[RFC v3 12/12] Revert 'drm/amdgpu: annotate a false positive recursive locking'

Andrey Grodzovsky andrey.grodzovsky at amd.com
Tue Jan 25 22:37:52 UTC 2022


Since we have a single instance of reset semaphore which we
lock only once even for XGMI hive we don't need the nested
locking hint anymore.

Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 14 ++++----------
 1 file changed, 4 insertions(+), 10 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 31310922b6bf..b97992d62db5 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -4771,16 +4771,10 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle,
 	return r;
 }
 
-static void amdgpu_device_lock_reset_domain(struct amdgpu_reset_domain *reset_domain,
-					    struct amdgpu_hive_info *hive)
+static void amdgpu_device_lock_reset_domain(struct amdgpu_reset_domain *reset_domain)
 {
 	atomic_set(&reset_domain->in_gpu_reset, 1);
-
-	if (hive) {
-		down_write_nest_lock(&reset_domain->sem, &hive->hive_lock);
-	} else {
-		down_write(&reset_domain->sem);
-	}
+	down_write(&reset_domain->sem);
 }
 
 static void amdgpu_device_set_mp1_state(struct amdgpu_device *adev)
@@ -5017,7 +5011,7 @@ int amdgpu_device_gpu_recover_imp(struct amdgpu_device *adev,
 	/* We need to lock reset domain only once both for XGMI and single device */
 	tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device,
 				    reset_list);
-	amdgpu_device_lock_reset_domain(tmp_adev->reset_domain, hive);
+	amdgpu_device_lock_reset_domain(tmp_adev->reset_domain);
 
 	/* block all schedulers and reset given job's ring */
 	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
@@ -5441,7 +5435,7 @@ pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_sta
 		 * Locking adev->reset_domain->sem will prevent any external access
 		 * to GPU during PCI error recovery
 		 */
-		amdgpu_device_lock_reset_domain(adev->reset_domain, NULL);
+		amdgpu_device_lock_reset_domain(adev->reset_domain);
 		amdgpu_device_set_mp1_state(adev);
 
 		/*
-- 
2.25.1



More information about the dri-devel mailing list