Bug 90999

Summary: chkfontpath sends SIGUSR1 to PID 2 due to a programming flaw
Product: [Retired] Red Hat Linux Reporter: Mike A. Harris <mharris>
Component: chkfontpathAssignee: Mike A. Harris <mharris>
Status: CLOSED RAWHIDE QA Contact: Ben Levenson <benl>
Severity: high Docs Contact:
Priority: high    
Version: 9CC: mitr
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: 1.9.8-1 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2003-05-17 13:23:55 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Mike A. Harris 2003-05-16 09:04:17 UTC 
There is a rather serious bug in chkfontpath, which executes as part of an
attempt to send SIGUSR1 to xfs to force it to reload it's config file.  The bug
causes SIGUSR1 to also be sent to PID 2.  Here is the releavant broken code:
 
void restartXfs()
{
  struct stat st;
  /*  pid_t pid, wst;*/
 
  /* Stat /proc/version to see if /proc is mounted. */
  if (stat("/proc/version", &st) == 0) {
 
    /* It's there, we can do the pidof w/o fear of mounting /proc. */
    system("kill -USR1 `/sbin/pidof xfs` 2&1>/dev/null");
  }
}
 
system() above evaluates to:
 
kill -USR1 `/sbin/pidof xfs` 2  &  1>/dev/null
 
Which has the effect of USR1'ing PID 2 along with xfs, and running it in the
background.  PID 2 is almost certainly always going to be a kernel thread, which
means a random kernel thread is getting sent SIGUSR1 every single time
chkfontpath is executed.  This could potentially cause all sorts of problems
depending on how the receiving kernel thread handles SIGUSR1.
 
PID 2 on some of my machines here is:
 
 2 ?        SW     0:00 [migration_CPU0]
 2 ?        SW     0:00 [kflushd]

<pknirsch> root         2  0.0  0.0     0    0 ?        SW   May05   0:01 [keventd]
 
There is no knowing what if any problems may have been caused on live systems
because of this bug, but in my opinion it is serious enough to warrant an
erratum release of chkfontpath on all supported OS platforms that are affected.
Comment 1 Mike A. Harris 2003-05-17 13:23:55 UTC 
Fixed in version 1.9.8-1 in rawhide