> > Going back to my devel tree, based on Linus' 2.3.18, and the problem
> > goes away. It is apparently only repeatable by using e-mail 'til it
> > breaks again.
>
> Ok can you duplicate this first on:
>
>
> 2.3.18ac2
> - implies its the flushpage fix
>
> 2.3.18ac4
> - suggests its kupdated
>
> 2.3.18ac6
> - several candiates
I just noticed that with 2.3.18ac8, the dependency data appended to the
end of the Makefiles in wine became corrupted. It compiled the first
time fine, but when I went back about 10 minutes later and changed some of
the source code and tried to recompile, it gave messages about skipping
NULL characters in the file. This happened to three of the Makefiles. I
looked inside and sure enough, there were spurts of NULLs in the
dependency data.
The fact that it compiled fine the first time and then broke the next
time makes me think it's probably kupdated or some sort of writing to
disk problem rather than in-memory corruption.
I can try to reproduce this on older kernels if you'd like.
...Oy, actually, it just reproduced itself -- after having corrected the
files and let it recompile (a few hours ago), I just tried "make" again
now and got the same problem:
[sroot@oof:/usr/src/wine/wine-990923]# make
make[1]: Entering directory `/usr/src/wine/wine-990923/tools'
make[1]: Nothing to be done for `all'.
make[1]: Leaving directory `/usr/src/wine/wine-990923/tools'
make[1]: Entering directory `/usr/src/wine/wine-990923/tools/wrc'
make[1]: Nothing to be done for `all'.
make[1]: Leaving directory `/usr/src/wine/wine-990923/tools/wrc'
make[1]: Entering directory `/usr/src/wine/wine-990923/debugger'
Makefile:239: warning: NUL character seen; rest of line ignored
Makefile:242: warning: NUL character seen; rest of line ignored
Makefile:243: warning: NUL character seen; rest of line ignored
make[1]: *** No rule to make target `../inc', needed by `break.o'. Stop.
make[1]: Leaving directory `/usr/src/wine/wine-990923/debugger'
make: *** [debugger/debugger.o] Error 2
*blink*. :)
I attached the Makefile if it will help any (offset and length of
corruption or something, maybe). This is on an ext2 filesystem with 1k
blocks.
I'll test some older kernels tomorrow.
Simon-
| Simon Kirby | Systems Administration |
| mailto:sim@netnation.com | NetNation Communications |
| http://www.netnation.com/ | Tech: (604) 684-6892 |
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/