printk: Fix calculation of length used to discard records
authorJeff Mahoney <jeffm@suse.com>
Fri, 10 Aug 2012 19:07:09 +0000 (15:07 -0400)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sun, 12 Aug 2012 18:25:50 +0000 (21:25 +0300)
commite3756477aec028427fec767957c0d4b6cfb87208
treec92974b2e75dea5f050b2060cf181a44c2d49643
parentf4ba394c1b02e7fc2179fda8d3941a5b3b65efb6
printk: Fix calculation of length used to discard records

While tracking down a weird buffer overflow issue in a program that
looked to be sane, I started double checking the length returned by
syslog(SYSLOG_ACTION_READ_ALL, ...) to make sure it wasn't overflowing
the buffer.

Sure enough, it was.  I saw this in strace:

  11339 syslog(SYSLOG_ACTION_READ_ALL, "<5>[244017.708129] REISERFS (dev"..., 8192) = 8279

It turns out that the loops that calculate how much space the entries
will take when they're copied don't include the newlines and prefixes
that will be included in the final output since prev flags is passed as
zero.

This patch properly accounts for it and fixes the overflow.

CC: stable@kernel.org
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/printk.c