From: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
To: christian.koenig@amd.com, dri-devel@lists.freedesktop.org
Cc: Alexander.Deucher@amd.com, tzimmermann@suse.de,
amd-gfx@lists.freedesktop.org
Subject: Re: [PATCH] drm/amdgpu: Initialise drm_gem_object_funcs for imported BOs
Date: Tue, 8 Dec 2020 13:26:17 -0500 [thread overview]
Message-ID: <dde83a65-07ae-2285-9999-a5c0f09dc1d3@amd.com> (raw)
In-Reply-To: <76baebc8-6fd3-6b28-8c62-bd049875a8c5@gmail.com>
On 12/8/20 12:36 PM, Christian König wrote:
> Am 08.12.20 um 18:10 schrieb Andrey Grodzovsky:
>> For BOs imported from outside of amdgpu, setting of amdgpu_gem_object_funcs
>> was missing in amdgpu_dma_buf_create_obj. Fix by refactoring BO creation
>> and amdgpu_gem_object_funcs setting into single function called
>> from both code paths.
>
> Can you outline why we can't use amdgpu_gem_object_create() directly?
>
> I mean we have a bit of extra error handling in there and we need to grab the
> resv lock and set the domains after creation, but that shouldn't matter and I
> don't see why that should not work.
On top of what you mentioned you also have bp.domain/bp.preferred_domain being
set differently so you need to add another
argument to amdgpu_gem_object_create to reflect this difference which clutters
even more the already cluttered argument list.
Regarding the extra error handling - you have the 'retry' dance in
amdgpu_gem_object_create which jumps back to the middle of amdgpu_bo_param
initialization but you don't have it in amdgpu_dma_buf_create_obj which also
complicates the reuse of amdgpu_gem_object_create as is.
Andrey
>
> Thanks,
> Christian.
>
>>
>> This fixes null ptr regression casued by commit
>> d693def drm: Remove obsolete GEM and PRIME callbacks from struct drm_driver
>>
>> Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
>> ---
>> drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c | 13 ++++++-------
>> drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 22 +++++++++++++++++-----
>> drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h | 5 +++++
>> 3 files changed, 28 insertions(+), 12 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
>> index e5919ef..da4d0ab 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c
>> @@ -405,6 +405,7 @@ struct dma_buf *amdgpu_gem_prime_export(struct
>> drm_gem_object *gobj,
>> return buf;
>> }
>> +
>> /**
>> * amdgpu_dma_buf_create_obj - create BO for DMA-buf import
>> *
>> @@ -424,7 +425,7 @@ amdgpu_dma_buf_create_obj(struct drm_device *dev, struct
>> dma_buf *dma_buf)
>> struct amdgpu_device *adev = drm_to_adev(dev);
>> struct amdgpu_bo *bo;
>> struct amdgpu_bo_param bp;
>> - int ret;
>> + struct drm_gem_object *obj;
>> memset(&bp, 0, sizeof(bp));
>> bp.size = dma_buf->size;
>> @@ -434,21 +435,19 @@ amdgpu_dma_buf_create_obj(struct drm_device *dev,
>> struct dma_buf *dma_buf)
>> bp.type = ttm_bo_type_sg;
>> bp.resv = resv;
>> dma_resv_lock(resv, NULL);
>> - ret = amdgpu_bo_create(adev, &bp, &bo);
>> - if (ret)
>> + obj = amdgpu_gem_object_create_raw(adev, &bp);
>> + if (IS_ERR(obj))
>> goto error;
>> + bo = gem_to_amdgpu_bo(obj);
>> bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
>> bo->preferred_domains = AMDGPU_GEM_DOMAIN_GTT;
>> if (dma_buf->ops != &amdgpu_dmabuf_ops)
>> bo->prime_shared_count = 1;
>> - dma_resv_unlock(resv);
>> - return &bo->tbo.base;
>> -
>> error:
>> dma_resv_unlock(resv);
>> - return ERR_PTR(ret);
>> + return obj;
>> }
>> /**
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>> index c9f94fb..5f22ce6 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
>> @@ -52,13 +52,26 @@ static void amdgpu_gem_object_free(struct drm_gem_object
>> *gobj)
>> }
>> }
>> +struct drm_gem_object *amdgpu_gem_object_create_raw(struct amdgpu_device
>> *adev,
>> + struct amdgpu_bo_param *bp)
>> +{
>> + struct amdgpu_bo *bo;
>> + int r;
>> +
>> + r = amdgpu_bo_create(adev, bp, &bo);
>> + if (r)
>> + return ERR_PTR(r);
>> +
>> + bo->tbo.base.funcs = &amdgpu_gem_object_funcs;
>> + return &bo->tbo.base;
>> +}
>> +
>> int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
>> int alignment, u32 initial_domain,
>> u64 flags, enum ttm_bo_type type,
>> struct dma_resv *resv,
>> struct drm_gem_object **obj)
>> {
>> - struct amdgpu_bo *bo;
>> struct amdgpu_bo_param bp;
>> int r;
>> @@ -73,8 +86,9 @@ int amdgpu_gem_object_create(struct amdgpu_device *adev,
>> unsigned long size,
>> retry:
>> bp.flags = flags;
>> bp.domain = initial_domain;
>> - r = amdgpu_bo_create(adev, &bp, &bo);
>> - if (r) {
>> + *obj = amdgpu_gem_object_create_raw(adev, &bp);
>> + if (IS_ERR(*obj)) {
>> + r = PTR_ERR(*obj);
>> if (r != -ERESTARTSYS) {
>> if (flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED) {
>> flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
>> @@ -90,8 +104,6 @@ int amdgpu_gem_object_create(struct amdgpu_device *adev,
>> unsigned long size,
>> }
>> return r;
>> }
>> - *obj = &bo->tbo.base;
>> - (*obj)->funcs = &amdgpu_gem_object_funcs;
>> return 0;
>> }
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
>> index 637bf51..a6b90d3 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
>> @@ -38,12 +38,17 @@ unsigned long amdgpu_gem_timeout(uint64_t timeout_ns);
>> /*
>> * GEM objects.
>> */
>> +
>> +struct amdgpu_bo_param;
>> +
>> void amdgpu_gem_force_release(struct amdgpu_device *adev);
>> int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
>> int alignment, u32 initial_domain,
>> u64 flags, enum ttm_bo_type type,
>> struct dma_resv *resv,
>> struct drm_gem_object **obj);
>> +struct drm_gem_object *amdgpu_gem_object_create_raw(struct amdgpu_device *adev,
>> + struct amdgpu_bo_param *bp);
>> int amdgpu_mode_dumb_create(struct drm_file *file_priv,
>> struct drm_device *dev,
>
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
next prev parent reply other threads:[~2020-12-08 18:26 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-12-08 17:10 [PATCH] drm/amdgpu: Initialise drm_gem_object_funcs for imported BOs Andrey Grodzovsky
2020-12-08 17:23 ` Alex Deucher
2020-12-08 17:36 ` Christian König
2020-12-08 18:26 ` Andrey Grodzovsky [this message]
2020-12-08 18:29 ` Christian König
2020-12-08 18:44 ` Andrey Grodzovsky
2020-12-08 18:47 ` Christian König
2020-12-08 18:52 ` Andrey Grodzovsky
2020-12-08 19:01 ` Christian König
2020-12-08 19:11 ` Andrey Grodzovsky
2020-12-08 19:16 ` Christian König
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=dde83a65-07ae-2285-9999-a5c0f09dc1d3@amd.com \
--to=andrey.grodzovsky@amd.com \
--cc=Alexander.Deucher@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=christian.koenig@amd.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=tzimmermann@suse.de \
/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