Re: [PATCH v4 2/2] drm/amdgpu: Free VGA stolen memory as soon as possible.

[Alex Deucher <alexdeucher-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

On Thu, Apr 12, 2018 at 12:08 AM, Andrey Grodzovsky
<andrey.grodzovsky@amd.com> wrote:
> Reserved VRAM is used to avoid overriding pre OS FB.
> Once our display stack takes over we don't need the reserved
> VRAM anymore.
>
> v2:
> Remove comment, we know actually why we need to reserve the stolen VRAM.
> Fix return type for amdgpu_ttm_late_init.
> v3:
> Return 0 in amdgpu_bo_late_init, rebase on changes to previous patch
> v4:
> Don't release stolen memory for GMC9 ASICs untill GART corruption
> on S3 resume is resolved.
>
> Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Looks like you used the original version of this patch as well.
Updated version here:
https://patchwork.freedesktop.org/patch/215567/
more comments below.

> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_object.c |  7 +++++++
>  drivers/gpu/drm/amd/amdgpu/amdgpu_object.h |  1 +
>  drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c    |  8 ++++++--
>  drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h    |  1 +
>  drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c      |  2 ++
>  drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c      |  2 ++
>  drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c      |  2 ++
>  drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c      | 14 ++++++++++----
>  8 files changed, 31 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
> index 9e23d6f..a160ef0 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
> @@ -852,6 +852,13 @@ int amdgpu_bo_init(struct amdgpu_device *adev)
>         return amdgpu_ttm_init(adev);
>  }
>
> +int amdgpu_bo_late_init(struct amdgpu_device *adev)
> +{
> +       amdgpu_ttm_late_init(adev);
> +
> +       return 0;
> +}
> +
>  void amdgpu_bo_fini(struct amdgpu_device *adev)
>  {
>         amdgpu_ttm_fini(adev);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
> index 3bee133..1e9fe85 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
> @@ -251,6 +251,7 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
>  int amdgpu_bo_unpin(struct amdgpu_bo *bo);
>  int amdgpu_bo_evict_vram(struct amdgpu_device *adev);
>  int amdgpu_bo_init(struct amdgpu_device *adev);
> +int amdgpu_bo_late_init(struct amdgpu_device *adev);
>  void amdgpu_bo_fini(struct amdgpu_device *adev);
>  int amdgpu_bo_fbdev_mmap(struct amdgpu_bo *bo,
>                                 struct vm_area_struct *vma);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> index 0555821..7a608cf 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> @@ -1517,14 +1517,18 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
>         return 0;
>  }
>
> +void amdgpu_ttm_late_init(struct amdgpu_device *adev)
> +{
> +       if (adev->gmc.stolen_size)

no need to check for NULL here.  amdgpu_bo_free_kernel() will do the
right thing even if stolen_vga_memory is NULL.

> +               amdgpu_bo_free_kernel(&adev->stolen_vga_memory, NULL, NULL);
> +}
> +
>  void amdgpu_ttm_fini(struct amdgpu_device *adev)
>  {
>         if (!adev->mman.initialized)
>                 return;
>
>         amdgpu_ttm_debugfs_fini(adev);
> -       if (adev->gmc.stolen_size)
> -               amdgpu_bo_free_kernel(&adev->stolen_vga_memory, NULL, NULL);
>         amdgpu_ttm_fw_reserve_vram_fini(adev);
>         if (adev->mman.aper_base_kaddr)
>                 iounmap(adev->mman.aper_base_kaddr);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
> index 6ea7de8..e969c87 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
> @@ -77,6 +77,7 @@ uint64_t amdgpu_vram_mgr_usage(struct ttm_mem_type_manager *man);
>  uint64_t amdgpu_vram_mgr_vis_usage(struct ttm_mem_type_manager *man);
>
>  int amdgpu_ttm_init(struct amdgpu_device *adev);
> +void amdgpu_ttm_late_init(struct amdgpu_device *adev);
>  void amdgpu_ttm_fini(struct amdgpu_device *adev);
>  void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev,
>                                         bool enable);
> diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
> index 63f0b65..4a8f9bd 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
> @@ -819,6 +819,8 @@ static int gmc_v6_0_late_init(void *handle)
>  {
>         struct amdgpu_device *adev = (struct amdgpu_device *)handle;
>
> +       amdgpu_bo_late_init(adev);
> +
>         if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
>                 return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
>         else
> diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
> index 2deb5c9..189fdf9 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
> @@ -958,6 +958,8 @@ static int gmc_v7_0_late_init(void *handle)
>  {
>         struct amdgpu_device *adev = (struct amdgpu_device *)handle;
>
> +       amdgpu_bo_late_init(adev);
> +
>         if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
>                 return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
>         else
> diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
> index 04b00df..19e153f 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
> @@ -1049,6 +1049,8 @@ static int gmc_v8_0_late_init(void *handle)
>  {
>         struct amdgpu_device *adev = (struct amdgpu_device *)handle;
>
> +       amdgpu_bo_late_init(adev);
> +
>         if (amdgpu_vm_fault_stop != AMDGPU_VM_FAULT_STOP_ALWAYS)
>                 return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
>         else
> diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> index 252a6c69..099e3ce5 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> @@ -659,6 +659,16 @@ static int gmc_v9_0_late_init(void *handle)
>         unsigned i;
>         int r;
>
> +       /*
> +        * TODO:
> +        * Currently there is a bug where some memory client outside
> +        * of the driver writes to first 8M of VRAM on S3 resume,
> +        * this overrides GART which by default gets placed in first 8M and
> +        * causes VM_FAULTS once GTT is accessed.
> +        * Keep the stolen memory reservation until this solved.
> +        */
> +       /* amdgpu_bo_late_init(adev); /
> +

We still need to free this somewhere.  I'd suggest calling it in
gmc_v9_0_sw_fini() and add a comment there about moving it when we fix
the issue.

>         for(i = 0; i < adev->num_rings; ++i) {
>                 struct amdgpu_ring *ring = adev->rings[i];
>                 unsigned vmhub = ring->funcs->vmhub;
> @@ -884,10 +894,6 @@ static int gmc_v9_0_sw_init(void *handle)
>          */
>         adev->gmc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */
>
> -       /*
> -        * It needs to reserve 8M stolen memory for vega10
> -        * TODO: Figure out how to avoid that...
> -        */
>         adev->gmc.stolen_size = gmc_v9_0_get_vbios_fb_size(adev);

We may also just want to return 8MB or 9MB temporarily in
gmc_v9_0_get_vbios_fb_size until we sort out the root cause of the S3
issue otherwise we're potentially wasting a lot more memory.

Alex


>
>         /* set DMA mask + need_dma32 flags.
> --
> 2.7.4
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx