dri-devel.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: Thomas Zimmermann <tzimmermann@suse.de>
To: "Christian König" <christian.koenig@amd.com>,
	oushixiong <oushixiong@kylinos.cn>,
	"Dave Airlie" <airlied@redhat.com>
Cc: David Airlie <airlied@linux.ie>,
	linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org,
	linaro-mm-sig@lists.linaro.org,
	Sumit Semwal <sumit.semwal@linaro.org>,
	linux-media@vger.kernel.org
Subject: Re: [PATCH v3] drm/ast: add dmabuf/prime buffer sharing support
Date: Wed, 7 Sep 2022 11:40:34 +0200	[thread overview]
Message-ID: <f078d10f-9613-d6b1-0ee8-50feaf7d5299@suse.de> (raw)
In-Reply-To: <ee27c832-a1fd-bc93-9f1b-33f828195e83@amd.com>


[-- Attachment #1.1: Type: text/plain, Size: 10060 bytes --]

Hi

Am 07.09.22 um 10:10 schrieb Christian König:
> Hi Thomas,
> 
> I was wondering pretty much the same thing, but then thought that this 
> might be the first step to direct scanout from DMA-bufs.
> 
> If this isn't the case then I to see this rather critically since that 
> functionality belongs into userspace.

With GEM VRAM helpers, ast currently doesn't support dma-buf sharing. I 
do have patches that convert it to GEM SHMEM (for other reasons), which 
would also add this functionality.

I intent to post these patches in the coming days. My suggestion is to 
merge them first and then see how to go from there.

Best regards
Thomas

> 
> Regards,
> Christian.
> 
> Am 07.09.22 um 09:50 schrieb Thomas Zimmermann:
>> Hi,
>>
>> on a more general note, let me say that your patch doesn't seem to fit 
>> the ideas of how buffer sharing is supposed to work. Your patch does 
>> the BMC screen update 'behind the scenes.'
>>
>> Shouldn't userspace set up the DRM state for mirroring the output of 
>> the discrete card to the BMC?
>>
>> Best regards
>> Thomas
>>
>> Am 01.09.22 um 14:44 schrieb oushixiong:
>>>
>>> This patch adds ast specific codes for DRM prime feature, this is to
>>> allow for offloading of rending in one direction and outputs in other.
>>>
>>> This patch is designed to solve the problem that the AST is not 
>>> displayed
>>> when the server plug in a discrete graphics card at the same time.
>>> We call the dirty callback function to copy the rendering results of the
>>> discrete graphics card to the ast side by dma-buf.
>>>
>>> v1->v2:
>>>    - Fix the comment.
>>> v2->v3:
>>>    - we remove the gem_prime_import_sg_table callback and use the
>>>      gem_prime_import callback, because it just map and access the 
>>> buffer
>>>      with the CPU. and do not to pin the buffer.
>>>
>>> Signed-off-by: oushixiong <oushixiong@kylinos.cn>
>>> Acked-by: Christian König <christian.koenig@amd.com>
>>> ---
>>>   drivers/gpu/drm/ast/ast_drv.c  |  27 +++++++
>>>   drivers/gpu/drm/ast/ast_mode.c | 125 ++++++++++++++++++++++++++++++++-
>>>   2 files changed, 151 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/gpu/drm/ast/ast_drv.c 
>>> b/drivers/gpu/drm/ast/ast_drv.c
>>> index 7465c4f0156a..fd3c4bad2eb4 100644
>>> --- a/drivers/gpu/drm/ast/ast_drv.c
>>> +++ b/drivers/gpu/drm/ast/ast_drv.c
>>> @@ -28,6 +28,7 @@
>>>     #include <linux/module.h>
>>>   #include <linux/pci.h>
>>> +#include <linux/dma-buf.h>
>>>     #include <drm/drm_aperture.h>
>>>   #include <drm/drm_atomic_helper.h>
>>> @@ -50,6 +51,29 @@ module_param_named(modeset, ast_modeset, int, 0400);
>>>     DEFINE_DRM_GEM_FOPS(ast_fops);
>>>   +static struct drm_gem_object *ast_gem_prime_import(struct 
>>> drm_device *dev,
>>> +                        struct dma_buf *dma_buf)
>>> +{
>>> +    struct drm_gem_vram_object *gbo;
>>> +
>>> +    gbo = drm_gem_vram_of_gem(dma_buf->priv);
>>> +    if (gbo->bo.base.dev == dev) {
>>> +        /*
>>> +         * Importing dmabuf exported from out own gem increases
>>> +         * refcount on gem itself instead of f_count of dmabuf.
>>> +         */
>>> +        drm_gem_object_get(&gbo->bo.base);
>>> +        return &gbo->bo.base;
>>> +    }
>>> +
>>> +    gbo = drm_gem_vram_create(dev, dma_buf->size, 0);
>>> +    if (IS_ERR(gbo))
>>> +        return NULL;
>>> +
>>> +    get_dma_buf(dma_buf);
>>> +    return &gbo->bo.base;
>>> +}
>>> +
>>>   static const struct drm_driver ast_driver = {
>>>       .driver_features = DRIVER_ATOMIC |
>>>                  DRIVER_GEM |
>>> @@ -63,6 +87,9 @@ static const struct drm_driver ast_driver = {
>>>       .minor = DRIVER_MINOR,
>>>       .patchlevel = DRIVER_PATCHLEVEL,
>>>   +    .prime_fd_to_handle = drm_gem_prime_fd_to_handle,
>>> +    .gem_prime_import = ast_gem_prime_import,
>>> +
>>>       DRM_GEM_VRAM_DRIVER
>>>   };
>>>   diff --git a/drivers/gpu/drm/ast/ast_mode.c 
>>> b/drivers/gpu/drm/ast/ast_mode.c
>>> index 45b56b39ad47..65a4342c5622 100644
>>> --- a/drivers/gpu/drm/ast/ast_mode.c
>>> +++ b/drivers/gpu/drm/ast/ast_mode.c
>>> @@ -48,6 +48,8 @@
>>>   #include "ast_drv.h"
>>>   #include "ast_tables.h"
>>>   +MODULE_IMPORT_NS(DMA_BUF);
>>> +
>>>   static inline void ast_load_palette_index(struct ast_private *ast,
>>>                        u8 index, u8 red, u8 green,
>>>                        u8 blue)
>>> @@ -1535,8 +1537,129 @@ static const struct 
>>> drm_mode_config_helper_funcs ast_mode_config_helper_funcs =
>>>       .atomic_commit_tail = drm_atomic_helper_commit_tail_rpm,
>>>   };
>>>   +static int ast_handle_damage(struct drm_framebuffer *fb, int x, 
>>> int y,
>>> +                    int width, int height)
>>> +{
>>> +    struct drm_gem_vram_object *dst_bo = NULL;
>>> +    void *dst = NULL;
>>> +    int ret = 0, i;
>>> +    unsigned long offset = 0;
>>> +    bool unmap = false;
>>> +    unsigned int bytesPerPixel;
>>> +    struct iosys_map map;
>>> +    struct iosys_map dmabuf_map;
>>> +
>>> +    bytesPerPixel = fb->format->cpp[0];
>>> +
>>> +    if (!fb->obj[0]->dma_buf)
>>> +        return -EINVAL;
>>> +
>>> +    if (!fb->obj[0]->dma_buf->vmap_ptr.vaddr) {
>>> +        ret = dma_buf_vmap(fb->obj[0]->dma_buf, &dmabuf_map);
>>> +        if (ret)
>>> +            return ret;
>>> +    } else
>>> +        dmabuf_map.vaddr = fb->obj[0]->dma_buf->vmap_ptr.vaddr;
>>> +
>>> +    dst_bo = drm_gem_vram_of_gem(fb->obj[0]);
>>> +
>>> +    ret = drm_gem_vram_pin(dst_bo, 0);
>>> +    if (ret) {
>>> +        DRM_ERROR("ast_bo_pin failed\n");
>>> +        return ret;
>>> +    }
>>> +
>>> +    if (!dst_bo->map.vaddr) {
>>> +        ret = drm_gem_vram_vmap(dst_bo, &map);
>>> +        if (ret) {
>>> +            drm_gem_vram_unpin(dst_bo);
>>> +            DRM_ERROR("failed to vmap fbcon\n");
>>> +            return ret;
>>> +        }
>>> +        unmap = true;
>>> +    }
>>> +    dst = dst_bo->map.vaddr;
>>> +
>>> +    for (i = y; i < y + height; i++) {
>>> +        offset = i * fb->pitches[0] + (x * bytesPerPixel);
>>> +        memcpy_toio(dst + offset, dmabuf_map.vaddr + offset,
>>> +            width * bytesPerPixel);
>>> +    }
>>> +
>>> +    if (unmap)
>>> +        drm_gem_vram_vunmap(dst_bo, &map);
>>> +
>>> +    drm_gem_vram_unpin(dst_bo);
>>> +
>>> +    return 0;
>>> +}
>>> +
>>> +
>>> +static int ast_user_framebuffer_dirty(struct drm_framebuffer *fb,
>>> +                struct drm_file *file,
>>> +                unsigned int flags,
>>> +                unsigned int color,
>>> +                struct drm_clip_rect *clips,
>>> +                unsigned int num_clips)
>>> +{
>>> +    int i, ret = 0;
>>> +
>>> +    drm_modeset_lock_all(fb->dev);
>>> +    if (fb->obj[0]->dma_buf) {
>>> +        ret = dma_buf_begin_cpu_access(fb->obj[0]->dma_buf,
>>> +                DMA_FROM_DEVICE);
>>> +        if (ret)
>>> +            goto unlock;
>>> +    }
>>> +
>>> +    for (i = 0; i < num_clips; i++) {
>>> +        ret = ast_handle_damage(fb, clips[i].x1, clips[i].y1,
>>> +                clips[i].x2 - clips[i].x1, clips[i].y2 - clips[i].y1);
>>> +        if (ret)
>>> +            break;
>>> +    }
>>> +
>>> +    if (fb->obj[0]->dma_buf) {
>>> +        dma_buf_end_cpu_access(fb->obj[0]->dma_buf,
>>> +                DMA_FROM_DEVICE);
>>> +    }
>>> +
>>> +unlock:
>>> +    drm_modeset_unlock_all(fb->dev);
>>> +
>>> +    return ret;
>>> +}
>>> +
>>> +static void ast_user_framebuffer_destroy(struct drm_framebuffer *fb)
>>> +{
>>> +    struct iosys_map dmabuf_map;
>>> +
>>> +    if (fb->obj[0]->dma_buf) {
>>> +        dmabuf_map.is_iomem = fb->obj[0]->dma_buf->vmap_ptr.is_iomem;
>>> +        dmabuf_map.vaddr = fb->obj[0]->dma_buf->vmap_ptr.vaddr;
>>> +        if (dmabuf_map.vaddr)
>>> +            dma_buf_vunmap(fb->obj[0]->dma_buf, &dmabuf_map);
>>> +    }
>>> +
>>> +    drm_gem_fb_destroy(fb);
>>> +}
>>> +
>>> +static const struct drm_framebuffer_funcs ast_gem_fb_funcs_dirtyfb = {
>>> +    .destroy    = ast_user_framebuffer_destroy,
>>> +    .create_handle    = drm_gem_fb_create_handle,
>>> +    .dirty        = ast_user_framebuffer_dirty,
>>> +};
>>> +
>>> +static struct drm_framebuffer *
>>> +ast_gem_fb_create_with_dirty(struct drm_device *dev, struct drm_file 
>>> *file,
>>> +                const struct drm_mode_fb_cmd2 *mode_cmd)
>>> +{
>>> +    return drm_gem_fb_create_with_funcs(dev, file, mode_cmd,
>>> +                    &ast_gem_fb_funcs_dirtyfb);
>>> +}
>>> +
>>>   static const struct drm_mode_config_funcs ast_mode_config_funcs = {
>>> -    .fb_create = drm_gem_fb_create,
>>> +    .fb_create = ast_gem_fb_create_with_dirty,
>>>       .mode_valid = drm_vram_helper_mode_valid,
>>>       .atomic_check = drm_atomic_helper_check,
>>>       .atomic_commit = drm_atomic_helper_commit,
>>>
>>>
>>> Content-type: Text/plain
>>>
>>> No virus found
>>>         Checked by Hillstone Network AntiVirus
>>
> 

-- 
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
(HRB 36809, AG Nürnberg)
Geschäftsführer: Ivo Totev

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 840 bytes --]

  reply	other threads:[~2022-09-07  9:40 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-01 12:44 [PATCH v3] drm/ast: add dmabuf/prime buffer sharing support oushixiong
2022-09-05  9:19 ` Thomas Zimmermann
2022-09-07  7:50 ` Thomas Zimmermann
2022-09-07  8:10   ` Christian König
2022-09-07  9:40     ` Thomas Zimmermann [this message]
2022-09-07 10:14       ` oushixiong
  -- strict thread matches above, loose matches on Subject: below --
2022-08-29  5:59 oushixiong

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=f078d10f-9613-d6b1-0ee8-50feaf7d5299@suse.de \
    --to=tzimmermann@suse.de \
    --cc=airlied@linux.ie \
    --cc=airlied@redhat.com \
    --cc=christian.koenig@amd.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=linaro-mm-sig@lists.linaro.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=oushixiong@kylinos.cn \
    --cc=sumit.semwal@linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).