Re: [rfc] x86: optimise page fault path a little

[Nick Piggin <npiggin@suse.de>]

On Thu, Nov 13, 2008 at 05:06:33PM +0100, Ingo Molnar wrote:
> 
> * Nick Piggin <npiggin@suse.de> wrote:
> 
> > > 32-bit should use oops_begin() too. Solves the previous comment as 
> > > well.
> > > 
> > > > +#ifdef CONFIG_X86_32
> > > > +	die("Oops", regs, error_code);
> > > > +	bust_spinlocks(0);
> > > > +	do_exit(SIGKILL);
> > > > +#else
> > > > +	if (__die("Oops", regs, error_code))
> > > > +		regs = NULL;
> > > > +	/* Executive summary in case the body of the oops scrolled away */
> > > > +	printk(KERN_EMERG "CR2: %016lx\n", address);
> > > > +	oops_end(flags, regs, SIGKILL);
> > > > +#endif
> > > 
> > > this difference seems unnecessary too - 32-bit should use oops_end() 
> > > too.
> > 
> > Probably all 3 good comments, but I didn't want to be tempted into 
> > changing behaviour (modulo adding bugs). Easy to merge them up in a 
> > subsequent patch, however...
> 
> please do feel tempted to clean this code up - and if it changes 
> behavior, split it up into smaller steps. The patch is already quite 
> large with a flux of 430 lines:
> 
>  1 file changed, 251 insertions(+), 188 deletions(-)
> 
> ... your primary goal is faster code, we also want cleaner code. I 
> think we can meet in the middle, have faster _and_ cleaner code, and 
> it's a done deal ;)

I might. But the flux is "supposed" to just be moving things around so
it compiles better. Any behaviour change should be a bug.

So I definitely won't add any of these types of cleanups in the same
patch. And with or without cleanups, the patch should stand on its own
to get merged -- if it's good, it's good; if not, not :)
 
That, and, I don't really know the subtlties of this code or what
made it so tricky that such a cleanup wasn't done when merging the
two files.