| Message ID | 1349298662-16666-1-git-send-email-sw@weilnetz.de |
|---|---|
| State | Accepted |
| Headers | show |
On Wed, Oct 03, 2012 at 11:11:02PM +0200, Stefan Weil wrote: > The current check will give a wrong result for gcc-5.x with x < 4. > Using QEMU_GNUC_PREREQ is simpler and fixes that issue. > > Signed-off-by: Stefan Weil <sw@weilnetz.de> > --- > > With current gcc versions (gcc-4.x) there is no problem. > Nevertheless, I suggest to fix this in stable versions, too. > > Regards > > Stefan Weil > > qemu-barrier.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) Thanks, applied to the trivial patches tree: https://github.com/stefanha/qemu/commits/trivial-patches Stefan
On Wed, Oct 03, 2012 at 11:11:02PM +0200, Stefan Weil wrote: > The current check will give a wrong result for gcc-5.x with x < 4. > Using QEMU_GNUC_PREREQ is simpler and fixes that issue. > > Signed-off-by: Stefan Weil <sw@weilnetz.de> > --- > > With current gcc versions (gcc-4.x) there is no problem. > Nevertheless, I suggest to fix this in stable versions, too. > > Regards > > Stefan Weil ACK for stable tree. > qemu-barrier.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/qemu-barrier.h b/qemu-barrier.h > index 7e11197..16f0942 100644 > --- a/qemu-barrier.h > +++ b/qemu-barrier.h > @@ -19,7 +19,7 @@ > * mfence on 32 bit as well, e.g. if built with -march=pentium-m. > * However, on i386, there seem to be known bugs as recently as 4.3. > * */ > -#if defined(__GNUC__) && __GNUC__ >= 4 && __GNUC_MINOR__ >= 4 > +#if QEMU_GNUC_PREREQ(4, 4) > #define smp_mb() __sync_synchronize() > #else > #define smp_mb() asm volatile("lock; addl $0,0(%%esp) " ::: "memory") > -- > 1.7.10
Am 08.10.2012 22:01, schrieb Michael S. Tsirkin: > On Wed, Oct 03, 2012 at 11:11:02PM +0200, Stefan Weil wrote: >> The current check will give a wrong result for gcc-5.x with x < 4. >> Using QEMU_GNUC_PREREQ is simpler and fixes that issue. >> >> Signed-off-by: Stefan Weil <sw@weilnetz.de> >> --- >> >> With current gcc versions (gcc-4.x) there is no problem. >> Nevertheless, I suggest to fix this in stable versions, too. >> >> Regards >> >> Stefan Weil > ACK for stable tree. > >> qemu-barrier.h | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/qemu-barrier.h b/qemu-barrier.h >> index 7e11197..16f0942 100644 >> --- a/qemu-barrier.h >> +++ b/qemu-barrier.h >> @@ -19,7 +19,7 @@ >> * mfence on 32 bit as well, e.g. if built with -march=pentium-m. >> * However, on i386, there seem to be known bugs as recently as 4.3. >> * */ >> -#if defined(__GNUC__) && __GNUC__ >= 4 && __GNUC_MINOR__ >= 4 >> +#if QEMU_GNUC_PREREQ(4, 4) >> #define smp_mb() __sync_synchronize() >> #else >> #define smp_mb() asm volatile("lock; addl $0,0(%%esp) " ::: "memory") >> -- >> 1.7.10 My patch was incomplete: an include statement for compiler.h is needed (see the patches which were applied to the same file in git master. Regards Stefan Weil
diff --git a/qemu-barrier.h b/qemu-barrier.h index 7e11197..16f0942 100644 --- a/qemu-barrier.h +++ b/qemu-barrier.h @@ -19,7 +19,7 @@ * mfence on 32 bit as well, e.g. if built with -march=pentium-m. * However, on i386, there seem to be known bugs as recently as 4.3. * */ -#if defined(__GNUC__) && __GNUC__ >= 4 && __GNUC_MINOR__ >= 4 +#if QEMU_GNUC_PREREQ(4, 4) #define smp_mb() __sync_synchronize() #else #define smp_mb() asm volatile("lock; addl $0,0(%%esp) " ::: "memory")
The current check will give a wrong result for gcc-5.x with x < 4. Using QEMU_GNUC_PREREQ is simpler and fixes that issue. Signed-off-by: Stefan Weil <sw@weilnetz.de> --- With current gcc versions (gcc-4.x) there is no problem. Nevertheless, I suggest to fix this in stable versions, too. Regards Stefan Weil qemu-barrier.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)