Re: [PATCH 2/5] drm/amdkfd: check child range to drain retry fault

[Felix Kuehling <felix.kuehling@amd.com>]

On 2021-11-09 6:04 p.m., Philip Yang wrote:
> If unmapping partial range, the parent prange list op is update
> notifier, child range list op is unmap range, need check child range to
> set drain retry fault flag.
>
> Signed-off-by: Philip Yang <Philip.Yang@amd.com>

I think this could be simplified by simply setting 
svms->drain_pagefaults in svm_range_unmap_from_cpu. The mmap lock 
ensures that this is serialized with the deferred list worker reading 
and clearing svms->drain_pagefaults. You can also use READ_ONCE and 
WRITE_ONCE to be safe.

Regards,
   Felix


> ---
>   drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 13 ++++++++++++-
>   1 file changed, 12 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
> index 77239b06b236..64f642935600 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
> @@ -2049,8 +2049,19 @@ svm_range_add_list_work(struct svm_range_list *svms, struct svm_range *prange,
>   	 * before the range is freed to avoid straggler interrupts on
>   	 * unmapped memory causing "phantom faults".
>   	 */
> -	if (op == SVM_OP_UNMAP_RANGE)
> +	if (op == SVM_OP_UNMAP_RANGE) {
> +		pr_debug("set range drain_pagefaults true\n");
>   		svms->drain_pagefaults = true;
> +	} else {
> +		struct svm_range *pchild;
> +
> +		list_for_each_entry(pchild, &prange->child_list, child_list)
> +			if (pchild->work_item.op == SVM_OP_UNMAP_RANGE) {
> +				pr_debug("set child drain_pagefaults true\n");
> +				svms->drain_pagefaults = true;
> +			}
> +	}
> +
>   	/* if prange is on the deferred list */
>   	if (!list_empty(&prange->deferred_list)) {
>   		pr_debug("update exist prange 0x%p work op %d\n", prange, op);