Re: Linux 2.6.27.27

[Marc Dionne <marc.c.dionne@gmail.com>]

On 07/20/2009 09:05 PM, Linus Torvalds wrote:
>
> On Mon, 20 Jul 2009, Marc Dionne wrote:
>>> It could be ccache too, of course.
>> Actually in my case it turns out that it is ccache after all - if I remove it
>> from the picture everything is fine.  If I re-enable it, even with a clean
>> cache, I get the problem.
>>
>> It might just be a coincidence that it's triggered by the -fwrapv change.
>
> Btw, do you find any core-files lying around if you enable them before the
> build with
>
> 	ulimit -c unlimited
>
> or similar?
>
> And how did you clean ccache? There's "-c" and then there's "-C".
>
> 		Linus

Unfortunately I'm not able to reproduce anymore, after clearing 
/var/cache/ccache completely (rm -rf).  Earlier I had done ccache -C, 
which didn't help.  I didn't think to copy the contents for more analysis.

So perhaps a combination of some odd ccache state along with changing 
gcc, binutils (which I updated today) and the compile flag.  Revving 
binutils and gcc back and forth didn't reproduce it.

What is odd though is that when I straced a single gcc command line that 
produced an empty .o file, it looked normal - a series of successful 
writes with the correct amount of data to a temp file, close, unlink .o 
file, rename temp file -> .o file.  But the resulting file was empty. 
Make me wonder if there was something filesystem/caching related to it.

Marc