Minki/linux
1
0
Fork 1
mirror of https://github.com/torvalds/linux.git synced 2024-12-02 00:51:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

drm/amdgpu: Use per device reset_domain for XGMI on sriov configuration

Browse Source 
For SRIOV configuration, host driver control the reset method(either FLR or
heavier chain reset). The host will notify the guest individually with FLR
message if individual GPU within the hive need to be reset. So for guest
side, no need to use hive->reset_domain to replace the original per
device reset_domain

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
shaoyunl 2022-09-06 23:05:11 -04:00 committed by Alex Deucher
parent 1ed1f6be6e
commit 46c676600c
2 changed files with 35 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
View File

@ -2453,17 +2453,19 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
if (amdgpu_xgmi_add_device(adev) == 0) {
struct amdgpu_hive_info *hive = amdgpu_get_xgmi_hive(adev);
if (!hive->reset_domain ||
!amdgpu_reset_get_reset_domain(hive->reset_domain)) {
r = -ENOENT;
amdgpu_put_xgmi_hive(hive);
goto init_failed;
}
if (!amdgpu_sriov_vf(adev)) {
if (!hive->reset_domain ||
!amdgpu_reset_get_reset_domain(hive->reset_domain)) {
r = -ENOENT;
amdgpu_put_xgmi_hive(hive);
goto init_failed;
}
/* Drop the early temporary reset domain we created for device */
amdgpu_reset_put_reset_domain(adev->reset_domain);
adev->reset_domain = hive->reset_domain;
amdgpu_put_xgmi_hive(hive);
/* Drop the early temporary reset domain we created for device */
amdgpu_reset_put_reset_domain(adev->reset_domain);
adev->reset_domain = hive->reset_domain;
amdgpu_put_xgmi_hive(hive);
}
}
}

37
drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
View File

@ -391,24 +391,33 @@ struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev)
goto pro_end;
}
/**
* Only init hive->reset_domain for none SRIOV configuration. For SRIOV,
* Host driver decide how to reset the GPU either through FLR or chain reset.
* Guest side will get individual notifications from the host for the FLR
* if necessary.
*/
if (!amdgpu_sriov_vf(adev)) {
/**
* Avoid recreating reset domain when hive is reconstructed for the case
* of reset the devices in the XGMI hive during probe for SRIOV
* of reset the devices in the XGMI hive during probe for passthrough GPU
* See https://www.spinics.net/lists/amd-gfx/msg58836.html
*/
if (adev->reset_domain->type != XGMI_HIVE) {
hive->reset_domain = amdgpu_reset_create_reset_domain(XGMI_HIVE, "amdgpu-reset-hive");
if (!hive->reset_domain) {
dev_err(adev->dev, "XGMI: failed initializing reset domain for xgmi hive\n");
ret = -ENOMEM;
kobject_put(&hive->kobj);
kfree(hive);
hive = NULL;
goto pro_end;
}
} else {
amdgpu_reset_get_reset_domain(adev->reset_domain);
hive->reset_domain = adev->reset_domain;
if (adev->reset_domain->type != XGMI_HIVE) {
hive->reset_domain =
amdgpu_reset_create_reset_domain(XGMI_HIVE, "amdgpu-reset-hive");
if (!hive->reset_domain) {
dev_err(adev->dev, "XGMI: failed initializing reset domain for xgmi hive\n");
ret = -ENOMEM;
kobject_put(&hive->kobj);
kfree(hive);
hive = NULL;
goto pro_end;
}
} else {
amdgpu_reset_get_reset_domain(adev->reset_domain);
hive->reset_domain = adev->reset_domain;
}
}
hive->hive_id = adev->gmc.xgmi.hive_id;