Re: [GIT PULL] Additional x86 fixes for 2.6.31-rc5

[Paul Mackerras <paulus@samba.org>]

Linus Torvalds writes:

> On Sat, 1 Aug 2009, H. Peter Anvin wrote:
> > In particular, if I remember right the problem with using __thread for 
> > percpu was exactly that the current cpuness can change almost anywhere, 
> > unless preemption is disabled.
> 
> That shouldn't matter. If it uses '%gs', it should all just work 
> automatically. But if gcc does something different for thread-local, it's 
> basically useless.

When I tried using __thread for per-cpu variables on ppc64, I found
that gcc would sometimes precompute and cache the address of a per-cpu
variable even though it could always access the variable using an
offset from r13.  The cached address was wrong if the task got moved
to a different cpu, of course, but there was no way to tell gcc that.
Compiler barriers don't help because they say that the contents of
variables might have changed, but not their addresses.

So on x86 the concern would be that gcc might do lea %gs:foo,%rbx and
then use (%rbx) to refer to foo later on.  It would be possible to use
__thread for per-task variables rather than having to put all per-task
things in the task_struct, but __thread doesn't work for per-cpu
variables in my experience.

Paul.