| Message ID | 20201207171940.25240-1-anton.ivanov@cambridgegreys.com |
|---|---|
| State | Accepted |
| Headers | show |
| Series | [1/3] um: Monitor error events in IRQ controller | expand |
On Mon, 2020-12-07 at 17:19 +0000, anton.ivanov@cambridgegreys.com wrote: > From: Anton Ivanov <anton.ivanov@cambridgegreys.com> > > Ensure that file closes, connection closes, etc are propagated > as interrupts in the interrupt controller. > > Fixes: ff6a17989c08 ("Epoll based IRQ controller") > Signed-off-by: Anton Ivanov <anton.ivanov@cambridgegreys.com> > --- > arch/um/os-Linux/irq.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/arch/um/os-Linux/irq.c b/arch/um/os-Linux/irq.c > index d508310ee5e1..f1732c308c61 100644 > --- a/arch/um/os-Linux/irq.c > +++ b/arch/um/os-Linux/irq.c > @@ -48,7 +48,7 @@ int os_epoll_triggered(int index, int events) > int os_event_mask(int irq_type) > { > if (irq_type == IRQ_READ) > - return EPOLLIN | EPOLLPRI; > + return EPOLLIN | EPOLLPRI | EPOLLERR | EPOLLHUP | EPOLLRDHUP; > if (irq_type == IRQ_WRITE) > return EPOLLOUT; Why not monitor it also for write? johannes
On 07/12/2020 17:35, Johannes Berg wrote: > On Mon, 2020-12-07 at 17:19 +0000, anton.ivanov@cambridgegreys.com > wrote: >> From: Anton Ivanov <anton.ivanov@cambridgegreys.com> >> >> Ensure that file closes, connection closes, etc are propagated >> as interrupts in the interrupt controller. >> >> Fixes: ff6a17989c08 ("Epoll based IRQ controller") >> Signed-off-by: Anton Ivanov <anton.ivanov@cambridgegreys.com> >> --- >> arch/um/os-Linux/irq.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/arch/um/os-Linux/irq.c b/arch/um/os-Linux/irq.c >> index d508310ee5e1..f1732c308c61 100644 >> --- a/arch/um/os-Linux/irq.c >> +++ b/arch/um/os-Linux/irq.c >> @@ -48,7 +48,7 @@ int os_epoll_triggered(int index, int events) >> int os_event_mask(int irq_type) >> { >> if (irq_type == IRQ_READ) >> - return EPOLLIN | EPOLLPRI; >> + return EPOLLIN | EPOLLPRI | EPOLLERR | EPOLLHUP | EPOLLRDHUP; >> if (irq_type == IRQ_WRITE) >> return EPOLLOUT; > > Why not monitor it also for write? The write code in most drivers has no error checking and no close upon error. I will have to go through them and ensure they behave correctly first. So - yes, it's a good idea and we will add it later, once I have double-checked all drivers. > > johannes > > > _______________________________________________ > linux-um mailing list > linux-um@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-um >
On 07/12/2020 18:11, Anton Ivanov wrote: > On 07/12/2020 17:35, Johannes Berg wrote: >> On Mon, 2020-12-07 at 17:19 +0000, anton.ivanov@cambridgegreys.com >> wrote: >>> From: Anton Ivanov <anton.ivanov@cambridgegreys.com> >>> >>> Ensure that file closes, connection closes, etc are propagated >>> as interrupts in the interrupt controller. >>> >>> Fixes: ff6a17989c08 ("Epoll based IRQ controller") >>> Signed-off-by: Anton Ivanov <anton.ivanov@cambridgegreys.com> >>> --- >>> arch/um/os-Linux/irq.c | 2 +- >>> 1 file changed, 1 insertion(+), 1 deletion(-) >>> >>> diff --git a/arch/um/os-Linux/irq.c b/arch/um/os-Linux/irq.c >>> index d508310ee5e1..f1732c308c61 100644 >>> --- a/arch/um/os-Linux/irq.c >>> +++ b/arch/um/os-Linux/irq.c >>> @@ -48,7 +48,7 @@ int os_epoll_triggered(int index, int events) >>> int os_event_mask(int irq_type) >>> { >>> if (irq_type == IRQ_READ) >>> - return EPOLLIN | EPOLLPRI; >>> + return EPOLLIN | EPOLLPRI | EPOLLERR | EPOLLHUP | EPOLLRDHUP; >>> if (irq_type == IRQ_WRITE) >>> return EPOLLOUT; >> >> Why not monitor it also for write? > > The write code in most drivers has no error checking and no close upon > error. > > I will have to go through them and ensure they behave correctly first. > > So - yes, it's a good idea and we will add it later, once I have > double-checked all drivers. I have a couple of further clean-up patches to follow, but they need both your series and my series to go in first. They are mostly cosmetic - ensuring correct behavior on epoll errors and suppressing messages for conditions which are not really errors, but "business as usual". A. > >> >> johannes >> >> >> _______________________________________________ >> linux-um mailing list >> linux-um@lists.infradead.org >> http://lists.infradead.org/mailman/listinfo/linux-um >> > >
diff --git a/arch/um/os-Linux/irq.c b/arch/um/os-Linux/irq.c index d508310ee5e1..f1732c308c61 100644 --- a/arch/um/os-Linux/irq.c +++ b/arch/um/os-Linux/irq.c @@ -48,7 +48,7 @@ int os_epoll_triggered(int index, int events) int os_event_mask(int irq_type) { if (irq_type == IRQ_READ) - return EPOLLIN | EPOLLPRI; + return EPOLLIN | EPOLLPRI | EPOLLERR | EPOLLHUP | EPOLLRDHUP; if (irq_type == IRQ_WRITE) return EPOLLOUT; return 0;