* [PATCH 6.12.y 2/3] efi/cper, cxl: Make definitions and structures global
2026-09-08 1:21 ` [PATCH 6.12.y 1/3] efi/cper, cxl: Prefix protocol error struct and function names with cxl_ Sasha Levin
@ 2026-09-08 1:21 ` Sasha Levin
2026-09-08 1:21 ` [PATCH 6.12.y 3/3] acpi/apei/ghes: Use raw_spinlock_t for CXL CPER work locks Sasha Levin
1 sibling, 0 replies; 4+ messages in thread
From: Sasha Levin @ 2026-09-08 1:21 UTC (permalink / raw)
To: stable
Cc: Smita Koralahalli, Jonathan Cameron, Ira Weiny, Dave Jiang,
Fan Ni, Gregory Price, Dan Williams, Sasha Levin
From: Smita Koralahalli <Smita.KoralahalliChannabasappa@amd.com>
[ Upstream commit 958c3a6706863f9f77b21c90d2428473441cd8a1 ]
In preparation to add tracepoint support, move protocol error UUID
definition to a common location, Also, make struct CXL RAS capability,
cxl_cper_sec_prot_err and CPER validation flags global for use across
different modules.
Signed-off-by: Smita Koralahalli <Smita.KoralahalliChannabasappa@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Gregory Price <gourry@gourry.net>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Link: https://patch.msgid.link/20250123084421.127697-3-Smita.KoralahalliChannabasappa@amd.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Stable-dep-of: 6625ca499c31 ("acpi/apei/ghes: Use raw_spinlock_t for CXL CPER work locks")
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
drivers/firmware/efi/cper.c | 1 +
drivers/firmware/efi/cper_cxl.c | 35 +--------------
drivers/firmware/efi/cper_cxl.h | 51 ---------------------
include/cxl/event.h | 80 +++++++++++++++++++++++++++++++++
include/linux/cper.h | 4 ++
5 files changed, 86 insertions(+), 85 deletions(-)
diff --git a/drivers/firmware/efi/cper.c b/drivers/firmware/efi/cper.c
index e0adbd6df9b53..7fe10a06e68c3 100644
--- a/drivers/firmware/efi/cper.c
+++ b/drivers/firmware/efi/cper.c
@@ -25,6 +25,7 @@
#include <linux/bcd.h>
#include <acpi/ghes.h>
#include <ras/ras_event.h>
+#include <cxl/event.h>
#include "cper_cxl.h"
/*
diff --git a/drivers/firmware/efi/cper_cxl.c b/drivers/firmware/efi/cper_cxl.c
index cbaabcb7382d9..64c0dd27be6ed 100644
--- a/drivers/firmware/efi/cper_cxl.c
+++ b/drivers/firmware/efi/cper_cxl.c
@@ -8,27 +8,9 @@
*/
#include <linux/cper.h>
+#include <cxl/event.h>
#include "cper_cxl.h"
-#define PROT_ERR_VALID_AGENT_TYPE BIT_ULL(0)
-#define PROT_ERR_VALID_AGENT_ADDRESS BIT_ULL(1)
-#define PROT_ERR_VALID_DEVICE_ID BIT_ULL(2)
-#define PROT_ERR_VALID_SERIAL_NUMBER BIT_ULL(3)
-#define PROT_ERR_VALID_CAPABILITY BIT_ULL(4)
-#define PROT_ERR_VALID_DVSEC BIT_ULL(5)
-#define PROT_ERR_VALID_ERROR_LOG BIT_ULL(6)
-
-/* CXL RAS Capability Structure, CXL v3.0 sec 8.2.4.16 */
-struct cxl_ras_capability_regs {
- u32 uncor_status;
- u32 uncor_mask;
- u32 uncor_severity;
- u32 cor_status;
- u32 cor_mask;
- u32 cap_control;
- u32 header_log[16];
-};
-
static const char * const prot_err_agent_type_strs[] = {
"Restricted CXL Device",
"Restricted CXL Host Downstream Port",
@@ -40,21 +22,6 @@ static const char * const prot_err_agent_type_strs[] = {
"CXL Upstream Switch Port",
};
-/*
- * The layout of the enumeration and the values matches CXL Agent Type
- * field in the UEFI 2.10 Section N.2.13,
- */
-enum {
- RCD, /* Restricted CXL Device */
- RCH_DP, /* Restricted CXL Host Downstream Port */
- DEVICE, /* CXL Device */
- LD, /* CXL Logical Device */
- FMLD, /* CXL Fabric Manager managed Logical Device */
- RP, /* CXL Root Port */
- DSP, /* CXL Downstream Switch Port */
- USP, /* CXL Upstream Switch Port */
-};
-
void cxl_cper_print_prot_err(const char *pfx,
const struct cxl_cper_sec_prot_err *prot_err)
{
diff --git a/drivers/firmware/efi/cper_cxl.h b/drivers/firmware/efi/cper_cxl.h
index 0e3ab0ba17c36..5ce1401ee17af 100644
--- a/drivers/firmware/efi/cper_cxl.h
+++ b/drivers/firmware/efi/cper_cxl.h
@@ -10,57 +10,6 @@
#ifndef LINUX_CPER_CXL_H
#define LINUX_CPER_CXL_H
-/* CXL Protocol Error Section */
-#define CPER_SEC_CXL_PROT_ERR \
- GUID_INIT(0x80B9EFB4, 0x52B5, 0x4DE3, 0xA7, 0x77, 0x68, 0x78, \
- 0x4B, 0x77, 0x10, 0x48)
-
-#pragma pack(1)
-
-/* Compute Express Link Protocol Error Section, UEFI v2.10 sec N.2.13 */
-struct cxl_cper_sec_prot_err {
- u64 valid_bits;
- u8 agent_type;
- u8 reserved[7];
-
- /*
- * Except for RCH Downstream Port, all the remaining CXL Agent
- * types are uniquely identified by the PCIe compatible SBDF number.
- */
- union {
- u64 rcrb_base_addr;
- struct {
- u8 function;
- u8 device;
- u8 bus;
- u16 segment;
- u8 reserved_1[3];
- };
- } agent_addr;
-
- struct {
- u16 vendor_id;
- u16 device_id;
- u16 subsystem_vendor_id;
- u16 subsystem_id;
- u8 class_code[2];
- u16 slot;
- u8 reserved_1[4];
- } device_id;
-
- struct {
- u32 lower_dw;
- u32 upper_dw;
- } dev_serial_num;
-
- u8 capability[60];
- u16 dvsec_len;
- u16 err_len;
- u8 reserved_2[4];
-};
-
-#pragma pack()
-
void cxl_cper_print_prot_err(const char *pfx,
const struct cxl_cper_sec_prot_err *prot_err);
diff --git a/include/cxl/event.h b/include/cxl/event.h
index 0bea1afbd747c..66d85fc87701d 100644
--- a/include/cxl/event.h
+++ b/include/cxl/event.h
@@ -152,6 +152,86 @@ struct cxl_cper_work_data {
struct cxl_cper_event_rec rec;
};
+#define PROT_ERR_VALID_AGENT_TYPE BIT_ULL(0)
+#define PROT_ERR_VALID_AGENT_ADDRESS BIT_ULL(1)
+#define PROT_ERR_VALID_DEVICE_ID BIT_ULL(2)
+#define PROT_ERR_VALID_SERIAL_NUMBER BIT_ULL(3)
+#define PROT_ERR_VALID_CAPABILITY BIT_ULL(4)
+#define PROT_ERR_VALID_DVSEC BIT_ULL(5)
+#define PROT_ERR_VALID_ERROR_LOG BIT_ULL(6)
+
+/*
+ * The layout of the enumeration and the values matches CXL Agent Type
+ * field in the UEFI 2.10 Section N.2.13,
+ */
+enum {
+ RCD, /* Restricted CXL Device */
+ RCH_DP, /* Restricted CXL Host Downstream Port */
+ DEVICE, /* CXL Device */
+ LD, /* CXL Logical Device */
+ FMLD, /* CXL Fabric Manager managed Logical Device */
+ RP, /* CXL Root Port */
+ DSP, /* CXL Downstream Switch Port */
+ USP, /* CXL Upstream Switch Port */
+};
+
+#pragma pack(1)
+
+/* Compute Express Link Protocol Error Section, UEFI v2.10 sec N.2.13 */
+struct cxl_cper_sec_prot_err {
+ u64 valid_bits;
+ u8 agent_type;
+ u8 reserved[7];
+
+ /*
+ * Except for RCH Downstream Port, all the remaining CXL Agent
+ * types are uniquely identified by the PCIe compatible SBDF number.
+ */
+ union {
+ u64 rcrb_base_addr;
+ struct {
+ u8 function;
+ u8 device;
+ u8 bus;
+ u16 segment;
+ u8 reserved_1[3];
+ };
+ } agent_addr;
+
+ struct {
+ u16 vendor_id;
+ u16 device_id;
+ u16 subsystem_vendor_id;
+ u16 subsystem_id;
+ u8 class_code[2];
+ u16 slot;
+ u8 reserved_1[4];
+ } device_id;
+
+ struct {
+ u32 lower_dw;
+ u32 upper_dw;
+ } dev_serial_num;
+
+ u8 capability[60];
+ u16 dvsec_len;
+ u16 err_len;
+ u8 reserved_2[4];
+};
+
+#pragma pack()
+
+/* CXL RAS Capability Structure, CXL v3.0 sec 8.2.4.16 */
+struct cxl_ras_capability_regs {
+ u32 uncor_status;
+ u32 uncor_mask;
+ u32 uncor_severity;
+ u32 cor_status;
+ u32 cor_mask;
+ u32 cap_control;
+ u32 header_log[16];
+};
+
#ifdef CONFIG_ACPI_APEI_GHES
int cxl_cper_register_work(struct work_struct *work);
int cxl_cper_unregister_work(struct work_struct *work);
diff --git a/include/linux/cper.h b/include/linux/cper.h
index 951291fa50d4f..e9c0331fe204a 100644
--- a/include/linux/cper.h
+++ b/include/linux/cper.h
@@ -89,6 +89,10 @@ enum {
#define CPER_NOTIFY_DMAR \
GUID_INIT(0x667DD791, 0xC6B3, 0x4c27, 0x8A, 0x6B, 0x0F, 0x8E, \
0x72, 0x2D, 0xEB, 0x41)
+/* CXL Protocol Error Section */
+#define CPER_SEC_CXL_PROT_ERR \
+ GUID_INIT(0x80B9EFB4, 0x52B5, 0x4DE3, 0xA7, 0x77, 0x68, 0x78, \
+ 0x4B, 0x77, 0x10, 0x48)
/* CXL Event record UUIDs are formatted as GUIDs and reported in section type */
/*
--
2.53.0
^ permalink raw reply related [flat|nested] 4+ messages in thread* [PATCH 6.12.y 3/3] acpi/apei/ghes: Use raw_spinlock_t for CXL CPER work locks
2026-09-08 1:21 ` [PATCH 6.12.y 1/3] efi/cper, cxl: Prefix protocol error struct and function names with cxl_ Sasha Levin
2026-09-08 1:21 ` [PATCH 6.12.y 2/3] efi/cper, cxl: Make definitions and structures global Sasha Levin
@ 2026-09-08 1:21 ` Sasha Levin
1 sibling, 0 replies; 4+ messages in thread
From: Sasha Levin @ 2026-09-08 1:21 UTC (permalink / raw)
To: stable
Cc: Terry Bowman, Sashiko, Dave Jiang, Jonathan Cameron, Tony Luck,
Sasha Levin
From: Terry Bowman <terry.bowman@amd.com>
[ Upstream commit 6625ca499c3131ef63be3215f8f942d7a097ea3a ]
The CXL CPER work registration and unregistration helpers acquire
cxl_cper_work_lock and cxl_cper_prot_err_work_lock with a spinlock
guard(), which leaves local interrupts enabled. The corresponding post
paths (cxl_cper_post_event(), cxl_cper_post_prot_err()) execute in hard
IRQ context (they are called from the GHES error notification path) and
acquire the same locks with an irqsave guard().
If a CPU is holding one of these locks via a spinlock guard() when a GHES
interrupt arrives on the same CPU, the IRQ handler spins on the held lock
waiting for it to release, while the lock holder is preempted by the IRQ.
The result is a deadlock.
Convert both locks from spinlock_t to raw_spinlock_t and use guard() at
all call sites. On PREEMPT_RT kernels spinlock_t is backed by rt_mutex and
sleeping from hard IRQ context is not permitted; raw_spinlock_t is safe in
both contexts.
Add WARN_ONCE to both register functions to surface double-registration
bugs at runtime.
Restructure both unregister functions to clear the global work pointer
under the lock before calling cancel_work_sync(), closing the window
where a CPER interrupt could schedule work on a pointer about to be
freed. Add kfifo_reset() after cancel_work_sync() so stale entries
are not replayed on next module load.
Both kfifos are single-consumer: only one work_struct is registered at
a time, enforced by the WARN_ONCE guard in the register functions.
kfifo_reset() is safe outside the lock because cancel_work_sync() has
already quiesced the consumer, and no new consumer can register until
the current module exit completes and a fresh module init runs.
Remove the redundant cancel_work_sync() call from cxl_ras_exit() and
cxl_pci_driver_exit(). The CPER unregister functions now quiesce
the work internally.
Reported-by: Sashiko <sashiko@linuxfoundation.org>
Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Fixes: 5e4a264bf8b5 ("acpi/ghes: Process CXL Component Events")
Fixes: 36f257e3b0ba ("acpi/ghes, cxl/pci: Process CXL CPER Protocol Errors")
Cc: stable@vger.kernel.org
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <jonathan.cameron@oss.qualcomm.com>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Link: https://patch.msgid.link/20260803221810.3685703-4-terry.bowman@amd.com
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
[ omitted protocol-error changes because the corresponding consumer infrastructure is absent in 6.12 ]
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
drivers/acpi/apei/ghes.c | 24 ++++++++++++++++--------
drivers/cxl/pci.c | 1 -
2 files changed, 16 insertions(+), 9 deletions(-)
diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 25163eb9b8053..e1cbc35919cbf 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -710,7 +710,7 @@ static void ghes_defer_non_standard_event(struct acpi_hest_generic_data *gdata,
DEFINE_KFIFO(cxl_cper_fifo, struct cxl_cper_work_data, CXL_CPER_FIFO_DEPTH);
/* Synchronize schedule_work() with cxl_cper_work changes */
-static DEFINE_SPINLOCK(cxl_cper_work_lock);
+static DEFINE_RAW_SPINLOCK(cxl_cper_work_lock);
struct work_struct *cxl_cper_work;
static void cxl_cper_post_event(enum cxl_event_type event_type,
@@ -730,7 +730,7 @@ static void cxl_cper_post_event(enum cxl_event_type event_type,
return;
}
- guard(spinlock_irqsave)(&cxl_cper_work_lock);
+ guard(raw_spinlock_irqsave)(&cxl_cper_work_lock);
if (!cxl_cper_work)
return;
@@ -748,10 +748,11 @@ static void cxl_cper_post_event(enum cxl_event_type event_type,
int cxl_cper_register_work(struct work_struct *work)
{
- if (cxl_cper_work)
+ guard(raw_spinlock_irqsave)(&cxl_cper_work_lock);
+ if (WARN_ONCE(cxl_cper_work,
+ "CXL CPER kfifo consumer already registered\n"))
return -EINVAL;
- guard(spinlock)(&cxl_cper_work_lock);
cxl_cper_work = work;
return 0;
}
@@ -759,11 +760,18 @@ EXPORT_SYMBOL_NS_GPL(cxl_cper_register_work, CXL);
int cxl_cper_unregister_work(struct work_struct *work)
{
- if (cxl_cper_work != work)
- return -EINVAL;
+ scoped_guard(raw_spinlock_irqsave, &cxl_cper_work_lock) {
+ if (WARN_ONCE(cxl_cper_work != work,
+ "CXL CPER kfifo consumer mismatch on unregister\n"))
+ return -EINVAL;
+ cxl_cper_work = NULL;
+ }
+
+ cancel_work_sync(work);
+
+ /* Discard stale entries so they are not replayed on next module load */
+ kfifo_reset(&cxl_cper_fifo);
- guard(spinlock)(&cxl_cper_work_lock);
- cxl_cper_work = NULL;
return 0;
}
EXPORT_SYMBOL_NS_GPL(cxl_cper_unregister_work, CXL);
diff --git a/drivers/cxl/pci.c b/drivers/cxl/pci.c
index 09a2bc817d645..4c307ee6cd28e 100644
--- a/drivers/cxl/pci.c
+++ b/drivers/cxl/pci.c
@@ -1087,7 +1087,6 @@ static int __init cxl_pci_driver_init(void)
static void __exit cxl_pci_driver_exit(void)
{
cxl_cper_unregister_work(&cxl_cper_work);
- cancel_work_sync(&cxl_cper_work);
pci_unregister_driver(&cxl_pci_driver);
}
--
2.53.0
^ permalink raw reply related [flat|nested] 4+ messages in thread