[wplug-bsd] Frequent page faults -- getting on my nerves

Bill Moran wmoran at potentialtech.com
Tue Apr 12 08:33:38 EDT 2005


Brandon Kuczenski <brandon at 301south.net> wrote:
> I've been running freeBSD 4.10 ("stable") since around August 1.  My 
> longest uptime is about 30 days.  Every time there is a crash due to a 
> page fault by a process associated with httpd.  At the time of the last 
> crash, about an hour ago, apache was version 2.0.52_3.  I've since 
> upgraded to the latest: 2.0.53_1, but in every case in the past, upgrading 
> has not made the problem go away.
> 
> The error is always the same: "Fatal trap 12: page fault while in kernel 
> mode" and the current process is always httpd.  I would have suspected 
> hardware except that this has happened 5 times now and httpd has always 
> been the current process.
> 
> My httpd.conf is nothing if not unremarkable -- I think I'm using the 
> defaults for pretty much everything.  The only notable thing I can think 
> of is that /usr/local/www/html is a symlink to /usr/local/www/data-dist, 
> which I don't know if that's conventional or not.
> 
> Here's a sample of the entry in /var/log/messages:
> 
> Apr 11 20:06:14 ocean /kernel: 
> Apr 11 20:06:14 ocean /kernel: 
> Apr 11 20:06:14 ocean /kernel: Fatal trap 12: page fault while in kernel mode
> Apr 11 20:06:14 ocean /kernel: fault virtual address	= 0x0
> Apr 11 20:06:14 ocean /kernel: fault code		= supervisor read, page not present
> Apr 11 20:06:14 ocean /kernel: instruction pointer	= 0x8:0xc169ba17
> Apr 11 20:06:14 ocean /kernel: stack pointer	        = 0x10:0xd1907c34
> Apr 11 20:06:14 ocean /kernel: frame pointer	        = 0x10:0xd1907c64
> Apr 11 20:06:14 ocean /kernel: code segment		= base 0x0, limit 0xfffff, type 0x1b
> Apr 11 20:06:14 ocean /kernel: = DPL 0, pres 1, def32 1, gran 1
> Apr 11 20:06:14 ocean /kernel: processor eflags	= interrupt enabled, resume, IOPL = 0
> Apr 11 20:06:14 ocean /kernel: current process		= 188 (httpd)
> Apr 11 20:06:14 ocean /kernel: interrupt mask		= 
> Apr 11 20:06:14 ocean /kernel: trap number		= 12
> Apr 11 20:06:14 ocean /kernel: panic: page fault
> Apr 11 20:06:14 ocean /kernel: 
> Apr 11 20:06:14 ocean /kernel: syncing disks... 6 
> Apr 11 20:06:14 ocean /kernel: done
> Apr 11 20:06:14 ocean /kernel: Uptime: 8d22h56m44s
> Apr 11 20:06:14 ocean /kernel: Automatic reboot in 15 seconds - press a key on the console to abort
> .....
> 
> The only remarkable thing in httpd-error.log is a whole bunch of requests 
> for mp3 files that once existed (but have since been removed) just before 
> the crash.  I seem to remember that this is consistent with behavior 
> around the other page faults too.
> 
> Is this something for the httpd list?  How do I go about finding more 
> information?

If this is happening this often, you can debug the kernel.
First, subscribe to one of the FreeBSD developer lists.  Then update to the
current 4.x-STABLE (4.10 has been superceded by 4.11).  Then, follow the
instructions below to set up your system to debug the problem:

http://www.freebsd.org/doc/en_US.ISO8859-1/books/developers-handbook/kerneldebug.html

Report your findings to the developers, and (hopefully) the problem will be
tracked down and fixed.

-- 
Bill Moran
Potential Technologies
http://www.potentialtech.com


More information about the wplug-bsd mailing list