Am 13.04.22 um 21:31 schrieb Andrey Grodzovsky:
Lock reset domain unconditionally because on resume
we unlock it unconditionally.
This solved mutex deadlock when handling both FATAL
and non FATAL PCI errors one after another.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@xxxxxxx>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 1cc488a767d8..c65f25e3a0fc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -5531,18 +5531,18 @@ pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_sta
adev->pci_channel_state = state;
+ /*
+ * Locking adev->reset_domain->sem will prevent any external access
+ * to GPU during PCI error recovery
+ */
+ amdgpu_device_lock_reset_domain(adev->reset_domain);
+ amdgpu_device_set_mp1_state(adev);
+
switch (state) {
case pci_channel_io_normal:
return PCI_ERS_RESULT_CAN_RECOVER;
BTW: Where are we unlocking that again?
/* Fatal error, prepare for slot reset */
case pci_channel_io_frozen:
- /*
- * Locking adev->reset_domain->sem will prevent any external access
- * to GPU during PCI error recovery
- */
- amdgpu_device_lock_reset_domain(adev->reset_domain);
- amdgpu_device_set_mp1_state(adev);
-
/*
* Block any work scheduling as we do for regular GPU reset
* for the duration of the recovery
