Re: Very weird bug in fs/exec.c

[Jordan Breeding <jordan.breeding@attbi.com>]

Felipe Contreras wrote:
> Hi,
> 
> I've found a weird bug that seems to only happend in my system. It makes
> recursive makes segfault, like:
> 
> test:
> 	( make -v )
> 
> After a lot of work tracking it I finally found what causes it, I'm attaching
> the patch that generates the bug, it's a diff from 2.5.18 to 2.5.19.
> 
> I'm saying it's weird because just adding a printk before do_execve returns
> successfully makes the bug dissapear.
> 
> BTW, yes, my system is very special.
> 


I can verify that a lot of weird problems (compiles failing with seg 
faults, `java -version` not running, other java programs not running and 
in most cases the program in question would succeed if I ran it through 
strace first) I was seeing in kernels after 2.5.18 (ie. 2.5.20-dj1, 
2.5.20-dj4, 2.5.23-dj1 and 2.5.24-dj1) go away if I reverse the patch 
included in the original email for this thread.  Glad to find out what 
had been causing that mess!  Thanks.

Jordan