| Message ID | 1369753257-11091-1-git-send-email-mpa@pengutronix.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, May 28, 2013 at 05:00:57PM +0200, Markus Pargmann wrote: > The mxs interrupt controller does not support polling for interrupts, > but the driver still does it, which is a relict from > pre-MULTI_IRQ_HANDLER times. > > The existing code assumes that 0x7f means no interrupt, but this value > is an actually valid irq number, namely gpio bank 0's irq. This results > in the driver not detecting when irq 0x7f is active which makes the > machine effectively dead lock. > > This patch removes the interrupt poll loop and allows usage of gpio0 > interrupt without an infinite loop. > > Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> > Signed-off-by: Markus Pargmann <mpa@pengutronix.de> > Cc: stable@vger.kernel.org Thomas, can you please help handle the fix? If yes, here is my ack. Acked-by: Shawn Guo <shawn.guo@linaro.org> Otherwise, I can send it through arm-soc tree if you're fine with it. Shawn > --- > drivers/irqchip/irq-mxs.c | 14 ++++---------- > 1 file changed, 4 insertions(+), 10 deletions(-) > > diff --git a/drivers/irqchip/irq-mxs.c b/drivers/irqchip/irq-mxs.c > index 29889bb..63b3d4e 100644 > --- a/drivers/irqchip/irq-mxs.c > +++ b/drivers/irqchip/irq-mxs.c > @@ -76,16 +76,10 @@ asmlinkage void __exception_irq_entry icoll_handle_irq(struct pt_regs *regs) > { > u32 irqnr; > > - do { > - irqnr = __raw_readl(icoll_base + HW_ICOLL_STAT_OFFSET); > - if (irqnr != 0x7f) { > - __raw_writel(irqnr, icoll_base + HW_ICOLL_VECTOR); > - irqnr = irq_find_mapping(icoll_domain, irqnr); > - handle_IRQ(irqnr, regs); > - continue; > - } > - break; > - } while (1); > + irqnr = __raw_readl(icoll_base + HW_ICOLL_STAT_OFFSET); > + __raw_writel(irqnr, icoll_base + HW_ICOLL_VECTOR); > + irqnr = irq_find_mapping(icoll_domain, irqnr); > + handle_IRQ(irqnr, regs); > } > > static int icoll_irq_domain_map(struct irq_domain *d, unsigned int virq, > -- > 1.8.2.rc2 >
On Wed, May 29, 2013 at 10:42:58PM +0800, Shawn Guo wrote: > On Tue, May 28, 2013 at 05:00:57PM +0200, Markus Pargmann wrote: > > The mxs interrupt controller does not support polling for interrupts, > > but the driver still does it, which is a relict from > > pre-MULTI_IRQ_HANDLER times. > > > > The existing code assumes that 0x7f means no interrupt, but this value > > is an actually valid irq number, namely gpio bank 0's irq. This results > > in the driver not detecting when irq 0x7f is active which makes the > > machine effectively dead lock. > > > > This patch removes the interrupt poll loop and allows usage of gpio0 > > interrupt without an infinite loop. > > > > Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> > > Signed-off-by: Markus Pargmann <mpa@pengutronix.de> > > Cc: stable@vger.kernel.org > > Thomas, can you please help handle the fix? If yes, here is my ack. > > Acked-by: Shawn Guo <shawn.guo@linaro.org> > > Otherwise, I can send it through arm-soc tree if you're fine with it. Thomas, I'm sending the fix through arm-soc, and let me know if you have problem with that. Shawn
diff --git a/drivers/irqchip/irq-mxs.c b/drivers/irqchip/irq-mxs.c index 29889bb..63b3d4e 100644 --- a/drivers/irqchip/irq-mxs.c +++ b/drivers/irqchip/irq-mxs.c @@ -76,16 +76,10 @@ asmlinkage void __exception_irq_entry icoll_handle_irq(struct pt_regs *regs) { u32 irqnr; - do { - irqnr = __raw_readl(icoll_base + HW_ICOLL_STAT_OFFSET); - if (irqnr != 0x7f) { - __raw_writel(irqnr, icoll_base + HW_ICOLL_VECTOR); - irqnr = irq_find_mapping(icoll_domain, irqnr); - handle_IRQ(irqnr, regs); - continue; - } - break; - } while (1); + irqnr = __raw_readl(icoll_base + HW_ICOLL_STAT_OFFSET); + __raw_writel(irqnr, icoll_base + HW_ICOLL_VECTOR); + irqnr = irq_find_mapping(icoll_domain, irqnr); + handle_IRQ(irqnr, regs); } static int icoll_irq_domain_map(struct irq_domain *d, unsigned int virq,