Skip to contentSkip to conversation

kmsg_dump: Don't run on non-error paths by default

PATCH In review Matthew Garrett 2 participants 4 messages Feb 10, 2012 – Feb 14, 2012 Documentation

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 kmsg_dump: Don't run on non-error paths by default

Since 04c6862c055fb687c90d9652f32c11a063df15cf kmsg_dump() gets run on normal paths including poweroff and reboot. This is less than ideal given pstore implementations that can only represent single backtraces, since a reboot may overwrite a stored oops before it's been picked up by userspace. In addition, some pstore backends may have low performance and provide a significant delay in reboot as a result.

This patch adds a printk.always_kmsg_dump kernel parameter (which can also be changed from userspace). Without it, the code will only be run on failure paths rather than on normal paths. The option can be enabled in environments where there's a desire to attempt to audit whether or not a reboot was cleanly requested or not.

---
 Documentation/kernel-parameters.txt |    6 ++++++
 include/linux/kmsg_dump.h           |    9 +++++++--
 kernel/printk.c                     |    6 ++++++
 3 files changed, 19 insertions(+), 2 deletions(-)
Documentation/kernel-parameters.txt +6 -0
index 033d4e6..d99fd9c 100644
@@ -2211,6 +2211,12 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
 
 			default: off.
 
+	printk.always_kmsg_dump=
+			Trigger kmsg_dump for cases other than kernel oops or
+			panics
+			Format: <bool>  (1/Y/y=enable, 0/N/n=disable)
+			default: disabled
+
 	printk.time=	Show timing data prefixed to each printk message line
 			Format: <bool>  (1/Y/y=enable, 0/N/n=disable)
 
include/linux/kmsg_dump.h +7 -2
index fee6631..35f7237 100644
@@ -15,13 +15,18 @@
 #include <linux/errno.h>
 #include <linux/list.h>
 
+/*
+ * Keep this list arranged in rough order of priority. Anything listed after
+ * KMSG_DUMP_OOPS will not be logged by default unless printk.always_kmsg_dump
+ * is passed to the kernel.
+ */
 enum kmsg_dump_reason {
-	KMSG_DUMP_OOPS,
 	KMSG_DUMP_PANIC,
+	KMSG_DUMP_OOPS,
+	KMSG_DUMP_EMERG,
 	KMSG_DUMP_RESTART,
 	KMSG_DUMP_HALT,
 	KMSG_DUMP_POWEROFF,
-	KMSG_DUMP_EMERG,
 };
 
 /**
kernel/printk.c +6 -0
index 13c0a11..32690a0 100644
@@ -702,6 +702,9 @@ static bool printk_time = 0;
 #endif
 module_param_named(time, printk_time, bool, S_IRUGO | S_IWUSR);
 
+static bool always_kmsg_dump;
+module_param_named(always_kmsg_dump, always_kmsg_dump, bool, S_IRUGO | S_IWUSR);
+
 /* Check if we have any console registered that can be called early in boot. */
 static int have_callable_console(void)
 {
@@ -1732,6 +1735,9 @@ void kmsg_dump(enum kmsg_dump_reason reason)
 	unsigned long l1, l2;
 	unsigned long flags;
 
+	if ((reason > KMSG_DUMP_OOPS) && !always_kmsg_dump)
+		return;
+
 	/* Theoretically, the log could move on after we do this, but
 	   there's not a lot we can do about that. The new messages
 	   will overwrite the start of what we dump. */
Signature · 1 line
1.7.7.6
#1 lore raw
Quoted 7 lines · Matthew Garrett

Since 04c6862c055fb687c90d9652f32c11a063df15cf kmsg_dump() gets run on normal paths including poweroff and reboot. This is less than ideal given pstore implementations that can only represent single backtraces, since a reboot may overwrite a stored oops before it's been picked up by userspace. In addition, some pstore backends may have low performance and provide a significant delay in reboot as a result.

Quoted 5 lines

This patch adds a printk.always_kmsg_dump kernel parameter (which can also be changed from userspace). Without it, the code will only be run on failure paths rather than on normal paths. The option can be enabled in environments where there's a desire to attempt to audit whether or not a reboot was cleanly requested or not.

Anyone? Whose tree should this go through?

Signature · 1 line
Matthew Garrett | [email protected]
#2 lore raw
In reply to Matthew Garrett
Quoted 7 lines · On Fri, 10 Feb 2012 10:45:15 -0500 Matthew Garrett <[email protected]> wrote:

Since 04c6862c055fb687c90d9652f32c11a063df15cf kmsg_dump() gets run on normal paths including poweroff and reboot. This is less than ideal given pstore implementations that can only represent single backtraces, since a reboot may overwrite a stored oops before it's been picked up by userspace. In addition, some pstore backends may have low performance and provide a significant delay in reboot as a result.

Quoted 5 lines

This patch adds a printk.always_kmsg_dump kernel parameter (which can also be changed from userspace). Without it, the code will only be run on failure paths rather than on normal paths. The option can be enabled in environments where there's a desire to attempt to audit whether or not a reboot was cleanly requested or not.

So if I'm understanding it correctly, this patch reverts the kernel behaviour to what it was before 04c6862c055fb687c9 ("kmsg_dump: add kmsg_dump() calls to the reboot, halt, poweroff and emergency_restart paths"). The user must now set printk.always_kmsg_dump to get the 04c6862c055fb687c9 behaviour.

Correct? If so, we should get this into 3.3 so we don't have a oddball behaviour in one particular kernel release.

#3 lore raw
Quoted 5 lines · Andrew Morton

So if I'm understanding it correctly, this patch reverts the kernel behaviour to what it was before 04c6862c055fb687c9 ("kmsg_dump: add kmsg_dump() calls to the reboot, halt, poweroff and emergency_restart paths"). The user must now set printk.always_kmsg_dump to get the 04c6862c055fb687c9 behaviour.

Correct.

Quoted 2 lines

Correct? If so, we should get this into 3.3 so we don't have a oddball behaviour in one particular kernel release.

That would be my preference.

Signature · 1 line
Matthew Garrett | [email protected]