegra_video_frame *frame,
enum dma_data_direction dma_dir,
bool baseline_profile,
bool release)
@@ -703,106 +265,22 @@ static void tegra_vde_release_frame_dmabufs(struct tegra_vde *vde,
tegra_vde_dmabuf_cache_unmap(vde, frame->y_dmabuf_attachment, release);
}
-static int tegra_vde_validate_frame(struct device *dev,
- struct tegra_vde_h264_frame *frame)
-{
- if (frame->frame_num > 0x7FFFFF) {
- dev_err(dev, "Bad frame_num %u\n", frame->frame_num);
- return -EINVAL;
- }
-
- return 0;
-}
-
-static int tegra_vde_validate_h264_ctx(struct device *dev,
- struct tegra_vde_h264_decoder_ctx *ctx)
-{
- if (ctx->dpb_frames_nb == 0 || ctx->dpb_frames_nb > 17) {
- dev_err(dev, "Bad DPB size %u\n", ctx->dpb_frames_nb);
- return -EINVAL;
- }
-
- if (ctx->level_idc > 15) {
- dev_err(dev, "Bad level value %u\n", ctx->level_idc);
- return -EINVAL;
- }
-
- if (ctx->pic_init_qp > 52) {
- dev_err(dev, "Bad pic_init_qp value %u\n", ctx->pic_init_qp);
- return -EINVAL;
- }
-
- if (ctx->log2_max_pic_order_cnt_lsb > 16) {
- dev_err(dev, "Bad log2_max_pic_order_cnt_lsb value %u\n",
- ctx->log2_max_pic_order_cnt_lsb);
- return -EINVAL;
- }
-
- if (ctx->log2_max_frame_num > 16) {
- dev_err(dev, "Bad log2_max_frame_num value %u\n",
- ctx->log2_max_frame_num);
- return -EINVAL;
- }
-
- if (ctx->chroma_qp_index_offset > 31) {
- dev_err(dev, "Bad chroma_qp_index_offset value %u\n",
- ctx->chroma_qp_index_offset);
- return -EINVAL;
- }
-
- if (ctx->pic_order_cnt_type > 2) {
- dev_err(dev, "Bad pic_order_cnt_type value %u\n",
- ctx->pic_order_cnt_type);
- return -EINVAL;
- }
-
- if (ctx->num_ref_idx_l0_active_minus1 > 15) {
- dev_err(dev, "Bad num_ref_idx_l0_active_minus1 value %u\n",
- ctx->num_ref_idx_l0_active_minus1);
- return -EINVAL;
- }
-
- if (ctx->num_ref_idx_l1_active_minus1 > 15) {
- dev_err(dev, "Bad num_ref_idx_l1_active_minus1 value %u\n",
- ctx->num_ref_idx_l1_active_minus1);
- return -EINVAL;
- }
-
- if (!ctx->pic_width_in_mbs || ctx->pic_width_in_mbs > 127) {
- dev_err(dev, "Bad pic_width_in_mbs value %u\n",
- ctx->pic_width_in_mbs);
- return -EINVAL;
- }
-
- if (!ctx->pic_height_in_mbs || ctx->pic_height_in_mbs > 127) {
- dev_err(dev, "Bad pic_height_in_mbs value %u\n",
- ctx->pic_height_in_mbs);
- return -EINVAL;
- }
-
- return 0;
-}
-
static int tegra_vde_ioctl_decode_h264(struct tegra_vde *vde,
unsigned long vaddr)
{
+ struct dma_buf_attachment *bitstream_data_dmabuf_attachment;
+ struct tegra_vde_h264_frame __user *frames_user;
+ size_t bitstream_data_size, lsize, csize;
struct device *dev = vde->miscdev.parent;
struct tegra_vde_h264_decoder_ctx ctx;
+ struct tegra_video_frame *dpb_frames;
struct tegra_vde_h264_frame *frames;
- struct tegra_vde_h264_frame __user *frames_user;
- struct video_frame *dpb_frames;
- struct dma_buf_attachment *bitstream_data_dmabuf_attachment;
enum dma_data_direction dma_dir;
dma_addr_t bitstream_data_addr;
- dma_addr_t bsev_ptr;
- size_t lsize, csize;
- size_t bitstream_data_size;
unsigned int macroblocks_nb;
- unsigned int read_bytes;
unsigned int cstride;
unsigned int i;
- long timeout;
- int ret, err;
+ int ret;
if (copy_from_user(&ctx, (void __user *)vaddr, sizeof(ctx)))
return -EFAULT;
@@ -848,7 +326,7 @@ static int tegra_vde_ioctl_decode_h264(struct tegra_vde *vde,
lsize = macroblocks_nb * 256;
for (i = 0; i < ctx.dpb_frames_nb; i++) {
- ret = tegra_vde_validate_frame(dev, &frames[i]);
+ ret = tegra_vde_validate_h264_frame(dev, &frames[i]);
if (ret)
goto release_dpb_frames;
@@ -865,81 +343,8 @@ static int tegra_vde_ioctl_decode_h264(struct tegra_vde *vde,
goto release_dpb_frames;
}
- ret = mutex_lock_interruptible(&vde->lock);
- if (ret)
- goto release_dpb_frames;
-
- ret = pm_runtime_resume_and_get(dev);
- if (ret < 0)
- goto unlock;
-
- /*
- * We rely on the VDE registers reset value, otherwise VDE
- * causes bus lockup.
- */
- ret = reset_control_assert(vde->rst_mc);
- if (ret) {
- dev_err(dev, "DEC start: Failed to assert MC reset: %d\n",
- ret);
- goto put_runtime_pm;
- }
-
- ret = reset_control_reset(vde->rst);
- if (ret) {
- dev_err(dev, "DEC start: Failed to reset HW: %d\n", ret);
- goto put_runtime_pm;
- }
-
- ret = reset_control_deassert(vde->rst_mc);
- if (ret) {
- dev_err(dev, "DEC start: Failed to deassert MC reset: %d\n",
- ret);
- goto put_runtime_pm;
- }
-
- ret = tegra_vde_setup_hw_context(vde, &ctx, dpb_frames,
- bitstream_data_addr,
- bitstream_data_size,
- macroblocks_nb);
- if (ret)
- goto put_runtime_pm;
-
- tegra_vde_decode_frame(vde, macroblocks_nb);
-
- timeout = wait_for_completion_interruptible_timeout(
- &vde->decode_completion, msecs_to_jiffies(1000));
- if (timeout == 0) {
- bsev_ptr = tegra_vde_readl(vde, vde->bsev, 0x10);
- macroblocks_nb = tegra_vde_readl(vde, vde->sxe, 0xC8) & 0x1FFF;
- read_bytes = bsev_ptr ? bsev_ptr - bitstream_data_addr : 0;
-
- dev_err(dev, "Decoding failed: read 0x%X bytes, %u macroblocks parsed\n",
- read_bytes, macroblocks_nb);
-
- ret = -EIO;
- } else if (timeout < 0) {
- ret = timeout;
- }
-
- /*
- * At first reset memory client to avoid resetting VDE HW in the
- * middle of DMA which could result into memory corruption or hang
- * the whole system.
- */
- err = reset_control_assert(vde->rst_mc);
- if (err)
- dev_err(dev, "DEC end: Failed to assert MC reset: %d\n", err);
-
- err = reset_control_assert(vde->rst);
- if (err)
- dev_err(dev, "DEC end: Failed to assert HW reset: %d\n", err);
-
-put_runtime_pm:
- pm_runtime_mark_last_busy(dev);
- pm_runtime_put_autosuspend(dev);
-
-unlock:
- mutex_unlock(&vde->lock);
+ ret = tegra_vde_decode_h264(vde, &ctx, dpb_frames,
+ bitstream_data_addr, bitstream_data_size);
release_dpb_frames:
while (i--) {
@@ -1088,6 +493,7 @@ static int tegra_vde_probe(struct platform_device *pdev)
platform_set_drvdata(pdev, vde);
vde->soc = of_device_get_match_data(&pdev->dev);
+ vde->dev = dev;
vde->sxe = devm_platform_ioremap_resource_byname(pdev, "sxe");
if (IS_ERR(vde->sxe))
diff --git a/drivers/staging/media/tegra-vde/vde.h b/drivers/staging/media/tegra-vde/vde.h
index bbd42b8d9991..8ba6a71e3e40 100644
--- a/drivers/staging/media/tegra-vde/vde.h
+++ b/drivers/staging/media/tegra-vde/vde.h
@@ -16,6 +16,15 @@
#include <linux/mutex.h>
#include <linux/types.h>
+#define ICMDQUE_WR 0x00
+#define CMDQUE_CONTROL 0x08
+#define INTR_STATUS 0x18
+#define BSE_INT_ENB 0x40
+#define BSE_CONFIG 0x44
+
+#define BSE_ICMDQUE_EMPTY BIT(3)
+#define BSE_DMA_BUSY BIT(23)
+
struct clk;
struct dma_buf;
struct gen_pool;
@@ -23,6 +32,21 @@ struct iommu_group;
struct iommu_domain;
struct reset_control;
struct dma_buf_attachment;
+struct tegra_vde_h264_frame;
+struct tegra_vde_h264_decoder_ctx;
+
+struct tegra_video_frame {
+ struct dma_buf_attachment *y_dmabuf_attachment;
+ struct dma_buf_attachment *cb_dmabuf_attachment;
+ struct dma_buf_attachment *cr_dmabuf_attachment;
+ struct dma_buf_attachment *aux_dmabuf_attachment;
+ dma_addr_t y_addr;
+ dma_addr_t cb_addr;
+ dma_addr_t cr_addr;
+ dma_addr_t aux_addr;
+ u32 frame_num;
+ u32 flags;
+};
struct tegra_vde_soc {
bool supports_ref_pic_marking;
@@ -50,6 +74,7 @@ struct tegra_vde {
void __iomem *ppb;
void __iomem *vdma;
void __iomem *frameid;
+ struct device *dev;
struct mutex lock;
struct mutex map_lock;
struct list_head map_list;
@@ -66,10 +91,27 @@ struct tegra_vde {
struct iova *iova_resv_last_page;
const struct tegra_vde_soc *soc;
struct tegra_vde_bo *secure_bo;
+ dma_addr_t bitstream_data_addr;
dma_addr_t iram_lists_addr;
u32 *iram;
};
+void tegra_vde_writel(struct tegra_vde *vde, u32 value, void __iomem *base,
+ u32 offset);
+u32 tegra_vde_readl(struct tegra_vde *vde, void __iomem *base, u32 offset);
+void tegra_vde_set_bits(struct tegra_vde *vde, u32 mask, void __iomem *base,
+ u32 offset);
+
+int tegra_vde_validate_h264_frame(struct device *dev,
+ struct tegra_vde_h264_frame *frame);
+int tegra_vde_validate_h264_ctx(struct device *dev,
+ struct tegra_vde_h264_decoder_ctx *ctx);
+int tegra_vde_decode_h264(struct tegra_vde *vde,
+ struct tegra_vde_h264_decoder_ctx *ctx,
+ struct tegra_video_frame *dpb_frames,
+ dma_addr_t bitstream_data_addr,
+ size_t bitstream_data_size);
+
int tegra_vde_iommu_init(struct tegra_vde *vde);
void tegra_vde_iommu_deinit(struct tegra_vde *vde);
int tegra_vde_iommu_map(struct tegra_vde *vde,
--
2.34.1
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH v3 3/4] media: staging: tegra-vde: Support V4L stateless video decoder API
2022-02-07 14:19 [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Dmitry Osipenko
2022-02-07 14:19 ` [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags Dmitry Osipenko
2022-02-07 14:19 ` [PATCH v3 2/4] media: staging: tegra-vde: Factor out H.264 code Dmitry Osipenko
@ 2022-02-07 14:19 ` Dmitry Osipenko
2022-02-07 14:19 ` [PATCH v3 4/4] media: staging: tegra-vde: Remove legacy UAPI support Dmitry Osipenko
2022-02-18 15:04 ` [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Hans Verkuil
4 siblings, 0 replies; 10+ messages in thread
From: Dmitry Osipenko @ 2022-02-07 14:19 UTC (permalink / raw)
To: Thierry Reding, Jonathan Hunter, Mauro Carvalho Chehab,
Hans Verkuil, Nicolas Dufresne, Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
Expose Tegra video decoder as a generic V4L M2M stateless video decoder.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
drivers/staging/media/tegra-vde/Kconfig | 7 +
drivers/staging/media/tegra-vde/Makefile | 2 +-
drivers/staging/media/tegra-vde/h264.c | 304 ++++++-
drivers/staging/media/tegra-vde/v4l2.c | 1018 ++++++++++++++++++++++
drivers/staging/media/tegra-vde/vde.c | 74 +-
drivers/staging/media/tegra-vde/vde.h | 86 ++
6 files changed, 1483 insertions(+), 8 deletions(-)
create mode 100644 drivers/staging/media/tegra-vde/v4l2.c
diff --git a/drivers/staging/media/tegra-vde/Kconfig b/drivers/staging/media/tegra-vde/Kconfig
index 0dc78afd09e0..07dbc1f44ca8 100644
--- a/drivers/staging/media/tegra-vde/Kconfig
+++ b/drivers/staging/media/tegra-vde/Kconfig
@@ -2,9 +2,16 @@
config TEGRA_VDE
tristate "NVIDIA Tegra Video Decoder Engine driver"
depends on ARCH_TEGRA || COMPILE_TEST
+ depends on VIDEO_DEV && VIDEO_V4L2
select DMA_SHARED_BUFFER
select IOMMU_IOVA
+ select MEDIA_CONTROLLER
+ select MEDIA_CONTROLLER_REQUEST_API
select SRAM
+ select VIDEOBUF2_DMA_CONTIG
+ select VIDEOBUF2_DMA_SG
+ select V4L2_H264
+ select V4L2_MEM2MEM_DEV
help
Say Y here to enable support for the NVIDIA Tegra video decoder
driver.
diff --git a/drivers/staging/media/tegra-vde/Makefile b/drivers/staging/media/tegra-vde/Makefile
index 43525b08b3b0..1758aa201baa 100644
--- a/drivers/staging/media/tegra-vde/Makefile
+++ b/drivers/staging/media/tegra-vde/Makefile
@@ -1,3 +1,3 @@
# SPDX-License-Identifier: GPL-2.0
-tegra-vde-y := vde.o iommu.o dmabuf-cache.o h264.o
+tegra-vde-y := vde.o iommu.o dmabuf-cache.o h264.o v4l2.o
obj-$(CONFIG_TEGRA_VDE) += tegra-vde.o
diff --git a/drivers/staging/media/tegra-vde/h264.c b/drivers/staging/media/tegra-vde/h264.c
index 03faa705bf71..cbf27f0cfdb1 100644
--- a/drivers/staging/media/tegra-vde/h264.c
+++ b/drivers/staging/media/tegra-vde/h264.c
@@ -11,10 +11,18 @@
#include <linux/reset.h>
#include <linux/slab.h>
+#include <media/v4l2-h264.h>
+
#include "trace.h"
#include "uapi.h"
#include "vde.h"
+struct h264_reflists {
+ u8 p[V4L2_H264_NUM_DPB_ENTRIES];
+ u8 b0[V4L2_H264_NUM_DPB_ENTRIES];
+ u8 b1[V4L2_H264_NUM_DPB_ENTRIES];
+};
+
static int tegra_vde_wait_mbe(struct tegra_vde *vde)
{
u32 tmp;
@@ -125,8 +133,8 @@ static void tegra_vde_setup_frameid(struct tegra_vde *vde,
u32 y_addr = frame ? frame->y_addr : 0x6CDEAD00;
u32 cb_addr = frame ? frame->cb_addr : 0x6CDEAD00;
u32 cr_addr = frame ? frame->cr_addr : 0x6CDEAD00;
- u32 value1 = frame ? ((mbs_width << 16) | mbs_height) : 0;
- u32 value2 = frame ? ((((mbs_width + 1) >> 1) << 6) | 1) : 0;
+ u32 value1 = frame ? ((frame->luma_atoms_pitch << 16) | mbs_height) : 0;
+ u32 value2 = frame ? ((frame->chroma_atoms_pitch << 6) | 1) : 0;
tegra_vde_writel(vde, y_addr >> 8, vde->frameid, 0x000 + frameid * 4);
tegra_vde_writel(vde, cb_addr >> 8, vde->frameid, 0x100 + frameid * 4);
@@ -645,3 +653,295 @@ int tegra_vde_decode_h264(struct tegra_vde *vde,
return tegra_vde_decode_end(vde);
}
+
+static struct vb2_buffer *get_ref_buf(struct tegra_ctx *ctx,
+ struct vb2_v4l2_buffer *dst,
+ unsigned int dpb_idx)
+{
+ const struct v4l2_h264_dpb_entry *dpb = ctx->h264.decode_params->dpb;
+ struct vb2_queue *cap_q = &ctx->fh.m2m_ctx->cap_q_ctx.q;
+ int buf_idx = -1;
+
+ if (dpb[dpb_idx].flags & V4L2_H264_DPB_ENTRY_FLAG_ACTIVE)
+ buf_idx = vb2_find_timestamp(cap_q,
+ dpb[dpb_idx].reference_ts, 0);
+
+ /*
+ * If a DPB entry is unused or invalid, address of current destination
+ * buffer is returned.
+ */
+ if (buf_idx < 0)
+ return &dst->vb2_buf;
+
+ return vb2_get_buffer(cap_q, buf_idx);
+}
+
+static int tegra_vde_validate_vb_size(struct tegra_ctx *ctx,
+ struct vb2_buffer *vb,
+ unsigned int plane_id,
+ size_t min_size)
+{
+ u64 offset = vb->planes[plane_id].data_offset;
+ struct device *dev = ctx->vde->dev;
+
+ if (offset + min_size > vb2_plane_size(vb, plane_id)) {
+ dev_err(dev, "Too small plane[%u] size %lu @0x%llX, should be at least %zu\n",
+ plane_id, vb2_plane_size(vb, plane_id), offset, min_size);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int tegra_vde_h264_setup_frame(struct tegra_ctx *ctx,
+ struct tegra_vde_h264_decoder_ctx *h264,
+ struct v4l2_h264_reflist_builder *b,
+ struct vb2_buffer *vb,
+ unsigned int ref_id,
+ unsigned int id)
+{
+ struct v4l2_pix_format_mplane *pixfmt = &ctx->decoded_fmt.fmt.pix_mp;
+ struct tegra_m2m_buffer *tb = vb_to_tegra_buf(vb);
+ struct tegra_ctx_h264 *h = &ctx->h264;
+ struct tegra_vde *vde = ctx->vde;
+ struct device *dev = vde->dev;
+ unsigned int cstride, lstride;
+ unsigned int flags = 0;
+ size_t lsize, csize;
+ int err, frame_num;
+
+ lsize = h264->pic_width_in_mbs * 16 * h264->pic_height_in_mbs * 16;
+ csize = h264->pic_width_in_mbs * 8 * h264->pic_height_in_mbs * 8;
+ lstride = pixfmt->plane_fmt[0].bytesperline;
+ cstride = pixfmt->plane_fmt[1].bytesperline;
+
+ err = tegra_vde_validate_vb_size(ctx, vb, 0, lsize);
+ if (err)
+ return err;
+
+ err = tegra_vde_validate_vb_size(ctx, vb, 1, csize);
+ if (err)
+ return err;
+
+ err = tegra_vde_validate_vb_size(ctx, vb, 2, csize);
+ if (err)
+ return err;
+
+ if (!tb->aux || tb->aux->size < csize) {
+ dev_err(dev, "Too small aux size %zd, should be at least %zu\n",
+ tb->aux ? tb->aux->size : -1, csize);
+ return -EINVAL;
+ }
+
+ if (id == 0) {
+ frame_num = h->decode_params->frame_num;
+
+ if (h->decode_params->nal_ref_idc)
+ flags |= FLAG_REFERENCE;
+ } else {
+ frame_num = b->refs[ref_id].frame_num;
+ }
+
+ if (tb->b_frame)
+ flags |= FLAG_B_FRAME;
+
+ vde->frames[id].flags = flags;
+ vde->frames[id].y_addr = tb->dma_addr[0];
+ vde->frames[id].cb_addr = tb->dma_addr[1];
+ vde->frames[id].cr_addr = tb->dma_addr[2];
+ vde->frames[id].aux_addr = tb->aux->dma_addr;
+ vde->frames[id].frame_num = frame_num & 0x7fffff;
+ vde->frames[id].luma_atoms_pitch = lstride / VDE_ATOM;
+ vde->frames[id].chroma_atoms_pitch = cstride / VDE_ATOM;
+
+ return 0;
+}
+
+static int tegra_vde_h264_setup_frames(struct tegra_ctx *ctx,
+ struct tegra_vde_h264_decoder_ctx *h264)
+{
+ struct vb2_v4l2_buffer *src = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx);
+ struct vb2_v4l2_buffer *dst = v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx);
+ const struct v4l2_h264_dpb_entry *dpb = ctx->h264.decode_params->dpb;
+ struct tegra_m2m_buffer *tb = vb_to_tegra_buf(&dst->vb2_buf);
+ struct tegra_ctx_h264 *h = &ctx->h264;
+ struct v4l2_h264_reflist_builder b;
+ struct h264_reflists reflists;
+ struct vb2_buffer *ref;
+ unsigned int i;
+ u8 *dpb_id;
+ int err;
+
+ /*
+ * Tegra hardware requires information about frame's type, assuming
+ * that frame consists of the same type slices. Userspace must tag
+ * frame's type appropriately.
+ *
+ * Decoding of a non-uniform frames isn't supported by hardware and
+ * require software preprocessing that we don't implement. Decoding
+ * is expected to fail in this case. Such video streams are rare in
+ * practice, so not a big deal.
+ *
+ * If userspace doesn't tell us frame's type, then we will try decode
+ * as-is.
+ */
+ v4l2_m2m_buf_copy_metadata(src, dst, true);
+
+ if (h->decode_params->flags & V4L2_H264_DECODE_PARAM_FLAG_BFRAME)
+ tb->b_frame = true;
+ else
+ tb->b_frame = false;
+
+ err = tegra_vde_h264_setup_frame(ctx, h264, NULL, &dst->vb2_buf, 0,
+ h264->dpb_frames_nb++);
+ if (err)
+ return err;
+
+ if (!(h->decode_params->flags & (V4L2_H264_DECODE_PARAM_FLAG_PFRAME |
+ V4L2_H264_DECODE_PARAM_FLAG_BFRAME)))
+ return 0;
+
+ v4l2_h264_init_reflist_builder(&b, h->decode_params, h->sps, dpb);
+
+ if (h->decode_params->flags & V4L2_H264_DECODE_PARAM_FLAG_BFRAME) {
+ v4l2_h264_build_b_ref_lists(&b, reflists.b0, reflists.b1);
+ dpb_id = reflists.b0;
+ } else {
+ v4l2_h264_build_p_ref_list(&b, reflists.p);
+ dpb_id = reflists.p;
+ }
+
+ for (i = 0; i < b.num_valid; i++) {
+ ref = get_ref_buf(ctx, dst, dpb_id[i]);
+
+ err = tegra_vde_h264_setup_frame(ctx, h264, &b, ref, dpb_id[i],
+ h264->dpb_frames_nb++);
+ if (err)
+ return err;
+
+ if (b.refs[dpb_id[i]].pic_order_count < b.cur_pic_order_count)
+ h264->dpb_ref_frames_with_earlier_poc_nb++;
+ }
+
+ return 0;
+}
+
+static unsigned int to_tegra_vde_h264_level_idc(unsigned int level_idc)
+{
+ switch (level_idc) {
+ case 11:
+ return 2;
+ case 12:
+ return 3;
+ case 13:
+ return 4;
+ case 20:
+ return 5;
+ case 21:
+ return 6;
+ case 22:
+ return 7;
+ case 30:
+ return 8;
+ case 31:
+ return 9;
+ case 32:
+ return 10;
+ case 40:
+ return 11;
+ case 41:
+ return 12;
+ case 42:
+ return 13;
+ case 50:
+ return 14;
+ default:
+ break;
+ }
+
+ return 15;
+}
+
+static int tegra_vde_h264_setup_context(struct tegra_ctx *ctx,
+ struct tegra_vde_h264_decoder_ctx *h264)
+{
+ struct tegra_ctx_h264 *h = &ctx->h264;
+ struct tegra_vde *vde = ctx->vde;
+ struct device *dev = vde->dev;
+ int err;
+
+ memset(h264, 0, sizeof(*h264));
+ memset(vde->frames, 0, sizeof(vde->frames));
+
+ tegra_vde_prepare_control_data(ctx, V4L2_CID_STATELESS_H264_DECODE_PARAMS);
+ tegra_vde_prepare_control_data(ctx, V4L2_CID_STATELESS_H264_SPS);
+ tegra_vde_prepare_control_data(ctx, V4L2_CID_STATELESS_H264_PPS);
+
+ /* CABAC unsupported by hardware, requires software preprocessing */
+ if (h->pps->flags & V4L2_H264_PPS_FLAG_ENTROPY_CODING_MODE)
+ return -EOPNOTSUPP;
+
+ if (h->sps->profile_idc == 66)
+ h264->baseline_profile = 1;
+
+ if (h->sps->flags & V4L2_H264_SPS_FLAG_DIRECT_8X8_INFERENCE)
+ h264->direct_8x8_inference_flag = 1;
+
+ if (h->pps->flags & V4L2_H264_PPS_FLAG_CONSTRAINED_INTRA_PRED)
+ h264->constrained_intra_pred_flag = 1;
+
+ if (h->pps->flags & V4L2_H264_PPS_FLAG_DEBLOCKING_FILTER_CONTROL_PRESENT)
+ h264->deblocking_filter_control_present_flag = 1;
+
+ if (h->pps->flags & V4L2_H264_PPS_FLAG_BOTTOM_FIELD_PIC_ORDER_IN_FRAME_PRESENT)
+ h264->pic_order_present_flag = 1;
+
+ h264->level_idc = to_tegra_vde_h264_level_idc(h->sps->level_idc);
+ h264->log2_max_pic_order_cnt_lsb = h->sps->log2_max_pic_order_cnt_lsb_minus4 + 4;
+ h264->log2_max_frame_num = h->sps->log2_max_frame_num_minus4 + 4;
+ h264->pic_order_cnt_type = h->sps->pic_order_cnt_type;
+ h264->pic_width_in_mbs = h->sps->pic_width_in_mbs_minus1 + 1;
+ h264->pic_height_in_mbs = h->sps->pic_height_in_map_units_minus1 + 1;
+
+ h264->num_ref_idx_l0_active_minus1 = h->pps->num_ref_idx_l0_default_active_minus1;
+ h264->num_ref_idx_l1_active_minus1 = h->pps->num_ref_idx_l1_default_active_minus1;
+ h264->chroma_qp_index_offset = h->pps->chroma_qp_index_offset & 0x1f;
+ h264->pic_init_qp = h->pps->pic_init_qp_minus26 + 26;
+
+ err = tegra_vde_h264_setup_frames(ctx, h264);
+ if (err)
+ return err;
+
+ err = tegra_vde_validate_h264_ctx(dev, h264);
+ if (err)
+ return err;
+
+ return 0;
+}
+
+int tegra_vde_h264_decode_run(struct tegra_ctx *ctx)
+{
+ struct vb2_v4l2_buffer *src = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx);
+ struct tegra_m2m_buffer *bitstream = vb_to_tegra_buf(&src->vb2_buf);
+ size_t bitstream_size = vb2_get_plane_payload(&src->vb2_buf, 0);
+ struct tegra_vde_h264_decoder_ctx h264;
+ struct tegra_vde *vde = ctx->vde;
+ int err;
+
+ err = tegra_vde_h264_setup_context(ctx, &h264);
+ if (err)
+ return err;
+
+ err = tegra_vde_decode_begin(vde, &h264, vde->frames,
+ bitstream->dma_addr[0],
+ bitstream_size);
+ if (err)
+ return err;
+
+ return 0;
+}
+
+int tegra_vde_h264_decode_wait(struct tegra_ctx *ctx)
+{
+ return tegra_vde_decode_end(ctx->vde);
+}
diff --git a/drivers/staging/media/tegra-vde/v4l2.c b/drivers/staging/media/tegra-vde/v4l2.c
new file mode 100644
index 000000000000..bd8c207d5b54
--- /dev/null
+++ b/drivers/staging/media/tegra-vde/v4l2.c
@@ -0,0 +1,1018 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * NVIDIA Tegra Video decoder driver
+ *
+ * Copyright (C) 2019-2022 Dmitry Osipenko <digetx@gmail.com>
+ *
+ * Based on Cedrus driver by Bootlin.
+ * Copyright (C) 2016 Florent Revest <florent.revest@free-electrons.com>
+ * Copyright (C) 2018 Paul Kocialkowski <paul.kocialkowski@bootlin.com>
+ *
+ * Based on Rockchip driver by Collabora.
+ * Copyright (C) 2019 Boris Brezillon <boris.brezillon@collabora.com>
+ */
+
+#include <linux/err.h>
+#include <linux/slab.h>
+
+#include "vde.h"
+
+static const struct v4l2_ctrl_config ctrl_cfgs[] = {
+ { .id = V4L2_CID_STATELESS_H264_DECODE_PARAMS, },
+ { .id = V4L2_CID_STATELESS_H264_SPS, },
+ { .id = V4L2_CID_STATELESS_H264_PPS, },
+ {
+ .id = V4L2_CID_STATELESS_H264_DECODE_MODE,
+ .min = V4L2_STATELESS_H264_DECODE_MODE_FRAME_BASED,
+ .max = V4L2_STATELESS_H264_DECODE_MODE_FRAME_BASED,
+ .def = V4L2_STATELESS_H264_DECODE_MODE_FRAME_BASED,
+ },
+ {
+ .id = V4L2_CID_STATELESS_H264_START_CODE,
+ .min = V4L2_STATELESS_H264_START_CODE_ANNEX_B,
+ .max = V4L2_STATELESS_H264_START_CODE_ANNEX_B,
+ .def = V4L2_STATELESS_H264_START_CODE_ANNEX_B,
+ },
+ {
+ .id = V4L2_CID_MPEG_VIDEO_H264_PROFILE,
+ .min = V4L2_MPEG_VIDEO_H264_PROFILE_BASELINE,
+ .max = V4L2_MPEG_VIDEO_H264_PROFILE_MAIN,
+ .def = V4L2_MPEG_VIDEO_H264_PROFILE_MAIN,
+ },
+ {
+ .id = V4L2_CID_MPEG_VIDEO_H264_LEVEL,
+ .min = V4L2_MPEG_VIDEO_H264_LEVEL_1_0,
+ .max = V4L2_MPEG_VIDEO_H264_LEVEL_5_1,
+ },
+};
+
+static inline struct tegra_ctx *fh_to_tegra_ctx(struct v4l2_fh *fh)
+{
+ return container_of(fh, struct tegra_ctx, fh);
+}
+
+static void tegra_set_control_data(struct tegra_ctx *ctx, void *data, u32 id)
+{
+ switch (id) {
+ case V4L2_CID_STATELESS_H264_DECODE_PARAMS:
+ ctx->h264.decode_params = data;
+ break;
+ case V4L2_CID_STATELESS_H264_SPS:
+ ctx->h264.sps = data;
+ break;
+ case V4L2_CID_STATELESS_H264_PPS:
+ ctx->h264.pps = data;
+ break;
+ }
+}
+
+void tegra_vde_prepare_control_data(struct tegra_ctx *ctx, u32 id)
+{
+ unsigned int i;
+
+ for (i = 0; i < ARRAY_SIZE(ctrl_cfgs); i++) {
+ if (ctx->ctrls[i]->id == id) {
+ tegra_set_control_data(ctx, ctx->ctrls[i]->p_cur.p, id);
+ return;
+ }
+ }
+
+ tegra_set_control_data(ctx, NULL, id);
+}
+
+static int tegra_queue_setup(struct vb2_queue *vq,
+ unsigned int *nbufs,
+ unsigned int *num_planes,
+ unsigned int sizes[],
+ struct device *alloc_devs[])
+{
+ struct tegra_ctx *ctx = vb2_get_drv_priv(vq);
+ struct v4l2_format *f;
+ unsigned int i;
+
+ if (V4L2_TYPE_IS_OUTPUT(vq->type))
+ f = &ctx->coded_fmt;
+ else
+ f = &ctx->decoded_fmt;
+
+ if (*num_planes) {
+ if (*num_planes != f->fmt.pix_mp.num_planes)
+ return -EINVAL;
+
+ for (i = 0; i < f->fmt.pix_mp.num_planes; i++) {
+ if (sizes[i] < f->fmt.pix_mp.plane_fmt[i].sizeimage)
+ return -EINVAL;
+ }
+ } else {
+ *num_planes = f->fmt.pix_mp.num_planes;
+
+ for (i = 0; i < f->fmt.pix_mp.num_planes; i++)
+ sizes[i] = f->fmt.pix_mp.plane_fmt[i].sizeimage;
+ }
+
+ return 0;
+}
+
+static int tegra_buf_out_validate(struct vb2_buffer *vb)
+{
+ struct vb2_v4l2_buffer *vbuf = to_vb2_v4l2_buffer(vb);
+
+ vbuf->field = V4L2_FIELD_NONE;
+ return 0;
+}
+
+static void __tegra_buf_cleanup(struct vb2_buffer *vb, unsigned int i)
+{
+ struct vb2_queue *vq = vb->vb2_queue;
+ struct tegra_ctx *ctx = vb2_get_drv_priv(vq);
+ struct tegra_m2m_buffer *tb = vb_to_tegra_buf(vb);
+
+ while (i--) {
+ if (tb->a[i]) {
+ tegra_vde_dmabuf_cache_unmap(ctx->vde, tb->a[i], true);
+ tb->a[i] = NULL;
+ }
+
+ if (tb->iova[i]) {
+ tegra_vde_iommu_unmap(ctx->vde, tb->iova[i]);
+ tb->iova[i] = NULL;
+ }
+ }
+
+ if (tb->aux) {
+ tegra_vde_free_bo(tb->aux);
+ tb->aux = NULL;
+ }
+}
+
+static int tegra_buf_init(struct vb2_buffer *vb)
+{
+ struct vb2_queue *vq = vb->vb2_queue;
+ struct tegra_ctx *ctx = vb2_get_drv_priv(vq);
+ struct tegra_m2m_buffer *tb = vb_to_tegra_buf(vb);
+ struct tegra_vde *vde = ctx->vde;
+ enum dma_data_direction dma_dir;
+ struct sg_table *sgt;
+ unsigned int i;
+ int err;
+
+ if (V4L2_TYPE_IS_CAPTURE(vq->type) && vb->num_planes > 1) {
+ /*
+ * Tegra decoder writes auxiliary data for I/P frames.
+ * This data is needed for decoding of B frames.
+ */
+ err = tegra_vde_alloc_bo(vde, &tb->aux, DMA_FROM_DEVICE,
+ vb2_plane_size(vb, 1));
+ if (err)
+ return err;
+ }
+
+ if (V4L2_TYPE_IS_OUTPUT(vq->type))
+ dma_dir = DMA_TO_DEVICE;
+ else
+ dma_dir = DMA_FROM_DEVICE;
+
+ for (i = 0; i < vb->num_planes; i++) {
+ if (vq->memory == VB2_MEMORY_DMABUF) {
+ get_dma_buf(vb->planes[i].dbuf);
+
+ err = tegra_vde_dmabuf_cache_map(vde, vb->planes[i].dbuf,
+ dma_dir, &tb->a[i],
+ &tb->dma_base[i]);
+ if (err) {
+ dma_buf_put(vb->planes[i].dbuf);
+ goto cleanup;
+ }
+
+ continue;
+ }
+
+ if (vde->domain) {
+ sgt = vb2_dma_sg_plane_desc(vb, i);
+
+ err = tegra_vde_iommu_map(vde, sgt, &tb->iova[i],
+ vb2_plane_size(vb, i));
+ if (err)
+ goto cleanup;
+
+ tb->dma_base[i] = iova_dma_addr(&vde->iova, tb->iova[i]);
+ } else {
+ tb->dma_base[i] = vb2_dma_contig_plane_dma_addr(vb, i);
+ }
+ }
+
+ return 0;
+
+cleanup:
+ __tegra_buf_cleanup(vb, i);
+
+ return err;
+}
+
+static void tegra_buf_cleanup(struct vb2_buffer *vb)
+{
+ __tegra_buf_cleanup(vb, vb->num_planes);
+}
+
+static int tegra_buf_prepare(struct vb2_buffer *vb)
+{
+ struct vb2_queue *vq = vb->vb2_queue;
+ struct tegra_ctx *ctx = vb2_get_drv_priv(vq);
+ struct tegra_m2m_buffer *tb = vb_to_tegra_buf(vb);
+ size_t hw_align, hw_size, hw_payload, size, offset;
+ struct v4l2_pix_format_mplane *pixfmt;
+ unsigned int i;
+ void *vb_data;
+
+ if (V4L2_TYPE_IS_OUTPUT(vq->type)) {
+ hw_align = BSEV_ALIGN;
+ pixfmt = &ctx->coded_fmt.fmt.pix_mp;
+ } else {
+ hw_align = FRAMEID_ALIGN;
+ pixfmt = &ctx->decoded_fmt.fmt.pix_mp;
+ }
+
+ for (i = 0; i < vb->num_planes; i++) {
+ offset = vb->planes[i].data_offset;
+
+ if (offset & (hw_align - 1))
+ return -EINVAL;
+
+ if (V4L2_TYPE_IS_CAPTURE(vq->type)) {
+ size = pixfmt->plane_fmt[i].sizeimage;
+ hw_payload = ALIGN(size, VDE_ATOM);
+ } else {
+ size = vb2_get_plane_payload(vb, i) - offset;
+ hw_payload = ALIGN(size + VDE_ATOM, SXE_BUFFER);
+ }
+
+ hw_size = offset + hw_payload;
+
+ if (vb2_plane_size(vb, i) < hw_size)
+ return -EINVAL;
+
+ vb2_set_plane_payload(vb, i, hw_payload);
+
+ if (V4L2_TYPE_IS_OUTPUT(vq->type)) {
+ vb_data = vb2_plane_vaddr(vb, i);
+
+ /*
+ * Hardware requires zero-padding of coded data.
+ * Otherwise it will fail to parse the trailing
+ * data and abort the decoding.
+ */
+ if (vb_data)
+ memset(vb_data + offset + size, 0,
+ hw_size - offset - size);
+ }
+
+ tb->dma_addr[i] = tb->dma_base[i] + offset;
+ }
+
+ switch (pixfmt->pixelformat) {
+ case V4L2_PIX_FMT_YVU420M:
+ swap(tb->dma_addr[1], tb->dma_addr[2]);
+ break;
+ }
+
+ return 0;
+}
+
+static void tegra_buf_queue(struct vb2_buffer *vb)
+{
+ struct tegra_ctx *ctx = vb2_get_drv_priv(vb->vb2_queue);
+ struct vb2_v4l2_buffer *vbuf = to_vb2_v4l2_buffer(vb);
+
+ v4l2_m2m_buf_queue(ctx->fh.m2m_ctx, vbuf);
+}
+
+static void tegra_buf_request_complete(struct vb2_buffer *vb)
+{
+ struct tegra_ctx *ctx = vb2_get_drv_priv(vb->vb2_queue);
+
+ v4l2_ctrl_request_complete(vb->req_obj.req, &ctx->hdl);
+}
+
+static int tegra_start_streaming(struct vb2_queue *vq, unsigned int count)
+{
+ return 0;
+}
+
+static void tegra_stop_streaming(struct vb2_queue *vq)
+{
+ struct tegra_ctx *ctx = vb2_get_drv_priv(vq);
+
+ while (true) {
+ struct vb2_v4l2_buffer *vbuf;
+
+ if (V4L2_TYPE_IS_OUTPUT(vq->type))
+ vbuf = v4l2_m2m_src_buf_remove(ctx->fh.m2m_ctx);
+ else
+ vbuf = v4l2_m2m_dst_buf_remove(ctx->fh.m2m_ctx);
+
+ if (!vbuf)
+ break;
+
+ v4l2_ctrl_request_complete(vbuf->vb2_buf.req_obj.req, &ctx->hdl);
+ v4l2_m2m_buf_done(vbuf, VB2_BUF_STATE_ERROR);
+ }
+}
+
+static const struct vb2_ops tegra_qops = {
+ .queue_setup = tegra_queue_setup,
+ .buf_init = tegra_buf_init,
+ .buf_cleanup = tegra_buf_cleanup,
+ .buf_prepare = tegra_buf_prepare,
+ .buf_queue = tegra_buf_queue,
+ .buf_out_validate = tegra_buf_out_validate,
+ .buf_request_complete = tegra_buf_request_complete,
+ .start_streaming = tegra_start_streaming,
+ .stop_streaming = tegra_stop_streaming,
+ .wait_prepare = vb2_ops_wait_prepare,
+ .wait_finish = vb2_ops_wait_finish,
+};
+
+static int tegra_queue_init(void *priv,
+ struct vb2_queue *src_vq,
+ struct vb2_queue *dst_vq)
+{
+ struct tegra_ctx *ctx = priv;
+ struct tegra_vde *vde = ctx->vde;
+ const struct vb2_mem_ops *mem_ops;
+ unsigned long dma_attrs;
+ int err;
+
+ /*
+ * TODO: Switch to use of vb2_dma_contig_memops uniformly once we
+ * will add IOMMU_DOMAIN support for video decoder to tegra-smmu
+ * driver. For now we need to stick with SG ops in order to be able
+ * to get SGT table easily. This is suboptimal since SG mappings are
+ * wasting CPU cache and we don't need that caching.
+ */
+ if (vde->domain)
+ mem_ops = &vb2_dma_sg_memops;
+ else
+ mem_ops = &vb2_dma_contig_memops;
+
+ dma_attrs = DMA_ATTR_WRITE_COMBINE;
+
+ src_vq->buf_struct_size = sizeof(struct tegra_m2m_buffer);
+ src_vq->timestamp_flags = V4L2_BUF_FLAG_TIMESTAMP_COPY;
+ src_vq->type = V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE;
+ src_vq->io_modes = VB2_DMABUF | VB2_MMAP;
+ src_vq->supports_requests = true;
+ src_vq->requires_requests = true;
+ src_vq->lock = &vde->v4l2_lock;
+ src_vq->dma_attrs = dma_attrs;
+ src_vq->mem_ops = mem_ops;
+ src_vq->ops = &tegra_qops;
+ src_vq->drv_priv = ctx;
+ src_vq->dev = vde->dev;
+
+ err = vb2_queue_init(src_vq);
+ if (err) {
+ v4l2_err(&vde->v4l2_dev,
+ "failed to initialize src queue: %d\n", err);
+ return err;
+ }
+
+ /*
+ * We may need to zero the end of bitstream in kernel if userspace
+ * doesn't do that, hence kmap is needed for the coded data. It's not
+ * needed for framebuffers.
+ */
+ dma_attrs |= DMA_ATTR_NO_KERNEL_MAPPING;
+
+ dst_vq->buf_struct_size = sizeof(struct tegra_m2m_buffer);
+ dst_vq->timestamp_flags = V4L2_BUF_FLAG_TIMESTAMP_COPY;
+ dst_vq->type = V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE;
+ dst_vq->io_modes = VB2_DMABUF | VB2_MMAP;
+ dst_vq->lock = &vde->v4l2_lock;
+ dst_vq->dma_attrs = dma_attrs;
+ dst_vq->mem_ops = mem_ops;
+ dst_vq->ops = &tegra_qops;
+ dst_vq->drv_priv = ctx;
+ dst_vq->dev = vde->dev;
+
+ err = vb2_queue_init(dst_vq);
+ if (err) {
+ v4l2_err(&vde->v4l2_dev,
+ "failed to initialize dst queue: %d\n", err);
+ return err;
+ }
+
+ return 0;
+}
+
+static void tegra_reset_fmt(struct tegra_ctx *ctx, struct v4l2_format *f,
+ u32 fourcc)
+{
+ memset(f, 0, sizeof(*f));
+ f->fmt.pix_mp.pixelformat = fourcc;
+ f->fmt.pix_mp.field = V4L2_FIELD_NONE;
+ f->fmt.pix_mp.xfer_func = V4L2_XFER_FUNC_DEFAULT;
+ f->fmt.pix_mp.ycbcr_enc = V4L2_YCBCR_ENC_DEFAULT;
+ f->fmt.pix_mp.colorspace = V4L2_COLORSPACE_REC709;
+ f->fmt.pix_mp.quantization = V4L2_QUANTIZATION_DEFAULT;
+}
+
+static void tegra_reset_coded_fmt(struct tegra_ctx *ctx)
+{
+ const struct tegra_vde_soc *soc = ctx->vde->soc;
+ struct v4l2_format *f = &ctx->coded_fmt;
+
+ ctx->coded_fmt_desc = &soc->coded_fmts[0];
+ tegra_reset_fmt(ctx, f, ctx->coded_fmt_desc->fourcc);
+
+ f->type = V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE;
+ f->fmt.pix_mp.width = ctx->coded_fmt_desc->frmsize.min_width;
+ f->fmt.pix_mp.height = ctx->coded_fmt_desc->frmsize.min_height;
+}
+
+static void tegra_fill_pixfmt_mp(struct v4l2_pix_format_mplane *pixfmt,
+ u32 pixelformat, u32 width, u32 height)
+{
+ const struct v4l2_format_info *info = v4l2_format_info(pixelformat);
+ struct v4l2_plane_pix_format *plane;
+ unsigned int i;
+
+ switch (pixelformat) {
+ case V4L2_PIX_FMT_YUV420M:
+ case V4L2_PIX_FMT_YVU420M:
+ pixfmt->width = width;
+ pixfmt->height = height;
+ pixfmt->pixelformat = pixelformat;
+ pixfmt->num_planes = info->mem_planes;
+
+ for (i = 0; i < pixfmt->num_planes; i++) {
+ unsigned int hdiv = (i == 0) ? 1 : 2;
+ unsigned int vdiv = (i == 0) ? 1 : 2;
+
+ /*
+ * VDE is connected to Graphics Memory using 128bit port,
+ * all memory accesses are made using 16B atoms.
+ *
+ * V4L requires Cb/Cr strides to be exactly half of the
+ * Y stride, hence we're aligning Y to 16B x 2.
+ */
+ plane = &pixfmt->plane_fmt[i];
+ plane->bytesperline = ALIGN(width, VDE_ATOM * 2) / hdiv;
+ plane->sizeimage = plane->bytesperline * height / vdiv;
+ }
+
+ break;
+ }
+}
+
+static void tegra_reset_decoded_fmt(struct tegra_ctx *ctx)
+{
+ struct v4l2_format *f = &ctx->decoded_fmt;
+
+ tegra_reset_fmt(ctx, f, ctx->coded_fmt_desc->decoded_fmts[0]);
+ f->type = V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE;
+ tegra_fill_pixfmt_mp(&f->fmt.pix_mp,
+ ctx->coded_fmt_desc->decoded_fmts[0],
+ ctx->coded_fmt.fmt.pix_mp.width,
+ ctx->coded_fmt.fmt.pix_mp.height);
+}
+
+static void tegra_job_finish(struct tegra_ctx *ctx,
+ enum vb2_buffer_state result)
+{
+ v4l2_m2m_buf_done_and_job_finish(ctx->vde->m2m, ctx->fh.m2m_ctx,
+ result);
+}
+
+static void tegra_decode_complete(struct work_struct *work)
+{
+ struct tegra_ctx *ctx = container_of(work, struct tegra_ctx, work);
+ int err;
+
+ err = ctx->coded_fmt_desc->decode_wait(ctx);
+ if (err)
+ tegra_job_finish(ctx, VB2_BUF_STATE_ERROR);
+ else
+ tegra_job_finish(ctx, VB2_BUF_STATE_DONE);
+}
+
+static int tegra_querycap(struct file *file, void *priv,
+ struct v4l2_capability *cap)
+{
+ strscpy(cap->bus_info, "platform:tegra-vde", sizeof(cap->bus_info));
+ strscpy(cap->driver, "tegra-vde", sizeof(cap->driver));
+ strscpy(cap->card, "tegra-vde", sizeof(cap->card));
+
+ return 0;
+}
+
+static int tegra_enum_decoded_fmt(struct file *file, void *priv,
+ struct v4l2_fmtdesc *f)
+{
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(priv);
+
+ if (WARN_ON(!ctx->coded_fmt_desc))
+ return -EINVAL;
+
+ if (f->index >= ctx->coded_fmt_desc->num_decoded_fmts)
+ return -EINVAL;
+
+ f->pixelformat = ctx->coded_fmt_desc->decoded_fmts[f->index];
+
+ return 0;
+}
+
+static int tegra_g_decoded_fmt(struct file *file, void *priv,
+ struct v4l2_format *f)
+{
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(priv);
+
+ *f = ctx->decoded_fmt;
+ return 0;
+}
+
+static int tegra_try_decoded_fmt(struct file *file, void *priv,
+ struct v4l2_format *f)
+{
+ struct v4l2_pix_format_mplane *pix_mp = &f->fmt.pix_mp;
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(priv);
+ const struct tegra_coded_fmt_desc *coded_desc;
+ unsigned int i;
+
+ /*
+ * The codec context should point to a coded format desc, if the format
+ * on the coded end has not been set yet, it should point to the
+ * default value.
+ */
+ coded_desc = ctx->coded_fmt_desc;
+ if (WARN_ON(!coded_desc))
+ return -EINVAL;
+
+ if (!coded_desc->num_decoded_fmts)
+ return -EINVAL;
+
+ for (i = 0; i < coded_desc->num_decoded_fmts; i++) {
+ if (coded_desc->decoded_fmts[i] == pix_mp->pixelformat)
+ break;
+ }
+
+ if (i == coded_desc->num_decoded_fmts)
+ pix_mp->pixelformat = coded_desc->decoded_fmts[0];
+
+ /* always apply the frmsize constraint of the coded end */
+ v4l2_apply_frmsize_constraints(&pix_mp->width,
+ &pix_mp->height,
+ &coded_desc->frmsize);
+
+ tegra_fill_pixfmt_mp(pix_mp, pix_mp->pixelformat,
+ pix_mp->width, pix_mp->height);
+ pix_mp->field = V4L2_FIELD_NONE;
+
+ return 0;
+}
+
+static int tegra_s_decoded_fmt(struct file *file, void *priv,
+ struct v4l2_format *f)
+{
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(priv);
+ struct vb2_queue *vq;
+ int err;
+
+ /* change not allowed if queue is busy */
+ vq = v4l2_m2m_get_vq(ctx->fh.m2m_ctx,
+ V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
+ if (vb2_is_busy(vq))
+ return -EBUSY;
+
+ err = tegra_try_decoded_fmt(file, priv, f);
+ if (err)
+ return err;
+
+ ctx->decoded_fmt = *f;
+
+ return 0;
+}
+
+static int tegra_enum_coded_fmt(struct file *file, void *priv,
+ struct v4l2_fmtdesc *f)
+{
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(priv);
+ const struct tegra_vde_soc *soc = ctx->vde->soc;
+
+ if (f->index >= soc->num_coded_fmts)
+ return -EINVAL;
+
+ f->pixelformat = soc->coded_fmts[f->index].fourcc;
+
+ return 0;
+}
+
+static int tegra_g_coded_fmt(struct file *file, void *priv,
+ struct v4l2_format *f)
+{
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(priv);
+
+ *f = ctx->coded_fmt;
+ return 0;
+}
+
+static const struct tegra_coded_fmt_desc *
+tegra_find_coded_fmt_desc(struct tegra_ctx *ctx, u32 fourcc)
+{
+ const struct tegra_vde_soc *soc = ctx->vde->soc;
+ unsigned int i;
+
+ for (i = 0; i < soc->num_coded_fmts; i++) {
+ if (soc->coded_fmts[i].fourcc == fourcc)
+ return &soc->coded_fmts[i];
+ }
+
+ return NULL;
+}
+
+static int tegra_try_coded_fmt(struct file *file, void *priv,
+ struct v4l2_format *f)
+{
+ struct v4l2_pix_format_mplane *pix_mp = &f->fmt.pix_mp;
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(priv);
+ const struct tegra_vde_soc *soc = ctx->vde->soc;
+ int size = pix_mp->plane_fmt[0].sizeimage;
+ const struct tegra_coded_fmt_desc *desc;
+
+ desc = tegra_find_coded_fmt_desc(ctx, pix_mp->pixelformat);
+ if (!desc) {
+ pix_mp->pixelformat = soc->coded_fmts[0].fourcc;
+ desc = &soc->coded_fmts[0];
+ }
+
+ v4l2_apply_frmsize_constraints(&pix_mp->width,
+ &pix_mp->height,
+ &desc->frmsize);
+
+ pix_mp->plane_fmt[0].sizeimage = max(ALIGN(size, SXE_BUFFER), SZ_2M);
+ pix_mp->field = V4L2_FIELD_NONE;
+ pix_mp->num_planes = 1;
+
+ return 0;
+}
+
+static int tegra_s_coded_fmt(struct file *file, void *priv,
+ struct v4l2_format *f)
+{
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(priv);
+ struct v4l2_m2m_ctx *m2m_ctx = ctx->fh.m2m_ctx;
+ const struct tegra_coded_fmt_desc *desc;
+ struct vb2_queue *peer_vq, *vq;
+ struct v4l2_format *cap_fmt;
+ int err;
+
+ /*
+ * In order to support dynamic resolution change, the decoder admits
+ * a resolution change, as long as the pixelformat remains. Can't be
+ * done if streaming.
+ */
+ vq = v4l2_m2m_get_vq(m2m_ctx, V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE);
+ if (vb2_is_streaming(vq) ||
+ (vb2_is_busy(vq) &&
+ f->fmt.pix_mp.pixelformat != ctx->coded_fmt.fmt.pix_mp.pixelformat))
+ return -EBUSY;
+
+ /*
+ * Since format change on the OUTPUT queue will reset the CAPTURE
+ * queue, we can't allow doing so when the CAPTURE queue has buffers
+ * allocated.
+ */
+ peer_vq = v4l2_m2m_get_vq(m2m_ctx, V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
+ if (vb2_is_busy(peer_vq))
+ return -EBUSY;
+
+ err = tegra_try_coded_fmt(file, priv, f);
+ if (err)
+ return err;
+
+ desc = tegra_find_coded_fmt_desc(ctx, f->fmt.pix_mp.pixelformat);
+ if (!desc)
+ return -EINVAL;
+
+ ctx->coded_fmt_desc = desc;
+ ctx->coded_fmt = *f;
+
+ /*
+ * Current decoded format might have become invalid with newly
+ * selected codec, so reset it to default just to be safe and
+ * keep internal driver state sane. User is mandated to set
+ * the decoded format again after we return, so we don't need
+ * anything smarter.
+ *
+ * Note that this will propagates any size changes to the decoded format.
+ */
+ tegra_reset_decoded_fmt(ctx);
+
+ /* propagate colorspace information to capture */
+ cap_fmt = &ctx->decoded_fmt;
+ cap_fmt->fmt.pix_mp.xfer_func = f->fmt.pix_mp.xfer_func;
+ cap_fmt->fmt.pix_mp.ycbcr_enc = f->fmt.pix_mp.ycbcr_enc;
+ cap_fmt->fmt.pix_mp.colorspace = f->fmt.pix_mp.colorspace;
+ cap_fmt->fmt.pix_mp.quantization = f->fmt.pix_mp.quantization;
+
+ return 0;
+}
+
+static int tegra_enum_framesizes(struct file *file, void *priv,
+ struct v4l2_frmsizeenum *fsize)
+{
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(priv);
+ const struct tegra_coded_fmt_desc *fmt;
+
+ if (fsize->index)
+ return -EINVAL;
+
+ fmt = tegra_find_coded_fmt_desc(ctx, fsize->pixel_format);
+ if (!fmt)
+ return -EINVAL;
+
+ fsize->type = V4L2_FRMSIZE_TYPE_STEPWISE;
+ fsize->stepwise = fmt->frmsize;
+
+ return 0;
+}
+
+static const struct v4l2_ioctl_ops tegra_v4l2_ioctl_ops = {
+ .vidioc_querycap = tegra_querycap,
+ .vidioc_enum_framesizes = tegra_enum_framesizes,
+
+ .vidioc_try_fmt_vid_out_mplane = tegra_try_coded_fmt,
+ .vidioc_g_fmt_vid_out_mplane = tegra_g_coded_fmt,
+ .vidioc_s_fmt_vid_out_mplane = tegra_s_coded_fmt,
+ .vidioc_enum_fmt_vid_out = tegra_enum_coded_fmt,
+
+ .vidioc_try_fmt_vid_cap_mplane = tegra_try_decoded_fmt,
+ .vidioc_g_fmt_vid_cap_mplane = tegra_g_decoded_fmt,
+ .vidioc_s_fmt_vid_cap_mplane = tegra_s_decoded_fmt,
+ .vidioc_enum_fmt_vid_cap = tegra_enum_decoded_fmt,
+
+ .vidioc_reqbufs = v4l2_m2m_ioctl_reqbufs,
+ .vidioc_querybuf = v4l2_m2m_ioctl_querybuf,
+ .vidioc_qbuf = v4l2_m2m_ioctl_qbuf,
+ .vidioc_dqbuf = v4l2_m2m_ioctl_dqbuf,
+ .vidioc_prepare_buf = v4l2_m2m_ioctl_prepare_buf,
+ .vidioc_create_bufs = v4l2_m2m_ioctl_create_bufs,
+ .vidioc_expbuf = v4l2_m2m_ioctl_expbuf,
+
+ .vidioc_streamon = v4l2_m2m_ioctl_streamon,
+ .vidioc_streamoff = v4l2_m2m_ioctl_streamoff,
+
+ .vidioc_subscribe_event = v4l2_ctrl_subscribe_event,
+ .vidioc_unsubscribe_event = v4l2_event_unsubscribe,
+};
+
+static int tegra_init_ctrls(struct tegra_ctx *ctx)
+{
+ unsigned int i;
+ int err;
+
+ err = v4l2_ctrl_handler_init(&ctx->hdl, ARRAY_SIZE(ctrl_cfgs));
+ if (err)
+ return err;
+
+ for (i = 0; i < ARRAY_SIZE(ctrl_cfgs); i++) {
+ ctx->ctrls[i] = v4l2_ctrl_new_custom(&ctx->hdl, &ctrl_cfgs[i],
+ NULL);
+ if (ctx->hdl.error) {
+ err = ctx->hdl.error;
+ goto free_ctrls;
+ }
+ }
+
+ err = v4l2_ctrl_handler_setup(&ctx->hdl);
+ if (err)
+ goto free_ctrls;
+
+ ctx->fh.ctrl_handler = &ctx->hdl;
+
+ return 0;
+
+free_ctrls:
+ v4l2_ctrl_handler_free(&ctx->hdl);
+
+ return err;
+}
+
+static int tegra_init_m2m(struct tegra_ctx *ctx)
+{
+ ctx->fh.m2m_ctx = v4l2_m2m_ctx_init(ctx->vde->m2m,
+ ctx, tegra_queue_init);
+ if (IS_ERR(ctx->fh.m2m_ctx))
+ return PTR_ERR(ctx->fh.m2m_ctx);
+
+ return 0;
+}
+
+static int tegra_open(struct file *file)
+{
+ struct tegra_vde *vde = video_drvdata(file);
+ struct tegra_ctx *ctx;
+ int err;
+
+ ctx = kzalloc(offsetof(struct tegra_ctx, ctrls[ARRAY_SIZE(ctrl_cfgs)]),
+ GFP_KERNEL);
+ if (!ctx)
+ return -ENOMEM;
+
+ ctx->vde = vde;
+ v4l2_fh_init(&ctx->fh, video_devdata(file));
+ INIT_WORK(&ctx->work, tegra_decode_complete);
+
+ err = tegra_init_ctrls(ctx);
+ if (err) {
+ v4l2_err(&vde->v4l2_dev, "failed to add controls: %d\n", err);
+ goto free_ctx;
+ }
+
+ err = tegra_init_m2m(ctx);
+ if (err) {
+ v4l2_err(&vde->v4l2_dev, "failed to initialize m2m: %d\n", err);
+ goto free_ctrls;
+ }
+
+ file->private_data = &ctx->fh;
+ v4l2_fh_add(&ctx->fh);
+
+ tegra_reset_coded_fmt(ctx);
+ tegra_try_coded_fmt(file, file->private_data, &ctx->coded_fmt);
+
+ tegra_reset_decoded_fmt(ctx);
+ tegra_try_decoded_fmt(file, file->private_data, &ctx->decoded_fmt);
+
+ return 0;
+
+free_ctrls:
+ v4l2_ctrl_handler_free(&ctx->hdl);
+free_ctx:
+ kfree(ctx);
+
+ return err;
+}
+
+static int tegra_release(struct file *file)
+{
+ struct v4l2_fh *fh = file->private_data;
+ struct tegra_ctx *ctx = fh_to_tegra_ctx(fh);
+ struct tegra_vde *vde = ctx->vde;
+
+ v4l2_fh_del(fh);
+ v4l2_m2m_ctx_release(fh->m2m_ctx);
+ v4l2_ctrl_handler_free(&ctx->hdl);
+ v4l2_fh_exit(fh);
+ kfree(ctx);
+
+ tegra_vde_dmabuf_cache_unmap_sync(vde);
+
+ return 0;
+}
+
+static const struct v4l2_file_operations tegra_v4l2_fops = {
+ .owner = THIS_MODULE,
+ .open = tegra_open,
+ .poll = v4l2_m2m_fop_poll,
+ .mmap = v4l2_m2m_fop_mmap,
+ .release = tegra_release,
+ .unlocked_ioctl = video_ioctl2,
+};
+
+static void tegra_device_run(void *priv)
+{
+ struct tegra_ctx *ctx = priv;
+ struct vb2_v4l2_buffer *src = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx);
+ struct media_request *src_req = src->vb2_buf.req_obj.req;
+ int err;
+
+ v4l2_ctrl_request_setup(src_req, &ctx->hdl);
+
+ err = ctx->coded_fmt_desc->decode_run(ctx);
+
+ v4l2_ctrl_request_complete(src_req, &ctx->hdl);
+
+ if (err)
+ tegra_job_finish(ctx, VB2_BUF_STATE_ERROR);
+ else
+ queue_work(ctx->vde->wq, &ctx->work);
+}
+
+static const struct v4l2_m2m_ops tegra_v4l2_m2m_ops = {
+ .device_run = tegra_device_run,
+};
+
+static int tegra_request_validate(struct media_request *req)
+{
+ unsigned int count;
+
+ count = vb2_request_buffer_cnt(req);
+ if (!count)
+ return -ENOENT;
+ else if (count > 1)
+ return -EINVAL;
+
+ return vb2_request_validate(req);
+}
+
+static const struct media_device_ops tegra_media_device_ops = {
+ .req_validate = tegra_request_validate,
+ .req_queue = v4l2_m2m_request_queue,
+};
+
+int tegra_vde_v4l2_init(struct tegra_vde *vde)
+{
+ struct device *dev = vde->dev;
+ int err;
+
+ mutex_init(&vde->v4l2_lock);
+ media_device_init(&vde->mdev);
+ video_set_drvdata(&vde->vdev, vde);
+
+ vde->vdev.lock = &vde->v4l2_lock,
+ vde->vdev.fops = &tegra_v4l2_fops,
+ vde->vdev.vfl_dir = VFL_DIR_M2M,
+ vde->vdev.release = video_device_release_empty,
+ vde->vdev.v4l2_dev = &vde->v4l2_dev;
+ vde->vdev.ioctl_ops = &tegra_v4l2_ioctl_ops,
+ vde->vdev.device_caps = V4L2_CAP_VIDEO_M2M_MPLANE | V4L2_CAP_STREAMING,
+
+ vde->v4l2_dev.mdev = &vde->mdev;
+ vde->mdev.ops = &tegra_media_device_ops;
+ vde->mdev.dev = dev;
+
+ strscpy(vde->mdev.model, "tegra-vde", sizeof(vde->mdev.model));
+ strscpy(vde->vdev.name, "tegra-vde", sizeof(vde->vdev.name));
+ strscpy(vde->mdev.bus_info, "platform:tegra-vde",
+ sizeof(vde->mdev.bus_info));
+
+ vde->wq = create_workqueue("tegra-vde");
+ if (!vde->wq)
+ return -ENOMEM;
+
+ err = media_device_register(&vde->mdev);
+ if (err) {
+ dev_err(dev, "failed to register media device: %d\n", err);
+ goto clean_up_media_device;
+ }
+
+ err = v4l2_device_register(dev, &vde->v4l2_dev);
+ if (err) {
+ dev_err(dev, "failed to register v4l2 device: %d\n", err);
+ goto unreg_media_device;
+ }
+
+ err = video_register_device(&vde->vdev, VFL_TYPE_VIDEO, -1);
+ if (err) {
+ dev_err(dev, "failed to register video device: %d\n", err);
+ goto unreg_v4l2;
+ }
+
+ vde->m2m = v4l2_m2m_init(&tegra_v4l2_m2m_ops);
+ err = PTR_ERR_OR_ZERO(vde->m2m);
+ if (err) {
+ dev_err(dev, "failed to initialize m2m device: %d\n", err);
+ goto unreg_video_device;
+ }
+
+ err = v4l2_m2m_register_media_controller(vde->m2m, &vde->vdev,
+ MEDIA_ENT_F_PROC_VIDEO_DECODER);
+ if (err) {
+ dev_err(dev, "failed to register media controller: %d\n", err);
+ goto release_m2m;
+ }
+
+ v4l2_info(&vde->v4l2_dev, "v4l2 device registered as /dev/video%d\n",
+ vde->vdev.num);
+
+ return 0;
+
+release_m2m:
+ v4l2_m2m_release(vde->m2m);
+unreg_video_device:
+ video_unregister_device(&vde->vdev);
+unreg_v4l2:
+ v4l2_device_unregister(&vde->v4l2_dev);
+unreg_media_device:
+ media_device_unregister(&vde->mdev);
+clean_up_media_device:
+ media_device_cleanup(&vde->mdev);
+
+ destroy_workqueue(vde->wq);
+
+ return err;
+}
+
+void tegra_vde_v4l2_deinit(struct tegra_vde *vde)
+{
+ v4l2_m2m_unregister_media_controller(vde->m2m);
+ v4l2_m2m_release(vde->m2m);
+
+ video_unregister_device(&vde->vdev);
+ v4l2_device_unregister(&vde->v4l2_dev);
+
+ media_device_unregister(&vde->mdev);
+ media_device_cleanup(&vde->mdev);
+
+ destroy_workqueue(vde->wq);
+}
diff --git a/drivers/staging/media/tegra-vde/vde.c b/drivers/staging/media/tegra-vde/vde.c
index 36f5595c0fd8..c147d58c3bfb 100644
--- a/drivers/staging/media/tegra-vde/vde.c
+++ b/drivers/staging/media/tegra-vde/vde.c
@@ -53,10 +53,10 @@ void tegra_vde_set_bits(struct tegra_vde *vde, u32 mask,
tegra_vde_writel(vde, value | mask, base, offset);
}
-static int tegra_vde_alloc_bo(struct tegra_vde *vde,
- struct tegra_vde_bo **ret_bo,
- enum dma_data_direction dma_dir,
- size_t size)
+int tegra_vde_alloc_bo(struct tegra_vde *vde,
+ struct tegra_vde_bo **ret_bo,
+ enum dma_data_direction dma_dir,
+ size_t size)
{
struct device *dev = vde->miscdev.parent;
struct tegra_vde_bo *bo;
@@ -126,7 +126,7 @@ static int tegra_vde_alloc_bo(struct tegra_vde *vde,
return err;
}
-static void tegra_vde_free_bo(struct tegra_vde_bo *bo)
+void tegra_vde_free_bo(struct tegra_vde_bo *bo)
{
struct tegra_vde *vde = bo->vde;
struct device *dev = vde->miscdev.parent;
@@ -332,6 +332,8 @@ static int tegra_vde_ioctl_decode_h264(struct tegra_vde *vde,
dpb_frames[i].flags = frames[i].flags;
dpb_frames[i].frame_num = frames[i].frame_num;
+ dpb_frames[i].luma_atoms_pitch = ctx.pic_width_in_mbs;
+ dpb_frames[i].chroma_atoms_pitch = cstride / VDE_ATOM;
dma_dir = (i == 0) ? DMA_FROM_DEVICE : DMA_TO_DEVICE;
@@ -626,8 +628,16 @@ static int tegra_vde_probe(struct platform_device *pdev)
goto err_free_secure_bo;
}
+ err = tegra_vde_v4l2_init(vde);
+ if (err) {
+ dev_err(dev, "Failed to initialize V4L2: %d\n", err);
+ goto misc_unreg;
+ }
+
return 0;
+misc_unreg:
+ misc_deregister(&vde->miscdev);
err_free_secure_bo:
tegra_vde_free_bo(vde->secure_bo);
err_pm_runtime:
@@ -648,6 +658,7 @@ static int tegra_vde_remove(struct platform_device *pdev)
struct tegra_vde *vde = platform_get_drvdata(pdev);
struct device *dev = &pdev->dev;
+ tegra_vde_v4l2_deinit(vde);
misc_deregister(&vde->miscdev);
tegra_vde_free_bo(vde->secure_bo);
@@ -722,20 +733,73 @@ static const struct dev_pm_ops tegra_vde_pm_ops = {
tegra_vde_pm_resume)
};
+static const u32 tegra124_decoded_fmts[] = {
+ /* TBD: T124 supports only a non-standard Tegra tiled format */
+};
+
+static const struct tegra_coded_fmt_desc tegra124_coded_fmts[] = {
+ {
+ .fourcc = V4L2_PIX_FMT_H264_SLICE,
+ .frmsize = {
+ .min_width = 16,
+ .max_width = 1920,
+ .step_width = 16,
+ .min_height = 16,
+ .max_height = 2032,
+ .step_height = 16,
+ },
+ .num_decoded_fmts = ARRAY_SIZE(tegra124_decoded_fmts),
+ .decoded_fmts = tegra124_decoded_fmts,
+ .decode_run = tegra_vde_h264_decode_run,
+ .decode_wait = tegra_vde_h264_decode_wait,
+ },
+};
+
+static const u32 tegra20_decoded_fmts[] = {
+ V4L2_PIX_FMT_YUV420M,
+ V4L2_PIX_FMT_YVU420M,
+};
+
+static const struct tegra_coded_fmt_desc tegra20_coded_fmts[] = {
+ {
+ .fourcc = V4L2_PIX_FMT_H264_SLICE,
+ .frmsize = {
+ .min_width = 16,
+ .max_width = 1920,
+ .step_width = 16,
+ .min_height = 16,
+ .max_height = 2032,
+ .step_height = 16,
+ },
+ .num_decoded_fmts = ARRAY_SIZE(tegra20_decoded_fmts),
+ .decoded_fmts = tegra20_decoded_fmts,
+ .decode_run = tegra_vde_h264_decode_run,
+ .decode_wait = tegra_vde_h264_decode_wait,
+ },
+};
+
static const struct tegra_vde_soc tegra124_vde_soc = {
.supports_ref_pic_marking = true,
+ .coded_fmts = tegra124_coded_fmts,
+ .num_coded_fmts = ARRAY_SIZE(tegra124_coded_fmts),
};
static const struct tegra_vde_soc tegra114_vde_soc = {
.supports_ref_pic_marking = true,
+ .coded_fmts = tegra20_coded_fmts,
+ .num_coded_fmts = ARRAY_SIZE(tegra20_coded_fmts),
};
static const struct tegra_vde_soc tegra30_vde_soc = {
.supports_ref_pic_marking = false,
+ .coded_fmts = tegra20_coded_fmts,
+ .num_coded_fmts = ARRAY_SIZE(tegra20_coded_fmts),
};
static const struct tegra_vde_soc tegra20_vde_soc = {
.supports_ref_pic_marking = false,
+ .coded_fmts = tegra20_coded_fmts,
+ .num_coded_fmts = ARRAY_SIZE(tegra20_coded_fmts),
};
static const struct of_device_id tegra_vde_of_match[] = {
diff --git a/drivers/staging/media/tegra-vde/vde.h b/drivers/staging/media/tegra-vde/vde.h
index 8ba6a71e3e40..e0396bb0b986 100644
--- a/drivers/staging/media/tegra-vde/vde.h
+++ b/drivers/staging/media/tegra-vde/vde.h
@@ -15,6 +15,16 @@
#include <linux/miscdevice.h>
#include <linux/mutex.h>
#include <linux/types.h>
+#include <linux/workqueue.h>
+
+#include <media/media-device.h>
+#include <media/videobuf2-dma-contig.h>
+#include <media/videobuf2-dma-sg.h>
+#include <media/v4l2-ctrls.h>
+#include <media/v4l2-device.h>
+#include <media/v4l2-event.h>
+#include <media/v4l2-ioctl.h>
+#include <media/v4l2-mem2mem.h>
#define ICMDQUE_WR 0x00
#define CMDQUE_CONTROL 0x08
@@ -25,9 +35,15 @@
#define BSE_ICMDQUE_EMPTY BIT(3)
#define BSE_DMA_BUSY BIT(23)
+#define BSEV_ALIGN SZ_1
+#define FRAMEID_ALIGN SZ_256
+#define SXE_BUFFER SZ_32K
+#define VDE_ATOM SZ_16
+
struct clk;
struct dma_buf;
struct gen_pool;
+struct tegra_ctx;
struct iommu_group;
struct iommu_domain;
struct reset_control;
@@ -46,10 +62,23 @@ struct tegra_video_frame {
dma_addr_t aux_addr;
u32 frame_num;
u32 flags;
+ u32 luma_atoms_pitch;
+ u32 chroma_atoms_pitch;
+};
+
+struct tegra_coded_fmt_desc {
+ u32 fourcc;
+ struct v4l2_frmsize_stepwise frmsize;
+ unsigned int num_decoded_fmts;
+ const u32 *decoded_fmts;
+ int (*decode_run)(struct tegra_ctx *ctx);
+ int (*decode_wait)(struct tegra_ctx *ctx);
};
struct tegra_vde_soc {
bool supports_ref_pic_marking;
+ const struct tegra_coded_fmt_desc *coded_fmts;
+ u32 num_coded_fmts;
};
struct tegra_vde_bo {
@@ -94,8 +123,60 @@ struct tegra_vde {
dma_addr_t bitstream_data_addr;
dma_addr_t iram_lists_addr;
u32 *iram;
+ struct v4l2_device v4l2_dev;
+ struct v4l2_m2m_dev *m2m;
+ struct media_device mdev;
+ struct video_device vdev;
+ struct mutex v4l2_lock;
+ struct workqueue_struct *wq;
+ struct tegra_video_frame frames[V4L2_H264_NUM_DPB_ENTRIES + 1];
+};
+
+int tegra_vde_alloc_bo(struct tegra_vde *vde,
+ struct tegra_vde_bo **ret_bo,
+ enum dma_data_direction dma_dir,
+ size_t size);
+void tegra_vde_free_bo(struct tegra_vde_bo *bo);
+
+struct tegra_ctx_h264 {
+ const struct v4l2_ctrl_h264_decode_params *decode_params;
+ const struct v4l2_ctrl_h264_sps *sps;
+ const struct v4l2_ctrl_h264_pps *pps;
+};
+
+struct tegra_ctx {
+ struct tegra_vde *vde;
+ struct tegra_ctx_h264 h264;
+ struct work_struct work;
+ struct v4l2_fh fh;
+ struct v4l2_ctrl_handler hdl;
+ struct v4l2_format coded_fmt;
+ struct v4l2_format decoded_fmt;
+ const struct tegra_coded_fmt_desc *coded_fmt_desc;
+ struct v4l2_ctrl *ctrls[];
+};
+
+struct tegra_m2m_buffer {
+ struct v4l2_m2m_buffer m2m;
+ struct dma_buf_attachment *a[VB2_MAX_PLANES];
+ dma_addr_t dma_base[VB2_MAX_PLANES];
+ dma_addr_t dma_addr[VB2_MAX_PLANES];
+ struct iova *iova[VB2_MAX_PLANES];
+ struct tegra_vde_bo *aux;
+ bool b_frame;
};
+static inline struct tegra_m2m_buffer *
+vb_to_tegra_buf(struct vb2_buffer *vb)
+{
+ struct v4l2_m2m_buffer *m2m = container_of(vb, struct v4l2_m2m_buffer,
+ vb.vb2_buf);
+
+ return container_of(m2m, struct tegra_m2m_buffer, m2m);
+}
+
+void tegra_vde_prepare_control_data(struct tegra_ctx *ctx, u32 id);
+
void tegra_vde_writel(struct tegra_vde *vde, u32 value, void __iomem *base,
u32 offset);
u32 tegra_vde_readl(struct tegra_vde *vde, void __iomem *base, u32 offset);
@@ -111,6 +192,8 @@ int tegra_vde_decode_h264(struct tegra_vde *vde,
struct tegra_video_frame *dpb_frames,
dma_addr_t bitstream_data_addr,
size_t bitstream_data_size);
+int tegra_vde_h264_decode_run(struct tegra_ctx *ctx);
+int tegra_vde_h264_decode_wait(struct tegra_ctx *ctx);
int tegra_vde_iommu_init(struct tegra_vde *vde);
void tegra_vde_iommu_deinit(struct tegra_vde *vde);
@@ -164,4 +247,7 @@ tegra_vde_reg_base_name(struct tegra_vde *vde, void __iomem *base)
return "???";
}
+int tegra_vde_v4l2_init(struct tegra_vde *vde);
+void tegra_vde_v4l2_deinit(struct tegra_vde *vde);
+
#endif /* TEGRA_VDE_H */
--
2.34.1
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH v3 4/4] media: staging: tegra-vde: Remove legacy UAPI support
2022-02-07 14:19 [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Dmitry Osipenko
` (2 preceding siblings ...)
2022-02-07 14:19 ` [PATCH v3 3/4] media: staging: tegra-vde: Support V4L stateless video decoder API Dmitry Osipenko
@ 2022-02-07 14:19 ` Dmitry Osipenko
2022-02-18 15:04 ` [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Hans Verkuil
4 siblings, 0 replies; 10+ messages in thread
From: Dmitry Osipenko @ 2022-02-07 14:19 UTC (permalink / raw)
To: Thierry Reding, Jonathan Hunter, Mauro Carvalho Chehab,
Hans Verkuil, Nicolas Dufresne, Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
The legacy UAPI became unnecessary with the V4L stateless decoder API
support addition to the Tegra decoder driver. Remove legacy UAPI support.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
.../staging/media/tegra-vde/dmabuf-cache.c | 2 +-
drivers/staging/media/tegra-vde/h264.c | 61 ++--
drivers/staging/media/tegra-vde/iommu.c | 2 +-
drivers/staging/media/tegra-vde/uapi.h | 73 -----
drivers/staging/media/tegra-vde/vde.c | 283 +-----------------
drivers/staging/media/tegra-vde/vde.h | 11 -
6 files changed, 35 insertions(+), 397 deletions(-)
delete mode 100644 drivers/staging/media/tegra-vde/uapi.h
diff --git a/drivers/staging/media/tegra-vde/dmabuf-cache.c b/drivers/staging/media/tegra-vde/dmabuf-cache.c
index a98d03419b8f..69c346148070 100644
--- a/drivers/staging/media/tegra-vde/dmabuf-cache.c
+++ b/drivers/staging/media/tegra-vde/dmabuf-cache.c
@@ -66,9 +66,9 @@ int tegra_vde_dmabuf_cache_map(struct tegra_vde *vde,
struct dma_buf_attachment **ap,
dma_addr_t *addrp)
{
- struct device *dev = vde->miscdev.parent;
struct dma_buf_attachment *attachment;
struct tegra_vde_cache_entry *entry;
+ struct device *dev = vde->dev;
struct sg_table *sgt;
struct iova *iova;
int err;
diff --git a/drivers/staging/media/tegra-vde/h264.c b/drivers/staging/media/tegra-vde/h264.c
index cbf27f0cfdb1..a46c648a26c6 100644
--- a/drivers/staging/media/tegra-vde/h264.c
+++ b/drivers/staging/media/tegra-vde/h264.c
@@ -14,9 +14,36 @@
#include <media/v4l2-h264.h>
#include "trace.h"
-#include "uapi.h"
#include "vde.h"
+#define FLAG_B_FRAME 0x1
+#define FLAG_REFERENCE 0x2
+
+struct tegra_vde_h264_frame {
+ unsigned int frame_num;
+ unsigned int flags;
+};
+
+struct tegra_vde_h264_decoder_ctx {
+ unsigned int dpb_frames_nb;
+ unsigned int dpb_ref_frames_with_earlier_poc_nb;
+ unsigned int baseline_profile;
+ unsigned int level_idc;
+ unsigned int log2_max_pic_order_cnt_lsb;
+ unsigned int log2_max_frame_num;
+ unsigned int pic_order_cnt_type;
+ unsigned int direct_8x8_inference_flag;
+ unsigned int pic_width_in_mbs;
+ unsigned int pic_height_in_mbs;
+ unsigned int pic_init_qp;
+ unsigned int deblocking_filter_control_present_flag;
+ unsigned int constrained_intra_pred_flag;
+ unsigned int chroma_qp_index_offset;
+ unsigned int pic_order_present_flag;
+ unsigned int num_ref_idx_l0_active_minus1;
+ unsigned int num_ref_idx_l1_active_minus1;
+};
+
struct h264_reflists {
u8 p[V4L2_H264_NUM_DPB_ENTRIES];
u8 b0[V4L2_H264_NUM_DPB_ENTRIES];
@@ -438,19 +465,8 @@ static void tegra_vde_decode_frame(struct tegra_vde *vde,
vde->sxe, 0x00);
}
-int tegra_vde_validate_h264_frame(struct device *dev,
- struct tegra_vde_h264_frame *frame)
-{
- if (frame->frame_num > 0x7FFFFF) {
- dev_err(dev, "Bad frame_num %u\n", frame->frame_num);
- return -EINVAL;
- }
-
- return 0;
-}
-
-int tegra_vde_validate_h264_ctx(struct device *dev,
- struct tegra_vde_h264_decoder_ctx *ctx)
+static int tegra_vde_validate_h264_ctx(struct device *dev,
+ struct tegra_vde_h264_decoder_ctx *ctx)
{
if (ctx->dpb_frames_nb == 0 || ctx->dpb_frames_nb > 17) {
dev_err(dev, "Bad DPB size %u\n", ctx->dpb_frames_nb);
@@ -637,23 +653,6 @@ static int tegra_vde_decode_end(struct tegra_vde *vde)
return ret;
}
-int tegra_vde_decode_h264(struct tegra_vde *vde,
- struct tegra_vde_h264_decoder_ctx *ctx,
- struct tegra_video_frame *dpb_frames,
- dma_addr_t bitstream_data_addr,
- size_t bitstream_data_size)
-{
- int err;
-
- err = tegra_vde_decode_begin(vde, ctx, dpb_frames,
- bitstream_data_addr,
- bitstream_data_size);
- if (err)
- return err;
-
- return tegra_vde_decode_end(vde);
-}
-
static struct vb2_buffer *get_ref_buf(struct tegra_ctx *ctx,
struct vb2_v4l2_buffer *dst,
unsigned int dpb_idx)
diff --git a/drivers/staging/media/tegra-vde/iommu.c b/drivers/staging/media/tegra-vde/iommu.c
index adf8dc7ee25c..5521ed3e465f 100644
--- a/drivers/staging/media/tegra-vde/iommu.c
+++ b/drivers/staging/media/tegra-vde/iommu.c
@@ -60,7 +60,7 @@ void tegra_vde_iommu_unmap(struct tegra_vde *vde, struct iova *iova)
int tegra_vde_iommu_init(struct tegra_vde *vde)
{
- struct device *dev = vde->miscdev.parent;
+ struct device *dev = vde->dev;
struct iova *iova;
unsigned long order;
unsigned long shift;
diff --git a/drivers/staging/media/tegra-vde/uapi.h b/drivers/staging/media/tegra-vde/uapi.h
deleted file mode 100644
index ffb4983e5bb6..000000000000
--- a/drivers/staging/media/tegra-vde/uapi.h
+++ /dev/null
@@ -1,73 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0+ */
-/* Copyright (C) 2016-2017 Dmitry Osipenko <digetx@gmail.com> */
-#ifndef _UAPI_TEGRA_VDE_H_
-#define _UAPI_TEGRA_VDE_H_
-
-#include <linux/types.h>
-#include <asm/ioctl.h>
-
-#define FLAG_B_FRAME 0x1
-#define FLAG_REFERENCE 0x2
-
-struct tegra_vde_h264_frame {
- __s32 y_fd;
- __s32 cb_fd;
- __s32 cr_fd;
- __s32 aux_fd;
- __u32 y_offset;
- __u32 cb_offset;
- __u32 cr_offset;
- __u32 aux_offset;
- __u32 frame_num;
- __u32 flags;
-
- // Must be zero'ed
- __u32 reserved[6];
-};
-
-struct tegra_vde_h264_decoder_ctx {
- __s32 bitstream_data_fd;
- __u32 bitstream_data_offset;
-
- __u64 dpb_frames_ptr;
- __u32 dpb_frames_nb;
- __u32 dpb_ref_frames_with_earlier_poc_nb;
-
- // SPS
- __u32 baseline_profile;
- __u32 level_idc;
- __u32 log2_max_pic_order_cnt_lsb;
- __u32 log2_max_frame_num;
- __u32 pic_order_cnt_type;
- __u32 direct_8x8_inference_flag;
- __u32 pic_width_in_mbs;
- __u32 pic_height_in_mbs;
-
- // PPS
- __u32 pic_init_qp;
- __u32 deblocking_filter_control_present_flag;
- __u32 constrained_intra_pred_flag;
- __u32 chroma_qp_index_offset;
- __u32 pic_order_present_flag;
-
- // Slice header
- __u32 num_ref_idx_l0_active_minus1;
- __u32 num_ref_idx_l1_active_minus1;
-
- // Must be zero'ed
- __u32 reserved[11];
-};
-
-#define VDE_IOCTL_BASE ('v' + 0x20)
-
-#define VDE_IO(nr) _IO(VDE_IOCTL_BASE, nr)
-#define VDE_IOR(nr, type) _IOR(VDE_IOCTL_BASE, nr, type)
-#define VDE_IOW(nr, type) _IOW(VDE_IOCTL_BASE, nr, type)
-#define VDE_IOWR(nr, type) _IOWR(VDE_IOCTL_BASE, nr, type)
-
-#define TEGRA_VDE_DECODE_H264 0x00
-
-#define TEGRA_VDE_IOCTL_DECODE_H264 \
- VDE_IOW(TEGRA_VDE_DECODE_H264, struct tegra_vde_h264_decoder_ctx)
-
-#endif // _UAPI_TEGRA_VDE_H_
diff --git a/drivers/staging/media/tegra-vde/vde.c b/drivers/staging/media/tegra-vde/vde.c
index c147d58c3bfb..f3e863a94c5a 100644
--- a/drivers/staging/media/tegra-vde/vde.c
+++ b/drivers/staging/media/tegra-vde/vde.c
@@ -11,7 +11,6 @@
#include <linux/genalloc.h>
#include <linux/interrupt.h>
#include <linux/list.h>
-#include <linux/miscdevice.h>
#include <linux/module.h>
#include <linux/of_device.h>
#include <linux/pm_runtime.h>
@@ -22,7 +21,6 @@
#include <soc/tegra/common.h>
#include <soc/tegra/pmc.h>
-#include "uapi.h"
#include "vde.h"
#define CREATE_TRACE_POINTS
@@ -58,7 +56,7 @@ int tegra_vde_alloc_bo(struct tegra_vde *vde,
enum dma_data_direction dma_dir,
size_t size)
{
- struct device *dev = vde->miscdev.parent;
+ struct device *dev = vde->dev;
struct tegra_vde_bo *bo;
int err;
@@ -129,7 +127,7 @@ int tegra_vde_alloc_bo(struct tegra_vde *vde,
void tegra_vde_free_bo(struct tegra_vde_bo *bo)
{
struct tegra_vde *vde = bo->vde;
- struct device *dev = vde->miscdev.parent;
+ struct device *dev = vde->dev;
if (vde->domain)
tegra_vde_iommu_unmap(vde, bo->iova);
@@ -143,266 +141,6 @@ void tegra_vde_free_bo(struct tegra_vde_bo *bo)
kfree(bo);
}
-static int tegra_vde_attach_dmabuf(struct tegra_vde *vde,
- int fd,
- unsigned long offset,
- size_t min_size,
- size_t align_size,
- struct dma_buf_attachment **a,
- dma_addr_t *addrp,
- size_t *size,
- enum dma_data_direction dma_dir)
-{
- struct device *dev = vde->miscdev.parent;
- struct dma_buf *dmabuf;
- int err;
-
- dmabuf = dma_buf_get(fd);
- if (IS_ERR(dmabuf)) {
- dev_err(dev, "Invalid dmabuf FD\n");
- return PTR_ERR(dmabuf);
- }
-
- if (dmabuf->size & (align_size - 1)) {
- dev_err(dev, "Unaligned dmabuf 0x%zX, should be aligned to 0x%zX\n",
- dmabuf->size, align_size);
- return -EINVAL;
- }
-
- if ((u64)offset + min_size > dmabuf->size) {
- dev_err(dev, "Too small dmabuf size %zu @0x%lX, should be at least %zu\n",
- dmabuf->size, offset, min_size);
- return -EINVAL;
- }
-
- err = tegra_vde_dmabuf_cache_map(vde, dmabuf, dma_dir, a, addrp);
- if (err)
- goto err_put;
-
- *addrp = *addrp + offset;
-
- if (size)
- *size = dmabuf->size - offset;
-
- return 0;
-
-err_put:
- dma_buf_put(dmabuf);
-
- return err;
-}
-
-static int tegra_vde_attach_dmabufs_to_frame(struct tegra_vde *vde,
- struct tegra_video_frame *frame,
- struct tegra_vde_h264_frame *src,
- enum dma_data_direction dma_dir,
- bool baseline_profile,
- size_t lsize, size_t csize)
-{
- int err;
-
- err = tegra_vde_attach_dmabuf(vde, src->y_fd,
- src->y_offset, lsize, SZ_256,
- &frame->y_dmabuf_attachment,
- &frame->y_addr,
- NULL, dma_dir);
- if (err)
- return err;
-
- err = tegra_vde_attach_dmabuf(vde, src->cb_fd,
- src->cb_offset, csize, SZ_256,
- &frame->cb_dmabuf_attachment,
- &frame->cb_addr,
- NULL, dma_dir);
- if (err)
- goto err_release_y;
-
- err = tegra_vde_attach_dmabuf(vde, src->cr_fd,
- src->cr_offset, csize, SZ_256,
- &frame->cr_dmabuf_attachment,
- &frame->cr_addr,
- NULL, dma_dir);
- if (err)
- goto err_release_cb;
-
- if (baseline_profile) {
- frame->aux_addr = 0x64DEAD00;
- return 0;
- }
-
- err = tegra_vde_attach_dmabuf(vde, src->aux_fd,
- src->aux_offset, csize, SZ_256,
- &frame->aux_dmabuf_attachment,
- &frame->aux_addr,
- NULL, dma_dir);
- if (err)
- goto err_release_cr;
-
- return 0;
-
-err_release_cr:
- tegra_vde_dmabuf_cache_unmap(vde, frame->cr_dmabuf_attachment, true);
-err_release_cb:
- tegra_vde_dmabuf_cache_unmap(vde, frame->cb_dmabuf_attachment, true);
-err_release_y:
- tegra_vde_dmabuf_cache_unmap(vde, frame->y_dmabuf_attachment, true);
-
- return err;
-}
-
-static void tegra_vde_release_frame_dmabufs(struct tegra_vde *vde,
- struct tegra_video_frame *frame,
- enum dma_data_direction dma_dir,
- bool baseline_profile,
- bool release)
-{
- if (!baseline_profile)
- tegra_vde_dmabuf_cache_unmap(vde, frame->aux_dmabuf_attachment,
- release);
-
- tegra_vde_dmabuf_cache_unmap(vde, frame->cr_dmabuf_attachment, release);
- tegra_vde_dmabuf_cache_unmap(vde, frame->cb_dmabuf_attachment, release);
- tegra_vde_dmabuf_cache_unmap(vde, frame->y_dmabuf_attachment, release);
-}
-
-static int tegra_vde_ioctl_decode_h264(struct tegra_vde *vde,
- unsigned long vaddr)
-{
- struct dma_buf_attachment *bitstream_data_dmabuf_attachment;
- struct tegra_vde_h264_frame __user *frames_user;
- size_t bitstream_data_size, lsize, csize;
- struct device *dev = vde->miscdev.parent;
- struct tegra_vde_h264_decoder_ctx ctx;
- struct tegra_video_frame *dpb_frames;
- struct tegra_vde_h264_frame *frames;
- enum dma_data_direction dma_dir;
- dma_addr_t bitstream_data_addr;
- unsigned int macroblocks_nb;
- unsigned int cstride;
- unsigned int i;
- int ret;
-
- if (copy_from_user(&ctx, (void __user *)vaddr, sizeof(ctx)))
- return -EFAULT;
-
- ret = tegra_vde_validate_h264_ctx(dev, &ctx);
- if (ret)
- return ret;
-
- ret = tegra_vde_attach_dmabuf(vde, ctx.bitstream_data_fd,
- ctx.bitstream_data_offset,
- SZ_16K, SZ_16K,
- &bitstream_data_dmabuf_attachment,
- &bitstream_data_addr,
- &bitstream_data_size,
- DMA_TO_DEVICE);
- if (ret)
- return ret;
-
- frames = kmalloc_array(ctx.dpb_frames_nb, sizeof(*frames), GFP_KERNEL);
- if (!frames) {
- ret = -ENOMEM;
- goto release_bitstream_dmabuf;
- }
-
- dpb_frames = kcalloc(ctx.dpb_frames_nb, sizeof(*dpb_frames),
- GFP_KERNEL);
- if (!dpb_frames) {
- ret = -ENOMEM;
- goto free_frames;
- }
-
- macroblocks_nb = ctx.pic_width_in_mbs * ctx.pic_height_in_mbs;
- frames_user = u64_to_user_ptr(ctx.dpb_frames_ptr);
-
- if (copy_from_user(frames, frames_user,
- ctx.dpb_frames_nb * sizeof(*frames))) {
- ret = -EFAULT;
- goto free_dpb_frames;
- }
-
- cstride = ALIGN(ctx.pic_width_in_mbs * 8, 16);
- csize = cstride * ctx.pic_height_in_mbs * 8;
- lsize = macroblocks_nb * 256;
-
- for (i = 0; i < ctx.dpb_frames_nb; i++) {
- ret = tegra_vde_validate_h264_frame(dev, &frames[i]);
- if (ret)
- goto release_dpb_frames;
-
- dpb_frames[i].flags = frames[i].flags;
- dpb_frames[i].frame_num = frames[i].frame_num;
- dpb_frames[i].luma_atoms_pitch = ctx.pic_width_in_mbs;
- dpb_frames[i].chroma_atoms_pitch = cstride / VDE_ATOM;
-
- dma_dir = (i == 0) ? DMA_FROM_DEVICE : DMA_TO_DEVICE;
-
- ret = tegra_vde_attach_dmabufs_to_frame(vde, &dpb_frames[i],
- &frames[i], dma_dir,
- ctx.baseline_profile,
- lsize, csize);
- if (ret)
- goto release_dpb_frames;
- }
-
- ret = tegra_vde_decode_h264(vde, &ctx, dpb_frames,
- bitstream_data_addr, bitstream_data_size);
-
-release_dpb_frames:
- while (i--) {
- dma_dir = (i == 0) ? DMA_FROM_DEVICE : DMA_TO_DEVICE;
-
- tegra_vde_release_frame_dmabufs(vde, &dpb_frames[i], dma_dir,
- ctx.baseline_profile, ret != 0);
- }
-
-free_dpb_frames:
- kfree(dpb_frames);
-
-free_frames:
- kfree(frames);
-
-release_bitstream_dmabuf:
- tegra_vde_dmabuf_cache_unmap(vde, bitstream_data_dmabuf_attachment,
- ret != 0);
-
- return ret;
-}
-
-static long tegra_vde_unlocked_ioctl(struct file *filp,
- unsigned int cmd, unsigned long arg)
-{
- struct miscdevice *miscdev = filp->private_data;
- struct tegra_vde *vde = container_of(miscdev, struct tegra_vde,
- miscdev);
-
- switch (cmd) {
- case TEGRA_VDE_IOCTL_DECODE_H264:
- return tegra_vde_ioctl_decode_h264(vde, arg);
- }
-
- dev_err(miscdev->parent, "Invalid IOCTL command %u\n", cmd);
-
- return -ENOTTY;
-}
-
-static int tegra_vde_release_file(struct inode *inode, struct file *filp)
-{
- struct miscdevice *miscdev = filp->private_data;
- struct tegra_vde *vde = container_of(miscdev, struct tegra_vde,
- miscdev);
-
- tegra_vde_dmabuf_cache_unmap_sync(vde);
-
- return 0;
-}
-
-static const struct file_operations tegra_vde_fops = {
- .owner = THIS_MODULE,
- .unlocked_ioctl = tegra_vde_unlocked_ioctl,
- .release = tegra_vde_release_file,
-};
-
static irqreturn_t tegra_vde_isr(int irq, void *data)
{
struct tegra_vde *vde = data;
@@ -590,11 +328,6 @@ static int tegra_vde_probe(struct platform_device *pdev)
mutex_init(&vde->lock);
init_completion(&vde->decode_completion);
- vde->miscdev.minor = MISC_DYNAMIC_MINOR;
- vde->miscdev.name = "tegra_vde";
- vde->miscdev.fops = &tegra_vde_fops;
- vde->miscdev.parent = dev;
-
err = tegra_vde_iommu_init(vde);
if (err) {
dev_err(dev, "Failed to initialize IOMMU: %d\n", err);
@@ -622,22 +355,14 @@ static int tegra_vde_probe(struct platform_device *pdev)
goto err_pm_runtime;
}
- err = misc_register(&vde->miscdev);
- if (err) {
- dev_err(dev, "Failed to register misc device: %d\n", err);
- goto err_free_secure_bo;
- }
-
err = tegra_vde_v4l2_init(vde);
if (err) {
dev_err(dev, "Failed to initialize V4L2: %d\n", err);
- goto misc_unreg;
+ goto err_free_secure_bo;
}
return 0;
-misc_unreg:
- misc_deregister(&vde->miscdev);
err_free_secure_bo:
tegra_vde_free_bo(vde->secure_bo);
err_pm_runtime:
@@ -659,8 +384,6 @@ static int tegra_vde_remove(struct platform_device *pdev)
struct device *dev = &pdev->dev;
tegra_vde_v4l2_deinit(vde);
- misc_deregister(&vde->miscdev);
-
tegra_vde_free_bo(vde->secure_bo);
/*
diff --git a/drivers/staging/media/tegra-vde/vde.h b/drivers/staging/media/tegra-vde/vde.h
index e0396bb0b986..0fbb1f3d2c88 100644
--- a/drivers/staging/media/tegra-vde/vde.h
+++ b/drivers/staging/media/tegra-vde/vde.h
@@ -12,7 +12,6 @@
#include <linux/dma-direction.h>
#include <linux/iova.h>
#include <linux/list.h>
-#include <linux/miscdevice.h>
#include <linux/mutex.h>
#include <linux/types.h>
#include <linux/workqueue.h>
@@ -107,7 +106,6 @@ struct tegra_vde {
struct mutex lock;
struct mutex map_lock;
struct list_head map_list;
- struct miscdevice miscdev;
struct reset_control *rst;
struct reset_control *rst_mc;
struct gen_pool *iram_pool;
@@ -183,15 +181,6 @@ u32 tegra_vde_readl(struct tegra_vde *vde, void __iomem *base, u32 offset);
void tegra_vde_set_bits(struct tegra_vde *vde, u32 mask, void __iomem *base,
u32 offset);
-int tegra_vde_validate_h264_frame(struct device *dev,
- struct tegra_vde_h264_frame *frame);
-int tegra_vde_validate_h264_ctx(struct device *dev,
- struct tegra_vde_h264_decoder_ctx *ctx);
-int tegra_vde_decode_h264(struct tegra_vde *vde,
- struct tegra_vde_h264_decoder_ctx *ctx,
- struct tegra_video_frame *dpb_frames,
- dma_addr_t bitstream_data_addr,
- size_t bitstream_data_size);
int tegra_vde_h264_decode_run(struct tegra_ctx *ctx);
int tegra_vde_h264_decode_wait(struct tegra_ctx *ctx);
--
2.34.1
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags
2022-02-07 14:19 ` [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags Dmitry Osipenko
@ 2022-02-11 13:29 ` Nicolas Dufresne
2022-02-11 14:21 ` Dmitry Osipenko
2022-02-18 14:53 ` Nicolas Dufresne
1 sibling, 1 reply; 10+ messages in thread
From: Nicolas Dufresne @ 2022-02-11 13:29 UTC (permalink / raw)
To: Dmitry Osipenko, Thierry Reding, Jonathan Hunter,
Mauro Carvalho Chehab, Hans Verkuil, Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
Le lundi 07 février 2022 à 17:19 +0300, Dmitry Osipenko a écrit :
> Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags that are needed by
> NVIDIA Tegra video decoder. Userspace will have to set these flags in
> accordance to the type of a decoded frame.
>
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
> ---
Can you share here (out of commit) a userland implementation that was used to
demonstrate the new API ? It does not matter which project (GStreamer, FFMPEG,
Chromium, your_test_progtram). Should be fairly straightforward. This is just to
be transparent.
> .../userspace-api/media/v4l/ext-ctrls-codec-stateless.rst | 6 ++++++
> include/uapi/linux/v4l2-controls.h | 2 ++
> 2 files changed, 8 insertions(+)
>
> diff --git a/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst b/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst
> index cc080c4257d0..f87584ad90ba 100644
> --- a/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst
> +++ b/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst
> @@ -616,6 +616,12 @@ Stateless Codec Control ID
> * - ``V4L2_H264_DECODE_PARAM_FLAG_BOTTOM_FIELD``
> - 0x00000004
> -
> + * - ``V4L2_H264_DECODE_PARAM_FLAG_PFRAME``
> + - 0x00000008
> + -
> + * - ``V4L2_H264_DECODE_PARAM_FLAG_BFRAME``
> + - 0x00000010
> + -
>
> .. raw:: latex
>
> diff --git a/include/uapi/linux/v4l2-controls.h b/include/uapi/linux/v4l2-controls.h
> index c8e0f84d204d..e3d48d571062 100644
> --- a/include/uapi/linux/v4l2-controls.h
> +++ b/include/uapi/linux/v4l2-controls.h
> @@ -1563,6 +1563,8 @@ struct v4l2_h264_dpb_entry {
> #define V4L2_H264_DECODE_PARAM_FLAG_IDR_PIC 0x01
> #define V4L2_H264_DECODE_PARAM_FLAG_FIELD_PIC 0x02
> #define V4L2_H264_DECODE_PARAM_FLAG_BOTTOM_FIELD 0x04
> +#define V4L2_H264_DECODE_PARAM_FLAG_PFRAME 0x08
> +#define V4L2_H264_DECODE_PARAM_FLAG_BFRAME 0x10
>
> #define V4L2_CID_STATELESS_H264_DECODE_PARAMS (V4L2_CID_CODEC_STATELESS_BASE + 7)
> /**
Please update the doc too, see:
Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst:
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags
2022-02-11 13:29 ` Nicolas Dufresne
@ 2022-02-11 14:21 ` Dmitry Osipenko
0 siblings, 0 replies; 10+ messages in thread
From: Dmitry Osipenko @ 2022-02-11 14:21 UTC (permalink / raw)
To: Nicolas Dufresne, Dmitry Osipenko, Thierry Reding,
Jonathan Hunter, Mauro Carvalho Chehab, Hans Verkuil,
Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
11.02.2022 16:29, Nicolas Dufresne пишет:
> Le lundi 07 février 2022 à 17:19 +0300, Dmitry Osipenko a écrit :
>> Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags that are needed by
>> NVIDIA Tegra video decoder. Userspace will have to set these flags in
>> accordance to the type of a decoded frame.
>>
>> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
>> ---
> Can you share here (out of commit) a userland implementation that was used to
> demonstrate the new API ? It does not matter which project (GStreamer, FFMPEG,
> Chromium, your_test_progtram). Should be fairly straightforward. This is just to
> be transparent.
>
You may find all the links in the cover letter, please see the
grate-driver's GStreamer [1] and libvdpau-tegra [2]:
[1] https://github.com/grate-driver/gstreamer/commits/main
[2] https://github.com/grate-driver/libvdpau-tegra/commits/master
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags
2022-02-07 14:19 ` [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags Dmitry Osipenko
2022-02-11 13:29 ` Nicolas Dufresne
@ 2022-02-18 14:53 ` Nicolas Dufresne
1 sibling, 0 replies; 10+ messages in thread
From: Nicolas Dufresne @ 2022-02-18 14:53 UTC (permalink / raw)
To: Dmitry Osipenko, Thierry Reding, Jonathan Hunter,
Mauro Carvalho Chehab, Hans Verkuil, Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
Le lundi 07 février 2022 à 17:19 +0300, Dmitry Osipenko a écrit :
> Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags that are needed by
> NVIDIA Tegra video decoder. Userspace will have to set these flags in
> accordance to the type of a decoded frame.
>
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Reviewed-by: Nicolas Dufresne <nicolas@collabora.com>
> ---
> .../userspace-api/media/v4l/ext-ctrls-codec-stateless.rst | 6 ++++++
> include/uapi/linux/v4l2-controls.h | 2 ++
> 2 files changed, 8 insertions(+)
>
> diff --git a/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst b/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst
> index cc080c4257d0..f87584ad90ba 100644
> --- a/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst
> +++ b/Documentation/userspace-api/media/v4l/ext-ctrls-codec-stateless.rst
> @@ -616,6 +616,12 @@ Stateless Codec Control ID
> * - ``V4L2_H264_DECODE_PARAM_FLAG_BOTTOM_FIELD``
> - 0x00000004
> -
> + * - ``V4L2_H264_DECODE_PARAM_FLAG_PFRAME``
> + - 0x00000008
> + -
> + * - ``V4L2_H264_DECODE_PARAM_FLAG_BFRAME``
> + - 0x00000010
> + -
>
> .. raw:: latex
>
> diff --git a/include/uapi/linux/v4l2-controls.h b/include/uapi/linux/v4l2-controls.h
> index c8e0f84d204d..e3d48d571062 100644
> --- a/include/uapi/linux/v4l2-controls.h
> +++ b/include/uapi/linux/v4l2-controls.h
> @@ -1563,6 +1563,8 @@ struct v4l2_h264_dpb_entry {
> #define V4L2_H264_DECODE_PARAM_FLAG_IDR_PIC 0x01
> #define V4L2_H264_DECODE_PARAM_FLAG_FIELD_PIC 0x02
> #define V4L2_H264_DECODE_PARAM_FLAG_BOTTOM_FIELD 0x04
> +#define V4L2_H264_DECODE_PARAM_FLAG_PFRAME 0x08
> +#define V4L2_H264_DECODE_PARAM_FLAG_BFRAME 0x10
>
> #define V4L2_CID_STATELESS_H264_DECODE_PARAMS (V4L2_CID_CODEC_STATELESS_BASE + 7)
> /**
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver
2022-02-07 14:19 [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Dmitry Osipenko
` (3 preceding siblings ...)
2022-02-07 14:19 ` [PATCH v3 4/4] media: staging: tegra-vde: Remove legacy UAPI support Dmitry Osipenko
@ 2022-02-18 15:04 ` Hans Verkuil
2022-02-18 17:48 ` Dmitry Osipenko
4 siblings, 1 reply; 10+ messages in thread
From: Hans Verkuil @ 2022-02-18 15:04 UTC (permalink / raw)
To: Dmitry Osipenko, Thierry Reding, Jonathan Hunter,
Mauro Carvalho Chehab, Nicolas Dufresne, Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
Dmitry,
On 07/02/2022 15:19, Dmitry Osipenko wrote:
> Support V4L stateless video decoder API by NVIDIA Tegra decoder driver.
> Tested using GStreamer [1] and libvdpau-tegra [2][8].
>
> [1] https://github.com/grate-driver/gstreamer/commit/b8509bdbb69b534e61419ea1798f32f9ad2f3597
> [2] https://github.com/grate-driver/libvdpau-tegra/commit/f822e95911e5e0c39f8ba19f843ddc1e0138d5ce
> [8] https://github.com/grate-driver/libvdpau-tegra/commit/80db4d02369f2a984ce3173d6bc305f32e9fdb97
Nice work!
Can you make a v4 that just moves the whole driver to drivers/media/platform? I see no
reason for keeping this in staging.
Once I have a v4 that does that I plan to make a PR for it.
For the v4 don't forget to update MAINTAINERS with the new path and to drop the TODO
file in staging.
Regards,
Hans
>
> Changelog:
>
> v3: - Added new decode_params flags [7] instead of V4L2_BUF_FLAG_*FRAME flags,
> as was suggested by Nicolas Dufresne.
>
> [7] https://github.com/grate-driver/gstreamer/commit/c5cd847f9c26b7669720ae58f9058de2515f51a2
>
> - Added new patch that removes legacy UAPI.
>
> v2: - Made V4L2_BUF_FLAG_*FRAME flags mandatory [3] and dropped reading
> of raw bitstream from the driver code, as was suggested by
> Nicolas Dufresne.
>
> [3] https://github.com/grate-driver/gstreamer/commit/aee292f0f2e84b7654a314dd7e63f916888ffaa5
>
> - Ran v4l2-compliance [4] and fluster [5][6] tests, like was suggested by
> Nicolas Dufresne. Fixed minor v4l2-compliance errors that were related
> to a partial initialization of the coded format and were harmless in
> practice, but made compliance checker unhappy.
>
> [4] https://gist.github.com/digetx/5d6bcdab633488f1dcc7c141ab90d30e
> [5] https://gist.github.com/digetx/b06c5d779e9d25afa41d9f46946fe399
> [6] https://gist.github.com/digetx/ac4198bc340e5065aa8ec3288bb21356
>
> Dmitry Osipenko (4):
> media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags
> media: staging: tegra-vde: Factor out H.264 code
> media: staging: tegra-vde: Support V4L stateless video decoder API
> media: staging: tegra-vde: Remove legacy UAPI support
>
> .../media/v4l/ext-ctrls-codec-stateless.rst | 6 +
> drivers/staging/media/tegra-vde/Kconfig | 7 +
> drivers/staging/media/tegra-vde/Makefile | 2 +-
> .../staging/media/tegra-vde/dmabuf-cache.c | 2 +-
> drivers/staging/media/tegra-vde/h264.c | 946 +++++++++++++++
> drivers/staging/media/tegra-vde/iommu.c | 2 +-
> drivers/staging/media/tegra-vde/uapi.h | 73 --
> drivers/staging/media/tegra-vde/v4l2.c | 1018 +++++++++++++++++
> drivers/staging/media/tegra-vde/vde.c | 945 ++-------------
> drivers/staging/media/tegra-vde/vde.h | 121 +-
> include/uapi/linux/v4l2-controls.h | 2 +
> 11 files changed, 2170 insertions(+), 954 deletions(-)
> create mode 100644 drivers/staging/media/tegra-vde/h264.c
> delete mode 100644 drivers/staging/media/tegra-vde/uapi.h
> create mode 100644 drivers/staging/media/tegra-vde/v4l2.c
>
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver
2022-02-18 15:04 ` [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Hans Verkuil
@ 2022-02-18 17:48 ` Dmitry Osipenko
0 siblings, 0 replies; 10+ messages in thread
From: Dmitry Osipenko @ 2022-02-18 17:48 UTC (permalink / raw)
To: Hans Verkuil, Thierry Reding, Jonathan Hunter,
Mauro Carvalho Chehab, Nicolas Dufresne, Ezequiel Garcia
Cc: linux-media, linux-staging, linux-tegra, linux-kernel
18.02.2022 18:04, Hans Verkuil пишет:
> Dmitry,
>
> On 07/02/2022 15:19, Dmitry Osipenko wrote:
>> Support V4L stateless video decoder API by NVIDIA Tegra decoder driver.
>> Tested using GStreamer [1] and libvdpau-tegra [2][8].
>>
>> [1] https://github.com/grate-driver/gstreamer/commit/b8509bdbb69b534e61419ea1798f32f9ad2f3597
>> [2] https://github.com/grate-driver/libvdpau-tegra/commit/f822e95911e5e0c39f8ba19f843ddc1e0138d5ce
>> [8] https://github.com/grate-driver/libvdpau-tegra/commit/80db4d02369f2a984ce3173d6bc305f32e9fdb97
> Nice work!
>
> Can you make a v4 that just moves the whole driver to drivers/media/platform? I see no
> reason for keeping this in staging.
>
> Once I have a v4 that does that I plan to make a PR for it.
>
> For the v4 don't forget to update MAINTAINERS with the new path and to drop the TODO
> file in staging.
I'll prepare v4 over the weekend, thanks to you and Nicolas.
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2022-02-18 17:48 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-02-07 14:19 [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Dmitry Osipenko
2022-02-07 14:19 ` [PATCH v3 1/4] media: v4l2-ctrls: Add new V4L2_H264_DECODE_PARAM_FLAG_P/BFRAME flags Dmitry Osipenko
2022-02-11 13:29 ` Nicolas Dufresne
2022-02-11 14:21 ` Dmitry Osipenko
2022-02-18 14:53 ` Nicolas Dufresne
2022-02-07 14:19 ` [PATCH v3 2/4] media: staging: tegra-vde: Factor out H.264 code Dmitry Osipenko
2022-02-07 14:19 ` [PATCH v3 3/4] media: staging: tegra-vde: Support V4L stateless video decoder API Dmitry Osipenko
2022-02-07 14:19 ` [PATCH v3 4/4] media: staging: tegra-vde: Remove legacy UAPI support Dmitry Osipenko
2022-02-18 15:04 ` [PATCH v3 0/4] Add V4L stateless video decoder API support to NVIDIA Tegra driver Hans Verkuil
2022-02-18 17:48 ` Dmitry Osipenko
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.