[prev in list] [next in list] [prev in thread] [next in thread] 

List:       busybox
Subject:    Re: syslogd config patch
From:       Sergey Naumov <sknaumov () gmail ! com>
Date:       2010-11-29 8:49:05
Message-ID: AANLkTi=Mww_GsdJhGngJFTU5j1dS6w-Vn8Ezeu5ywyiA () mail ! gmail ! com
[Download RAW message or body]

> What "killall busybox" says?
> What strace on klogd or sysklogd show?
> --
> vda

Killall busybox shows nothing.
Output of strace for killall busybox for 1.18.0 and 1.17.1 does not
differ (difference only in pids and addresses).

In 1.17.1 there was a killing signal handler.
In 1.18.0 signal is saved with record_signo() form libbb and then
bb_got_signal variable is tested in a loop, but when I launch busybox
klogd, it passes check for bb_got_signal and hangs (it seems that it
hangs at klogd_read(), because it does not enter to the nested loop).
So in 1.18.0 klogd can be instantly killed only with killall -9. With
killall it will wait for some log messages, then process them and only
then exit.


About syslog.conf: have you received my last mail with new version 2 patch?

-l option affects log files, specified in syslog.conf, because I have
placed parsing of fac/prio in timestamp_and_log(), because this
function is called from timestamp_and_log_internal() and from
split_escape_and_log(). But in split_escape_and_log() there is a test
for LOG_PRI(pri) < G.logLevel. From my point of view, -l option should
affect only default log, so this should be placed into
timestamp_and_log() function after fac/prio checks for entries of
syslog.conf.
What do you think about it?

Sergey Naumov.
_______________________________________________
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
[prev in list] [next in list] [prev in thread] [next in thread]