Re: [PATCH 1/2] drm/sched: add drm_sched_prealloc_dependency_slots

["Christian König" <ckoenig.leichtzumerken@gmail.com>]

Am 18.03.25 um 13:39 schrieb Danilo Krummrich:
> On Tue, Mar 18, 2025 at 01:03:12PM +0100, Christian König wrote:
>>  /**
>>   * drm_sched_job_add_dependency - adds the fence as a job dependency
>>   * @job: scheduler job to add the dependencies to
>> @@ -878,10 +910,12 @@ int drm_sched_job_add_dependency(struct drm_sched_job *job,
>>  	 * engines involved, rather than the number of BOs.
>>  	 */
>>  	xa_for_each(&job->dependencies, index, entry) {
>> -		if (entry->context != fence->context)
>> +		bool signaled = dma_fence_is_signaled(entry);
>> +
>> +		if (!signaled && entry->context != fence->context)
>>  			continue;
>>  
>> -		if (dma_fence_is_later(fence, entry)) {
>> +		if (signaled || dma_fence_is_later(fence, entry)) {
>>  			dma_fence_put(entry);
>>  			xa_store(&job->dependencies, index, fence, GFP_KERNEL);
>>  		} else {
>> @@ -890,7 +924,8 @@ int drm_sched_job_add_dependency(struct drm_sched_job *job,
>>  		return 0;
>>  	}
>>  
>> -	ret = xa_alloc(&job->dependencies, &id, fence, xa_limit_32b, GFP_KERNEL);
>> +	ret = xa_alloc(&job->dependencies, &id, fence, xa_limit_32b,
>> +		       GFP_KERNEL);
>>  	if (ret != 0)
>>  		dma_fence_put(fence);
> Those changes seem unrelated, aren't they?

Ah, yes that was just a leftover from a previous try to fix this.

Thanks,
Christian.