From: "Christian König" <christian.koenig@amd.com>
To: Thomas Zimmermann <tzimmermann@suse.de>,
alexander.deucher@amd.com, airlied@linux.ie, daniel@ffwll.ch,
bskeggs@redhat.com, ray.huang@amd.com,
linux-graphics-maintainer@vmware.com, sroland@vmware.com,
zackr@vmware.com, shashank.sharma@amd.com, sam@ravnborg.org,
emil.velikov@collabora.com, Felix.Kuehling@amd.com,
nirmoy.das@amd.com
Cc: nouveau@lists.freedesktop.org, amd-gfx@lists.freedesktop.org,
dri-devel@lists.freedesktop.org
Subject: Re: [PATCH v2 2/7] drm/amdgpu: Implement mmap as GEM object function
Date: Thu, 15 Apr 2021 14:16:13 +0200 [thread overview]
Message-ID: <5f669683-90fb-4a5a-0537-2a9f06cdeeb3@amd.com> (raw)
In-Reply-To: <ceb6e5df-25ba-a31b-5a24-44a7bb964009@suse.de>
Am 15.04.21 um 14:11 schrieb Thomas Zimmermann:
> Hi
>
> Am 15.04.21 um 13:38 schrieb Christian König:
>> Am 15.04.21 um 12:17 schrieb Thomas Zimmermann:
>>> Moving the driver-specific mmap code into a GEM object function allows
>>> for using DRM helpers for various mmap callbacks.
>>>
>>> This change resolves several inconsistencies between regular mmap and
>>> prime-based mmap. The vm_ops field in vma is now set for all mmap'ed
>>> areas. Previously it way only set for regular mmap calls, prime-based
>>> mmap used TTM's default vm_ops. The function amdgpu_verify_access() is
>>> no longer being called and therefore removed by this patch.
>>>
>>> As a side effect, amdgpu_ttm_vm_ops and amdgpu_ttm_fault() are now
>>> implemented in amdgpu's GEM code.
>>>
>>> v2:
>>> * rename amdgpu_ttm_vm_ops and amdgpu_ttm_fault() to
>>> amdgpu_gem_vm_ops and amdgpu_gem_fault() (Christian)
>>> * the check for kfd_bo has meanwhile been removed
>>>
>>> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
>>
>> [SNIP]
>>> +static int amdgpu_gem_prime_mmap(struct drm_gem_object *obj, struct
>>> vm_area_struct *vma)
>>> +{
>>
>> Mhm, just double checking this function is now a core GEM function
>> and not prime specific?
>>
>> If yes maybe drop the _prime part.
>
> No problem, but other functions in the GEM callbacks are also named
> _prime_. Probably needs a clean-up.
Yeah, some still needs to be renamed. But there shouldn't be any left in
amdgpu_gem.c
>
>>
>>> + struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj);
>>> + struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
>>> + unsigned long asize = amdgpu_bo_size(bo);
>>> +
>>> + if (!vma->vm_file)
>>> + return -ENODEV;
>>> +
>>> + if (!adev)
>>> + return -ENODEV;
>>> +
>>> + /* Check for valid size. */
>>> + if (asize < vma->vm_end - vma->vm_start)
>>> + return -EINVAL;
>>
>> Shouldn't we have that check in the common code?
>
> It's at [1]. I didn't really bother about tidying up the checks
> themselves. We can drop all these except for the usermm branch below.
> Same for the radeon patch.
In this case I would just drop that from both amdgpu and radeon.
Thanks,
Christian.
>
> Best regards
> Thomas
>
> [1]
> https://elixir.bootlin.com/linux/v5.12-rc7/source/drivers/gpu/drm/drm_gem.c#L1059
>
>>
>> Apart from that looks good to me.
>>
>> Christian.
>>
>>> +
>>> + if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm) ||
>>> + (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)) {
>>> + return -EPERM;
>>> + }
>>> +
>>> + return drm_gem_ttm_mmap(obj, vma);
>>> +}
>>> +
>>> static const struct drm_gem_object_funcs amdgpu_gem_object_funcs = {
>>> .free = amdgpu_gem_object_free,
>>> .open = amdgpu_gem_object_open,
>>> @@ -212,6 +266,8 @@ static const struct drm_gem_object_funcs
>>> amdgpu_gem_object_funcs = {
>>> .export = amdgpu_gem_prime_export,
>>> .vmap = drm_gem_ttm_vmap,
>>> .vunmap = drm_gem_ttm_vunmap,
>>> + .mmap = amdgpu_gem_prime_mmap,
>>> + .vm_ops = &amdgpu_gem_vm_ops,
>>> };
>>> /*
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
>>> index 1485f33c3cc7..d4083c19402b 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
>>> @@ -152,25 +152,6 @@ static void amdgpu_evict_flags(struct
>>> ttm_buffer_object *bo,
>>> *placement = abo->placement;
>>> }
>>> -/**
>>> - * amdgpu_verify_access - Verify access for a mmap call
>>> - *
>>> - * @bo: The buffer object to map
>>> - * @filp: The file pointer from the process performing the mmap
>>> - *
>>> - * This is called by ttm_bo_mmap() to verify whether a process
>>> - * has the right to mmap a BO to their process space.
>>> - */
>>> -static int amdgpu_verify_access(struct ttm_buffer_object *bo,
>>> struct file *filp)
>>> -{
>>> - struct amdgpu_bo *abo = ttm_to_amdgpu_bo(bo);
>>> -
>>> - if (amdgpu_ttm_tt_get_usermm(bo->ttm))
>>> - return -EPERM;
>>> - return drm_vma_node_verify_access(&abo->tbo.base.vma_node,
>>> - filp->private_data);
>>> -}
>>> -
>>> /**
>>> * amdgpu_ttm_map_buffer - Map memory into the GART windows
>>> * @bo: buffer object to map
>>> @@ -1522,7 +1503,6 @@ static struct ttm_device_funcs
>>> amdgpu_bo_driver = {
>>> .eviction_valuable = amdgpu_ttm_bo_eviction_valuable,
>>> .evict_flags = &amdgpu_evict_flags,
>>> .move = &amdgpu_bo_move,
>>> - .verify_access = &amdgpu_verify_access,
>>> .delete_mem_notify = &amdgpu_bo_delete_mem_notify,
>>> .release_notify = &amdgpu_bo_release_notify,
>>> .io_mem_reserve = &amdgpu_ttm_io_mem_reserve,
>>> @@ -1897,50 +1877,6 @@ void
>>> amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool
>>> enable)
>>> adev->mman.buffer_funcs_enabled = enable;
>>> }
>>> -static vm_fault_t amdgpu_ttm_fault(struct vm_fault *vmf)
>>> -{
>>> - struct ttm_buffer_object *bo = vmf->vma->vm_private_data;
>>> - vm_fault_t ret;
>>> -
>>> - ret = ttm_bo_vm_reserve(bo, vmf);
>>> - if (ret)
>>> - return ret;
>>> -
>>> - ret = amdgpu_bo_fault_reserve_notify(bo);
>>> - if (ret)
>>> - goto unlock;
>>> -
>>> - ret = ttm_bo_vm_fault_reserved(vmf, vmf->vma->vm_page_prot,
>>> - TTM_BO_VM_NUM_PREFAULT, 1);
>>> - if (ret == VM_FAULT_RETRY && !(vmf->flags &
>>> FAULT_FLAG_RETRY_NOWAIT))
>>> - return ret;
>>> -
>>> -unlock:
>>> - dma_resv_unlock(bo->base.resv);
>>> - return ret;
>>> -}
>>> -
>>> -static const struct vm_operations_struct amdgpu_ttm_vm_ops = {
>>> - .fault = amdgpu_ttm_fault,
>>> - .open = ttm_bo_vm_open,
>>> - .close = ttm_bo_vm_close,
>>> - .access = ttm_bo_vm_access
>>> -};
>>> -
>>> -int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma)
>>> -{
>>> - struct drm_file *file_priv = filp->private_data;
>>> - struct amdgpu_device *adev = drm_to_adev(file_priv->minor->dev);
>>> - int r;
>>> -
>>> - r = ttm_bo_mmap(filp, vma, &adev->mman.bdev);
>>> - if (unlikely(r != 0))
>>> - return r;
>>> -
>>> - vma->vm_ops = &amdgpu_ttm_vm_ops;
>>> - return 0;
>>> -}
>>> -
>>> int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
>>> uint64_t dst_offset, uint32_t byte_count,
>>> struct dma_resv *resv,
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
>>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
>>> index dec0db8b0b13..6e51faad7371 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
>>> @@ -146,7 +146,6 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
>>> struct dma_resv *resv,
>>> struct dma_fence **fence);
>>> -int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma);
>>> int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo);
>>> int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo);
>>> uint64_t amdgpu_ttm_domain_start(struct amdgpu_device *adev,
>>> uint32_t type);
>>
>
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel
next prev parent reply other threads:[~2021-04-15 12:16 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-04-15 10:17 [PATCH v2 0/7] drm: Clean up mmap for TTM-based GEM drivers Thomas Zimmermann
2021-04-15 10:17 ` [PATCH v2 1/7] drm/ttm: Don't override vm_ops callbacks, if set Thomas Zimmermann
2021-04-15 13:21 ` Daniel Vetter
2021-04-15 10:17 ` [PATCH v2 2/7] drm/amdgpu: Implement mmap as GEM object function Thomas Zimmermann
2021-04-15 11:38 ` Christian König
2021-04-15 12:11 ` Thomas Zimmermann
2021-04-15 12:16 ` Christian König [this message]
2021-04-15 10:17 ` [PATCH v2 3/7] drm/radeon: " Thomas Zimmermann
2021-04-15 10:17 ` [PATCH v2 4/7] drm/nouveau: " Thomas Zimmermann
2021-04-15 10:17 ` [PATCH v2 5/7] drm/vmwgfx: Inline ttm_bo_mmap() into vmwgfx driver Thomas Zimmermann
2021-04-15 10:17 ` [PATCH v2 6/7] drm/vmwgfx: Inline vmw_verify_access() Thomas Zimmermann
2021-04-15 10:17 ` [PATCH v2 7/7] drm/ttm: Remove ttm_bo_mmap() and friends Thomas Zimmermann
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=5f669683-90fb-4a5a-0537-2a9f06cdeeb3@amd.com \
--to=christian.koenig@amd.com \
--cc=Felix.Kuehling@amd.com \
--cc=airlied@linux.ie \
--cc=alexander.deucher@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=bskeggs@redhat.com \
--cc=daniel@ffwll.ch \
--cc=dri-devel@lists.freedesktop.org \
--cc=emil.velikov@collabora.com \
--cc=linux-graphics-maintainer@vmware.com \
--cc=nirmoy.das@amd.com \
--cc=nouveau@lists.freedesktop.org \
--cc=ray.huang@amd.com \
--cc=sam@ravnborg.org \
--cc=shashank.sharma@amd.com \
--cc=sroland@vmware.com \
--cc=tzimmermann@suse.de \
--cc=zackr@vmware.com \
/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).