Messages in this thread | | | Date | Sun, 18 Jan 2009 18:38:20 +0100 | From | Ingo Molnar <> | Subject | Re: [PATCH 12/17] x86-64: Use absolute displacements for per-cpu accesses. |
| |
* Ingo Molnar <mingo@elte.hu> wrote:
> > * Brian Gerst <brgerst@gmail.com> wrote: > > > On Sat, Jan 17, 2009 at 11:22 PM, Tejun Heo <tj@kernel.org> wrote: > > > > > Hello, Brian. > > > > > > Brian Gerst wrote: > > > > Accessing memory through %gs should not use rip-relative addressing. > > > > Adding a P prefix for the argument tells gcc to not add (%rip) to > > > > the memory references. > > > > > > Nice catch. I dind't know about the P prefix thing. It also is used > > > in other places too. Hmmm... I can't find anything about the P > > > argument prefix in the gcc info page (4.3). Any ideas where I can > > > find some information about it? It's a bit weird that it's not a > > > constraint prefix but an argument one. > > > > The only place I could confirm that it works is in the gcc source > > itself, and even there it's not well documented. > > does %P support go back as far as gcc 3.2 (the earliest GCC we still > support)?
update: Brian pointed it out off-list that switch_to() already uses %P, so we already rely on it.
Ingo
| |