Re: [Qemu-devel] [PATCH 1/7] coroutine-ucontext: use __thread

[Kevin Wolf]

Am 28.11.2014 um 15:45 hat Markus Armbruster geschrieben:
> Paolo Bonzini <address@hidden> writes:
> 
> > ELF thread local storage is about 10% faster on tests/test-coroutine's
> > perf/cost test.  The timing on my machine is 160ns per iteration with
> > pthread TLS, 145 with ELF TLS.
> >
> > Based on a patch by Kevin Wolf and Peter Lieven, but redone to follow
> > the model of coroutine-win32.c (including the important "noinline"
> > attribute!!!).
> >
> > Platforms without thread-local storage (OpenBSD probably?) will need
> > a new-enough GCC for this to compile, in order to use the same emutls
> > support that Windows already relies on.
> [...]
> > @@ -193,15 +155,22 @@ void qemu_coroutine_delete(Coroutine *co_)
> >      g_free(co);
> >  }
> >  
> > +/* This function is marked noinline to prevent GCC from inlining it
> > + * into coroutine_trampoline(). If we allow it to do that then it
> > + * hoists the code to get the address of the TLS variable "current"
> > + * out of the while() loop. This is an invalid transformation because
> > + * the SwitchToFiber() call may be called when running thread A but
> > + * return in thread B, and so we might be in a different thread
> > + * context each time round the loop.
> > + */
> >  CoroutineAction qemu_coroutine_switch(Coroutine *from_, Coroutine *to_,
> >                                        CoroutineAction action)
> 
> Err, did you forget the actual __attribute__((noinline))?

The comment needs updating, too. There's no SwitchToFiber() in the
ucontext implementation.

Kevin