| Message ID | 1367264377-7107-1-git-send-email-julien.grall@linaro.org |
|---|---|
| State | Changes Requested, archived |
| Headers | show |
On Mon, 29 Apr 2013, Julien Grall wrote: > Reset the IRQ_NOAUTOEN and IRQ_NOREQUEST flags that are enabled by > default on ARM. If IRQ_NOAUTOEN is set, __setup_irq doesn't call > irq_startup, that is responsible for calling irq_unmask at startup time. > As a result event channels remain masked. > > The clear is already made in bind_evtchn_to_irq with commit a8636c0 but was > missing in bind_virq_to_irq. > > Signed-off-by: Julien Grall <julien.grall@linaro.org> As in the original commit, you should point out that this change does not have any effects on x86 (where IRQ_NOREQUEST and IRQ_NOAUTOEN are cleared by default). At this point we might as well do this consistently everywhere we allocate a new evtchn irq, including pirqs and ipis, even though we don't actually use them on ARM. If the call to irq_clear_status_flags can be moved earlier, a good place for it could be xen_irq_init. > drivers/xen/events.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/drivers/xen/events.c b/drivers/xen/events.c > index d8cc812..b0ad226 100644 > --- a/drivers/xen/events.c > +++ b/drivers/xen/events.c > @@ -994,6 +994,7 @@ int bind_virq_to_irq(unsigned int virq, unsigned int cpu) > WARN_ON(info == NULL || info->type != IRQT_VIRQ); > } > > + irq_clear_status_flags(irq, IRQ_NOREQUEST|IRQ_NOAUTOEN); > out: > mutex_unlock(&irq_mapping_update_lock);
On 04/30/2013 04:02 PM, Stefano Stabellini wrote: > On Mon, 29 Apr 2013, Julien Grall wrote: >> Reset the IRQ_NOAUTOEN and IRQ_NOREQUEST flags that are enabled by >> default on ARM. If IRQ_NOAUTOEN is set, __setup_irq doesn't call >> irq_startup, that is responsible for calling irq_unmask at startup time. >> As a result event channels remain masked. >> >> The clear is already made in bind_evtchn_to_irq with commit a8636c0 but was >> missing in bind_virq_to_irq. >> >> Signed-off-by: Julien Grall <julien.grall@linaro.org> > > As in the original commit, you should point out that this change does > not have any effects on x86 (where IRQ_NOREQUEST and IRQ_NOAUTOEN are > cleared by default). > > At this point we might as well do this consistently everywhere we > allocate a new evtchn irq, including pirqs and ipis, even though we don't > actually use them on ARM. > > If the call to irq_clear_status_flags can be moved earlier, a good place > for it could be xen_irq_init. Thanks for the review. I will give a try and send a version with it if the solution works.
diff --git a/drivers/xen/events.c b/drivers/xen/events.c index d8cc812..b0ad226 100644 --- a/drivers/xen/events.c +++ b/drivers/xen/events.c @@ -994,6 +994,7 @@ int bind_virq_to_irq(unsigned int virq, unsigned int cpu) WARN_ON(info == NULL || info->type != IRQT_VIRQ); } + irq_clear_status_flags(irq, IRQ_NOREQUEST|IRQ_NOAUTOEN); out: mutex_unlock(&irq_mapping_update_lock);
Reset the IRQ_NOAUTOEN and IRQ_NOREQUEST flags that are enabled by default on ARM. If IRQ_NOAUTOEN is set, __setup_irq doesn't call irq_startup, that is responsible for calling irq_unmask at startup time. As a result event channels remain masked. The clear is already made in bind_evtchn_to_irq with commit a8636c0 but was missing in bind_virq_to_irq. Signed-off-by: Julien Grall <julien.grall@linaro.org> --- drivers/xen/events.c | 1 + 1 file changed, 1 insertion(+)