Re: [PATCH] Removes extra checking in kernel/cpuset.c



Rakib Mullick wrote:
Hello guys, this patch removes an extra checking over 'cs' in
functions 'guarantee_online_cpus' and 'guarantee_online_mems'.

Thanks.

Signed-off-by: Md.Rakib H. Mullick (rakib.mullick@xxxxxxxxx)


I'd have to NAK this one as well. 2-reasons:
1) Your patch is guaranteed to generate a NULL pointer dereference at some point. You keep using cs; and assigning cs = cs->parent in the loop, at some point you may end up at the TOP of the tree where cs->parent == NULL.
2) The change breaks the actual meaning of the code. Which is suggesting my the short snippet, find an entry (while loop)... if the entry is found (if cs test), do this otherwise BUG out. Your code is saying, if cs is not NULL initially... find someplace then do something. The function never BUGS out unless someone calls with cs NULL initially.

James Kosin

Attachment: signature.asc
Description: OpenPGP digital signature



Relevant Pages

  • SUMMARY: changing password on NIS problems
    ... My problem was that I was trying to use the C1Crypt (The entry ... u_newcrypt set to 3 in the /etc/auth/default file), this encryption ... algorithm as a bug that led me to the problem. ... Is this a bug from NIS or passwd, ...
    (Tru64-UNIX-Managers)
  • RE: duration problem in WM 11
    ... "Chris" wrote: ... Is there a bug in WMP 11 relating to the length of the file? ... This problem has never occurred for us when using WMP version 10. ...
    (microsoft.public.windowsmedia.sdk)
  • OOPS: 2.6.24-rc5-mm1 -- EIP is at r_show+0x2a/0x70
    ... BUG: unable to handle kernel NULL pointer dereference ... last sysfs file: ...
    (Linux-Kernel)
  • [2.6.30-rc6] cifs_close: NULL pointer dereference
    ... The bug is reproducible, a simple: ... touch foo; cat foo ... unable to handle kernel NULL pointer dereference at 0000000000000008 ... RSP ...
    (Linux-Kernel)
  • [patch 1/1 FINAL] cpufreq_conservative/ondemand: invert meaning of ignore nice
    ... Also fixed the compiling bug that made it obviously I slacked on the testing. ... this obvious breaks any userland tools that expected ignore_nice' ... list that the entry should be removed altogether so the userland app breaks ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)