On 02/25/2010 02:04 PM, Michael S. Tsirkin wrote:
Clobbering memory
> is not necessary since we are only caring about blocking assignments of
> "env", which is by definition in a register
Then I think you should add that as a clobber. Otherwise what prevents the
compiler from reordering this
On Thu, Feb 25, 2010 at 01:50:56PM +0100, Paolo Bonzini wrote:
> On 02/25/2010 12:40 PM, Michael S. Tsirkin wrote:
>> On Thu, Feb 18, 2010 at 11:28:14PM +0200, Blue Swirl wrote:
/* restore global registers */
-#include "hostregs_helper.h"
+asm("");
+env = (void *) s
On 02/25/2010 12:40 PM, Michael S. Tsirkin wrote:
On Thu, Feb 18, 2010 at 11:28:14PM +0200, Blue Swirl wrote:
/* restore global registers */
-#include "hostregs_helper.h"
+asm("");
+env = (void *) saved_env_reg;
Is this sufficient?
I see __asm__ __volatile__("": : :"memory") in v
On Thu, Feb 18, 2010 at 11:28:14PM +0200, Blue Swirl wrote:
> > /* restore global registers */
> > -#include "hostregs_helper.h"
> > + asm("");
> > + env = (void *) saved_env_reg;
> >
Is this sufficient?
I see __asm__ __volatile__("": : :"memory") in virtio.
Is memory clobber implied? Wh
Thanks, applied both.
On Thu, Feb 18, 2010 at 10:25 PM, Paolo Bonzini wrote:
> Since b567b38 (target-arm: remove T0 and T1, 2009-10-16) the only global
> register that is used is AREG0, so the complexity of hostregs_helper.h
> is unused. Use regular assignments and a compiler optimization barrie
