When an eGPU is unplugged the KFD topology should also be destroyed for that GPU. This never happens because the fini_sw callbacks never get to run. Run them manually before calling amdgpu_device_ip_fini_early() when a device has already been disconnected.
This location is intentionally chosen to make sure that the kfd locking refcount doesn't get incremented unintentionally. Cc: [email protected] Closes: https://community.frame.work/t/amd-egpu-on-linux/8691/33 Signed-off-by: Mario Limonciello (AMD) <[email protected]> --- v2: * Move the call earlier in amdgpu_device_fini_hw() to fix locking refcount issues --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 021ecc988ff79..f167ba1b6ffcb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -5251,6 +5251,14 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev) amdgpu_ttm_set_buffer_funcs_status(adev, false); + /* + * device went through surprise hotplug; we need to destroy topology + * before ip_fini_early to prevent kfd locking refcount issues by calling + * amdgpu_amdkfd_suspend() + */ + if (drm_dev_is_unplugged(adev_to_drm(adev))) + amdgpu_amdkfd_device_fini_sw(adev); + amdgpu_device_ip_fini_early(adev); amdgpu_irq_fini_hw(adev); -- 2.43.0
