[PATCH v4 3/9] drm/xe: Add mem_region to properties line in VM snapshot capture

[Cavitt, Jonathan]

-----Original Message-----
From: Intel-xe <intel-xe-bounces at lists.freedesktop.org> On Behalf Of Matthew Brost
Sent: Wednesday, May 7, 2025 1:16 PM
To: intel-xe at lists.freedesktop.org
Cc: Souza, Jose <jose.souza at intel.com>; Santa, Carlos <carlos.santa at intel.com>; Venkatesh Reddy, Sushma <sushma.venkatesh.reddy at intel.com>
Subject: [PATCH v4 3/9] drm/xe: Add mem_region to properties line in VM snapshot capture
> 
> Add memory region to properties line in VM snapshot capture indicating
> where the memory is located. The memory region corresponds to regions in
> the uAPI. This is useful information for debug and will help build a
> robust GPU hang replay tool.
> 
> The current format is:
> 
> [<vma address>]: <permissions>|<type>|mem_region=0x%x
> 
> Permissions has two options, either "read_only" or "read_write".
> 
> Type has three options, either "userptr", "null_sparse", or "bo".
> 
> Memory region is a bit mask of where the memory is located.
> 
> Cc: José Roberto de Souza <jose.souza at intel.com>
> Signed-off-by: Matthew Brost <matthew.brost at intel.com>

LGTM.

Reviewed-by: Jonathan Cavitt <jonathan.cavitt at intel.com>
-Jonathan Cavitt

> ---
>  drivers/gpu/drm/xe/xe_vm.c | 23 +++++++++++++++++++++--
>  1 file changed, 21 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
> index 95c51834c620..842e482b60b2 100644
> --- a/drivers/gpu/drm/xe/xe_vm.c
> +++ b/drivers/gpu/drm/xe/xe_vm.c
> @@ -3731,6 +3731,7 @@ struct xe_vm_snapshot {
>  #define XE_VM_SNAP_FLAG_READ_ONLY	BIT(1)
>  #define XE_VM_SNAP_FLAG_IS_NULL		BIT(2)
>  		unsigned long flags;
> +		int uapi_mem_region;
>  		struct xe_bo *bo;
>  		void *data;
>  		struct mm_struct *mm;
> @@ -3776,6 +3777,18 @@ struct xe_vm_snapshot *xe_vm_snapshot_capture(struct xe_vm *vm)
>  		if (bo) {
>  			snap->snap[i].bo = xe_bo_get(bo);
>  			snap->snap[i].bo_ofs = xe_vma_bo_offset(vma);
> +			switch (bo->ttm.resource->mem_type) {
> +			case XE_PL_SYSTEM:
> +			case XE_PL_TT:
> +				snap->snap[i].uapi_mem_region = 0;
> +				break;
> +			case XE_PL_VRAM0:
> +				snap->snap[i].uapi_mem_region = 1;
> +				break;
> +			case XE_PL_VRAM1:
> +				snap->snap[i].uapi_mem_region = 2;
> +				break;
> +			}
>  		} else if (xe_vma_is_userptr(vma)) {
>  			struct mm_struct *mm =
>  				to_userptr_vma(vma)->userptr.notifier.mm;
> @@ -3787,10 +3800,13 @@ struct xe_vm_snapshot *xe_vm_snapshot_capture(struct xe_vm *vm)
>  
>  			snap->snap[i].bo_ofs = xe_vma_userptr(vma);
>  			snap->snap[i].flags |= XE_VM_SNAP_FLAG_USERPTR;
> +			snap->snap[i].uapi_mem_region = 0;
>  		} else if (xe_vma_is_null(vma)) {
>  			snap->snap[i].flags |= XE_VM_SNAP_FLAG_IS_NULL;
> +			snap->snap[i].uapi_mem_region = -1;
>  		} else {
>  			snap->snap[i].data = ERR_PTR(-ENOENT);
> +			snap->snap[i].uapi_mem_region = -1;
>  		}
>  		i++;
>  	}
> @@ -3859,13 +3875,16 @@ void xe_vm_snapshot_print(struct xe_vm_snapshot *snap, struct drm_printer *p)
>  	for (i = 0; i < snap->num_snaps; i++) {
>  		drm_printf(p, "[%llx].length: 0x%lx\n", snap->snap[i].ofs, snap->snap[i].len);
>  
> -		drm_printf(p, "[%llx].properties: %s|%s\n", snap->snap[i].ofs,
> +		drm_printf(p, "[%llx].properties: %s|%s|mem_region=0x%lx\n",
> +			   snap->snap[i].ofs,
>  			   snap->snap[i].flags & XE_VM_SNAP_FLAG_READ_ONLY ?
>  			   "read_only" : "read_write",
>  			   snap->snap[i].flags & XE_VM_SNAP_FLAG_IS_NULL ?
>  			   "null_sparse" :
>  			   snap->snap[i].flags & XE_VM_SNAP_FLAG_USERPTR ?
> -			   "userptr" : "bo");
> +			   "userptr" : "bo",
> +			   snap->snap[i].uapi_mem_region == -1 ? 0 :
> +			   BIT(snap->snap[i].uapi_mem_region));
>  
>  		if (IS_ERR(snap->snap[i].data)) {
>  			drm_printf(p, "[%llx].error: %li\n", snap->snap[i].ofs,
> -- 
> 2.34.1
> 
>