Re: [patch] Prevent dead code/data removal with gcc 3.4

[Daniel Jacobowitz <dan@debian.org>]

On Sat, Feb 14, 2004 at 07:28:49AM +0100, Thiemo Seufer wrote:
> Jun Sun wrote:
> > On Sat, Feb 14, 2004 at 03:17:40AM +0100, Thiemo Seufer wrote:
> > > Ralf Baechle wrote:
> > > [snip]
> > > > Anyway, gcc could load next weeks lucky lottery numbers into the
> > > > s-registers after saving them.  That'd break save_static but not the
> > > > ABI which only promises to restore the old values in s-registers on
> > > > return.
> > > 
> > > Ok, it could, but adding such insns to the prologue wouldn't make
> > > sense at all, so this is unlikely to happen.
> > > 
> > 
> > OS people who have been around long enough know "unlikely" things
> > always end up happening. :)
> [snip]
> > sys_sigsuspend(struct pt_regs regs)
> > {
> >     8008e280:   27bdffc0        addiu   $sp,$sp,-64
> >     8008e284:   afb00030        sw      $s0,48($sp)
> >         sigset_t *uset, saveset, newset;
> > 
> >         save_static(&regs);
> 
> Which is a compiler bug, because it schedules around __asm__ __volatile__,
> but not a breakage caused by the prologue.
> 
> There's no way to be safe from broken compilers.

Not at all.  It's not code being scheduled around the asm - it's _part
of the prologue_ to save $s0 to the stack.  Register saves are
considered part of the prologue.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer