From 028cfb2444b94d4f394a6fa4ca46182481236e91 Mon Sep 17 00:00:00 2001 From: Evan Quan Date: Fri, 10 Apr 2020 15:38:44 +0800 Subject: drm/amdgpu: fix wrong vram lost counter increment V2 Vram lost counter is wrongly increased by two during baco reset. V2: assumed vram lost for mode1 reset on all ASICs Signed-off-by: Evan Quan Acked-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_device.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 559dc24ef436..7d35b0a366a2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -2008,8 +2008,24 @@ static void amdgpu_device_fill_reset_magic(struct amdgpu_device *adev) */ static bool amdgpu_device_check_vram_lost(struct amdgpu_device *adev) { - return !!memcmp(adev->gart.ptr, adev->reset_magic, - AMDGPU_RESET_MAGIC_NUM); + if (memcmp(adev->gart.ptr, adev->reset_magic, + AMDGPU_RESET_MAGIC_NUM)) + return true; + + if (!adev->in_gpu_reset) + return false; + + /* + * For all ASICs with baco/mode1 reset, the VRAM is + * always assumed to be lost. + */ + switch (amdgpu_asic_reset_method(adev)) { + case AMD_RESET_METHOD_BACO: + case AMD_RESET_METHOD_MODE1: + return true; + default: + return false; + } } /** -- cgit v1.2.3 From b2a7e9735ab2864330be9d00d7f38c961c28de5d Mon Sep 17 00:00:00 2001 From: Prike Liang Date: Mon, 13 Apr 2020 21:41:14 +0800 Subject: drm/amdgpu: fix the hw hang during perform system reboot and reset The system reboot failed as some IP blocks enter power gate before perform hw resource destory. Meanwhile use unify interface to set device CGPG to ungate state can simplify the amdgpu poweroff or reset ungate guard. Fixes: 487eca11a321ef ("drm/amdgpu: fix gfx hang during suspend with video playback (v2)") Signed-off-by: Prike Liang Tested-by: Mengbing Wang Tested-by: Paul Menzel Acked-by: Alex Deucher Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_device.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 7d35b0a366a2..f84f9e35a73b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -2356,6 +2356,8 @@ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev) { int i, r; + amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE); + amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE); for (i = adev->num_ip_blocks - 1; i >= 0; i--) { if (!adev->ip_blocks[i].status.valid) -- cgit v1.2.3 From c457a273e118bb96e1db8d1825f313e6cafe4258 Mon Sep 17 00:00:00 2001 From: Evan Quan Date: Fri, 24 Apr 2020 15:32:41 +0800 Subject: drm/amdgpu: move kfd suspend after ip_suspend_phase1 This sequence change should be safe as what did in ip_suspend_phase1 is to suspend DCE only. And this is a prerequisite for coming redundant cg/pg ungate dropping. Fixes: 487eca11a321ef ("drm/amdgpu: fix gfx hang during suspend with video playback (v2)") Signed-off-by: Evan Quan Acked-by: Alex Deucher Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_device.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index f84f9e35a73b..957d4872a855 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -3375,12 +3375,12 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon) amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE); amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE); - amdgpu_amdkfd_suspend(adev, !fbcon); - amdgpu_ras_suspend(adev); r = amdgpu_device_ip_suspend_phase1(adev); + amdgpu_amdkfd_suspend(adev, !fbcon); + /* evict vram memory */ amdgpu_bo_evict_vram(adev); -- cgit v1.2.3 From f7b52890daba570bc8162d43c96b5583bbdd4edd Mon Sep 17 00:00:00 2001 From: Evan Quan Date: Fri, 24 Apr 2020 15:36:22 +0800 Subject: drm/amdgpu: drop redundant cg/pg ungate on runpm enter CG/PG ungate is already performed in ip_suspend_phase1. Otherwise, the CG/PG ungate will be performed twice. That will cause gfxoff disablement is performed twice also on runpm enter while gfxoff enablemnt once on rump exit. That will put gfxoff into disabled state. Fixes: b2a7e9735ab286 ("drm/amdgpu: fix the hw hang during perform system reboot and reset") Signed-off-by: Evan Quan Acked-by: Alex Deucher Signed-off-by: Alex Deucher Cc: stable@vger.kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'drivers/gpu/drm/amd/amdgpu/amdgpu_device.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 957d4872a855..affde2de2a0d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -3372,9 +3372,6 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon) } } - amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE); - amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE); - amdgpu_ras_suspend(adev); r = amdgpu_device_ip_suspend_phase1(adev); -- cgit v1.2.3