[lkml]   [2000]   [Nov]   [26]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: 2.4.0-test5 bug: invalid "shmid_kernel" passed to "shm_nopage_core"
Christoph Rohland <> writes:
> > I use a SysReq patch to do an oops-style dump instead of the usual
> > "showPc" function, so I was able to copy a stack dump down.
> Could you send me the patch? Does it do the dump on all cpus?

You can grab it at:

It doesn't dump all CPUs; it just dumps whichever one handles the
SysReq request, so I just keep doing it until I get them both.

> I would be happy to help debug the shm code if you find a way to
> reproduce it.

Okay. I've actually determined that my window manager (Enlightenment)
creates a shared memory segment for the X server to map and unmap
anywhere from 25 to 100 times a second; the segment appears in the X
server's memory map at the address 0x40014000 that "shm_nopage" was
trying to fault in when my lockup occurred.

I didn't notice it before because the time the page is mapped is
small. To catch it, I had to do

while true; do egrep 40014000 /proc/xxx/maps; done

and wait a bit. A "printk" for shm_mmapping at address 0x40014004
gave me the 25-100 times per second figure.

The fact that this has crashed once in all the time I've been using
this setup would seem to imply a very subtle race condition. Ugh.

Can you offer me a tutorial on the SHM locking? What's supposed to
protect against what?

In the meantime, I'll keep plugging away at it.

Kevin <>
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

 \ /
  Last update: 2005-03-22 12:47    [W:2.748 / U:0.012 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site