[PATCH] drm/amdgpu: Disable GPU reset on SRIOV before remove pci.

Gavin Wan Gavin.Wan at amd.com
Tue Oct 25 14:16:00 UTC 2022


  The change of the commit <f5c7e7797060> ("Adjust removal control
  flow for smu v13_0_2") brought a bug on SRIOV envrionment. It
  caused unloading amdgpu failed on Guest VM. The reason is that
  the VF FLR was requested while unloading amdgpu driver, but the
  VF FLR of SRIOV sequence is wrong while removing PCI device.

Signed-off-by: Gavin Wan <Gavin.Wan at amd.com>
Fixes: f5c7e7797060 ("drm/amdgpu: Adjust removal control flow
       for smu v13_0_2")
Acked-by: Alex Deucher <alexander.deucher at amd.com>
Change-Id: I1ff8dcbffd85d7f3d8267d660fd8292423d2f70f
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 16f6a313335e..8e97e95aca8c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -2187,7 +2187,8 @@ amdgpu_pci_remove(struct pci_dev *pdev)
 		pm_runtime_forbid(dev->dev);
 	}
 
-	if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2)) {
+	if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2) &&
+	    !amdgpu_sriov_vf(adev)) {
 		bool need_to_reset_gpu = false;
 
 		if (adev->gmc.xgmi.num_physical_nodes > 1) {
-- 
2.34.1



More information about the amd-gfx mailing list