Hi Stable team,
The mips build for qemu (and possibly other targets) crashes when
doing any irq probe.
Specifically this happens on boot when brining up the ne driver.
This is fixed by the following two patches from linus's tree
46f4f8f665080900e865392f4b3593be463bf0d8 - IRQ_NOPROBE helper functions
24649c00ca334955ac7d8a79f5a7834fc7ea441d - MIPS: Mark all but i8259
interrupts as no-probe.
They have no impact on any other architecture so I hope are simple to
include in the stable tree.
Thanks
Samuel
ps: Included below for reference but gmail will probably mangle the
white spaces...
commit 46f4f8f665080900e865392f4b3593be463bf0d8
Author: Ralf Baechle <[email protected]>
Date: Fri Feb 8 04:22:01 2008 -0800
IRQ_NOPROBE helper functions
Probing non-ISA interrupts using the handle_percpu_irq as their handle_irq
method may crash the system because handle_percpu_irq does not check
IRQ_WAITING. This for example hits the MIPS Qemu configuration.
This patch provides two helper functions set_irq_noprobe and
set_irq_probe to
set rsp. clear the IRQ_NOPROBE flag. The only current caller is MIPS code
but this really belongs into generic code.
As an aside, interrupt probing these days has become a mostly
obsolete if not
dangerous art. I think Linux interrupts should be changed to default to
non-probing but that's subject of this patch.
Signed-off-by: Ralf Baechle <[email protected]>
Acked-and-tested-by: Rob Landley <[email protected]>
Cc: Alan Cox <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
diff --git a/include/linux/irq.h b/include/linux/irq.h
index a19b381..bfd9efb 100644
--- a/include/linux/irq.h
+++ b/include/linux/irq.h
@@ -367,6 +367,9 @@ set_irq_chained_handler(unsigned int irq,
__set_irq_handler(irq, handle, 1, NULL);
}
+extern void set_irq_noprobe(unsigned int irq);
+extern void set_irq_probe(unsigned int irq);
+
/* Handle dynamic irq creation and destruction */
extern int create_irq(void);
extern void destroy_irq(unsigned int irq);
diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c
index 10e0066..cc54c62 100644
--- a/kernel/irq/chip.c
+++ b/kernel/irq/chip.c
@@ -589,3 +589,39 @@ set_irq_chip_and_handler_name(unsigned int irq,
struct irq_chip *chip,
set_irq_chip(irq, chip);
__set_irq_handler(irq, handle, 0, name);
}
+
+void __init set_irq_noprobe(unsigned int irq)
+{
+ struct irq_desc *desc;
+ unsigned long flags;
+
+ if (irq >= NR_IRQS) {
+ printk(KERN_ERR "Trying to mark IRQ%d non-probeable¥n", irq);
+
+ return;
+ }
+
+ desc = irq_desc + irq;
+
+ spin_lock_irqsave(&desc->lock, flags);
+ desc->status |= IRQ_NOPROBE;
+ spin_unlock_irqrestore(&desc->lock, flags);
+}
+
+void __init set_irq_probe(unsigned int irq)
+{
+ struct irq_desc *desc;
+ unsigned long flags;
+
+ if (irq >= NR_IRQS) {
+ printk(KERN_ERR "Trying to mark IRQ%d probeable¥n", irq);
+
+ return;
+ }
+
+ desc = irq_desc + irq;
+
+ spin_lock_irqsave(&desc->lock, flags);
+ desc->status &= ‾IRQ_NOPROBE;
+ spin_unlock_irqrestore(&desc->lock, flags);
+}
commit 24649c00ca334955ac7d8a79f5a7834fc7ea441d
Author: Ralf Baechle <[email protected]>
Date: Fri Feb 8 04:22:02 2008 -0800
MIPS: Mark all but i8259 interrupts as no-probe.
Use set_irq_noprobe() to mark all MIPS interrupts as non-probe.
Override that
default for i8259 interrupts.
Signed-off-by: Ralf Baechle <[email protected]>
Acked-and-tested-by: Rob Landley <[email protected]>
Cc: Alan Cox <[email protected]>
Cc: Ingo Molnar <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
diff --git a/arch/mips/kernel/i8259.c b/arch/mips/kernel/i8259.c
index 197d797..413bd1d 100644
--- a/arch/mips/kernel/i8259.c
+++ b/arch/mips/kernel/i8259.c
@@ -338,8 +338,10 @@ void __init init_i8259_irqs(void)
init_8259A(0);
- for (i = I8259A_IRQ_BASE; i < I8259A_IRQ_BASE + 16; i++)
+ for (i = I8259A_IRQ_BASE; i < I8259A_IRQ_BASE + 16; i++) {
set_irq_chip_and_handler(i, &i8259A_chip, handle_level_irq);
+ set_irq_probe(i);
+ }
setup_irq(I8259A_IRQ_BASE + PIC_CASCADE_IR, &irq2);
}
diff --git a/arch/mips/kernel/irq.c b/arch/mips/kernel/irq.c
index d06e9c9..e3309ff 100644
--- a/arch/mips/kernel/irq.c
+++ b/arch/mips/kernel/irq.c
@@ -145,6 +145,11 @@ __setup("nokgdb", nokgdb);
void __init init_IRQ(void)
{
+ int i;
+
+ for (i = 0; i < NR_IRQS; i++)
+ set_irq_noprobe(i);
+
arch_init_irq();
#ifdef CONFIG_KGDB
On Mon, Feb 25, 2008 at 10:36:59AM +0900, Samuel Masham wrote:
> Hi Stable team,
>
> The mips build for qemu (and possibly other targets) crashes when
> doing any irq probe.
>
> Specifically this happens on boot when brining up the ne driver.
>
> This is fixed by the following two patches from linus's tree
>
> 46f4f8f665080900e865392f4b3593be463bf0d8 - IRQ_NOPROBE helper functions
> 24649c00ca334955ac7d8a79f5a7834fc7ea441d - MIPS: Mark all but i8259
> interrupts as no-probe.
>
> They have no impact on any other architecture so I hope are simple to
> include in the stable tree.
ACK.
Ralf
On Sunday 24 February 2008 20:08:17 Ralf Baechle wrote:
> On Mon, Feb 25, 2008 at 10:36:59AM +0900, Samuel Masham wrote:
> > Hi Stable team,
> >
> > The mips build for qemu (and possibly other targets) crashes when
> > doing any irq probe.
> >
> > Specifically this happens on boot when brining up the ne driver.
> >
> > This is fixed by the following two patches from linus's tree
> >
> > 46f4f8f665080900e865392f4b3593be463bf0d8 - IRQ_NOPROBE helper functions
> > 24649c00ca334955ac7d8a79f5a7834fc7ea441d - MIPS: Mark all but i8259
> > interrupts as no-probe.
> >
> > They have no impact on any other architecture so I hope are simple to
> > include in the stable tree.
>
> ACK.
I'm using it against 2.6.24.2 here. It applies cleanly and Works For Me (tm).
> Ralf
Rob
--
"One of my most productive days was throwing away 1000 lines of code."
- Ken Thompson.
Hi Stable, Greg, all,
Just a quick ping.
Still not seeing these on the stable queue? Any issue or am i just
looking in the wrong place:-
http://www.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git
2008/2/25 Samuel Masham <[email protected]>:
>
> 46f4f8f665080900e865392f4b3593be463bf0d8 - IRQ_NOPROBE helper functions
> 24649c00ca334955ac7d8a79f5a7834fc7ea441d - MIPS: Mark all but i8259
> interrupts as no-probe.
>
> They have no impact on any other architecture so I hope are simple to
> include in the stable tree.
thanks,
Samuel
On Tue, Mar 11, 2008 at 11:13:02AM +0900, Samuel Masham wrote:
> Hi Stable, Greg, all,
>
> Just a quick ping.
>
> Still not seeing these on the stable queue? Any issue or am i just
> looking in the wrong place:-
>
> http://www.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git
You're looking in the right place, I'm just slow these days :)
thanks,
greg k-h
On Monday 10 March 2008 22:33:06 Greg KH wrote:
> On Tue, Mar 11, 2008 at 11:13:02AM +0900, Samuel Masham wrote:
> > Hi Stable, Greg, all,
> >
> > Just a quick ping.
> >
> > Still not seeing these on the stable queue? Any issue or am i just
> > looking in the wrong place:-
> >
> > http://www.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git
>
> You're looking in the right place, I'm just slow these days :)
>
> thanks,
>
> greg k-h
Speaking of the stable queue, I just tried 2.6.24.3 and "make headers_install"
now dies with:
make[2]: *** No rule to make target
`/home/landley/firmware/firmware/build/temp-i686/linux/include/linux/if_addrlabel.h',
needed by
`/home/landley/firmware/firmware/build/cross-compiler-i686/include/linux/if_addrlabel.h'.
Stop.
I note that 2.6.24.2 didn't do that.
The problem seems t be this hunk:
diff -ru build.bak/sources/linux/include/linux/Kbuild
build/sources/linux/includ
e/linux/Kbuild
--- build.bak/sources/linux/include/linux/Kbuild 2008-02-10
23:51:11.0000
00000 -0600
+++ build/sources/linux/include/linux/Kbuild 2008-02-25
18:20:20.000000000 -0
600
@@ -217,6 +217,7 @@
unifdef-y += icmp.h
unifdef-y += icmpv6.h
unifdef-y += if_addr.h
+unifdef-y += if_addrlabel.h
unifdef-y += if_arp.h
unifdef-y += if_bridge.h
unifdef-y += if_ec.h
The corresponding if_addrlabel.h file isn't in the 2.6.24.3 tarball, that I
can find...
Rob
--
"One of my most productive days was throwing away 1000 lines of code."
- Ken Thompson.
On Sun, 16 Mar 2008 17:09:51 -0500 Rob Landley wrote:
> On Monday 10 March 2008 22:33:06 Greg KH wrote:
> > On Tue, Mar 11, 2008 at 11:13:02AM +0900, Samuel Masham wrote:
> > > Hi Stable, Greg, all,
> > >
> > > Just a quick ping.
> > >
> > > Still not seeing these on the stable queue? Any issue or am i just
> > > looking in the wrong place:-
> > >
> > > http://www.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git
> >
> > You're looking in the right place, I'm just slow these days :)
> >
> > thanks,
> >
> > greg k-h
>
> Speaking of the stable queue, I just tried 2.6.24.3 and "make headers_install"
> now dies with:
>
> make[2]: *** No rule to make target
> `/home/landley/firmware/firmware/build/temp-i686/linux/include/linux/if_addrlabel.h',
> needed by
> `/home/landley/firmware/firmware/build/cross-compiler-i686/include/linux/if_addrlabel.h'.
> Stop.
>
> I note that 2.6.24.2 didn't do that.
>
> The problem seems t be this hunk:
>
> diff -ru build.bak/sources/linux/include/linux/Kbuild
> build/sources/linux/includ
> e/linux/Kbuild
> --- build.bak/sources/linux/include/linux/Kbuild 2008-02-10
> 23:51:11.0000
> 00000 -0600
> +++ build/sources/linux/include/linux/Kbuild 2008-02-25
> 18:20:20.000000000 -0
> 600
> @@ -217,6 +217,7 @@
> unifdef-y += icmp.h
> unifdef-y += icmpv6.h
> unifdef-y += if_addr.h
> +unifdef-y += if_addrlabel.h
> unifdef-y += if_arp.h
> unifdef-y += if_bridge.h
> unifdef-y += if_ec.h
>
> The corresponding if_addrlabel.h file isn't in the 2.6.24.3 tarball, that I
> can find...
Yes, it's been reported a few times already and should be fixed/reverted
in the next stable release.
---
~Randy
* Randy Dunlap ([email protected]) wrote:
> Yes, it's been reported a few times already and should be fixed/reverted
> in the next stable release.
Yup, fix is queued up.
thanks,
-chris