Skip to contentSkip to conversation

Add memory page offlining support

PATCH Superseded Tejas Upadhyay 5 participants 38 messages Sep 2 – Sep 3

Subsystems (3)

v20 of 21 Superseded Latest is v21 →

lore
Share & tools
lore
#0 lore raw

v20 cover Add memory page offlining support

This functionality represents a significant step in making the xe driver gracefully handle hardware memory degradation. By integrating with the DRM Buddy allocator, the driver can permanently "carve out" faulty memory so it isn't reused by subsequent allocations.

IGT tests for testing this via injecting simple single address and duplicate address fault to unit test functionality:
https://patchwork.freedesktop.org/series/170555/#rev3

v20(Sashiko/MattB):
- Retry failed queued pages during xe_ttm_vram_mgr_del()
- xe_bo::q lifecycle fix
- hold down_write while traversing exec_queue list
- fix configfs default value + improve confusing read policy
v19:
- Sashiko review comment closure
- Sliting of making xe_bo_ttm_purge() export and NULL handling
- Remove redundant code in xe_ttm_vram_reserve_page_at_addr()
- Resolve nit comments provided in configfs and debugfs
v18:
- Splitting out the sysfs patch; it will come later once it is redesigned
- Address sashiko reivew comments
- Add debugfs and adjust igt tests and fault injection
v17:
- Rebase on drm-tip
- Enforce CONFIG_FAULT_INJECTION_DEBUG_FS for fault inject
v16:
- Correct sysfs patches with moving code with rcu lock
- In case purge fail let next alloc decide final failure
- Remove addr_to_block API, its being pulled from drm-tip
- Remove some unused code and replace where existing API can be used v15:
- Split few big patches into small
- Avoid vram_mgr lock in sysfs
- fix missing queue_pages counter increment
v14:
- Solve sashiko reviews
- Remove SOFT->HARD offline patch, decision is taken based on -EEXIST
- Dump gpu buddy allocated patch dropped
v13:
- Add fault inject and remove standlone debugfs
v12:
- Fix Sashiko review comments
v11:
- Add BAN reason for UMD to know about offlining
- Add support for soft offline mode
- Rebase and remove dummy lockdep annotation patch, as it merged from upstream v10:
- Remove RFC
v7:
- Improve debugfs warning messages
- Use scope_guard for locking(MattB)
- Adapt addition of queue member of LRC BO(MattB)
- Extend and use xe_ttm_bo_purge API for vram pages(MattB)
- Handle dma_buf_map requests for native and remote(MattB)
- Address if in never initialized block, set block to NULL
- Add lockdep in gpu buddy (MattB)
- Correct allocated_addr_to_block logic (MattA)
V6:
- Add more specific tests to noncritical bo sections
- Handle smooth exit of user created exec queues
- Break code and make purge specific static API
V5:
- Sysfs "max_pages" addition
- Reset block->private NULL post purge

Tejas Upadhyay (15):
drm/xe: Link VRAM object with gpu buddy
drm/xe: Link LRC BO and its execution queue with safe lifetime rules drm/xe: Export xe_ttm_bo_purge()
drm/xe: Handle NULL resource and allow purging of VRAM pages drm/xe/bo: Make xe_bo_is_user() public
drm/xe: Guard teardown paths against purged BOs drm/xe/vram: Extract buddy allocation and free helpers drm/xe/vram: Add page offline data structures and lifecycle drm/xe/vram: Add VRAM page offline fault handler drm/xe/configfs: Add disable_vram_page_offline attribute drm/xe/ras: Cache disable_vram_page_offline policy at init drm/xe/vram: Check disable_vram_page_offline policy in fault handler drm/xe: Expose bad VRAM pages via debugfs
drm/xe/uapi: Expose ban reason in EXEC_QUEUE_GET_PROPERTY_BAN drm/xe: Add fault-inject based VRAM page offline injection

 drivers/gpu/drm/xe/xe_bo.c                 |  25 +-
 drivers/gpu/drm/xe/xe_bo.h                 |   2 +
 drivers/gpu/drm/xe/xe_bo_types.h           |   8 +
 drivers/gpu/drm/xe/xe_configfs.c           |  72 ++-
 drivers/gpu/drm/xe/xe_configfs.h           |   2 +
 drivers/gpu/drm/xe/xe_debugfs.c            |  51 ++
 drivers/gpu/drm/xe/xe_debugfs.h            |   2 +
 drivers/gpu/drm/xe/xe_dma_buf.c            |   3 +
 drivers/gpu/drm/xe/xe_drm_ras_types.h      |   3 +
 drivers/gpu/drm/xe/xe_exec_queue.c         |  72 ++-
 drivers/gpu/drm/xe/xe_exec_queue_types.h   |   7 +-
 drivers/gpu/drm/xe/xe_execlist.c           |   4 +-
 drivers/gpu/drm/xe/xe_guc_submit.c         |  36 +-
 drivers/gpu/drm/xe/xe_ras.c                |   9 +
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c       | 590 ++++++++++++++++++++-
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.h       |   4 +
 drivers/gpu/drm/xe/xe_ttm_vram_mgr_types.h |  40 ++
 include/uapi/drm/xe_drm.h                  |  18 +-
 18 files changed, 902 insertions(+), 46 deletions(-)
Signature · 1 line
2.52.0
#1 lore raw

PATCH v20 1/15 drm/xe: Link VRAM object with gpu buddy

Setup to link TTM buffer object inside gpu buddy. This functionality is critical for supporting the memory page offline feature on CRI, where identified faulty pages must be traced back to their originating buffer for safe removal.

V2(MattB): Clear block->private in xe_ttm_vram_mgr_del as well

---
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 6 ++++++
 1 file changed, 6 insertions(+)
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +6 -0
index 05911904c1f9..51e983ee3bad 100644
@@ -57,6 +57,7 @@ static int xe_ttm_vram_mgr_new(struct ttm_resource_manager *man,
 	struct xe_ttm_vram_mgr *mgr = to_xe_ttm_vram_mgr(man);
 	struct xe_ttm_vram_mgr_resource *vres;
 	struct gpu_buddy *mm = &mgr->mm;
+	struct gpu_buddy_block *block;
 	u64 size, min_page_size;
 	unsigned long lpfn;
 	int err;
@@ -141,6 +142,8 @@ static int xe_ttm_vram_mgr_new(struct ttm_resource_manager *man,
 	}
 
 	mgr->visible_avail -= vres->used_visible_size;
+	list_for_each_entry(block, &vres->blocks, link)
+		block->private = tbo;
 	mutex_unlock(&mgr->lock);
 
 	if (!(vres->base.placement & TTM_PL_FLAG_CONTIGUOUS) &&
@@ -179,8 +182,11 @@ static void xe_ttm_vram_mgr_del(struct ttm_resource_manager *man,
 		to_xe_ttm_vram_mgr_resource(res);
 	struct xe_ttm_vram_mgr *mgr = to_xe_ttm_vram_mgr(man);
 	struct gpu_buddy *mm = &mgr->mm;
+	struct gpu_buddy_block *block;
 
 	mutex_lock(&mgr->lock);
+	list_for_each_entry(block, &vres->blocks, link)
+		block->private = NULL;
 	gpu_buddy_free_list(mm, &vres->blocks, 0);
 	mgr->visible_avail += vres->used_visible_size;
 	mutex_unlock(&mgr->lock);
Signature · 1 line
2.52.0
#2 lore raw

PATCH v20 2/15 drm/xe: Link LRC BO and its execution queue with safe lifetime rules

In reply to Tejas Upadhyay

Introduce an execution queue back-pointer (q) within struct xe_bo, primarily for Logical Ring Context (LRC) Buffer Objects. This back-pointer allows the driver to identify and execute targeted corrective actions on a specific queue if its associated LRC BO encounters errors like memory corruption or eviction.

Because this back-pointer takes no reference on its target execution queue, strict lifetime and serialization rules are implemented to prevent concurrent readers from encountering use-after-free or dangling pointer bugs:

- Encapsulate tracking logic inside xe_exec_queue_set_lrc_bo_backpointer() and xe_exec_queue_clear_lrc_bo_backpointer().

- Explicitly wrap all back-pointer writes and clears under the BO's dma_resv lock via xe_bo_lock(). Readers must hold this same lock across both the pointer read and its subsequent xe_exec_queue_get_unless_zero() call to guarantee serialization against teardown.

- Defer publishing the back-pointer until the very end of xe_exec_queue_create(). This ensures that early initialization failure paths (which bypass the kref mechanism and immediately free the queue structure) never leak a transient pointer to concurrent readers.

- Clear the back-pointer at the absolute top of __xe_exec_queue_fini(). This strips the pointer before q->ops->fini() destroys the hardware backend, ensuring that any reader holding the BO lock either observes a fully functional queue or NULL.

For multi-queue engines, secondary LRC BOs safely point to the primary queue, which is guaranteed to outlive the teardown pass due to active references held by its secondaries.

V4 (MattB):
- Add LRC BO's execution queue safe lifetime rules
V3 (Sashiko):
- Use 8-byte placeholder structure compatibility for non-LRC BO cases.
- Wrap assignments and clears securely under dma_resv locks.
V2 (Matt B):
- Add native support handling multi-queue configuration tracking.

---
 drivers/gpu/drm/xe/xe_bo_types.h   |  8 ++++
 drivers/gpu/drm/xe/xe_exec_queue.c | 64 ++++++++++++++++++++++++++++++
 2 files changed, 72 insertions(+)
drivers/gpu/drm/xe/xe_bo_types.h +8 -0
index e45f24301050..0eb93052c1d5 100644
@@ -20,6 +20,7 @@
 struct xe_device;
 struct xe_mem_pool_node;
 struct xe_vm;
+struct xe_exec_queue;
 
 #define XE_BO_MAX_PLACEMENTS	3
 
@@ -42,6 +43,13 @@ struct xe_bo {
 	u32 flags;
 	/** @vm: VM this BO is attached to, for extobj this will be NULL */
 	struct xe_vm *vm;
+	/**
+	 * @q: Queue this BO is attached to, mostly for LRC BO, NULL otherwise.
+	 * Protected by the BO dma_resv: readers must hold it across both the
+	 * read and xe_exec_queue_get_unless_zero(). The BO holds no reference
+	 * on the queue.
+	 */
+	struct xe_exec_queue *q;
 	/** @tile: Tile this BO is attached to (kernel BO only) */
 	struct xe_tile *tile;
 	/** @placements: valid placements for this BO */
drivers/gpu/drm/xe/xe_exec_queue.c +64 -0

80 lines not on this page · raw message

Signature · 1 line
2.52.0
#3 lore raw

PATCH v20 3/15 drm/xe: Export xe_ttm_bo_purge()

In reply to Tejas Upadhyay

Remove the static qualifier from xe_ttm_bo_purge() and add its prototype to xe_bo.h. This allows the function to be called from other parts of the driver outside of xe_bo.c, specifically needed for the memory page offline feature.

---
 drivers/gpu/drm/xe/xe_bo.c | 2 +-
 drivers/gpu/drm/xe/xe_bo.h | 1 +
 2 files changed, 2 insertions(+), 1 deletion(-)
drivers/gpu/drm/xe/xe_bo.c +1 -1
index dde309821237..8f93e09e51bb 100644
@@ -921,7 +921,7 @@ void xe_bo_set_purgeable_state(struct xe_bo *bo,
  *
  * Return: 0 on success, negative error code on failure
  */
-static int xe_ttm_bo_purge(struct ttm_buffer_object *ttm_bo, struct ttm_operation_ctx *ctx)
+int xe_ttm_bo_purge(struct ttm_buffer_object *ttm_bo, struct ttm_operation_ctx *ctx)
 {
 	struct xe_bo *bo = ttm_to_xe_bo(ttm_bo);
 	struct ttm_placement place = {};
drivers/gpu/drm/xe/xe_bo.h +1 -0
index e8081af5bfc1..90b15fff36c7 100644
@@ -600,6 +600,7 @@ struct xe_bo_shrink_flags {
 long xe_bo_shrink(struct ttm_operation_ctx *ctx, struct ttm_buffer_object *bo,
 		  const struct xe_bo_shrink_flags flags,
 		  unsigned long *scanned);
+int xe_ttm_bo_purge(struct ttm_buffer_object *ttm_bo, struct ttm_operation_ctx *ctx);
 
 /**
  * xe_bo_is_mem_type - Whether the bo currently resides in the given
Signature · 1 line
2.52.0
#4 lore raw

PATCH v20 4/15 drm/xe: Handle NULL resource and allow purging of VRAM pages

In reply to Tejas Upadhyay

Add NULL checks for the buffer object's resource pointer in xe_bo_is_vram() and xe_bo_is_stolen() to prevent potential NULL pointer dereferences when checking memory location during page offlining.

Additionally, remove the !ttm_bo->ttm check from xe_ttm_bo_purge(). This restriction is stripped away to explicitly allow the driver to purge VRAM pages during the memory page offline sequence.

v2(Sashiko):
- Split export and this patch also Check NULL res in xe_bo_is_vram() and xe_bo_is_stolen()

---
 drivers/gpu/drm/xe/xe_bo.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/gpu/drm/xe/xe_bo.c +6 -6
index 8f93e09e51bb..d30a8411d2a3 100644
@@ -104,13 +104,16 @@ static bool resource_is_vram(struct ttm_resource *res)
 
 bool xe_bo_is_vram(struct xe_bo *bo)
 {
-	return resource_is_vram(bo->ttm.resource) ||
-		resource_is_stolen_vram(xe_bo_device(bo), bo->ttm.resource);
+	struct ttm_resource *res = bo->ttm.resource;
+
+	return  res && (resource_is_vram(res) || resource_is_stolen_vram(xe_bo_device(bo), res));
 }
 
 bool xe_bo_is_stolen(struct xe_bo *bo)
 {
-	return bo->ttm.resource->mem_type == XE_PL_STOLEN;
+	struct ttm_resource *res = bo->ttm.resource;
+
+	return res && res->mem_type == XE_PL_STOLEN;
 }
 
 /**
@@ -929,9 +932,6 @@ int xe_ttm_bo_purge(struct ttm_buffer_object *ttm_bo, struct ttm_operation_ctx *
 
 	xe_bo_assert_held(bo);
 
-	if (!ttm_bo->ttm)
-		return 0;
-
 	if (!xe_bo_madv_is_dontneed(bo))
 		return 0;
 
Signature · 1 line
2.52.0
#5 lore raw

Re: [PATCH V20 04/15] drm/xe: Handle NULL resource and allow purging of VRAM pages

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Null pointer dereference in xe_vm_snapshot_capture when capturing a VM containing a purged BO. --

commit f809a0c4f30d1158f4546c8b29fd7d07fc7912ae
Author: Tejas Upadhyay <[email protected]>

drm/xe: Handle NULL resource and allow purging of VRAM pages

This patch adds NULL checks for the buffer object's resource pointer in xe_bo_is_vram() and xe_bo_is_stolen() to prevent potential NULL pointer dereferences. It also removes the restriction in xe_ttm_bo_purge() to explicitly allow the driver to purge VRAM pages.

Quoted 4 lines
drivers/gpu/drm/xe/xe_bo.c +0 -0
index 8f93e09e51bb8..d30a8411d2a33 100644

[ ... ]

drivers/gpu/drm/xe/xe_bo.c int xe_ttm_bo_purge(struct ttm_buffer_object *ttm_bo, struct ttm_operation_ctx * · line 934

Quoted 7 lines
drivers/gpu/drm/xe/xe_bo.c +0 -3
@@ -929,9 +932,6 @@ int xe_ttm_bo_purge(struct ttm_buffer_object *ttm_bo, struct ttm_operation_ctx *
 
 	xe_bo_assert_held(bo);
 
-	if (!ttm_bo->ttm)
-		return 0;
-

[Severity: High]
This is a pre-existing issue, but by removing this check to allow VRAM pages to be purged, does it expose a NULL pointer dereference in xe_vm_snapshot_capture()?

If a BO mapped in a VM with the XE_VMA_DUMPABLE flag is purged (e.g., via VRAM page offlining or memory pressure after being marked DONTNEED), its backing store is freed and bo->ttm.resource is set to NULL.

If a GPU hang subsequently triggers a devcoredump, xe_vm_snapshot_capture() iterates over the VMAs and unconditionally dereferences bo->ttm.resource without holding the BO's dma-resv lock or checking for NULL:

drivers/gpu/drm/xe/xe_vm.c:xe_vm_snapshot_capture() {
    ...
    if (bo) {
        snap->snap[i].cpu_caching = bo->cpu_caching;
        snap->snap[i].bo = xe_bo_get(bo);
        snap->snap[i].bo_ofs = xe_vma_bo_offset(vma);
        switch (bo->ttm.resource->mem_type) {
    ...
}

Could this cause a kernel panic if a local user marks a dumpable BO as DONTNEED, waits for it to be purged, and then intentionally hangs the GPU?

Quoted 3 lines
drivers/gpu/drm/xe/xe_bo.c +0 -0
 	if (!xe_bo_madv_is_dontneed(bo))
 		return 0;
 
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=4
#6 lore raw

PATCH v20 5/15 drm/xe/bo: Make xe_bo_is_user() public

In reply to Tejas Upadhyay

Export xe_bo_is_user() so it can be used by the VRAM page offline code to distinguish user-created BOs from kernel BOs when deciding whether a faulty page can be safely purged or requires a full reset.

---
 drivers/gpu/drm/xe/xe_bo.c | 8 +++++++-
 drivers/gpu/drm/xe/xe_bo.h | 1 +
 2 files changed, 8 insertions(+), 1 deletion(-)
drivers/gpu/drm/xe/xe_bo.c +7 -1
index d30a8411d2a3..c255327669da 100644
@@ -161,7 +161,13 @@ bool xe_bo_is_vm_bound(struct xe_bo *bo)
 	return !list_empty(&bo->ttm.base.gpuva.list);
 }
 
-static bool xe_bo_is_user(struct xe_bo *bo)
+/**
+ * xe_bo_is_user - Check if BO is user-created
+ * @bo: The BO
+ *
+ * Returns: true if @bo was created by userspace
+ */
+bool xe_bo_is_user(struct xe_bo *bo)
 {
 	return bo->flags & XE_BO_FLAG_USER;
 }
drivers/gpu/drm/xe/xe_bo.h +1 -0
index 90b15fff36c7..eede678ad303 100644
@@ -601,6 +601,7 @@ long xe_bo_shrink(struct ttm_operation_ctx *ctx, struct ttm_buffer_object *bo,
 		  const struct xe_bo_shrink_flags flags,
 		  unsigned long *scanned);
 int xe_ttm_bo_purge(struct ttm_buffer_object *ttm_bo, struct ttm_operation_ctx *ctx);
+bool xe_bo_is_user(struct xe_bo *bo);
 
 /**
  * xe_bo_is_mem_type - Whether the bo currently resides in the given
Signature · 1 line
2.52.0
#7 lore raw

PATCH v20 6/15 drm/xe: Guard teardown paths against purged BOs

In reply to Tejas Upadhyay

VRAM page offlining can purge BOs that are still referenced by page tables, exec queues, and DMA-buf exports. Add xe_bo_is_purged() guards in the teardown paths to prevent unpinning or mapping an already-purged BO:

- xe_bo_unpin_map_no_vm(): skip unpin if purged
- xe_dma_buf_map(): return -ENOENT early if purged
- xe_exec_queue_update_run_ticks(): skip LRC timestamp read if purged
- xe_pt_destroy(): skip unpin if purged

v4(Sashiko):
- Move bo purge check in xe_bo_unpin() to handle all instances v3(Sashiko):
- Remove dma_resv is already held
v2(Himal):
- take dma_resv lock before calling xe_bo_is_purged()

---
 drivers/gpu/drm/xe/xe_bo.c         | 3 +++
 drivers/gpu/drm/xe/xe_dma_buf.c    | 3 +++
 drivers/gpu/drm/xe/xe_exec_queue.c | 8 ++++++--
 3 files changed, 12 insertions(+), 2 deletions(-)
drivers/gpu/drm/xe/xe_bo.c +3 -0
index c255327669da..b162753cebb7 100644
@@ -3267,6 +3267,9 @@ void xe_bo_unpin(struct xe_bo *bo)
 	struct ttm_place *place = &bo->placements[0];
 	struct xe_device *xe = xe_bo_device(bo);
 
+	if (xe_bo_is_purged(bo))
+		return;
+
 	xe_assert(xe, !bo->ttm.base.import_attach);
 	xe_assert(xe, xe_bo_is_pinned(bo));
 
drivers/gpu/drm/xe/xe_dma_buf.c +3 -0
index bf0728838ead..5d9f1cd24b7f 100644
@@ -104,6 +104,9 @@ static struct sg_table *xe_dma_buf_map(struct dma_buf_attachment *attach,
 	struct sg_table *sgt;
 	int r = 0;
 
+	if (xe_bo_is_purged(bo))
+		return ERR_PTR(-ENOENT);
+
 	if (!attach->peer2peer && !xe_bo_can_migrate(bo, XE_PL_TT))
 		return ERR_PTR(-EOPNOTSUPP);
 
drivers/gpu/drm/xe/xe_exec_queue.c +6 -2
index f3ea4807ac33..e63559a2f582 100644
@@ -1630,8 +1630,12 @@ void xe_exec_queue_update_run_ticks(struct xe_exec_queue *q)
 	 * errors.
 	 */
 	lrc = q->lrc[0];
-	new_ts = xe_lrc_update_timestamp(lrc, &old_ts);
-	q->xef->run_ticks[q->class] += (new_ts - old_ts) * q->width;
+	xe_bo_lock(lrc->bo, false);
+	if (!xe_bo_is_purged(lrc->bo)) {
+		new_ts = xe_lrc_update_timestamp(lrc, &old_ts);
+		q->xef->run_ticks[q->class] += (new_ts - old_ts) * q->width;
+	}
+	xe_bo_unlock(lrc->bo);
 
 	drm_dev_exit(idx);
 }
Signature · 1 line
2.52.0
#8 lore raw

Re: [PATCH V20 06/15] drm/xe: Guard teardown paths against purged BOs

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] xe_bo_unpin_external() is missing the xe_bo_is_purged() guard, which will cause a crash when…

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] xe_bo_unpin_external() is missing the xe_bo_is_purged() guard, which will cause a crash when unpinning purged DMA-buf exports.
- [High] xe_dma_buf_begin_cpu_access() fails to check if the BO is purged.
--

commit 58df7bf49c93c1c7134f73fd8224de6fa0f907f3
Author: Tejas Upadhyay <[email protected]>
Subject: drm/xe: Guard teardown paths against purged BOs

This commit adds xe_bo_is_purged guards to various teardown and mapping paths in the xe driver. It prevents unpinning or mapping a buffer object that has already been purged by VRAM page offlining.

drivers/gpu/drm/xe/xe_bo.c void xe_bo_unpin(struct xe_bo *bo) · line 3272

Quoted 11 lines
drivers/gpu/drm/xe/xe_bo.c +3 -0
index c255327669da5..b162753cebb70 100644
@@ -3267,6 +3267,9 @@ void xe_bo_unpin(struct xe_bo *bo)
 	struct ttm_place *place = &bo->placements[0];
 	struct xe_device *xe = xe_bo_device(bo);
 
+	if (xe_bo_is_purged(bo))
+		return;
+

[Severity: High]
Does xe_bo_unpin_external also need a check for purged BOs?

When a DMA-buf exported BO is purged, closing the DMA-buf file descriptor from userspace will call xe_dma_buf_unpin, which calls xe_bo_unpin_external:

drivers/gpu/drm/xe/xe_dma_buf.c:xe_dma_buf_unpin() {
	...
	xe_bo_unpin_external(bo);
}

Because xe_bo_unpin_external lacks the xe_bo_is_purged check, it will attempt to unpin an already-unpinned BO. This leads to a pin_count underflow and fails the assertion:

drivers/gpu/drm/xe/xe_bo.c:xe_bo_unpin_external() {
	...
	xe_assert(xe, !bo->vm);
	xe_assert(xe, xe_bo_is_pinned(bo));
	...
}
Quoted 8 lines
drivers/gpu/drm/xe/xe_bo.c +0 -0
 	xe_assert(xe, !bo->ttm.base.import_attach);
 	xe_assert(xe, xe_bo_is_pinned(bo));
 
drivers/gpu/drm/xe/xe_dma_buf.c +0 -0
index bf0728838ead5..5d9f1cd24b7f3 100644

drivers/gpu/drm/xe/xe_dma_buf.c static struct sg_table *xe_dma_buf_map(struct dma_buf_attachment *attach, · line 109

Quoted 6 lines
drivers/gpu/drm/xe/xe_dma_buf.c +3 -0
@@ -104,6 +104,9 @@ static struct sg_table *xe_dma_buf_map(struct dma_buf_attachment *attach,
 	struct sg_table *sgt;
 	int r = 0;
 
+	if (xe_bo_is_purged(bo))
+		return ERR_PTR(-ENOENT);
+

[Severity: High]
Does xe_dma_buf_begin_cpu_access need a similar guard against purged BOs?

If a DMA-buf exported BO is purged, and userspace invokes the DMA_BUF_IOCTL_SYNC ioctl to begin CPU access, xe_dma_buf_begin_cpu_access will call xe_bo_migrate:

drivers/gpu/drm/xe/xe_dma_buf.c:xe_dma_buf_begin_cpu_access() {
	...
	xe_validation_guard(...) {
		ret = drm_exec_lock_obj(&exec, &bo->ttm.base);
		...
		ret = xe_bo_migrate(bo, XE_PL_TT, NULL, &exec);
	...
}

Without an xe_bo_is_purged check before migrating, xe_bo_migrate will try to migrate a destroyed buffer object. This can cause a NULL pointer dereference since bo->ttm.resource->mem_type can be NULL for purged BOs, or it will silently migrate the empty backing store to XE_PL_TT, providing userspace with garbage memory.

Quoted 3 lines
drivers/gpu/drm/xe/xe_dma_buf.c +0 -0
 	if (!attach->peer2peer && !xe_bo_can_migrate(bo, XE_PL_TT))
 		return ERR_PTR(-EOPNOTSUPP);
 

[ ... ]

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=6
#9 lore raw

RE: [PATCH V20 06/15] drm/xe: Guard teardown paths against purged BOs

Quoted 52 lines

-----Original Message-----
From: [email protected] <[email protected]>
Sent: 02 September 2026 21:28
To: Upadhyay, Tejas <[email protected]>
Cc: [email protected]
Subject: Re: [PATCH V20 06/15] drm/xe: Guard teardown paths against purged BOs

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] xe_bo_unpin_external() is missing the xe_bo_is_purged() guard, which will cause a crash when unpinning purged DMA-buf exports.
- [High] xe_dma_buf_begin_cpu_access() fails to check if the BO is purged. --

commit 58df7bf49c93c1c7134f73fd8224de6fa0f907f3
Author: Tejas Upadhyay <[email protected]>
Subject: drm/xe: Guard teardown paths against purged BOs

This commit adds xe_bo_is_purged guards to various teardown and mapping paths in the xe driver. It prevents unpinning or mapping a buffer object that has already been purged by VRAM page offlining.

drivers/gpu/drm/xe/xe_bo.c void xe_bo_unpin(struct xe_bo *bo) · line 3272

Quoted 11 lines
drivers/gpu/drm/xe/xe_bo.c +3 -0
index c255327669da5..b162753cebb70 100644
@@ -3267,6 +3267,9 @@ void xe_bo_unpin(struct xe_bo *bo)
 	struct ttm_place *place = &bo->placements[0];
 	struct xe_device *xe = xe_bo_device(bo);
 
+	if (xe_bo_is_purged(bo))
+		return;
+

[Severity: High]
Does xe_bo_unpin_external also need a check for purged BOs?

When a DMA-buf exported BO is purged, closing the DMA-buf file descriptor from userspace will call xe_dma_buf_unpin, which calls xe_bo_unpin_external:

drivers/gpu/drm/xe/xe_dma_buf.c:xe_dma_buf_unpin() {
	...
	xe_bo_unpin_external(bo);
}

Because xe_bo_unpin_external lacks the xe_bo_is_purged check, it will attempt to unpin an already-unpinned BO. This leads to a pin_count underflow and fails the assertion:

Quoted 9 lines
drivers/gpu/drm/xe/xe_bo.c:xe_bo_unpin_external() {
	...
	xe_assert(xe, !bo->vm);
	xe_assert(xe, xe_bo_is_pinned(bo));
	...
}
Quoted 2 lines
drivers/gpu/drm/xe/xe_bo.c +0 -0
 	xe_assert(xe, !bo->ttm.base.import_attach);
 	xe_assert(xe, xe_bo_is_pinned(bo));

Exported Bos are run though SBR not purged.

Quoted 33 lines
Quoted 3 lines
(quoted diff) +0 -0

b/drivers/gpu/drm/xe/xe_dma_buf.c index

bf0728838ead5..5d9f1cd24b7f3

drivers/gpu/drm/xe/xe_dma_buf.c static struct sg_table *xe_dma_buf_map(struct

Quoted 4 lines

100644

drivers/gpu/drm/xe/xe_dma_buf.c +0 -0
@@ -104,6 +104,9 @@ static struct sg_table *xe_dma_buf_map(struct

dma_buf_attachment *attach,

drivers/gpu/drm/xe/xe_dma_buf.c static struct sg_table *xe_dma_buf_map(struct dma_buf_attachment *attach, · line 109

Quoted 6 lines
drivers/gpu/drm/xe/xe_dma_buf.c +3 -0
 	struct sg_table *sgt;
 	int r = 0;
 
+	if (xe_bo_is_purged(bo))
+		return ERR_PTR(-ENOENT);
+

[Severity: High]
Does xe_dma_buf_begin_cpu_access need a similar guard against purged BOs?

If a DMA-buf exported BO is purged, and userspace invokes the DMA_BUF_IOCTL_SYNC ioctl to begin CPU access, xe_dma_buf_begin_cpu_access will call xe_bo_migrate:

drivers/gpu/drm/xe/xe_dma_buf.c:xe_dma_buf_begin_cpu_access() {
	...
	xe_validation_guard(...) {
		ret = drm_exec_lock_obj(&exec, &bo->ttm.base);
		...
		ret = xe_bo_migrate(bo, XE_PL_TT, NULL, &exec);
	...
}
Quoted 5 lines

Without an xe_bo_is_purged check before migrating, xe_bo_migrate will try to migrate a destroyed buffer object. This can cause a NULL pointer dereference since bo->ttm.resource->mem_type can be NULL for purged BOs, or it will silently migrate the empty backing store to XE_PL_TT, providing userspace with garbage memory.

Exported Bos are run though SBR not purged.

Tejas

Quoted 11 lines

drivers/gpu/drm/xe/xe_dma_buf.c static struct sg_table *xe_dma_buf_map(struct dma_buf_attachment *attach, · line 112

Quoted 4 lines
drivers/gpu/drm/xe/xe_dma_buf.c +0 -0
 	if (!attach->peer2peer && !xe_bo_can_migrate(bo, XE_PL_TT))
 		return ERR_PTR(-EOPNOTSUPP);
 

[ ... ]

--
Sashiko AI review ·
https://sashiko.dev/#/patchset/20260902145343.465686-17-
[email protected]?part=6

#10 lore raw

PATCH v20 7/15 drm/xe/vram: Extract buddy allocation and free helpers

In reply to Tejas Upadhyay

Factor out xe_ttm_vram_buddy_alloc() and xe_ttm_vram_buddy_free() from xe_ttm_vram_mgr_new() and xe_ttm_vram_mgr_del().

These new helpers consolidate block allocation/deallocation with visible-size tracking. This modularization makes the logic reusable by the upcoming VRAM page offlining reservation path.

This update includes a functional change: block->private now tracks the owning ttm_resource instead of the tbo. Tracking the resource avoids stale pointer bugs when TTM hands resources to a ghost object during accelerated moves or pipelined gutting, ensuring the cached pointer shares the exact lifetime of the blocks.

v2 (Matt B):
- Pass and store ttm_resource pointer rather than tbo to avoid stale BO backpointers.

---
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 90 ++++++++++++++++++----------
 1 file changed, 59 insertions(+), 31 deletions(-)
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +59 -31
index 51e983ee3bad..a5116dc05166 100644
@@ -49,6 +49,48 @@ static inline bool xe_is_vram_mgr_blocks_contiguous(struct gpu_buddy *mm,
 	return true;
 }
 
+static int xe_ttm_vram_buddy_alloc(struct xe_ttm_vram_mgr *mgr, u64 start,
+				   u64 end, u64 size, u64 min_page_size,
+				   struct list_head *blocks, unsigned long flags,
+				   struct ttm_resource *res, u64 *used_visible)
+{
+	struct gpu_buddy *mm = &mgr->mm;
+	struct gpu_buddy_block *block;
+	int err;
+
+	err = gpu_buddy_alloc_blocks(mm, start, end, size, min_page_size, blocks, flags);
+	if (err)
+		return err;
+
+	/*
+	 * Track the owning resource, never the owning BO. A BO backpointer
+	 * cached here goes stale the moment TTM hands the resource to a ghost
+	 * object (ttm_buffer_object_transfer()), which happens on every
+	 * accelerated move and on pipelined gutting. The resource, in
+	 * contrast, has exactly the same lifetime as these blocks and TTM
+	 * keeps &ttm_resource.bo pointing at the current owner for us.
+	 */
+	list_for_each_entry(block, blocks, link)
+		block->private = res;
+
+	if (end <= mgr->visible_size) {
+		*used_visible = size;
+	} else {
+		list_for_each_entry(block, blocks, link) {
+			u64 blk_start = gpu_buddy_block_offset(block);
+
+			if (blk_start < mgr->visible_size) {
+				u64 blk_end = blk_start + gpu_buddy_block_size(mm, block);
+
+				*used_visible += min(blk_end, mgr->visible_size) - blk_start;
+			}
+		}
+	}
+
+	mgr->visible_avail -= *used_visible;
+	return 0;
+}
+
 static int xe_ttm_vram_mgr_new(struct ttm_resource_manager *man,
 			       struct ttm_buffer_object *tbo,
 			       const struct ttm_place *place,
@@ -57,7 +99,6 @@ static int xe_ttm_vram_mgr_new(struct ttm_resource_manager *man,
 	struct xe_ttm_vram_mgr *mgr = to_xe_ttm_vram_mgr(man);
 	struct xe_ttm_vram_mgr_resource *vres;
 	struct gpu_buddy *mm = &mgr->mm;
-	struct gpu_buddy_block *block;
 	u64 size, min_page_size;
 	unsigned long lpfn;
 	int err;
@@ -118,32 +159,12 @@ static int xe_ttm_vram_mgr_new(struct ttm_resource_manager *man,
 		goto error_unlock;
 	}
 
-	err = gpu_buddy_alloc_blocks(mm, (u64)place->fpfn << PAGE_SHIFT,
-				     (u64)lpfn << PAGE_SHIFT, size,
-				     min_page_size, &vres->blocks, vres->flags);
+	err = xe_ttm_vram_buddy_alloc(mgr, (u64)place->fpfn << PAGE_SHIFT,
+				      (u64)lpfn << PAGE_SHIFT, size,
+				      min_page_size, &vres->blocks, vres->flags,
+				      &vres->base, &vres->used_visible_size);
 	if (err)
 		goto error_unlock;
-
-	if (lpfn <= mgr->visible_size >> PAGE_SHIFT) {
-		vres->used_visible_size = size;
-	} else {
-		struct gpu_buddy_block *block;
-
-		list_for_each_entry(block, &vres->blocks, link) {
-			u64 start = gpu_buddy_block_offset(block);
-
-			if (start < mgr->visible_size) {
-				u64 end = start + gpu_buddy_block_size(mm, block);
-
-				vres->used_visible_size +=
-					min(end, mgr->visible_size) - start;
-			}
-		}
-	}
-
-	mgr->visible_avail -= vres->used_visible_size;
-	list_for_each_entry(block, &vres->blocks, link)
-		block->private = tbo;
 	mutex_unlock(&mgr->lock);
 
 	if (!(vres->base.placement & TTM_PL_FLAG_CONTIGUOUS) &&
@@ -175,20 +196,27 @@ static int xe_ttm_vram_mgr_new(struct ttm_resource_manager *man,
 	return err;
 }
 
+static void xe_ttm_vram_buddy_free(struct xe_ttm_vram_mgr *mgr,
+				   struct list_head *blocks,
+				   u64 used_visible)
+{
+	struct gpu_buddy_block *block;
+
+	list_for_each_entry(block, blocks, link)
+		block->private = NULL;
+	gpu_buddy_free_list(&mgr->mm, blocks, 0);
+	mgr->visible_avail += used_visible;
+}
+
 static void xe_ttm_vram_mgr_del(struct ttm_resource_manager *man,
 				struct ttm_resource *res)
 {
 	struct xe_ttm_vram_mgr_resource *vres =
 		to_xe_ttm_vram_mgr_resource(res);
 	struct xe_ttm_vram_mgr *mgr = to_xe_ttm_vram_mgr(man);
-	struct gpu_buddy *mm = &mgr->mm;
-	struct gpu_buddy_block *block;
 
 	mutex_lock(&mgr->lock);
-	list_for_each_entry(block, &vres->blocks, link)
-		block->private = NULL;
-	gpu_buddy_free_list(mm, &vres->blocks, 0);
-	mgr->visible_avail += vres->used_visible_size;
+	xe_ttm_vram_buddy_free(mgr, &vres->blocks, vres->used_visible_size);
 	mutex_unlock(&mgr->lock);
 
 	ttm_resource_fini(man, res);
Signature · 1 line
2.52.0
#11 lore raw

PATCH v20 8/15 drm/xe/vram: Add page offline data structures and lifecycle

In reply to Tejas Upadhyay

Add xe_ttm_vram_offline_resource to track individual offlined VRAM pages, and extend xe_ttm_vram_mgr with offlined_pages/queued_pages lists and their counters.

Initialize the lists in __xe_ttm_vram_mgr_init() and add xe_ttm_vram_free_bad_pages() to release all tracked pages during xe_ttm_vram_mgr_fini() teardown.

v3(Sashiko):
- Reorder xe_ttm_vram_buddy_free and list_del_rcu
- Introduce reservation status
v2(Himal):
- Address possible leak in xe_ttm_vram_mgr_fini()
- Remove unused dev and add comment for used_visible_size 0

---
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c       | 25 ++++++++++++++
 drivers/gpu/drm/xe/xe_ttm_vram_mgr_types.h | 38 ++++++++++++++++++++++
 2 files changed, 63 insertions(+)
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +25 -0
index a5116dc05166..d97739a84a2d 100644
@@ -346,12 +346,35 @@ static void xe_ttm_vram_mgr_set_unused(struct drm_device *dev, void *arg)
 	ttm_resource_manager_set_used(man, false);
 }
 
+static void xe_ttm_vram_free_bad_pages(struct xe_ttm_vram_mgr *mgr)
+{
+	struct xe_ttm_vram_offline_resource *pos, *n;
+
+	list_for_each_entry_safe(pos, n, &mgr->offlined_pages, offlined_link) {
+		list_del_rcu(&pos->offlined_link);
+		xe_ttm_vram_buddy_free(mgr, &pos->blocks, pos->used_visible_size);
+		--mgr->n_offlined_pages;
+		kfree_rcu(pos, rcu);
+	}
+	list_for_each_entry_safe(pos, n, &mgr->queued_pages, queued_link) {
+		list_del_rcu(&pos->queued_link);
+		/* queued entries have no buddy reservation yet */
+		xe_ttm_vram_buddy_free(mgr, &pos->blocks, 0);
+		--mgr->n_queued_pages;
+		kfree_rcu(pos, rcu);
+	}
+}
+
 static void xe_ttm_vram_mgr_fini(struct drm_device *dev, void *arg)
 {
 	struct xe_device *xe = to_xe_device(dev);
 	struct xe_ttm_vram_mgr *mgr = arg;
 	struct ttm_resource_manager *man = &mgr->manager;
 
+	mutex_lock(&mgr->lock);
+	xe_ttm_vram_free_bad_pages(mgr);
+	mutex_unlock(&mgr->lock);
+
 	if (ttm_resource_manager_evict_all(&xe->ttm, man))
 		return;
 
@@ -378,6 +401,8 @@ int __xe_ttm_vram_mgr_init(struct xe_device *xe, struct xe_ttm_vram_mgr *mgr,
 	err = drmm_mutex_init(&xe->drm, &mgr->lock);
 	if (err)
 		return err;
+	INIT_LIST_HEAD(&mgr->offlined_pages);
+	INIT_LIST_HEAD(&mgr->queued_pages);
 	mgr->default_page_size = default_page_size;
 	mgr->visible_size = io_size;
 	mgr->visible_avail = io_size;
drivers/gpu/drm/xe/xe_ttm_vram_mgr_types.h +38 -0
index 9106da056b49..dc97b0ad0e51 100644
@@ -19,6 +19,14 @@ struct xe_ttm_vram_mgr {
 	struct ttm_resource_manager manager;
 	/** @mm: DRM buddy allocator which manages the VRAM */
 	struct gpu_buddy mm;
+	/** @offlined_pages: List of offlined pages */
+	struct list_head offlined_pages;
+	/** @n_offlined_pages: Number of offlined pages */
+	u16 n_offlined_pages;
+	/** @queued_pages: List of queued pages */
+	struct list_head queued_pages;
+	/** @n_queued_pages: Number of queued pages */
+	u16 n_queued_pages;
 	/** @visible_size: Proped size of the CPU visible portion */
 	u64 visible_size;
 	/** @visible_avail: CPU visible portion still unallocated */
@@ -45,4 +53,34 @@ struct xe_ttm_vram_mgr_resource {
 	unsigned long flags;
 };
 
+/**
+ * enum xe_page_reserve_status - Buddy reservation status
+ * @XE_PAGE_RESERVE_PENDING: reservation in progress
+ * @XE_PAGE_RESERVE_FAIL: reservation failed
+ */
+enum xe_page_reserve_status {
+	XE_PAGE_RESERVE_PENDING = 0,
+	XE_PAGE_RESERVE_FAIL,
+};
+
+/**
+ * struct xe_ttm_vram_offline_resource - Tracks a single offlined VRAM page
+ */
+struct xe_ttm_vram_offline_resource {
+	/** @offlined_link: Link into mgr->offlined_pages */
+	struct list_head offlined_link;
+	/** @queued_link: Link into mgr->queued_pages */
+	struct list_head queued_link;
+	/** @blocks: Buddy blocks reserved for this page */
+	struct list_head blocks;
+	/** @used_visible_size: CPU-visible bytes consumed */
+	u64 used_visible_size;
+	/** @addr: Faulty DPA reported by HW */
+	u64 addr;
+	/** @status: buddy reservation status */
+	enum xe_page_reserve_status status;
+	/** @rcu: RCU head for deferred freeing */
+	struct rcu_head rcu;
+};
+
 #endif
Signature · 1 line
2.52.0
#12 lore raw

PATCH v20 9/15 drm/xe/vram: Add VRAM page offline fault handler

In reply to Tejas Upadhyay

Add the core VRAM page offlining logic to handle HW-reported faulty physical addresses:

- xe_ttm_vram_purge_page(): Purges the BO containing the faulty address. Bans the associated VM (if page table BO) and exec queue (if LRC BO). Moves xe_exec_queue_kill() outside xe_bo_lock() to avoid AB-BA deadlock with vm->lock. Uses READ_ONCE(bo->q) to safely access the exec queue pointer.

- xe_ttm_vram_page_already_processed(): Checks if an address is already tracked in offlined_pages or queued_pages lists to avoid double-processing.

- xe_ttm_vram_reserve_page_at_addr(): Two-phase reservation that first queues the page, purges the BO outside the lock, then reserves the buddy block. Handles both allocated (BO present) and free page cases. Returns -EIO for critical kernel BOs to trigger system reset.

- xe_ttm_vram_addr_to_region(): Maps a DPA to its corresponding VRAM region. Checks if the address falls within usable space, or infrastructure zones (CCS, GSM, DSM) where it returns NULL to flag a reset path. If the target address is outside any known region returns ERR_PTR(-EOPNOTSUPP)

- xe_ttm_vram_handle_addr_fault(): Entry point called by RAS. Returns -EEXIST if already processed, -EIO for GSM/critical BO, -EOPNOTSUPP if out of bounds.

v14(Sashiko/MattB):
- Retry queued pages during xe_ttm_vram_mgr_del()
- Move all changes to find owner of bo to its API v13(Himal):
- Remove redundant code
v12(Sashiko):
- Handle multi tile and add assert for 4K align
- Remove unaligned action comment for ENXIO v11(Himal):
- match everywhere with enum vs bool for status member
- Fix comment and remove unused var
- if purge fail let next alloc confirm failure
- pass absolute address, useful for multi tile

---
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 330 +++++++++++++++++++++++++++
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.h |   1 +
 2 files changed, 331 insertions(+)
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +330 -0
index d97739a84a2d..0ab92bc3c699 100644
@@ -13,10 +13,16 @@
 #include <drm/ttm/ttm_placement.h>
 #include <drm/ttm/ttm_range_manager.h>
 
+#include "regs/xe_regs.h"
 #include "xe_bo.h"
+#include "xe_configfs.h"
 #include "xe_device.h"
+#include "xe_exec_queue.h"
+#include "xe_lrc.h"
+#include "xe_mmio.h"
 #include "xe_pm.h"
 #include "xe_res_cursor.h"
+#include "xe_ttm_stolen_mgr.h"
 #include "xe_ttm_vram_mgr.h"
 #include "xe_vram_types.h"
 
@@ -208,6 +214,36 @@ static void xe_ttm_vram_buddy_free(struct xe_ttm_vram_mgr *mgr,
 	mgr->visible_avail += used_visible;
 }
 
+/*
+ * Retry pending page-offline reservations.
+ *
+ * A reservation can fail because the blocks backing the bad page are still
+ * allocated: either the owning BO could not be purged, or the purge was
+ * pipelined and TTM handed the resource to a ghost object which frees it
+ * only once the move fences signal. Rather than giving up, entries stay on
+ * @queued_pages and are retried here every time VRAM blocks come back.
+ *
+ * Called with @mgr->lock held.
+ */
+static void xe_ttm_vram_retry_queued_pages(struct xe_ttm_vram_mgr *mgr)
+{
+	struct xe_ttm_vram_offline_resource *pos, *n;
+
+	lockdep_assert_held(&mgr->lock);
+
+	list_for_each_entry_safe(pos, n, &mgr->queued_pages, queued_link) {
+		if (xe_ttm_vram_buddy_alloc(mgr, pos->addr, pos->addr + SZ_4K,
+					    SZ_4K, SZ_4K, &pos->blocks,
+					    GPU_BUDDY_RANGE_ALLOCATION, NULL,
+					    &pos->used_visible_size))
+			continue;
+		--mgr->n_queued_pages;
+		list_del_rcu(&pos->queued_link);
+		++mgr->n_offlined_pages;
+		list_add_rcu(&pos->offlined_link, &mgr->offlined_pages);
+	}
+}
+
 static void xe_ttm_vram_mgr_del(struct ttm_resource_manager *man,
 				struct ttm_resource *res)
 {
@@ -217,6 +253,8 @@ static void xe_ttm_vram_mgr_del(struct ttm_resource_manager *man,
 
 	mutex_lock(&mgr->lock);
 	xe_ttm_vram_buddy_free(mgr, &vres->blocks, vres->used_visible_size);
+	if (unlikely(!list_empty(&mgr->queued_pages)))
+		xe_ttm_vram_retry_queued_pages(mgr);
 	mutex_unlock(&mgr->lock);
 
 	ttm_resource_fini(man, res);
@@ -580,3 +618,295 @@ u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man)
 
 	return avail;
 }
+
+static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo)
+{
+	struct ttm_operation_ctx ctx = {};
+	struct xe_exec_queue *q_to_put = NULL;
+	struct xe_exec_queue *q = NULL;
+	struct xe_vm *vm = NULL;
+	u32	flags;
+	int ret = 0;
+
+	xe_bo_lock(bo, false);
+	if (bo->vm)
+		vm = xe_vm_get(bo->vm);
+	flags = bo->flags;
+	xe_bo_unlock(bo);
+	/*  Ban VM if BO is PPGTT */
+	if (vm && (flags & XE_BO_FLAG_PAGETABLE)) {
+		down_write(&vm->lock);
+		xe_vm_kill(vm, true);
+		up_write(&vm->lock);
+	}
+	if (vm)
+		xe_vm_put(vm);
+
+	xe_bo_lock(bo, false);
+	q = READ_ONCE(bo->q);
+	/*  Ban exec queue if BO is lrc */
+	if (q && xe_exec_queue_get_unless_zero(q)) {
+		/* ban queue */
+		q_to_put = q;
+	}
+
+	if (bo->purgeable.state == XE_MADV_PURGEABLE_PURGED) {
+		/* Already purged by shrinker during unlocked window — nothing to do */
+		xe_bo_unlock(bo);
+		goto out;
+	}
+
+	xe_bo_set_purgeable_state(bo, XE_MADV_PURGEABLE_DONTNEED);
+	ttm_bo_unmap_virtual(&bo->ttm);   /* nuke CPU mmap + VRAM IO mappings */
+	if (xe_bo_is_pinned(bo))
+		xe_bo_unpin(bo);
+	ret = xe_ttm_bo_purge(&bo->ttm, &ctx);
+	xe_bo_unlock(bo);
+
+out:
+	if (q_to_put) {
+		xe_exec_queue_kill(q_to_put);
+		xe_exec_queue_put(q_to_put);
+	}
+
+	return ret;
+}
+
+static bool xe_ttm_vram_page_already_processed(struct xe_ttm_vram_mgr *mgr,
+					       u64 addr)
+{
+	struct xe_ttm_vram_offline_resource *pos;
+
+	lockdep_assert_held(&mgr->lock);
+
+	list_for_each_entry(pos, &mgr->offlined_pages, offlined_link) {
+		if (pos->addr == addr)
+			return true;
+	}
+
+	list_for_each_entry(pos, &mgr->queued_pages, queued_link) {
+		if (pos->addr == addr)
+			return true;
+	}
+
+	return false;
+}
+
+/*
+ * Resolve the BO currently owning @block and take a reference on it.
+ *
+ * Called with @mgr->lock held, which serializes against
+ * xe_ttm_vram_buddy_free() clearing block->private.
+ *
+ * Returns NULL when there is no xe_bo we can act on: either the block is
+ * free, or the resource is temporarily owned by a TTM ghost object because
+ * a move or a pipelined gutting is still in flight. In both cases the
+ * blocks will hit xe_ttm_vram_mgr_del() on their own and the pending
+ * reservation is retried from there.
+ */
+static struct xe_bo *xe_ttm_vram_block_owner_get(struct gpu_buddy_block *block)
+{
+	struct ttm_resource *res = block->private;
+	struct ttm_buffer_object *tbo;
+	struct xe_bo *bo;
+
+	if (!res)
+		return NULL;
+
+	/*
+	 * res->bo is updated under bdev->lru_lock by ttm_resource_set_bo().
+	 * Racing with a ghost transfer here is benign: we either see the old
+	 * owner (whose purge is a no-op and the retry path recovers) or the
+	 * ghost (rejected below).
+	 *
+	 * A ghost is a bare ttm_transfer_obj, not an xe_bo, so ttm_to_xe_bo()
+	 * on one would be out of bounds. xe_bo_is_xe_bo() rejects it since
+	 * only our own BOs carry xe_ttm_bo_destroy().
+	 */
+	tbo = READ_ONCE(res->bo);
+	if (!tbo || !xe_bo_is_xe_bo(tbo))
+		return NULL;
+
+	bo = ttm_to_xe_bo(tbo);
+
+	/* The BO may already be in teardown with a zero refcount */
+	return xe_bo_get_unless_zero(bo) ? bo : NULL;
+}
+
+static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,
+					    struct xe_ttm_vram_mgr *vram_mgr, struct gpu_buddy *mm)
+{
+	struct xe_ttm_vram_offline_resource *pos, *n;
+	struct xe_ttm_vram_offline_resource *nentry;
+	struct xe_bo *pbo_to_put = NULL;
+	struct xe_bo *pbo = NULL;
+	struct gpu_buddy_block *block;
+	u64 size = SZ_4K;
+	int ret = 0;
+
+	scoped_guard(mutex, &vram_mgr->lock) {
+		if (xe_ttm_vram_page_already_processed(vram_mgr, addr))
+			return -EEXIST;
+		block = gpu_buddy_allocated_addr_to_block(mm, addr);
+		if (WARN_ON(IS_ERR(block)))
+			return PTR_ERR(block);
+
+		nentry = kzalloc_obj(*nentry);
+		if (!nentry)
+			return -ENOMEM;
+		INIT_LIST_HEAD(&nentry->blocks);
+		nentry->status = XE_PAGE_RESERVE_PENDING;
+		nentry->addr = addr;
+
+		if (block) {
+			pbo = xe_ttm_vram_block_owner_get(block);
+
+			/*
+			 * Critical kernel BO? Best-effort check without resv lock;
+			 * worst case a concurrent pin causes reset path unnecessarily.
+			 */
+			if (pbo && ((pbo->ttm.type == ttm_bo_type_kernel &&
+				     !(pbo->flags & XE_BO_FLAG_PINNED_LATE_RESTORE)) ||
+				    (xe_bo_is_user(pbo) && xe_bo_is_pinned(pbo)))) {
+				kfree(nentry);
+				pbo_to_put = pbo;
+				drm_err(&xe->drm,
+					"%s: addr: 0x%llx is critical kernel bo, requesting SBR\n",
+					__func__, addr);
+				break;
+			}
+		}
+		/* Queue both free and occupied (to-be-purged) pages */
+		++vram_mgr->n_queued_pages;
+		list_add_rcu(&nentry->queued_link, &vram_mgr->queued_pages);
+	}
+
+	/* Deferred put outside lock to avoid recursive deadlock */
+	if (pbo_to_put) {
+		xe_bo_put(pbo_to_put);
+		/* Hint System controller driver for reset with -EIO  */
+		return -EIO;
+	}
+
+	if (pbo) {
+		/*
+		 * Purge BO containing address - reference held from above.
+		 * This does not necessarily free the blocks synchronously: if
+		 * the BO is not idle, ttm_bo_pipeline_gutting() hands the
+		 * resource to a ghost object and it is released only once the
+		 * move fences signal. The reservation below then fails and is
+		 * retried from xe_ttm_vram_mgr_del().
+		 */
+		ret = xe_ttm_vram_purge_page(xe, pbo);
+		xe_bo_put(pbo);
+		if (ret)
+			drm_warn(&xe->drm, "Purge failed at addr:0x%llx, ret:%d\n", addr, ret);
+	}
+
+	scoped_guard(mutex, &vram_mgr->lock) {
+		ret = xe_ttm_vram_buddy_alloc(vram_mgr, addr, addr + size,
+					      size, size, &nentry->blocks,
+					      GPU_BUDDY_RANGE_ALLOCATION,
+					      NULL, &nentry->used_visible_size);
+		if (ret) {
+			nentry->status = XE_PAGE_RESERVE_FAIL;
+			drm_dbg(&xe->drm,
+				"Page at addr:0x%llx still busy (%d), deferring reservation\n",
+				addr, ret);
+			return 0;
+		}
+
+		list_for_each_entry_safe(pos, n, &vram_mgr->queued_pages, queued_link) {
+			if (pos->addr == nentry->addr) {
+				--vram_mgr->n_queued_pages;
+				list_del_rcu(&pos->queued_link);
+				break;
+			}
+		}
+		++vram_mgr->n_offlined_pages;
+		list_add_rcu(&nentry->offlined_link, &vram_mgr->offlined_pages);
+		/* RAS will send command to FW for offlining page based on ret value */
+	}
+	/* Success */
+	return ret;
+}
+
+static struct xe_vram_region *xe_ttm_vram_addr_to_region(struct xe_device *xe, u64 addr)
+{
+	struct xe_tile *tile;
+	u8 id;
+
+	for_each_tile(tile, xe, id) {
+		struct xe_vram_region *vr = tile->mem.vram;
+
+		if (!vr)
+			continue;
+
+		if (addr >= vr->dpa_base && addr < (vr->dpa_base + vr->usable_size))
+			return vr;
+
+		/* CCS, GSM, or DSM — infrastructure zone, needs reset */
+		if (addr >= (vr->dpa_base + vr->usable_size) &&
+		    addr < (vr->dpa_base + vr->actual_physical_size))
+			return NULL;
+	}
+
+	/*
+	 * Return an explicit error pointer so the caller knows the addr
+	 * is invalid and should be ignored, NOT SBR.
+	 */
+	return ERR_PTR(-EOPNOTSUPP);
+}
+
+/**
+ * xe_ttm_vram_handle_addr_fault - Handle vram physical address error flaged
+ * @xe: pointer to parent device
+ * @addr: physical faulty address
+ *
+ * Handle the physcial faulty address error on specific tile.
+ *
+ * Returns 0 for success, negative error code otherwise as follow:
+ * * %-EIO - critical BO or address outside any VRAM region; next action is reset.
+ * * %-EOPNOTSUPP - log-only policy or unknown address; no further action.
+ * * %-ENOMEM - allocation failure; next action is reset.
+ * * %-ENXIO - address not found in buddy; no further action.
+ * * %-EEXIST - address already processed; no further action.
+ *
+ * A return of 0 means the page is tracked. It may still be listed as
+ * pending if the blocks backing it could not be freed immediately; the
+ * reservation is then completed from xe_ttm_vram_mgr_del().
+ */
+int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)
+{
+	struct xe_ttm_vram_mgr *vram_mgr;
+	struct xe_vram_region *vr;
+	struct gpu_buddy *mm;
+
+	/* Assert that the address is 4K aligned */
+	if (WARN_ON_ONCE(!IS_ALIGNED(addr, SZ_4K))) {
+		drm_err(&xe->drm, "Address %llx is not 4K aligned!\n", addr);
+		return -EINVAL;
+	}
+
+	vr = xe_ttm_vram_addr_to_region(xe, addr);
+	if (IS_ERR(vr)) {
+		/*
+		 * The addr is outside VRAM and GSM.
+		 * Log a debug message if needed, and safely exit/ignore.
+		 */
+		drm_dbg(&xe->drm, "Address %llx is out of bounds, ignoring fault.\n", addr);
+		return PTR_ERR(vr);
+	}
+	if (!vr) {
+		drm_err(&xe->drm, "%s:%d GSM addr:%llx error requesting SBR\n",
+			__func__, __LINE__, addr);
+		/* Hint System controller driver for reset with -EIO  */
+		return -EIO;
+	}
+	vram_mgr = &vr->ttm;
+	mm = &vram_mgr->mm;
+
+	/* Reserve page at address */
+	return xe_ttm_vram_reserve_page_at_addr(xe, addr - vr->dpa_base, vram_mgr, mm);
+}
+EXPORT_SYMBOL(xe_ttm_vram_handle_addr_fault);
drivers/gpu/drm/xe/xe_ttm_vram_mgr.h +1 -0
index 87b7fae5edba..d5392beff30c 100644
@@ -31,6 +31,7 @@ u64 xe_ttm_vram_get_cpu_visible_size(struct ttm_resource_manager *man);
 void xe_ttm_vram_get_used(struct ttm_resource_manager *man,
 			  u64 *used, u64 *used_visible);
 
+int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr);
 static inline struct xe_ttm_vram_mgr_resource *
 to_xe_ttm_vram_mgr_resource(struct ttm_resource *res)
 {
Signature · 1 line
2.52.0
#13 lore raw

Re: [PATCH V20 09/15] drm/xe/vram: Add VRAM page offline fault handler

Thank you for your contribution! Sashiko AI review found 5 potential issue(s) to consider: - [High] Lockless read of `res->bo` in `xe_ttm_vram_block_owner_get` leads to a Use-After-Free when racing…

Thank you for your contribution! Sashiko AI review found 5 potential issue(s) to consider:
- [High] Lockless read of res->bo in xe_ttm_vram_block_owner_get leads to a Use-After-Free when racing with a TTM ghost object transfer.
- [High] Race condition when offlining a free VRAM page allows the faulty page to be allocated and never purged.
- [High] Forcefully unpinning BOs in xe_ttm_vram_purge_page violates pin counting semantics and creates TOCTOU vulnerabilities.
- [High] xe_ttm_vram_addr_to_region ignores the kernel_vram region, causing hardware faults in critical kernel memory to be silently ignored.
- [Medium] A race condition between two-phase page reservation and synchronous BO eviction falsely marks successfully offlined pages as failed. --

commit 23a2bf9a5461265765230ffd0e7a9d24c27ca918
Author: Tejas Upadhyay <[email protected]>
Subject: drm/xe/vram: Add VRAM page offline fault handler

This commit introduces a hardware VRAM page offline fault handler for the xe driver. It adds functionality to purge the buffer object containing the faulty address, ban the associated VM or exec queue, and reserve the hardware page in the buddy allocator to prevent future use.

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index d97739a84a2d1..0ab92bc3c6998 100644

[ ... ]

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
@@ -580,3 +618,295 @@ u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man)

[ ... ]

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo)
+{

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man) · line 662

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+	xe_bo_set_purgeable_state(bo, XE_MADV_PURGEABLE_DONTNEED);
+	ttm_bo_unmap_virtual(&bo->ttm);   /* nuke CPU mmap + VRAM IO mappings */
+	if (xe_bo_is_pinned(bo))
+		xe_bo_unpin(bo);

[Severity: High]
Does this forcefully break the pin counting contract by unpinning a BO on behalf of an unknown owner?

In xe_ttm_vram_purge_page(), this unconditionally unpins the BO if it is pinned, stealing a pin reference belonging to another subsystem (such as the display controller). When the true owner later unpins the BO, could this trigger an underflow crash?

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+	ret = xe_ttm_bo_purge(&bo->ttm, &ctx);
+	xe_bo_unlock(bo);
+
+out:

[ ... ]

Quoted 14 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +14 -0
+static struct xe_bo *xe_ttm_vram_block_owner_get(struct gpu_buddy_block *block)
+{
+	struct ttm_resource *res = block->private;
+	struct ttm_buffer_object *tbo;
+	struct xe_bo *bo;
+
+	if (!res)
+		return NULL;
+
+	/*
+	 * res->bo is updated under bdev->lru_lock by ttm_resource_set_bo().
+	 * Racing with a ghost transfer here is benign: we either see the old
+	 * owner (whose purge is a no-op and the retry path recovers) or the
+	 * ghost (rejected below).

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man) · line 730

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+	 *
+	 * A ghost is a bare ttm_transfer_obj, not an xe_bo, so ttm_to_xe_bo()
+	 * on one would be out of bounds. xe_bo_is_xe_bo() rejects it since
+	 * only our own BOs carry xe_ttm_bo_destroy().
+	 */
+	tbo = READ_ONCE(res->bo);
+	if (!tbo || !xe_bo_is_xe_bo(tbo))
+		return NULL;
+
+	bo = ttm_to_xe_bo(tbo);

[Severity: High]
Is it safe to read res->bo locklessly and pass it to xe_bo_is_xe_bo()?

In xe_ttm_vram_block_owner_get(), tbo is read locklessly. A concurrent accelerated move or pipelined gutting can transfer the resource to a TTM ghost object under bdev->lru_lock. This drops the last reference to the original BO, freeing it via kfree().

If the thread is preempted and resumes after the original BO is freed, passing the stale tbo pointer to xe_bo_is_xe_bo(tbo) will dereference tbo->destroy on freed memory, causing a use-after-free regression.

Quoted 8 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +8 -0
+
+	/* The BO may already be in teardown with a zero refcount */
+	return xe_bo_get_unless_zero(bo) ? bo : NULL;
+}
+
+static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,
+					    struct xe_ttm_vram_mgr *vram_mgr, struct gpu_buddy *mm)
+{

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man) · line 770

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+		if (block) {
+			pbo = xe_ttm_vram_block_owner_get(block);
+
+			/*
+			 * Critical kernel BO? Best-effort check without resv lock;
+			 * worst case a concurrent pin causes reset path unnecessarily.
+			 */
+			if (pbo && ((pbo->ttm.type == ttm_bo_type_kernel &&
+				     !(pbo->flags & XE_BO_FLAG_PINNED_LATE_RESTORE)) ||
+				    (xe_bo_is_user(pbo) && xe_bo_is_pinned(pbo)))) {

[Severity: High]
Does this lockless pin check create a time-of-check to time-of-use race condition?

In xe_ttm_vram_reserve_page_at_addr(), xe_bo_is_pinned(pbo) is checked without holding the appropriate locks. A user BO could be pinned right after this check (e.g., for display scanout), and then forcefully unpinned and purged.

The display hardware could then continue scanning out the old physical address, potentially leaking data of whatever new BO is allocated there.

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+				kfree(nentry);
+				pbo_to_put = pbo;
+				drm_err(&xe->drm,
+					"%s: addr: 0x%llx is critical kernel bo, requesting SBR\n",

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man) · line 784

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+					__func__, addr);
+				break;
+			}
+		}
+		/* Queue both free and occupied (to-be-purged) pages */
+		++vram_mgr->n_queued_pages;
+		list_add_rcu(&nentry->queued_link, &vram_mgr->queued_pages);
+	}
+
+	/* Deferred put outside lock to avoid recursive deadlock */

[Severity: High]
Can a concurrent allocation claim a faulty free page when the lock is dropped?

In xe_ttm_vram_reserve_page_at_addr(), when a hardware fault occurs at a free address, gpu_buddy_allocated_addr_to_block() returns NULL, meaning pbo is NULL. The page is queued and vram_mgr->lock is dropped to prevent deadlocks.

During this lockless window, a concurrent TTM allocation can claim this exact free buddy block. When the fault handler later re-acquires the lock and attempts xe_ttm_vram_buddy_alloc(), it will fail, returning 0 without purging (since pbo was NULL). This leaves the newly allocated BO using a faulty hardware page, which will cause persistent hardware errors.

Quoted 23 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +23 -0
+	if (pbo_to_put) {
+		xe_bo_put(pbo_to_put);
+		/* Hint System controller driver for reset with -EIO  */
+		return -EIO;
+	}
+
+	if (pbo) {
+		/*
+		 * Purge BO containing address - reference held from above.
+		 * This does not necessarily free the blocks synchronously: if
+		 * the BO is not idle, ttm_bo_pipeline_gutting() hands the
+		 * resource to a ghost object and it is released only once the
+		 * move fences signal. The reservation below then fails and is
+		 * retried from xe_ttm_vram_mgr_del().
+		 */
+		ret = xe_ttm_vram_purge_page(xe, pbo);
+		xe_bo_put(pbo);
+		if (ret)
+			drm_warn(&xe->drm, "Purge failed at addr:0x%llx, ret:%d\n", addr, ret);
+	}
+
+	scoped_guard(mutex, &vram_mgr->lock) {
+		ret = xe_ttm_vram_buddy_alloc(vram_mgr, addr, addr + size,

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man) · line 817

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+					      size, size, &nentry->blocks,
+					      GPU_BUDDY_RANGE_ALLOCATION,
+					      NULL, &nentry->used_visible_size);
+		if (ret) {
+			nentry->status = XE_PAGE_RESERVE_FAIL;
+			drm_dbg(&xe->drm,
+				"Page at addr:0x%llx still busy (%d), deferring reservation\n",
+				addr, ret);
+			return 0;
+		}

[Severity: Medium]
Does this unconditionally attempt buddy allocation without checking if the retry path already succeeded?

In xe_ttm_vram_reserve_page_at_addr(), if xe_ttm_vram_purge_page() frees the BO's VRAM resource synchronously, it invokes xe_ttm_vram_mgr_del() which triggers xe_ttm_vram_retry_queued_pages(). The retry logic will successfully allocate the buddy block.

When the original thread re-acquires the lock here and blindly calls xe_ttm_vram_buddy_alloc(), it will fail with -ENOSPC because the block is already reserved. The page's status is then falsely marked as XE_PAGE_RESERVE_FAIL, leading to misleading internal state.

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+
+		list_for_each_entry_safe(pos, n, &vram_mgr->queued_pages, queued_link) {

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man) · line 843

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+static struct xe_vram_region *xe_ttm_vram_addr_to_region(struct xe_device *xe, u64 addr)
+{
+	struct xe_tile *tile;
+	u8 id;
+
+	for_each_tile(tile, xe, id) {
+		struct xe_vram_region *vr = tile->mem.vram;
+
+		if (!vr)
+			continue;

[Severity: High]
Does this skip checking the kernel-dedicated VRAM region?

In xe_ttm_vram_addr_to_region(), the address lookup loop only checks tile->mem.vram and neglects tile->mem.kernel_vram. If a hardware fault occurs within kernel_vram, the address mapping fails and returns ERR_PTR(-EOPNOTSUPP). This causes faults in kernel memory to be silently ignored instead of triggering a system reset.

Quoted 3 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+
+		if (addr >= vr->dpa_base && addr < (vr->dpa_base + vr->usable_size))
+			return vr;

[ ... ]

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=9
maintainer
#14 lore raw

Re: [PATCH V20 09/15] drm/xe/vram: Add VRAM page offline fault handler

Quoted 74 lines · On Wed, Sep 02, 2026 at 04:25:40PM +0000, [email protected] wrote:

Thank you for your contribution! Sashiko AI review found 5 potential issue(s) to consider:
- [High] Lockless read of res->bo in xe_ttm_vram_block_owner_get leads to a Use-After-Free when racing with a TTM ghost object transfer.
- [High] Race condition when offlining a free VRAM page allows the faulty page to be allocated and never purged.
- [High] Forcefully unpinning BOs in xe_ttm_vram_purge_page violates pin counting semantics and creates TOCTOU vulnerabilities.
- [High] xe_ttm_vram_addr_to_region ignores the kernel_vram region, causing hardware faults in critical kernel memory to be silently ignored.
- [Medium] A race condition between two-phase page reservation and synchronous BO eviction falsely marks successfully offlined pages as failed. --

commit 23a2bf9a5461265765230ffd0e7a9d24c27ca918
Author: Tejas Upadhyay <[email protected]>
Subject: drm/xe/vram: Add VRAM page offline fault handler

This commit introduces a hardware VRAM page offline fault handler for the xe driver. It adds functionality to purge the buffer object containing the faulty address, ban the associated VM or exec queue, and reserve the hardware page in the buddy allocator to prevent future use.

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index d97739a84a2d1..0ab92bc3c6998 100644

[ ... ]

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
@@ -580,3 +618,295 @@ u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man)

[ ... ]

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo)
+{

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+	xe_bo_set_purgeable_state(bo, XE_MADV_PURGEABLE_DONTNEED);
+	ttm_bo_unmap_virtual(&bo->ttm);   /* nuke CPU mmap + VRAM IO mappings */
+	if (xe_bo_is_pinned(bo))
+		xe_bo_unpin(bo);

[Severity: High]
Does this forcefully break the pin counting contract by unpinning a BO on behalf of an unknown owner?

In xe_ttm_vram_purge_page(), this unconditionally unpins the BO if it is pinned, stealing a pin reference belonging to another subsystem (such as the display controller). When the true owner later unpins the BO, could this trigger an underflow crash?

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+	ret = xe_ttm_bo_purge(&bo->ttm, &ctx);
+	xe_bo_unlock(bo);
+
+out:

[ ... ]

Quoted 14 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +14 -0
+static struct xe_bo *xe_ttm_vram_block_owner_get(struct gpu_buddy_block *block)
+{
+	struct ttm_resource *res = block->private;
+	struct ttm_buffer_object *tbo;
+	struct xe_bo *bo;
+
+	if (!res)
+		return NULL;
+
+	/*
+	 * res->bo is updated under bdev->lru_lock by ttm_resource_set_bo().
+	 * Racing with a ghost transfer here is benign: we either see the old
+	 * owner (whose purge is a no-op and the retry path recovers) or the
+	 * ghost (rejected below).

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+	 *
+	 * A ghost is a bare ttm_transfer_obj, not an xe_bo, so ttm_to_xe_bo()
+	 * on one would be out of bounds. xe_bo_is_xe_bo() rejects it since
+	 * only our own BOs carry xe_ttm_bo_destroy().
+	 */
+	tbo = READ_ONCE(res->bo);
+	if (!tbo || !xe_bo_is_xe_bo(tbo))
+		return NULL;
+
+	bo = ttm_to_xe_bo(tbo);

[Severity: High]
Is it safe to read res->bo locklessly and pass it to xe_bo_is_xe_bo()?

Quoted 9 lines

In xe_ttm_vram_block_owner_get(), tbo is read locklessly. A concurrent accelerated move or pipelined gutting can transfer the resource to a TTM ghost object under bdev->lru_lock. This drops the last reference to the original BO, freeing it via kfree().

If the thread is preempted and resumes after the original BO is freed, passing the stale tbo pointer to xe_bo_is_xe_bo(tbo) will dereference tbo->destroy on freed memory, causing a use-after-free regression.

This one is correct and this memory unsafe here.

Luckily, this fix is easy...

This one will need to be fixed, snippet below:

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +5 -2
index 6a954018012a..ff937ab3837a 100644
@@ -733,7 +733,8 @@ static bool xe_ttm_vram_page_already_processed(struct xe_ttm_vram_mgr *mgr,
  * blocks will hit xe_ttm_vram_mgr_del() on their own and the pending
  * reservation is retried from there.
  */
-static struct xe_bo *xe_ttm_vram_block_owner_get(struct gpu_buddy_block *block)
+static struct xe_bo *xe_ttm_vram_block_owner_get(struct xe_device *xe,
+                                                struct gpu_buddy_block *block)
 {
        struct ttm_resource *res = block->private;
        struct ttm_buffer_object *tbo;
@@ -742,6 +743,8 @@ static struct xe_bo *xe_ttm_vram_block_owner_get(struct gpu_buddy_block *block)
        if (!res)
                return NULL;
 
+       guard(spinlock)(&xe->ttm.lru_lock);
+
        /*
         * res->bo is updated under bdev->lru_lock by ttm_resource_set_bo().
         * Racing with a ghost transfer here is benign: we either see the old
@@ -788,7 +791,7 @@ static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,
                nentry->addr = addr;
 
                if (block) {
-                       pbo = xe_ttm_vram_block_owner_get(block);
+                       pbo = xe_ttm_vram_block_owner_get(xe, block);
 
                        /*
                         * Critical kernel BO? Best-effort check without resv lock;
Quoted 23 lines
Quoted 8 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +8 -0
+
+	/* The BO may already be in teardown with a zero refcount */
+	return xe_bo_get_unless_zero(bo) ? bo : NULL;
+}
+
+static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,
+					    struct xe_ttm_vram_mgr *vram_mgr, struct gpu_buddy *mm)
+{

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+		if (block) {
+			pbo = xe_ttm_vram_block_owner_get(block);
+
+			/*
+			 * Critical kernel BO? Best-effort check without resv lock;
+			 * worst case a concurrent pin causes reset path unnecessarily.
+			 */
+			if (pbo && ((pbo->ttm.type == ttm_bo_type_kernel &&
+				     !(pbo->flags & XE_BO_FLAG_PINNED_LATE_RESTORE)) ||
+				    (xe_bo_is_user(pbo) && xe_bo_is_pinned(pbo)))) {

[Severity: High]
Does this lockless pin check create a time-of-check to time-of-use race condition?

Quoted 8 lines

In xe_ttm_vram_reserve_page_at_addr(), xe_bo_is_pinned(pbo) is checked without holding the appropriate locks. A user BO could be pinned right after this check (e.g., for display scanout), and then forcefully unpinned and purged.

The display hardware could then continue scanning out the old physical address, potentially leaking data of whatever new BO is allocated there.

We should probably file a Jira for this, as I agree this possible with the right race + NIC trying to pin the BO, but unsure how to handle one properly.

Quoted 22 lines
Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+				kfree(nentry);
+				pbo_to_put = pbo;
+				drm_err(&xe->drm,
+					"%s: addr: 0x%llx is critical kernel bo, requesting SBR\n",

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+					__func__, addr);
+				break;
+			}
+		}
+		/* Queue both free and occupied (to-be-purged) pages */
+		++vram_mgr->n_queued_pages;
+		list_add_rcu(&nentry->queued_link, &vram_mgr->queued_pages);
+	}
+
+	/* Deferred put outside lock to avoid recursive deadlock */

[Severity: High]
Can a concurrent allocation claim a faulty free page when the lock is dropped?

In xe_ttm_vram_reserve_page_at_addr(), when a hardware fault occurs at a free address, gpu_buddy_allocated_addr_to_block() returns NULL, meaning pbo is NULL. The page is queued and vram_mgr->lock is dropped to prevent deadlocks.

Quoted 6 lines

During this lockless window, a concurrent TTM allocation can claim this exact free buddy block. When the fault handler later re-acquires the lock and attempts xe_ttm_vram_buddy_alloc(), it will fail, returning 0 without purging (since pbo was NULL). This leaves the newly allocated BO using a faulty hardware page, which will cause persistent hardware errors.

This also looks right. So I think:

@@ -804,10 +807,17 @@ static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,

                                        __func__, addr);
                                break;
                        }
+
+                       /* Queue both free and occupied (to-be-purged) pages */
+                       ++vram_mgr->n_queued_pages;
+                       list_add_rcu(&nentry->queued_link, &vram_mgr->queued_pages);
+               } else {
+                       /* Immediately offline */
+                       return xe_ttm_vram_buddy_alloc(vram_mgr, addr, addr + size,
+                                                      size, size, &nentry->blocks,
+                                                      GPU_BUDDY_RANGE_ALLOCATION,
+                                                      NULL, &nentry->used_visible_size);
                }
-               /* Queue both free and occupied (to-be-purged) pages */
-               ++vram_mgr->n_queued_pages;
-               list_add_rcu(&nentry->queued_link, &vram_mgr->queued_pages);
        }
Quoted 65 lines
Quoted 23 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +23 -0
+	if (pbo_to_put) {
+		xe_bo_put(pbo_to_put);
+		/* Hint System controller driver for reset with -EIO  */
+		return -EIO;
+	}
+
+	if (pbo) {
+		/*
+		 * Purge BO containing address - reference held from above.
+		 * This does not necessarily free the blocks synchronously: if
+		 * the BO is not idle, ttm_bo_pipeline_gutting() hands the
+		 * resource to a ghost object and it is released only once the
+		 * move fences signal. The reservation below then fails and is
+		 * retried from xe_ttm_vram_mgr_del().
+		 */
+		ret = xe_ttm_vram_purge_page(xe, pbo);
+		xe_bo_put(pbo);
+		if (ret)
+			drm_warn(&xe->drm, "Purge failed at addr:0x%llx, ret:%d\n", addr, ret);
+	}
+
+	scoped_guard(mutex, &vram_mgr->lock) {
+		ret = xe_ttm_vram_buddy_alloc(vram_mgr, addr, addr + size,

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+					      size, size, &nentry->blocks,
+					      GPU_BUDDY_RANGE_ALLOCATION,
+					      NULL, &nentry->used_visible_size);
+		if (ret) {
+			nentry->status = XE_PAGE_RESERVE_FAIL;
+			drm_dbg(&xe->drm,
+				"Page at addr:0x%llx still busy (%d), deferring reservation\n",
+				addr, ret);
+			return 0;
+		}

[Severity: Medium]
Does this unconditionally attempt buddy allocation without checking if the retry path already succeeded?

In xe_ttm_vram_reserve_page_at_addr(), if xe_ttm_vram_purge_page() frees the BO's VRAM resource synchronously, it invokes xe_ttm_vram_mgr_del() which triggers xe_ttm_vram_retry_queued_pages(). The retry logic will successfully allocate the buddy block.

When the original thread re-acquires the lock here and blindly calls xe_ttm_vram_buddy_alloc(), it will fail with -ENOSPC because the block is already reserved. The page's status is then falsely marked as XE_PAGE_RESERVE_FAIL, leading to misleading internal state.

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+
+		list_for_each_entry_safe(pos, n, &vram_mgr->queued_pages, queued_link) {

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+static struct xe_vram_region *xe_ttm_vram_addr_to_region(struct xe_device *xe, u64 addr)
+{
+	struct xe_tile *tile;
+	u8 id;
+
+	for_each_tile(tile, xe, id) {
+		struct xe_vram_region *vr = tile->mem.vram;
+
+		if (!vr)
+			continue;

[Severity: High]
Does this skip checking the kernel-dedicated VRAM region?

Quoted 6 lines

In xe_ttm_vram_addr_to_region(), the address lookup loop only checks tile->mem.vram and neglects tile->mem.kernel_vram. If a hardware fault occurs within kernel_vram, the address mapping fails and returns ERR_PTR(-EOPNOTSUPP). This causes faults in kernel memory to be silently ignored instead of triggering a system reset.

tile->mem.kernel_vram = tile->mem.vram in the current code but for completeness, this likely right too. I'd check the kernel_vram if tile->mem.kernel_vram != tile->mem.vram.

Matt

Quoted 7 lines
Quoted 3 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+
+		if (addr >= vr->dpa_base && addr < (vr->dpa_base + vr->usable_size))
+			return vr;

[ ... ]

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=9
maintainer
#15 lore raw

Re: [PATCH V20 09/15] drm/xe/vram: Add VRAM page offline fault handler

On Wed, Sep 02, 2026 at 04:04:09PM -0700, Matthew Brost wrote:

One mistake in a suggestion. Inline below.

Quoted 196 lines · On Wed, Sep 02, 2026 at 04:25:40PM +0000, [email protected] wrote:
Quoted 74 lines

Thank you for your contribution! Sashiko AI review found 5 potential issue(s) to consider:
- [High] Lockless read of res->bo in xe_ttm_vram_block_owner_get leads to a Use-After-Free when racing with a TTM ghost object transfer.
- [High] Race condition when offlining a free VRAM page allows the faulty page to be allocated and never purged.
- [High] Forcefully unpinning BOs in xe_ttm_vram_purge_page violates pin counting semantics and creates TOCTOU vulnerabilities.
- [High] xe_ttm_vram_addr_to_region ignores the kernel_vram region, causing hardware faults in critical kernel memory to be silently ignored.
- [Medium] A race condition between two-phase page reservation and synchronous BO eviction falsely marks successfully offlined pages as failed. --

commit 23a2bf9a5461265765230ffd0e7a9d24c27ca918
Author: Tejas Upadhyay <[email protected]>
Subject: drm/xe/vram: Add VRAM page offline fault handler

This commit introduces a hardware VRAM page offline fault handler for the xe driver. It adds functionality to purge the buffer object containing the faulty address, ban the associated VM or exec queue, and reserve the hardware page in the buddy allocator to prevent future use.

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index d97739a84a2d1..0ab92bc3c6998 100644

[ ... ]

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
@@ -580,3 +618,295 @@ u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man)

[ ... ]

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo)
+{

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+	xe_bo_set_purgeable_state(bo, XE_MADV_PURGEABLE_DONTNEED);
+	ttm_bo_unmap_virtual(&bo->ttm);   /* nuke CPU mmap + VRAM IO mappings */
+	if (xe_bo_is_pinned(bo))
+		xe_bo_unpin(bo);

[Severity: High]
Does this forcefully break the pin counting contract by unpinning a BO on behalf of an unknown owner?

In xe_ttm_vram_purge_page(), this unconditionally unpins the BO if it is pinned, stealing a pin reference belonging to another subsystem (such as the display controller). When the true owner later unpins the BO, could this trigger an underflow crash?

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+	ret = xe_ttm_bo_purge(&bo->ttm, &ctx);
+	xe_bo_unlock(bo);
+
+out:

[ ... ]

Quoted 14 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +14 -0
+static struct xe_bo *xe_ttm_vram_block_owner_get(struct gpu_buddy_block *block)
+{
+	struct ttm_resource *res = block->private;
+	struct ttm_buffer_object *tbo;
+	struct xe_bo *bo;
+
+	if (!res)
+		return NULL;
+
+	/*
+	 * res->bo is updated under bdev->lru_lock by ttm_resource_set_bo().
+	 * Racing with a ghost transfer here is benign: we either see the old
+	 * owner (whose purge is a no-op and the retry path recovers) or the
+	 * ghost (rejected below).

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+	 *
+	 * A ghost is a bare ttm_transfer_obj, not an xe_bo, so ttm_to_xe_bo()
+	 * on one would be out of bounds. xe_bo_is_xe_bo() rejects it since
+	 * only our own BOs carry xe_ttm_bo_destroy().
+	 */
+	tbo = READ_ONCE(res->bo);
+	if (!tbo || !xe_bo_is_xe_bo(tbo))
+		return NULL;
+
+	bo = ttm_to_xe_bo(tbo);

[Severity: High]
Is it safe to read res->bo locklessly and pass it to xe_bo_is_xe_bo()?

Quoted 9 lines

In xe_ttm_vram_block_owner_get(), tbo is read locklessly. A concurrent accelerated move or pipelined gutting can transfer the resource to a TTM ghost object under bdev->lru_lock. This drops the last reference to the original BO, freeing it via kfree().

If the thread is preempted and resumes after the original BO is freed, passing the stale tbo pointer to xe_bo_is_xe_bo(tbo) will dereference tbo->destroy on freed memory, causing a use-after-free regression.

This one is correct and this memory unsafe here.

Luckily, this fix is easy...

This one will need to be fixed, snippet below:

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +5 -2
index 6a954018012a..ff937ab3837a 100644
@@ -733,7 +733,8 @@ static bool xe_ttm_vram_page_already_processed(struct xe_ttm_vram_mgr *mgr,
  * blocks will hit xe_ttm_vram_mgr_del() on their own and the pending
  * reservation is retried from there.
  */
-static struct xe_bo *xe_ttm_vram_block_owner_get(struct gpu_buddy_block *block)
+static struct xe_bo *xe_ttm_vram_block_owner_get(struct xe_device *xe,
+                                                struct gpu_buddy_block *block)
 {
        struct ttm_resource *res = block->private;
        struct ttm_buffer_object *tbo;
@@ -742,6 +743,8 @@ static struct xe_bo *xe_ttm_vram_block_owner_get(struct gpu_buddy_block *block)
        if (!res)
                return NULL;
 
+       guard(spinlock)(&xe->ttm.lru_lock);
+
        /*
         * res->bo is updated under bdev->lru_lock by ttm_resource_set_bo().
         * Racing with a ghost transfer here is benign: we either see the old
@@ -788,7 +791,7 @@ static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,
                nentry->addr = addr;
 
                if (block) {
-                       pbo = xe_ttm_vram_block_owner_get(block);
+                       pbo = xe_ttm_vram_block_owner_get(xe, block);
 
                        /*
                         * Critical kernel BO? Best-effort check without resv lock;
Quoted 25 lines
Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +8 -0
+
+	/* The BO may already be in teardown with a zero refcount */
+	return xe_bo_get_unless_zero(bo) ? bo : NULL;
+}
+
+static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,
+					    struct xe_ttm_vram_mgr *vram_mgr, struct gpu_buddy *mm)
+{

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+		if (block) {
+			pbo = xe_ttm_vram_block_owner_get(block);
+
+			/*
+			 * Critical kernel BO? Best-effort check without resv lock;
+			 * worst case a concurrent pin causes reset path unnecessarily.
+			 */
+			if (pbo && ((pbo->ttm.type == ttm_bo_type_kernel &&
+				     !(pbo->flags & XE_BO_FLAG_PINNED_LATE_RESTORE)) ||
+				    (xe_bo_is_user(pbo) && xe_bo_is_pinned(pbo)))) {

[Severity: High]
Does this lockless pin check create a time-of-check to time-of-use race condition?

Quoted 8 lines

In xe_ttm_vram_reserve_page_at_addr(), xe_bo_is_pinned(pbo) is checked without holding the appropriate locks. A user BO could be pinned right after this check (e.g., for display scanout), and then forcefully unpinned and purged.

The display hardware could then continue scanning out the old physical address, potentially leaking data of whatever new BO is allocated there.

We should probably file a Jira for this, as I agree this possible with the right race + NIC trying to pin the BO, but unsure how to handle one properly.

Quoted 22 lines
Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+				kfree(nentry);
+				pbo_to_put = pbo;
+				drm_err(&xe->drm,
+					"%s: addr: 0x%llx is critical kernel bo, requesting SBR\n",

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+					__func__, addr);
+				break;
+			}
+		}
+		/* Queue both free and occupied (to-be-purged) pages */
+		++vram_mgr->n_queued_pages;
+		list_add_rcu(&nentry->queued_link, &vram_mgr->queued_pages);
+	}
+
+	/* Deferred put outside lock to avoid recursive deadlock */

[Severity: High]
Can a concurrent allocation claim a faulty free page when the lock is dropped?

In xe_ttm_vram_reserve_page_at_addr(), when a hardware fault occurs at a free address, gpu_buddy_allocated_addr_to_block() returns NULL, meaning pbo is NULL. The page is queued and vram_mgr->lock is dropped to prevent deadlocks.

Quoted 6 lines

During this lockless window, a concurrent TTM allocation can claim this exact free buddy block. When the fault handler later re-acquires the lock and attempts xe_ttm_vram_buddy_alloc(), it will fail, returning 0 without purging (since pbo was NULL). This leaves the newly allocated BO using a faulty hardware page, which will cause persistent hardware errors.

This also looks right. So I think:

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
@@ -804,10 +807,17 @@ static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,
                                        __func__, addr);
                                break;
                        }

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr, · line 819

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+
+                       /* Queue both free and occupied (to-be-purged) pages */
+                       ++vram_mgr->n_queued_pages;
+                       list_add_rcu(&nentry->queued_link, &vram_mgr->queued_pages);
+               } else {
+                       /* Immediately offline */
+                       return xe_ttm_vram_buddy_alloc(vram_mgr, addr, addr + size,
+                                                      size, size, &nentry->blocks,
+                                                      GPU_BUDDY_RANGE_ALLOCATION,
+                                                      NULL, &nentry->used_visible_size);

I missed adjusting n_offlined_pages and adding nentry to offlined_pages here.

So I think check the return of xe_ttm_vram_buddy_alloc, then adjust n_offlined_pages + add nentry to offlined_pages.

Matt

Quoted 92 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -3
                }
-               /* Queue both free and occupied (to-be-purged) pages */
-               ++vram_mgr->n_queued_pages;
-               list_add_rcu(&nentry->queued_link, &vram_mgr->queued_pages);
        }
 
 
Quoted 65 lines
Quoted 23 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +23 -0
+	if (pbo_to_put) {
+		xe_bo_put(pbo_to_put);
+		/* Hint System controller driver for reset with -EIO  */
+		return -EIO;
+	}
+
+	if (pbo) {
+		/*
+		 * Purge BO containing address - reference held from above.
+		 * This does not necessarily free the blocks synchronously: if
+		 * the BO is not idle, ttm_bo_pipeline_gutting() hands the
+		 * resource to a ghost object and it is released only once the
+		 * move fences signal. The reservation below then fails and is
+		 * retried from xe_ttm_vram_mgr_del().
+		 */
+		ret = xe_ttm_vram_purge_page(xe, pbo);
+		xe_bo_put(pbo);
+		if (ret)
+			drm_warn(&xe->drm, "Purge failed at addr:0x%llx, ret:%d\n", addr, ret);
+	}
+
+	scoped_guard(mutex, &vram_mgr->lock) {
+		ret = xe_ttm_vram_buddy_alloc(vram_mgr, addr, addr + size,

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+					      size, size, &nentry->blocks,
+					      GPU_BUDDY_RANGE_ALLOCATION,
+					      NULL, &nentry->used_visible_size);
+		if (ret) {
+			nentry->status = XE_PAGE_RESERVE_FAIL;
+			drm_dbg(&xe->drm,
+				"Page at addr:0x%llx still busy (%d), deferring reservation\n",
+				addr, ret);
+			return 0;
+		}

[Severity: Medium]
Does this unconditionally attempt buddy allocation without checking if the retry path already succeeded?

In xe_ttm_vram_reserve_page_at_addr(), if xe_ttm_vram_purge_page() frees the BO's VRAM resource synchronously, it invokes xe_ttm_vram_mgr_del() which triggers xe_ttm_vram_retry_queued_pages(). The retry logic will successfully allocate the buddy block.

When the original thread re-acquires the lock here and blindly calls xe_ttm_vram_buddy_alloc(), it will fail with -ENOSPC because the block is already reserved. The page's status is then falsely marked as XE_PAGE_RESERVE_FAIL, leading to misleading internal state.

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+
+		list_for_each_entry_safe(pos, n, &vram_mgr->queued_pages, queued_link) {

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+static struct xe_vram_region *xe_ttm_vram_addr_to_region(struct xe_device *xe, u64 addr)
+{
+	struct xe_tile *tile;
+	u8 id;
+
+	for_each_tile(tile, xe, id) {
+		struct xe_vram_region *vr = tile->mem.vram;
+
+		if (!vr)
+			continue;

[Severity: High]
Does this skip checking the kernel-dedicated VRAM region?

Quoted 6 lines

In xe_ttm_vram_addr_to_region(), the address lookup loop only checks tile->mem.vram and neglects tile->mem.kernel_vram. If a hardware fault occurs within kernel_vram, the address mapping fails and returns ERR_PTR(-EOPNOTSUPP). This causes faults in kernel memory to be silently ignored instead of triggering a system reset.

tile->mem.kernel_vram = tile->mem.vram in the current code but for completeness, this likely right too. I'd check the kernel_vram if tile->mem.kernel_vram != tile->mem.vram.

Matt

Quoted 7 lines
Quoted 3 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+
+		if (addr >= vr->dpa_base && addr < (vr->dpa_base + vr->usable_size))
+			return vr;

[ ... ]

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=9
#16 lore raw

RE: [PATCH V20 09/15] drm/xe/vram: Add VRAM page offline fault handler

Quoted 125 lines

-----Original Message-----
From: Brost, Matthew <[email protected]>
Sent: 03 September 2026 09:52
To: [email protected]
Cc: Upadhyay, Tejas <[email protected]>; intel- [email protected]
Subject: Re: [PATCH V20 09/15] drm/xe/vram: Add VRAM page offline fault handler

On Wed, Sep 02, 2026 at 04:04:09PM -0700, Matthew Brost wrote:

One mistake in a suggestion. Inline below.

Quoted 1 line

On Wed, Sep 02, 2026 at 04:25:40PM +0000, [email protected]

Quoted 1 line · wrote:
Quoted 1 line

Thank you for your contribution! Sashiko AI review found 5 potential

issue(s) to consider:

Quoted 1 line
Quoted 1 line

- [High] Lockless read of res->bo in xe_ttm_vram_block_owner_get leads

to a Use-After-Free when racing with a TTM ghost object transfer.

Quoted 1 line
Quoted 1 line

- [High] Race condition when offlining a free VRAM page allows the faulty

page to be allocated and never purged.

Quoted 1 line
Quoted 1 line

- [High] Forcefully unpinning BOs in xe_ttm_vram_purge_page violates

pin counting semantics and creates TOCTOU vulnerabilities.

Quoted 1 line
Quoted 1 line

- [High] xe_ttm_vram_addr_to_region ignores the kernel_vram region,

causing hardware faults in critical kernel memory to be silently ignored.

Quoted 1 line
Quoted 1 line

- [Medium] A race condition between two-phase page reservation and

synchronous BO eviction falsely marks successfully offlined pages as failed.

Quoted 10 lines
Quoted 10 lines

--

commit 23a2bf9a5461265765230ffd0e7a9d24c27ca918
Author: Tejas Upadhyay <[email protected]>
Subject: drm/xe/vram: Add VRAM page offline fault handler

This commit introduces a hardware VRAM page offline fault handler for the xe driver. It adds functionality to purge the buffer object containing the faulty address, ban the associated VM or exec queue, and reserve the hardware page in the buddy allocator to prevent future

use.

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index d97739a84a2d1..0ab92bc3c6998 100644

[ ... ]

Quoted 8 lines
Quoted 8 lines
Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
@@ -580,3 +618,295 @@ u64 xe_ttm_vram_get_avail(struct
 ttm_resource_manager *man)

[ ... ]

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+static int xe_ttm_vram_purge_page(struct xe_device *xe, struct
+xe_bo *bo) {

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+	xe_bo_set_purgeable_state(bo, XE_MADV_PURGEABLE_DONTNEED);
+	ttm_bo_unmap_virtual(&bo->ttm);   /* nuke CPU mmap + VRAM IO

mappings */

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c static bool

Quoted 68 lines
Quoted 50 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+	if (xe_bo_is_pinned(bo))
+		xe_bo_unpin(bo);

[Severity: High]
Does this forcefully break the pin counting contract by unpinning a BO on behalf of an unknown owner?

In xe_ttm_vram_purge_page(), this unconditionally unpins the BO if it is pinned, stealing a pin reference belonging to another subsystem (such as the display controller). When the true owner later unpins the BO, could this trigger an underflow crash?

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+	ret = xe_ttm_bo_purge(&bo->ttm, &ctx);
+	xe_bo_unlock(bo);
+
+out:

[ ... ]

Quoted 14 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +14 -0
+static struct xe_bo *xe_ttm_vram_block_owner_get(struct
+gpu_buddy_block *block) {
+	struct ttm_resource *res = block->private;
+	struct ttm_buffer_object *tbo;
+	struct xe_bo *bo;
+
+	if (!res)
+		return NULL;
+
+	/*
+	 * res->bo is updated under bdev->lru_lock by ttm_resource_set_bo().
+	 * Racing with a ghost transfer here is benign: we either see the old
+	 * owner (whose purge is a no-op and the retry path recovers) or the
+	 * ghost (rejected below).

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+	 *
+	 * A ghost is a bare ttm_transfer_obj, not an xe_bo, so ttm_to_xe_bo()
+	 * on one would be out of bounds. xe_bo_is_xe_bo() rejects it since
+	 * only our own BOs carry xe_ttm_bo_destroy().
+	 */
+	tbo = READ_ONCE(res->bo);
+	if (!tbo || !xe_bo_is_xe_bo(tbo))
+		return NULL;
+
+	bo = ttm_to_xe_bo(tbo);

[Severity: High]
Is it safe to read res->bo locklessly and pass it to xe_bo_is_xe_bo()?

In xe_ttm_vram_block_owner_get(), tbo is read locklessly. A concurrent accelerated move or pipelined gutting can transfer the resource to a TTM ghost object under bdev->lru_lock. This drops the last reference to the original BO, freeing it via kfree().

Quoted 5 lines

If the thread is preempted and resumes after the original BO is freed, passing the stale tbo pointer to xe_bo_is_xe_bo(tbo) will dereference
tbo->destroy on freed memory, causing a use-after-free regression.

This one is correct and this memory unsafe here.

Luckily, this fix is easy...

This one will need to be fixed, snippet below:

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index 6a954018012a..ff937ab3837a 100644
@@ -733,7 +733,8 @@ static bool

xe_ttm_vram_page_already_processed(struct xe_ttm_vram_mgr *mgr,

Quoted 18 lines
Quoted 11 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -1
  * blocks will hit xe_ttm_vram_mgr_del() on their own and the pending
  * reservation is retried from there.
  */
-static struct xe_bo *xe_ttm_vram_block_owner_get(struct
 gpu_buddy_block *block)
+static struct xe_bo *xe_ttm_vram_block_owner_get(struct xe_device *xe,
+                                                struct
+gpu_buddy_block *block)
 {
        struct ttm_resource *res = block->private;
        struct ttm_buffer_object *tbo; @@ -742,6 +743,8 @@ static
Quoted 1 line

struct xe_bo *xe_ttm_vram_block_owner_get(struct gpu_buddy_block

*block)

Quoted 5 lines
        if (!res)
                return NULL;
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+       guard(spinlock)(&xe->ttm.lru_lock);
+

Sure

Quoted 46 lines
Quoted 4 lines

/*
* res->bo is updated under bdev->lru_lock by ttm_resource_set_bo().
* Racing with a ghost transfer here is benign: we either see
the old @@ -788,7 +791,7 @@ static int

xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,

Quoted 11 lines
                nentry->addr = addr;

                if (block) {
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -1
-                       pbo = xe_ttm_vram_block_owner_get(block);
+                       pbo = xe_ttm_vram_block_owner_get(xe, block);
 
                        /*
                         * Critical kernel BO? Best-effort check

without resv lock;

Quoted 5 lines
Quoted 5 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +5 -0
+
+	/* The BO may already be in teardown with a zero refcount */
+	return xe_bo_get_unless_zero(bo) ? bo : NULL; }
+
+static int xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64

addr,

Quoted 1 line
Quoted 1 line

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -0
+					    struct xe_ttm_vram_mgr

*vram_mgr, struct gpu_buddy *mm)

Quoted 7 lines
Quoted 7 lines
Quoted 1 line

+{

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +5 -0
+		if (block) {
+			pbo = xe_ttm_vram_block_owner_get(block);
+
+			/*
+			 * Critical kernel BO? Best-effort check without resv

lock;

Quoted 1 line
Quoted 1 line

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -0
+			 * worst case a concurrent pin causes reset path

unnecessarily.

Quoted 2 lines
Quoted 2 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+			 */
+			if (pbo && ((pbo->ttm.type == ttm_bo_type_kernel

&&

Quoted 1 line
Quoted 1 line

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -0
+				     !(pbo->flags &

XE_BO_FLAG_PINNED_LATE_RESTORE)) ||

Quoted 1 line
Quoted 1 line

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -0
+				    (xe_bo_is_user(pbo) &&

xe_bo_is_pinned(pbo)))) {

Quoted 3 lines
Quoted 3 lines

[Severity: High]
Does this lockless pin check create a time-of-check to time-of-use race

condition?

Quoted 1 line
Quoted 1 line
Quoted 13 lines
Quoted 7 lines
Quoted 7 lines

In xe_ttm_vram_reserve_page_at_addr(), xe_bo_is_pinned(pbo) is checked without holding the appropriate locks. A user BO could be pinned right after this check (e.g., for display scanout), and then forcefully unpinned and purged.

The display hardware could then continue scanning out the old physical address, potentially leaking data of whatever new BO is allocated

there.

Quoted 5 lines
Quoted 1 line

We should probably file a Jira for this, as I agree this possible with the right race + NIC trying to pin the BO, but unsure how to handle one properly.

Sure

Quoted 55 lines
Quoted 5 lines
Quoted 5 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+				kfree(nentry);
+				pbo_to_put = pbo;
+				drm_err(&xe->drm,
+					"%s: addr: 0x%llx is critical kernel bo,

requesting SBR\n",

Quoted 14 lines
Quoted 7 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 7 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +7 -0
+					__func__, addr);
+				break;
+			}
+		}
+		/* Queue both free and occupied (to-be-purged) pages */
+		++vram_mgr->n_queued_pages;
+		list_add_rcu(&nentry->queued_link, &vram_mgr-

queued_pages);

Quoted 6 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 3 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+	}
+
+	/* Deferred put outside lock to avoid recursive deadlock */

[Severity: High]
Can a concurrent allocation claim a faulty free page when the lock is

dropped?

Quoted 4 lines
Quoted 4 lines

In xe_ttm_vram_reserve_page_at_addr(), when a hardware fault occurs at a free address, gpu_buddy_allocated_addr_to_block() returns NULL, meaning pbo is NULL. The page is queued and vram_mgr->lock is dropped

to prevent deadlocks.

Quoted 6 lines
Quoted 6 lines

During this lockless window, a concurrent TTM allocation can claim this exact free buddy block. When the fault handler later re-acquires the lock and attempts xe_ttm_vram_buddy_alloc(), it will fail, returning 0 without purging (since pbo was NULL). This leaves the newly allocated BO using a faulty hardware page, which will cause

persistent hardware errors.

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c static int

Quoted 5 lines
Quoted 1 line

This also looks right. So I think:

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
@@ -804,10 +807,17 @@ static int

xe_ttm_vram_reserve_page_at_addr(struct xe_device *xe, u64 addr,

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
                                        __func__, addr);
                                break;
                        }
+
+                       /* Queue both free and occupied (to-be-purged) pages */

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c static int · line 821

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +9 -0
+                       ++vram_mgr->n_queued_pages;
+                       list_add_rcu(&nentry->queued_link, &vram_mgr-
 queued_pages);
+               } else {
+                       /* Immediately offline */
+                       return xe_ttm_vram_buddy_alloc(vram_mgr, addr, addr + size,
+                                                      size, size, &nentry->blocks,
+                                                      GPU_BUDDY_RANGE_ALLOCATION,
+                                                      NULL,
+ &nentry->used_visible_size);
Quoted 4 lines

I missed adjusting n_offlined_pages and adding nentry to offlined_pages here.

So I think check the return of xe_ttm_vram_buddy_alloc, then adjust n_offlined_pages + add nentry to offlined_pages.

Sure

Quoted 51 lines

Matt

Quoted 18 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -3
                }
-               /* Queue both free and occupied (to-be-purged) pages */
-               ++vram_mgr->n_queued_pages;
-               list_add_rcu(&nentry->queued_link, &vram_mgr->queued_pages);
        }
Quoted 13 lines
Quoted 13 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +11 -0
+	if (pbo_to_put) {
+		xe_bo_put(pbo_to_put);
+		/* Hint System controller driver for reset with -EIO  */
+		return -EIO;
+	}
+
+	if (pbo) {
+		/*
+		 * Purge BO containing address - reference held from above.
+		 * This does not necessarily free the blocks synchronously: if
+		 * the BO is not idle, ttm_bo_pipeline_gutting() hands the
Quoted 8 lines
Quoted 8 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 8 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +8 -0
+		 * resource to a ghost object and it is released only once the
+		 * move fences signal. The reservation below then fails and is
+		 * retried from xe_ttm_vram_mgr_del().
+		 */
+		ret = xe_ttm_vram_purge_page(xe, pbo);
+		xe_bo_put(pbo);
+		if (ret)
+			drm_warn(&xe->drm, "Purge failed at addr:0x%llx,

ret:%d\n", addr, ret);

Quoted 6 lines
Quoted 6 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 6 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +6 -0
+	}
+
+	scoped_guard(mutex, &vram_mgr->lock) {
+		ret = xe_ttm_vram_buddy_alloc(vram_mgr, addr, addr + size,
+					      size, size, &nentry->blocks,
+

GPU_BUDDY_RANGE_ALLOCATION,

Quoted 5 lines
Quoted 5 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +5 -0
+					      NULL, &nentry->used_visible_size);
+		if (ret) {
+			nentry->status = XE_PAGE_RESERVE_FAIL;
+			drm_dbg(&xe->drm,
+				"Page at addr:0x%llx still busy (%d), deferring

reservation\n",

Quoted 8 lines
Quoted 8 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 3 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+				addr, ret);
+			return 0;
+		}

[Severity: Medium]
Does this unconditionally attempt buddy allocation without checking if the retry path already succeeded?

Quoted 40 lines
Quoted 4 lines
Quoted 4 lines

In xe_ttm_vram_reserve_page_at_addr(), if xe_ttm_vram_purge_page() frees the BO's VRAM resource synchronously, it invokes xe_ttm_vram_mgr_del() which triggers xe_ttm_vram_retry_queued_pages(). The retry logic will successfully

allocate the buddy block.

Quoted 35 lines
Quoted 24 lines

When the original thread re-acquires the lock here and blindly calls xe_ttm_vram_buddy_alloc(), it will fail with -ENOSPC because the block is already reserved. The page's status is then falsely marked as XE_PAGE_RESERVE_FAIL, leading to misleading internal state.

Quoted 3 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+
+		list_for_each_entry_safe(pos, n, &vram_mgr->queued_pages,
+queued_link) {

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+static struct xe_vram_region *xe_ttm_vram_addr_to_region(struct
+xe_device *xe, u64 addr) {
+	struct xe_tile *tile;
+	u8 id;
+
+	for_each_tile(tile, xe, id) {
+		struct xe_vram_region *vr = tile->mem.vram;
+
+		if (!vr)
+			continue;

[Severity: High]
Does this skip checking the kernel-dedicated VRAM region?

Quoted 7 lines

In xe_ttm_vram_addr_to_region(), the address lookup loop only checks tile->mem.vram and neglects tile->mem.kernel_vram. If a hardware tile->fault occurs
within kernel_vram, the address mapping fails and returns ERR_PTR(-EOPNOTSUPP). This causes faults in kernel memory to be silently ignored instead of triggering a system reset.

tile->mem.kernel_vram = tile->mem.vram in the current code but for completeness, this likely right too. I'd check the kernel_vram if tile->mem.kernel_vram != tile->mem.vram.

We have no use of kernel_vram on drm-tip right now. First use may be in EXI, so as discussed let defer this change for later.

Tejas

Quoted 14 lines
Quoted 12 lines

Matt

Quoted 2 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+
+		if (addr >= vr->dpa_base && addr < (vr->dpa_base + vr-

usable_size))

Quoted 6 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -0
+			return vr;

[ ... ]

--
Sashiko AI review *
https://sashiko.dev/#/patchset/20260902145343.465686-17-

tejas.upadhy

Quoted 1 line
Quoted 1 line

[email protected]?part=9

#17 lore raw

PATCH v20 10/15 drm/xe/configfs: Add disable_vram_page_offline attribute

In reply to Tejas Upadhyay

Add a new configfs attribute 'disable_vram_page_offline' to control how bad VRAM pages are handled:
0, n, N, false - Do not disable (Offlining is active - default) 1, y, Y, true - Disable vram page offline (Logging only)

The attribute can only be set before binding to the device and defaults to false (offlining enabled). This gives administrators control over whether corrupted VRAM pages detected by hardware (e.g., ECC errors) are actively offlined or only logged.

v2(Sashiko):
- keep default value in case configfs not present

---
 drivers/gpu/drm/xe/xe_configfs.c | 72 +++++++++++++++++++++++++++++++-
 drivers/gpu/drm/xe/xe_configfs.h |  2 +
 2 files changed, 73 insertions(+), 1 deletion(-)
drivers/gpu/drm/xe/xe_configfs.c +71 -1
index 052cce962161..a9c72504c8a4 100644
@@ -61,7 +61,8 @@
  *	    ├── survivability_mode
  *	    ├── gt_types_allowed
  *	    ├── engines_allowed
- *	    └── enable_psmi
+ *	    ├── enable_psmi
+ *	    └── disable_vram_page_offline
  *
  * After configuring the attributes as per next section, the device can be
  * probed with::
@@ -157,6 +158,18 @@
  *
  *	# echo 1 > /sys/kernel/config/xe/0000:03:00.0/enable_psmi
  *
+ * This attribute can only be set on CRI before binding to the device.
+ *
+ * Disable VRAM page offline:
+ * ----------------------------
+ *
+ *  0, n, N, false - Do not disable (Offlining is active - default)
+ *  1, y, Y, true  - Disable vram page offline (Logging only)
+ *
+ *  Example to disable VRAM offline::
+ *
+ *      # echo 1 > /sys/kernel/config/xe/0000:03:00.0/disable_vram_page_offline
+ *
  * This attribute can only be set before binding to the device.
  *
  * Context restore BB
@@ -275,6 +288,7 @@ struct xe_config_group_device {
 		bool survivability_mode;
 		bool enable_psmi;
 		bool enable_multi_queue;
+		bool disable_vram_page_offline;
 		struct {
 			unsigned int max_vfs;
 			bool admin_only_pf;
@@ -295,6 +309,7 @@ static const struct xe_config_device device_defaults = {
 	.survivability_mode = false,
 	.enable_psmi = false,
 	.enable_multi_queue = true,
+	.disable_vram_page_offline = false,
 	.sriov = {
 		.max_vfs = XE_DEFAULT_MAX_VFS,
 		.admin_only_pf = XE_DEFAULT_ADMIN_ONLY_PF,
@@ -616,6 +631,33 @@ static ssize_t enable_multi_queue_store(struct config_item *item, const char *pa
 	return len;
 }
 
+static ssize_t disable_vram_page_offline_show(struct config_item *item, char *page)
+{
+	struct xe_config_device *dev = to_xe_config_device(item);
+
+	return sprintf(page, "%s\n", str_yes_no(dev->disable_vram_page_offline));
+}
+
+static ssize_t disable_vram_page_offline_store(struct config_item *item,
+					       const char *page, size_t len)
+{
+	struct xe_config_group_device *dev = to_xe_config_group_device(item);
+	bool val;
+	int ret;
+
+	ret = kstrtobool(page, &val);
+	if (ret)
+		return ret;
+
+	guard(mutex)(&dev->lock);
+	if (is_bound(dev))
+		return -EBUSY;
+
+	dev->config.disable_vram_page_offline = val;
+
+	return len;
+}
+
 static bool wa_bb_read_advance(bool dereference, char **p,
 			       const char *append, size_t len,
 			       size_t *max_size)
@@ -855,6 +897,7 @@ CONFIGFS_ATTR(, ctx_restore_mid_bb);
 CONFIGFS_ATTR(, ctx_restore_post_bb);
 CONFIGFS_ATTR(, enable_multi_queue);
 CONFIGFS_ATTR(, enable_psmi);
+CONFIGFS_ATTR(, disable_vram_page_offline);
 CONFIGFS_ATTR(, engines_allowed);
 CONFIGFS_ATTR(, gt_types_allowed);
 CONFIGFS_ATTR(, survivability_mode);
@@ -864,6 +907,7 @@ static struct configfs_attribute *xe_config_device_attrs[] = {
 	&attr_ctx_restore_post_bb,
 	&attr_enable_multi_queue,
 	&attr_enable_psmi,
+	&attr_disable_vram_page_offline,
 	&attr_engines_allowed,
 	&attr_gt_types_allowed,
 	&attr_survivability_mode,
@@ -895,6 +939,11 @@ static bool xe_config_device_is_visible(struct config_item *item,
 			return false;
 	}
 
+	if (attr == &attr_disable_vram_page_offline) {
+		if (!dev->desc->is_dgfx || dev->desc->platform != XE_CRESCENTISLAND)
+			return false;
+	}
+
 	return true;
 }
 
@@ -1142,6 +1191,7 @@ static void dump_custom_dev_config(struct pci_dev *pdev,
 	PRI_CUSTOM_ATTR("%llx", engines_allowed);
 	PRI_CUSTOM_ATTR("%d", enable_multi_queue);
 	PRI_CUSTOM_ATTR("%d", enable_psmi);
+	PRI_CUSTOM_ATTR("%d", disable_vram_page_offline);
 	PRI_CUSTOM_ATTR("%d", survivability_mode);
 	PRI_CUSTOM_ATTR("%u", sriov.admin_only_pf);
 
@@ -1290,6 +1340,26 @@ bool xe_configfs_get_enable_multi_queue(struct pci_dev *pdev)
 	return ret;
 }
 
+/**
+ * xe_configfs_get_disable_vram_page_offline - get configfs disable_vram_page_offline setting
+ * @pdev: pci device
+ *
+ * Return: disable_vram_page_offline setting in configfs
+ */
+bool xe_configfs_get_disable_vram_page_offline(struct pci_dev *pdev)
+{
+	struct xe_config_group_device *dev = find_xe_config_group_device(pdev);
+	bool ret;
+
+	if (!dev)
+		return device_defaults.disable_vram_page_offline;
+
+	ret = dev->config.disable_vram_page_offline;
+	config_group_put(&dev->group);
+
+	return ret;
+}
+
 /**
  * xe_configfs_get_ctx_restore_mid_bb - get configfs ctx_restore_mid_bb setting
  * @pdev: pci device
drivers/gpu/drm/xe/xe_configfs.h +2 -0
index 4fbbeafba473..42cd1a491d01 100644
@@ -24,6 +24,7 @@ bool xe_configfs_media_gt_allowed(struct pci_dev *pdev);
 u64 xe_configfs_get_engines_allowed(struct pci_dev *pdev);
 bool xe_configfs_get_psmi_enabled(struct pci_dev *pdev);
 bool xe_configfs_get_enable_multi_queue(struct pci_dev *pdev);
+bool xe_configfs_get_disable_vram_page_offline(struct pci_dev *pdev);
 u32 xe_configfs_get_ctx_restore_mid_bb(struct pci_dev *pdev,
 				       enum xe_engine_class class,
 				       const u32 **cs);
@@ -44,6 +45,7 @@ static inline bool xe_configfs_media_gt_allowed(struct pci_dev *pdev) { return t
 static inline u64 xe_configfs_get_engines_allowed(struct pci_dev *pdev) { return U64_MAX; }
 static inline bool xe_configfs_get_psmi_enabled(struct pci_dev *pdev) { return false; }
 static inline bool xe_configfs_get_enable_multi_queue(struct pci_dev *pdev) { return true; }
+static inline bool xe_configfs_get_disable_vram_page_offline(struct pci_dev *pdev) { return false; }
 static inline u32 xe_configfs_get_ctx_restore_mid_bb(struct pci_dev *pdev,
 						     enum xe_engine_class class,
 						     const u32 **cs) { return 0; }
Signature · 1 line
2.52.0
#18 lore raw

Re: [PATCH V20 10/15] drm/xe/configfs: Add disable_vram_page_offline attribute

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] Misplaced documentation comment restricting the attribute to CRI. --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] Misplaced documentation comment restricting the attribute to CRI. --

commit 4299f8f46041fce362473361938b350fd32fe701
Author: Tejas Upadhyay <[email protected]>

drm/xe/configfs: Add disable_vram_page_offline attribute

This commit adds a new configfs attribute disable_vram_page_offline to control how bad VRAM pages are handled. It allows administrators to disable active VRAM page offlining on Crescent Island (CRI) platforms, falling back to logging only.

Quoted 4 lines
drivers/gpu/drm/xe/xe_configfs.c +0 -0
index 052cce9621611..a9c72504c8a44 100644

[ ... ]

drivers/gpu/drm/xe/xe_configfs.c line 161

Quoted 5 lines
drivers/gpu/drm/xe/xe_configfs.c +1 -0
@@ -157,6 +158,18 @@
  *
  *	# echo 1 > /sys/kernel/config/xe/0000:03:00.0/enable_psmi
  *
+ * This attribute can only be set on CRI before binding to the device.

[Severity: Low]
Does this documentation string apply to enable_psmi or the newly added disable_vram_page_offline attribute?

Because it is placed above the "Disable VRAM page offline:" heading, it appears to document enable_psmi. However, since the code restricts disable_vram_page_offline to XE_CRESCENTISLAND (CRI), it seems this restriction text was intended for the new attribute instead.

Additionally, placing the new text here pushes the original "This attribute can only be set before binding to the device." text down into the disable_vram_page_offline section, which removes it from the enable_psmi section where it previously belonged.

Quoted 14 lines
drivers/gpu/drm/xe/xe_configfs.c +11 -0
+ *
+ * Disable VRAM page offline:
+ * ----------------------------
+ *
+ *  0, n, N, false - Do not disable (Offlining is active - default)
+ *  1, y, Y, true  - Disable vram page offline (Logging only)
+ *
+ *  Example to disable VRAM offline::
+ *
+ *      # echo 1 > /sys/kernel/config/xe/0000:03:00.0/disable_vram_page_offline
+ *
  * This attribute can only be set before binding to the device.
  *
  * Context restore BB
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=10
#19 lore raw

RE: [PATCH V20 10/15] drm/xe/configfs: Add disable_vram_page_offline attribute

Quoted 46 lines

-----Original Message-----
From: [email protected] <[email protected]>
Sent: 02 September 2026 22:05
To: Upadhyay, Tejas <[email protected]>
Cc: [email protected]
Subject: Re: [PATCH V20 10/15] drm/xe/configfs: Add disable_vram_page_offline attribute

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] Misplaced documentation comment restricting the attribute to CRI. --

commit 4299f8f46041fce362473361938b350fd32fe701
Author: Tejas Upadhyay <[email protected]>

drm/xe/configfs: Add disable_vram_page_offline attribute

This commit adds a new configfs attribute disable_vram_page_offline to control how bad VRAM pages are handled. It allows administrators to disable active VRAM page offlining on Crescent Island (CRI) platforms, falling back to logging only.

Quoted 5 lines
drivers/gpu/drm/xe/xe_configfs.c +0 -0
index 052cce9621611..a9c72504c8a44 100644

[ ... ]

drivers/gpu/drm/xe/xe_configfs.c line 161

Quoted 5 lines
drivers/gpu/drm/xe/xe_configfs.c +1 -0
@@ -157,6 +158,18 @@
  *
  *	# echo 1 > /sys/kernel/config/xe/0000:03:00.0/enable_psmi
  *
+ * This attribute can only be set on CRI before binding to the device.

[Severity: Low]
Does this documentation string apply to enable_psmi or the newly added disable_vram_page_offline attribute?

Because it is placed above the "Disable VRAM page offline:" heading, it appears to document enable_psmi. However, since the code restricts disable_vram_page_offline to XE_CRESCENTISLAND (CRI), it seems this restriction text was intended for the new attribute instead.

Quoted 4 lines

Additionally, placing the new text here pushes the original "This attribute can only be set before binding to the device." text down into the disable_vram_page_offline section, which removes it from the enable_psmi section where it previously belonged.

Nit, will not block. Will fix in follow up series if this is not blocked.

Tejas

Quoted 21 lines

drivers/gpu/drm/xe/xe_configfs.c

Quoted 11 lines
drivers/gpu/drm/xe/xe_configfs.c +10 -0
+ *
+ * Disable VRAM page offline:
+ * ----------------------------
+ *
+ *  0, n, N, false - Do not disable (Offlining is active - default)
+ *  1, y, Y, true  - Disable vram page offline (Logging only)
+ *
+ *  Example to disable VRAM offline::
+ *
+ *      # echo 1 >

/sys/kernel/config/xe/0000:03:00.0/disable_vram_page_offline

drivers/gpu/drm/xe/xe_configfs.c line 175

Quoted 4 lines
drivers/gpu/drm/xe/xe_configfs.c +1 -0
+ *
  * This attribute can only be set before binding to the device.
  *
  * Context restore BB

--
Sashiko AI review ·
https://sashiko.dev/#/patchset/20260902145343.465686-17-
[email protected]?part=10

#20 lore raw

PATCH v20 11/15 drm/xe/ras: Cache disable_vram_page_offline policy at init

In reply to Tejas Upadhyay

The configfs disable_vram_page_offline attribute can only be set before device bind, so its value is immutable at runtime. Cache it in struct xe_drm_ras during xe_ras_init() to avoid repeated configfs lookups on every fault.

---
 drivers/gpu/drm/xe/xe_drm_ras_types.h | 3 +++
 drivers/gpu/drm/xe/xe_ras.c           | 9 +++++++++
 2 files changed, 12 insertions(+)
drivers/gpu/drm/xe/xe_drm_ras_types.h +3 -0
index 8d729ad6a264..0be218ba2db7 100644
@@ -43,6 +43,9 @@ struct xe_drm_ras {
 
 	/** @info: info array for all types of errors */
 	struct xe_drm_ras_counter *info[DRM_XE_RAS_ERR_SEV_MAX];
+
+	/** @disable_vram_page_offline: cached configfs policy, immutable after init */
+	bool disable_vram_page_offline;
 };
 
 #endif
drivers/gpu/drm/xe/xe_ras.c +9 -0
index de4cb9ef7355..7a85735c57d5 100644
@@ -3,6 +3,7 @@
  * Copyright © 2026 Intel Corporation
  */
 
+#include "xe_configfs.h"
 #include "xe_debugfs.h"
 #include "xe_device.h"
 #include "xe_drm_ras.h"
@@ -915,6 +916,14 @@ void xe_ras_init(struct xe_device *xe)
 {
 	int ret;
 
+	/*
+	 * TODO: Replace platform check with xe->info.has_disable_vram_page_offline
+	 * once the feature flag is plumbed through device info.
+	 */
+	if (xe->info.platform == XE_CRESCENTISLAND)
+		xe->ras.disable_vram_page_offline =
+			xe_configfs_get_disable_vram_page_offline(to_pci_dev(xe->drm.dev));
+
 	xe_drm_ras_init(xe);
 
 	if (!xe->info.has_sysctrl)
Signature · 1 line
2.52.0
#21 lore raw

PATCH v20 12/15 drm/xe/vram: Check disable_vram_page_offline policy in fault handler

In reply to Tejas Upadhyay

Before reserving a page at a faulting address, check the cached disable_vram_page_offline policy from xe->ras. If the policy is enabled (logging only), log the corrupted address and return -EOPNOTSUPP so that RAS can report to firmware to drop the address from the SRAM queue without attempting to offline the page.

---
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +7 -0
index 0ab92bc3c699..00569c5279a8 100644
@@ -21,6 +21,7 @@
 #include "xe_lrc.h"
 #include "xe_mmio.h"
 #include "xe_pm.h"
+#include "xe_printk.h"
 #include "xe_res_cursor.h"
 #include "xe_ttm_stolen_mgr.h"
 #include "xe_ttm_vram_mgr.h"
@@ -906,6 +907,12 @@ int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)
 	vram_mgr = &vr->ttm;
 	mm = &vram_mgr->mm;
 
+	if (xe->ras.disable_vram_page_offline) {
+		xe_err(xe, "0x%llx is reported as corrupted address by HW\n",
+		       addr);
+		return -EOPNOTSUPP;
+	}
+
 	/* Reserve page at address */
 	return xe_ttm_vram_reserve_page_at_addr(xe, addr - vr->dpa_base, vram_mgr, mm);
 }
Signature · 1 line
2.52.0
#22 lore raw

Re: [PATCH V20 12/15] drm/xe/vram: Check disable_vram_page_offline policy in fault handler

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Medium] Unratelimited error logging in the VRAM address fault handler can cause dmesg flooding and DoS. --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Unratelimited error logging in the VRAM address fault handler can cause dmesg flooding and DoS. --

commit 5bfee849427fab2b54c95eec2934afe083fd4e87
Author: Tejas Upadhyay <[email protected]>

drm/xe/vram: Check disable_vram_page_offline policy in fault handler

This commit introduces a check in the VRAM address fault handler to verify the disable_vram_page_offline policy. If enabled, it logs the corrupted address and returns -EOPNOTSUPP so that RAS can report it without attempting to offline the page.

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index 0ab92bc3c6998..00569c5279a83 100644

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 912

Quoted 7 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
@@ -906,6 +907,12 @@ int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)
 	vram_mgr = &vr->ttm;
 	mm = &vram_mgr->mm;
 
+	if (xe->ras.disable_vram_page_offline) {
+		xe_err(xe, "0x%llx is reported as corrupted address by HW\n",
+		       addr);

[Severity: Medium]
Since this page is intentionally kept online and not offlined, can a corrupted VRAM page accessed in a tight loop by a userspace application, or polled continuously by the hardware, cause a log flood here?

Could this use a rate-limited logging function to prevent a potential local Denial of Service (DoS) from continuous dmesg writes, especially on systems with serial consoles?

Quoted 6 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+		return -EOPNOTSUPP;
+	}
+
 	/* Reserve page at address */
 	return xe_ttm_vram_reserve_page_at_addr(xe, addr - vr->dpa_base, vram_mgr, mm);
 }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=12
#23 lore raw

RE: [PATCH V20 12/15] drm/xe/vram: Check disable_vram_page_offline policy in fault handler

Quoted 46 lines

-----Original Message-----
From: [email protected] <[email protected]>
Sent: 02 September 2026 22:18
To: Upadhyay, Tejas <[email protected]>
Cc: [email protected]
Subject: Re: [PATCH V20 12/15] drm/xe/vram: Check disable_vram_page_offline policy in fault handler

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Unratelimited error logging in the VRAM address fault handler can cause dmesg flooding and DoS.
--

commit 5bfee849427fab2b54c95eec2934afe083fd4e87
Author: Tejas Upadhyay <[email protected]>

drm/xe/vram: Check disable_vram_page_offline policy in fault handler

This commit introduces a check in the VRAM address fault handler to verify the disable_vram_page_offline policy. If enabled, it logs the corrupted address and returns -EOPNOTSUPP so that RAS can report it without attempting to offline the page.

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index 0ab92bc3c6998..00569c5279a83 100644

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
@@ -906,6 +907,12 @@ int xe_ttm_vram_handle_addr_fault(struct

xe_device *xe, u64 addr)

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 912

Quoted 6 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
 	vram_mgr = &vr->ttm;
 	mm = &vram_mgr->mm;
 
+	if (xe->ras.disable_vram_page_offline) {
+		xe_err(xe, "0x%llx is reported as corrupted address by HW\n",
+		       addr);

[Severity: Medium]
Since this page is intentionally kept online and not offlined, can a corrupted VRAM page accessed in a tight loop by a userspace application, or polled continuously by the hardware, cause a log flood here?

Quoted 3 lines

Could this use a rate-limited logging function to prevent a potential local Denial of Service (DoS) from continuous dmesg writes, especially on systems with serial consoles?

Not expecting amount of errors, which can flood. So what we have looks ok.

Tejas

Quoted 12 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)

Quoted 7 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+		return -EOPNOTSUPP;
+	}
+
 	/* Reserve page at address */
 	return xe_ttm_vram_reserve_page_at_addr(xe, addr - vr->dpa_base,
 vram_mgr, mm);  }

--
Sashiko AI review ·
https://sashiko.dev/#/patchset/20260902145343.465686-17-
[email protected]?part=12

#24 lore raw

Re: [PATCH V20 12/15] drm/xe/vram: Check disable_vram_page_offline policy in fault handler

In reply to Tejas Upadhyay
Quoted 27 lines · Tejas Upadhyay

Before reserving a page at a faulting address, check the cached disable_vram_page_offline policy from xe->ras. If the policy is enabled (logging only), log the corrupted address and return -EOPNOTSUPP so that RAS can report to firmware to drop the address from the SRAM queue without attempting to offline the page.

---
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -0
index 0ab92bc3c699..00569c5279a8 100644
@@ -21,6 +21,7 @@
 #include "xe_lrc.h"
 #include "xe_mmio.h"
 #include "xe_pm.h"
+#include "xe_printk.h"
 #include "xe_res_cursor.h"
 #include "xe_ttm_stolen_mgr.h"
 #include "xe_ttm_vram_mgr.h"

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 910

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -0
@@ -906,6 +907,12 @@ int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)
 	vram_mgr = &vr->ttm;
 	mm = &vram_mgr->mm;
 
+	if (xe->ras.disable_vram_page_offline) {

shouldn't this be checked by the caller?
here it looks like a layer violation

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 912

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+		xe_err(xe, "0x%llx is reported as corrupted address by HW\n",
+		       addr);

shouldn't we use xe_log() for this?
and also done by the caller?

Quoted 6 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+		return -EOPNOTSUPP;
+	}
+
 	/* Reserve page at address */
 	return xe_ttm_vram_reserve_page_at_addr(xe, addr - vr->dpa_base, vram_mgr, mm);
 }
#25 lore raw

PATCH v20 13/15 drm/xe: Expose bad VRAM pages via debugfs

In reply to Tejas Upadhyay

Add a debugfs file "vram_bad_pages" that shows offlined and queued VRAM pages across all tiles. Each entry displays the page frame number, GPU page size, and status flag (R=reserved, P=pending, F=failed).

example,
cat /sys/kernel/debug/dri/0/vram_bad_pages

max_pages: 10000
0x0000000000000000 : 0x0000000000001000 : R
0x0000000000001234 : 0x0000000000001000 : P
0x0000000000080000 : 0x0000000000001000 : R ← tile 1 addr

v2(Sashikoi/Michal/Himal):
-Remove block iteration, use offline and queue list only -Move platform check inside api

---
 drivers/gpu/drm/xe/xe_debugfs.c            |  3 ++
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c       | 59 ++++++++++++++++++++++
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.h       |  2 +
 drivers/gpu/drm/xe/xe_ttm_vram_mgr_types.h |  2 +
 4 files changed, 66 insertions(+)
drivers/gpu/drm/xe/xe_debugfs.c +3 -0
index 28135f84e286..cb24e001b142 100644
@@ -32,6 +32,7 @@
 #include "xe_sriov_vf.h"
 #include "xe_step.h"
 #include "xe_tile_debugfs.h"
+#include "xe_ttm_vram_mgr.h"
 #include "xe_vsec.h"
 #include "xe_wa.h"
 
@@ -773,6 +774,8 @@ void xe_debugfs_register(struct xe_device *xe)
 	if (man)
 		ttm_resource_manager_create_debugfs(man, root, "stolen_mm");
 
+	xe_ttm_vram_debugfs_init(xe, root);
+
 	for_each_tile(tile, xe, tile_id)
 		xe_tile_debugfs_register(tile);
 
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +59 -0

68 lines not on this page · raw message

drivers/gpu/drm/xe/xe_ttm_vram_mgr.h +2 -0
index d5392beff30c..f354c26c4257 100644
@@ -9,6 +9,7 @@
 #include "xe_ttm_vram_mgr_types.h"
 
 enum dma_data_direction;
+struct dentry;
 struct xe_device;
 struct xe_tile;
 struct xe_vram_region;
@@ -32,6 +33,7 @@ void xe_ttm_vram_get_used(struct ttm_resource_manager *man,
 			  u64 *used, u64 *used_visible);
 
 int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr);
+void xe_ttm_vram_debugfs_init(struct xe_device *xe, struct dentry *root);
 static inline struct xe_ttm_vram_mgr_resource *
 to_xe_ttm_vram_mgr_resource(struct ttm_resource *res)
 {
drivers/gpu/drm/xe/xe_ttm_vram_mgr_types.h +2 -0
index dc97b0ad0e51..efcf3e1d4e80 100644
@@ -37,6 +37,8 @@ struct xe_ttm_vram_mgr {
 	struct mutex lock;
 	/** @mem_type: The TTM memory type */
 	u32 mem_type;
+	/** @max_pages: max pages that can be in offline queue retrieved from FW */
+	u16 max_pages;
 };
 
 /**
Signature · 1 line
2.52.0
#26 lore raw

Re: [PATCH V20 13/15] drm/xe: Expose bad VRAM pages via debugfs

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [Low] Debugfs output computes GPU VRAM page frame numbers and sizes using host CPU macros (PAGE_SHIFT and…

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Low] Debugfs output computes GPU VRAM page frame numbers and sizes using host CPU macros (PAGE_SHIFT and PAGE_SIZE), causing silent truncation and incorrect reporting on architectures with non-4K CPU pages.
- [Low] The VRAM max_pages limit is printed globally using only tile 0's manager, ignoring limits of other tiles on multi-tile devices.
--

commit c12841d073a619e5edeb8c8306adf6fd03042edd
Author: Tejas Upadhyay <[email protected]>

drm/xe: Expose bad VRAM pages via debugfs

This commit adds a debugfs file to show offlined and queued VRAM pages across all tiles, displaying the page frame number, GPU page size, and status flag.

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index 00569c5279a83..d9da2454d968e 100644

[ ... ]

Quoted 9 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +5 -0
@@ -917,3 +918,61 @@ int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)
 	return xe_ttm_vram_reserve_page_at_addr(xe, addr - vr->dpa_base, vram_mgr, mm);
 }
 EXPORT_SYMBOL(xe_ttm_vram_handle_addr_fault);
+
+static int vram_bad_pages_show(struct seq_file *m, void *unused)
+{
+	struct xe_device *xe = m->private;
+	struct xe_ttm_vram_offline_resource *pos;

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 935

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+	struct ttm_resource_manager *man;
+	struct xe_ttm_vram_mgr *mgr;
+	struct xe_tile *tile;
+	u8 id;
+
+	man = ttm_manager_type(&xe->ttm, XE_PL_VRAM0);
+	if (man)
+		/* TODO Hook with RAS to show max_pages fetched from FW */
+		seq_printf(m, "max_pages: %d\n",
+			   to_xe_ttm_vram_mgr(man)->max_pages);

[Severity: Low]
Does this correctly represent the limit for multi-tile devices?

By querying only the manager for XE_PL_VRAM0, it seems the max_pages value is printed as a global limit, which might ignore independent offline queue limits configured on subsequent tiles.

Quoted 6 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +6 -0
+
+	for_each_tile(tile, xe, id) {
+		struct xe_vram_region *vr = tile->mem.vram;
+
+		man = ttm_manager_type(&xe->ttm, XE_PL_VRAM0 + id);
+		if (!man || !vr)

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 951

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+			continue;
+		mgr = to_xe_ttm_vram_mgr(man);
+
+		rcu_read_lock();
+
+		list_for_each_entry_rcu(pos, &mgr->offlined_pages, offlined_link) {
+			u64 pfn;
+
+			pfn = (pos->addr + vr->dpa_base) >> PAGE_SHIFT;
+			seq_printf(m, "0x%016llx : 0x%016lx : R\n", pfn, PAGE_SIZE);

[Severity: Low]
Can this cause incorrect PFN calculation and page size reporting on architectures with a non-4K CPU page size?

Using the host CPU macros PAGE_SHIFT and PAGE_SIZE to compute GPU VRAM addresses couples the hardware memory topology to host CPU page boundaries. On a system with 64K CPU pages, shifting the 4K-aligned GPU address by 16 truncates the lower 4 bits of the GPU PFN, and reports 64K as the VRAM page size instead of 4K.

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 959

Quoted 8 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +8 -0
+		}
+
+		list_for_each_entry_rcu(pos, &mgr->queued_pages, queued_link) {
+			u64 pfn;
+
+			pfn = (pos->addr + vr->dpa_base) >> PAGE_SHIFT;
+			seq_printf(m, "0x%016llx : 0x%016lx : %c\n",
+				   pfn, PAGE_SIZE, pos->status ? 'F' : 'P');

[Severity: Low]
This has the same PAGE_SHIFT and PAGE_SIZE host CPU macro coupling as above, which may truncate the PFN and report the wrong size on non-4K CPU architectures.

Quoted 7 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +7 -0
+		}
+
+		rcu_read_unlock();
+	}
+
+	return 0;
+}

[ ... ]

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=13
#27 lore raw

RE: [PATCH V20 13/15] drm/xe: Expose bad VRAM pages via debugfs

Quoted 58 lines

-----Original Message-----
From: [email protected] <[email protected]>
Sent: 02 September 2026 22:27
To: Upadhyay, Tejas <[email protected]>
Cc: [email protected]
Subject: Re: [PATCH V20 13/15] drm/xe: Expose bad VRAM pages via debugfs

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Low] Debugfs output computes GPU VRAM page frame numbers and sizes using host CPU macros (PAGE_SHIFT and PAGE_SIZE), causing silent truncation and incorrect reporting on architectures with non-4K CPU pages.
- [Low] The VRAM max_pages limit is printed globally using only tile 0's manager, ignoring limits of other tiles on multi-tile devices.
--

commit c12841d073a619e5edeb8c8306adf6fd03042edd
Author: Tejas Upadhyay <[email protected]>

drm/xe: Expose bad VRAM pages via debugfs

This commit adds a debugfs file to show offlined and queued VRAM pages across all tiles, displaying the page frame number, GPU page size, and status flag.

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index 00569c5279a83..d9da2454d968e 100644

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
@@ -917,3 +918,61 @@ int xe_ttm_vram_handle_addr_fault(struct

xe_device *xe, u64 addr)

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
 	return xe_ttm_vram_reserve_page_at_addr(xe, addr - vr->dpa_base,
 vram_mgr, mm);  }  EXPORT_SYMBOL(xe_ttm_vram_handle_addr_fault);
+
+static int vram_bad_pages_show(struct seq_file *m, void *unused) {

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+	struct xe_device *xe = m->private;
+	struct xe_ttm_vram_offline_resource *pos;
+	struct ttm_resource_manager *man;
+	struct xe_ttm_vram_mgr *mgr;
+	struct xe_tile *tile;
+	u8 id;
+
+	man = ttm_manager_type(&xe->ttm, XE_PL_VRAM0);
+	if (man)
+		/* TODO Hook with RAS to show max_pages fetched from FW

*/

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 935

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+		seq_printf(m, "max_pages: %d\n",
+			   to_xe_ttm_vram_mgr(man)->max_pages);

[Severity: Low]
Does this correctly represent the limit for multi-tile devices?

Quoted 3 lines

By querying only the manager for XE_PL_VRAM0, it seems the max_pages value is printed as a global limit, which might ignore independent offline queue limits configured on subsequent tiles.

This will come from RAS post integration with RAS and it will be same for all tile.

Quoted 24 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)

Quoted 13 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +12 -0
+
+	for_each_tile(tile, xe, id) {
+		struct xe_vram_region *vr = tile->mem.vram;
+
+		man = ttm_manager_type(&xe->ttm, XE_PL_VRAM0 + id);
+		if (!man || !vr)
+			continue;
+		mgr = to_xe_ttm_vram_mgr(man);
+
+		rcu_read_lock();
+
+		list_for_each_entry_rcu(pos, &mgr->offlined_pages,

offlined_link) {

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+			u64 pfn;
+
+			pfn = (pos->addr + vr->dpa_base) >> PAGE_SHIFT;
+			seq_printf(m, "0x%016llx : 0x%016lx : R\n", pfn,

PAGE_SIZE);

[Severity: Low]
Can this cause incorrect PFN calculation and page size reporting on architectures with a non-4K CPU page size?

Quoted 6 lines

Using the host CPU macros PAGE_SHIFT and PAGE_SIZE to compute GPU VRAM addresses couples the hardware memory topology to host CPU page boundaries.
On a system with 64K CPU pages, shifting the 4K-aligned GPU address by 16 truncates the lower 4 bits of the GPU PFN, and reports 64K as the VRAM page size instead of 4K.

We have 4K aligned pages everywhere so for CRI it covers alright.

Tejas

Quoted 30 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+		}
+
+		list_for_each_entry_rcu(pos, &mgr->queued_pages,

queued_link) {

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 959

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +5 -0
+			u64 pfn;
+
+			pfn = (pos->addr + vr->dpa_base) >> PAGE_SHIFT;
+			seq_printf(m, "0x%016llx : 0x%016lx : %c\n",
+				   pfn, PAGE_SIZE, pos->status ? 'F' : 'P');

[Severity: Low]
This has the same PAGE_SHIFT and PAGE_SIZE host CPU macro coupling as above, which may truncate the PFN and report the wrong size on non-4K CPU architectures.

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 966

Quoted 7 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +7 -0
+		}
+
+		rcu_read_unlock();
+	}
+
+	return 0;
+}

[ ... ]

--
Sashiko AI review ·
https://sashiko.dev/#/patchset/20260902145343.465686-17-
[email protected]?part=13

maintainer
#28 lore raw

Re: [PATCH V20 13/15] drm/xe: Expose bad VRAM pages via debugfs

Quoted 88 lines · Tejas
Quoted 58 lines

-----Original Message-----
From: [email protected] <[email protected]>
Sent: 02 September 2026 22:27
To: Upadhyay, Tejas <[email protected]>
Cc: [email protected]
Subject: Re: [PATCH V20 13/15] drm/xe: Expose bad VRAM pages via debugfs

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Low] Debugfs output computes GPU VRAM page frame numbers and sizes using host CPU macros (PAGE_SHIFT and PAGE_SIZE), causing silent truncation and incorrect reporting on architectures with non-4K CPU pages.
- [Low] The VRAM max_pages limit is printed globally using only tile 0's manager, ignoring limits of other tiles on multi-tile devices.
--

commit c12841d073a619e5edeb8c8306adf6fd03042edd
Author: Tejas Upadhyay <[email protected]>

drm/xe: Expose bad VRAM pages via debugfs

This commit adds a debugfs file to show offlined and queued VRAM pages across all tiles, displaying the page frame number, GPU page size, and status flag.

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
index 00569c5279a83..d9da2454d968e 100644

[ ... ]

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct

Quoted 1 line
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0
@@ -917,3 +918,61 @@ int xe_ttm_vram_handle_addr_fault(struct

xe_device *xe, u64 addr)

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
 	return xe_ttm_vram_reserve_page_at_addr(xe, addr - vr->dpa_base,
 vram_mgr, mm);  }  EXPORT_SYMBOL(xe_ttm_vram_handle_addr_fault);
+
+static int vram_bad_pages_show(struct seq_file *m, void *unused) {

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+	struct xe_device *xe = m->private;
+	struct xe_ttm_vram_offline_resource *pos;
+	struct ttm_resource_manager *man;
+	struct xe_ttm_vram_mgr *mgr;
+	struct xe_tile *tile;
+	u8 id;
+
+	man = ttm_manager_type(&xe->ttm, XE_PL_VRAM0);
+	if (man)
+		/* TODO Hook with RAS to show max_pages fetched from FW

*/

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 935

Quoted 2 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +2 -0
+		seq_printf(m, "max_pages: %d\n",
+			   to_xe_ttm_vram_mgr(man)->max_pages);

[Severity: Low]
Does this correctly represent the limit for multi-tile devices?

Quoted 3 lines

By querying only the manager for XE_PL_VRAM0, it seems the max_pages value is printed as a global limit, which might ignore independent offline queue limits configured on subsequent tiles.

This will come from RAS post integration with RAS and it will be same for all tile.

Quoted 24 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)

Quoted 13 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +12 -0
+
+	for_each_tile(tile, xe, id) {
+		struct xe_vram_region *vr = tile->mem.vram;
+
+		man = ttm_manager_type(&xe->ttm, XE_PL_VRAM0 + id);
+		if (!man || !vr)
+			continue;
+		mgr = to_xe_ttm_vram_mgr(man);
+
+		rcu_read_lock();
+
+		list_for_each_entry_rcu(pos, &mgr->offlined_pages,

offlined_link) {

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +4 -0
+			u64 pfn;
+
+			pfn = (pos->addr + vr->dpa_base) >> PAGE_SHIFT;
+			seq_printf(m, "0x%016llx : 0x%016lx : R\n", pfn,

PAGE_SIZE);

[Severity: Low]
Can this cause incorrect PFN calculation and page size reporting on architectures with a non-4K CPU page size?

Quoted 9 lines
Quoted 6 lines

Using the host CPU macros PAGE_SHIFT and PAGE_SIZE to compute GPU VRAM addresses couples the hardware memory topology to host CPU page boundaries.
On a system with 64K CPU pages, shifting the 4K-aligned GPU address by 16 truncates the lower 4 bits of the GPU PFN, and reports 64K as the VRAM page size instead of 4K.

We have 4K aligned pages everywhere so for CRI it covers alright.

I think you use SZ_4K in a bunch of places in this series where you actually want PAGE_SIZE.

xe_ttm_vram_mgr_init uses PAGE_SIZE no SZ_4K, in practice these are value but using PAGE_SIZE is correct for future proofing or possibly allowing non-x86 CPUs to plug into a dGPU.

Matt

Quoted 31 lines

Tejas

Quoted 30 lines

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +3 -0
+		}
+
+		list_for_each_entry_rcu(pos, &mgr->queued_pages,

queued_link) {

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 959

Quoted 5 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +5 -0
+			u64 pfn;
+
+			pfn = (pos->addr + vr->dpa_base) >> PAGE_SHIFT;
+			seq_printf(m, "0x%016llx : 0x%016lx : %c\n",
+				   pfn, PAGE_SIZE, pos->status ? 'F' : 'P');

[Severity: Low]
This has the same PAGE_SHIFT and PAGE_SIZE host CPU macro coupling as above, which may truncate the PFN and report the wrong size on non-4K CPU architectures.

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 966

Quoted 7 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +7 -0
+		}
+
+		rcu_read_unlock();
+	}
+
+	return 0;
+}

[ ... ]

--
Sashiko AI review ·
https://sashiko.dev/#/patchset/20260902145343.465686-17-
[email protected]?part=13

#29 lore raw

Re: [PATCH V20 13/15] drm/xe: Expose bad VRAM pages via debugfs

In reply to Tejas Upadhyay
Quoted 102 lines · Tejas Upadhyay

Add a debugfs file "vram_bad_pages" that shows offlined and queued VRAM pages across all tiles. Each entry displays the page frame number, GPU page size, and status flag (R=reserved, P=pending, F=failed).

example,
cat /sys/kernel/debug/dri/0/vram_bad_pages

max_pages: 10000
0x0000000000000000 : 0x0000000000001000 : R
0x0000000000001234 : 0x0000000000001000 : P
0x0000000000080000 : 0x0000000000001000 : R ← tile 1 addr

v2(Sashikoi/Michal/Himal):

(quoted diff) +0 -2
-Remove block iteration, use offline and queue list only
-Move platform check inside api
---
 drivers/gpu/drm/xe/xe_debugfs.c            |  3 ++
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c       | 59 ++++++++++++++++++++++
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.h       |  2 +
 drivers/gpu/drm/xe/xe_ttm_vram_mgr_types.h |  2 +
 4 files changed, 66 insertions(+)
drivers/gpu/drm/xe/xe_debugfs.c +3 -0
index 28135f84e286..cb24e001b142 100644
@@ -32,6 +32,7 @@
 #include "xe_sriov_vf.h"
 #include "xe_step.h"
 #include "xe_tile_debugfs.h"
+#include "xe_ttm_vram_mgr.h"
 #include "xe_vsec.h"
 #include "xe_wa.h"
 
@@ -773,6 +774,8 @@ void xe_debugfs_register(struct xe_device *xe)
 	if (man)
 		ttm_resource_manager_create_debugfs(man, root, "stolen_mm");
 
+	xe_ttm_vram_debugfs_init(xe, root);
+
 	for_each_tile(tile, xe, tile_id)
 		xe_tile_debugfs_register(tile);
 
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +39 -0
index 00569c5279a8..d9da2454d968 100644
@@ -5,6 +5,7 @@
  */
 
 #include <linux/cgroup_dmem.h>
+#include <linux/debugfs.h>
 
 #include <drm/drm_managed.h>
 #include <drm/drm_drv.h>
@@ -917,3 +918,61 @@ int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr)
 	return xe_ttm_vram_reserve_page_at_addr(xe, addr - vr->dpa_base, vram_mgr, mm);
 }
 EXPORT_SYMBOL(xe_ttm_vram_handle_addr_fault);
+
+static int vram_bad_pages_show(struct seq_file *m, void *unused)
+{
+	struct xe_device *xe = m->private;
+	struct xe_ttm_vram_offline_resource *pos;
+	struct ttm_resource_manager *man;
+	struct xe_ttm_vram_mgr *mgr;
+	struct xe_tile *tile;
+	u8 id;
+
+	man = ttm_manager_type(&xe->ttm, XE_PL_VRAM0);
+	if (man)
+		/* TODO Hook with RAS to show max_pages fetched from FW */
+		seq_printf(m, "max_pages: %d\n",
+			   to_xe_ttm_vram_mgr(man)->max_pages);
+
+	for_each_tile(tile, xe, id) {
+		struct xe_vram_region *vr = tile->mem.vram;
+
+		man = ttm_manager_type(&xe->ttm, XE_PL_VRAM0 + id);
+		if (!man || !vr)
+			continue;
+		mgr = to_xe_ttm_vram_mgr(man);
+
+		rcu_read_lock();
+
+		list_for_each_entry_rcu(pos, &mgr->offlined_pages, offlined_link) {
+			u64 pfn;
+
+			pfn = (pos->addr + vr->dpa_base) >> PAGE_SHIFT;
+			seq_printf(m, "0x%016llx : 0x%016lx : R\n", pfn, PAGE_SIZE);
+		}
+
+		list_for_each_entry_rcu(pos, &mgr->queued_pages, queued_link) {
+			u64 pfn;
+
+			pfn = (pos->addr + vr->dpa_base) >> PAGE_SHIFT;
+			seq_printf(m, "0x%016llx : 0x%016lx : %c\n",

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr) · line 968

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+				   pfn, PAGE_SIZE, pos->status ? 'F' : 'P');
+		}
+
+		rcu_read_unlock();
+	}
+
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(vram_bad_pages);
+

missing kernel-doc

Quoted 23 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+void xe_ttm_vram_debugfs_init(struct xe_device *xe, struct dentry *root)
+{
+	/*
+	 * TODO: Replace platform check with xe->info
+	 * once the feature flag is plumbed through device info.
+	 */
+	if (xe->info.platform != XE_CRESCENTISLAND)
+		return;
+	debugfs_create_file("vram_bad_pages", 0444, root, xe, &vram_bad_pages_fops);
+}
drivers/gpu/drm/xe/xe_ttm_vram_mgr.h +1 -0
index d5392beff30c..f354c26c4257 100644
@@ -9,6 +9,7 @@
 #include "xe_ttm_vram_mgr_types.h"
 
 enum dma_data_direction;
+struct dentry;
 struct xe_device;
 struct xe_tile;
 struct xe_vram_region;

drivers/gpu/drm/xe/xe_ttm_vram_mgr.h void xe_ttm_vram_get_used(struct ttm_resource_manager *man, · line 36

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.h +1 -0
@@ -32,6 +33,7 @@ void xe_ttm_vram_get_used(struct ttm_resource_manager *man,
 			  u64 *used, u64 *used_visible);
 
 int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr);
+void xe_ttm_vram_debugfs_init(struct xe_device *xe, struct dentry *root);

nit: I would put this declaration closer to other init() functions

Quoted 16 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.h +0 -0
 static inline struct xe_ttm_vram_mgr_resource *
 to_xe_ttm_vram_mgr_resource(struct ttm_resource *res)
 {
drivers/gpu/drm/xe/xe_ttm_vram_mgr_types.h +2 -0
index dc97b0ad0e51..efcf3e1d4e80 100644
@@ -37,6 +37,8 @@ struct xe_ttm_vram_mgr {
 	struct mutex lock;
 	/** @mem_type: The TTM memory type */
 	u32 mem_type;
+	/** @max_pages: max pages that can be in offline queue retrieved from FW */
+	u16 max_pages;
 };
 
 /**
#30 lore raw

PATCH v20 14/15 drm/xe/uapi: Expose ban reason in EXEC_QUEUE_GET_PROPERTY_BAN

In reply to Tejas Upadhyay

Extend DRM_XE_EXEC_QUEUE_GET_PROPERTY_BAN to return a bitmask indicating the reason for the ban, rather than a simple boolean. This allows userspace to distinguish between different ban causes:

- DRM_XE_EXEC_QUEUE_BAN_REASON_GPU_HANG (bit 0): exec queue was banned due to a GPU hang or job timeout detected by the TDR.
- DRM_XE_EXEC_QUEUE_BAN_REASON_PAGE_OFFLINE (bit 1): exec queue was banned because a VRAM page backing its resources was taken offline.

The ban_reason field is added to struct xe_exec_queue and set at the point where the ban is triggered:
- In guc_exec_queue_timedout_job() for GPU hang.
- In xe_ttm_vram_purge_page() for memory page offline, before calling xe_exec_queue_kill() or xe_vm_kill().

The reset_status op is updated to return u64 with the reason bitmask. When a queue is banned but no explicit reason was recorded (e.g., from a generic CAT error), it defaults to GPU_HANG for backward compatibility. A value of 0 means the exec queue is not banned.

v5 (Sashiko/MattB):
- Take the write lock for the traversal to tag ban_reason v4(Sashiko):
- Add ban reason for non-LR exec queues
- Add TODO for multiqueue
v3(Rodrigo):
- Add doc in xe_drm.h
v2(Sashiko):
- Use atomic_t for ban_reason to fix concurrent updates from TDR and page-offline
- Guard GPU_HANG bit with !exec_queue_killed to avoid masking page-offline reason
- Clear ban_reason on queue recovery (clear_exec_queue_banned path)
- Use atomic_read in guc_exec_queue_reset_status for lockless read

---
 drivers/gpu/drm/xe/xe_exec_queue_types.h |  7 +++--
 drivers/gpu/drm/xe/xe_execlist.c         |  4 +--
 drivers/gpu/drm/xe/xe_guc_submit.c       | 36 ++++++++++++++++++++----
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c     | 27 ++++++++++++++++++
 include/uapi/drm/xe_drm.h                | 18 +++++++++++-
 5 files changed, 82 insertions(+), 10 deletions(-)
drivers/gpu/drm/xe/xe_exec_queue_types.h +5 -2
index 95f75d61a647..836f88fc0faa 100644
@@ -154,6 +154,9 @@ struct xe_exec_queue {
 	 */
 	unsigned long flags;
 
+	/** @ban_reason: Bitmask of ban reasons (DRM_XE_EXEC_QUEUE_BAN_REASON_*) */
+	atomic_t ban_reason;
+
 	union {
 		/** @multi_gt_list: list head for VM bind engines if multi-GT */
 		struct list_head multi_gt_list;
@@ -348,8 +351,8 @@ struct xe_exec_queue_ops {
 	 * signalled when this function is called.
 	 */
 	void (*resume)(struct xe_exec_queue *q);
-	/** @reset_status: check exec queue reset status */
-	bool (*reset_status)(struct xe_exec_queue *q);
+	/** @reset_status: check exec queue ban status, returns ban reason bitmask */
+	u64 (*reset_status)(struct xe_exec_queue *q);
 };
 
 #endif
drivers/gpu/drm/xe/xe_execlist.c +2 -2
index 0d0db66c6ea2..a36db39dcda8 100644
@@ -453,10 +453,10 @@ static void execlist_exec_queue_resume(struct xe_exec_queue *q)
 	/* NIY */
 }
 
-static bool execlist_exec_queue_reset_status(struct xe_exec_queue *q)
+static u64 execlist_exec_queue_reset_status(struct xe_exec_queue *q)
 {
 	/* NIY */
-	return false;
+	return 0;
 }
 
 static const struct xe_exec_queue_ops execlist_exec_queue_ops = {
drivers/gpu/drm/xe/xe_guc_submit.c +31 -5
index 99d8c807ff05..0a6e2b81b5a5 100644
@@ -6,6 +6,7 @@
 #include "xe_guc_submit.h"
 
 #include <linux/bitfield.h>
+#include <uapi/drm/xe_drm.h>
 #include <linux/bitmap.h>
 #include <linux/circ_buf.h>
 #include <linux/dma-fence-array.h>
@@ -1599,6 +1600,12 @@ guc_exec_queue_timedout_job(struct drm_sched_job *drm_job)
 	else
 		wedged = xe_device_wedged(xe);
 
+	/*
+	 * Only tag as GPU hang if this is the original timeout, not a
+	 * consequence of a prior kill (e.g., page-offline).
+	 */
+	if (!exec_queue_killed(q))
+		atomic_or(DRM_XE_EXEC_QUEUE_BAN_REASON_GPU_HANG, &q->ban_reason);
 	set_exec_queue_banned(q);
 
 	/* Kick job / queue off hardware */
@@ -1682,6 +1689,9 @@ guc_exec_queue_timedout_job(struct drm_sched_job *drm_job)
 		if (timeout_needs_gt_reset(q, job, skip_timeout_check)) {
 			if (!xe_sched_invalidate_job(job, 2)) {
 				clear_exec_queue_banned(q);
+				/* protect concurrent page offline reasons */
+				atomic_andnot(DRM_XE_EXEC_QUEUE_BAN_REASON_GPU_HANG,
+					      &q->ban_reason);
 				xe_gt_reset_async(q->gt);
 				goto rearm;
 			}
@@ -2580,13 +2590,29 @@ static void guc_exec_queue_multi_queue_drop_suspend(struct xe_exec_queue *q)
 	}
 }
 
-static bool guc_exec_queue_reset_status(struct xe_exec_queue *q)
+static u64 guc_exec_queue_reset_status(struct xe_exec_queue *q)
 {
-	if (xe_exec_queue_is_multi_queue_secondary(q) &&
-	    guc_exec_queue_reset_status(xe_exec_queue_multi_queue_primary(q)))
-		return true;
+	/* TODO: In case of multiqueue, if a secondary queue is banned due to
+	 * page offlining, checking only the primary queue's GuC reset status
+	 * may mask the true reason or race with it.
+	 */
+	if (xe_exec_queue_is_multi_queue_secondary(q)) {
+		u64 status = guc_exec_queue_reset_status(xe_exec_queue_multi_queue_primary(q));
 
-	return exec_queue_reset(q) || exec_queue_killed_or_banned_or_wedged(q);
+		if (status)
+			return status;
+	}
+
+	if (exec_queue_reset(q) || exec_queue_killed_or_banned_or_wedged(q)) {
+		u64 reason = atomic_read_acquire(&q->ban_reason);
+
+		/* If no specific reason was recorded, default to GPU hang */
+		if (!reason)
+			reason = DRM_XE_EXEC_QUEUE_BAN_REASON_GPU_HANG;
+		return reason;
+	}
+
+	return 0;
 }
 
 /*
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +27 -0
index d9da2454d968..3c17f906a549 100644
@@ -10,6 +10,7 @@
 #include <drm/drm_managed.h>
 #include <drm/drm_drv.h>
 #include <drm/drm_buddy.h>
+#include <uapi/drm/xe_drm.h>
 
 #include <drm/ttm/ttm_placement.h>
 #include <drm/ttm/ttm_range_manager.h>
@@ -623,6 +624,7 @@ u64 xe_ttm_vram_get_avail(struct ttm_resource_manager *man)
 
 static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo)
 {
+	u32 q_flag = DRM_XE_EXEC_QUEUE_BAN_REASON_PAGE_OFFLINE;
 	struct ttm_operation_ctx ctx = {};
 	struct xe_exec_queue *q_to_put = NULL;
 	struct xe_exec_queue *q = NULL;
@@ -637,7 +639,30 @@ static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo)
 	xe_bo_unlock(bo);
 	/*  Ban VM if BO is PPGTT */
 	if (vm && (flags & XE_BO_FLAG_PAGETABLE)) {
+		struct xe_exec_queue *eq;
+		int id;
+
 		down_write(&vm->lock);
+		if (xe->info.has_ctx_tlb_inval) {
+			/*
+			 * Must be the write lock: send_tlb_inval_ctx_ppgtt()
+			 * mutates this list (list_move_tail() onto an on-stack
+			 * head) while holding only the read lock, relying on
+			 * tlb_inval->seqno_lock to keep itself the sole
+			 * mutator. Traversing it under down_read() would let
+			 * this walk follow entries onto that stack list.
+			 */
+			down_write(&vm->exec_queues.lock);
+			for (id = 0; id < ARRAY_SIZE(vm->exec_queues.list); id++)
+				list_for_each_entry(eq, &vm->exec_queues.list[id],
+						    vm_exec_queue_link)
+					atomic_or(q_flag, &eq->ban_reason);
+			up_write(&vm->exec_queues.lock);
+		} else {
+			list_for_each_entry(eq, &vm->preempt.exec_queues, lr.link)
+				atomic_or(q_flag, &eq->ban_reason);
+		}
+		smp_wmb(); /* Force all queue bits to be visible before killing the VM */
 		xe_vm_kill(vm, true);
 		up_write(&vm->lock);
 	}
@@ -649,6 +674,8 @@ static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo)
 	/*  Ban exec queue if BO is lrc */
 	if (q && xe_exec_queue_get_unless_zero(q)) {
 		/* ban queue */
+		atomic_or(q_flag, &q->ban_reason);
+		smp_wmb(); /* Force bit change to finish before state change triggers */
 		q_to_put = q;
 	}
 
include/uapi/drm/xe_drm.h +17 -1
index 509202a7b13e..ee4a921b2e6e 100644
@@ -1491,6 +1491,12 @@ struct drm_xe_exec_queue_destroy {
  *
  * The @property can be:
  *  - %DRM_XE_EXEC_QUEUE_GET_PROPERTY_BAN
+ *
+ * For %DRM_XE_EXEC_QUEUE_GET_PROPERTY_BAN, @value is a bitmask of ban reasons:
+ *  - %DRM_XE_EXEC_QUEUE_BAN_REASON_GPU_HANG - banned due to GPU hang/timeout
+ *  - %DRM_XE_EXEC_QUEUE_BAN_REASON_PAGE_OFFLINE - banned due to memory page offline
+ *
+ * A @value of 0 means the exec queue is not banned.
  */
 struct drm_xe_exec_queue_get_property {
 	/** @extensions: Pointer to the first extension struct, if any */
@@ -1503,7 +1509,17 @@ struct drm_xe_exec_queue_get_property {
 	/** @property: property to get */
 	__u32 property;
 
-	/** @value: property value */
+	/**
+	 * @value: property value
+	 *
+	 * For %DRM_XE_EXEC_QUEUE_GET_PROPERTY_BAN, this is a bitmask of:
+	 *  - %DRM_XE_EXEC_QUEUE_BAN_REASON_GPU_HANG - banned due to GPU hang/timeout
+	 *  - %DRM_XE_EXEC_QUEUE_BAN_REASON_PAGE_OFFLINE - banned due to memory page offline
+	 *
+	 * Value of 0 means the exec queue is not banned.
+	 */
+#define DRM_XE_EXEC_QUEUE_BAN_REASON_GPU_HANG		(1 << 0)
+#define DRM_XE_EXEC_QUEUE_BAN_REASON_PAGE_OFFLINE	(1 << 1)
 	__u64 value;
 
 	/** @reserved: Reserved */
Signature · 1 line
2.52.0
#31 lore raw

Re: [PATCH V20 14/15] drm/xe/uapi: Expose ban reason in EXEC_QUEUE_GET_PROPERTY_BAN

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider: - [High] Incomplete and inconsistent queue tagging for `PAGE_OFFLINE` ban reason due to mutually exclusive…

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
- [High] Incomplete and inconsistent queue tagging for PAGE_OFFLINE ban reason due to mutually exclusive list iteration in xe_ttm_vram_purge_page().
- [Medium] Missing read memory barrier (smp_rmb) in guc_exec_queue_reset_status() allows speculative loads to mask the PAGE_OFFLINE ban reason.
- [Medium] Multi-queue primary status check in guc_exec_queue_reset_status() unconditionally masks the PAGE_OFFLINE reason for secondary queues. --

commit 461fac78c1238596fc6141993bb5f0e31fd92773
Author: Tejas Upadhyay <[email protected]>
Subject: drm/xe/uapi: Expose ban reason in EXEC_QUEUE_GET_PROPERTY_BAN

This commit extends the queue ban property UAPI to return a bitmask of ban reasons rather than a boolean. It introduces GPU_HANG and PAGE_OFFLINE reasons, tracking these via a new ban_reason atomic variable in xe_exec_queue.

Quoted 2 lines
drivers/gpu/drm/xe/xe_guc_submit.c +0 -0

[ ... ]

Quoted 11 lines
drivers/gpu/drm/xe/xe_guc_submit.c +2 -4
@@ -2580,13 +2590,29 @@ static void guc_exec_queue_multi_queue_drop_suspend(struct xe_exec_queue *q)
 	}
 }
 
-static bool guc_exec_queue_reset_status(struct xe_exec_queue *q)
+static u64 guc_exec_queue_reset_status(struct xe_exec_queue *q)
 {
-	if (xe_exec_queue_is_multi_queue_secondary(q) &&
-	    guc_exec_queue_reset_status(xe_exec_queue_multi_queue_primary(q)))
-		return true;
+	/* TODO: In case of multiqueue, if a secondary queue is banned due to

drivers/gpu/drm/xe/xe_guc_submit.c static void guc_exec_queue_multi_queue_drop_suspend(struct xe_exec_queue *q) · line 2604

Quoted 10 lines
drivers/gpu/drm/xe/xe_guc_submit.c +8 -1
+	 * page offlining, checking only the primary queue's GuC reset status
+	 * may mask the true reason or race with it.
+	 */
+	if (xe_exec_queue_is_multi_queue_secondary(q)) {
+		u64 status = guc_exec_queue_reset_status(xe_exec_queue_multi_queue_primary(q));
 
-	return exec_queue_reset(q) || exec_queue_killed_or_banned_or_wedged(q);
+		if (status)
+			return status;
+	}

[Severity: Medium]
When querying a secondary queue that was individually purged and killed due to page offlining, won't this check unconditionally return the primary queue's default status (GPU_HANG)? Because the primary queue wasn't explicitly tagged by the purge, the secondary queue's PAGE_OFFLINE reason gets masked.

drivers/gpu/drm/xe/xe_guc_submit.c static void guc_exec_queue_multi_queue_drop_suspend(struct xe_exec_queue *q) · line 2611

Quoted 7 lines
drivers/gpu/drm/xe/xe_guc_submit.c +7 -0
+
+	if (exec_queue_reset(q) || exec_queue_killed_or_banned_or_wedged(q)) {
+		u64 reason = atomic_read_acquire(&q->ban_reason);
+
+		/* If no specific reason was recorded, default to GPU hang */
+		if (!reason)
+			reason = DRM_XE_EXEC_QUEUE_BAN_REASON_GPU_HANG;

[Severity: Medium]
Is a read memory barrier (smp_rmb) missing here?

In xe_ttm_vram_purge_page() located in drivers/gpu/drm/xe/xe_ttm_vram_mgr.c, ban_reason is updated and then smp_wmb() is called before killing the VM or queue. Concurrently, guc_exec_queue_reset_status() reads the state with a relaxed load in exec_queue_killed_or_banned_or_wedged() and conditionally reads ban_reason with an acquire load.

Without an smp_rmb() between the state load and the ban_reason load, the CPU could speculatively load the old ban_reason (0) before observing the killed state, resulting in userspace receiving a false GPU_HANG.

Quoted 8 lines
drivers/gpu/drm/xe/xe_guc_submit.c +4 -0
+		return reason;
+	}
+
+	return 0;
 }
 
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0

[ ... ]

Quoted 17 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +11 -0
@@ -637,7 +639,30 @@ static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo)
 	xe_bo_unlock(bo);
 	/*  Ban VM if BO is PPGTT */
 	if (vm && (flags & XE_BO_FLAG_PAGETABLE)) {
+		struct xe_exec_queue *eq;
+		int id;
 
 		down_write(&vm->lock);
+		if (xe->info.has_ctx_tlb_inval) {
+			/*
+			 * Must be the write lock: send_tlb_inval_ctx_ppgtt()
+			 * mutates this list (list_move_tail() onto an on-stack
+			 * head) while holding only the read lock, relying on
+			 * tlb_inval->seqno_lock to keep itself the sole
+			 * mutator. Traversing it under down_read() would let
+			 * this walk follow entries onto that stack list.
+			 */

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo) · line 664

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+			down_write(&vm->exec_queues.lock);
+			for (id = 0; id < ARRAY_SIZE(vm->exec_queues.list); id++)
+				list_for_each_entry(eq, &vm->exec_queues.list[id],
+						    vm_exec_queue_link)
+					atomic_or(q_flag, &eq->ban_reason);
+			up_write(&vm->exec_queues.lock);
+		} else {
+			list_for_each_entry(eq, &vm->preempt.exec_queues, lr.link)
+				atomic_or(q_flag, &eq->ban_reason);
+		}

[Severity: High]
Does this mutually exclusive check cause PAGE_OFFLINE to be missed for certain queues?

The lists vm->exec_queues.list and vm->preempt.exec_queues contain disjoint sets of queues. VM_BIND queues are explicitly excluded from vm->exec_queues.list, and standard short-running queues are not in vm->preempt.exec_queues.

By iterating only one list based on the unrelated has_ctx_tlb_inval hardware feature, the driver selectively tags only a subset of queues, leaving the remaining offlined queues to falsely report a GPU_HANG.

Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -0
+		smp_wmb(); /* Force all queue bits to be visible before killing the VM */
 		xe_vm_kill(vm, true);
 		up_write(&vm->lock);
 	}
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=14
maintainer
#32 lore raw

Re: [PATCH V20 14/15] drm/xe/uapi: Expose ban reason in EXEC_QUEUE_GET_PROPERTY_BAN

Quoted 107 lines · On Wed, Sep 02, 2026 at 05:10:59PM +0000, [email protected] wrote:

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
- [High] Incomplete and inconsistent queue tagging for PAGE_OFFLINE ban reason due to mutually exclusive list iteration in xe_ttm_vram_purge_page().
- [Medium] Missing read memory barrier (smp_rmb) in guc_exec_queue_reset_status() allows speculative loads to mask the PAGE_OFFLINE ban reason.
- [Medium] Multi-queue primary status check in guc_exec_queue_reset_status() unconditionally masks the PAGE_OFFLINE reason for secondary queues. --

commit 461fac78c1238596fc6141993bb5f0e31fd92773
Author: Tejas Upadhyay <[email protected]>
Subject: drm/xe/uapi: Expose ban reason in EXEC_QUEUE_GET_PROPERTY_BAN

This commit extends the queue ban property UAPI to return a bitmask of ban reasons rather than a boolean. It introduces GPU_HANG and PAGE_OFFLINE reasons, tracking these via a new ban_reason atomic variable in xe_exec_queue.

Quoted 2 lines
drivers/gpu/drm/xe/xe_guc_submit.c +0 -0

[ ... ]

Quoted 11 lines
drivers/gpu/drm/xe/xe_guc_submit.c +2 -4
@@ -2580,13 +2590,29 @@ static void guc_exec_queue_multi_queue_drop_suspend(struct xe_exec_queue *q)
 	}
 }
 
-static bool guc_exec_queue_reset_status(struct xe_exec_queue *q)
+static u64 guc_exec_queue_reset_status(struct xe_exec_queue *q)
 {
-	if (xe_exec_queue_is_multi_queue_secondary(q) &&
-	    guc_exec_queue_reset_status(xe_exec_queue_multi_queue_primary(q)))
-		return true;
+	/* TODO: In case of multiqueue, if a secondary queue is banned due to

drivers/gpu/drm/xe/xe_guc_submit.c static void guc_exec_queue_multi_queue_drop_suspend(struct xe_exec_queue *q) · line 2604

Quoted 10 lines
drivers/gpu/drm/xe/xe_guc_submit.c +8 -1
+	 * page offlining, checking only the primary queue's GuC reset status
+	 * may mask the true reason or race with it.
+	 */
+	if (xe_exec_queue_is_multi_queue_secondary(q)) {
+		u64 status = guc_exec_queue_reset_status(xe_exec_queue_multi_queue_primary(q));
 
-	return exec_queue_reset(q) || exec_queue_killed_or_banned_or_wedged(q);
+		if (status)
+			return status;
+	}

[Severity: Medium]
When querying a secondary queue that was individually purged and killed due to page offlining, won't this check unconditionally return the primary queue's default status (GPU_HANG)? Because the primary queue wasn't explicitly tagged by the purge, the secondary queue's PAGE_OFFLINE reason gets masked.

drivers/gpu/drm/xe/xe_guc_submit.c static void guc_exec_queue_multi_queue_drop_suspend(struct xe_exec_queue *q) · line 2611

Quoted 7 lines
drivers/gpu/drm/xe/xe_guc_submit.c +7 -0
+
+	if (exec_queue_reset(q) || exec_queue_killed_or_banned_or_wedged(q)) {
+		u64 reason = atomic_read_acquire(&q->ban_reason);
+
+		/* If no specific reason was recorded, default to GPU hang */
+		if (!reason)
+			reason = DRM_XE_EXEC_QUEUE_BAN_REASON_GPU_HANG;

[Severity: Medium]
Is a read memory barrier (smp_rmb) missing here?

In xe_ttm_vram_purge_page() located in drivers/gpu/drm/xe/xe_ttm_vram_mgr.c, ban_reason is updated and then smp_wmb() is called before killing the VM or queue. Concurrently, guc_exec_queue_reset_status() reads the state with a relaxed load in exec_queue_killed_or_banned_or_wedged() and conditionally reads ban_reason with an acquire load.

Without an smp_rmb() between the state load and the ban_reason load, the CPU could speculatively load the old ban_reason (0) before observing the killed state, resulting in userspace receiving a false GPU_HANG.

Quoted 8 lines
drivers/gpu/drm/xe/xe_guc_submit.c +4 -0
+		return reason;
+	}
+
+	return 0;
 }
 
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +0 -0

[ ... ]

Quoted 17 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +11 -0
@@ -637,7 +639,30 @@ static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo)
 	xe_bo_unlock(bo);
 	/*  Ban VM if BO is PPGTT */
 	if (vm && (flags & XE_BO_FLAG_PAGETABLE)) {
+		struct xe_exec_queue *eq;
+		int id;
 
 		down_write(&vm->lock);
+		if (xe->info.has_ctx_tlb_inval) {
+			/*
+			 * Must be the write lock: send_tlb_inval_ctx_ppgtt()
+			 * mutates this list (list_move_tail() onto an on-stack
+			 * head) while holding only the read lock, relying on
+			 * tlb_inval->seqno_lock to keep itself the sole
+			 * mutator. Traversing it under down_read() would let
+			 * this walk follow entries onto that stack list.
+			 */

drivers/gpu/drm/xe/xe_ttm_vram_mgr.c static int xe_ttm_vram_purge_page(struct xe_device *xe, struct xe_bo *bo) · line 664

Quoted 10 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +10 -0
+			down_write(&vm->exec_queues.lock);
+			for (id = 0; id < ARRAY_SIZE(vm->exec_queues.list); id++)
+				list_for_each_entry(eq, &vm->exec_queues.list[id],
+						    vm_exec_queue_link)
+					atomic_or(q_flag, &eq->ban_reason);
+			up_write(&vm->exec_queues.lock);
+		} else {
+			list_for_each_entry(eq, &vm->preempt.exec_queues, lr.link)
+				atomic_or(q_flag, &eq->ban_reason);
+		}

[Severity: High]
Does this mutually exclusive check cause PAGE_OFFLINE to be missed for certain queues?

Quoted 8 lines

The lists vm->exec_queues.list and vm->preempt.exec_queues contain disjoint sets of queues. VM_BIND queues are explicitly excluded from vm->exec_queues.list, and standard short-running queues are not in vm->preempt.exec_queues.

By iterating only one list based on the unrelated has_ctx_tlb_inval hardware feature, the driver selectively tags only a subset of queues, leaving the remaining offlined queues to falsely report a GPU_HANG.

This is right - we should probably just use exec_queues.list. I have various code floating around which use exec_queues.list regardless if info.has_ctx_tlb_inval is enabled. We should probably extract that into a standalone patch and then fixup this code. Another thing we mostly likely want to do is kill all exec queues too.

Both can be done in a follow up. Can we get a Jira filed for this?

Matt

Quoted 7 lines
Quoted 4 lines
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +1 -0
+		smp_wmb(); /* Force all queue bits to be visible before killing the VM */
 		xe_vm_kill(vm, true);
 		up_write(&vm->lock);
 	}
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=14
#33 lore raw

PATCH v20 15/15 drm/xe: Add fault-inject based VRAM page offline injection

In reply to Tejas Upadhyay

Add a fault-inject based debugfs interface for testing VRAM page offlining. This replaces the previous standalone debugfs approach with the standard kernel fault-inject infrastructure.

Two debugfs entries are created under the xe debugfs root for CRI platforms:
- inject_mempage_offline/: Standard fault-inject knobs (probability, times, interval, etc.) created by fault_create_debugfs_attr(). Without CONFIG_FAULT_INJECTION_DEBUG_FS, the stub returns ERR_PTR(-ENODEV) and no knobs are created, making the trigger effectively a no-op.
- inject_mempage_offline_trigger: Write a PFN value to inject a specific page, or write "0" to auto-pick the last unallocated VRAM page

The trigger accepts:
- "0" : auto-pick last unallocated page
- "0xPFN" : inject fault at a specific PFN address

Usage:
echo 100 > inject_mempage_offline/probability
echo 1 > inject_mempage_offline/times
echo 0 > inject_mempage_offline_trigger

probability: likelihood of should_fail() returning true (0-100)
times: number of times injection is allowed (-1 for unlimited)

v6(Himal):
- Add warning to rebind driver post test run v5(Sashiko):
- exclude SRIOV and remove dpa_base addition, already absolute dpa v4(Himal):
- Use xe_fault_mempage_offline() instead of IS_ENABLED() + direct should_fail(). CONFIG_FAULT_INJECTION_DEBUG_FS is now an implicit requirement for the trigger to function.
v3(Himal):
- Use FAULT_ACTION
v2(sashiko):
- use cond_resched()
- validate input first and fix addr < 0 case
- validate vr, move block, found var as local to scope_guard

---
 drivers/gpu/drm/xe/xe_debugfs.c      | 48 +++++++++++++++++++++++
 drivers/gpu/drm/xe/xe_debugfs.h      |  2 +
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 58 ++++++++++++++++++++++++++++
 drivers/gpu/drm/xe/xe_ttm_vram_mgr.h |  1 +
 4 files changed, 109 insertions(+)
drivers/gpu/drm/xe/xe_debugfs.c +48 -0
index cb24e001b142..80f62634fae5 100644
@@ -45,12 +45,18 @@
 DECLARE_FAULT_ATTR(gt_reset_failure);
 DECLARE_FAULT_ATTR(inject_csc_hw_error);
 DECLARE_FAULT_ATTR(wedge_cold_reset);
+DECLARE_FAULT_ATTR(inject_mempage_offline);
 
 static bool csc_hw_error_available(struct xe_device *xe)
 {
 	return !IS_SRIOV_VF(xe) && xe->info.platform == XE_BATTLEMAGE;
 }
 
+static bool is_crescent_island_pf(struct xe_device *xe)
+{
+	return !IS_SRIOV_VF(xe) && xe->info.platform == XE_CRESCENTISLAND;
+}
+
 /*
  * Fault injection table.  Each entry registers a debugfs attribute; add a
  * matching FAULT_ACTION() below for every entry added here.
@@ -67,6 +73,9 @@ static struct {
 	  .is_visible = csc_hw_error_available },
 	{ .name = "wedge_cold_reset",
 	  .attr = &wedge_cold_reset },
+	{ .name = "inject_mempage_offline",
+	  .attr = &inject_mempage_offline,
+	  .is_visible = is_crescent_island_pf },
 };
 
 /*
@@ -82,6 +91,40 @@ bool xe_fault_##name(void)				\
 FAULT_ACTION(gt_reset, gt_reset_failure)
 FAULT_ACTION(csc_hw_error, inject_csc_hw_error)
 FAULT_ACTION(wedge_cold_reset, wedge_cold_reset)
+FAULT_ACTION(mempage_offline, inject_mempage_offline)
+
+static ssize_t inject_mempage_offline_trigger(struct file *f,
+					      const char __user *ubuf,
+					      size_t size, loff_t *pos)
+{
+	struct xe_device *xe = file_inode(f)->i_private;
+	struct xe_tile *tile = xe_device_get_root_tile(xe);
+	struct xe_vram_region *vr = tile->mem.vram;
+	u64 pfn;
+	int ret;
+
+	if (!vr)
+		return -ENODEV;
+
+	ret = kstrtou64_from_user(ubuf, size, 0, &pfn);
+	if (ret)
+		return ret;
+
+	if (!xe_fault_mempage_offline())
+		return size;
+
+	xe_warn(xe, "Page offlining test interface accessed. Notice: Offlined or reserved memory pages cannot be reclaimed dynamically. A driver rebind (unbind and bind loop) is required post-test to clean up.\n");
+	if (pfn == 0)
+		return xe_ttm_vram_inject_fault(xe) ?: size;
+
+	/* User provided PFN - convert to DPA and inject */
+	return xe_ttm_vram_handle_addr_fault(xe, pfn << PAGE_SHIFT) ?: size;
+}
+
+static const struct file_operations inject_mempage_offline_fops = {
+	.owner = THIS_MODULE,
+	.write = inject_mempage_offline_trigger,
+};
 
 static void xe_fault_inject_debugfs_register(struct xe_device *xe,
 					     struct dentry *root)
@@ -96,6 +139,11 @@ static void xe_fault_inject_debugfs_register(struct xe_device *xe,
 		fault_create_debugfs_attr(xe_fault_inject_entry[i].name, root,
 					  xe_fault_inject_entry[i].attr);
 	}
+
+	if (is_crescent_island_pf(xe)) {
+		debugfs_create_file("inject_mempage_offline_trigger", 0200,
+				    root, xe, &inject_mempage_offline_fops);
+	}
 }
 
 static void read_residency_counter(struct xe_device *xe, struct xe_mmio *mmio,
drivers/gpu/drm/xe/xe_debugfs.h +2 -0
index 0dcd28fd7dc0..88d91c78036b 100644
@@ -14,11 +14,13 @@ struct xe_device;
 bool xe_fault_gt_reset(void);
 bool xe_fault_csc_hw_error(void);
 bool xe_fault_wedge_cold_reset(void);
+bool xe_fault_mempage_offline(void);
 void xe_debugfs_register(struct xe_device *xe);
 #else
 static inline bool xe_fault_gt_reset(void) { return false; }
 static inline bool xe_fault_csc_hw_error(void) { return false; }
 static inline bool xe_fault_wedge_cold_reset(void) { return false; }
+static inline bool xe_fault_mempage_offline(void) { return false; }
 static inline void xe_debugfs_register(struct xe_device *xe) { }
 #endif
 
drivers/gpu/drm/xe/xe_ttm_vram_mgr.c +58 -0

64 lines not on this page · raw message

drivers/gpu/drm/xe/xe_ttm_vram_mgr.h +1 -0
index f354c26c4257..8878e36292b2 100644
@@ -33,6 +33,7 @@ void xe_ttm_vram_get_used(struct ttm_resource_manager *man,
 			  u64 *used, u64 *used_visible);
 
 int xe_ttm_vram_handle_addr_fault(struct xe_device *xe, u64 addr);
+int xe_ttm_vram_inject_fault(struct xe_device *xe);
 void xe_ttm_vram_debugfs_init(struct xe_device *xe, struct dentry *root);
 static inline struct xe_ttm_vram_mgr_resource *
 to_xe_ttm_vram_mgr_resource(struct ttm_resource *res)
Signature · 1 line
2.52.0
CI Reviewed-by Himal Prasad Ghimiray on the v20 series
#34 lore raw

✗ CI.checkpatch: warning for Add memory page offlining support (rev24)

In reply to Tejas Upadhyay
== Series Details ==

== Series Details ==

Series: Add memory page offlining support (rev24)
URL : https://patchwork.freedesktop.org/series/161473/
State : warning

== Summary ==

+ KERNEL=/kernel
+ git clone https://gitlab.freedesktop.org/drm/maintainer-tools mt Cloning into 'mt'...
warning: redirecting to https://gitlab.freedesktop.org/drm/maintainer-tools.git/
+ git -C mt rev-list -n1 origin/master d875049d2b299159a272bd5151994970cdcd1e31
+ cd /kernel
+ git config --global --add safe.directory /kernel
+ git log -n1
commit ee8bd1bfff88c1a26b8f98014bbd82c47ecb2375

Author: Tejas Upadhyay <[email protected]>
Date:   Wed Sep 2 20:23:58 2026 +0530

drm/xe: Add fault-inject based VRAM page offline injection

Add a fault-inject based debugfs interface for testing VRAM page offlining. This replaces the previous standalone debugfs approach with the standard kernel fault-inject infrastructure.

Two debugfs entries are created under the xe debugfs root for CRI platforms:
- inject_mempage_offline/: Standard fault-inject knobs (probability, times, interval, etc.) created by fault_create_debugfs_attr(). Without CONFIG_FAULT_INJECTION_DEBUG_FS, the stub returns ERR_PTR(-ENODEV) and no knobs are created, making the trigger effectively a no-op.
- inject_mempage_offline_trigger: Write a PFN value to inject a specific page, or write "0" to auto-pick the last unallocated VRAM page

The trigger accepts:
- "0" : auto-pick last unallocated page
- "0xPFN" : inject fault at a specific PFN address

Usage:
echo 100 > inject_mempage_offline/probability echo 1 > inject_mempage_offline/times echo 0 > inject_mempage_offline_trigger

probability: likelihood of should_fail() returning true (0-100)
times: number of times injection is allowed (-1 for unlimited)

v6(Himal):
- Add warning to rebind driver post test run v5(Sashiko):
- exclude SRIOV and remove dpa_base addition, already absolute dpa v4(Himal):
- Use xe_fault_mempage_offline() instead of IS_ENABLED() + direct should_fail(). CONFIG_FAULT_INJECTION_DEBUG_FS is now an implicit requirement for the trigger to function.
v3(Himal):
- Use FAULT_ACTION
v2(sashiko):
- use cond_resched()
- validate input first and fix addr < 0 case
- validate vr, move block, found var as local to scope_guard

+ /mt/dim checkpatch 1ddb465787c65331f609fbe02998ce1643b041c7 drm-intel
dff3e0e16c13 drm/xe: Link VRAM object with gpu buddy
cd41da48d7b0 drm/xe: Link LRC BO and its execution queue with safe lifetime rules
-:7: WARNING:COMMIT_LOG_LONG_LINE: Prefer a maximum 75 chars per line (possible unwrapped commit description?) #7:
Introduce an execution queue back-pointer (q) within struct xe_bo, primarily

-:45: WARNING:BAD_SIGN_OFF: Non-standard signature: Co-authored-by:
#45:
Co-authored-by: Copilot

-:45: ERROR:BAD_SIGN_OFF: Unrecognized email address: 'Copilot' #45:
Co-authored-by: Copilot

total: 1 errors, 2 warnings, 0 checks, 100 lines checked 6cefc976e143 drm/xe: Export xe_ttm_bo_purge()
f6c1071d5cee drm/xe: Handle NULL resource and allow purging of VRAM pages 182cce6f4c1f drm/xe/bo: Make xe_bo_is_user() public a8d067c4d6e0 drm/xe: Guard teardown paths against purged BOs 8f351bbcecbf drm/xe/vram: Extract buddy allocation and free helpers 5bd9540dc81e drm/xe/vram: Add page offline data structures and lifecycle 8be2e05c08eb drm/xe/vram: Add VRAM page offline fault handler -:49: WARNING:BAD_SIGN_OFF: Non-standard signature: Co-authored-by:
#49:
Co-authored-by: Copilot

-:49: ERROR:BAD_SIGN_OFF: Unrecognized email address: 'Copilot' #49:
Co-authored-by: Copilot

total: 1 errors, 1 warnings, 0 checks, 362 lines checked 7aad507c44de drm/xe/configfs: Add disable_vram_page_offline attribute 45d735196c5d drm/xe/ras: Cache disable_vram_page_offline policy at init 5beddc6e63e3 drm/xe/vram: Check disable_vram_page_offline policy in fault handler d19a8863a078 drm/xe: Expose bad VRAM pages via debugfs
2379b91471c9 drm/xe/uapi: Expose ban reason in EXEC_QUEUE_GET_PROPERTY_BAN ee8bd1bfff88 drm/xe: Add fault-inject based VRAM page offline injection

#35 lore raw

✓ CI.KUnit: success for Add memory page offlining support (rev24)

In reply to Tejas Upadhyay
== Series Details ==

== Series Details ==

Series: Add memory page offlining support (rev24)
URL : https://patchwork.freedesktop.org/series/161473/
State : success

== Summary ==

+ trap cleanup EXIT
+ kunitconfigs=('/kernel/drivers/gpu/tests/.kunitconfig' '/kernel/drivers/gpu/drm/xe/.kunitconfig' '/kernel/drivers/gpu/drm/tests/.kunitconfig' '/kernel/drivers/gpu/drm/ttm/tests/.kunitconfig' '/kernel/drivers/dma-buf/.kunitconfig')
+ for kcfg in "${kunitconfigs[@]}"
+ [[ ! -f /kernel/drivers/gpu/tests/.kunitconfig ]]
+ /kernel/tools/testing/kunit/kunit.py run --kunitconfig /kernel/drivers/gpu/tests/.kunitconfig
[17:02:43] Configuring KUnit Kernel ...
Generating .config ...
Populating config with:

$ make ARCH=um O=.kunit olddefconfig
[17:02:48] Building KUnit Kernel ...
Populating config with:
$ make ARCH=um O=.kunit olddefconfig
Building with:
$ make all compile_commands.json scripts_gdb ARCH=um O=.kunit --jobs=48
[17:03:08] Starting KUnit Kernel (1/1)...
[17:03:08] ============================================================
Running tests with:
$ .kunit/linux kunit.enable=1 mem=1G console=tty kunit_shutdown=halt
[17:03:08] ============= refcount_interrupt (4 subtests) ==============
[17:03:08] [PASSED] test_single_irq_change
[17:03:08] [PASSED] test_nested_irq_change
[17:03:08] [PASSED] test_multiple_irq_change
[17:03:08] [PASSED] test_irq_save
[17:03:08] =============== [PASSED] refcount_interrupt ================
[17:03:08] ================= gpu_buddy (14 subtests) ==================
[17:03:08] [PASSED] gpu_test_buddy_alloc_limit
[17:03:08] [PASSED] gpu_test_buddy_alloc_optimistic
[17:03:08] [PASSED] gpu_test_buddy_alloc_pessimistic
[17:03:08] [PASSED] gpu_test_buddy_alloc_pathological
[17:03:08] [PASSED] gpu_test_buddy_alloc_contiguous
[17:03:08] [PASSED] gpu_test_buddy_alloc_clear
[17:03:08] [PASSED] gpu_test_buddy_alloc_range
[17:03:08] [PASSED] gpu_test_buddy_alloc_range_bias
[17:03:09] [PASSED] gpu_test_buddy_fragmentation_performance
[17:03:10] [PASSED] gpu_test_buddy_dirty_tracker_performance
[17:03:10] [PASSED] gpu_test_buddy_alloc_exceeds_max_order
[17:03:10] [PASSED] gpu_test_buddy_offset_aligned_allocation
[17:03:10] [PASSED] gpu_test_buddy_subtree_offset_alignment_stress
[17:03:10] [PASSED] gpu_test_buddy_addr_to_block
[17:03:10] ==================== [PASSED] gpu_buddy ====================
[17:03:10] ============================================================
[17:03:10] Testing complete. Ran 18 tests: passed: 18
[17:03:10] Elapsed time: 26.773s total, 4.351s configuring, 20.504s building, 1.886s running

+ for kcfg in "${kunitconfigs[@]}"
+ [[ ! -f /kernel/drivers/gpu/drm/xe/.kunitconfig ]]
+ /kernel/tools/testing/kunit/kunit.py run --kunitconfig /kernel/drivers/gpu/drm/xe/.kunitconfig [17:03:10] Configuring KUnit Kernel ...
Regenerating .config ...
Populating config with:

$ make ARCH=um O=.kunit olddefconfig
[17:03:12] Building KUnit Kernel ...
Populating config with:
$ make ARCH=um O=.kunit olddefconfig
Building with:
$ make all compile_commands.json scripts_gdb ARCH=um O=.kunit --jobs=48
[17:03:46] Starting KUnit Kernel (1/1)...
[17:03:46] ============================================================
Running tests with:
$ .kunit/linux kunit.enable=1 mem=1G console=tty kunit_shutdown=halt
[17:03:46] ============= refcount_interrupt (4 subtests) ==============
[17:03:46] [PASSED] test_single_irq_change
[17:03:46] [PASSED] test_nested_irq_change
[17:03:46] [PASSED] test_multiple_irq_change
[17:03:46] [PASSED] test_irq_save
[17:03:46] =============== [PASSED] refcount_interrupt ================
[17:03:46] ================== guc_buf (11 subtests) ===================
[17:03:46] [PASSED] test_smallest
[17:03:46] [PASSED] test_largest
[17:03:46] [PASSED] test_granular
[17:03:46] [PASSED] test_unique
[17:03:46] [PASSED] test_overlap
[17:03:46] [PASSED] test_reusable
[17:03:46] [PASSED] test_too_big
[17:03:46] [PASSED] test_flush
[17:03:46] [PASSED] test_lookup
[17:03:46] [PASSED] test_data
[17:03:46] [PASSED] test_class
[17:03:46] ===================== [PASSED] guc_buf =====================
[17:03:46] =================== guc_dbm (7 subtests) ===================
[17:03:46] [PASSED] test_empty
[17:03:46] [PASSED] test_default
[17:03:46] ======================== test_size  ========================
[17:03:46] [PASSED] 4
[17:03:46] [PASSED] 8
[17:03:46] [PASSED] 32
[17:03:46] [PASSED] 256
[17:03:46] ==================== [PASSED] test_size ====================
[17:03:46] ======================= test_reuse  ========================
[17:03:46] [PASSED] 4
[17:03:46] [PASSED] 8
[17:03:46] [PASSED] 32
[17:03:46] [PASSED] 256
[17:03:46] =================== [PASSED] test_reuse ====================
[17:03:46] =================== test_range_overlap  ====================
[17:03:46] [PASSED] 4
[17:03:46] [PASSED] 8
[17:03:46] [PASSED] 32
[17:03:46] [PASSED] 256
[17:03:46] =============== [PASSED] test_range_overlap ================
[17:03:46] =================== test_range_compact  ====================
[17:03:46] [PASSED] 4
[17:03:46] [PASSED] 8
[17:03:46] [PASSED] 32
[17:03:46] [PASSED] 256
[17:03:46] =============== [PASSED] test_range_compact ================
[17:03:46] ==================== test_range_spare  =====================
[17:03:46] [PASSED] 4
[17:03:46] [PASSED] 8
[17:03:46] [PASSED] 32
[17:03:46] [PASSED] 256
[17:03:46] ================ [PASSED] test_range_spare =================
[17:03:46] ===================== [PASSED] guc_dbm =====================
[17:03:46] =================== guc_idm (6 subtests) ===================
[17:03:46] [PASSED] bad_init
[17:03:46] [PASSED] no_init
[17:03:46] [PASSED] init_fini
[17:03:46] [PASSED] check_used
[17:03:46] [PASSED] check_quota
[17:03:46] [PASSED] check_all
[17:03:46] ===================== [PASSED] guc_idm =====================
[17:03:46] =============== guc_klv_helpers (9 subtests) ===============
[17:03:46] [PASSED] test_count
[17:03:46] [PASSED] test_encode_u32
[17:03:46] [PASSED] test_encode_u64
[17:03:46] [PASSED] test_encode_string
[17:03:46] [PASSED] test_encode_object_raw
[17:03:46] [PASSED] test_encode_object_klv
[17:03:46] [PASSED] test_encode_object_nested
[17:03:46] [PASSED] test_encode_object_basic
[17:03:46] [PASSED] test_print
[17:03:46] ================= [PASSED] guc_klv_helpers =================
[17:03:46] =================== xe_log (4 subtests) ====================
[17:03:46] [PASSED] demo_cper
[17:03:46] [PASSED] demo_dmesg
[17:03:46] ======================= test_dmesg  ========================
[17:03:46] [PASSED] test_fatal
[17:03:46] [PASSED] test_fatal_tile
[17:03:46] [PASSED] test_fatal_gt
[17:03:46] [PASSED] test_fatal_comp
[17:03:46] [PASSED] test_fatal_comp_tile
[17:03:46] [PASSED] test_fatal_comp_gt
[17:03:46] [PASSED] test_fatal_all
[17:03:46] [PASSED] test_recoverable
[17:03:46] [PASSED] test_recoverable_tile
[17:03:46] [PASSED] test_recoverable_gt
[17:03:46] [PASSED] test_recoverable_comp
[17:03:46] [PASSED] test_recoverable_comp_tile
[17:03:46] [PASSED] test_recoverable_comp_gt
[17:03:46] [PASSED] test_recoverable_all
[17:03:46] [PASSED] test_info
[17:03:46] [PASSED] test_info_tile
[17:03:46] [PASSED] test_info_gt
[17:03:46] [PASSED] test_info_err
[17:03:46] [PASSED] test_info_comp
[17:03:46] [PASSED] test_info_comp_tile
[17:03:46] [PASSED] test_info_comp_gt
[17:03:46] [PASSED] test_info_all
[17:03:46] [PASSED] test_hw_fatal
[17:03:46] [PASSED] test_hw_recoverable
[17:03:46] [PASSED] test_hw_corrected
[17:03:46] [PASSED] test_hw_informational
[17:03:46] =================== [PASSED] test_dmesg ====================
[17:03:46] ====================== test_invalid  =======================
[17:03:46] [SKIPPED] no-component no-location no-warn (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] [SKIPPED] reserved location (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] [SKIPPED] unknown location (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] [SKIPPED] nonzero-device-id location (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] [SKIPPED] invalid-tile-id location (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] [SKIPPED] invalid-gt-id location (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] [SKIPPED] unknown component class (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] [SKIPPED] unknown system component (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] [SKIPPED] unknown hardware component (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] [SKIPPED] unknown component and location (requires CONFIG_DRM_XE_DEBUG)
[17:03:46] ================== [SKIPPED] test_invalid ==================
[17:03:46] ===================== [PASSED] xe_log ======================
[17:03:46] ================== no_relay (3 subtests) ===================
[17:03:46] [PASSED] xe_drops_guc2pf_if_not_ready
[17:03:46] [PASSED] xe_drops_guc2vf_if_not_ready
[17:03:46] [PASSED] xe_rejects_send_if_not_ready
[17:03:46] ==================== [PASSED] no_relay =====================
[17:03:46] ================== pf_relay (14 subtests) ==================
[17:03:46] [PASSED] pf_rejects_guc2pf_too_short
[17:03:46] [PASSED] pf_rejects_guc2pf_too_long
[17:03:46] [PASSED] pf_rejects_guc2pf_no_payload
[17:03:46] [PASSED] pf_fails_no_payload
[17:03:46] [PASSED] pf_fails_bad_origin
[17:03:46] [PASSED] pf_fails_bad_type
[17:03:46] [PASSED] pf_txn_reports_error
[17:03:46] [PASSED] pf_txn_sends_pf2guc
[17:03:46] [PASSED] pf_sends_pf2guc
[17:03:46] [SKIPPED] pf_loopback_nop (requires CONFIG_DRM_XE_DEBUG_SRIOV)
[17:03:46] [SKIPPED] pf_loopback_echo (requires CONFIG_DRM_XE_DEBUG_SRIOV)
[17:03:46] [SKIPPED] pf_loopback_fail (requires CONFIG_DRM_XE_DEBUG_SRIOV)
[17:03:46] [SKIPPED] pf_loopback_busy (requires CONFIG_DRM_XE_DEBUG_SRIOV)
[17:03:46] [SKIPPED] pf_loopback_retry (requires CONFIG_DRM_XE_DEBUG_SRIOV)
[17:03:46] ==================== [PASSED] pf_relay =====================
[17:03:46] ================== vf_relay (3 subtests) ===================
[17:03:46] [PASSED] vf_rejects_guc2vf_too_short
[17:03:46] [PASSED] vf_rejects_guc2vf_too_long
[17:03:46] [PASSED] vf_rejects_guc2vf_no_payload
[17:03:46] ==================== [PASSED] vf_relay =====================
[17:03:46] ================ pf_gt_config (9 subtests) =================
[17:03:46] [PASSED] fair_contexts_1vf
[17:03:46] [PASSED] fair_doorbells_1vf
[17:03:46] [PASSED] fair_ggtt_1vf
[17:03:46] ====================== fair_vram_1vf  ======================
[17:03:46] [PASSED] 3.50 GiB
[17:03:46] [PASSED] 11.5 GiB
[17:03:46] [PASSED] 15.5 GiB
[17:03:46] [PASSED] 31.5 GiB
[17:03:46] [PASSED] 63.5 GiB
[17:03:46] [PASSED] 1.91 GiB
[17:03:46] ================== [PASSED] fair_vram_1vf ==================
[17:03:46] ================ fair_vram_1vf_admin_only  =================
[17:03:46] [PASSED] 3.50 GiB
[17:03:46] [PASSED] 11.5 GiB
[17:03:46] [PASSED] 15.5 GiB
[17:03:46] [PASSED] 31.5 GiB
[17:03:46] [PASSED] 63.5 GiB
[17:03:46] [PASSED] 1.91 GiB
[17:03:46] ============ [PASSED] fair_vram_1vf_admin_only =============
[17:03:46] ====================== fair_contexts  ======================
[17:03:46] [PASSED] 1 VF
[17:03:46] [PASSED] 2 VFs
[17:03:46] [PASSED] 3 VFs
[17:03:46] [PASSED] 4 VFs
[17:03:46] [PASSED] 5 VFs
[17:03:46] [PASSED] 6 VFs
[17:03:46] [PASSED] 7 VFs
[17:03:46] [PASSED] 8 VFs
[17:03:46] [PASSED] 9 VFs
[17:03:46] [PASSED] 10 VFs
[17:03:46] [PASSED] 11 VFs
[17:03:46] [PASSED] 12 VFs
[17:03:46] [PASSED] 13 VFs
[17:03:46] [PASSED] 14 VFs
[17:03:46] [PASSED] 15 VFs
[17:03:46] [PASSED] 16 VFs
[17:03:46] [PASSED] 17 VFs
[17:03:46] [PASSED] 18 VFs
[17:03:46] [PASSED] 19 VFs
[17:03:46] [PASSED] 20 VFs
[17:03:46] [PASSED] 21 VFs
[17:03:46] [PASSED] 22 VFs
[17:03:46] [PASSED] 23 VFs
[17:03:46] [PASSED] 24 VFs
[17:03:46] [PASSED] 25 VFs
[17:03:46] [PASSED] 26 VFs
[17:03:46] [PASSED] 27 VFs
[17:03:46] [PASSED] 28 VFs
[17:03:46] [PASSED] 29 VFs
[17:03:46] [PASSED] 30 VFs
[17:03:46] [PASSED] 31 VFs
[17:03:46] [PASSED] 32 VFs
[17:03:46] [PASSED] 33 VFs
[17:03:46] [PASSED] 34 VFs
[17:03:46] [PASSED] 35 VFs
[17:03:47] [PASSED] 36 VFs
[17:03:47] [PASSED] 37 VFs
[17:03:47] [PASSED] 38 VFs
[17:03:47] [PASSED] 39 VFs
[17:03:47] [PASSED] 40 VFs
[17:03:47] [PASSED] 41 VFs
[17:03:47] [PASSED] 42 VFs
[17:03:47] [PASSED] 43 VFs
[17:03:47] [PASSED] 44 VFs
[17:03:47] [PASSED] 45 VFs
[17:03:47] [PASSED] 46 VFs
[17:03:47] [PASSED] 47 VFs
[17:03:47] [PASSED] 48 VFs
[17:03:47] [PASSED] 49 VFs
[17:03:47] [PASSED] 50 VFs
[17:03:47] [PASSED] 51 VFs
[17:03:47] [PASSED] 52 VFs
[17:03:47] [PASSED] 53 VFs
[17:03:47] [PASSED] 54 VFs
[17:03:47] [PASSED] 55 VFs
[17:03:47] [PASSED] 56 VFs
[17:03:47] [PASSED] 57 VFs
[17:03:47] [PASSED] 58 VFs
[17:03:47] [PASSED] 59 VFs
[17:03:47] [PASSED] 60 VFs
[17:03:47] [PASSED] 61 VFs
[17:03:47] [PASSED] 62 VFs
[17:03:47] [PASSED] 63 VFs
[17:03:47] ================== [PASSED] fair_contexts ==================
[17:03:47] ===================== fair_doorbells  ======================
[17:03:47] [PASSED] 1 VF
[17:03:47] [PASSED] 2 VFs
[17:03:47] [PASSED] 3 VFs
[17:03:47] [PASSED] 4 VFs
[17:03:47] [PASSED] 5 VFs
[17:03:47] [PASSED] 6 VFs
[17:03:47] [PASSED] 7 VFs
[17:03:47] [PASSED] 8 VFs
[17:03:47] [PASSED] 9 VFs
[17:03:47] [PASSED] 10 VFs
[17:03:47] [PASSED] 11 VFs
[17:03:47] [PASSED] 12 VFs
[17:03:47] [PASSED] 13 VFs
[17:03:47] [PASSED] 14 VFs
[17:03:47] [PASSED] 15 VFs
[17:03:47] [PASSED] 16 VFs
[17:03:47] [PASSED] 17 VFs
[17:03:47] [PASSED] 18 VFs
[17:03:47] [PASSED] 19 VFs
[17:03:47] [PASSED] 20 VFs
[17:03:47] [PASSED] 21 VFs
[17:03:47] [PASSED] 22 VFs
[17:03:47] [PASSED] 23 VFs
[17:03:47] [PASSED] 24 VFs
[17:03:47] [PASSED] 25 VFs
[17:03:47] [PASSED] 26 VFs
[17:03:47] [PASSED] 27 VFs
[17:03:47] [PASSED] 28 VFs
[17:03:47] [PASSED] 29 VFs
[17:03:47] [PASSED] 30 VFs
[17:03:47] [PASSED] 31 VFs
[17:03:47] [PASSED] 32 VFs
[17:03:47] [PASSED] 33 VFs
[17:03:47] [PASSED] 34 VFs
[17:03:47] [PASSED] 35 VFs
[17:03:47] [PASSED] 36 VFs
[17:03:47] [PASSED] 37 VFs
[17:03:47] [PASSED] 38 VFs
[17:03:47] [PASSED] 39 VFs
[17:03:47] [PASSED] 40 VFs
[17:03:47] [PASSED] 41 VFs
[17:03:47] [PASSED] 42 VFs
[17:03:47] [PASSED] 43 VFs
[17:03:47] [PASSED] 44 VFs
[17:03:47] [PASSED] 45 VFs
[17:03:47] [PASSED] 46 VFs
[17:03:47] [PASSED] 47 VFs
[17:03:47] [PASSED] 48 VFs
[17:03:47] [PASSED] 49 VFs
[17:03:47] [PASSED] 50 VFs
[17:03:47] [PASSED] 51 VFs
[17:03:47] [PASSED] 52 VFs
[17:03:47] [PASSED] 53 VFs
[17:03:47] [PASSED] 54 VFs
[17:03:47] [PASSED] 55 VFs
[17:03:47] [PASSED] 56 VFs
[17:03:47] [PASSED] 57 VFs
[17:03:47] [PASSED] 58 VFs
[17:03:47] [PASSED] 59 VFs
[17:03:47] [PASSED] 60 VFs
[17:03:47] [PASSED] 61 VFs
[17:03:47] [PASSED] 62 VFs
[17:03:47] [PASSED] 63 VFs
[17:03:47] ================= [PASSED] fair_doorbells ==================
[17:03:47] ======================== fair_ggtt  ========================
[17:03:47] [PASSED] 1 VF
[17:03:47] [PASSED] 2 VFs
[17:03:47] [PASSED] 3 VFs
[17:03:47] [PASSED] 4 VFs
[17:03:47] [PASSED] 5 VFs
[17:03:47] [PASSED] 6 VFs
[17:03:47] [PASSED] 7 VFs
[17:03:47] [PASSED] 8 VFs
[17:03:47] [PASSED] 9 VFs
[17:03:47] [PASSED] 10 VFs
[17:03:47] [PASSED] 11 VFs
[17:03:47] [PASSED] 12 VFs
[17:03:47] [PASSED] 13 VFs
[17:03:47] [PASSED] 14 VFs
[17:03:47] [PASSED] 15 VFs
[17:03:47] [PASSED] 16 VFs
[17:03:47] [PASSED] 17 VFs
[17:03:47] [PASSED] 18 VFs
[17:03:47] [PASSED] 19 VFs
[17:03:47] [PASSED] 20 VFs
[17:03:47] [PASSED] 21 VFs
[17:03:47] [PASSED] 22 VFs
[17:03:47] [PASSED] 23 VFs
[17:03:47] [PASSED] 24 VFs
[17:03:47] [PASSED] 25 VFs
[17:03:47] [PASSED] 26 VFs
[17:03:47] [PASSED] 27 VFs
[17:03:47] [PASSED] 28 VFs
[17:03:47] [PASSED] 29 VFs
[17:03:47] [PASSED] 30 VFs
[17:03:47] [PASSED] 31 VFs
[17:03:47] [PASSED] 32 VFs
[17:03:47] [PASSED] 33 VFs
[17:03:47] [PASSED] 34 VFs
[17:03:47] [PASSED] 35 VFs
[17:03:47] [PASSED] 36 VFs
[17:03:47] [PASSED] 37 VFs
[17:03:47] [PASSED] 38 VFs
[17:03:47] [PASSED] 39 VFs
[17:03:47] [PASSED] 40 VFs
[17:03:47] [PASSED] 41 VFs
[17:03:47] [PASSED] 42 VFs
[17:03:47] [PASSED] 43 VFs
[17:03:47] [PASSED] 44 VFs
[17:03:47] [PASSED] 45 VFs
[17:03:47] [PASSED] 46 VFs
[17:03:47] [PASSED] 47 VFs
[17:03:47] [PASSED] 48 VFs
[17:03:47] [PASSED] 49 VFs
[17:03:47] [PASSED] 50 VFs
[17:03:47] [PASSED] 51 VFs
[17:03:47] [PASSED] 52 VFs
[17:03:47] [PASSED] 53 VFs
[17:03:47] [PASSED] 54 VFs
[17:03:47] [PASSED] 55 VFs
[17:03:47] [PASSED] 56 VFs
[17:03:47] [PASSED] 57 VFs
[17:03:47] [PASSED] 58 VFs
[17:03:47] [PASSED] 59 VFs
[17:03:47] [PASSED] 60 VFs
[17:03:47] [PASSED] 61 VFs
[17:03:47] [PASSED] 62 VFs
[17:03:47] [PASSED] 63 VFs
[17:03:47] ==================== [PASSED] fair_ggtt ====================
[17:03:47] ======================== fair_vram  ========================
[17:03:47] [PASSED] 1 VF
[17:03:47] [PASSED] 2 VFs
[17:03:47] [PASSED] 3 VFs
[17:03:47] [PASSED] 4 VFs
[17:03:47] [PASSED] 5 VFs
[17:03:47] [PASSED] 6 VFs
[17:03:47] [PASSED] 7 VFs
[17:03:47] [PASSED] 8 VFs
[17:03:47] [PASSED] 9 VFs
[17:03:47] [PASSED] 10 VFs
[17:03:47] [PASSED] 11 VFs
[17:03:47] [PASSED] 12 VFs
[17:03:47] [PASSED] 13 VFs
[17:03:47] [PASSED] 14 VFs
[17:03:47] [PASSED] 15 VFs
[17:03:47] [PASSED] 16 VFs
[17:03:47] [PASSED] 17 VFs
[17:03:47] [PASSED] 18 VFs
[17:03:47] [PASSED] 19 VFs
[17:03:47] [PASSED] 20 VFs
[17:03:47] [PASSED] 21 VFs
[17:03:47] [PASSED] 22 VFs
[17:03:47] [PASSED] 23 VFs
[17:03:47] [PASSED] 24 VFs
[17:03:47] [PASSED] 25 VFs
[17:03:47] [PASSED] 26 VFs
[17:03:47] [PASSED] 27 VFs
[17:03:47] [PASSED] 28 VFs
[17:03:47] [PASSED] 29 VFs
[17:03:47] [PASSED] 30 VFs
[17:03:47] [PASSED] 31 VFs
[17:03:47] [PASSED] 32 VFs
[17:03:47] [PASSED] 33 VFs
[17:03:47] [PASSED] 34 VFs
[17:03:47] [PASSED] 35 VFs
[17:03:47] [PASSED] 36 VFs
[17:03:47] [PASSED] 37 VFs
[17:03:47] [PASSED] 38 VFs
[17:03:47] [PASSED] 39 VFs
[17:03:47] [PASSED] 40 VFs
[17:03:47] [PASSED] 41 VFs
[17:03:47] [PASSED] 42 VFs
[17:03:47] [PASSED] 43 VFs
[17:03:47] [PASSED] 44 VFs
[17:03:47] [PASSED] 45 VFs
[17:03:47] [PASSED] 46 VFs
[17:03:47] [PASSED] 47 VFs
[17:03:47] [PASSED] 48 VFs
[17:03:47] [PASSED] 49 VFs
[17:03:47] [PASSED] 50 VFs
[17:03:47] [PASSED] 51 VFs
[17:03:47] [PASSED] 52 VFs
[17:03:47] [PASSED] 53 VFs
[17:03:47] [PASSED] 54 VFs
[17:03:47] [PASSED] 55 VFs
[17:03:47] [PASSED] 56 VFs
[17:03:47] [PASSED] 57 VFs
[17:03:47] [PASSED] 58 VFs
[17:03:47] [PASSED] 59 VFs
[17:03:47] [PASSED] 60 VFs
[17:03:47] [PASSED] 61 VFs
[17:03:47] [PASSED] 62 VFs
[17:03:47] [PASSED] 63 VFs
[17:03:47] ==================== [PASSED] fair_vram ====================
[17:03:47] ================== [PASSED] pf_gt_config ===================
[17:03:47] ===================== lmtt (1 subtest) =====================
[17:03:47] ======================== test_ops  =========================
[17:03:47] [PASSED] 2-level
[17:03:47] [PASSED] multi-level
[17:03:47] ==================== [PASSED] test_ops =====================
[17:03:47] ====================== [PASSED] lmtt =======================
[17:03:47] ================= sriov_packet (1 subtest) =================
[17:03:47] [PASSED] test_descriptor_init
[17:03:47] ================== [PASSED] sriov_packet ===================
[17:03:47] ================= pf_service (11 subtests) =================
[17:03:47] [PASSED] pf_negotiate_any
[17:03:47] [PASSED] pf_negotiate_base_match
[17:03:47] [PASSED] pf_negotiate_base_newer
[17:03:47] [PASSED] pf_negotiate_base_next
[17:03:47] [SKIPPED] pf_negotiate_base_older (no older minor)
[17:03:47] [PASSED] pf_negotiate_base_prev
[17:03:47] [PASSED] pf_negotiate_latest_match
[17:03:47] [PASSED] pf_negotiate_latest_newer
[17:03:47] [PASSED] pf_negotiate_latest_next
[17:03:47] [SKIPPED] pf_negotiate_latest_older (no older minor)
[17:03:47] [SKIPPED] pf_negotiate_latest_prev (no prev major)
[17:03:47] =================== [PASSED] pf_service ====================
[17:03:47] ================= xe_guc_g2g (2 subtests) ==================
[17:03:47] ============== xe_live_guc_g2g_kunit_default  ==============
[17:03:47] ========= [SKIPPED] xe_live_guc_g2g_kunit_default ==========
[17:03:47] ============== xe_live_guc_g2g_kunit_allmem  ===============
[17:03:47] ========== [SKIPPED] xe_live_guc_g2g_kunit_allmem ==========
[17:03:47] =================== [SKIPPED] xe_guc_g2g ===================
[17:03:47] =================== xe_mocs (2 subtests) ===================
[17:03:47] ================ xe_live_mocs_kernel_kunit  ================
[17:03:47] =========== [SKIPPED] xe_live_mocs_kernel_kunit ============
[17:03:47] ================ xe_live_mocs_reset_kunit  =================
[17:03:47] ============ [SKIPPED] xe_live_mocs_reset_kunit ============
[17:03:47] ==================== [SKIPPED] xe_mocs =====================
[17:03:47] ================= xe_migrate (2 subtests) ==================
[17:03:47] ================= xe_migrate_sanity_kunit  =================
[17:03:47] ============ [SKIPPED] xe_migrate_sanity_kunit =============
[17:03:47] ================== xe_validate_ccs_kunit  ==================
[17:03:47] ============= [SKIPPED] xe_validate_ccs_kunit ==============
[17:03:47] =================== [SKIPPED] xe_migrate ===================
[17:03:47] ================== xe_dma_buf (1 subtest) ==================
[17:03:47] ==================== xe_dma_buf_kunit  =====================
[17:03:47] ================ [SKIPPED] xe_dma_buf_kunit ================
[17:03:47] =================== [SKIPPED] xe_dma_buf ===================
[17:03:47] ================= xe_bo_shrink (1 subtest) =================
[17:03:47] =================== xe_bo_shrink_kunit  ====================
[17:03:47] =============== [SKIPPED] xe_bo_shrink_kunit ===============
[17:03:47] ================== [SKIPPED] xe_bo_shrink ==================
[17:03:47] ==================== xe_bo (2 subtests) ====================
[17:03:47] ================== xe_ccs_migrate_kunit  ===================
[17:03:47] ============== [SKIPPED] xe_ccs_migrate_kunit ==============
[17:03:47] ==================== xe_bo_evict_kunit  ====================
[17:03:47] =============== [SKIPPED] xe_bo_evict_kunit ================
[17:03:47] ===================== [SKIPPED] xe_bo ======================
[17:03:47] =================== xe_any (9 subtests) ====================
[17:03:47] [PASSED] test_to_xe
[17:03:47] [PASSED] test_to_dev
[17:03:47] [PASSED] test_to_pdev
[17:03:47] [PASSED] test_to_drm
[17:03:47] [PASSED] test_if_pdev
[17:03:47] [PASSED] test_if_xe
[17:03:47] [PASSED] test_if_tile
[17:03:47] [PASSED] test_if_gt
[17:03:47] [PASSED] test_to_id
[17:03:47] ===================== [PASSED] xe_any ======================
[17:03:47] ==================== args (13 subtests) ====================
[17:03:47] [PASSED] count_args_test
[17:03:47] [PASSED] call_args_example
[17:03:47] [PASSED] call_args_test
[17:03:47] [PASSED] drop_first_arg_example
[17:03:47] [PASSED] drop_first_arg_test
[17:03:47] [PASSED] first_arg_example
[17:03:47] [PASSED] first_arg_test
[17:03:47] [PASSED] last_arg_example
[17:03:47] [PASSED] last_arg_test
[17:03:47] [PASSED] pick_arg_example
[17:03:47] [PASSED] if_args_example
[17:03:47] [PASSED] if_args_test
[17:03:47] [PASSED] sep_comma_example
[17:03:47] ====================== [PASSED] args =======================
[17:03:47] =================== xe_pci (3 subtests) ====================
[17:03:47] ==================== check_graphics_ip  ====================
[17:03:47] [PASSED] 12.00 Xe_LP
[17:03:47] [PASSED] 12.10 Xe_LP+
[17:03:47] [PASSED] 12.55 Xe_HPG
[17:03:47] [PASSED] 12.60 Xe_HPC
[17:03:47] [PASSED] 12.70 Xe_LPG
[17:03:47] [PASSED] 12.71 Xe_LPG
[17:03:47] [PASSED] 12.74 Xe_LPG+
[17:03:47] [PASSED] 20.01 Xe2_HPG
[17:03:47] [PASSED] 20.02 Xe2_HPG
[17:03:47] [PASSED] 20.04 Xe2_LPG
[17:03:47] [PASSED] 30.00 Xe3_LPG
[17:03:47] [PASSED] 30.01 Xe3_LPG
[17:03:47] [PASSED] 30.03 Xe3_LPG
[17:03:47] [PASSED] 30.04 Xe3_LPG
[17:03:47] [PASSED] 30.05 Xe3_LPG
[17:03:47] [PASSED] 35.10 Xe3p_LPG
[17:03:47] [PASSED] 35.11 Xe3p_XPC
[17:03:47] ================ [PASSED] check_graphics_ip ================
[17:03:47] ===================== check_media_ip  ======================
[17:03:47] [PASSED] 12.00 Xe_M
[17:03:47] [PASSED] 12.55 Xe_HPM
[17:03:47] [PASSED] 13.00 Xe_LPM+
[17:03:47] [PASSED] 13.01 Xe2_HPM
[17:03:47] [PASSED] 20.00 Xe2_LPM
[17:03:47] [PASSED] 30.00 Xe3_LPM
[17:03:47] [PASSED] 30.02 Xe3_LPM
[17:03:47] [PASSED] 35.00 Xe3p_LPM
[17:03:47] [PASSED] 35.03 Xe3p_HPM
[17:03:47] ================= [PASSED] check_media_ip ==================
[17:03:47] =================== check_platform_desc  ===================
[17:03:47] [PASSED] 0x9A60 (TIGERLAKE)
[17:03:47] [PASSED] 0x9A68 (TIGERLAKE)
[17:03:47] [PASSED] 0x9A70 (TIGERLAKE)
[17:03:47] [PASSED] 0x9A40 (TIGERLAKE)
[17:03:47] [PASSED] 0x9A49 (TIGERLAKE)
[17:03:47] [PASSED] 0x9A59 (TIGERLAKE)
[17:03:47] [PASSED] 0x9A78 (TIGERLAKE)
[17:03:47] [PASSED] 0x9AC0 (TIGERLAKE)
[17:03:47] [PASSED] 0x9AC9 (TIGERLAKE)
[17:03:47] [PASSED] 0x9AD9 (TIGERLAKE)
[17:03:47] [PASSED] 0x9AF8 (TIGERLAKE)
[17:03:47] [PASSED] 0x4C80 (ROCKETLAKE)
[17:03:47] [PASSED] 0x4C8A (ROCKETLAKE)
[17:03:47] [PASSED] 0x4C8B (ROCKETLAKE)
[17:03:47] [PASSED] 0x4C8C (ROCKETLAKE)
[17:03:47] [PASSED] 0x4C90 (ROCKETLAKE)
[17:03:47] [PASSED] 0x4C9A (ROCKETLAKE)
[17:03:47] [PASSED] 0x4680 (ALDERLAKE_S)
[17:03:47] [PASSED] 0x4682 (ALDERLAKE_S)
[17:03:47] [PASSED] 0x4688 (ALDERLAKE_S)
[17:03:47] [PASSED] 0x468A (ALDERLAKE_S)
[17:03:47] [PASSED] 0x468B (ALDERLAKE_S)
[17:03:47] [PASSED] 0x4690 (ALDERLAKE_S)
[17:03:47] [PASSED] 0x4692 (ALDERLAKE_S)
[17:03:47] [PASSED] 0x4693 (ALDERLAKE_S)
[17:03:47] [PASSED] 0x46A0 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46A1 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46A2 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46A3 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46A6 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46A8 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46AA (ALDERLAKE_P)
[17:03:47] [PASSED] 0x462A (ALDERLAKE_P)
[17:03:47] [PASSED] 0x4626 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x4628 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46B0 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46B1 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46B2 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46B3 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46C0 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46C1 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46C2 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46C3 (ALDERLAKE_P)
[17:03:47] [PASSED] 0x46D0 (ALDERLAKE_N)
[17:03:47] [PASSED] 0x46D1 (ALDERLAKE_N)
[17:03:47] [PASSED] 0x46D2 (ALDERLAKE_N)
[17:03:47] [PASSED] 0x46D3 (ALDERLAKE_N)
[17:03:47] [PASSED] 0x46D4 (ALDERLAKE_N)
[17:03:47] [PASSED] 0xA721 (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA7A1 (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA7A9 (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA7AC (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA7AD (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA720 (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA7A0 (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA7A8 (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA7AA (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA7AB (ALDERLAKE_P)
[17:03:47] [PASSED] 0xA780 (ALDERLAKE_S)
[17:03:47] [PASSED] 0xA781 (ALDERLAKE_S)
[17:03:47] [PASSED] 0xA782 (ALDERLAKE_S)
[17:03:47] [PASSED] 0xA783 (ALDERLAKE_S)
[17:03:47] [PASSED] 0xA788 (ALDERLAKE_S)
[17:03:47] [PASSED] 0xA789 (ALDERLAKE_S)
[17:03:47] [PASSED] 0xA78A (ALDERLAKE_S)
[17:03:47] [PASSED] 0xA78B (ALDERLAKE_S)
[17:03:47] [PASSED] 0x4905 (DG1)
[17:03:47] [PASSED] 0x4906 (DG1)
[17:03:47] [PASSED] 0x4907 (DG1)
[17:03:47] [PASSED] 0x4908 (DG1)
[17:03:47] [PASSED] 0x4909 (DG1)
[17:03:47] [PASSED] 0x56C0 (DG2)
[17:03:47] [PASSED] 0x56C2 (DG2)
[17:03:47] [PASSED] 0x56C1 (DG2)
[17:03:47] [PASSED] 0x7D51 (METEORLAKE)
[17:03:47] [PASSED] 0x7DD1 (METEORLAKE)
[17:03:47] [PASSED] 0x7D41 (METEORLAKE)
[17:03:47] [PASSED] 0x7D67 (METEORLAKE)
[17:03:47] [PASSED] 0xB640 (METEORLAKE)
[17:03:47] [PASSED] 0x56A0 (DG2)
[17:03:47] [PASSED] 0x56A1 (DG2)
[17:03:47] [PASSED] 0x56A2 (DG2)
[17:03:47] [PASSED] 0x56BE (DG2)
[17:03:47] [PASSED] 0x56BF (DG2)
[17:03:47] [PASSED] 0x5690 (DG2)
[17:03:47] [PASSED] 0x5691 (DG2)
[17:03:47] [PASSED] 0x5692 (DG2)
[17:03:47] [PASSED] 0x56A5 (DG2)
[17:03:47] [PASSED] 0x56A6 (DG2)
[17:03:47] [PASSED] 0x56B0 (DG2)
[17:03:47] [PASSED] 0x56B1 (DG2)
[17:03:47] [PASSED] 0x56BA (DG2)
[17:03:47] [PASSED] 0x56BB (DG2)
[17:03:47] [PASSED] 0x56BC (DG2)
[17:03:47] [PASSED] 0x56BD (DG2)
[17:03:47] [PASSED] 0x5693 (DG2)
[17:03:47] [PASSED] 0x5694 (DG2)
[17:03:47] [PASSED] 0x5695 (DG2)
[17:03:47] [PASSED] 0x56A3 (DG2)
[17:03:47] [PASSED] 0x56A4 (DG2)
[17:03:47] [PASSED] 0x56B2 (DG2)
[17:03:47] [PASSED] 0x56B3 (DG2)
[17:03:47] [PASSED] 0x5696 (DG2)
[17:03:47] [PASSED] 0x5697 (DG2)
[17:03:47] [PASSED] 0xB69 (PVC)
[17:03:47] [PASSED] 0xB6E (PVC)
[17:03:47] [PASSED] 0xBD4 (PVC)
[17:03:47] [PASSED] 0xBD5 (PVC)
[17:03:47] [PASSED] 0xBD6 (PVC)
[17:03:47] [PASSED] 0xBD7 (PVC)
[17:03:47] [PASSED] 0xBD8 (PVC)
[17:03:47] [PASSED] 0xBD9 (PVC)
[17:03:47] [PASSED] 0xBDA (PVC)
[17:03:47] [PASSED] 0xBDB (PVC)
[17:03:47] [PASSED] 0xBE0 (PVC)
[17:03:47] [PASSED] 0xBE1 (PVC)
[17:03:47] [PASSED] 0xBE5 (PVC)
[17:03:47] [PASSED] 0x7D40 (METEORLAKE)
[17:03:47] [PASSED] 0x7D45 (METEORLAKE)
[17:03:47] [PASSED] 0x7D55 (METEORLAKE)
[17:03:47] [PASSED] 0x7D60 (METEORLAKE)
[17:03:47] [PASSED] 0x7DD5 (METEORLAKE)
[17:03:47] [PASSED] 0x6420 (LUNARLAKE)
[17:03:47] [PASSED] 0x64A0 (LUNARLAKE)
[17:03:47] [PASSED] 0x64B0 (LUNARLAKE)
[17:03:47] [PASSED] 0xE202 (BATTLEMAGE)
[17:03:47] [PASSED] 0xE209 (BATTLEMAGE)
[17:03:47] [PASSED] 0xE20B (BATTLEMAGE)
[17:03:47] [PASSED] 0xE20C (BATTLEMAGE)
[17:03:47] [PASSED] 0xE20D (BATTLEMAGE)
[17:03:47] [PASSED] 0xE210 (BATTLEMAGE)
[17:03:47] [PASSED] 0xE211 (BATTLEMAGE)
[17:03:47] [PASSED] 0xE212 (BATTLEMAGE)
[17:03:47] [PASSED] 0xE216 (BATTLEMAGE)
[17:03:47] [PASSED] 0xE220 (BATTLEMAGE)
[17:03:47] [PASSED] 0xE221 (BATTLEMAGE)
[17:03:47] [PASSED] 0xE222 (BATTLEMAGE)
[17:03:47] [PASSED] 0xE223 (BATTLEMAGE)
[17:03:47] [PASSED] 0xB080 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB081 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB082 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB083 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB084 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB085 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB086 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB087 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB08F (PANTHERLAKE)
[17:03:47] [PASSED] 0xB090 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB0A0 (PANTHERLAKE)
[17:03:47] [PASSED] 0xB0B0 (PANTHERLAKE)
[17:03:47] [PASSED] 0xFD80 (PANTHERLAKE)
[17:03:47] [PASSED] 0xFD81 (PANTHERLAKE)
[17:03:47] [PASSED] 0xD740 (NOVALAKE_S)
[17:03:47] [PASSED] 0xD741 (NOVALAKE_S)
[17:03:47] [PASSED] 0xD742 (NOVALAKE_S)
[17:03:47] [PASSED] 0xD743 (NOVALAKE_S)
[17:03:47] [PASSED] 0xD745 (NOVALAKE_S)
[17:03:47] [PASSED] 0xD74A (NOVALAKE_S)
[17:03:47] [PASSED] 0xD74B (NOVALAKE_S)
[17:03:47] [PASSED] 0x674C (CRESCENTISLAND)
[17:03:47] [PASSED] 0x674D (CRESCENTISLAND)
[17:03:47] [PASSED] 0x674E (CRESCENTISLAND)
[17:03:47] [PASSED] 0x674F (CRESCENTISLAND)
[17:03:47] [PASSED] 0x6750 (CRESCENTISLAND)
[17:03:47] [PASSED] 0xD750 (NOVALAKE_P)
[17:03:47] [PASSED] 0xD751 (NOVALAKE_P)
[17:03:47] [PASSED] 0xD752 (NOVALAKE_P)
[17:03:47] [PASSED] 0xD753 (NOVALAKE_P)
[17:03:47] [PASSED] 0xD754 (NOVALAKE_P)
[17:03:47] [PASSED] 0xD755 (NOVALAKE_P)
[17:03:47] [PASSED] 0xD756 (NOVALAKE_P)
[17:03:47] [PASSED] 0xD757 (NOVALAKE_P)
[17:03:47] [PASSED] 0xD75F (NOVALAKE_P)
[17:03:47] =============== [PASSED] check_platform_desc ===============
[17:03:47] ===================== [PASSED] xe_pci ======================
[17:03:47] ============= xe_rtp_tables_test (5 subtests) ==============
[17:03:47] ================== xe_rtp_table_gt_test  ===================
[17:03:47] [PASSED] gt_was/14011060649
[17:03:47] [PASSED] gt_was/14011059788
[17:03:47] [PASSED] gt_was/14015795083
[17:03:47] [PASSED] gt_was/16021867713
[17:03:47] [PASSED] gt_was/14019449301
[17:03:47] [PASSED] gt_was/16028005424
[17:03:47] [PASSED] gt_was/14026578760
[17:03:47] [PASSED] gt_was/1409420604
[17:03:47] [PASSED] gt_was/1408615072
[17:03:47] [PASSED] gt_was/22010523718
[17:03:47] [PASSED] gt_was/14011006942
[17:03:47] [PASSED] gt_was/14014830051
[17:03:47] [PASSED] gt_was/18018781329
[17:03:47] [PASSED] gt_was/1509235366
[17:03:47] [PASSED] gt_was/18018781329
[17:03:47] [PASSED] gt_was/16016694945
[17:03:47] [PASSED] gt_was/14018575942
[17:03:47] [PASSED] gt_was/22016670082
[17:03:47] [PASSED] gt_was/22016670082
[17:03:47] [PASSED] gt_was/14017421178
[17:03:47] [PASSED] gt_was/16025250150
[17:03:47] [PASSED] gt_was/14021871409
[17:03:47] [PASSED] gt_was/16021865536
[17:03:47] [PASSED] gt_was/14021486841
[17:03:47] [PASSED] gt_was/14025160223
[17:03:47] [PASSED] gt_was/14026144927, 16029437861, 14026127056
[17:03:47] [PASSED] gt_was/14025635424
[17:03:47] [PASSED] gt_was/16028005424
[17:03:47] ============== [PASSED] xe_rtp_table_gt_test ===============
[17:03:47] ================== xe_rtp_table_gt_test  ===================
[17:03:47] [PASSED] gt_tunings/Tuning: Blend Fill Caching Optimization Disable
[17:03:47] [PASSED] gt_tunings/Tuning: 32B Access Enable
[17:03:47] [PASSED] gt_tunings/Tuning: L3 cache
[17:03:47] [PASSED] gt_tunings/Tuning: L3 cache - media
[17:03:47] [PASSED] gt_tunings/Tuning: Compression Overfetch
[17:03:47] [PASSED] gt_tunings/Tuning: Compression Overfetch - media
[17:03:47] [PASSED] gt_tunings/Tuning: Enable compressible partial write overfetch in L3
[17:03:47] [PASSED] gt_tunings/Tuning: Enable compressible partial write overfetch in L3 - media
[17:03:47] [PASSED] gt_tunings/Tuning: L2 Overfetch Compressible Only
[17:03:47] [PASSED] gt_tunings/Tuning: L2 Overfetch Compressible Only - media
[17:03:47] [PASSED] gt_tunings/Tuning: Stateless compression control
[17:03:47] [PASSED] gt_tunings/Tuning: Stateless compression control - media
[17:03:47] [PASSED] gt_tunings/Tuning: L3 RW flush all Cache
[17:03:47] [PASSED] gt_tunings/Tuning: L3 RW flush all cache - media
[17:03:47] [PASSED] gt_tunings/Tuning: Set STLB Bank Hash Mode to 4KB
[17:03:47] ============== [PASSED] xe_rtp_table_gt_test ===============
[17:03:47] ================== xe_rtp_table_oob_test  ==================
[17:03:47] [PASSED] oob_was/1607983814
[17:03:47] [PASSED] oob_was/16010904313
[17:03:47] [PASSED] oob_was/18022495364
[17:03:47] [PASSED] oob_was/22012773006
[17:03:47] [PASSED] oob_was/14014475959
[17:03:47] [PASSED] oob_was/22011391025
[17:03:47] [PASSED] oob_was/22012727170
[17:03:47] [PASSED] oob_was/22012727685
[17:03:47] [PASSED] oob_was/22016596838
[17:03:47] [PASSED] oob_was/18020744125
[17:03:47] [PASSED] oob_was/1409600907
[17:03:47] [PASSED] oob_was/22014953428
[17:03:47] [PASSED] oob_was/16017236439
[17:03:47] [PASSED] oob_was/14019821291
[17:03:47] [PASSED] oob_was/14015076503
[17:03:47] [PASSED] oob_was/14018913170
[17:03:47] [PASSED] oob_was/14018094691
[17:03:47] [PASSED] oob_was/18024947630
[17:03:47] [PASSED] oob_was/16022287689
[17:03:47] [PASSED] oob_was/13011645652
[17:03:47] [PASSED] oob_was/14022293748
[17:03:47] [PASSED] oob_was/22019794406
[17:03:47] [PASSED] oob_was/22019338487
[17:03:47] [PASSED] oob_was/16023588340
[17:03:47] [PASSED] oob_was/14019789679
[17:03:47] [PASSED] oob_was/14022866841
[17:03:47] [PASSED] oob_was/16021333562
[17:03:47] [PASSED] oob_was/14016712196
[17:03:47] [PASSED] oob_was/14015568240
[17:03:47] [PASSED] oob_was/18013179988
[17:03:47] [PASSED] oob_was/1508761755
[17:03:47] [PASSED] oob_was/16023105232
[17:03:47] [PASSED] oob_was/16026508708
[17:03:47] [PASSED] oob_was/14020001231
[17:03:47] [PASSED] oob_was/16023683509
[17:03:47] [PASSED] oob_was/14025515070
[17:03:47] [PASSED] oob_was/15015404425_disable
[17:03:47] [PASSED] oob_was/16026007364
[17:03:47] [PASSED] oob_was/14020316580
[17:03:47] [PASSED] oob_was/14025883347
[17:03:47] [PASSED] oob_was/16029380221
[17:03:47] [PASSED] oob_was/22022079272
[17:03:47] [PASSED] oob_was/16029897822
[17:03:47] [PASSED] oob_was/14027054324
[17:03:47] ============== [PASSED] xe_rtp_table_oob_test ==============
[17:03:47] ================ xe_rtp_table_dev_oob_test  ================
[17:03:47] [PASSED] device_oob_was/22010954014
[17:03:47] [PASSED] device_oob_was/15015404425
[17:03:47] [PASSED] device_oob_was/22019338487_display
[17:03:47] [PASSED] device_oob_was/14022085890
[17:03:47] [PASSED] device_oob_was/14026539277
[17:03:47] [PASSED] device_oob_was/14026633728
[17:03:47] [PASSED] device_oob_was/14026746987
[17:03:47] [PASSED] device_oob_was/14026779378
[17:03:47] ============ [PASSED] xe_rtp_table_dev_oob_test ============
[17:03:47] ========== xe_rtp_table_missing_upper_bound_test  ==========
[17:03:47] [PASSED] register_whitelist/WaAllowPMDepthAndInvocationCountAccessFromUMD, 1408556865
[17:03:47] [PASSED] register_whitelist/1508744258, 14012131227, 1808121037
[17:03:47] [PASSED] register_whitelist/1806527549
[17:03:47] [PASSED] register_whitelist/allow_read_ctx_timestamp
[17:03:47] [PASSED] register_whitelist/allow_read_queue_timestamp
[17:03:47] [PASSED] register_whitelist/16014440446
[17:03:47] [PASSED] register_whitelist/16017236439
[17:03:47] [PASSED] register_whitelist/16020183090
[17:03:47] [PASSED] register_whitelist/14024997852
[17:03:47] [PASSED] register_whitelist/14024997852
[17:03:47] ====== [PASSED] xe_rtp_table_missing_upper_bound_test ======
[17:03:47] =============== [PASSED] xe_rtp_tables_test ================
[17:03:47] =================== xe_rtp (3 subtests) ====================
[17:03:47] =================== xe_rtp_rules_tests  ====================
[17:03:47] [PASSED] no
[17:03:47] [PASSED] yes
[17:03:47] [PASSED] no-and-no
[17:03:47] [PASSED] no-and-yes
[17:03:47] [PASSED] yes-and-no
[17:03:47] [PASSED] yes-and-yes
[17:03:47] [PASSED] no-or-no
[17:03:47] [PASSED] no-or-yes
[17:03:47] [PASSED] yes-or-no
[17:03:47] [PASSED] yes-or-yes
[17:03:47] [PASSED] no-yes-or-yes-no
[17:03:47] [PASSED] no-yes-or-yes-yes
[17:03:47] [PASSED] yes-yes-or-no-yes
[17:03:47] [PASSED] yes-yes-or-yes-yes
[17:03:47] [PASSED] no-no-or-yes-or-no
[17:03:47] [PASSED] or
[17:03:47] [PASSED] or-yes
[17:03:47] [PASSED] or-no
[17:03:47] [PASSED] yes-or
[17:03:47] [PASSED] no-or
[17:03:47] [PASSED] no-or-or-yes
[17:03:47] [PASSED] yes-or-or-no
[17:03:47] [PASSED] no-or-or-no
[17:03:47] [PASSED] missing-context-engine-class
[17:03:47] [PASSED] missing-context-engine-class-or-yes
[17:03:47] [PASSED] missing-context-engine-class-or-or-yes
[17:03:47] =============== [PASSED] xe_rtp_rules_tests ================
[17:03:47] =============== xe_rtp_process_to_sr_tests  ================
[17:03:47] [PASSED] coalesce-same-reg
[17:03:47] [PASSED] coalesce-same-reg-literal-and-func
[17:03:47] [PASSED] no-match-no-add
[17:03:47] [PASSED] two-regs-two-entries
[17:03:47] [PASSED] clr-one-set-other
[17:03:47] [PASSED] set-field
[17:03:47] [PASSED] conflict-duplicate
[17:03:47] [PASSED] conflict-not-disjoint
[17:03:47] [PASSED] conflict-not-disjoint-literal-and-func
[17:03:47] [PASSED] conflict-reg-type
[17:03:47] [PASSED] bad-mcr-reg-forced-to-regular
[17:03:47] [PASSED] bad-regular-reg-forced-to-mcr
[17:03:47] =========== [PASSED] xe_rtp_process_to_sr_tests ============
[17:03:47] ================== xe_rtp_process_tests  ===================
[17:03:47] [PASSED] active1
[17:03:47] [PASSED] active2
[17:03:47] [PASSED] active-inactive
[17:03:47] [PASSED] inactive-active
[17:03:47] [PASSED] inactive-active-inactive
[17:03:47] [PASSED] inactive-inactive-inactive
[17:03:47] ============== [PASSED] xe_rtp_process_tests ===============
[17:03:47] ===================== [PASSED] xe_rtp ======================
[17:03:47] ==================== xe_wa (1 subtest) =====================
[17:03:47] ======================== xe_wa_gt  =========================
[17:03:47] [PASSED] TIGERLAKE B0
[17:03:47] [PASSED] DG1 A0
[17:03:47] [PASSED] DG1 B0
[17:03:47] [PASSED] ALDERLAKE_S A0
[17:03:47] [PASSED] ALDERLAKE_S B0
[17:03:47] [PASSED] ALDERLAKE_S C0
[17:03:47] [PASSED] ALDERLAKE_S D0
[17:03:47] [PASSED] ALDERLAKE_P A0
[17:03:47] [PASSED] ALDERLAKE_P B0
[17:03:47] [PASSED] ALDERLAKE_P C0
[17:03:47] [PASSED] ALDERLAKE_S RPLS D0
[17:03:47] [PASSED] ALDERLAKE_P RPLU E0
[17:03:47] [PASSED] DG2 G10 C0
[17:03:47] [PASSED] DG2 G11 B1
[17:03:47] [PASSED] DG2 G12 A1
[17:03:47] [PASSED] METEORLAKE 12.70(Xe_LPG) A0 13.00(Xe_LPM+) A0
[17:03:47] [PASSED] METEORLAKE 12.71(Xe_LPG) A0 13.00(Xe_LPM+) A0
[17:03:47] [PASSED] METEORLAKE 12.74(Xe_LPG+) A0 13.00(Xe_LPM+) A0
[17:03:47] [PASSED] LUNARLAKE 20.04(Xe2_LPG) A0 20.00(Xe2_LPM) A0
[17:03:47] [PASSED] LUNARLAKE 20.04(Xe2_LPG) B0 20.00(Xe2_LPM) A0
[17:03:47] [PASSED] BATTLEMAGE 20.01(Xe2_HPG) A0 13.01(Xe2_HPM) A1
[17:03:47] [PASSED] PANTHERLAKE 30.00(Xe3_LPG) A0 30.00(Xe3_LPM) A0
[17:03:47] ==================== [PASSED] xe_wa_gt =====================
[17:03:47] ====================== [PASSED] xe_wa ======================
[17:03:47] ============================================================
[17:03:47] Testing complete. Ran 793 tests: passed: 765, skipped: 28
[17:03:47] Elapsed time: 36.735s total, 1.806s configuring, 34.213s building, 0.696s running

+ for kcfg in "${kunitconfigs[@]}"
+ [[ ! -f /kernel/drivers/gpu/drm/tests/.kunitconfig ]]
+ /kernel/tools/testing/kunit/kunit.py run --kunitconfig /kernel/drivers/gpu/drm/tests/.kunitconfig [17:03:47] Configuring KUnit Kernel ...
Regenerating .config ...
Populating config with:

$ make ARCH=um O=.kunit olddefconfig
[17:03:49] Building KUnit Kernel ...
Populating config with:
$ make ARCH=um O=.kunit olddefconfig
Building with:
$ make all compile_commands.json scripts_gdb ARCH=um O=.kunit --jobs=48
[17:04:14] Starting KUnit Kernel (1/1)...
[17:04:14] ============================================================
Running tests with:
$ .kunit/linux kunit.enable=1 mem=1G console=tty kunit_shutdown=halt
[17:04:14] ============= refcount_interrupt (4 subtests) ==============
[17:04:14] [PASSED] test_single_irq_change
[17:04:14] [PASSED] test_nested_irq_change
[17:04:14] [PASSED] test_multiple_irq_change
[17:04:14] [PASSED] test_irq_save
[17:04:14] =============== [PASSED] refcount_interrupt ================
[17:04:14] ============ drm_test_pick_cmdline (2 subtests) ============
[17:04:14] [PASSED] drm_test_pick_cmdline_res_1920_1080_60
[17:04:14] =============== drm_test_pick_cmdline_named  ===============
[17:04:14] [PASSED] NTSC
[17:04:14] [PASSED] NTSC-J
[17:04:14] [PASSED] PAL
[17:04:14] [PASSED] PAL-M
[17:04:14] =========== [PASSED] drm_test_pick_cmdline_named ===========
[17:04:14] ============== [PASSED] drm_test_pick_cmdline ==============
[17:04:14] == drm_test_atomic_get_connector_for_encoder (1 subtest) ===
[17:04:14] [PASSED] drm_test_drm_atomic_get_connector_for_encoder
[17:04:14] ==== [PASSED] drm_test_atomic_get_connector_for_encoder ====
[17:04:14] =========== drm_validate_clone_mode (2 subtests) ===========
[17:04:14] ============== drm_test_check_in_clone_mode  ===============
[17:04:14] [PASSED] in_clone_mode
[17:04:14] [PASSED] not_in_clone_mode
[17:04:14] ========== [PASSED] drm_test_check_in_clone_mode ===========
[17:04:14] =============== drm_test_check_valid_clones  ===============
[17:04:14] [PASSED] not_in_clone_mode
[17:04:14] [PASSED] valid_clone
[17:04:14] [PASSED] invalid_clone
[17:04:14] =========== [PASSED] drm_test_check_valid_clones ===========
[17:04:14] ============= [PASSED] drm_validate_clone_mode =============
[17:04:14] ============= drm_validate_modeset (1 subtest) =============
[17:04:14] [PASSED] drm_test_check_connector_changed_modeset
[17:04:14] ============== [PASSED] drm_validate_modeset ===============
[17:04:14] ====== drm_test_bridge_get_current_state (1 subtest) =======
[17:04:14] [PASSED] drm_test_drm_bridge_get_current_state_atomic
[17:04:14] ======== [PASSED] drm_test_bridge_get_current_state ========
[17:04:14] ====== drm_test_bridge_helper_reset_crtc (3 subtests) ======
[17:04:14] [PASSED] drm_test_drm_bridge_helper_reset_crtc_atomic
[17:04:14] [PASSED] drm_test_drm_bridge_helper_reset_crtc_atomic_disabled
[17:04:14] [PASSED] drm_test_drm_bridge_helper_hdmi_output_bus_fmts
[17:04:14] ======== [PASSED] drm_test_bridge_helper_reset_crtc ========
[17:04:14] ============== drm_bridge_alloc (2 subtests) ===============
[17:04:14] [PASSED] drm_test_drm_bridge_alloc_basic
[17:04:14] [PASSED] drm_test_drm_bridge_alloc_get_put
[17:04:14] ================ [PASSED] drm_bridge_alloc =================
[17:04:14] ============= drm_bridge_bus_fmt (5 subtests) ==============
[17:04:14] [PASSED] drm_test_bridge_rgb_yuv_rgb
[17:04:14] [PASSED] drm_test_bridge_must_convert_to_yuv444
[17:04:14] [PASSED] drm_test_bridge_hdmi_auto_rgb
[17:04:14] [PASSED] drm_test_bridge_auto_first
[17:04:14] [PASSED] drm_test_bridge_rgb_yuv_no_path
[17:04:14] =============== [PASSED] drm_bridge_bus_fmt ================
[17:04:14] ============= drm_cmdline_parser (40 subtests) =============
[17:04:14] [PASSED] drm_test_cmdline_force_d_only
[17:04:14] [PASSED] drm_test_cmdline_force_D_only_dvi
[17:04:14] [PASSED] drm_test_cmdline_force_D_only_hdmi
[17:04:14] [PASSED] drm_test_cmdline_force_D_only_not_digital
[17:04:14] [PASSED] drm_test_cmdline_force_e_only
[17:04:14] [PASSED] drm_test_cmdline_res
[17:04:14] [PASSED] drm_test_cmdline_res_vesa
[17:04:14] [PASSED] drm_test_cmdline_res_vesa_rblank
[17:04:14] [PASSED] drm_test_cmdline_res_rblank
[17:04:14] [PASSED] drm_test_cmdline_res_bpp
[17:04:14] [PASSED] drm_test_cmdline_res_refresh
[17:04:14] [PASSED] drm_test_cmdline_res_bpp_refresh
[17:04:14] [PASSED] drm_test_cmdline_res_bpp_refresh_interlaced
[17:04:14] [PASSED] drm_test_cmdline_res_bpp_refresh_margins
[17:04:14] [PASSED] drm_test_cmdline_res_bpp_refresh_force_off
[17:04:14] [PASSED] drm_test_cmdline_res_bpp_refresh_force_on
[17:04:14] [PASSED] drm_test_cmdline_res_bpp_refresh_force_on_analog
[17:04:14] [PASSED] drm_test_cmdline_res_bpp_refresh_force_on_digital
[17:04:14] [PASSED] drm_test_cmdline_res_bpp_refresh_interlaced_margins_force_on
[17:04:14] [PASSED] drm_test_cmdline_res_margins_force_on
[17:04:14] [PASSED] drm_test_cmdline_res_vesa_margins
[17:04:14] [PASSED] drm_test_cmdline_name
[17:04:14] [PASSED] drm_test_cmdline_name_bpp
[17:04:14] [PASSED] drm_test_cmdline_name_option
[17:04:14] [PASSED] drm_test_cmdline_name_bpp_option
[17:04:14] [PASSED] drm_test_cmdline_rotate_0
[17:04:14] [PASSED] drm_test_cmdline_rotate_90
[17:04:14] [PASSED] drm_test_cmdline_rotate_180
[17:04:14] [PASSED] drm_test_cmdline_rotate_270
[17:04:14] [PASSED] drm_test_cmdline_hmirror
[17:04:14] [PASSED] drm_test_cmdline_vmirror
[17:04:14] [PASSED] drm_test_cmdline_margin_options
[17:04:14] [PASSED] drm_test_cmdline_multiple_options
[17:04:14] [PASSED] drm_test_cmdline_bpp_extra_and_option
[17:04:14] [PASSED] drm_test_cmdline_extra_and_option
[17:04:14] [PASSED] drm_test_cmdline_freestanding_options
[17:04:14] [PASSED] drm_test_cmdline_freestanding_force_e_and_options
[17:04:14] [PASSED] drm_test_cmdline_panel_orientation
[17:04:14] ================ drm_test_cmdline_invalid  =================
[17:04:14] [PASSED] margin_only
[17:04:14] [PASSED] interlace_only
[17:04:14] [PASSED] res_missing_x
[17:04:14] [PASSED] res_missing_y
[17:04:14] [PASSED] res_bad_y
[17:04:14] [PASSED] res_missing_y_bpp
[17:04:14] [PASSED] res_bad_bpp
[17:04:14] [PASSED] res_bad_refresh
[17:04:14] [PASSED] res_bpp_refresh_force_on_off
[17:04:14] [PASSED] res_invalid_mode
[17:04:14] [PASSED] res_bpp_wrong_place_mode
[17:04:14] [PASSED] name_bpp_refresh
[17:04:14] [PASSED] name_refresh
[17:04:14] [PASSED] name_refresh_wrong_mode
[17:04:14] [PASSED] name_refresh_invalid_mode
[17:04:14] [PASSED] rotate_multiple
[17:04:14] [PASSED] rotate_invalid_val
[17:04:14] [PASSED] rotate_truncated
[17:04:14] [PASSED] invalid_option
[17:04:14] [PASSED] invalid_tv_option
[17:04:14] [PASSED] truncated_tv_option
[17:04:14] ============ [PASSED] drm_test_cmdline_invalid =============
[17:04:14] =============== drm_test_cmdline_tv_options  ===============
[17:04:14] [PASSED] NTSC
[17:04:14] [PASSED] NTSC_443
[17:04:14] [PASSED] NTSC_J
[17:04:14] [PASSED] PAL
[17:04:14] [PASSED] PAL_M
[17:04:14] [PASSED] PAL_N
[17:04:14] [PASSED] SECAM
[17:04:14] [PASSED] MONO_525
[17:04:14] [PASSED] MONO_625
[17:04:14] =========== [PASSED] drm_test_cmdline_tv_options ===========
[17:04:14] =============== [PASSED] drm_cmdline_parser ================
[17:04:14] ========== drmm_connector_hdmi_init (20 subtests) ==========
[17:04:14] [PASSED] drm_test_connector_hdmi_init_valid
[17:04:14] [PASSED] drm_test_connector_hdmi_init_bpc_8
[17:04:14] [PASSED] drm_test_connector_hdmi_init_bpc_10
[17:04:14] [PASSED] drm_test_connector_hdmi_init_bpc_12
[17:04:14] [PASSED] drm_test_connector_hdmi_init_bpc_invalid
[17:04:14] [PASSED] drm_test_connector_hdmi_init_bpc_null
[17:04:14] [PASSED] drm_test_connector_hdmi_init_formats_empty
[17:04:14] [PASSED] drm_test_connector_hdmi_init_formats_no_rgb
[17:04:14] === drm_test_connector_hdmi_init_formats_yuv420_allowed  ===
[17:04:14] [PASSED] supported_formats=0x9 yuv420_allowed=1
[17:04:14] [PASSED] supported_formats=0x9 yuv420_allowed=0
[17:04:14] [PASSED] supported_formats=0x5 yuv420_allowed=1
[17:04:14] [PASSED] supported_formats=0x5 yuv420_allowed=0
[17:04:14] === [PASSED] drm_test_connector_hdmi_init_formats_yuv420_allowed ===
[17:04:14] [PASSED] drm_test_connector_hdmi_init_null_ddc
[17:04:14] [PASSED] drm_test_connector_hdmi_init_null_product
[17:04:14] [PASSED] drm_test_connector_hdmi_init_null_vendor
[17:04:14] [PASSED] drm_test_connector_hdmi_init_product_length_exact
[17:04:14] [PASSED] drm_test_connector_hdmi_init_product_length_too_long
[17:04:14] [PASSED] drm_test_connector_hdmi_init_product_valid
[17:04:14] [PASSED] drm_test_connector_hdmi_init_vendor_length_exact
[17:04:14] [PASSED] drm_test_connector_hdmi_init_vendor_length_too_long
[17:04:14] [PASSED] drm_test_connector_hdmi_init_vendor_valid
[17:04:14] ========= drm_test_connector_hdmi_init_type_valid  =========
[17:04:14] [PASSED] HDMI-A
[17:04:14] [PASSED] HDMI-B
[17:04:14] ===== [PASSED] drm_test_connector_hdmi_init_type_valid =====
[17:04:14] ======== drm_test_connector_hdmi_init_type_invalid  ========
[17:04:14] [PASSED] Unknown
[17:04:14] [PASSED] VGA
[17:04:14] [PASSED] DVI-I
[17:04:14] [PASSED] DVI-D
[17:04:14] [PASSED] DVI-A
[17:04:14] [PASSED] Composite
[17:04:14] [PASSED] SVIDEO
[17:04:14] [PASSED] LVDS
[17:04:14] [PASSED] Component
[17:04:14] [PASSED] DIN
[17:04:14] [PASSED] DP
[17:04:14] [PASSED] TV
[17:04:14] [PASSED] eDP
[17:04:14] [PASSED] Virtual
[17:04:14] [PASSED] DSI
[17:04:14] [PASSED] DPI
[17:04:14] [PASSED] Writeback
[17:04:14] [PASSED] SPI
[17:04:14] [PASSED] USB
[17:04:14] ==== [PASSED] drm_test_connector_hdmi_init_type_invalid ====
[17:04:14] ============ [PASSED] drmm_connector_hdmi_init =============
[17:04:14] ============= drmm_connector_init (3 subtests) =============
[17:04:14] [PASSED] drm_test_drmm_connector_init
[17:04:14] [PASSED] drm_test_drmm_connector_init_null_ddc
[17:04:14] ========= drm_test_drmm_connector_init_type_valid  =========
[17:04:14] [PASSED] Unknown
[17:04:14] [PASSED] VGA
[17:04:14] [PASSED] DVI-I
[17:04:14] [PASSED] DVI-D
[17:04:14] [PASSED] DVI-A
[17:04:14] [PASSED] Composite
[17:04:14] [PASSED] SVIDEO
[17:04:14] [PASSED] LVDS
[17:04:14] [PASSED] Component
[17:04:14] [PASSED] DIN
[17:04:14] [PASSED] DP
[17:04:14] [PASSED] HDMI-A
[17:04:14] [PASSED] HDMI-B
[17:04:14] [PASSED] TV
[17:04:14] [PASSED] eDP
[17:04:14] [PASSED] Virtual
[17:04:14] [PASSED] DSI
[17:04:14] [PASSED] DPI
[17:04:14] [PASSED] Writeback
[17:04:14] [PASSED] SPI
[17:04:14] [PASSED] USB
[17:04:14] ===== [PASSED] drm_test_drmm_connector_init_type_valid =====
[17:04:14] =============== [PASSED] drmm_connector_init ===============
[17:04:14] ========= drm_connector_dynamic_init (6 subtests) ==========
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_init
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_init_null_ddc
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_init_not_added
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_init_properties
[17:04:14] ===== drm_test_drm_connector_dynamic_init_type_valid  ======
[17:04:14] [PASSED] Unknown
[17:04:14] [PASSED] VGA
[17:04:14] [PASSED] DVI-I
[17:04:14] [PASSED] DVI-D
[17:04:14] [PASSED] DVI-A
[17:04:14] [PASSED] Composite
[17:04:14] [PASSED] SVIDEO
[17:04:14] [PASSED] LVDS
[17:04:14] [PASSED] Component
[17:04:14] [PASSED] DIN
[17:04:14] [PASSED] DP
[17:04:14] [PASSED] HDMI-A
[17:04:14] [PASSED] HDMI-B
[17:04:14] [PASSED] TV
[17:04:14] [PASSED] eDP
[17:04:14] [PASSED] Virtual
[17:04:14] [PASSED] DSI
[17:04:14] [PASSED] DPI
[17:04:14] [PASSED] Writeback
[17:04:14] [PASSED] SPI
[17:04:14] [PASSED] USB
[17:04:14] = [PASSED] drm_test_drm_connector_dynamic_init_type_valid ==
[17:04:14] ======== drm_test_drm_connector_dynamic_init_name  =========
[17:04:14] [PASSED] Unknown
[17:04:14] [PASSED] VGA
[17:04:14] [PASSED] DVI-I
[17:04:14] [PASSED] DVI-D
[17:04:14] [PASSED] DVI-A
[17:04:14] [PASSED] Composite
[17:04:14] [PASSED] SVIDEO
[17:04:14] [PASSED] LVDS
[17:04:14] [PASSED] Component
[17:04:14] [PASSED] DIN
[17:04:14] [PASSED] DP
[17:04:14] [PASSED] HDMI-A
[17:04:14] [PASSED] HDMI-B
[17:04:14] [PASSED] TV
[17:04:14] [PASSED] eDP
[17:04:14] [PASSED] Virtual
[17:04:14] [PASSED] DSI
[17:04:14] [PASSED] DPI
[17:04:14] [PASSED] Writeback
[17:04:14] [PASSED] SPI
[17:04:14] [PASSED] USB
[17:04:14] ==== [PASSED] drm_test_drm_connector_dynamic_init_name =====
[17:04:14] =========== [PASSED] drm_connector_dynamic_init ============
[17:04:14] ==== drm_connector_dynamic_register_early (4 subtests) =====
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_early_on_list
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_early_defer
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_early_no_init
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_early_no_mode_object
[17:04:14] ====== [PASSED] drm_connector_dynamic_register_early =======
[17:04:14] ======= drm_connector_dynamic_register (7 subtests) ========
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_on_list
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_no_defer
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_no_init
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_mode_object
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_sysfs
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_sysfs_name
[17:04:14] [PASSED] drm_test_drm_connector_dynamic_register_debugfs
[17:04:14] ========= [PASSED] drm_connector_dynamic_register ==========
[17:04:14] = drm_connector_attach_broadcast_rgb_property (2 subtests) =
[17:04:14] [PASSED] drm_test_drm_connector_attach_broadcast_rgb_property
[17:04:14] [PASSED] drm_test_drm_connector_attach_broadcast_rgb_property_hdmi_connector
[17:04:14] === [PASSED] drm_connector_attach_broadcast_rgb_property ===
[17:04:14] ========== drm_get_tv_mode_from_name (2 subtests) ==========
[17:04:14] ========== drm_test_get_tv_mode_from_name_valid  ===========
[17:04:14] [PASSED] NTSC
[17:04:14] [PASSED] NTSC-443
[17:04:14] [PASSED] NTSC-J
[17:04:14] [PASSED] PAL
[17:04:14] [PASSED] PAL-M
[17:04:14] [PASSED] PAL-N
[17:04:14] [PASSED] SECAM
[17:04:14] [PASSED] Mono
[17:04:14] ====== [PASSED] drm_test_get_tv_mode_from_name_valid =======
[17:04:14] [PASSED] drm_test_get_tv_mode_from_name_truncated
[17:04:14] ============ [PASSED] drm_get_tv_mode_from_name ============
[17:04:14] = drm_test_connector_hdmi_compute_mode_clock (12 subtests) =
[17:04:14] [PASSED] drm_test_drm_hdmi_compute_mode_clock_rgb
[17:04:14] [PASSED] drm_test_drm_hdmi_compute_mode_clock_rgb_10bpc
[17:04:14] [PASSED] drm_test_drm_hdmi_compute_mode_clock_rgb_10bpc_vic_1
[17:04:14] [PASSED] drm_test_drm_hdmi_compute_mode_clock_rgb_12bpc
[17:04:14] [PASSED] drm_test_drm_hdmi_compute_mode_clock_rgb_12bpc_vic_1
[17:04:14] [PASSED] drm_test_drm_hdmi_compute_mode_clock_rgb_double
[17:04:14] = drm_test_connector_hdmi_compute_mode_clock_yuv420_valid  =
[17:04:14] [PASSED] VIC 96
[17:04:14] [PASSED] VIC 97
[17:04:14] [PASSED] VIC 101
[17:04:14] [PASSED] VIC 102
[17:04:14] [PASSED] VIC 106
[17:04:14] [PASSED] VIC 107
[17:04:14] === [PASSED] drm_test_connector_hdmi_compute_mode_clock_yuv420_valid ===
[17:04:14] [PASSED] drm_test_connector_hdmi_compute_mode_clock_yuv420_10_bpc
[17:04:14] [PASSED] drm_test_connector_hdmi_compute_mode_clock_yuv420_12_bpc
[17:04:14] [PASSED] drm_test_connector_hdmi_compute_mode_clock_yuv422_8_bpc
[17:04:14] [PASSED] drm_test_connector_hdmi_compute_mode_clock_yuv422_10_bpc
[17:04:14] [PASSED] drm_test_connector_hdmi_compute_mode_clock_yuv422_12_bpc
[17:04:14] === [PASSED] drm_test_connector_hdmi_compute_mode_clock ====
[17:04:14] == drm_hdmi_connector_get_broadcast_rgb_name (2 subtests) ==
[17:04:14] === drm_test_drm_hdmi_connector_get_broadcast_rgb_name  ====
[17:04:14] [PASSED] Automatic
[17:04:14] [PASSED] Full
[17:04:14] [PASSED] Limited 16:235
[17:04:14] === [PASSED] drm_test_drm_hdmi_connector_get_broadcast_rgb_name ===
[17:04:14] [PASSED] drm_test_drm_hdmi_connector_get_broadcast_rgb_name_invalid
[17:04:14] ==== [PASSED] drm_hdmi_connector_get_broadcast_rgb_name ====
[17:04:14] == drm_hdmi_connector_get_output_format_name (2 subtests) ==
[17:04:14] === drm_test_drm_hdmi_connector_get_output_format_name  ====
[17:04:14] [PASSED] RGB
[17:04:14] [PASSED] YUV 4:2:0
[17:04:14] [PASSED] YUV 4:2:2
[17:04:14] [PASSED] YUV 4:4:4
[17:04:14] === [PASSED] drm_test_drm_hdmi_connector_get_output_format_name ===
[17:04:14] [PASSED] drm_test_drm_hdmi_connector_get_output_format_name_invalid
[17:04:14] ==== [PASSED] drm_hdmi_connector_get_output_format_name ====
[17:04:14] ============= drm_damage_helper (21 subtests) ==============
[17:04:14] [PASSED] drm_test_damage_iter_no_damage
[17:04:14] [PASSED] drm_test_damage_iter_no_damage_fractional_src
[17:04:14] [PASSED] drm_test_damage_iter_no_damage_src_moved
[17:04:14] [PASSED] drm_test_damage_iter_no_damage_fractional_src_moved
[17:04:14] [PASSED] drm_test_damage_iter_no_damage_not_visible
[17:04:14] [PASSED] drm_test_damage_iter_no_damage_no_crtc
[17:04:14] [PASSED] drm_test_damage_iter_no_damage_no_fb
[17:04:14] [PASSED] drm_test_damage_iter_simple_damage
[17:04:14] [PASSED] drm_test_damage_iter_single_damage
[17:04:14] [PASSED] drm_test_damage_iter_single_damage_intersect_src
[17:04:14] [PASSED] drm_test_damage_iter_single_damage_outside_src
[17:04:14] [PASSED] drm_test_damage_iter_single_damage_fractional_src
[17:04:14] [PASSED] drm_test_damage_iter_single_damage_intersect_fractional_src
[17:04:14] [PASSED] drm_test_damage_iter_single_damage_outside_fractional_src
[17:04:14] [PASSED] drm_test_damage_iter_single_damage_src_moved
[17:04:14] [PASSED] drm_test_damage_iter_single_damage_fractional_src_moved
[17:04:14] [PASSED] drm_test_damage_iter_damage
[17:04:14] [PASSED] drm_test_damage_iter_damage_one_intersect
[17:04:14] [PASSED] drm_test_damage_iter_damage_one_outside
[17:04:14] [PASSED] drm_test_damage_iter_damage_src_moved
[17:04:14] [PASSED] drm_test_damage_iter_damage_not_visible
[17:04:14] ================ [PASSED] drm_damage_helper ================
[17:04:14] ============== drm_dp_mst_helper (3 subtests) ==============
[17:04:14] ============== drm_test_dp_mst_calc_pbn_mode  ==============
[17:04:14] [PASSED] Clock 154000 BPP 30 DSC disabled
[17:04:14] [PASSED] Clock 234000 BPP 30 DSC disabled
[17:04:14] [PASSED] Clock 297000 BPP 24 DSC disabled
[17:04:14] [PASSED] Clock 332880 BPP 24 DSC enabled
[17:04:14] [PASSED] Clock 324540 BPP 24 DSC enabled
[17:04:14] ========== [PASSED] drm_test_dp_mst_calc_pbn_mode ==========
[17:04:14] ============== drm_test_dp_mst_calc_pbn_div  ===============
[17:04:14] [PASSED] Link rate 2000000 lane count 4
[17:04:14] [PASSED] Link rate 2000000 lane count 2
[17:04:14] [PASSED] Link rate 2000000 lane count 1
[17:04:14] [PASSED] Link rate 1350000 lane count 4
[17:04:14] [PASSED] Link rate 1350000 lane count 2
[17:04:14] [PASSED] Link rate 1350000 lane count 1
[17:04:14] [PASSED] Link rate 1000000 lane count 4
[17:04:14] [PASSED] Link rate 1000000 lane count 2
[17:04:14] [PASSED] Link rate 1000000 lane count 1
[17:04:14] [PASSED] Link rate 810000 lane count 4
[17:04:14] [PASSED] Link rate 810000 lane count 2
[17:04:14] [PASSED] Link rate 810000 lane count 1
[17:04:14] [PASSED] Link rate 540000 lane count 4
[17:04:14] [PASSED] Link rate 540000 lane count 2
[17:04:14] [PASSED] Link rate 540000 lane count 1
[17:04:14] [PASSED] Link rate 270000 lane count 4
[17:04:14] [PASSED] Link rate 270000 lane count 2
[17:04:14] [PASSED] Link rate 270000 lane count 1
[17:04:14] [PASSED] Link rate 162000 lane count 4
[17:04:14] [PASSED] Link rate 162000 lane count 2
[17:04:14] [PASSED] Link rate 162000 lane count 1
[17:04:14] ========== [PASSED] drm_test_dp_mst_calc_pbn_div ===========
[17:04:14] ========= drm_test_dp_mst_sideband_msg_req_decode  =========
[17:04:14] [PASSED] DP_ENUM_PATH_RESOURCES with port number
[17:04:14] [PASSED] DP_POWER_UP_PHY with port number
[17:04:14] [PASSED] DP_POWER_DOWN_PHY with port number
[17:04:14] [PASSED] DP_ALLOCATE_PAYLOAD with SDP stream sinks
[17:04:14] [PASSED] DP_ALLOCATE_PAYLOAD with port number
[17:04:14] [PASSED] DP_ALLOCATE_PAYLOAD with VCPI
[17:04:14] [PASSED] DP_ALLOCATE_PAYLOAD with PBN
[17:04:14] [PASSED] DP_QUERY_PAYLOAD with port number
[17:04:14] [PASSED] DP_QUERY_PAYLOAD with VCPI
[17:04:14] [PASSED] DP_REMOTE_DPCD_READ with port number
[17:04:14] [PASSED] DP_REMOTE_DPCD_READ with DPCD address
[17:04:14] [PASSED] DP_REMOTE_DPCD_READ with max number of bytes
[17:04:14] [PASSED] DP_REMOTE_DPCD_WRITE with port number
[17:04:14] [PASSED] DP_REMOTE_DPCD_WRITE with DPCD address
[17:04:14] [PASSED] DP_REMOTE_DPCD_WRITE with data array
[17:04:14] [PASSED] DP_REMOTE_I2C_READ with port number
[17:04:14] [PASSED] DP_REMOTE_I2C_READ with I2C device ID
[17:04:14] [PASSED] DP_REMOTE_I2C_READ with transactions array
[17:04:14] [PASSED] DP_REMOTE_I2C_WRITE with port number
[17:04:14] [PASSED] DP_REMOTE_I2C_WRITE with I2C device ID
[17:04:14] [PASSED] DP_REMOTE_I2C_WRITE with data array
[17:04:14] [PASSED] DP_QUERY_STREAM_ENC_STATUS with stream ID
[17:04:14] [PASSED] DP_QUERY_STREAM_ENC_STATUS with client ID
[17:04:14] [PASSED] DP_QUERY_STREAM_ENC_STATUS with stream event
[17:04:14] [PASSED] DP_QUERY_STREAM_ENC_STATUS with valid stream event
[17:04:14] [PASSED] DP_QUERY_STREAM_ENC_STATUS with stream behavior
[17:04:14] [PASSED] DP_QUERY_STREAM_ENC_STATUS with a valid stream behavior
[17:04:14] ===== [PASSED] drm_test_dp_mst_sideband_msg_req_decode =====
[17:04:14] ================ [PASSED] drm_dp_mst_helper ================
[17:04:14] ================== drm_exec (7 subtests) ===================
[17:04:14] [PASSED] sanitycheck
[17:04:14] [PASSED] test_lock
[17:04:14] [PASSED] test_lock_unlock
[17:04:14] [PASSED] test_duplicates
[17:04:14] [PASSED] test_prepare
[17:04:14] [PASSED] test_prepare_array
[17:04:14] [PASSED] test_multiple_loops
[17:04:14] ==================== [PASSED] drm_exec =====================
[17:04:14] =========== drm_format_helper_test (17 subtests) ===========
[17:04:14] ============== drm_test_fb_xrgb8888_to_gray8  ==============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ========== [PASSED] drm_test_fb_xrgb8888_to_gray8 ==========
[17:04:14] ============= drm_test_fb_xrgb8888_to_rgb332  ==============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ========= [PASSED] drm_test_fb_xrgb8888_to_rgb332 ==========
[17:04:14] ============= drm_test_fb_xrgb8888_to_rgb565  ==============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ========= [PASSED] drm_test_fb_xrgb8888_to_rgb565 ==========
[17:04:14] ============ drm_test_fb_xrgb8888_to_xrgb1555  =============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ======== [PASSED] drm_test_fb_xrgb8888_to_xrgb1555 =========
[17:04:14] ============ drm_test_fb_xrgb8888_to_argb1555  =============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ======== [PASSED] drm_test_fb_xrgb8888_to_argb1555 =========
[17:04:14] ============ drm_test_fb_xrgb8888_to_rgba5551  =============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ======== [PASSED] drm_test_fb_xrgb8888_to_rgba5551 =========
[17:04:14] ============= drm_test_fb_xrgb8888_to_rgb888  ==============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ========= [PASSED] drm_test_fb_xrgb8888_to_rgb888 ==========
[17:04:14] ============= drm_test_fb_xrgb8888_to_bgr888  ==============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ========= [PASSED] drm_test_fb_xrgb8888_to_bgr888 ==========
[17:04:14] ============ drm_test_fb_xrgb8888_to_argb8888  =============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ======== [PASSED] drm_test_fb_xrgb8888_to_argb8888 =========
[17:04:14] =========== drm_test_fb_xrgb8888_to_xrgb2101010  ===========
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ======= [PASSED] drm_test_fb_xrgb8888_to_xrgb2101010 =======
[17:04:14] =========== drm_test_fb_xrgb8888_to_argb2101010  ===========
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ======= [PASSED] drm_test_fb_xrgb8888_to_argb2101010 =======
[17:04:14] ============== drm_test_fb_xrgb8888_to_mono  ===============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ========== [PASSED] drm_test_fb_xrgb8888_to_mono ===========
[17:04:14] ==================== drm_test_fb_swab  =====================
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ================ [PASSED] drm_test_fb_swab =================
[17:04:14] ============ drm_test_fb_xrgb8888_to_xbgr8888  =============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ======== [PASSED] drm_test_fb_xrgb8888_to_xbgr8888 =========
[17:04:14] ============ drm_test_fb_xrgb8888_to_abgr8888  =============
[17:04:14] [PASSED] single_pixel_source_buffer
[17:04:14] [PASSED] single_pixel_clip_rectangle
[17:04:14] [PASSED] well_known_colors
[17:04:14] [PASSED] destination_pitch
[17:04:14] ======== [PASSED] drm_test_fb_xrgb8888_to_abgr8888 =========
[17:04:14] ================= drm_test_fb_clip_offset  =================
[17:04:14] [PASSED] pass through
[17:04:14] [PASSED] horizontal offset
[17:04:14] [PASSED] vertical offset
[17:04:14] [PASSED] horizontal and vertical offset
[17:04:14] [PASSED] horizontal offset (custom pitch)
[17:04:14] [PASSED] vertical offset (custom pitch)
[17:04:14] [PASSED] horizontal and vertical offset (custom pitch)
[17:04:14] ============= [PASSED] drm_test_fb_clip_offset =============
[17:04:14] =================== drm_test_fb_memcpy  ====================
[17:04:14] [PASSED] single_pixel_source_buffer: XR24 little-endian (0x34325258)
[17:04:14] [PASSED] single_pixel_source_buffer: XRA8 little-endian (0x38415258)
[17:04:14] [PASSED] single_pixel_source_buffer: YU24 little-endian (0x34325559)
[17:04:14] [PASSED] single_pixel_clip_rectangle: XB24 little-endian (0x34324258)
[17:04:14] [PASSED] single_pixel_clip_rectangle: XRA8 little-endian (0x38415258)
[17:04:14] [PASSED] single_pixel_clip_rectangle: YU24 little-endian (0x34325559)
[17:04:14] [PASSED] well_known_colors: XB24 little-endian (0x34324258)
[17:04:14] [PASSED] well_known_colors: XRA8 little-endian (0x38415258)
[17:04:14] [PASSED] well_known_colors: YU24 little-endian (0x34325559)
[17:04:14] [PASSED] destination_pitch: XB24 little-endian (0x34324258)
[17:04:14] [PASSED] destination_pitch: XRA8 little-endian (0x38415258)
[17:04:14] [PASSED] destination_pitch: YU24 little-endian (0x34325559)
[17:04:14] =============== [PASSED] drm_test_fb_memcpy ================
[17:04:14] ============= [PASSED] drm_format_helper_test ==============
[17:04:14] ================= drm_format (18 subtests) =================
[17:04:14] [PASSED] drm_test_format_block_width_invalid
[17:04:14] [PASSED] drm_test_format_block_width_one_plane
[17:04:14] [PASSED] drm_test_format_block_width_two_plane
[17:04:14] [PASSED] drm_test_format_block_width_three_plane
[17:04:14] [PASSED] drm_test_format_block_width_tiled
[17:04:14] [PASSED] drm_test_format_block_height_invalid
[17:04:14] [PASSED] drm_test_format_block_height_one_plane
[17:04:14] [PASSED] drm_test_format_block_height_two_plane
[17:04:14] [PASSED] drm_test_format_block_height_three_plane
[17:04:14] [PASSED] drm_test_format_block_height_tiled
[17:04:14] [PASSED] drm_test_format_min_pitch_invalid
[17:04:14] [PASSED] drm_test_format_min_pitch_one_plane_8bpp
[17:04:14] [PASSED] drm_test_format_min_pitch_one_plane_16bpp
[17:04:14] [PASSED] drm_test_format_min_pitch_one_plane_24bpp
[17:04:14] [PASSED] drm_test_format_min_pitch_one_plane_32bpp
[17:04:14] [PASSED] drm_test_format_min_pitch_two_plane
[17:04:14] [PASSED] drm_test_format_min_pitch_three_plane_8bpp
[17:04:14] [PASSED] drm_test_format_min_pitch_tiled
[17:04:14] =================== [PASSED] drm_format ====================
[17:04:14] ============== drm_framebuffer (10 subtests) ===============
[17:04:14] ========== drm_test_framebuffer_check_src_coords  ==========
[17:04:14] [PASSED] Success: source fits into fb
[17:04:14] [PASSED] Fail: overflowing fb with x-axis coordinate
[17:04:14] [PASSED] Fail: overflowing fb with y-axis coordinate
[17:04:14] [PASSED] Fail: overflowing fb with source width
[17:04:14] [PASSED] Fail: overflowing fb with source height
[17:04:14] ====== [PASSED] drm_test_framebuffer_check_src_coords ======
[17:04:14] [PASSED] drm_test_framebuffer_cleanup
[17:04:14] =============== drm_test_framebuffer_create  ===============
[17:04:14] [PASSED] ABGR8888 normal sizes
[17:04:14] [PASSED] ABGR8888 max sizes
[17:04:14] [PASSED] ABGR8888 pitch greater than min required
[17:04:14] [PASSED] ABGR8888 pitch less than min required
[17:04:14] [PASSED] ABGR8888 Invalid width
[17:04:14] [PASSED] ABGR8888 Invalid buffer handle
[17:04:14] [PASSED] No pixel format
[17:04:14] [PASSED] ABGR8888 Width 0
[17:04:14] [PASSED] ABGR8888 Height 0
[17:04:14] [PASSED] ABGR8888 Out of bound height * pitch combination
[17:04:14] [PASSED] ABGR8888 Large buffer offset
[17:04:14] [PASSED] ABGR8888 Buffer offset for inexistent plane
[17:04:14] [PASSED] ABGR8888 Invalid flag
[17:04:14] [PASSED] ABGR8888 Set DRM_MODE_FB_MODIFIERS without modifiers
[17:04:14] [PASSED] ABGR8888 Valid buffer modifier
[17:04:14] [PASSED] ABGR8888 Invalid buffer modifier(DRM_FORMAT_MOD_SAMSUNG_64_32_TILE)
[17:04:14] [PASSED] ABGR8888 Extra pitches without DRM_MODE_FB_MODIFIERS
[17:04:14] [PASSED] ABGR8888 Extra pitches with DRM_MODE_FB_MODIFIERS
[17:04:14] [PASSED] NV12 Normal sizes
[17:04:14] [PASSED] NV12 Max sizes
[17:04:14] [PASSED] NV12 Invalid pitch
[17:04:14] [PASSED] NV12 Invalid modifier/missing DRM_MODE_FB_MODIFIERS flag
[17:04:14] [PASSED] NV12 different  modifier per-plane
[17:04:14] [PASSED] NV12 with DRM_FORMAT_MOD_SAMSUNG_64_32_TILE
[17:04:14] [PASSED] NV12 Valid modifiers without DRM_MODE_FB_MODIFIERS
[17:04:14] [PASSED] NV12 Modifier for inexistent plane
[17:04:14] [PASSED] NV12 Handle for inexistent plane
[17:04:14] [PASSED] NV12 Handle for inexistent plane without DRM_MODE_FB_MODIFIERS
[17:04:14] [PASSED] YVU420 DRM_MODE_FB_MODIFIERS set without modifier
[17:04:14] [PASSED] YVU420 Normal sizes
[17:04:14] [PASSED] YVU420 Max sizes
[17:04:14] [PASSED] YVU420 Invalid pitch
[17:04:14] [PASSED] YVU420 Different pitches
[17:04:14] [PASSED] YVU420 Different buffer offsets/pitches
[17:04:14] [PASSED] YVU420 Modifier set just for plane 0, without DRM_MODE_FB_MODIFIERS
[17:04:14] [PASSED] YVU420 Modifier set just for planes 0, 1, without DRM_MODE_FB_MODIFIERS
[17:04:14] [PASSED] YVU420 Modifier set just for plane 0, 1, with DRM_MODE_FB_MODIFIERS
[17:04:14] [PASSED] YVU420 Valid modifier
[17:04:14] [PASSED] YVU420 Different modifiers per plane
[17:04:14] [PASSED] YVU420 Modifier for inexistent plane
[17:04:14] [PASSED] YUV420_10BIT Invalid modifier(DRM_FORMAT_MOD_LINEAR)
[17:04:14] [PASSED] X0L2 Normal sizes
[17:04:14] [PASSED] X0L2 Max sizes
[17:04:14] [PASSED] X0L2 Invalid pitch
[17:04:14] [PASSED] X0L2 Pitch greater than minimum required
[17:04:14] [PASSED] X0L2 Handle for inexistent plane
[17:04:14] [PASSED] X0L2 Offset for inexistent plane, without DRM_MODE_FB_MODIFIERS set
[17:04:14] [PASSED] X0L2 Modifier without DRM_MODE_FB_MODIFIERS set
[17:04:14] [PASSED] X0L2 Valid modifier
[17:04:14] [PASSED] X0L2 Modifier for inexistent plane
[17:04:14] =========== [PASSED] drm_test_framebuffer_create ===========
[17:04:14] [PASSED] drm_test_framebuffer_free
[17:04:14] [PASSED] drm_test_framebuffer_init
[17:04:14] [PASSED] drm_test_framebuffer_init_bad_format
[17:04:14] [PASSED] drm_test_framebuffer_init_dev_mismatch
[17:04:14] [PASSED] drm_test_framebuffer_lookup
[17:04:14] [PASSED] drm_test_framebuffer_lookup_inexistent
[17:04:14] [PASSED] drm_test_framebuffer_modifiers_not_supported
[17:04:14] ================= [PASSED] drm_framebuffer =================
[17:04:14] ================ drm_gem_shmem (8 subtests) ================
[17:04:14] [PASSED] drm_gem_shmem_test_obj_create
[17:04:14] [PASSED] drm_gem_shmem_test_obj_create_private
[17:04:14] [PASSED] drm_gem_shmem_test_pin_pages
[17:04:14] [PASSED] drm_gem_shmem_test_vmap
[17:04:14] [PASSED] drm_gem_shmem_test_get_sg_table
[17:04:14] [PASSED] drm_gem_shmem_test_get_pages_sgt
[17:04:14] [PASSED] drm_gem_shmem_test_madvise
[17:04:14] [PASSED] drm_gem_shmem_test_purge
[17:04:14] ================== [PASSED] drm_gem_shmem ==================
[17:04:14] === drm_atomic_helper_connector_hdmi_check (29 subtests) ===
[17:04:14] [PASSED] drm_test_check_broadcast_rgb_auto_cea_mode
[17:04:14] [PASSED] drm_test_check_broadcast_rgb_auto_cea_mode_vic_1
[17:04:14] [PASSED] drm_test_check_broadcast_rgb_full_cea_mode
[17:04:14] [PASSED] drm_test_check_broadcast_rgb_full_cea_mode_vic_1
[17:04:14] [PASSED] drm_test_check_broadcast_rgb_limited_cea_mode
[17:04:14] [PASSED] drm_test_check_broadcast_rgb_limited_cea_mode_vic_1
[17:04:14] ====== drm_test_check_broadcast_rgb_cea_mode_yuv420  =======
[17:04:14] [PASSED] Automatic
[17:04:14] [PASSED] Full
[17:04:14] [PASSED] Limited 16:235
[17:04:14] == [PASSED] drm_test_check_broadcast_rgb_cea_mode_yuv420 ===
[17:04:14] [PASSED] drm_test_check_broadcast_rgb_crtc_mode_changed
[17:04:14] [PASSED] drm_test_check_broadcast_rgb_crtc_mode_not_changed
[17:04:14] [PASSED] drm_test_check_disable_connector
[17:04:14] [PASSED] drm_test_check_hdmi_funcs_reject_rate
[17:04:14] [PASSED] drm_test_check_max_tmds_rate_bpc_fallback_rgb
[17:04:14] [PASSED] drm_test_check_max_tmds_rate_bpc_fallback_yuv420
[17:04:14] [PASSED] drm_test_check_max_tmds_rate_bpc_fallback_ignore_yuv422
[17:04:14] [PASSED] drm_test_check_max_tmds_rate_bpc_fallback_ignore_yuv420
[17:04:14] [PASSED] drm_test_check_driver_unsupported_fallback_yuv420
[17:04:14] [PASSED] drm_test_check_output_bpc_crtc_mode_changed
[17:04:14] [PASSED] drm_test_check_output_bpc_crtc_mode_not_changed
[17:04:14] [PASSED] drm_test_check_output_bpc_dvi
[17:04:14] [PASSED] drm_test_check_output_bpc_format_vic_1
[17:04:14] [PASSED] drm_test_check_output_bpc_format_display_8bpc_only
[17:04:14] [PASSED] drm_test_check_output_bpc_format_display_rgb_only
[17:04:14] [PASSED] drm_test_check_output_bpc_format_driver_8bpc_only
[17:04:14] [PASSED] drm_test_check_output_bpc_format_driver_rgb_only
[17:04:14] [PASSED] drm_test_check_tmds_char_rate_rgb_8bpc
[17:04:14] [PASSED] drm_test_check_tmds_char_rate_rgb_10bpc
[17:04:14] [PASSED] drm_test_check_tmds_char_rate_rgb_12bpc
[17:04:14] ============ drm_test_check_hdmi_color_format  =============
[17:04:14] [PASSED] AUTO -> RGB
[17:04:14] [PASSED] YCBCR422 -> YUV422
[17:04:14] [PASSED] YCBCR420 -> YUV420
[17:04:14] [PASSED] YCBCR444 -> YUV444
[17:04:14] [PASSED] RGB -> RGB
[17:04:14] ======== [PASSED] drm_test_check_hdmi_color_format =========
[17:04:14] ======== drm_test_check_hdmi_color_format_420_only  ========
[17:04:14] [PASSED] RGB should fail
[17:04:14] [PASSED] YUV444 should fail
[17:04:14] [PASSED] YUV422 should fail
[17:04:14] [PASSED] YUV420 should work
[17:04:14] ==== [PASSED] drm_test_check_hdmi_color_format_420_only ====
[17:04:14] ===== [PASSED] drm_atomic_helper_connector_hdmi_check ======
[17:04:14] === drm_atomic_helper_connector_hdmi_reset (6 subtests) ====
[17:04:14] [PASSED] drm_test_check_broadcast_rgb_value
[17:04:14] [PASSED] drm_test_check_bpc_8_value
[17:04:14] [PASSED] drm_test_check_bpc_10_value
[17:04:14] [PASSED] drm_test_check_bpc_12_value
[17:04:14] [PASSED] drm_test_check_format_value
[17:04:14] [PASSED] drm_test_check_tmds_char_value
[17:04:14] ===== [PASSED] drm_atomic_helper_connector_hdmi_reset ======
[17:04:14] = drm_atomic_helper_connector_hdmi_mode_valid (7 subtests) =
[17:04:14] [PASSED] drm_test_check_mode_valid
[17:04:14] [PASSED] drm_test_check_mode_valid_reject
[17:04:14] [PASSED] drm_test_check_mode_valid_reject_rate
[17:04:14] [PASSED] drm_test_check_mode_valid_reject_max_clock
[17:04:14] [PASSED] drm_test_check_mode_valid_yuv420_only_max_clock
[17:04:14] [PASSED] drm_test_check_mode_valid_reject_yuv420_only_connector
[17:04:14] [PASSED] drm_test_check_mode_valid_accept_yuv420_also_connector_rgb
[17:04:14] === [PASSED] drm_atomic_helper_connector_hdmi_mode_valid ===
[17:04:14] = drm_atomic_helper_connector_hdmi_infoframes (5 subtests) =
[17:04:14] [PASSED] drm_test_check_infoframes
[17:04:14] [PASSED] drm_test_check_reject_avi_infoframe
[17:04:14] [PASSED] drm_test_check_reject_hdr_infoframe_bpc_8
[17:04:14] [PASSED] drm_test_check_reject_hdr_infoframe_bpc_10
[17:04:14] [PASSED] drm_test_check_reject_audio_infoframe
[17:04:14] === [PASSED] drm_atomic_helper_connector_hdmi_infoframes ===
[17:04:14] ================= drm_managed (2 subtests) =================
[17:04:14] [PASSED] drm_test_managed_release_action
[17:04:14] [PASSED] drm_test_managed_run_action
[17:04:14] =================== [PASSED] drm_managed ===================
[17:04:14] =================== drm_mm (6 subtests) ====================
[17:04:14] [PASSED] drm_test_mm_init
[17:04:14] [PASSED] drm_test_mm_debug
[17:04:14] [PASSED] drm_test_mm_align32
[17:04:14] [PASSED] drm_test_mm_align64
[17:04:14] [PASSED] drm_test_mm_lowest
[17:04:14] [PASSED] drm_test_mm_highest
[17:04:14] ===================== [PASSED] drm_mm ======================
[17:04:14] ============= drm_modes_analog_tv (5 subtests) =============
[17:04:14] [PASSED] drm_test_modes_analog_tv_mono_576i
[17:04:14] [PASSED] drm_test_modes_analog_tv_ntsc_480i
[17:04:14] [PASSED] drm_test_modes_analog_tv_ntsc_480i_inlined
[17:04:14] [PASSED] drm_test_modes_analog_tv_pal_576i
[17:04:14] [PASSED] drm_test_modes_analog_tv_pal_576i_inlined
[17:04:14] =============== [PASSED] drm_modes_analog_tv ===============
[17:04:14] ============== drm_plane_helper (2 subtests) ===============
[17:04:14] =============== drm_test_check_plane_state  ================
[17:04:14] [PASSED] clipping_simple
[17:04:14] [PASSED] clipping_rotate_reflect
[17:04:14] [PASSED] positioning_simple
[17:04:14] [PASSED] upscaling
[17:04:14] [PASSED] downscaling
[17:04:14] [PASSED] rounding1
[17:04:14] [PASSED] rounding2
[17:04:14] [PASSED] rounding3
[17:04:14] [PASSED] rounding4
[17:04:14] =========== [PASSED] drm_test_check_plane_state ============
[17:04:14] =========== drm_test_check_invalid_plane_state  ============
[17:04:14] [PASSED] positioning_invalid
[17:04:14] [PASSED] upscaling_invalid
[17:04:14] [PASSED] downscaling_invalid
[17:04:14] ======= [PASSED] drm_test_check_invalid_plane_state ========
[17:04:14] ================ [PASSED] drm_plane_helper =================
[17:04:14] ====== drm_connector_helper_tv_get_modes (1 subtest) =======
[17:04:14] ====== drm_test_connector_helper_tv_get_modes_check  =======
[17:04:14] [PASSED] None
[17:04:14] [PASSED] PAL
[17:04:14] [PASSED] NTSC
[17:04:14] [PASSED] Both, NTSC Default
[17:04:14] [PASSED] Both, PAL Default
[17:04:14] [PASSED] Both, NTSC Default, with PAL on command-line
[17:04:14] [PASSED] Both, PAL Default, with NTSC on command-line
[17:04:14] == [PASSED] drm_test_connector_helper_tv_get_modes_check ===
[17:04:14] ======== [PASSED] drm_connector_helper_tv_get_modes ========
[17:04:14] ================== drm_rect (9 subtests) ===================
[17:04:14] [PASSED] drm_test_rect_clip_scaled_div_by_zero
[17:04:14] [PASSED] drm_test_rect_clip_scaled_not_clipped
[17:04:14] [PASSED] drm_test_rect_clip_scaled_clipped
[17:04:14] [PASSED] drm_test_rect_clip_scaled_signed_vs_unsigned
[17:04:14] ================= drm_test_rect_intersect  =================
[17:04:14] [PASSED] top-left x bottom-right: 2x2+1+1 x 2x2+0+0
[17:04:14] [PASSED] top-right x bottom-left: 2x2+0+0 x 2x2+1-1
[17:04:14] [PASSED] bottom-left x top-right: 2x2+1-1 x 2x2+0+0
[17:04:14] [PASSED] bottom-right x top-left: 2x2+0+0 x 2x2+1+1
[17:04:14] [PASSED] right x left: 2x1+0+0 x 3x1+1+0
[17:04:14] [PASSED] left x right: 3x1+1+0 x 2x1+0+0
[17:04:14] [PASSED] up x bottom: 1x2+0+0 x 1x3+0-1
[17:04:14] [PASSED] bottom x up: 1x3+0-1 x 1x2+0+0
[17:04:14] [PASSED] touching corner: 1x1+0+0 x 2x2+1+1
[17:04:14] [PASSED] touching side: 1x1+0+0 x 1x1+1+0
[17:04:14] [PASSED] equal rects: 2x2+0+0 x 2x2+0+0
[17:04:14] [PASSED] inside another: 2x2+0+0 x 1x1+1+1
[17:04:14] [PASSED] far away: 1x1+0+0 x 1x1+3+6
[17:04:14] [PASSED] points intersecting: 0x0+5+10 x 0x0+5+10
[17:04:14] [PASSED] points not intersecting: 0x0+0+0 x 0x0+5+10
[17:04:14] ============= [PASSED] drm_test_rect_intersect =============
[17:04:14] ================ drm_test_rect_calc_hscale  ================
[17:04:14] [PASSED] normal use
[17:04:14] [PASSED] out of max range
[17:04:14] [PASSED] out of min range
[17:04:14] [PASSED] zero dst
[17:04:14] [PASSED] negative src
[17:04:14] [PASSED] negative dst
[17:04:14] ============ [PASSED] drm_test_rect_calc_hscale ============
[17:04:14] ================ drm_test_rect_calc_vscale  ================
[17:04:14] [PASSED] normal use
[17:04:14] [PASSED] out of max range
[17:04:14] [PASSED] out of min range
[17:04:14] [PASSED] zero dst
[17:04:14] [PASSED] negative src
[17:04:14] [PASSED] negative dst
[17:04:14] ============ [PASSED] drm_test_rect_calc_vscale ============
[17:04:14] ================== drm_test_rect_rotate  ===================
[17:04:14] [PASSED] reflect-x
[17:04:14] [PASSED] reflect-y
[17:04:14] [PASSED] rotate-0
[17:04:14] [PASSED] rotate-90
[17:04:14] [PASSED] rotate-180
[17:04:14] [PASSED] rotate-270
[17:04:14] ============== [PASSED] drm_test_rect_rotate ===============
[17:04:14] ================ drm_test_rect_rotate_inv  =================
[17:04:14] [PASSED] reflect-x
[17:04:14] [PASSED] reflect-y
[17:04:14] [PASSED] rotate-0
[17:04:14] [PASSED] rotate-90
[17:04:14] [PASSED] rotate-180
[17:04:14] [PASSED] rotate-270
[17:04:14] ============ [PASSED] drm_test_rect_rotate_inv =============
[17:04:14] ==================== [PASSED] drm_rect =====================
[17:04:14] ============ drm_sysfb_modeset_test (1 subtest) ============
[17:04:14] ============ drm_test_sysfb_build_fourcc_list  =============
[17:04:14] [PASSED] no native formats
[17:04:14] [PASSED] XRGB8888 as native format
[17:04:14] [PASSED] remove duplicates
[17:04:14] [PASSED] convert alpha formats
[17:04:14] [PASSED] random formats
[17:04:14] ======== [PASSED] drm_test_sysfb_build_fourcc_list =========
[17:04:14] ============= [PASSED] drm_sysfb_modeset_test ==============
[17:04:14] ================== drm_fixp (2 subtests) ===================
[17:04:14] [PASSED] drm_test_int2fixp
[17:04:14] [PASSED] drm_test_sm2fixp
[17:04:14] ==================== [PASSED] drm_fixp =====================
[17:04:14] ============================================================
[17:04:14] Testing complete. Ran 641 tests: passed: 641
[17:04:14] Elapsed time: 27.462s total, 1.830s configuring, 25.417s building, 0.183s running

+ for kcfg in "${kunitconfigs[@]}"
+ [[ ! -f /kernel/drivers/gpu/drm/ttm/tests/.kunitconfig ]]
+ /kernel/tools/testing/kunit/kunit.py run --kunitconfig /kernel/drivers/gpu/drm/ttm/tests/.kunitconfig [17:04:15] Configuring KUnit Kernel ...
Regenerating .config ...
Populating config with:

$ make ARCH=um O=.kunit olddefconfig
[17:04:16] Building KUnit Kernel ...
Populating config with:
$ make ARCH=um O=.kunit olddefconfig
Building with:
$ make all compile_commands.json scripts_gdb ARCH=um O=.kunit --jobs=48
[17:04:26] Starting KUnit Kernel (1/1)...
[17:04:26] ============================================================
Running tests with:
$ .kunit/linux kunit.enable=1 mem=1G console=tty kunit_shutdown=halt
[17:04:27] ============= refcount_interrupt (4 subtests) ==============
[17:04:27] [PASSED] test_single_irq_change
[17:04:27] [PASSED] test_nested_irq_change
[17:04:27] [PASSED] test_multiple_irq_change
[17:04:27] [PASSED] test_irq_save
[17:04:27] =============== [PASSED] refcount_interrupt ================
[17:04:27] ================= ttm_device (5 subtests) ==================
[17:04:27] [PASSED] ttm_device_init_basic
[17:04:27] [PASSED] ttm_device_init_multiple
[17:04:27] [PASSED] ttm_device_fini_basic
[17:04:27] [PASSED] ttm_device_init_no_vma_man
[17:04:27] ================== ttm_device_init_pools  ==================
[17:04:27] [PASSED] No DMA allocations, no DMA32 required
[17:04:27] [PASSED] DMA allocations, DMA32 required
[17:04:27] [PASSED] No DMA allocations, DMA32 required
[17:04:27] [PASSED] DMA allocations, no DMA32 required
[17:04:27] ============== [PASSED] ttm_device_init_pools ==============
[17:04:27] =================== [PASSED] ttm_device ====================
[17:04:27] ================== ttm_pool (8 subtests) ===================
[17:04:27] ================== ttm_pool_alloc_basic  ===================
[17:04:27] [PASSED] One page
[17:04:27] [PASSED] More than one page
[17:04:27] [PASSED] Above the allocation limit
[17:04:27] [PASSED] One page, with coherent DMA mappings enabled
[17:04:27] [PASSED] Above the allocation limit, with coherent DMA mappings enabled
[17:04:27] ============== [PASSED] ttm_pool_alloc_basic ===============
[17:04:27] ============== ttm_pool_alloc_basic_dma_addr  ==============
[17:04:27] [PASSED] One page
[17:04:27] [PASSED] More than one page
[17:04:27] [PASSED] Above the allocation limit
[17:04:27] [PASSED] One page, with coherent DMA mappings enabled
[17:04:27] [PASSED] Above the allocation limit, with coherent DMA mappings enabled
[17:04:27] ========== [PASSED] ttm_pool_alloc_basic_dma_addr ==========
[17:04:27] [PASSED] ttm_pool_alloc_order_caching_match
[17:04:27] [PASSED] ttm_pool_alloc_caching_mismatch
[17:04:27] [PASSED] ttm_pool_alloc_order_mismatch
[17:04:27] [PASSED] ttm_pool_free_dma_alloc
[17:04:27] [PASSED] ttm_pool_free_no_dma_alloc
[17:04:27] [PASSED] ttm_pool_fini_basic
[17:04:27] ==================== [PASSED] ttm_pool =====================
[17:04:27] ================ ttm_resource (8 subtests) =================
[17:04:27] ================= ttm_resource_init_basic  =================
[17:04:27] [PASSED] Init resource in TTM_PL_SYSTEM
[17:04:27] [PASSED] Init resource in TTM_PL_VRAM
[17:04:27] [PASSED] Init resource in a private placement
[17:04:27] [PASSED] Init resource in TTM_PL_SYSTEM, set placement flags
[17:04:27] ============= [PASSED] ttm_resource_init_basic =============
[17:04:27] [PASSED] ttm_resource_init_pinned
[17:04:27] [PASSED] ttm_resource_fini_basic
[17:04:27] [PASSED] ttm_resource_manager_init_basic
[17:04:27] [PASSED] ttm_resource_manager_usage_basic
[17:04:27] [PASSED] ttm_resource_manager_set_used_basic
[17:04:27] [PASSED] ttm_sys_man_alloc_basic
[17:04:27] [PASSED] ttm_sys_man_free_basic
[17:04:27] ================== [PASSED] ttm_resource ===================
[17:04:27] =================== ttm_tt (15 subtests) ===================
[17:04:27] ==================== ttm_tt_init_basic  ====================
[17:04:27] [PASSED] Page-aligned size
[17:04:27] [PASSED] Extra pages requested
[17:04:27] ================ [PASSED] ttm_tt_init_basic ================
[17:04:27] [PASSED] ttm_tt_init_misaligned
[17:04:27] [PASSED] ttm_tt_fini_basic
[17:04:27] [PASSED] ttm_tt_fini_sg
[17:04:27] [PASSED] ttm_tt_fini_shmem
[17:04:27] [PASSED] ttm_tt_create_basic
[17:04:27] [PASSED] ttm_tt_create_invalid_bo_type
[17:04:27] [PASSED] ttm_tt_create_ttm_exists
[17:04:27] [PASSED] ttm_tt_create_failed
[17:04:27] [PASSED] ttm_tt_destroy_basic
[17:04:27] [PASSED] ttm_tt_populate_null_ttm
[17:04:27] [PASSED] ttm_tt_populate_populated_ttm
[17:04:27] [PASSED] ttm_tt_unpopulate_basic
[17:04:27] [PASSED] ttm_tt_unpopulate_empty_ttm
[17:04:27] [PASSED] ttm_tt_swapin_basic
[17:04:27] ===================== [PASSED] ttm_tt ======================
[17:04:27] =================== ttm_bo (14 subtests) ===================
[17:04:27] =========== ttm_bo_reserve_optimistic_no_ticket  ===========
[17:04:27] [PASSED] Cannot be interrupted and sleeps
[17:04:27] [PASSED] Cannot be interrupted, locks straight away
[17:04:27] [PASSED] Can be interrupted, sleeps
[17:04:27] ======= [PASSED] ttm_bo_reserve_optimistic_no_ticket =======
[17:04:27] [PASSED] ttm_bo_reserve_locked_no_sleep
[17:04:27] [PASSED] ttm_bo_reserve_no_wait_ticket
[17:04:27] [PASSED] ttm_bo_reserve_double_resv
[17:04:27] [PASSED] ttm_bo_reserve_interrupted
[17:04:27] [PASSED] ttm_bo_reserve_deadlock
[17:04:27] [PASSED] ttm_bo_unreserve_basic
[17:04:27] [PASSED] ttm_bo_unreserve_pinned
[17:04:27] [PASSED] ttm_bo_unreserve_bulk
[17:04:27] [PASSED] ttm_bo_fini_basic
[17:04:27] [PASSED] ttm_bo_fini_shared_resv
[17:04:27] [PASSED] ttm_bo_pin_basic
[17:04:27] [PASSED] ttm_bo_pin_unpin_resource
[17:04:27] [PASSED] ttm_bo_multiple_pin_one_unpin
[17:04:27] ===================== [PASSED] ttm_bo ======================
[17:04:27] ============== ttm_bo_validate (22 subtests) ===============
[17:04:27] ============== ttm_bo_init_reserved_sys_man  ===============
[17:04:27] [PASSED] Buffer object for userspace
[17:04:27] [PASSED] Kernel buffer object
[17:04:27] [PASSED] Shared buffer object
[17:04:27] ========== [PASSED] ttm_bo_init_reserved_sys_man ===========
[17:04:27] ============== ttm_bo_init_reserved_mock_man  ==============
[17:04:27] [PASSED] Buffer object for userspace
[17:04:27] [PASSED] Kernel buffer object
[17:04:27] [PASSED] Shared buffer object
[17:04:27] ========== [PASSED] ttm_bo_init_reserved_mock_man ==========
[17:04:27] [PASSED] ttm_bo_init_reserved_resv
[17:04:27] ================== ttm_bo_validate_basic  ==================
[17:04:27] [PASSED] Buffer object for userspace
[17:04:27] [PASSED] Kernel buffer object
[17:04:27] [PASSED] Shared buffer object
[17:04:27] ============== [PASSED] ttm_bo_validate_basic ==============
[17:04:27] [PASSED] ttm_bo_validate_invalid_placement
[17:04:27] ============= ttm_bo_validate_same_placement  ==============
[17:04:27] [PASSED] System manager
[17:04:27] [PASSED] VRAM manager
[17:04:27] ========= [PASSED] ttm_bo_validate_same_placement ==========
[17:04:27] [PASSED] ttm_bo_validate_failed_alloc
[17:04:27] [PASSED] ttm_bo_validate_pinned
[17:04:27] [PASSED] ttm_bo_validate_busy_placement
[17:04:27] ================ ttm_bo_validate_multihop  =================
[17:04:27] [PASSED] Buffer object for userspace
[17:04:27] [PASSED] Kernel buffer object
[17:04:27] [PASSED] Shared buffer object
[17:04:27] ============ [PASSED] ttm_bo_validate_multihop =============
[17:04:27] ========== ttm_bo_validate_no_placement_signaled  ==========
[17:04:27] [PASSED] Buffer object in system domain, no page vector
[17:04:27] [PASSED] Buffer object in system domain with an existing page vector
[17:04:27] ====== [PASSED] ttm_bo_validate_no_placement_signaled ======
[17:04:27] ======== ttm_bo_validate_no_placement_not_signaled  ========
[17:04:27] [PASSED] Buffer object for userspace
[17:04:27] [PASSED] Kernel buffer object
[17:04:27] [PASSED] Shared buffer object
[17:04:27] ==== [PASSED] ttm_bo_validate_no_placement_not_signaled ====
[17:04:27] [PASSED] ttm_bo_validate_move_fence_signaled
[17:04:27] ========= ttm_bo_validate_move_fence_not_signaled  =========
[17:04:27] [PASSED] Waits for GPU
[17:04:27] [PASSED] Tries to lock straight away
[17:04:27] ===== [PASSED] ttm_bo_validate_move_fence_not_signaled =====
[17:04:27] [PASSED] ttm_bo_validate_swapout
[17:04:27] [PASSED] ttm_bo_validate_happy_evict
[17:04:27] [PASSED] ttm_bo_validate_all_pinned_evict
[17:04:27] [PASSED] ttm_bo_validate_allowed_only_evict
[17:04:27] [PASSED] ttm_bo_validate_deleted_evict
[17:04:27] [PASSED] ttm_bo_validate_busy_domain_evict
[17:04:27] [PASSED] ttm_bo_validate_evict_gutting
[17:04:27] [PASSED] ttm_bo_validate_recrusive_evict
[17:04:27] ================= [PASSED] ttm_bo_validate =================
[17:04:27] ============================================================
[17:04:27] Testing complete. Ran 106 tests: passed: 106
[17:04:27] Elapsed time: 12.162s total, 1.834s configuring, 10.063s building, 0.215s running

+ for kcfg in "${kunitconfigs[@]}"
+ [[ ! -f /kernel/drivers/dma-buf/.kunitconfig ]]
+ /kernel/tools/testing/kunit/kunit.py run --kunitconfig /kernel/drivers/dma-buf/.kunitconfig [17:04:27] Configuring KUnit Kernel ...
Regenerating .config ...
Populating config with:

$ make ARCH=um O=.kunit olddefconfig
[17:04:29] Building KUnit Kernel ...
Populating config with:
$ make ARCH=um O=.kunit olddefconfig
Building with:
$ make all compile_commands.json scripts_gdb ARCH=um O=.kunit --jobs=48
[17:04:37] Starting KUnit Kernel (1/1)...
[17:04:37] ============================================================
Running tests with:
$ .kunit/linux kunit.enable=1 mem=1G console=tty kunit_shutdown=halt
[17:04:37] ============= refcount_interrupt (4 subtests) ==============
[17:04:37] [PASSED] test_single_irq_change
[17:04:37] [PASSED] test_nested_irq_change
[17:04:37] [PASSED] test_multiple_irq_change
[17:04:37] [PASSED] test_irq_save
[17:04:37] =============== [PASSED] refcount_interrupt ================
[17:04:37] =============== dma-buf-fence (12 subtests) ================
[17:04:37] [PASSED] test_sanitycheck
[17:04:37] [PASSED] test_signaling
[17:04:37] [PASSED] test_add_callback
[17:04:37] [PASSED] test_late_add_callback
[17:04:37] [PASSED] test_rm_callback
[17:04:37] [PASSED] test_late_rm_callback
[17:04:37] [PASSED] test_status
[17:04:37] [PASSED] test_error
[17:04:37] [PASSED] test_wait
[17:04:37] [PASSED] test_wait_timeout
[17:04:37] [PASSED] test_stub
[17:04:37] [SKIPPED] test_race_signal_callback (requires at least 2 CPUs)
[17:04:37] ================== [PASSED] dma-buf-fence ==================
[17:04:37] ============ dma-buf-fence-chain (11 subtests) =============
[17:04:37] [PASSED] test_sanitycheck
[17:04:37] [PASSED] test_find_seqno
[17:04:37] [PASSED] test_find_signaled
[17:04:37] [PASSED] test_find_out_of_order
[17:04:42] [PASSED] test_find_gap
[17:04:42] [PASSED] test_find_race
[17:04:42] [PASSED] test_signal_forward
[17:04:43] [PASSED] test_signal_backward
[17:04:43] [PASSED] test_wait_forward
[17:04:43] [PASSED] test_wait_backward
[17:04:43] [PASSED] test_wait_random
[17:04:43] =============== [PASSED] dma-buf-fence-chain ===============
[17:04:43] ============ dma-buf-fence-unwrap (10 subtests) ============
[17:04:43] [PASSED] test_sanitycheck
[17:04:43] [PASSED] test_unwrap_array
[17:04:43] [PASSED] test_unwrap_chain
[17:04:43] [PASSED] test_unwrap_chain_array
[17:04:43] [PASSED] test_unwrap_merge
[17:04:43] [PASSED] test_unwrap_merge_duplicate
[17:04:43] [PASSED] test_unwrap_merge_seqno
[17:04:43] [PASSED] test_unwrap_merge_order
[17:04:43] [PASSED] test_unwrap_merge_complex
[17:04:43] [PASSED] test_unwrap_merge_complex_seqno
[17:04:43] ============== [PASSED] dma-buf-fence-unwrap ===============
[17:04:43] ================ dma-buf-resv (5 subtests) =================
[17:04:43] [PASSED] test_sanitycheck
[17:04:43] ===================== test_signaling  ======================
[17:04:43] [PASSED] kernel
[17:04:43] [PASSED] write
[17:04:43] [PASSED] read
[17:04:43] [PASSED] bookkeep
[17:04:43] ================= [PASSED] test_signaling ==================
[17:04:43] ====================== test_for_each  ======================
[17:04:43] [PASSED] kernel
[17:04:43] [PASSED] write
[17:04:43] [PASSED] read
[17:04:43] [PASSED] bookkeep
[17:04:43] ================== [PASSED] test_for_each ==================
[17:04:43] ================= test_for_each_unlocked  ==================
[17:04:43] [PASSED] kernel
[17:04:43] [PASSED] write
[17:04:43] [PASSED] read
[17:04:43] [PASSED] bookkeep
[17:04:43] ============= [PASSED] test_for_each_unlocked ==============
[17:04:43] ===================== test_get_fences  =====================
[17:04:43] [PASSED] kernel
[17:04:43] [PASSED] write
[17:04:43] [PASSED] read
[17:04:43] [PASSED] bookkeep
[17:04:43] ================= [PASSED] test_get_fences =================
[17:04:43] ================== [PASSED] dma-buf-resv ===================
[17:04:43] ============================================================
[17:04:43] Testing complete. Ran 54 tests: passed: 53, skipped: 1
[17:04:43] Elapsed time: 15.675s total, 1.767s configuring, 8.636s building, 5.260s running

+ cleanup
++ stat -c %u:%g /kernel
+ chown -R 1003:1003 /kernel

#36 lore raw

✓ Xe.CI.BAT: success for Add memory page offlining support (rev24)

In reply to Tejas Upadhyay
== Series Details ==

== Series Details ==

Series: Add memory page offlining support (rev24)
URL : https://patchwork.freedesktop.org/series/161473/
State : success

== Summary ==

CI Bug Log - changes from xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7_BAT -> xe-pw-161473v24_BAT
====================================================

Summary
-------

**SUCCESS**

No regressions found.

Participating hosts (12 -> 12)
------------------------------

No changes in participating hosts

Known issues
------------

Here are the changes found in xe-pw-161473v24_BAT that come from known issues:

### IGT changes ###

#### Possible fixes ####

* igt@xe_waitfence@engine:
- bat-ptl-2: [FAIL][1] -> [PASS][2]
[1]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/bat-ptl-2/igt@[email protected]
[2]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/bat-ptl-2/igt@[email protected]
- bat-wcl-1: [FAIL][3] ([Intel XE#9103]) -> [PASS][4]
[3]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/bat-wcl-1/igt@[email protected]
[4]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/bat-wcl-1/igt@[email protected]

* igt@xe_waitfence@reltime:
- bat-wcl-1: [FAIL][5] -> [PASS][6]
[5]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/bat-wcl-1/igt@[email protected]
[6]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/bat-wcl-1/igt@[email protected]

[Intel XE#9103]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/9103

Build changes
-------------

* Linux: xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7 -> xe-pw-161473v24

IGT_9080: 65c1a091179475c3f0cd925799b49f58720bcf43 @ https://gitlab.freedesktop.org/drm/igt-gpu-tools.git
xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7: 9a6566667a8b712e9b28a1469c282187a6d043c7
xe-pw-161473v24: 161473v24

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/index.html

#37 lore raw

✓ Xe.CI.FULL: success for Add memory page offlining support (rev24)

In reply to Tejas Upadhyay
== Series Details ==

== Series Details ==

Series: Add memory page offlining support (rev24)
URL : https://patchwork.freedesktop.org/series/161473/
State : success

== Summary ==

CI Bug Log - changes from xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7_FULL -> xe-pw-161473v24_FULL
====================================================

Summary
-------

**SUCCESS**

No regressions found.

Participating hosts (2 -> 2)
------------------------------

No changes in participating hosts

Known issues
------------

Here are the changes found in xe-pw-161473v24_FULL that come from known issues:

### IGT changes ###

#### Issues hit ####

* igt@kms_atomic_transition@plane-all-modeset-transition-internal-panels:
- shard-bmg: NOTRUN -> [SKIP][1] ([Intel XE#2370])
[1]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@kms_atomic_transition@plane-all-modeset-transition-internal-panels.html

* igt@kms_big_fb@linear-max-hw-stride-64bpp-rotate-0-hflip:
- shard-bmg: NOTRUN -> [SKIP][2] ([Intel XE#7059] / [Intel XE#7085])
[2]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_big_fb@x-tiled-32bpp-rotate-270:
- shard-bmg: NOTRUN -> [SKIP][3] ([Intel XE#2327])
[3]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_big_fb@y-tiled-addfb:
- shard-bmg: NOTRUN -> [SKIP][4] ([Intel XE#2328] / [Intel XE#7367])
[4]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_big_fb@yf-tiled-16bpp-rotate-180:
- shard-bmg: NOTRUN -> [SKIP][5] ([Intel XE#1124]) +3 other tests skip
[5]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_bw@connected-linear-tiling-3-displays-target-2560x1440p:
- shard-bmg: NOTRUN -> [SKIP][6] ([Intel XE#7679])
[6]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_bw@linear-tiling-1-displays-target-3840x2160p:
- shard-bmg: NOTRUN -> [SKIP][7] ([Intel XE#367])
[7]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_ccs@bad-pixel-format-4-tiled-dg2-mc-ccs:
- shard-bmg: NOTRUN -> [SKIP][8] ([Intel XE#2887]) +4 other tests skip
[8]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_ccs@crc-primary-suspend-4-tiled-bmg-ccs:
- shard-bmg: NOTRUN -> [INCOMPLETE][9] ([Intel XE#7084] / [Intel XE#8150]) +1 other test incomplete
[9]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_chamelium_audio@dp-audio-after-suspend:
- shard-bmg: NOTRUN -> [SKIP][10] ([Intel XE#2252]) +4 other tests skip
[10]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_content_protection@dp-mst-lic-type-1:
- shard-bmg: NOTRUN -> [SKIP][11] ([Intel XE#2390] / [Intel XE#6974])
[11]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_cursor_crc@cursor-random-256x85:
- shard-bmg: NOTRUN -> [SKIP][12] ([Intel XE#2320]) +2 other tests skip
[12]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_cursor_legacy@flip-vs-cursor-legacy:
- shard-bmg: [PASS][13] -> [FAIL][14] ([Intel XE#7809])
[13]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-bmg-3/igt@[email protected]
[14]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_cursor_legacy@short-busy-flip-before-cursor-atomic-transitions:
- shard-bmg: NOTRUN -> [SKIP][15] ([Intel XE#2286] / [Intel XE#6035])
[15]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@kms_cursor_legacy@short-busy-flip-before-cursor-atomic-transitions.html

* igt@kms_dsc@dsc-with-bpc-formats-ultrajoiner:
- shard-bmg: NOTRUN -> [SKIP][16] ([Intel XE#8265]) +1 other test skip
[16]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_flip@flip-vs-expired-vblank-interruptible@c-edp1:
- shard-lnl: [PASS][17] -> [FAIL][18] ([Intel XE#301] / [Intel XE#3149])
[17]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-lnl-5/igt@kms_flip@[email protected]
[18]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-lnl-8/igt@kms_flip@[email protected]

* igt@kms_flip_scaled_crc@flip-32bpp-4tile-to-32bpp-4tiledg2rcccs-downscaling:
- shard-bmg: NOTRUN -> [SKIP][19] ([Intel XE#7178] / [Intel XE#7349])
[19]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@kms_flip_scaled_crc@flip-32bpp-4tile-to-32bpp-4tiledg2rcccs-downscaling.html

* igt@kms_frontbuffer_tracking@drrs-abgr161616f-draw-render:
- shard-bmg: NOTRUN -> [SKIP][20] ([Intel XE#7061] / [Intel XE#7356]) +1 other test skip
[20]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_frontbuffer_tracking@drrs-rgb565-draw-render:
- shard-bmg: NOTRUN -> [SKIP][21] ([Intel XE#2311]) +21 other tests skip
[21]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_frontbuffer_tracking@drrshdr-abgr161616f-draw-blt:
- shard-bmg: NOTRUN -> [SKIP][22] ([Intel XE#7061])
[22]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_frontbuffer_tracking@fbc-rgb565-draw-render:
- shard-bmg: NOTRUN -> [SKIP][23] ([Intel XE#4141]) +7 other tests skip
[23]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_frontbuffer_tracking@fbcpsrhdr-2p-primscrn-cur-indfb-draw-render:
- shard-bmg: NOTRUN -> [SKIP][24] ([Intel XE#2313]) +26 other tests skip
[24]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@kms_frontbuffer_tracking@fbcpsrhdr-2p-primscrn-cur-indfb-draw-render.html

* igt@kms_frontbuffer_tracking@plane-fbc-rte:
- shard-bmg: NOTRUN -> [SKIP][25] ([Intel XE#2350] / [Intel XE#7503])
[25]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_joiner@invalid-modeset-ultra-joiner:
- shard-bmg: NOTRUN -> [SKIP][26] ([Intel XE#6911] / [Intel XE#7378])
[26]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_psr2_sf@psr2-primary-plane-update-sf-dmg-area-big-fb:
- shard-bmg: NOTRUN -> [SKIP][27] ([Intel XE#1489]) +2 other tests skip
[27]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_psr@fbc-pr-sprite-blt:
- shard-bmg: NOTRUN -> [SKIP][28] ([Intel XE#2234] / [Intel XE#2850]) +5 other tests skip
[28]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_rotation_crc@primary-y-tiled-reflect-x-0:
- shard-bmg: NOTRUN -> [SKIP][29] ([Intel XE#2330] / [Intel XE#5813])
[29]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_rotation_crc@primary-yf-tiled-reflect-x-270:
- shard-bmg: NOTRUN -> [SKIP][30] ([Intel XE#3904] / [Intel XE#7342])
[30]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_scaling_modes@scaling-mode-full-aspect:
- shard-bmg: NOTRUN -> [SKIP][31] ([Intel XE#2413])
[31]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_sharpness_filter@invalid-filter-with-scaling-mode:
- shard-bmg: NOTRUN -> [SKIP][32] ([Intel XE#6503]) +1 other test skip
[32]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@kms_vrr@lobf-dc3co:
- shard-bmg: NOTRUN -> [SKIP][33] ([Intel XE#8397])
[33]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@xe_compute@eu-busy-10s:
- shard-bmg: NOTRUN -> [SKIP][34] ([Intel XE#6599]) +1 other test skip
[34]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@xe_evict@evict-cm-threads-small-multi-queue:
- shard-bmg: NOTRUN -> [SKIP][35] ([Intel XE#8370])
[35]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@xe_exec_basic@multigpu-once-bindexecqueue-rebind:
- shard-bmg: NOTRUN -> [SKIP][36] ([Intel XE#2322] / [Intel XE#7372]) +1 other test skip
[36]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@xe_exec_fault_mode@twice-multi-queue-userptr-rebind-prefetch:
- shard-bmg: NOTRUN -> [SKIP][37] ([Intel XE#8374]) +1 other test skip
[37]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@xe_exec_fault_mode@twice-multi-queue-userptr-rebind-prefetch.html

* igt@xe_exec_multi_queue@few-execs-preempt-mode-priority:
- shard-bmg: NOTRUN -> [SKIP][38] ([Intel XE#8364]) +11 other tests skip
[38]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@xe_exec_reset@multi-queue-cat-error-on-secondary:
- shard-bmg: NOTRUN -> [SKIP][39] ([Intel XE#8369]) +2 other tests skip
[39]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@xe_exec_threads@threads-multi-queue-mixed-shared-vm-userptr-invalidate-race:
- shard-bmg: NOTRUN -> [SKIP][40] ([Intel XE#8378]) +3 other tests skip
[40]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@xe_exec_threads@threads-multi-queue-mixed-shared-vm-userptr-invalidate-race.html

* igt@xe_live_ktest@xe_bo@xe_ccs_migrate_kunit:
- shard-bmg: NOTRUN -> [SKIP][41] ([Intel XE#2229])
[41]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@xe_live_ktest@xe_bo@xe_ccs_migrate_kunit.html

* igt@xe_mmap@small-bar:
- shard-bmg: NOTRUN -> [SKIP][42] ([Intel XE#586] / [Intel XE#7323] / [Intel XE#7384])
[42]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@xe_multigpu_svm@mgpu-xgpu-access-prefetch:
- shard-bmg: NOTRUN -> [SKIP][43] ([Intel XE#6964])
[43]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@xe_page_reclaim@prl-invalidate-full:
- shard-bmg: NOTRUN -> [SKIP][44] ([Intel XE#7793])
[44]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@xe_pm@s2idle-d3cold-basic-exec:
- shard-bmg: NOTRUN -> [SKIP][45] ([Intel XE#2284] / [Intel XE#7370])
[45]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@xe_pxp@pxp-stale-bo-exec-post-rpm:
- shard-bmg: NOTRUN -> [SKIP][46] ([Intel XE#4733] / [Intel XE#7417])
[46]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-3/igt@[email protected]

* igt@xe_wedged@basic-wedged:
- shard-bmg: NOTRUN -> [DMESG-WARN][47] ([Intel XE#8963])
[47]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

#### Possible fixes ####

* igt@kms_hdr@invalid-hdr:
- shard-bmg: [SKIP][48] ([Intel XE#1503]) -> [PASS][49]
[48]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-bmg-9/igt@[email protected]
[49]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-7/igt@[email protected]

* igt@kms_pm_rpm@system-suspend-idle:
- shard-bmg: [ABORT][50] ([Intel XE#9093]) -> [PASS][51]
[50]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-bmg-6/igt@[email protected]
[51]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-4/igt@[email protected]

* igt@kms_setmode@basic:
- shard-bmg: [FAIL][52] ([Intel XE#8618]) -> [PASS][53] +4 other tests pass
[52]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-bmg-9/igt@[email protected]
[53]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-10/igt@[email protected]

* igt@kms_setmode@basic@pipe-a-dp-2:
- shard-bmg: [FAIL][54] -> [PASS][55] +1 other test pass
[54]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-bmg-9/igt@kms_setmode@[email protected]
[55]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-10/igt@kms_setmode@[email protected]

* igt@xe_pmu@engine-activity-accuracy-50:
- shard-bmg: [FAIL][56] ([Intel XE#8555]) -> [PASS][57] +1 other test pass
[56]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-bmg-4/igt@[email protected]
[57]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-2/igt@[email protected]

#### Warnings ####

* igt@kms_flip@flip-vs-expired-vblank-interruptible:
- shard-lnl: [FAIL][58] ([Intel XE#301]) -> [FAIL][59] ([Intel XE#301] / [Intel XE#3149])
[58]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-lnl-5/igt@[email protected]
[59]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-lnl-8/igt@[email protected]

* igt@kms_hdr@brightness-with-hdr:
- shard-bmg: [SKIP][60] ([Intel XE#3544]) -> [SKIP][61] ([Intel XE#3374] / [Intel XE#3544])
[60]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-bmg-3/igt@[email protected]
[61]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-2/igt@[email protected]

* igt@kms_tiled_display@basic-test-pattern:
- shard-bmg: [SKIP][62] ([Intel XE#2426] / [Intel XE#5848]) -> [FAIL][63] ([Intel XE#1729] / [Intel XE#7424])
[62]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7/shard-bmg-8/igt@[email protected]
[63]: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/shard-bmg-10/igt@[email protected]

[Intel XE#1124]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/1124
[Intel XE#1489]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/1489
[Intel XE#1503]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/1503
[Intel XE#1729]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/1729
[Intel XE#2229]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2229
[Intel XE#2234]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2234
[Intel XE#2252]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2252
[Intel XE#2284]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2284
[Intel XE#2286]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2286
[Intel XE#2311]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2311
[Intel XE#2313]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2313
[Intel XE#2320]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2320
[Intel XE#2322]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2322
[Intel XE#2327]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2327
[Intel XE#2328]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2328
[Intel XE#2330]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2330
[Intel XE#2350]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2350
[Intel XE#2370]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2370
[Intel XE#2390]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2390
[Intel XE#2413]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2413
[Intel XE#2426]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2426
[Intel XE#2850]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2850
[Intel XE#2887]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/2887
[Intel XE#301]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/301
[Intel XE#3149]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/3149
[Intel XE#3374]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/3374
[Intel XE#3544]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/3544
[Intel XE#367]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/367
[Intel XE#3904]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/3904
[Intel XE#4141]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/4141
[Intel XE#4733]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/4733
[Intel XE#5813]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/5813
[Intel XE#5848]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/5848
[Intel XE#586]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/586
[Intel XE#6035]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/6035
[Intel XE#6503]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/6503
[Intel XE#6599]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/6599
[Intel XE#6911]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/6911
[Intel XE#6964]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/6964
[Intel XE#6974]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/6974
[Intel XE#7059]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7059
[Intel XE#7061]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7061
[Intel XE#7084]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7084
[Intel XE#7085]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7085
[Intel XE#7178]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7178
[Intel XE#7323]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7323
[Intel XE#7342]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7342
[Intel XE#7349]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7349
[Intel XE#7356]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7356
[Intel XE#7367]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7367
[Intel XE#7370]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7370
[Intel XE#7372]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7372
[Intel XE#7378]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7378
[Intel XE#7384]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7384
[Intel XE#7417]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7417
[Intel XE#7424]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7424
[Intel XE#7503]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7503
[Intel XE#7679]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7679
[Intel XE#7793]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7793
[Intel XE#7809]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/7809
[Intel XE#8150]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8150
[Intel XE#8265]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8265
[Intel XE#8364]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8364
[Intel XE#8369]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8369
[Intel XE#8370]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8370
[Intel XE#8374]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8374
[Intel XE#8378]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8378
[Intel XE#8397]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8397
[Intel XE#8555]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8555
[Intel XE#8618]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8618
[Intel XE#8963]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/8963
[Intel XE#9093]: https://gitlab.freedesktop.org/drm/xe/kernel/issues/9093

Build changes
-------------

* Linux: xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7 -> xe-pw-161473v24

IGT_9080: 65c1a091179475c3f0cd925799b49f58720bcf43 @ https://gitlab.freedesktop.org/drm/igt-gpu-tools.git
xe-5676-9a6566667a8b712e9b28a1469c282187a6d043c7: 9a6566667a8b712e9b28a1469c282187a6d043c7
xe-pw-161473v24: 161473v24

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/intel-xe/xe-pw-161473v24/index.html