[PATCH] drm/amdgpu: grab the id mgr lock while accessing passid_mapping

Chunming Zhou zhoucm1 at amd.com
Tue Sep 10 12:22:12 UTC 2019


Reviewed-by: Chunming Zhou <david1.zhou at amd.com>

在 2019/9/10 16:56, Christian König 写道:
> Ping!
>
> Am 09.09.19 um 13:59 schrieb Christian König:
>> Need to make sure that we actually dropping the right fence.
>> Could be done with RCU as well, but to complicated for a fix.
>>
>> Signed-off-by: Christian König <christian.koenig at amd.com>
>> ---
>>   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 12 +++++++++---
>>   1 file changed, 9 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c 
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
>> index b285ab25146d..e11764164cbf 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
>> @@ -1036,10 +1036,8 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, 
>> struct amdgpu_job *job, bool need_
>>           id->oa_base != job->oa_base ||
>>           id->oa_size != job->oa_size);
>>       bool vm_flush_needed = job->vm_needs_flush;
>> -    bool pasid_mapping_needed = id->pasid != job->pasid ||
>> -        !id->pasid_mapping ||
>> -        !dma_fence_is_signaled(id->pasid_mapping);
>>       struct dma_fence *fence = NULL;
>> +    bool pasid_mapping_needed;
>>       unsigned patch_offset = 0;
>>       int r;
>>   @@ -1049,6 +1047,12 @@ int amdgpu_vm_flush(struct amdgpu_ring 
>> *ring, struct amdgpu_job *job, bool need_
>>           pasid_mapping_needed = true;
>>       }
>>   +    mutex_lock(&id_mgr->lock);
>> +    if (id->pasid != job->pasid || !id->pasid_mapping ||
>> +        !dma_fence_is_signaled(id->pasid_mapping))
>> +        pasid_mapping_needed = true;
>> +    mutex_unlock(&id_mgr->lock);
>> +
>>       gds_switch_needed &= !!ring->funcs->emit_gds_switch;
>>       vm_flush_needed &= !!ring->funcs->emit_vm_flush &&
>>               job->vm_pd_addr != AMDGPU_BO_INVALID_OFFSET;
>> @@ -1088,9 +1092,11 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, 
>> struct amdgpu_job *job, bool need_
>>       }
>>         if (pasid_mapping_needed) {
>> +        mutex_lock(&id_mgr->lock);
>>           id->pasid = job->pasid;
>>           dma_fence_put(id->pasid_mapping);
>>           id->pasid_mapping = dma_fence_get(fence);
>> +        mutex_unlock(&id_mgr->lock);
>>       }
>>       dma_fence_put(fence);
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx


More information about the amd-gfx mailing list