[PATCH 4/7] drm/amdgpu: Fix SMU error failure
Andrey Grodzovsky
andrey.grodzovsky at amd.com
Wed Aug 26 14:46:19 UTC 2020
Wait for HW/PSP initiated ASIC reset to complete before
starting the recovery operations.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky at amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 22 ++++++++++++++++++++--
1 file changed, 20 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 0a2130b..d9e3994 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -4846,13 +4846,31 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
{
struct drm_device *dev = pci_get_drvdata(pdev);
struct amdgpu_device *adev = drm_to_adev(dev);
- int r;
+ int r, i;
bool vram_lost;
+ u32 memsize;
DRM_INFO("PCI error: slot reset callback!!\n");
+ /* wait for asic to come out of reset */
+ msleep(500);
+
pci_restore_state(pdev);
+ /* confirm ASIC came out of reset */
+ for (i = 0; i < adev->usec_timeout; i++) {
+ memsize = amdgpu_asic_get_config_memsize(adev);
+
+ if (memsize != 0xffffffff)
+ break;
+ udelay(1);
+ }
+ if (memsize == 0xffffffff) {
+ r = -ETIME;
+ goto out;
+ }
+
+ /* TODO Call amdgpu_pre_asic_reset instead */
adev->in_dpc = true;
r = amdgpu_device_ip_suspend(adev);
adev->in_dpc = false;
@@ -4882,7 +4900,7 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
r = amdgpu_device_fw_loading(adev);
if (r)
- return r;
+ goto out;
r = amdgpu_device_ip_resume_phase2(adev);
if (r)
--
2.7.4
More information about the amd-gfx
mailing list