The interrupt is set using pseudo-NMI if it is available. Arm has a
diagnostic interrupt feature called "Arm Generic Diagnostic Dump and
Reset device", but the A64FX does not support this feature and instead
has its own device definition.
This patch series includes ones created by Sumit.[1]
I tested on FX700:
$ echo 1 > /proc/sys/kernel/sysrq
$ echo HARDLOCKUP > /sys/kernel/debug/provoke-crash/DIRECT
[ 124.712351] lkdtm: Performing direct entry HARDLOCKUP
[ 147.232096] rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
:
:
Send the "chassis power diag" command from the management server
using ipmitool, the following message is shown:
[ 206.061770] sysrq: Trigger a crash
[ 206.061779] Kernel panic - not syncing: sysrq triggered crash
:
:
Changes in V2:
- Include Sumit's patch.
- The handler calls handle_sysrq() to use the sysrq feature to cause
a panic.
- request_nmi() and request_irq() now use the same handler, and
the function name of the handler has also changed.
- Use readl()/writel() instead of readl_relaxed()/writel_relaxed().
V1:
https://lore.kernel.org/linux-arm-kernel/[email protected]/
[1] https://lore.kernel.org/all/[email protected]/
Sumit Garg (1):
tty/sysrq: Make sysrq handler NMI aware
Hitomi Hasegawa (1):
soc: fujitsu: Add A64FX diagnostic interrupt driver
MAINTAINERS | 5 +
drivers/soc/Kconfig | 1 +
drivers/soc/Makefile | 1 +
drivers/soc/fujitsu/Kconfig | 13 +++
drivers/soc/fujitsu/Makefile | 3 +
drivers/soc/fujitsu/a64fx-diag.c | 151 +++++++++++++++++++++++++++++++
drivers/tty/sysrq.c | 45 ++++++++-
include/linux/sysrq.h | 1 +
kernel/debug/debug_core.c | 1 +
9 files changed, 220 insertions(+), 1 deletion(-)
create mode 100644 drivers/soc/fujitsu/Kconfig
create mode 100644 drivers/soc/fujitsu/Makefile
create mode 100644 drivers/soc/fujitsu/a64fx-diag.c
--
2.27.0
From: Sumit Garg <[email protected]>
Allow a magic sysrq to be triggered from an NMI context. This is done
via marking some sysrq actions as NMI safe. Safe actions will be allowed
to run from NMI context whilst that cannot run from an NMI will be queued
as irq_work for later processing.
The major use-case is to add NMI debugging capabilities to the kernel
in order to debug scenarios such as:
- Primary CPU is stuck in deadlock with interrupts disabled and hence
doesn't honor serial device interrupt. So having magic sysrq triggered
as an NMI is helpful for debugging.
- Always enabled NMI based magic sysrq irrespective of whether the serial
TTY port is active or not.
- Apart from UART interrupts, it allows magic sysrq to be triggered from
a diagnostic NMI interrupt on systems such as A64FX.
A particular sysrq handler is only marked as NMI safe in case the handler
isn't contending for any synchronization primitives as in NMI context
they are expected to cause deadlocks. Note that the debug sysrq do not
contend for any synchronization primitives. It does call kgdb_breakpoint()
to provoke a trap but that trap handler should be NMI safe on
architectures that implement an NMI.
Signed-off-by: Sumit Garg <[email protected]>
Signed-off-by: Hitomi Hasegawa <[email protected]>
---
drivers/tty/sysrq.c | 45 ++++++++++++++++++++++++++++++++++++++-
include/linux/sysrq.h | 1 +
kernel/debug/debug_core.c | 1 +
3 files changed, 46 insertions(+), 1 deletion(-)
diff --git a/drivers/tty/sysrq.c b/drivers/tty/sysrq.c
index bbfd004449b5..40cd492fe6ec 100644
--- a/drivers/tty/sysrq.c
+++ b/drivers/tty/sysrq.c
@@ -51,6 +51,7 @@
#include <linux/syscalls.h>
#include <linux/of.h>
#include <linux/rcupdate.h>
+#include <linux/irq_work.h>
#include <asm/ptrace.h>
#include <asm/irq_regs.h>
@@ -112,6 +113,7 @@ static const struct sysrq_key_op sysrq_loglevel_op = {
.help_msg = "loglevel(0-9)",
.action_msg = "Changing Loglevel",
.enable_mask = SYSRQ_ENABLE_LOG,
+ .nmi_safe = true,
};
#ifdef CONFIG_VT
@@ -159,6 +161,7 @@ static const struct sysrq_key_op sysrq_crash_op = {
.help_msg = "crash(c)",
.action_msg = "Trigger a crash",
.enable_mask = SYSRQ_ENABLE_DUMP,
+ .nmi_safe = true,
};
static void sysrq_handle_reboot(int key)
@@ -172,6 +175,7 @@ static const struct sysrq_key_op sysrq_reboot_op = {
.help_msg = "reboot(b)",
.action_msg = "Resetting",
.enable_mask = SYSRQ_ENABLE_BOOT,
+ .nmi_safe = true,
};
const struct sysrq_key_op *__sysrq_reboot_op = &sysrq_reboot_op;
@@ -219,6 +223,7 @@ static const struct sysrq_key_op sysrq_showlocks_op = {
.handler = sysrq_handle_showlocks,
.help_msg = "show-all-locks(d)",
.action_msg = "Show Locks Held",
+ .nmi_safe = true,
};
#else
#define sysrq_showlocks_op (*(const struct sysrq_key_op *)NULL)
@@ -291,6 +296,7 @@ static const struct sysrq_key_op sysrq_showregs_op = {
.help_msg = "show-registers(p)",
.action_msg = "Show Regs",
.enable_mask = SYSRQ_ENABLE_DUMP,
+ .nmi_safe = true,
};
static void sysrq_handle_showstate(int key)
@@ -328,6 +334,7 @@ static const struct sysrq_key_op sysrq_ftrace_dump_op = {
.help_msg = "dump-ftrace-buffer(z)",
.action_msg = "Dump ftrace buffer",
.enable_mask = SYSRQ_ENABLE_DUMP,
+ .nmi_safe = true,
};
#else
#define sysrq_ftrace_dump_op (*(const struct sysrq_key_op *)NULL)
@@ -566,6 +573,32 @@ static void __sysrq_put_key_op(int key, const struct sysrq_key_op *op_p)
sysrq_key_table[i] = op_p;
}
+static int sysrq_nmi_key = -1;
+
+static void sysrq_do_irq_work(struct irq_work *work)
+{
+ const struct sysrq_key_op *op_p;
+ int orig_suppress_printk;
+
+ orig_suppress_printk = suppress_printk;
+ suppress_printk = 0;
+
+ rcu_sysrq_start();
+ rcu_read_lock();
+
+ op_p = __sysrq_get_key_op(sysrq_nmi_key);
+ if (op_p)
+ op_p->handler(sysrq_nmi_key);
+
+ rcu_read_unlock();
+ rcu_sysrq_end();
+
+ suppress_printk = orig_suppress_printk;
+ sysrq_nmi_key = -1;
+}
+
+static DEFINE_IRQ_WORK(sysrq_irq_work, sysrq_do_irq_work);
+
void __handle_sysrq(int key, bool check_mask)
{
const struct sysrq_key_op *op_p;
@@ -573,6 +606,10 @@ void __handle_sysrq(int key, bool check_mask)
int orig_suppress_printk;
int i;
+ /* Skip sysrq handling if one already in progress */
+ if (sysrq_nmi_key != -1)
+ return;
+
orig_suppress_printk = suppress_printk;
suppress_printk = 0;
@@ -596,7 +633,13 @@ void __handle_sysrq(int key, bool check_mask)
if (!check_mask || sysrq_on_mask(op_p->enable_mask)) {
pr_info("%s\n", op_p->action_msg);
console_loglevel = orig_log_level;
- op_p->handler(key);
+
+ if (in_nmi() && !op_p->nmi_safe) {
+ sysrq_nmi_key = key;
+ irq_work_queue(&sysrq_irq_work);
+ } else {
+ op_p->handler(key);
+ }
} else {
pr_info("This sysrq operation is disabled.\n");
console_loglevel = orig_log_level;
diff --git a/include/linux/sysrq.h b/include/linux/sysrq.h
index 3a582ec7a2f1..630b5b9dc225 100644
--- a/include/linux/sysrq.h
+++ b/include/linux/sysrq.h
@@ -34,6 +34,7 @@ struct sysrq_key_op {
const char * const help_msg;
const char * const action_msg;
const int enable_mask;
+ const bool nmi_safe;
};
#ifdef CONFIG_MAGIC_SYSRQ
diff --git a/kernel/debug/debug_core.c b/kernel/debug/debug_core.c
index da06a5553835..53b56114f59b 100644
--- a/kernel/debug/debug_core.c
+++ b/kernel/debug/debug_core.c
@@ -978,6 +978,7 @@ static const struct sysrq_key_op sysrq_dbg_op = {
.handler = sysrq_handle_dbg,
.help_msg = "debug(g)",
.action_msg = "DEBUG",
+ .nmi_safe = true,
};
#endif
--
2.27.0
Hi,
On Thu, Mar 3, 2022 at 10:45 PM Hitomi Hasegawa
<[email protected]> wrote:
>
> void __handle_sysrq(int key, bool check_mask)
> {
> const struct sysrq_key_op *op_p;
> @@ -573,6 +606,10 @@ void __handle_sysrq(int key, bool check_mask)
> int orig_suppress_printk;
> int i;
>
> + /* Skip sysrq handling if one already in progress */
> + if (sysrq_nmi_key != -1)
> + return;
Should this give a warning?
Also, can you remind me why this is safe if two CPUs both call
handle_sysrq() at the same time? Can't both of them make it past this?
That doesn't seem so great.
> @@ -596,7 +633,13 @@ void __handle_sysrq(int key, bool check_mask)
> if (!check_mask || sysrq_on_mask(op_p->enable_mask)) {
> pr_info("%s\n", op_p->action_msg);
> console_loglevel = orig_log_level;
> - op_p->handler(key);
> +
> + if (in_nmi() && !op_p->nmi_safe) {
> + sysrq_nmi_key = key;
> + irq_work_queue(&sysrq_irq_work);
It looks like irq_work_queue() returns false if it fails to queue.
Maybe it's worth checking and setting "sysrq_nmi_key" back to -1 if it
fails?
-Doug
Hi Doug,
On Fri, 4 Mar 2022 at 23:36, Doug Anderson <[email protected]> wrote:
>
> Hi,
>
> On Thu, Mar 3, 2022 at 10:45 PM Hitomi Hasegawa
> <[email protected]> wrote:
> >
> > void __handle_sysrq(int key, bool check_mask)
> > {
> > const struct sysrq_key_op *op_p;
> > @@ -573,6 +606,10 @@ void __handle_sysrq(int key, bool check_mask)
> > int orig_suppress_printk;
> > int i;
> >
> > + /* Skip sysrq handling if one already in progress */
> > + if (sysrq_nmi_key != -1)
> > + return;
>
> Should this give a warning?
>
> Also, can you remind me why this is safe if two CPUs both call
> handle_sysrq() at the same time? Can't both of them make it past this?
> That doesn't seem so great.
>
>
> > @@ -596,7 +633,13 @@ void __handle_sysrq(int key, bool check_mask)
> > if (!check_mask || sysrq_on_mask(op_p->enable_mask)) {
> > pr_info("%s\n", op_p->action_msg);
> > console_loglevel = orig_log_level;
> > - op_p->handler(key);
> > +
> > + if (in_nmi() && !op_p->nmi_safe) {
> > + sysrq_nmi_key = key;
> > + irq_work_queue(&sysrq_irq_work);
>
> It looks like irq_work_queue() returns false if it fails to queue.
> Maybe it's worth checking and setting "sysrq_nmi_key" back to -1 if it
> fails?
Thanks for your comments. I hope v4 here [1] addresses all of them.
Please have a look again.
[1] https://lkml.org/lkml/2022/3/7/1059
-Sumit