[PATCH v2] cxl/region: Fix self-deadlock when cxl_memdev_attach_region() fails
From: Junho Cha via B4 Relay
Date: Wed Nov 18 2026 - 20:44:43 EST
From: Junho Cha <junho.cha@xxxxxxxxx>
cxl_memdev_attach_region() holds cxl_rwsem.region for read while it
registers endpoint_unregister_region() with devm_add_action_or_reset().
If registering the action fails, devm_add_action_or_reset() runs the
action at once, which reaches cxl_decoder_detach() through
unregister_region() and detach_target():
cxl_memdev_attach_region()
down_read(&cxl_rwsem.region)
devm_add_action_or_reset() /* fails with -ENOMEM */
endpoint_unregister_region()
unregister_region()
detach_target()
cxl_decoder_detach()
down_write_killable(&cxl_rwsem.region)
The task then waits for a write lock while holding the same rwsem for
read, and never makes progress. Before it hangs, lockdep also reports
a circular dependency: endpoint_unregister_region() takes
cxlrd->regions_lock under cxl_rwsem.dpa, while cxl_add_to_region()
takes cxlrd->regions_lock before both rwsems.
Register the action with devm_add_action() instead, still under both
rwsems, and if that fails, run it only after the rwsems are dropped.
The endpoint device lock stays held throughout, as before.
Reproduced with cxl_test (type2_test=1) by failing the devres
allocation through failslab, with the stack filter limited to
cxl_memdev_attach_region(). Without this patch the probing task hangs
in cxl_decoder_detach(). With it, the cxl_mem probe fails with
-ENOMEM, the region is unregistered, and lockdep stays quiet.
Fixes: d8dcb0b74b04 ("cxl/region: Introduce devm_cxl_probe_mem()")
Reported-by: Sashiko <sashiko-bot@xxxxxxxxxx>
Link: https://lore.kernel.org/all/20260805080537.7489E1F000E9@xxxxxxxxxxxxxxx/
Assisted-by: LLM
Signed-off-by: Junho Cha <junho.cha@xxxxxxxxx>
---
Dave pointed at this reset path while reviewing the Type2 multipf
series [1]. This is a standalone fix for the existing code that keeps
the region teardown when registering the action fails.
[1] https://lore.kernel.org/linux-cxl/6bc33514-8bfb-44d6-8fde-28f45dff5eb9@xxxxxxxxx/
---
Changes in v2:
- Register the action under both rwsems as before and only run the
unwind after dropping them. v1 left a window where the region could
be detached before the action was registered (Ben).
- Add Reported-by/Link for Sashiko.
- Drop Alejandro's Reviewed-by, since the fix changed.
- Link to v1: https://patch.msgid.link/20261117-junhomax-fix-cxl-attach-region-deadlock-v1-1-f373a135edd2@xxxxxxxxx
---
drivers/cxl/core/region.c | 32 ++++++++++++++++++++++++--------
1 file changed, 24 insertions(+), 8 deletions(-)
diff --git a/drivers/cxl/core/region.c b/drivers/cxl/core/region.c
index 237172b33f55..f9538ac98a1f 100644
--- a/drivers/cxl/core/region.c
+++ b/drivers/cxl/core/region.c
@@ -4437,7 +4437,8 @@ static int first_mapped_decoder(struct device *dev, const void *data)
* Runs in cxl_mem_probe context after successful endpoint probe, assumes the
* simple case of single mapped decoder per memdev.
*/
-int cxl_memdev_attach_region(struct cxl_memdev *cxlmd)
+static int __cxl_memdev_attach_region(struct cxl_memdev *cxlmd,
+ struct cxl_region **unwind)
{
struct cxl_attach_region *attach =
container_of(cxlmd->attach, typeof(*attach), attach);
@@ -4446,10 +4447,6 @@ int cxl_memdev_attach_region(struct cxl_memdev *cxlmd)
struct cxl_region *cxlr;
int rc;
- /* hold endpoint lock to setup autoremove of the region */
- guard(device)(&endpoint->dev);
- if (!endpoint->dev.driver)
- return -ENXIO;
guard(rwsem_read)(&cxl_rwsem.region);
guard(rwsem_read)(&cxl_rwsem.dpa);
@@ -4484,10 +4481,11 @@ int cxl_memdev_attach_region(struct cxl_memdev *cxlmd)
/* Only teardown regions that pass validation, ignore the rest */
get_device(&cxlr->dev);
- rc = devm_add_action_or_reset(&endpoint->dev,
- endpoint_unregister_region, cxlr);
- if (rc)
+ rc = devm_add_action(&endpoint->dev, endpoint_unregister_region, cxlr);
+ if (rc) {
+ *unwind = cxlr;
return rc;
+ }
attach->hpa_range = (struct range) {
.start = cxlr->params.res->start,
@@ -4495,6 +4493,24 @@ int cxl_memdev_attach_region(struct cxl_memdev *cxlmd)
};
return 0;
}
+
+int cxl_memdev_attach_region(struct cxl_memdev *cxlmd)
+{
+ struct cxl_port *endpoint = cxlmd->endpoint;
+ struct cxl_region *unwind = NULL;
+ int rc;
+
+ /* hold endpoint lock to setup autoremove of the region */
+ guard(device)(&endpoint->dev);
+ if (!endpoint->dev.driver)
+ return -ENXIO;
+
+ /* Detaching the region takes cxl_rwsem.region, so unwind without it */
+ rc = __cxl_memdev_attach_region(cxlmd, &unwind);
+ if (unwind)
+ endpoint_unregister_region(unwind);
+ return rc;
+}
EXPORT_SYMBOL_FOR_MODULES(cxl_memdev_attach_region, "cxl_mem");
/*
---
base-commit: 32af7819352bae9c690ebe277a44eceb501d5c68
change-id: 20261113-junhomax-fix-cxl-attach-region-deadlock-fc0a425f7265
Best regards,
--
Junho Cha <junho.cha@xxxxxxxxx>