[parisc-linux] kernel BUG at mm/shmem.c:585!

[parisc-linux] kernel BUG at mm/shmem.c:585!

[Grant Grundler]

Just booted 2.6.1-pa9 (-pa8 + SBA change) and got:

kernel BUG at mm/shmem.c:585!
Kernel addresses on the stack:
 [<000000001016dc30>]  [<0000000010125cd8>]  [<0000000010111cd4>]  [<000000001010be4c>] 
 [<000000001010b094>]

those correspond to:
	0x1016dc30 unmap_page_range+a0
	0x10125cd8 sys32_rt_sigaction+90
	0x10111cd4 sys_mmap2+14
	0x1010be4c syscall_exit+0
	0x1010b094 intr_check_sig+0

This is on a500 (hppa64) with:
grundler <509>df
Filesystem           1K-blocks      Used Available Use% Mounted on
/dev/md2               1539008   1257752    265624  83% /
/dev/md3               6683304   3688352   2927052  56% /home
/dev/md0              52507076   3455824  46384048   7% /usr/src
grundler <510>mount
/dev/md2 on / type ext3 (rw,errors=remount-ro)
proc on /proc type proc (rw)
devpts on /dev/pts type devpts (rw,gid=5,mode=620)
/dev/md3 on /home type ext3 (rw,errors=remount-ro)
/dev/md0 on /usr/src type ext3 (rw,errors=remount-ro)


grant

Re: [parisc-linux] kernel BUG at mm/shmem.c:585!

[Grant Grundler]

On Thu, Jan 22, 2004 at 02:38:49PM -0700, Grant Grundler wrote:
> Just booted 2.6.1-pa9 (-pa8 + SBA change) and got:
> 
> kernel BUG at mm/shmem.c:585!
> Kernel addresses on the stack:
>  [<000000001016dc30>]  [<0000000010125cd8>]  [<0000000010111cd4>]  [<000000001010be4c>] 
>  [<000000001010b094>]

I've associated this with logging out of ssh sessions.
ie when I exit a remote login, the above output pops up in the console.

grant

Re: [parisc-linux] kernel BUG at mm/shmem.c:585!

[Carlos O'Donell]

On Thu, Jan 22, 2004 at 07:09:07PM -0700, Grant Grundler wrote:
> On Thu, Jan 22, 2004 at 02:38:49PM -0700, Grant Grundler wrote:
> > Just booted 2.6.1-pa9 (-pa8 + SBA change) and got:
> > 
> > kernel BUG at mm/shmem.c:585!
> > Kernel addresses on the stack:
> 
> I've associated this with logging out of ssh sessions.
> ie when I exit a remote login, the above output pops up in the console.

I've been looking at this bug for a while now, it crops up when I cross
compile a 64-bit kernel with a newer gcc. Randolph was saying that cross
compilng with 3.0.4 doesn't generate this problem. If I remember right
it's trying to free a shared memory inode that is still being used, and
that scenario shouldn't exist. Why this happens is a mystery. 

I see this twice, once for ssh, and something else generates the error.
It appears non-fatal, and I want to blame the compiler :)

http://lists.parisc-linux.org/pipermail/parisc-linux/2003-December/021783.html

c.

Re: [parisc-linux] kernel BUG at mm/shmem.c:585!

[John David Anglin]

> It appears non-fatal, and I want to blame the compiler :)

I agree but the proof is in the pudding ;-)

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

Re: [parisc-linux] kernel BUG at mm/shmem.c:585!

[Carlos O'Donell]

On Fri, Jan 23, 2004 at 02:40:27PM -0500, John David Anglin wrote:
> > It appears non-fatal, and I want to blame the compiler :)
> 
> I agree but the proof is in the pudding ;-)

I told Grant, that at my current schedule, it would be 2006 
before I fixed this problem :)

It looks like a reordering issue, perhaps taking a look at the
respective .o objects produced by the two compilers might reveal the
issue. I think that's the way we did it when the tty layer broke and we
had to have that console.o compiled with 3.0.4 by hand? I think Randolph
and Grant worked on that together. I might have the wrong names, but
we've successfully debugged a problem like this before using similar
methods.

c.

Re: [parisc-linux] kernel BUG at mm/shmem.c:585!

[John David Anglin]

> On Fri, Jan 23, 2004 at 02:40:27PM -0500, John David Anglin wrote:
> > > It appears non-fatal, and I want to blame the compiler :)
> > 
> > I agree but the proof is in the pudding ;-)
> 
> I told Grant, that at my current schedule, it would be 2006 
> before I fixed this problem :)

My schedule isn't much better.  I'm off to Tokyo at the end of the week
to waive the Canadian flag at the ISO LRG meetin next week.  After that,
my brother's company would like me to work on more firmware for a new
project.

Is this on a SMP kernel (i.e., could this be a locking issue)?

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

Re: [parisc-linux] kernel BUG at mm/shmem.c:585!

[Grant Grundler]

On Sun, Jan 25, 2004 at 09:35:07PM -0500, John David Anglin wrote:
> Is this on a SMP kernel (i.e., could this be a locking issue)?

no. I don't build or run SMP kernels on anything at the moment.

thanks,
grant