xfs
[Top] [All Lists]

Re: TAKE - work around gcc bug during xfs_growfs

To: Eric Sandeen <sandeen@xxxxxxx>, "linux-xfs@xxxxxxxxxxx" <linux-xfs@xxxxxxxxxxx>
Subject: Re: TAKE - work around gcc bug during xfs_growfs
From: Peter Wächtler <pwaechtler@xxxxxxxxxxxxx>
Date: Thu, 11 Oct 2001 17:58:28 +0200
Organization: LOEWE. Hannover
References: <200110102047.f9AKlqE00907@stout.americas.sgi.com>
Sender: owner-linux-xfs@xxxxxxxxxxx
Eric Sandeen wrote:
> 
> xfs_growfs was creating corrupted filesystems on kernels compiled with certain
> compilers.
> 
> A macro was passed to the function, but the function did not receive the
> correct value, i.e.:
> 
> function(MACRO(foo));
> 
> had a different result than
> 
> bar = MACRO(foo);
> function(bar);
> 
> The latter case yielded correct results.
> 
> This bug resulted in AGF headers being placed into the wrong filesystem
> blocks.
> 
> This was exibited on the 2.9x compilers in RH 7.1 and Mandrake 8.1.
> 
> "kgcc" and Mandrake's 3.0.1 compiler did not exhibit this problem.
> 
> Just another reminder that when we suggest the use of particular compilers,
> it's for a good reason!  :)
> 

Would you please provide code snippets or give a hint which portion in 
xfs_growfs
triggers this compiler bug?

Not that I don't believe you ;) It gives me really headaches..

Did you report this to the gcc maintainers?


<Prev in Thread] Current Thread [Next in Thread>