| Message ID | 1442962830-7437-3-git-send-email-eric.auger@linaro.org |
|---|---|
| State | New |
| Headers | show |
On Wed, 2015-09-23 at 00:00 +0100, Eric Auger wrote: > In irqfd mode, current code attempts to set a resamplefd whatever > the type of the IRQ. For an edge-sensitive IRQ this attempt fails > and as a consequence the whole irqfd setup fails and we fall back > to the slow mode. This patch bypasses the resamplefd setting for > non level-sentive IRQs. > > Signed-off-by: Eric Auger <eric.auger@linaro.org> > --- > hw/vfio/platform.c | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/hw/vfio/platform.c b/hw/vfio/platform.c > index 38eaccf..2c91650 100644 > --- a/hw/vfio/platform.c > +++ b/hw/vfio/platform.c > @@ -352,6 +352,10 @@ static int vfio_set_resample_eventfd(VFIOINTp *intp) > int argsz, ret; > int32_t *pfd; > > + if (!(intp->flags & VFIO_IRQ_INFO_AUTOMASKED)) { > + return 0; > + } > + > argsz = sizeof(*irq_set) + sizeof(*pfd); > irq_set = g_malloc0(argsz); > irq_set->argsz = argsz; Aren't we also wasting a file descriptor for intp->unmask on edge IRQs as well? I'd be tempted to make a helper function that does this test and use it both to avoid calling this in vfio_start_irqfd_injection() and also to avoid creating the unused eventfd in vfio_init_intp(). Thanks, Alex
On 09/23/2015 09:28 PM, Alex Williamson wrote: > On Wed, 2015-09-23 at 00:00 +0100, Eric Auger wrote: >> In irqfd mode, current code attempts to set a resamplefd whatever >> the type of the IRQ. For an edge-sensitive IRQ this attempt fails >> and as a consequence the whole irqfd setup fails and we fall back >> to the slow mode. This patch bypasses the resamplefd setting for >> non level-sentive IRQs. >> >> Signed-off-by: Eric Auger <eric.auger@linaro.org> >> --- >> hw/vfio/platform.c | 4 ++++ >> 1 file changed, 4 insertions(+) >> >> diff --git a/hw/vfio/platform.c b/hw/vfio/platform.c >> index 38eaccf..2c91650 100644 >> --- a/hw/vfio/platform.c >> +++ b/hw/vfio/platform.c >> @@ -352,6 +352,10 @@ static int vfio_set_resample_eventfd(VFIOINTp *intp) >> int argsz, ret; >> int32_t *pfd; >> >> + if (!(intp->flags & VFIO_IRQ_INFO_AUTOMASKED)) { >> + return 0; >> + } >> + >> argsz = sizeof(*irq_set) + sizeof(*pfd); >> irq_set = g_malloc0(argsz); >> irq_set->argsz = argsz; > > Aren't we also wasting a file descriptor for intp->unmask on edge IRQs > as well? I'd be tempted to make a helper function that does this test > and use it both to avoid calling this in vfio_start_irqfd_injection() > and also to avoid creating the unused eventfd in vfio_init_intp(). > Thanks, Makes sense Thanks Eric > > Alex >
diff --git a/hw/vfio/platform.c b/hw/vfio/platform.c index 38eaccf..2c91650 100644 --- a/hw/vfio/platform.c +++ b/hw/vfio/platform.c @@ -352,6 +352,10 @@ static int vfio_set_resample_eventfd(VFIOINTp *intp) int argsz, ret; int32_t *pfd; + if (!(intp->flags & VFIO_IRQ_INFO_AUTOMASKED)) { + return 0; + } + argsz = sizeof(*irq_set) + sizeof(*pfd); irq_set = g_malloc0(argsz); irq_set->argsz = argsz;
In irqfd mode, current code attempts to set a resamplefd whatever the type of the IRQ. For an edge-sensitive IRQ this attempt fails and as a consequence the whole irqfd setup fails and we fall back to the slow mode. This patch bypasses the resamplefd setting for non level-sentive IRQs. Signed-off-by: Eric Auger <eric.auger@linaro.org> --- hw/vfio/platform.c | 4 ++++ 1 file changed, 4 insertions(+)