Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Zhanjun Dong <zhanjun.dong@intel.com>
To: intel-xe@lists.freedesktop.org
Cc: Wajdeczko@freedesktop.org, Michal <Michal.Wajdeczko@intel.com>,
	Matthew Brost <matthew.brost@intel.com>,
	stable@vger.kernel.org
Subject: [PATCH v7 7/7] drm/xe: Open-code GGTT MMIO access protection
Date: Thu, 19 Feb 2026 13:07:01 -0500	[thread overview]
Message-ID: <20260219180701.2418453-8-zhanjun.dong@intel.com> (raw)
In-Reply-To: <20260219180701.2418453-1-zhanjun.dong@intel.com>

From: Matthew Brost <matthew.brost@intel.com>

GGTT MMIO access is currently protected by hotplug (drm_dev_enter),
which works correctly when the driver loads successfully and is later
unbound or unloaded. However, if driver load fails, this protection is
insufficient because drm_dev_unplug() is never called.

Additionally, devm release functions cannot guarantee that all BOs with
GGTT mappings are destroyed before the GGTT MMIO region is removed, as
some BOs may be freed asynchronously by worker threads.

To address this, introduce an open-coded flag, protected by the GGTT
lock, that guards GGTT MMIO access. The flag is cleared during the
dev_fini_ggtt devm release function to ensure MMIO access is disabled
once teardown begins.

Cc: stable@vger.kernel.org
Fixes: 919bb54e989c ("drm/xe: Fix missing runtime outer protection for ggtt_remove_node")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
---
 drivers/gpu/drm/xe/xe_ggtt.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/xe/xe_ggtt.c b/drivers/gpu/drm/xe/xe_ggtt.c
index 79310f565fe3..cf3311f5d140 100644
--- a/drivers/gpu/drm/xe/xe_ggtt.c
+++ b/drivers/gpu/drm/xe/xe_ggtt.c
@@ -66,6 +66,9 @@
  * give us the correct placement for free.
  */
 
+#define XE_GGTT_FLAGS_64K	BIT(0)
+#define XE_GGTT_FLAGS_ONLINE	BIT(1)
+
 /**
  * struct xe_ggtt_node - A node in GGTT.
  *
@@ -117,6 +120,8 @@ struct xe_ggtt {
 	 * @flags: Flags for this GGTT
 	 * Acceptable flags:
 	 * - %XE_GGTT_FLAGS_64K - if PTE size is 64K. Otherwise, regular is 4K.
+	 * - %XE_GGTT_FLAGS_ONLINE - is GGTT online, protected by ggtt->lock
+	 *   after init
 	 */
 	unsigned int flags;
 	/** @scratch: Internal object allocation used as a scratch page */
@@ -367,6 +372,8 @@ static void dev_fini_ggtt(void *arg)
 {
 	struct xe_ggtt *ggtt = arg;
 
+	scoped_guard(mutex, &ggtt->lock)
+		ggtt->flags &= ~XE_GGTT_FLAGS_ONLINE;
 	drain_workqueue(ggtt->wq);
 }
 
@@ -437,6 +444,7 @@ int xe_ggtt_init_early(struct xe_ggtt *ggtt)
 	if (err)
 		return err;
 
+	ggtt->flags |= XE_GGTT_FLAGS_ONLINE;
 	return devm_add_action_or_reset(xe->drm.dev, dev_fini_ggtt, ggtt);
 }
 ALLOW_ERROR_INJECTION(xe_ggtt_init_early, ERRNO); /* See xe_pci_probe() */
@@ -465,13 +473,10 @@ static void ggtt_node_fini(struct xe_ggtt_node *node)
 static void ggtt_node_remove(struct xe_ggtt_node *node)
 {
 	struct xe_ggtt *ggtt = node->ggtt;
-	struct xe_device *xe = tile_to_xe(ggtt->tile);
 	bool bound;
-	int idx;
-
-	bound = drm_dev_enter(&xe->drm, &idx);
 
 	mutex_lock(&ggtt->lock);
+	bound = ggtt->flags & XE_GGTT_FLAGS_ONLINE;
 	if (bound)
 		xe_ggtt_clear(ggtt, xe_ggtt_node_addr(node), xe_ggtt_node_size(node));
 	drm_mm_remove_node(&node->base);
@@ -484,8 +489,6 @@ static void ggtt_node_remove(struct xe_ggtt_node *node)
 	if (node->invalidate_on_remove)
 		xe_ggtt_invalidate(ggtt);
 
-	drm_dev_exit(idx);
-
 free_node:
 	ggtt_node_fini(node);
 }
-- 
2.34.1


  parent reply	other threads:[~2026-02-19 18:07 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-02-19 18:06 [PATCH v7 0/7] Attempt to fixup reset, wedge, unload corner cases Zhanjun Dong
2026-02-19 18:06 ` [PATCH v7 1/7] drm/xe: Always kill exec queues in xe_guc_submit_pause_abort Zhanjun Dong
2026-02-19 18:06 ` [PATCH v7 2/7] drm/xe: Forcefully tear down exec queues in GuC submit fini Zhanjun Dong
2026-02-20 20:36   ` Michal Wajdeczko
2026-02-20 23:36     ` Matthew Brost
2026-02-19 18:06 ` [PATCH v7 3/7] drm/xe: Trigger queue cleanup if not in wedged mode 2 Zhanjun Dong
2026-02-19 18:06 ` [PATCH v7 4/7] drm/xe: Use XE_WEDGED_MODE_UPON_ANY_HANG_NO_RESET enum instead of magic number Zhanjun Dong
2026-02-20 18:12   ` Matthew Brost
2026-02-19 18:06 ` [PATCH v7 5/7] drm/xe/guc: Ensure CT state transitions via STOP before DISABLED Zhanjun Dong
2026-02-19 18:07 ` [PATCH v7 6/7] drm/xe/uc: Drop xe_guc_sanitize in favor of managed cleanup Zhanjun Dong
2026-02-20 20:40   ` Michal Wajdeczko
2026-02-19 18:07 ` Zhanjun Dong [this message]
2026-02-19 18:34 ` ✓ CI.KUnit: success for Attempt to fixup reset, wedge, unload corner cases Patchwork
2026-02-20  8:37 ` ✓ Xe.CI.BAT: " Patchwork
2026-02-20 12:59 ` ✗ Xe.CI.FULL: failure " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260219180701.2418453-8-zhanjun.dong@intel.com \
    --to=zhanjun.dong@intel.com \
    --cc=Michal.Wajdeczko@intel.com \
    --cc=Wajdeczko@freedesktop.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=matthew.brost@intel.com \
    --cc=stable@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox