Re: [RFC] [PATCH 1/2 v2] x86: introduce int3-based instructionpatching

[Jiri Kosina]

On Thu, 11 Jul 2013, Steven Rostedt wrote:

[ .. snip .. ]
> > +	smp_wmb();
> > +
> > +	text_poke(addr, &int3, sizeof(int3));
> > +
> > +	if (len - sizeof(int3) > 0) {
> 
> I believe we need a sync here. Otherwise, if the instruction crosses
> cache lines, the original first byte could have been pulled in, and then
> after the text_poke() below, it gets the updated version, causing a
> crash on that CPU.
> 
> 		on_each_cpu(do_sync_core, NULL, 1);

Right you are.

OTOH we apparently don't need the one after the text_poke() below, as 
syncing the cores just after patching the first byte afterwards provides 
safe enough guard (at least according to hpa's words back in 2010 :) ).

Will change it for next respin of the patchset, thanks for review.

-- 
Jiri Kosina
SUSE Labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/