[PATCH v5 5/5] drm/xe: Refactor default device atomic settings

[Nirmoy Das]

Hi Jose,

On 4/26/2024 11:04 PM, Souza, Jose wrote:
> On Fri, 2024-04-26 at 12:56 +0200, Nirmoy Das wrote:
>> The default behavior of device atomics depends on the
>> VM type and buffer allocation types. Device atomics are
>> expected to function with all types of allocations for
>> traditional applications/APIs. Additionally, in compute/SVM
>> API scenarios with fault mode or LR mode VMs, device atomics
>> must work with single-region allocations. In all other cases
>> device atomics should be disabled by default also on platforms
>> where we know device atomics doesn't on work on particular
>> allocations types.
>>
>> v2: Fix platform checks to correct atomics behaviour on PVC.
>>
>> Signed-off-by: Nirmoy Das <nirmoy.das at intel.com>
>> ---
>>   drivers/gpu/drm/xe/xe_pt.c | 27 ++++++++++++++++++++++++---
>>   drivers/gpu/drm/xe/xe_vm.c |  2 +-
>>   2 files changed, 25 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/xe/xe_pt.c b/drivers/gpu/drm/xe/xe_pt.c
>> index 5b7930f46cf3..237e4a4985a4 100644
>> --- a/drivers/gpu/drm/xe/xe_pt.c
>> +++ b/drivers/gpu/drm/xe/xe_pt.c
>> @@ -619,9 +619,30 @@ xe_pt_stage_bind(struct xe_tile *tile, struct xe_vma *vma,
>>   	struct xe_pt *pt = xe_vma_vm(vma)->pt_root[tile->id];
>>   	int ret;
>>   
>> -	if ((vma->gpuva.flags & XE_VMA_ATOMIC_PTE_BIT) &&
>> -	    (is_devmem || !IS_DGFX(xe)))
>> -		xe_walk.default_pte |= XE_USM_PPGTT_PTE_AE;
>> +	/**
>> +	 * Default atomic expectations for different allocation scenarios are as follows:
>> +	 *
>> +	 * 1. Traditional API: When the VM is not in fault mode or LR mode:
>> +	 *    - Device atomics are expected to function with all allocations.
>> +	 *
>> +	 * 2. Compute/SVM API: When the VM is either in fault mode or LR mode:
>> +	 *    - Device atomics are the default behavior when the bo is placed in a single region.
>> +	 *    - In all other cases device atomics will be disabled with AE=0 until an application
>> +	 *      request differently using a ioctl like madvise.
>> +	 */
>> +	if (vma->gpuva.flags & XE_VMA_ATOMIC_PTE_BIT) {
>> +		if (xe_vm_in_fault_mode(xe_vma_vm(vma)) ||
>> +		    xe_vm_in_lr_mode(xe_vma_vm(vma))) {
> nit: xe_vm_in_fault_mode requires xe_vm_in_lr_mode, so you can just check for the later.

Will do that. I think scratch page is also part of non-traditional API, 
I will confirm that and add that if needed.


Thanks,

Nirmoy

>
>> +			if (bo && xe_bo_has_single_placement(bo))
>> +				xe_walk.default_pte |= XE_USM_PPGTT_PTE_AE;
>> +		} else {
>> +			xe_walk.default_pte |= XE_USM_PPGTT_PTE_AE;
>> +		}
>> +
>> +		/* Unset AE if the platform(PVC) doesn't support it */
>> +		if (!xe->info.has_device_atomics_on_smem && !is_devmem)
>> +			xe_walk.default_pte &= ~XE_USM_PPGTT_PTE_AE;
>> +	}
>>   
>>   	if (is_devmem) {
>>   		xe_walk.default_pte |= XE_PPGTT_PTE_DM;
>> diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
>> index 8fc37c5a0196..f795016a80d5 100644
>> --- a/drivers/gpu/drm/xe/xe_vm.c
>> +++ b/drivers/gpu/drm/xe/xe_vm.c
>> @@ -805,7 +805,7 @@ static struct xe_vma *xe_vma_create(struct xe_vm *vm,
>>   	for_each_tile(tile, vm->xe, id)
>>   		vma->tile_mask |= 0x1 << id;
>>   
>> -	if (GRAPHICS_VER(vm->xe) >= 20 || vm->xe->info.platform == XE_PVC)
>> +	if (vm->xe->info.has_atomic_enable_pte_bit)
>>   		vma->gpuva.flags |= XE_VMA_ATOMIC_PTE_BIT;
>>   
>>   	vma->pat_index = pat_index;