diff mbox

ARM: shmobile: r8a7790: skip set_wake for gic irq.

Message ID 1430737448-29223-1-git-send-email-ahaslam@baylibre.com
State Rejected
Delegated to: Simon Horman
Headers show

Commit Message

ahaslam@baylibre.com May 4, 2015, 11:04 a.m. UTC
From: Axel Haslam <ahaslam+renesas@baylibre.com>

when doing a suspend to ram with r8a7790 a warning starts appearing
since: commit ab82fa7da4dc ("gpio: rcar: Prevent module clock disable
when wake-up is enabled")

the call trace that ends up in the waring is:
	gpio_rcar_irq_set_wake()
		irq_set_irq_wake(p->irq_parent, on);
			gic_set_wake()

but since gic_arch_extn.irq_set_wake is not set, this returns
an error and  desc->wake_depth is set to 0, which generates warnings
on trying to disable the wake upon resume.

To avoid this warning set the IRQCHIP_SKIP_SET_WAKE flag for the gic
irq's. With this flag set_irq_wake_real will avoid returning
errors and the calls will be balanced.

usb usb1: root hub lost power or was reset
WARNING: CPU: 1 PID: 931 at kernel/irq/manage.c:572 irq_set_irq_wake+0x9c
Unbalanced IRQ 22 wake disable
CPU: 1 PID: 931 Comm: sh Not tainted 4.1.0-rc1-dirty #91
Hardware name: Generic R8A7790 (Flattened Device Tree)
Backtrace:
[<c0012f50>] (dump_backtrace) from [<c00130f8>] (show_stack+0x18/0x1c)
 r7:c05abf59 r6:ee161ac0 r5:00000009 r4:00000000
[<c00130e0>] (show_stack) from [<c049c714>] (dump_stack+0x78/0x94)
[<c049c69c>] (dump_stack) from [<c0027248>] (warn_slowpath_common+0x88/0xb4)
 r5:00000009 r4:ed8dfc98
[<c00271c0>] (warn_slowpath_common) from [<c00272e0>] (warn_slowpath_fmt+0x38/0x40)
 r9:00000000 r8:00000000 r7:00000010 r6:00000016 r5:ee095640 r4:00000000
[<c00272ac>] (warn_slowpath_fmt) from [<c005fbe8>] (irq_set_irq_wake+0x9c/0xf8)
 r3:00000016 r2:c05ac08a
[<c005fb4c>] (irq_set_irq_wake) from [<c01d34f0>] (gpio_rcar_irq_set_wake+0x1c/0x44)
 r7:00000010 r6:000000a6 r5:00000000 r4:ee276a30
[<c01d34d4>] (gpio_rcar_irq_set_wake) from [<c005ebe8>] (set_irq_wake_real+0x38/0x4c)
 r5:ee1177c0 r4:00000000
[<c005ebb0>] (set_irq_wake_real) from [<c005fc04>] (irq_set_irq_wake+0xb8/0xf8)
 r5:ee1177c0 r4:00000001
[<c005fb4c>] (irq_set_irq_wake) from [<c030a674>] (gpio_keys_resume+0x60/0xb8)
 r7:00000010 r6:00000000 r5:edba2c10 r4:edba2c10
[<c030a614>] (gpio_keys_resume) from [<c0261288>] (platform_pm_resume+0x48/0x54)
 r7:00000010 r6:ee28cc44 r5:ee28cc10 r4:c0261240
[<c0261240>] (platform_pm_resume) from [<c02699e0>] (dpm_run_callback.isra.10+0x24/0x50)
[<c02699bc>] (dpm_run_callback.isra.10) from [<c026a1c4>] (device_resume+0x10c/0x134)
 r5:ee28cc10 r4:00000001
[<c026a0b8>] (device_resume) from [<c026a2c0>] (dpm_resume+0xd4/0x200)
 r9:00000000 r8:00000010 r7:ee28cc10 r6:c0667cd8 r5:c06a1a44 r4:ee28cc78
[<c026a1ec>] (dpm_resume) from [<c026a400>] (dpm_resume_end+0x14/0x20)
 r10:00000000 r9:c067f6ec r8:c05a83f3 r7:00000000 r6:00000003 r5:c067f6ec
 r4:00000010
[<c026a3ec>] (dpm_resume_end) from [<c005ac14>] (suspend_devices_and_enter+0x230/0x3f4)
 r5:c067f6ec r4:00000000
[<c005a9e4>] (suspend_devices_and_enter) from [<c005ae84>] (pm_suspend+0xac/0x1c0)
 r9:c067f6f4 r8:c05a83f3 r7:edb6fa00 r6:00000003 r5:c06a1a48 r4:00000000
[<c005add8>] (pm_suspend) from [<c0059d74>] (state_store+0xb0/0xc4)
 r7:edb6fa00 r6:00000004 r5:00000003 r4:00000003
[<c0059cc4>] (state_store) from [<c01aa868>] (kobj_attr_store+0x1c/0x28)
 r9:000beb08 r8:ed8dff80 r7:ee10ee4c r6:ee10ee40 r5:edb6fa00 r4:00000004
[<c01aa84c>] (kobj_attr_store) from [<c011e970>] (sysfs_kf_write+0x40/0x4c)
[<c011e930>] (sysfs_kf_write) from [<c011dc24>] (kernfs_fop_write+0x120/0x180)
[<c011db04>] (kernfs_fop_write) from [<c00c8f80>] (__vfs_write+0x34/0xdc)
 r9:ed8de000 r8:c000fd44 r7:00000004 r6:ed8dff80 r5:c011db04 r4:edba9700
[<c00c8f4c>] (__vfs_write) from [<c00c91a4>] (vfs_write+0xb4/0x13c)
 r7:ed8dff80 r6:000beb08 r5:edba9700 r4:00000004
[<c00c90f0>] (vfs_write) from [<c00c930c>] (SyS_write+0x50/0x90)
 r9:ed8de000 r8:c000fd44 r7:00000004 r6:000beb08 r5:edba9700 r4:edba9700
[<c00c92bc>] (SyS_write) from [<c000fba0>] (ret_fast_syscall+0x0/0x34)
 r7:00000004 r6:000beb08 r5:00000001 r4:000bba10
---[ end trace 1c19f00acedc2bc7 ]---

Signed-off-by: Axel Haslam <ahaslam+renesas@baylibre.com>
---
 arch/arm/mach-shmobile/setup-r8a7790.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

Comments

Geert Uytterhoeven May 4, 2015, 12:30 p.m. UTC | #1
Hi Axel,

On Mon, May 4, 2015 at 1:04 PM,  <ahaslam@baylibre.com> wrote:
> From: Axel Haslam <ahaslam+renesas@baylibre.com>
>
> when doing a suspend to ram with r8a7790 a warning starts appearing
> since: commit ab82fa7da4dc ("gpio: rcar: Prevent module clock disable
> when wake-up is enabled")
>
> the call trace that ends up in the waring is:
>         gpio_rcar_irq_set_wake()
>                 irq_set_irq_wake(p->irq_parent, on);
>                         gic_set_wake()
>
> but since gic_arch_extn.irq_set_wake is not set, this returns
> an error and  desc->wake_depth is set to 0, which generates warnings
> on trying to disable the wake upon resume.
>
> To avoid this warning set the IRQCHIP_SKIP_SET_WAKE flag for the gic
> irq's. With this flag set_irq_wake_real will avoid returning
> errors and the calls will be balanced.

I tried to handle this in the GIC driver:
http://lists.infradead.org/pipermail/linux-arm-kernel/2015-March/331763.html
but that was rejected by Marc Zyngier:
http://lists.infradead.org/pipermail/linux-arm-kernel/2015-March/331813.html

I wrote a patch to fix this in gpio-rcar, but forgot to send it.
Sorry for that, will do later today.

> @@ -27,6 +30,12 @@ static const char * const r8a7790_boards_compat_dt[] __initconst = {
>         NULL,
>  };
>
> +void __init r8a7790_init_irq(void)
> +{
> +       gic_arch_extn.flags = IRQCHIP_SKIP_SET_WAKE;
> +       irqchip_init();
> +}

Please note that gic_arch_extn is cheduled for removal, cfr. "[PATCH] irqchip:
gic: Drop support for gic_arch_extn"
(http://www.spinics.net/lists/arm-kernel/msg413684.html)

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
--
To unsubscribe from this list: send the line "unsubscribe linux-sh" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
ahaslam@baylibre.com May 4, 2015, 12:54 p.m. UTC | #2
Hi Geert,

On Mon, May 4, 2015 at 2:30 PM, Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> Hi Axel,
>
> On Mon, May 4, 2015 at 1:04 PM,  <ahaslam@baylibre.com> wrote:
>> From: Axel Haslam <ahaslam+renesas@baylibre.com>
>>
>> when doing a suspend to ram with r8a7790 a warning starts appearing
>> since: commit ab82fa7da4dc ("gpio: rcar: Prevent module clock disable
>> when wake-up is enabled")
>>
>> the call trace that ends up in the waring is:
>>         gpio_rcar_irq_set_wake()
>>                 irq_set_irq_wake(p->irq_parent, on);
>>                         gic_set_wake()
>>
>> but since gic_arch_extn.irq_set_wake is not set, this returns
>> an error and  desc->wake_depth is set to 0, which generates warnings
>> on trying to disable the wake upon resume.
>>
>> To avoid this warning set the IRQCHIP_SKIP_SET_WAKE flag for the gic
>> irq's. With this flag set_irq_wake_real will avoid returning
>> errors and the calls will be balanced.
>
> I tried to handle this in the GIC driver:
> http://lists.infradead.org/pipermail/linux-arm-kernel/2015-March/331763.html
> but that was rejected by Marc Zyngier:
> http://lists.infradead.org/pipermail/linux-arm-kernel/2015-March/331813.html
>
> I wrote a patch to fix this in gpio-rcar, but forgot to send it.
> Sorry for that, will do later today.
>
>> @@ -27,6 +30,12 @@ static const char * const r8a7790_boards_compat_dt[] __initconst = {
>>         NULL,
>>  };
>>
>> +void __init r8a7790_init_irq(void)
>> +{
>> +       gic_arch_extn.flags = IRQCHIP_SKIP_SET_WAKE;
>> +       irqchip_init();
>> +}
>
> Please note that gic_arch_extn is cheduled for removal, cfr. "[PATCH] irqchip:
> gic: Drop support for gic_arch_extn"
> (http://www.spinics.net/lists/arm-kernel/msg413684.html)


I  missed your thread with Marc,
sorry for the noise.

Regards,
Axel

>
> Gr{oetje,eeting}s,
>
>                         Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
>                                 -- Linus Torvalds
--
To unsubscribe from this list: send the line "unsubscribe linux-sh" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/arch/arm/mach-shmobile/setup-r8a7790.c b/arch/arm/mach-shmobile/setup-r8a7790.c
index 3a18af4..84fa279 100644
--- a/arch/arm/mach-shmobile/setup-r8a7790.c
+++ b/arch/arm/mach-shmobile/setup-r8a7790.c
@@ -15,6 +15,9 @@ 
  */
 
 #include <linux/init.h>
+#include <linux/irq.h>
+#include <linux/irqchip.h>
+#include <linux/irqchip/arm-gic.h>
 
 #include <asm/mach/arch.h>
 
@@ -27,6 +30,12 @@  static const char * const r8a7790_boards_compat_dt[] __initconst = {
 	NULL,
 };
 
+void __init r8a7790_init_irq(void)
+{
+	gic_arch_extn.flags = IRQCHIP_SKIP_SET_WAKE;
+	irqchip_init();
+}
+
 DT_MACHINE_START(R8A7790_DT, "Generic R8A7790 (Flattened Device Tree)")
 	.smp		= smp_ops(r8a7790_smp_ops),
 	.init_early	= shmobile_init_delay,
@@ -34,4 +43,5 @@  DT_MACHINE_START(R8A7790_DT, "Generic R8A7790 (Flattened Device Tree)")
 	.init_late	= shmobile_init_late,
 	.reserve	= rcar_gen2_reserve,
 	.dt_compat	= r8a7790_boards_compat_dt,
+	.init_irq	= r8a7790_init_irq,
 MACHINE_END