| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <alpine.DEB.2.02.1311061149250.23353@ionos.tec.linutronix.de>
Date: Wed, 6 Nov 2013 12:30:07 +0100 (CET)
From: Thomas Gleixner <tglx@...utronix.de>
To: Geert Uytterhoeven <geert@...ux-m68k.org>
cc: Michael Schmitz <schmitzmic@...il.com>, linux-m68k@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH 09/17] [m68k] IRQ: add handle_polled_irq() for timer
based soft interrupts
On Wed, 20 Mar 2013, Geert Uytterhoeven wrote:
> On Sun, Feb 10, 2013 at 11:36 AM, Geert Uytterhoeven
> <geert@...ux-m68k.org> wrote:
> > On Thu, Jan 31, 2013 at 1:23 AM, Michael Schmitz <schmitzmic@...il.com> wrote:
> >> [PATCH 09/17] [m68k] IRQ: add handle_polled_irq() for timer based soft interrupts -
> >> experimental hack to avoid unhandled interrupt timer to fire
> >> on EtherNEC/NetUSBee cards that have no hardware interrupt
> >> and need to be polled from a timer
> >>
> >> This patch adds a special 'polled interrupt' handler for timer
> >> based software interrupts.
> >
> > Adding Thomas and lkml.
>
> No comments/suggestions?
> Thanks again!
Sorry for answering late. This slipped out of my view completely.
> >> handle_simple_irq() will respond to excessive unhandled
> >> interrupts (as are expected for a polling timer interrupt) by
> >> disabling the apparently unhandled interrupt source.
> >> handle_polled_irq() prevents this by setting the
> >> IRQS_POLL_INPROGRESS flag which will cause the unhandled
> >> interrupt events to be ignored.
> >> This is a temporary hack to allow timer based polling of the
> >> Atari ROM port network and USB cards only. Suggestions on how to
> >> properly handle this in the normal interrupt framework are most
> >> welcome.
So you're polling devices which have no hardware interrupt from the
timer interrupt. Of course if there is no interrupt pending on one of
these devices, this will trigger the spurious detector. By setting the
POLL_INPROGRESS flag, you're preventing that.
Reading the demultiplex handler it seems you have no way to figure out
which of the sub interrupts actually triggered the mfptimer_handler,
right?
I'm not too happy about the POLL flag "abuse". I'd rather have such
interrupts explicitely marked as polled by some other interrupt. That
also excludes such interrupts from the spurious mechanism completely.
Does the following patch solve the problem? You need to call
irq_set_status_flags(irq, IRQ_IS_POLLED);
when setting up the interrupt controller for those polled interrupt
lines.
Thanks,
tglx
-----------------
diff --git a/include/linux/irq.h b/include/linux/irq.h
index 56bb0dc..7dc1003 100644
--- a/include/linux/irq.h
+++ b/include/linux/irq.h
@@ -70,6 +70,9 @@ typedef void (*irq_preflow_handler_t)(struct irq_data *data);
* IRQ_MOVE_PCNTXT - Interrupt can be migrated from process context
* IRQ_NESTED_TRHEAD - Interrupt nests into another thread
* IRQ_PER_CPU_DEVID - Dev_id is a per-cpu variable
+ * IRQ_IS_POLLED - Always polled by another interrupt. Exclude
+ * it from the spurious interrupt detection
+ * mechanism and from core side polling.
*/
enum {
IRQ_TYPE_NONE = 0x00000000,
@@ -94,12 +97,14 @@ enum {
IRQ_NESTED_THREAD = (1 << 15),
IRQ_NOTHREAD = (1 << 16),
IRQ_PER_CPU_DEVID = (1 << 17),
+ IRQ_IS_POLLED = (1 << 18),
};
#define IRQF_MODIFY_MASK \
(IRQ_TYPE_SENSE_MASK | IRQ_NOPROBE | IRQ_NOREQUEST | \
IRQ_NOAUTOEN | IRQ_MOVE_PCNTXT | IRQ_LEVEL | IRQ_NO_BALANCING | \
- IRQ_PER_CPU | IRQ_NESTED_THREAD | IRQ_NOTHREAD | IRQ_PER_CPU_DEVID)
+ IRQ_PER_CPU | IRQ_NESTED_THREAD | IRQ_NOTHREAD | IRQ_PER_CPU_DEVID | \
+ IRQ_IS_POLLED)
#define IRQ_NO_BALANCING_MASK (IRQ_PER_CPU | IRQ_NO_BALANCING)
diff --git a/kernel/irq/settings.h b/kernel/irq/settings.h
index 1162f10..3320b84 100644
--- a/kernel/irq/settings.h
+++ b/kernel/irq/settings.h
@@ -14,6 +14,7 @@ enum {
_IRQ_NO_BALANCING = IRQ_NO_BALANCING,
_IRQ_NESTED_THREAD = IRQ_NESTED_THREAD,
_IRQ_PER_CPU_DEVID = IRQ_PER_CPU_DEVID,
+ _IRQ_IS_POLLED = IRQ_IS_POLLED,
_IRQF_MODIFY_MASK = IRQF_MODIFY_MASK,
};
@@ -26,6 +27,7 @@ enum {
#define IRQ_NOAUTOEN GOT_YOU_MORON
#define IRQ_NESTED_THREAD GOT_YOU_MORON
#define IRQ_PER_CPU_DEVID GOT_YOU_MORON
+#define IRQ_IS_POLLED GOT_YOU_MORON
#undef IRQF_MODIFY_MASK
#define IRQF_MODIFY_MASK GOT_YOU_MORON
@@ -147,3 +149,8 @@ static inline bool irq_settings_is_nested_thread(struct irq_desc *desc)
{
return desc->status_use_accessors & _IRQ_NESTED_THREAD;
}
+
+static inline bool irq_settings_is_polled(struct irq_desc *desc)
+{
+ return desc->status_use_accessors & _IRQ_IS_POLLED;
+}
diff --git a/kernel/irq/spurious.c b/kernel/irq/spurious.c
index 7b5f012..a1d8cc6 100644
--- a/kernel/irq/spurious.c
+++ b/kernel/irq/spurious.c
@@ -67,8 +67,13 @@ static int try_one_irq(int irq, struct irq_desc *desc, bool force)
raw_spin_lock(&desc->lock);
- /* PER_CPU and nested thread interrupts are never polled */
- if (irq_settings_is_per_cpu(desc) || irq_settings_is_nested_thread(desc))
+ /*
+ * PER_CPU, nested thread interrupts and interrupts explicitely
+ * marked polled are excluded from polling.
+ */
+ if (irq_settings_is_per_cpu(desc) ||
+ irq_settings_is_nested_thread(desc) ||
+ irq_settings_is_polled(desc))
goto out;
/*
@@ -268,7 +273,8 @@ try_misrouted_irq(unsigned int irq, struct irq_desc *desc,
void note_interrupt(unsigned int irq, struct irq_desc *desc,
irqreturn_t action_ret)
{
- if (desc->istate & IRQS_POLL_INPROGRESS)
+ if (desc->istate & IRQS_POLL_INPROGRESS ||
+ irq_settings_is_polled(desc))
return;
/* we get here again via the threaded handler */
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists