| Submitter | Richard Henderson |
|---|---|
| Date | Oct. 12, 2012, 4:26 p.m. |
| Message ID | <1350059215-7827-1-git-send-email-rth@twiddle.net> |
| Download | mbox | patch |
| Permalink | /patch/191155/ |
| State | New |
| Headers | show |
Comments
Il 12/10/2012 18:26, Richard Henderson ha scritto: > Older glibc (RHEL 5.x, Debian 5.x) does not have the _sigev_un._tid > member in its structure definition, while the accompanying kernel > headers do define SIGEV_THREAD_ID. We need configure to check for > both before using it. > > Cc: Paolo Bonzini <pbonzini@redhat.com> > Cc: Anthony Liguori <aliguori@us.ibm.com> > Signed-off-by: Richard Henderson <rth@twiddle.net> > --- > configure | 21 +++++++++++++++++++++ > qemu-timer.c | 4 ++-- > 2 files changed, 23 insertions(+), 2 deletions(-) > > FWIW, RHEL 5 is the last release for ia64. And for whatever reason, > the ultrasparc in the gcc compile farm is still running Lenny. I've > been hacking around this for some time, and now finally got around to > submitting a proper patch for it. Reviewed-by: Paolo Bonzini <pbonzini@redhat.com> > > > r~ > > > diff --git a/configure b/configure > index c4a7837..78d8819 100755 > --- a/configure > +++ b/configure > @@ -2841,6 +2841,23 @@ if compile_prog "" "" ; then > fi > > ########################################## > +# check if we have usable SIGEV_THREAD_ID > + > +sigev_thread_id=no > +cat > $TMPC << EOF > +#include <signal.h> > +int main(void) { > + struct sigevent ev; > + ev.sigev_notify = SIGEV_THREAD_ID; > + ev._sigev_un._tid = 0; > + return 0; > +} > +EOF > +if compile_prog "" "" ; then > + sigev_thread_id=yes > +fi > + > +########################################## > # check if trace backend exists > > $python "$source_path/scripts/tracetool.py" "--backend=$trace_backend" --check-backend > /dev/null 2> /dev/null > @@ -3186,6 +3203,7 @@ echo "preadv support $preadv" > echo "fdatasync $fdatasync" > echo "madvise $madvise" > echo "posix_madvise $posix_madvise" > +echo "sigev_thread_id $sigev_thread_id" > echo "uuid support $uuid" > echo "libcap-ng support $cap_ng" > echo "vhost-net support $vhost_net" > @@ -3469,6 +3487,9 @@ fi > if test "$posix_madvise" = "yes" ; then > echo "CONFIG_POSIX_MADVISE=y" >> $config_host_mak > fi > +if test "$sigev_thread_id" = "yes" ; then > + echo "CONFIG_SIGEV_THREAD_ID=y" >> $config_host_mak > +fi > > if test "$spice" = "yes" ; then > echo "CONFIG_SPICE=y" >> $config_host_mak > diff --git a/qemu-timer.c b/qemu-timer.c > index 908a103..ede84ff 100644 > --- a/qemu-timer.c > +++ b/qemu-timer.c > @@ -494,12 +494,12 @@ static int dynticks_start_timer(struct qemu_alarm_timer *t) > memset(&ev, 0, sizeof(ev)); > ev.sigev_value.sival_int = 0; > ev.sigev_notify = SIGEV_SIGNAL; > -#ifdef SIGEV_THREAD_ID > +#ifdef CONFIG_SIGEV_THREAD_ID > if (qemu_signalfd_available()) { > ev.sigev_notify = SIGEV_THREAD_ID; > ev._sigev_un._tid = qemu_get_thread_id(); > } > -#endif /* SIGEV_THREAD_ID */ > +#endif /* CONFIG_SIGEV_THREAD_ID */ > ev.sigev_signo = SIGALRM; > > if (timer_create(CLOCK_REALTIME, &ev, &host_timer)) { >
Ping? It's been Reviewed-by... r~ On 2012-10-13 02:26, Richard Henderson wrote: > Older glibc (RHEL 5.x, Debian 5.x) does not have the _sigev_un._tid > member in its structure definition, while the accompanying kernel > headers do define SIGEV_THREAD_ID. We need configure to check for > both before using it. > > Cc: Paolo Bonzini <pbonzini@redhat.com> > Cc: Anthony Liguori <aliguori@us.ibm.com> > Signed-off-by: Richard Henderson <rth@twiddle.net> > --- > configure | 21 +++++++++++++++++++++ > qemu-timer.c | 4 ++-- > 2 files changed, 23 insertions(+), 2 deletions(-) > > FWIW, RHEL 5 is the last release for ia64. And for whatever reason, > the ultrasparc in the gcc compile farm is still running Lenny. I've > been hacking around this for some time, and now finally got around to > submitting a proper patch for it. > > > r~ > > > diff --git a/configure b/configure > index c4a7837..78d8819 100755 > --- a/configure > +++ b/configure > @@ -2841,6 +2841,23 @@ if compile_prog "" "" ; then > fi > > ########################################## > +# check if we have usable SIGEV_THREAD_ID > + > +sigev_thread_id=no > +cat > $TMPC << EOF > +#include <signal.h> > +int main(void) { > + struct sigevent ev; > + ev.sigev_notify = SIGEV_THREAD_ID; > + ev._sigev_un._tid = 0; > + return 0; > +} > +EOF > +if compile_prog "" "" ; then > + sigev_thread_id=yes > +fi > + > +########################################## > # check if trace backend exists > > $python "$source_path/scripts/tracetool.py" "--backend=$trace_backend" --check-backend > /dev/null 2> /dev/null > @@ -3186,6 +3203,7 @@ echo "preadv support $preadv" > echo "fdatasync $fdatasync" > echo "madvise $madvise" > echo "posix_madvise $posix_madvise" > +echo "sigev_thread_id $sigev_thread_id" > echo "uuid support $uuid" > echo "libcap-ng support $cap_ng" > echo "vhost-net support $vhost_net" > @@ -3469,6 +3487,9 @@ fi > if test "$posix_madvise" = "yes" ; then > echo "CONFIG_POSIX_MADVISE=y" >> $config_host_mak > fi > +if test "$sigev_thread_id" = "yes" ; then > + echo "CONFIG_SIGEV_THREAD_ID=y" >> $config_host_mak > +fi > > if test "$spice" = "yes" ; then > echo "CONFIG_SPICE=y" >> $config_host_mak > diff --git a/qemu-timer.c b/qemu-timer.c > index 908a103..ede84ff 100644 > --- a/qemu-timer.c > +++ b/qemu-timer.c > @@ -494,12 +494,12 @@ static int dynticks_start_timer(struct qemu_alarm_timer *t) > memset(&ev, 0, sizeof(ev)); > ev.sigev_value.sival_int = 0; > ev.sigev_notify = SIGEV_SIGNAL; > -#ifdef SIGEV_THREAD_ID > +#ifdef CONFIG_SIGEV_THREAD_ID > if (qemu_signalfd_available()) { > ev.sigev_notify = SIGEV_THREAD_ID; > ev._sigev_un._tid = qemu_get_thread_id(); > } > -#endif /* SIGEV_THREAD_ID */ > +#endif /* CONFIG_SIGEV_THREAD_ID */ > ev.sigev_signo = SIGALRM; > > if (timer_create(CLOCK_REALTIME, &ev, &host_timer)) { >
On Fri, Oct 12, 2012 at 09:26:55AM -0700, Richard Henderson wrote: > Older glibc (RHEL 5.x, Debian 5.x) does not have the _sigev_un._tid > member in its structure definition, while the accompanying kernel > headers do define SIGEV_THREAD_ID. We need configure to check for > both before using it. > > Cc: Paolo Bonzini <pbonzini@redhat.com> > Cc: Anthony Liguori <aliguori@us.ibm.com> > Signed-off-by: Richard Henderson <rth@twiddle.net> > --- > configure | 21 +++++++++++++++++++++ > qemu-timer.c | 4 ++-- > 2 files changed, 23 insertions(+), 2 deletions(-) > > FWIW, RHEL 5 is the last release for ia64. And for whatever reason, > the ultrasparc in the gcc compile farm is still running Lenny. I've > been hacking around this for some time, and now finally got around to > submitting a proper patch for it. > > > r~ > > > diff --git a/configure b/configure > index c4a7837..78d8819 100755 > --- a/configure > +++ b/configure > @@ -2841,6 +2841,23 @@ if compile_prog "" "" ; then > fi > > ########################################## > +# check if we have usable SIGEV_THREAD_ID > + > +sigev_thread_id=no > +cat > $TMPC << EOF > +#include <signal.h> > +int main(void) { > + struct sigevent ev; > + ev.sigev_notify = SIGEV_THREAD_ID; > + ev._sigev_un._tid = 0; > + return 0; > +} > +EOF > +if compile_prog "" "" ; then > + sigev_thread_id=yes > +fi > + > +########################################## > # check if trace backend exists > > $python "$source_path/scripts/tracetool.py" "--backend=$trace_backend" --check-backend > /dev/null 2> /dev/null > @@ -3186,6 +3203,7 @@ echo "preadv support $preadv" > echo "fdatasync $fdatasync" > echo "madvise $madvise" > echo "posix_madvise $posix_madvise" > +echo "sigev_thread_id $sigev_thread_id" > echo "uuid support $uuid" > echo "libcap-ng support $cap_ng" > echo "vhost-net support $vhost_net" > @@ -3469,6 +3487,9 @@ fi > if test "$posix_madvise" = "yes" ; then > echo "CONFIG_POSIX_MADVISE=y" >> $config_host_mak > fi > +if test "$sigev_thread_id" = "yes" ; then > + echo "CONFIG_SIGEV_THREAD_ID=y" >> $config_host_mak > +fi > > if test "$spice" = "yes" ; then > echo "CONFIG_SPICE=y" >> $config_host_mak > diff --git a/qemu-timer.c b/qemu-timer.c > index 908a103..ede84ff 100644 > --- a/qemu-timer.c > +++ b/qemu-timer.c > @@ -494,12 +494,12 @@ static int dynticks_start_timer(struct qemu_alarm_timer *t) > memset(&ev, 0, sizeof(ev)); > ev.sigev_value.sival_int = 0; > ev.sigev_notify = SIGEV_SIGNAL; > -#ifdef SIGEV_THREAD_ID > +#ifdef CONFIG_SIGEV_THREAD_ID > if (qemu_signalfd_available()) { > ev.sigev_notify = SIGEV_THREAD_ID; > ev._sigev_un._tid = qemu_get_thread_id(); > } > -#endif /* SIGEV_THREAD_ID */ > +#endif /* CONFIG_SIGEV_THREAD_ID */ > ev.sigev_signo = SIGALRM; > > if (timer_create(CLOCK_REALTIME, &ev, &host_timer)) { Looks good on principle, but doesn't work for me. Running configure I get: | $./configure | ERROR: configure test passed without -Werror but failed with -Werror. | This is probably a bug in the configure script. The failing command | will be at the bottom of config.log. | You can run configure with --disable-werror to bypass this check. | $ And config.log contains: | /tmp/qemu-conf--6963-.c: In function ‘main’: | /tmp/qemu-conf--6963-.c:3:19: error: variable ‘ev’ set but not used [-Werror=unused-but-set-variable] | cc1: all warnings being treated as errors
Patch
diff --git a/configure b/configure index c4a7837..78d8819 100755 --- a/configure +++ b/configure @@ -2841,6 +2841,23 @@ if compile_prog "" "" ; then fi ########################################## +# check if we have usable SIGEV_THREAD_ID + +sigev_thread_id=no +cat > $TMPC << EOF +#include <signal.h> +int main(void) { + struct sigevent ev; + ev.sigev_notify = SIGEV_THREAD_ID; + ev._sigev_un._tid = 0; + return 0; +} +EOF +if compile_prog "" "" ; then + sigev_thread_id=yes +fi + +########################################## # check if trace backend exists $python "$source_path/scripts/tracetool.py" "--backend=$trace_backend" --check-backend > /dev/null 2> /dev/null @@ -3186,6 +3203,7 @@ echo "preadv support $preadv" echo "fdatasync $fdatasync" echo "madvise $madvise" echo "posix_madvise $posix_madvise" +echo "sigev_thread_id $sigev_thread_id" echo "uuid support $uuid" echo "libcap-ng support $cap_ng" echo "vhost-net support $vhost_net" @@ -3469,6 +3487,9 @@ fi if test "$posix_madvise" = "yes" ; then echo "CONFIG_POSIX_MADVISE=y" >> $config_host_mak fi +if test "$sigev_thread_id" = "yes" ; then + echo "CONFIG_SIGEV_THREAD_ID=y" >> $config_host_mak +fi if test "$spice" = "yes" ; then echo "CONFIG_SPICE=y" >> $config_host_mak diff --git a/qemu-timer.c b/qemu-timer.c index 908a103..ede84ff 100644 --- a/qemu-timer.c +++ b/qemu-timer.c @@ -494,12 +494,12 @@ static int dynticks_start_timer(struct qemu_alarm_timer *t) memset(&ev, 0, sizeof(ev)); ev.sigev_value.sival_int = 0; ev.sigev_notify = SIGEV_SIGNAL; -#ifdef SIGEV_THREAD_ID +#ifdef CONFIG_SIGEV_THREAD_ID if (qemu_signalfd_available()) { ev.sigev_notify = SIGEV_THREAD_ID; ev._sigev_un._tid = qemu_get_thread_id(); } -#endif /* SIGEV_THREAD_ID */ +#endif /* CONFIG_SIGEV_THREAD_ID */ ev.sigev_signo = SIGALRM; if (timer_create(CLOCK_REALTIME, &ev, &host_timer)) {
Older glibc (RHEL 5.x, Debian 5.x) does not have the _sigev_un._tid member in its structure definition, while the accompanying kernel headers do define SIGEV_THREAD_ID. We need configure to check for both before using it. Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Anthony Liguori <aliguori@us.ibm.com> Signed-off-by: Richard Henderson <rth@twiddle.net> --- configure | 21 +++++++++++++++++++++ qemu-timer.c | 4 ++-- 2 files changed, 23 insertions(+), 2 deletions(-) FWIW, RHEL 5 is the last release for ia64. And for whatever reason, the ultrasparc in the gcc compile farm is still running Lenny. I've been hacking around this for some time, and now finally got around to submitting a proper patch for it. r~