Re: race in ext2_file_write

[Alexander Viro]

On Wed, 15 Dec 1999, Oliver Neukum wrote:

> Hi,
> 
> is it possible, that there is a race in ext2_file_write ?
> 
> Are the following events possible ?
> 
> CPU1                          CPU2
> sys_write
> ext2_file_write
> generic_file_write
>                               executing binary written to by CPU1
> remove_suid
> 
> I checked there is no lock_kernel() in sys_write.

lock_kernel() wouldn't save you from that - the thing obviously can
block. Anyway, I'm taking the remove_suid() into generic_file_write(), so
it's going away.