drm/amdgpu: Use the default reset when loading or reloading the driver
authorlyndonli <Lyndon.Li@amd.com>
Sun, 23 Apr 2023 09:05:15 +0000 (17:05 +0800)
committerAlex Deucher <alexander.deucher@amd.com>
Thu, 4 May 2023 02:43:42 +0000 (22:43 -0400)
Below call trace and errors are observed when reloading
amdgpu driver with the module parameter reset_method=3.

It should do a default reset when loading or reloading the
driver, regardless of the module parameter reset_method.

v2: add comments inside and modify commit messages.

[  +2.180243] [drm] psp gfx command ID_LOAD_TOC(0x20) failed
and response status is (0x0)
[  +0.000011] [drm:psp_hw_start [amdgpu]] *ERROR* Failed to load toc
[  +0.000890] [drm:psp_hw_start [amdgpu]] *ERROR* PSP tmr init failed!
[  +0.020683] [drm:amdgpu_fill_buffer [amdgpu]] *ERROR* Trying to
clear memory with ring turned off.
[  +0.000003] RIP: 0010:amdgpu_bo_release_notify+0x1ef/0x210 [amdgpu]
[  +0.000004] Call Trace:
[  +0.000003]  <TASK>
[  +0.000008]  ttm_bo_release+0x2c4/0x330 [amdttm]
[  +0.000026]  amdttm_bo_put+0x3c/0x70 [amdttm]
[  +0.000020]  amdgpu_bo_free_kernel+0xe6/0x140 [amdgpu]
[  +0.000728]  psp_v11_0_ring_destroy+0x34/0x60 [amdgpu]
[  +0.000826]  psp_hw_init+0xe7/0x2f0 [amdgpu]
[  +0.000813]  amdgpu_device_fw_loading+0x1ad/0x2d0 [amdgpu]
[  +0.000731]  amdgpu_device_init.cold+0x108e/0x2002 [amdgpu]
[  +0.001071]  ? do_pci_enable_device+0xe1/0x110
[  +0.000011]  amdgpu_driver_load_kms+0x1a/0x160 [amdgpu]
[  +0.000729]  amdgpu_pci_probe+0x179/0x3a0 [amdgpu]

Signed-off-by: lyndonli <Lyndon.Li@amd.com>
Signed-off-by: Yunxiang Li <Yunxiang.Li@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

index 9b1eaba..981a9cf 100644 (file)
@@ -3578,6 +3578,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
        int r, i;
        bool px = false;
        u32 max_MBps;
+       int tmp;
 
        adev->shutdown = false;
        adev->flags = flags;
@@ -3799,7 +3800,13 @@ int amdgpu_device_init(struct amdgpu_device *adev,
                                }
                        }
                } else {
+                       tmp = amdgpu_reset_method;
+                       /* It should do a default reset when loading or reloading the driver,
+                        * regardless of the module parameter reset_method.
+                        */
+                       amdgpu_reset_method = AMD_RESET_METHOD_NONE;
                        r = amdgpu_asic_reset(adev);
+                       amdgpu_reset_method = tmp;
                        if (r) {
                                dev_err(adev->dev, "asic reset on init failed\n");
                                goto failed;