Convert ARM to generic irq subsystem
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 6 lines · Russell King - ARM Linux
Quoted 1 line · Thomas Gleixner
Our patchset, which converts ARM to the generic irq subsystem has reached a stable state and we ask for inclusion.
We require people's help on this list concerning these patches. The latest versions can be found at:
Quoted 8 lines
http://www.tglx.de/projects/armirq/
http://www.tglx.de/projects/armirq/2.6.17-rc1/patch-2.6.17-rc1-armirq5.patches.tar.bz2(please check for later versions! - note though that www.tglx.de's isp is having issues atm.)
Basically, there's two ways that we can merge these patches:
I am testing these on OMAP.
First pass shows that smc91x interrupts are getting lost. They are GPIO interrupts on most OMAP platforms and pass through a chained gpio handler. I'm still investigating, but I'm guessing it as something to do with SA_TRIGGER_* flags used in the smc91x driver for OMAP.
I don't see handling of SA_TRIGGER_* in the generic layer. Is there a new way of handling these?
Kevin
Quoted 9 lines · Kevin Hilman
R am testing these on OMAP.
First pass shows that smc91x interrupts are getting lost. They are GPIO interrupts on most OMAP platforms and pass through a chained gpio handler. I'm still investigating, but I'm guessing it as something to do with SA_TRIGGER_* flags used in the smc91x driver for OMAP.
I don't see handling of SA_TRIGGER_* in the generic layer. Is there a new way of handling these?
No. That chunk got lost unfortunately. Can you retry with the new patch against -rc3 please ?
http://www.tglx.de/projects/armirq/2.6.17-rc3/patch-2.6.17-rc3-armirq1.patches.tar.bz2
Thanks,
tglx
-------------------------------------------------------------------
List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel FAQ: http://www.arm.linux.org.uk/mailinglists/faq.php Etiquette: http://www.arm.linux.org.uk/mailinglists/etiquette.php
Quoted 12 lines · Thomas Gleixner
Quoted 5 lines · Kevin Hilman
R am testing these on OMAP.
First pass shows that smc91x interrupts are getting lost. They are GPIO interrupts on most OMAP platforms and pass through a chained gpio handler. I'm still investigating, but I'm guessing it as something to do with SA_TRIGGER_* flags used in the smc91x driver for OMAP.
I don't see handling of SA_TRIGGER_* in the generic layer. Is there a new way of handling these?
No. That chunk got lost unfortunately. Can you retry with the new patch against -rc3 please ?
http://www.tglx.de/projects/armirq/2.6.17-rc3/patch-2.6.17-rc3-armirq1.patches.tar.bz2
ok, I tried with -rc3 and I still dont see any GPIO interrupts.
After looking closer at the IRQ_TYPE_* flags, I noticed that they didn't match the SA_TRIGGER_* flags for _FALLING or _RISING. The following patch swaps the two and make OMAP happy. It's now booting, and things seem sane. Will do some more intensive testing
Kevin
include/linux/irq.h +2 -2
@@ -39,8 +39,8 @@* IRQ types, see also include/linux/interrupt.h */ #define IRQ_TYPE_NONE 0x0000/* Default, unspecified type */ -#define IRQ_TYPE_EDGE_FALLING 0x0001/* Edge falling type */ -#define IRQ_TYPE_EDGE_RISING 0x0002/* Edge rising type */ +#define IRQ_TYPE_EDGE_RISING 0x0001/* Edge rising type */ +#define IRQ_TYPE_EDGE_FALLING 0x0002/* Edge falling type */ #define IRQ_TYPE_EDGE_BOTH (IRQ_TYPE_EDGE_FALLING | IRQ_TYPE_EDGE_RISING)#define IRQ_TYPE_LEVEL_HIGH 0x0004/* Level high type */ #define IRQ_TYPE_LEVEL_LOW 0x0008/* Level low type */
Quoted 8 lines · Kevin Hilman
ok, I tried with -rc3 and I still dont see any GPIO interrupts.
After looking closer at the IRQ_TYPE_* flags, I noticed that they didn't match the SA_TRIGGER_* flags for
_FALLINGor_RISING. The following patch swaps the two and make OMAP happy. It's now booting, and things seem sane. Will do some more intensive testingKevin
Sigh. I knew I'd screw that one up.
Applied and released -rc3-armirq2.
Thanks,
tglx
Quoted 6 lines
Index:
linux-omap-2.6/include/linux/irq.h
===================================================================
include/linux/irq.h line 46
Quoted 10 lines
include/linux/irq.h +2
-2@@ -39,8 +39,8 @@* IRQ types, see also include/linux/interrupt.h */ #define IRQ_TYPE_NONE 0x0000/* Default, unspecified type */ -#define IRQ_TYPE_EDGE_FALLING 0x0001/* Edge falling type */ -#define IRQ_TYPE_EDGE_RISING 0x0002/* Edge rising type */ +#define IRQ_TYPE_EDGE_RISING 0x0001/* Edge rising type */ +#define IRQ_TYPE_EDGE_FALLING 0x0002/* Edge falling type */ #define IRQ_TYPE_EDGE_BOTH (IRQ_TYPE_EDGE_FALLING | IRQ_TYPE_EDGE_RISING)#define IRQ_TYPE_LEVEL_HIGH 0x0004/* Level high type */ #define IRQ_TYPE_LEVEL_LOW 0x0008/* Level low type */
-------------------------------------------------------------------
List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel FAQ: http://www.arm.linux.org.uk/mailinglists/faq.php Etiquette: http://www.arm.linux.org.uk/mailinglists/etiquette.php
Quoted 8 lines · Thomas Gleixner
Quoted 7 lines · Kevin Hilman
ok, I tried with -rc3 and I still dont see any GPIO interrupts.
After looking closer at the IRQ_TYPE_* flags, I noticed that they didn't match the SA_TRIGGER_* flags for
_FALLINGor_RISING. The following patch swaps the two and make OMAP happy. It's now booting, and things seem sane. Will do some more intensive testing
Quoted 5 lines
Quoted 1 line
Kevin
Sigh. I knew I'd screw that one up.
Applied and released -rc3-armirq2.
Patches seem to work for me too on various omap boards. Also dyntick works.
Should this part in kernel/irq/handle.c be CONFIG_ARM for now?
@@ -84,6 +229,15 @@ fastcall int handle_IRQ_event(unsigned i
if (!(action->flags & SA_INTERRUPT))
local_irq_enable();
+#if defined(CONFIG_NO_IDLE_HZ)
+ if (!(action->flags & SA_TIMER) && system_timer->dyn_tick !=
NULL) {
+ write_seqlock(&xtime_lock);
+ if (system_timer->dyn_tick->state & DYN_TICK_ENABLED)
+ system_timer->dyn_tick->handler(irq, 0, regs);
+ write_sequnlock(&xtime_lock);
+ }
+#endif
+
do {
ret = action->handler(irq, action->dev_id, regs); if (ret == IRQ_HANDLED)
Regards,
Tony
Quoted 4 lines · Tony Lindgren
Quoted 1 line
Applied and released -rc3-armirq2.
Patches seem to work for me too on various omap boards. Also dyntick works.
Thanks for testing.
Quoted 1 line
Should this part in
kernel/irq/handle.cbeCONFIG_ARMfor now?
Yes, until we find a unified solution for dynamic ticks.
tglx
-------------------------------------------------------------------
List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel FAQ: http://www.arm.linux.org.uk/mailinglists/faq.php Etiquette: http://www.arm.linux.org.uk/mailinglists/etiquette.php
Everything appears to be working on the IOP3xx boards. There is one thing I would like to confirm. None of the patches changed anything in the mach-iop3xx directory. Was this because no changes were required, or is our arch not actually using the new mechanism at this point?
-Adam Brooks
Quoted 17 lines · Thomas Gleixner
Quoted 4 lines · Tony Lindgren
Quoted 1 line
Applied and released -rc3-armirq2.
Patches seem to work for me too on various omap boards. Also dyntick works.
Thanks for testing.
Quoted 1 line
Should this part in
kernel/irq/handle.cbeCONFIG_ARMfor now?Yes, until we find a unified solution for dynamic ticks.
tglx
Quoted 5 lines
-------------------------------------------------------------------
List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel FAQ: http://www.arm.linux.org.uk/mailinglists/faq.php Etiquette: http://www.arm.linux.org.uk/mailinglists/etiquette.php
-------------------------------------------------------------------
List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel FAQ: http://www.arm.linux.org.uk/mailinglists/faq.php Etiquette: http://www.arm.linux.org.uk/mailinglists/etiquette.php