Shrink the printk timestamp field. Keep the printk timestamp from occupying more of the scarce, 80-column console line space than it really needs. We eliminate the excessive whitespace the field added to the line, and reduce timestamp precision from six digits (usecs) to three digits (msecs). msecs seems adequate for the purpose of tracking boot sequence timing issues. Signed-off-by: Joe Korty <joe.korty@ccur.com> Index: 2.6.27-rc4-git4/kernel/printk.c =================================================================== --- 2.6.27-rc4-git4.orig/kernel/printk.c 2008-08-26 18:23:03.000000000 -0400 +++ 2.6.27-rc4-git4/kernel/printk.c 2008-08-26 18:23:49.000000000 -0400 @@ -739,9 +739,9 @@ t = cpu_clock(printk_cpu); nanosec_rem = do_div(t, 1000000000); - tlen = sprintf(tbuf, "[%5lu.%06lu] ", + tlen = sprintf(tbuf, "[%3lu.%03lu] ", (unsigned long) t, - nanosec_rem / 1000); + nanosec_rem / 1000000); for (tp = tbuf; tp < tbuf + tlen; tp++) emit_log_char(*tp); --
applied to tip/core/printk, thanks Joe. Ingo --
On Wed, 27 Aug 2008 11:17:59 -0400 This is a significant loss in utility. [ 16.817285] md: Autodetecting RAID arrays. [ 16.817288] md: Scanned 0 and added 0 devices. [ 16.817290] md: autorun ... [ 16.817292] md: ... autorun DONE. This not-terribly-fast machine can emit printks into the log buffer within two microseconds. That's a pretty useful ad-hoc timing factility. This patch will reduce the precision by a factor of five hundred. --
I was looking at it from the point of view of finding out where the boot process was too slow. For that millisecs is enough. I am not sure where knowing printk output to the microsec would be useful for solving anything. Joe --
Of course it's useful. If you're working on performance or latency in a disk, network or USB driver, microsecond resolution is about right. --
Hello, On Sat, Aug 30, 2008 at 7:16 PM, Andrew Morton I second this. If we have timestamps enables, let it be useful for all current uses. The 3 digits extra are very cheap useful information in that area (without resorting to more elaborate methods like the recently merged latency tracer). Rather than cut 3 digits off, maybe fix some of the too-wide prints would solve the posters issue better. Can we please have this patch non-committed or reverted? Regards, -- Leon --
ok, i've zapped it from tip/core/printk. Ingo --
How about supporting a configurable format string either by .config and/or in runtime, e.g. given by a command line parameter? --
