Am 18.04.2018 um 06:14 schrieb Alex Deucher:
> On Tue, Apr 17, 2018 at 8:40 PM, Samuel Li <Samuel.Li at amd.com> wrote:
>> It's auto by default. For CZ/ST, auto setting enables sg display
>> when vram size is small; otherwise still uses vram.
>> This patch fixed some potention hang issue introduced by change
>> "allow framebuffer in GART memory as well" due to CZ/ST hardware
>> limitation.
>>
>> v2: Change default setting to auto.
>> v3: Move some logic from amdgpu_display_framebuffer_domains()
>> to pin function, suggested by Christian.
>> Signed-off-by: Samuel Li <Samuel.Li at amd.com>
>> ---
>> drivers/gpu/drm/amd/amdgpu/amdgpu.h | 2 ++
>> drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 4 ++--
>> drivers/gpu/drm/amd/amdgpu/amdgpu_display.h | 2 +-
>> drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 4 ++++
>> drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c | 2 +-
>> drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 25 +++++++++++++++++++++--
>> drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c | 2 +-
>> drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 3 +--
>> 8 files changed, 35 insertions(+), 9 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>> index b3d047d..26429de 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
>> @@ -129,6 +129,7 @@ extern int amdgpu_lbpw;
>> extern int amdgpu_compute_multipipe;
>> extern int amdgpu_gpu_recovery;
>> extern int amdgpu_emu_mode;
>> +extern int amdgpu_sg_display;
>>
>> #ifdef CONFIG_DRM_AMDGPU_SI
>> extern int amdgpu_si_support;
>> @@ -137,6 +138,7 @@ extern int amdgpu_si_support;
>> extern int amdgpu_cik_support;
>> #endif
>>
>> +#define AMDGPU_SG_THRESHOLD (256*1024*1024)
>> #define AMDGPU_DEFAULT_GTT_SIZE_MB 3072ULL /* 3GB by default */
>> #define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS 3000
>> #define AMDGPU_MAX_USEC_TIMEOUT 100000 /* 100 ms */
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
>> index 50f98df..0caa3d2 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
>> @@ -189,7 +189,7 @@ int amdgpu_display_crtc_page_flip_target(struct drm_crtc *crtc,
>> goto cleanup;
>> }
>>
>> - r = amdgpu_bo_pin(new_abo, amdgpu_display_framebuffer_domains(adev), &base);
>> + r = amdgpu_bo_pin(new_abo, amdgpu_display_supported_domains(adev), &base);
>> if (unlikely(r != 0)) {
>> DRM_ERROR("failed to pin new abo buffer before flip\n");
>> goto unreserve;
>> @@ -484,7 +484,7 @@ static const struct drm_framebuffer_funcs amdgpu_fb_funcs = {
>> .create_handle = drm_gem_fb_create_handle,
>> };
>>
>> -uint32_t amdgpu_display_framebuffer_domains(struct amdgpu_device *adev)
>> +uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev)
> This change should be a separate patch,
>
>> {
>> uint32_t domain = AMDGPU_GEM_DOMAIN_VRAM;
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h
>> index 2b11d80..f66e3e3 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h
>> @@ -23,7 +23,7 @@
>> #ifndef __AMDGPU_DISPLAY_H__
>> #define __AMDGPU_DISPLAY_H__
>>
>> -uint32_t amdgpu_display_framebuffer_domains(struct amdgpu_device *adev);
>> +uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev);
>> struct drm_framebuffer *
>> amdgpu_display_user_framebuffer_create(struct drm_device *dev,
>> struct drm_file *file_priv,
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
>> index 0b19482..85dcd1c 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
>> @@ -132,6 +132,7 @@ int amdgpu_lbpw = -1;
>> int amdgpu_compute_multipipe = -1;
>> int amdgpu_gpu_recovery = -1; /* auto */
>> int amdgpu_emu_mode = 0;
>> +int amdgpu_sg_display = -1;
>>
>> MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes");
>> module_param_named(vramlimit, amdgpu_vram_limit, int, 0600);
>> @@ -290,6 +291,9 @@ module_param_named(gpu_recovery, amdgpu_gpu_recovery, int, 0444);
>> MODULE_PARM_DESC(emu_mode, "Emulation mode, (1 = enable, 0 = disable)");
>> module_param_named(emu_mode, amdgpu_emu_mode, int, 0444);
>>
>> +MODULE_PARM_DESC(sg_display, "Enable scatter gather display, (1 = enable, 0 = disable, -1 = auto");
>> +module_param_named(sg_display, amdgpu_sg_display, int, 0444);
>> +
>> #ifdef CONFIG_DRM_AMDGPU_SI
>>
>> #if defined(CONFIG_DRM_RADEON) || defined(CONFIG_DRM_RADEON_MODULE)
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
>> index ff89e84..bc5fd8e 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
>> @@ -137,7 +137,7 @@ static int amdgpufb_create_pinned_object(struct amdgpu_fbdev *rfbdev,
>> /* need to align pitch with crtc limits */
>> mode_cmd->pitches[0] = amdgpu_align_pitch(adev, mode_cmd->width, cpp,
>> fb_tiled);
>> - domain = amdgpu_display_framebuffer_domains(adev);
>> + domain = amdgpu_display_supported_domains(adev);
>>
>> height = ALIGN(mode_cmd->height, 8);
>> size = mode_cmd->pitches[0] * height;
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
>> index 24f582c..f0f1f8a 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
>> @@ -689,8 +689,29 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
>> return -EINVAL;
>>
>> /* A shared bo cannot be migrated to VRAM */
>> - if (bo->prime_shared_count && (domain == AMDGPU_GEM_DOMAIN_VRAM))
>> - return -EINVAL;
>> + if (bo->prime_shared_count) {
>> + if (domain & AMDGPU_GEM_DOMAIN_GTT)
>> + domain = AMDGPU_GEM_DOMAIN_GTT;
>> + else
>> + return -EINVAL;
>> + }
> This is a bug fix and should be split out into a separate patch.
>
>> +
>> + /* display buffer */
>> +#if defined(CONFIG_DRM_AMD_DC)
Please drop that #if, we certainly don't want this to be depend on any
define.
>> + if (adev->asic_type >= CHIP_CARRIZO && adev->asic_type < CHIP_RAVEN &&
>> + adev->flags & AMD_IS_APU &&
>> + amdgpu_device_asic_has_dc_support(adev->asic_type) &&
Those checks are static and don't depend on the BO. So they should be in
amdgpu_display_supported_domains().
Regards,
Christian.
>> + domain == (AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT)) {
>> + if (amdgpu_sg_display == 1)
>> + domain = AMDGPU_GEM_DOMAIN_GTT;
>> + else if (amdgpu_sg_display == -1) {
>> + if (adev->gmc.real_vram_size < AMDGPU_SG_THRESHOLD)
>> + domain = AMDGPU_GEM_DOMAIN_GTT;
>> + else
>> + domain = AMDGPU_GEM_DOMAIN_VRAM;
>> + }
> I thought we were dropping the module parameter. Also, we had talked
> about taking preferred domains into account here as well, but that can
> be a follow on patch.
>
> Alex
>
>> + }
>> +#endif
>>
>> if (bo->pin_count) {
>> uint32_t mem_type = bo->tbo.mem.mem_type;
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
>> index 4b584cb7..cf0749f 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
>> @@ -209,7 +209,7 @@ static int amdgpu_gem_begin_cpu_access(struct dma_buf *dma_buf,
>> struct amdgpu_bo *bo = gem_to_amdgpu_bo(dma_buf->priv);
>> struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
>> struct ttm_operation_ctx ctx = { true, false };
>> - u32 domain = amdgpu_display_framebuffer_domains(adev);
>> + u32 domain = amdgpu_display_supported_domains(adev);
>> int ret;
>> bool reads = (direction == DMA_BIDIRECTIONAL ||
>> direction == DMA_FROM_DEVICE);
>> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> index 6f92a19..1f5603a 100644
>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> @@ -3109,12 +3109,11 @@ static int dm_plane_helper_prepare_fb(struct drm_plane *plane,
>> return r;
>>
>> if (plane->type != DRM_PLANE_TYPE_CURSOR)
>> - domain = amdgpu_display_framebuffer_domains(adev);
>> + domain = amdgpu_display_supported_domains(adev);
>> else
>> domain = AMDGPU_GEM_DOMAIN_VRAM;
>>
>> r = amdgpu_bo_pin(rbo, domain, &afb->address);
>> -
>> amdgpu_bo_unreserve(rbo);
>>
>> if (unlikely(r != 0)) {
>> --
>> 2.7.4
>>
>> _______________________________________________
>> amd-gfx mailing list
>> amd-gfx at lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
> _______________________________________________
> amd-gfx mailing list
> amd-gfx at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
