Re: [PATCH v8 8/8] drm/gpusvm: Use hmm_range_fault_unlocked_timeout() for range faults

[Matthew Brost <matthew.brost@intel.com>]

On Fri, Jul 10, 2026 at 02:27:19PM -0700, Stanislav Kinsburskii wrote:

Please send series like this to intel-xe@lists.freedesktop.org list too
as this will trigger our CI which expercises the change paths changed in
this series.

> Several GPU SVM paths take mmap_read_lock() only to call hmm_range_fault(),
> then retry -EBUSY until HMM_RANGE_DEFAULT_TIMEOUT expires. Those paths use
> MMU interval notifiers whose mm matches the mm that was locked for the HMM
> fault.
> 
> Use hmm_range_fault_unlocked_timeout() for those faults and pass the
> remaining retry budget to HMM. The helper owns mmap_lock acquisition and
> refreshes range->notifier_seq internally for each retry, while GPU SVM
> keeps its existing driver-lock validation with mmu_interval_read_retry()
> after a successful fault.
> 
> Leave drm_gpusvm_check_pages() on hmm_range_fault() because that path is
> called with the mmap lock already held by its caller.
> 
> Signed-off-by: Stanislav Kinsburskii <skinsburskii@gmail.com>
> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
> ---
>  drivers/gpu/drm/drm_gpusvm.c |   52 ++++++------------------------------------
>  1 file changed, 7 insertions(+), 45 deletions(-)
> 
> diff --git a/drivers/gpu/drm/drm_gpusvm.c b/drivers/gpu/drm/drm_gpusvm.c
> index 958cb605aedd..6b7a6eaebcd9 100644
> --- a/drivers/gpu/drm/drm_gpusvm.c
> +++ b/drivers/gpu/drm/drm_gpusvm.c
> @@ -788,22 +788,8 @@ enum drm_gpusvm_scan_result drm_gpusvm_scan_mm(struct drm_gpusvm_range *range,
>  	hmm_range.hmm_pfns = pfns;
>  
>  retry:
> -	hmm_range.notifier_seq = mmu_interval_read_begin(notifier);
> -	mmap_read_lock(range->gpusvm->mm);
> -
> -	while (true) {
> -		err = hmm_range_fault(&hmm_range);
> -		if (err == -EBUSY) {
> -			if (time_after(jiffies, timeout))
> -				break;
> -
> -			hmm_range.notifier_seq =
> -				mmu_interval_read_begin(notifier);
> -			continue;
> -		}
> -		break;
> -	}
> -	mmap_read_unlock(range->gpusvm->mm);
> +	err = hmm_range_fault_unlocked_timeout(&hmm_range,
> +					       max(timeout - jiffies, 1L));
>  	if (err)
>  		goto err_free;
>  
> @@ -1439,21 +1425,8 @@ int drm_gpusvm_get_pages(struct drm_gpusvm *gpusvm,
>  	}
>  
>  	hmm_range.hmm_pfns = pfns;
> -	while (true) {
> -		mmap_read_lock(mm);
> -		err = hmm_range_fault(&hmm_range);
> -		mmap_read_unlock(mm);
> -
> -		if (err == -EBUSY) {
> -			if (time_after(jiffies, timeout))
> -				break;
> -
> -			hmm_range.notifier_seq =
> -				mmu_interval_read_begin(notifier);
> -			continue;
> -		}
> -		break;
> -	}
> +	err = hmm_range_fault_unlocked_timeout(&hmm_range,
> +				max_t(long, timeout - jiffies, 1));

Unaligned indentation.

So I'd write this like this to avoid weird wraps:

ctimeout = max_t(long, timeout - jiffies, 1));
err = hmm_range_fault_unlocked_timeout(&hmm_range, ctimeout);

>  	mmput(mm);
>  	if (err)
>  		goto err_free;
> @@ -1736,24 +1709,13 @@ int drm_gpusvm_range_evict(struct drm_gpusvm *gpusvm,
>  		return -ENOMEM;
>  
>  	hmm_range.hmm_pfns = pfns;
> -	while (!time_after(jiffies, timeout)) {
> -		hmm_range.notifier_seq = mmu_interval_read_begin(notifier);
> -		if (time_after(jiffies, timeout)) {
> -			err = -ETIME;
> -			break;
> -		}
> -
> -		mmap_read_lock(mm);
> -		err = hmm_range_fault(&hmm_range);
> -		mmap_read_unlock(mm);
> -		if (err != -EBUSY)
> -			break;
> -	}
> +	err = hmm_range_fault_unlocked_timeout(&hmm_range,
> +				max_t(long, timeout - jiffies, 1));
>  

Same here.

Nits, aside LGTM.

Matt

>  	kvfree(pfns);
>  	mmput(mm);
>  
> -	return err;
> +	return err == -EBUSY ? -ETIME : err;
>  }
>  EXPORT_SYMBOL_GPL(drm_gpusvm_range_evict);
>  
> 
>