Re: Unable to handle kernel paging request [ext2 problem]

Ion Badulescu (
Mon, 24 Feb 1997 18:10:50 -0500 (EST)

On Mon, 24 Feb 1997, Ion Badulescu wrote:

> Really, folks, can you try out Bernd's code and see if it oopses on your
> particular configuration? Please? :) Drop me a message if you got rid of
> the message and I'll forward it to you.

Follow-up on my own message:

1. the problem only occurs if core is dumped (my tests had been done with
ulimit -c0 initially)
2. the 2.1.26 kernel catches an exception in ext2_file_write when running
the same code, so we definitely have a problem there. (this could also
explain some weird exceptions other people got in their logs)
3. the function which generates the oops is memcpy_fromfs, when it
dereferences the buf pointer. buf is not touched at all inside
ext2_file_write; also, there is no verify_area before the memcpy_fromfs
4. I got stack traces with do_truncate as well as __brelse as the
next return addresses on the stack, however neither of these two functions
call ext2_file_write directly:

do_truncate: call vmtruncate
__brelse : call refile_buffer

>From Jon's stack dump it would appear that the bogus value for buf is
passed by some higher-up function, but it doesn't appear at all in my own
stack dump:

Feb 24 17:52:09 moisil kernel: Unable to handle kernel paging request at virtual address 40007000
Feb 24 17:52:09 moisil kernel: Stack: 00001000 0000000a 00001000 0003f000 00000000 00000000 001a55f8 00000000
Feb 24 17:52:09 moisil kernel: 00000000 00000010 00004000 00000000 021a7e9c 00000000 0000010c 0271c68c
Feb 24 17:52:09 moisil kernel: 0271c5d0 0000000c 0271c688 0271c5d0 0271c5d0 00000000 00000000 021edd4c

Any ideas? Is some ext2 guru willing to look into this? :)


  It is better to keep your mouth shut and be thought a fool,
            than to open it and remove all doubt.