[Intel-gfx] [FIXES 2/3] drm/i915/userptr: Handle unlocked gup retries

[Tvrtko Ursulin]

On 11/11/2019 13:32, Chris Wilson wrote:
> Enable gup to retry and fault the pages outside of the mmap_sem lock in
> our worker. As we are inside our worker, outside of any critical path,
> we can allow the mmap_sem lock to be dropped in order to service a page
> fault; this in turn allows the mm to populate the page using a slow
> fault handler.
> 
> Testcase: igt/gem_userptr/userfault

There are no references or explanation on what does this fix?

Regards,

Tvrtko

> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
> ---
>   drivers/gpu/drm/i915/gem/i915_gem_userptr.c | 11 ++++++++---
>   1 file changed, 8 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
> index dd104b0e2071..54ebc7ab71bc 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c
> @@ -459,26 +459,31 @@ __i915_gem_userptr_get_pages_worker(struct work_struct *_work)
>   	if (pvec != NULL) {
>   		struct mm_struct *mm = obj->userptr.mm->mm;
>   		unsigned int flags = 0;
> +		int locked = 0;
>   
>   		if (!i915_gem_object_is_readonly(obj))
>   			flags |= FOLL_WRITE;
>   
>   		ret = -EFAULT;
>   		if (mmget_not_zero(mm)) {
> -			down_read(&mm->mmap_sem);
>   			while (pinned < npages) {
> +				if (!locked) {
> +					down_read(&mm->mmap_sem);
> +					locked = 1;
> +				}
>   				ret = get_user_pages_remote
>   					(work->task, mm,
>   					 obj->userptr.ptr + pinned * PAGE_SIZE,
>   					 npages - pinned,
>   					 flags,
> -					 pvec + pinned, NULL, NULL);
> +					 pvec + pinned, NULL, &locked);
>   				if (ret < 0)
>   					break;
>   
>   				pinned += ret;
>   			}
> -			up_read(&mm->mmap_sem);
> +			if (locked)
> +				up_read(&mm->mmap_sem);
>   			mmput(mm);
>   		}
>   	}
>