2013-05-07 06:30:58

by EUNBONG SONG

Subject: MIPS: Test reault for enable interrupts before WAIT instruction patch

Hello. I tested with two patches.
The first one is thomas gleixner's patch. The patch is as follow.
This patch works well without any problem.

Index: linux-2.6/arch/mips/kernel/process.c
===================================================================
--- linux-2.6.orig/arch/mips/kernel/process.c
+++ linux-2.6/arch/mips/kernel/process.c
@@ -50,13 +50,18 @@ void arch_cpu_idle_dead(void)
}
#endif

-void arch_cpu_idle(void)
+static void smtc_idle_hook(void)
{
#ifdef CONFIG_MIPS_MT_SMTC
extern void smtc_idle_loop_hook(void);
-
smtc_idle_loop_hook();
#endif
+}
+
+void arch_cpu_idle(void)
+{
+ local_irq_enable();
+ smtc_idle_hook();
if (cpu_wait)
(*cpu_wait)();
else
--

The second one is david daney's patch. The patch is as follow.
arch/mips/kernel/genex.S | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/arch/mips/kernel/genex.S b/arch/mips/kernel/genex.S
index ecb347c..57cda9a 100644
--- a/arch/mips/kernel/genex.S
+++ b/arch/mips/kernel/genex.S
@@ -132,12 +132,13 @@ LEAF(r4k_wait)
.set noreorder
/* start of rollback region */
LONG_L t0, TI_FLAGS($28)
- nop
andi t0, _TIF_NEED_RESCHED
bnez t0, 1f
nop
- nop
- nop
+ /* Enable interrupts so WAIT will complete */
+ mfc0 t0, CP0_STATUS
+ ori t0, ST0_IE
+ mtc0 t0, CP0_STATUS
.set mips3
wait
/* end of rollback region (the region size must be power of two) */

After apply this patch. I got two error message.
The first one is as follow
[ 124.661211] Checking for the daddi bug... no.
[ 124.665737] ------------[ cut here ]------------
[ 124.670187] WARNING: at kernel/cpu/idle.c:96 cpu_startup_entry+0x150/0x178()
[ 124.677209] Modules linked in:
[ 124.680251] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.9.0+ #40
[ 124.686237] Stack : 0000000000000004 0000000000000034 ffffffff80fa0000 ffffffff80292558
0000000000000000 ffffffff80fa0000 0000000000000001 ffffffff80293810
0000000000000000 0000000000000000 ffffffff81080000 ffffffff81080000
ffffffff80e2acf0 ffffffff80f8f977 ffffffff80f8fa80 ffffffff80e31730
0000000000000001 0000000000000004 ffffffff00000000 0000000000000004
ffffffffc05633f0 ffffffff806ef728 ffffffff80f57d08 ffffffff80290a74
ffffffffc05633f0 ffffffff80293c40 000000000000003e ffffffff80e2acf0
0000000000000000 ffffffff80f57c30 00ffffff80f8fdc0 ffffffff802908c0
0000000000000000 0000000000000000 0000000000000000 0000000000000000
0000000000000000 ffffffff80272498 0000000000000000 0000000000000000
...
[ 124.751163] Call Trace:
[ 124.753599] [<ffffffff80272498>] show_stack+0x68/0x80
[ 124.758634] [<ffffffff802908c0>] warn_slowpath_common+0x78/0xa8
[ 124.764533] [<ffffffff802d4448>] cpu_startup_entry+0x150/0x178
[ 124.770351] [<ffffffff80fd6b04>] start_kernel+0x440/0x45c
[ 124.775728]
[ 124.777219] ---[ end trace 9179e654e5693e72 ]---

After boot process is done the follow error message is printed periodically.

[ 284.751007] INFO: rcu_preempt detected stalls on CPUs/tasks: { 6} (detected by 1, t=14712 jiffies, g=18446744073709551344, c=18446744073709551343, q=2437)
[ 284.764878] Task dump for CPU 6:
[ 284.768105] swapper/6 R running task 0 0 1 0x00100000
[ 284.775174] Stack : 0000005311112000 ffffffff80f60000 ffffffff80f60000 a800000001d2d950
0000000000000018 ffffffff81080000 ffffffff81080000 0000000000000000
ffffffff81010000 ffffffff8030893c 4256e5715da6083d 800000040f800000
0000000000000018 ffffffff81080000 ffffffff81080000 ffffffff80264f3c
ffffffff80e31730 0000000000000000 ffffffff80e31730 ffffffff80f90000
ffffffff80fd0000 ffffffff8026c760 0000000000000000 0000000010008ce1
0000000000100000 a8000000414e4010 ffffffff80f8bb18 0000000000000000
0000005311112000 0000000000000001 0000000000000001 0000000000000000
ffffffff80f8bc58 a800000001d32c60 a8000000414e7fe0 0000000000008c00
a80000003f7d8000 0000000000000000 ffffffff80fd0000 ffffffff80e31730
...
[ 284.840704] Call Trace:
[ 284.843153] [<ffffffff806f1a48>] __schedule+0x3b0/0x938
[ 284.848377]


Thanks????{.n?+???????+%?????ݶ??w??{.n?+????{??G?????{ay?ʇڙ?,j??f???h?????????z_??(?階?ݢj"???m??????G????????????&???~???iO???z??v?^?m????????????I?

2013-05-07 17:07:33

by David Daney

Subject: Re: MIPS: Test reault for enable interrupts before WAIT instruction patch

On 05/06/2013 11:30 PM, EUNBONG SONG wrote:
>
> Hello. I tested with two patches.
> The first one is thomas gleixner's patch. The patch is as follow.
> This patch works well without any problem.

You don't need them both. When we fix it, it will be one or the other,
not both.


>
> Index: linux-2.6/arch/mips/kernel/process.c
> ===================================================================
> --- linux-2.6.orig/arch/mips/kernel/process.c
> +++ linux-2.6/arch/mips/kernel/process.c
> @@ -50,13 +50,18 @@ void arch_cpu_idle_dead(void)
> }
> #endif
>
> -void arch_cpu_idle(void)
> +static void smtc_idle_hook(void)
> {
> #ifdef CONFIG_MIPS_MT_SMTC
> extern void smtc_idle_loop_hook(void);
> -
> smtc_idle_loop_hook();
> #endif
> +}
> +
> +void arch_cpu_idle(void)
> +{
> + local_irq_enable();
> + smtc_idle_hook();
> if (cpu_wait)
> (*cpu_wait)();
> else

Although I wrote the other patch, I now think that Thomas Gleixner's
patch (above), is the best option.

There are many cpu_wait() implementations. Fixing a bunch of different
assembly implementations, most of which I cannot test, is a recipe for
disaster.


> --
>
> The second one is david daney's patch. The patch is as follow.
> arch/mips/kernel/genex.S | 7 ++++---
> 1 file changed, 4 insertions(+), 3 deletions(-)
>
> diff --git a/arch/mips/kernel/genex.S b/arch/mips/kernel/genex.S
> index ecb347c..57cda9a 100644
> --- a/arch/mips/kernel/genex.S
> +++ b/arch/mips/kernel/genex.S
> @@ -132,12 +132,13 @@ LEAF(r4k_wait)
> .set noreorder
> /* start of rollback region */
> LONG_L t0, TI_FLAGS($28)
> - nop
> andi t0, _TIF_NEED_RESCHED
> bnez t0, 1f
> nop
> - nop
> - nop
> + /* Enable interrupts so WAIT will complete */
> + mfc0 t0, CP0_STATUS
> + ori t0, ST0_IE
> + mtc0 t0, CP0_STATUS
> .set mips3
> wait
> /* end of rollback region (the region size must be power of two) */
>
> After apply this patch. I got two error message.
> The first one is as follow
> [ 124.661211] Checking for the daddi bug... no.
> [ 124.665737] ------------[ cut here ]------------
> [ 124.670187] WARNING: at kernel/cpu/idle.c:96 cpu_startup_entry+0x150/0x178()
> [ 124.677209] Modules linked in:
> [ 124.680251] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.9.0+ #40
> [ 124.686237] Stack : 0000000000000004 0000000000000034 ffffffff80fa0000 ffffffff80292558
> 0000000000000000 ffffffff80fa0000 0000000000000001 ffffffff80293810
> 0000000000000000 0000000000000000 ffffffff81080000 ffffffff81080000
> ffffffff80e2acf0 ffffffff80f8f977 ffffffff80f8fa80 ffffffff80e31730
> 0000000000000001 0000000000000004 ffffffff00000000 0000000000000004
> ffffffffc05633f0 ffffffff806ef728 ffffffff80f57d08 ffffffff80290a74
> ffffffffc05633f0 ffffffff80293c40 000000000000003e ffffffff80e2acf0
> 0000000000000000 ffffffff80f57c30 00ffffff80f8fdc0 ffffffff802908c0
> 0000000000000000 0000000000000000 0000000000000000 0000000000000000
> 0000000000000000 ffffffff80272498 0000000000000000 0000000000000000
> ...
> [ 124.751163] Call Trace:
> [ 124.753599] [<ffffffff80272498>] show_stack+0x68/0x80
> [ 124.758634] [<ffffffff802908c0>] warn_slowpath_common+0x78/0xa8
> [ 124.764533] [<ffffffff802d4448>] cpu_startup_entry+0x150/0x178
> [ 124.770351] [<ffffffff80fd6b04>] start_kernel+0x440/0x45c
> [ 124.775728]
> [ 124.777219] ---[ end trace 9179e654e5693e72 ]---
>
> After boot process is done the follow error message is printed periodically.
>
> [ 284.751007] INFO: rcu_preempt detected stalls on CPUs/tasks: { 6} (detected by 1, t=14712 jiffies, g=18446744073709551344, c=18446744073709551343, q=2437)
> [ 284.764878] Task dump for CPU 6:
> [ 284.768105] swapper/6 R running task 0 0 1 0x00100000
> [ 284.775174] Stack : 0000005311112000 ffffffff80f60000 ffffffff80f60000 a800000001d2d950
> 0000000000000018 ffffffff81080000 ffffffff81080000 0000000000000000
> ffffffff81010000 ffffffff8030893c 4256e5715da6083d 800000040f800000
> 0000000000000018 ffffffff81080000 ffffffff81080000 ffffffff80264f3c
> ffffffff80e31730 0000000000000000 ffffffff80e31730 ffffffff80f90000
> ffffffff80fd0000 ffffffff8026c760 0000000000000000 0000000010008ce1
> 0000000000100000 a8000000414e4010 ffffffff80f8bb18 0000000000000000
> 0000005311112000 0000000000000001 0000000000000001 0000000000000000
> ffffffff80f8bc58 a800000001d32c60 a8000000414e7fe0 0000000000008c00
> a80000003f7d8000 0000000000000000 ffffffff80fd0000 ffffffff80e31730
> ...
> [ 284.840704] Call Trace:
> [ 284.843153] [<ffffffff806f1a48>] __schedule+0x3b0/0x938
> [ 284.848377]
>
>
> Thanks
>

2013-05-07 23:12:28

by EUNBONG SONG

Subject: Re: Re: MIPS: Test reault for enable interrupts before WAIT instruction patch

>>
>> Hello. I tested with two patches.
>> The first one is thomas gleixner's patch. The patch is as follow.
>> This patch works well without any problem.

>You don't need them both. When we fix it, it will be one or the other,
>not both.

Hello. David.
Maybe i didn't write clearly for test result. I meant i applied only one patche for each test.

Thanks.

????{.n?+???????+%?????ݶ??w??{.n?+????{??G?????{ay?ʇڙ?,j??f???h?????????z_??(?階?ݢj"???m??????G????????????&???~???iO???z??v?^?m????????????I?