Commit 919c2b58 authored by Ilia Levi's avatar Ilia Levi Committed by Lucas De Marchi
Browse files

drm/xe: Use managed BO in memirq 



When probe has an error, the current scheme causes a page fault during
resource unwinding. This happens because GTTMMADR_BAR gets unmapped
before the allocated BO is released via drmm. Switching to a managed BO
that is released via devm solves this ordering issue.

Signed-off-by: default avatarIlia Levi <ilia.levi@intel.com>
Reviewed-by: default avatarLucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241204111655.1293-1-ilia.levi@intel.com


Signed-off-by: default avatarLucas De Marchi <lucas.demarchi@intel.com>
parent 7cbd0f7c

drivers/gpu/drm/xe/xe_memirq.c

+7 −16
Original line number Diff line number Diff line
@@ -155,13 +155,6 @@ static const char *guc_name(struct xe_guc *guc) 
 *

 */



static void __release_xe_bo(struct drm_device *drm, void *arg)

{

	struct xe_bo *bo = arg;



	xe_bo_unpin_map_no_vm(bo);

}



static inline bool hw_reports_to_instance_zero(struct xe_memirq *memirq)

{

	/*
@@ -184,9 +177,7 @@ static int memirq_alloc_pages(struct xe_memirq *memirq) 
	BUILD_BUG_ON(!IS_ALIGNED(XE_MEMIRQ_SOURCE_OFFSET(0), SZ_64));

	BUILD_BUG_ON(!IS_ALIGNED(XE_MEMIRQ_STATUS_OFFSET(0), SZ_4K));



	/* XXX: convert to managed bo */

	bo = xe_bo_create_pin_map(xe, tile, NULL, bo_size,

				  ttm_bo_type_kernel,

	bo = xe_managed_bo_create_pin_map(xe, tile, bo_size,

					  XE_BO_FLAG_SYSTEM |

					  XE_BO_FLAG_GGTT |

					  XE_BO_FLAG_GGTT_INVALIDATE |
@@ -215,7 +206,7 @@ static int memirq_alloc_pages(struct xe_memirq *memirq) 
		     xe_bo_ggtt_addr(bo), bo_size, XE_MEMIRQ_SOURCE_OFFSET(0),

		     XE_MEMIRQ_STATUS_OFFSET(0));



	return drmm_add_action_or_reset(&xe->drm, __release_xe_bo, memirq->bo);

	return 0;



out:

	memirq_err(memirq, "Failed to allocate memirq page (%pe)\n", ERR_PTR(err));