[PATCH v7 13/16] arm64: ras: Introduce ras inject interface
Ruidong Tian
tianruidong at linux.alibaba.com
Tue Jun 2 00:15:36 PDT 2026
End-to-end validation of the RAS driver requires the ability to
generate errors on demand. Driver gives two complementary paths,
which the inject interface deliberately keeps separate:
- Soft injection writes directly into the record's status registers
and is unconstrained by the spec. It is the tool of choice for
state-machine coverage: notifier delivery, ABI shape.
- Hard injection drives the architected inject registers and obeys
the hardware contract, exercising the real interrupt and decode
paths.
Signed-off-by: Ruidong Tian <tianruidong at linux.alibaba.com>
---
Documentation/ABI/testing/debugfs-arm64-ras | 39 +++++-
drivers/ras/arm64/Makefile | 1 +
drivers/ras/arm64/ras-core.c | 26 ++--
drivers/ras/arm64/ras-inject.c | 127 ++++++++++++++++++++
drivers/ras/arm64/ras-sysfs.c | 1 +
drivers/ras/arm64/ras.h | 2 +
6 files changed, 183 insertions(+), 13 deletions(-)
create mode 100644 drivers/ras/arm64/ras-inject.c
diff --git a/Documentation/ABI/testing/debugfs-arm64-ras b/Documentation/ABI/testing/debugfs-arm64-ras
index d86bde83d0b9..a14f481c0c04 100644
--- a/Documentation/ABI/testing/debugfs-arm64-ras
+++ b/Documentation/ABI/testing/debugfs-arm64-ras
@@ -47,4 +47,41 @@ KernelVersion: 6.19
Contact: Ruidong Tian <tianruidong at linux.alibaba.com>
Description:
(RW) Read and write the CE threshold to this record.
- Returns error if input exceeded the maximum threshold.
\ No newline at end of file
+ Returns error if input exceeded the maximum threshold.
+
+What: /sys/kernel/debug/ras/arm64/<node_name>/record<index>/inject/err_*
+Date: Dec 2025
+KernelVersion 6.19
+Contact: Ruidong Tian <tianruidong at linux.alibaba.com>
+Description:
+ (RW) These registers are used to simulate soft injection errors
+ by holding error register values. You can write any values
+ to them. To trigger the injection, you need to write soft_inject
+ at last. The validity of the injected error depends on the
+ value written to err_status.
+
+ Accepts values - any.
+
+What: /sys/kernel/debug/ras/arm64/<node_name>/record<index>/inject/soft_inject
+Date: Dec 2025
+KernelVersion 6.19
+Contact: Ruidong Tian <tianruidong at linux.alibaba.com>
+Description:
+ (WO) Write any value to this file to trigger the error
+ injection. Make sure you have specified all necessary error
+ parameters, i.e. this write should be the last step when
+ injecting errors.
+
+ Accepts values - any.
+
+What: /sys/kernel/debug/ras/arm64/<node_name>/record<index>/inject/hard_inject
+Date: Dec 2025
+KernelVersion 6.19
+Contact: Ruidong Tian <tianruidong at linux.alibaba.com>
+Description:
+ (WO) If the AEST table provides error injection registers,
+ you can write to them via this interface. For instance,
+ values can be written to the ERXPFGCTL register. The post-injection
+ behavior is then determined by the hardware specification.
+
+ Accepts values - any.
\ No newline at end of file
diff --git a/drivers/ras/arm64/Makefile b/drivers/ras/arm64/Makefile
index e13e223107dd..2f3119ac3ec5 100644
--- a/drivers/ras/arm64/Makefile
+++ b/drivers/ras/arm64/Makefile
@@ -4,3 +4,4 @@ obj-$(CONFIG_ARM64_RAS_DRIVER) += arm64_ras.o
arm64_ras-y := ras-core.o
arm64_ras-y += ras-sysfs.o
+arm64_ras-y += ras-inject.o
diff --git a/drivers/ras/arm64/ras-core.c b/drivers/ras/arm64/ras-core.c
index c427c131a862..8cde26ab95de 100644
--- a/drivers/ras/arm64/ras-core.c
+++ b/drivers/ras/arm64/ras-core.c
@@ -200,7 +200,7 @@ static void ras_panic(struct ras_record *record, struct ras_ext_regs *regs,
panic(msg);
}
-static void ras_proc_record(struct ras_record *record, void *data)
+void ras_proc_record(struct ras_record *record, void *data, bool fake)
{
struct ras_ext_regs regs = { 0 };
int *count = data;
@@ -240,13 +240,15 @@ static void ras_proc_record(struct ras_record *record, void *data)
ue = FIELD_GET(ERR_STATUS_UET, regs.err_status);
if ((regs.err_status & ERR_STATUS_UE) &&
(ue == ERR_STATUS_UET_UC || ue == ERR_STATUS_UET_UEU)) {
- if (!panic_on_ue)
- ras_record_err(record, "UE detected, panic suppressed\n");
- else
+ if (fake)
+ ras_record_info(record,
+ "Simulated error! Skip panic due to fault injection\n");
+ else if (panic_on_ue)
ras_panic(record, ®s,
- "AEST: unrecoverable error encountered");
+ "AEST: unrecoverable error encountered");
+ else
+ ras_record_err(record, "UE detected, panic suppressed\n");
}
-
ras_do_proc(record, ®s);
/* Write-one-to-clear the bits we've seen */
@@ -263,7 +265,7 @@ static void ras_proc_record(struct ras_record *record, void *data)
record_write(record, ERXSTATUS, regs.err_status);
}
-static void ras_node_foreach_record(void (*func)(struct ras_record *, void *),
+static void ras_node_foreach_record(void (*func)(struct ras_record *, void *, bool),
struct ras_node *node, void *data,
unsigned long *bitmap)
{
@@ -272,13 +274,13 @@ static void ras_node_foreach_record(void (*func)(struct ras_record *, void *),
for_each_clear_bit(i, bitmap, node->record_count) {
ras_select_record(node, i);
- func(&node->records[i], data);
+ func(&node->records[i], data, false);
ras_sync(node);
}
}
-static void ras_node_foreach_poll_record(void (*func)(struct ras_record *, void *),
+static void ras_node_foreach_poll_record(void (*func)(struct ras_record *, void *, bool),
struct ras_node *node, void *data)
{
int i;
@@ -300,7 +302,7 @@ static void ras_node_foreach_poll_record(void (*func)(struct ras_record *, void
ras_select_record(node, i);
- func(&node->records[i], data);
+ func(&node->records[i], data, false);
ras_sync(node);
}
@@ -329,7 +331,7 @@ static int ras_proc(struct ras_node *node)
*/
if (test_bit(i * BITS_PER_LONG + j, node->status_reporting))
continue;
- ras_proc_record(&node->records[j], &count);
+ ras_proc_record(&node->records[j], &count, false);
}
}
@@ -521,7 +523,7 @@ static int ras_init_record(struct ras_record *record, int i, struct ras_node *no
return 0;
}
-static void ras_online_record(struct ras_record *record, void *data)
+static void ras_online_record(struct ras_record *record, void *data, bool __unused)
{
ras_set_ce_threshold(record);
ras_enable_irq(record);
diff --git a/drivers/ras/arm64/ras-inject.c b/drivers/ras/arm64/ras-inject.c
new file mode 100644
index 000000000000..7fb522a845e7
--- /dev/null
+++ b/drivers/ras/arm64/ras-inject.c
@@ -0,0 +1,127 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * ARM Error Source Table Support
+ *
+ * Copyright (c) 2024, Alibaba Group.
+ */
+
+#include "ras.h"
+
+static struct ras_ext_regs regs_inj;
+
+struct inj_attr {
+ struct attribute attr;
+ ssize_t (*show)(struct ras_node *n, struct inj_attr *a, char *b);
+ ssize_t (*store)(struct ras_node *n, struct inj_attr *a, const char *b,
+ size_t c);
+};
+
+struct ras_inject {
+ struct ras_node *node;
+ struct kobject kobj;
+};
+
+#define to_inj(k) container_of(k, struct ras_inject, kobj)
+#define to_inj_attr(a) container_of(a, struct inj_attr, attr)
+
+static u64 ras_sysreg_read_inject(void *__unused, u32 offset)
+{
+ u64 *p = (u64 *)®s_inj;
+
+ return p[offset/8];
+}
+
+static void ras_sysreg_write_inject(void *base, u32 offset, u64 val)
+{
+ u64 *p = (u64 *)®s_inj;
+
+ p[offset/8] = val;
+}
+
+static u64 ras_iomem_read_inject(void *base, u32 offset)
+{
+ u64 *p = (u64 *)®s_inj;
+
+ return p[offset/8];
+}
+
+static void ras_iomem_write_inject(void *base, u32 offset, u64 val)
+{
+ u64 *p = (u64 *)®s_inj;
+
+ p[offset/8] = val;
+}
+
+static struct ras_access ras_access_inject[] = {
+ [ACPI_AEST_NODE_SYSTEM_REGISTER] = {
+ .read = ras_sysreg_read_inject,
+ .write = ras_sysreg_write_inject,
+ },
+
+ [ACPI_AEST_NODE_MEMORY_MAPPED] = {
+ .read = ras_iomem_read_inject,
+ .write = ras_iomem_write_inject,
+ },
+ [ACPI_AEST_NODE_SINGLE_RECORD_MEMORY_MAPPED] = {
+ .read = ras_iomem_read_inject,
+ .write = ras_iomem_write_inject,
+ },
+ { }
+};
+
+static int soft_inject_store(void *data, u64 val)
+{
+ int count = 0;
+ struct ras_record record_inj, *record = data;
+ struct ras_node *node = record->node;
+
+ memcpy(&record_inj, record, sizeof(*record));
+ record_inj.access = &ras_access_inject[node->access_type];
+
+ regs_inj.err_status |= ERR_STATUS_V;
+
+ ras_proc_record(&record_inj, &count, true);
+
+ if (count != 1)
+ return -EIO;
+
+ return 0;
+}
+DEFINE_DEBUGFS_ATTRIBUTE(soft_inject_ops, NULL, soft_inject_store, "%llu\n");
+
+static int hard_inject_store(void *data, u64 val)
+{
+ struct ras_record *record = data;
+ struct ras_node *node = record->node;
+
+ if (node->type != ACPI_AEST_PROCESSOR_ERROR_NODE && !node->inj)
+ return -EPERM;
+
+ ras_select_record(node, record->index);
+ record_write(record, ERXPFGCTL, val);
+ record_write(record, ERXPFGCDN, 0x100);
+ ras_sync(node);
+
+ return 0;
+}
+DEFINE_DEBUGFS_ATTRIBUTE(hard_inject_ops, NULL, hard_inject_store, "%llu\n");
+
+void ras_inject_init_debugfs(struct ras_record *record)
+{
+ struct dentry *inj;
+
+ inj = debugfs_create_dir("inject", record->debugfs);
+
+ debugfs_create_u64("err_fr", 0600, inj, ®s_inj.err_fr);
+ debugfs_create_u64("err_ctrl", 0600, inj, ®s_inj.err_ctlr);
+ debugfs_create_u64("err_status", 0600, inj, ®s_inj.err_status);
+ debugfs_create_u64("err_addr", 0600, inj, ®s_inj.err_addr);
+ debugfs_create_u64("err_misc0", 0600, inj, ®s_inj.err_misc[0]);
+ debugfs_create_u64("err_misc1", 0600, inj, ®s_inj.err_misc[1]);
+ debugfs_create_u64("err_misc2", 0600, inj, ®s_inj.err_misc[2]);
+ debugfs_create_u64("err_misc3", 0600, inj, ®s_inj.err_misc[3]);
+ debugfs_create_file("soft_inject", 0200, inj, record, &soft_inject_ops);
+
+ if (record->node->type == ACPI_AEST_PROCESSOR_ERROR_NODE || record->node->inj)
+ debugfs_create_file("hard_inject", 0200, inj, record, &hard_inject_ops);
+}
diff --git a/drivers/ras/arm64/ras-sysfs.c b/drivers/ras/arm64/ras-sysfs.c
index 03cc00b820e2..d8b351ee9aef 100644
--- a/drivers/ras/arm64/ras-sysfs.c
+++ b/drivers/ras/arm64/ras-sysfs.c
@@ -151,6 +151,7 @@ static void ras_record_init_debugfs(struct ras_record *record)
record, &ras_record_err_count_fops);
debugfs_create_file("ce_threshold", 0600, record->debugfs,
record, &record_ce_threshold_ops);
+ ras_inject_init_debugfs(record);
}
static void ras_init_records_debugfs(struct ras_node *node)
diff --git a/drivers/ras/arm64/ras.h b/drivers/ras/arm64/ras.h
index 92cbb975b4df..8a0d2909fe4b 100644
--- a/drivers/ras/arm64/ras.h
+++ b/drivers/ras/arm64/ras.h
@@ -302,5 +302,7 @@ static inline void ras_sync(struct ras_node *node)
}
void ras_node_init_debugfs(struct ras_node *node);
+void ras_inject_init_debugfs(struct ras_record *record);
+void ras_proc_record(struct ras_record *record, void *data, bool fake);
#endif /* _DRIVERS_RAS_ARM64_RAS_H_ */
--
2.51.2.612.gdc70283dfc
More information about the linux-arm-kernel
mailing list