Skip to contentSkip to conversation

printk timestamp post-boot suppression

PATCH In review Joe Korty 5 participants 9 messages Aug 27, 2008 – Aug 28, 2008 Library code

lore
Share & tools
lore

Joe Korty

author
#0 lore raw

PATCH 1/1 printk timestamp post-boot suppression

Suppress printk timestamping after system boot.

The timestamp printk prefix seems most useful during boot, where it easily shows where the boot sequence is spending its time.

Its utility after boot is questionable, since 1) the timestamp becomes a rather large, unreadable integer, as the hours, days and weeks go by, and 2) syslog does a proper TOD timestamp anyways on these later messages, as they go into /var/log/messages.

kernel/printk.c +1 -1
Index: 2.6.27-rc4-git4/kernel/printk.c
===================================================================
@@ -730,7 +730,7 @@
 			printed_len += 3;
 			new_text_line = 0;
 
-			if (printk_time) {
+			if (printk_time && system_state == SYSTEM_BOOTING) {
 				/* Follow the token with the time */
 				char tbuf[50], *tp;
 				unsigned tlen;
#1 lore raw
Quoted 5 lines · Joe Korty

Its utility after boot is questionable, since 1) the timestamp becomes a rather large, unreadable integer, as the hours, days and weeks go by, and 2) syslog does a proper TOD timestamp anyways on these later messages, as they go into /var/log/messages.

It'd be nicer if this were optional - syslog typically only logs at second resolution and not all systems are going to have logfiles.

author
#2 lore raw

PATCH 1/1 printk timestamp post-boot suppression, v2

Quoted 2 lines · Mark Brown

It'd be nicer if this were optional - syslog typically only logs at second resolution and not all systems are going to have logfiles.

Optionally suppress printk timestamping after system boot.

A new config option is introduced, which if selected suppresses printk timestamping after the boot sequence is completed.

kernel/printk.c +7 -1
Index: 2.6.27-rc4-git4/kernel/printk.c
===================================================================
@@ -560,6 +560,12 @@
 #endif
 module_param_named(time, printk_time, bool, S_IRUGO | S_IWUSR);
 
+#if defined(CONFIG_PRINTK_TIME_BOOTONLY)
+#define printk_time_bootonly (system_state == SYSTEM_BOOTING)
+#else
+#define printk_time_bootonly (1)
+#endif
+
 /* Check if we have any console registered that can be called early in boot. */
 static int have_callable_console(void)
 {
@@ -730,7 +736,7 @@
 			printed_len += 3;
 			new_text_line = 0;
 
-			if (printk_time) {
+			if (printk_time && printk_time_bootonly) {
 				/* Follow the token with the time */
 				char tbuf[50], *tp;
 				unsigned tlen;
lib/Kconfig.debug +11 -0
Index: 2.6.27-rc4-git4/lib/Kconfig.debug
===================================================================
@@ -9,6 +9,17 @@
 	  operations.  This is useful for identifying long delays
 	  in kernel startup.
 
+config PRINTK_TIME_BOOTONLY
+	bool "Only show printk timing during kernel startup"
+	depends on PRINTK_TIME
+	help
+	  If Y is selected, then timing information will
+	  be prefixed to printk output lines only as long
+	  as the kernel startup phase is in operation.
+
+	  If N is selected, then every printk line will
+	  be prefixed with timing information.
+
 config ENABLE_WARN_DEPRECATED
 	bool "Enable __deprecated logic"
 	default y
#3 lore raw

Re: [PATCH] printk timestamp post-boot suppression, v2

Quoted 7 lines · Joe Korty
Quoted 2 lines · Mark Brown

It'd be nicer if this were optional - syslog typically only logs at second resolution and not all systems are going to have logfiles.

Optionally suppress printk timestamping after system boot.

JOOI, what's wrong with having your init scripts contain a line like:
echo 0 > /sys/module/printk/parameters/time
?

This has the advantage that you can avoid turning the kernel's timestamps off until you know that syslog is running and timestamp events that it's receiving.

Signature · 1 line
Simon Farnsworth
author
#4 lore raw

Re: [PATCH] printk timestamp post-boot suppression, v2

Quoted 12 lines · Simon Farnsworth
Quoted 7 lines · Joe Korty
Quoted 2 lines · Mark Brown

It'd be nicer if this were optional - syslog typically only logs at second resolution and not all systems are going to have logfiles.

Optionally suppress printk timestamping after system boot.

JOOI, what's wrong with having your init scripts contain a line like:
echo 0 > /sys/module/printk/parameters/time
?

Quoted 5 lines

This has the advantage that you can avoid turning the kernel's timestamps off until you know that syslog is running and timestamp events that it's receiving.
--
Simon Farnsworth

That certainly is the better approach.
Joe

author
#5 lore raw

Re: [PATCH] printk timestamp post-boot suppression, v2

Quoted 4 lines · Joe Korty
Quoted 3 lines · Simon Farnsworth
Quoted 2 lines · Joe Korty

Optionally suppress printk timestamping after system boot.

Quoted 8 lines
Quoted 6 lines

JOOI, what's wrong with having your init scripts contain a line like:
echo 0 > /sys/module/printk/parameters/time

This has the advantage that you can avoid turning the kernel's timestamps off until you know that syslog is running and timestamp events that it's receiving.

That certainly is the better approach.

...and is already present in 2.6.27, the module_param_named(...) stmt in kernel/printk.c sets it all up.

Joe

#6 lore raw

Re: [PATCH] printk timestamp post-boot suppression, v2

In reply to Joe Korty
Quoted 6 lines · Joe Korty
Quoted 2 lines · Mark Brown

It'd be nicer if this were optional - syslog typically only logs at second resolution and not all systems are going to have logfiles.

Optionally suppress printk timestamping after system boot.

Such things shouldn't be CONFIGs, but boot options.

But a better option might be to just fix syslog or klogd to convert those time stamps into its own ones? Then your screen estate issue would disappear. That would be a user space fix only.

-Andi

Signature · 1 line
[email protected]
#7 lore raw

Re: [PATCH] printk timestamp post-boot suppression, v2

In reply to Joe Korty
Quoted 5 lines · Joe Korty
Quoted 2 lines · Mark Brown

It'd be nicer if this were optional - syslog typically only logs at second resolution and not all systems are going to have logfiles.

Optionally suppress printk timestamping after system boot.

i think the best approach is to write an initcall tracer - i think Arjan has some ideas in that area.

that would be infinitely higher quality than anything printk based anyway, as with ftrace you could see the reasons of why a boot delay happens - which tasks context-switch, in which code path, etc.

Ingo

#8 lore raw
In reply to Joe Korty
Quoted 6 lines · Joe Korty

Suppress printk timestamping after system boot.

The timestamp printk prefix seems most useful during boot, where it easily shows where the boot sequence is spending its time.

It's also very useful while the system is running. We've had problems with systems where the HDD or libata was "hiccuping" and stalling the entire system for 30 seconds; our logs were useless, as syslog was timestamping with the timestamp *after* the stall ended (as that was when it received the message), and dmesg had no timestamps.

In this case, our problem ran away and hid when we turned on printk timestamps, but if it recurred, we'd want the timestamps available to let us work out which operation was taking lots of time, without having to rebuild the kernel.

Signature · 1 line
Simon Farnsworth