Patchwork powerpc/time: When starting the decrementer don't zero the other bits in TCR

login
register
mail settings
Submitter Laurentiu TUDOR
Date Aug. 25, 2011, 12:19 p.m.
Message ID <1314274795-4966-1-git-send-email-Laurentiu.Tudor@freescale.com>
Download mbox | patch
Permalink /patch/111563/
State Changes Requested
Delegated to: Benjamin Herrenschmidt
Headers show

Comments

Laurentiu TUDOR - Aug. 25, 2011, 12:19 p.m.
Clearing the other TCR bits might break code that sets them (e.g. to setup
the watchdog or fixed interval timer) before start_cpu_decrementer() gets
called.

Signed-off-by: Laurentiu Tudor <Laurentiu.Tudor@freescale.com>
---
 arch/powerpc/kernel/time.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
Kumar Gala - Sept. 1, 2011, 6:27 a.m.
On Aug 25, 2011, at 7:19 AM, Laurentiu Tudor wrote:

> Clearing the other TCR bits might break code that sets them (e.g. to setup
> the watchdog or fixed interval timer) before start_cpu_decrementer() gets
> called.

Can you be more specific about the case you're concerned about.  I'm not sure we want to leave the FIT enabled if it was previously.  Watchdog I could possibly see leaving alone.
> 
> Signed-off-by: Laurentiu Tudor <Laurentiu.Tudor@freescale.com>
> ---
> arch/powerpc/kernel/time.c |    2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c
> index 03b29a6..e8b5cdc 100644
> --- a/arch/powerpc/kernel/time.c
> +++ b/arch/powerpc/kernel/time.c
> @@ -721,7 +721,7 @@ void start_cpu_decrementer(void)
> 	mtspr(SPRN_TSR, TSR_ENW | TSR_WIS | TSR_DIS | TSR_FIS);
> 
> 	/* Enable decrementer interrupt */
> -	mtspr(SPRN_TCR, TCR_DIE);
> +	mtspr(SPRN_TCR, mfspr(SPRN_TCR) | TCR_DIE);
> #endif /* defined(CONFIG_BOOKE) || defined(CONFIG_40x) */
> }
> 
> -- 
> 1.7.1
> 
> 
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev
Laurentiu TUDOR - Sept. 1, 2011, 8:31 a.m.
On 9/1/2011 9:27 AM, Kumar Gala wrote:
>
> On Aug 25, 2011, at 7:19 AM, Laurentiu Tudor wrote:
>
>> Clearing the other TCR bits might break code that sets them (e.g. to setup
>> the watchdog or fixed interval timer) before start_cpu_decrementer() gets
>> called.
>
> Can you be more specific about the case you're concerned about.  I'm not sure we want to leave the FIT enabled if it was previously.  Watchdog I could possibly see leaving alone.

My particular case was this: I was experimenting with having the 
watchdog enabled during the boot process, in case something goes wrong 
before the normal watchdog mechanism kicks in. That didn't work without 
making this modification.
Then, thinking that having a function start_cpu_decrementer() touching 
the other timers and not only the decrementer as its name implies might 
not be good. So I submitted this patch to the public mailing list to let 
the community decide if it's useful.

>> Signed-off-by: Laurentiu Tudor<Laurentiu.Tudor@freescale.com>
>> ---
>> arch/powerpc/kernel/time.c |    2 +-
>> 1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c
>> index 03b29a6..e8b5cdc 100644
>> --- a/arch/powerpc/kernel/time.c
>> +++ b/arch/powerpc/kernel/time.c
>> @@ -721,7 +721,7 @@ void start_cpu_decrementer(void)
>> 	mtspr(SPRN_TSR, TSR_ENW | TSR_WIS | TSR_DIS | TSR_FIS);
>>
>> 	/* Enable decrementer interrupt */
>> -	mtspr(SPRN_TCR, TCR_DIE);
>> +	mtspr(SPRN_TCR, mfspr(SPRN_TCR) | TCR_DIE);
>> #endif /* defined(CONFIG_BOOKE) || defined(CONFIG_40x) */
>> }
>>
>> --
>> 1.7.1
>>
>>
>> _______________________________________________
>> Linuxppc-dev mailing list
>> Linuxppc-dev@lists.ozlabs.org
>> https://lists.ozlabs.org/listinfo/linuxppc-dev
>
>
Kumar Gala - Sept. 1, 2011, 1:56 p.m.
On Sep 1, 2011, at 3:31 AM, Tudor Laurentiu wrote:

> On 9/1/2011 9:27 AM, Kumar Gala wrote:
>> 
>> On Aug 25, 2011, at 7:19 AM, Laurentiu Tudor wrote:
>> 
>>> Clearing the other TCR bits might break code that sets them (e.g. to setup
>>> the watchdog or fixed interval timer) before start_cpu_decrementer() gets
>>> called.
>> 
>> Can you be more specific about the case you're concerned about.  I'm not sure we want to leave the FIT enabled if it was previously.  Watchdog I could possibly see leaving alone.
> 
> My particular case was this: I was experimenting with having the watchdog enabled during the boot process, in case something goes wrong before the normal watchdog mechanism kicks in. That didn't work without making this modification.
> Then, thinking that having a function start_cpu_decrementer() touching the other timers and not only the decrementer as its name implies might not be good. So I submitted this patch to the public mailing list to let the community decide if it's useful.

So I think renaming the function is fine, and I think not touching the watchdog bits is also reasonable.  So if you update the patch along those lines I think we can push in.  I do want to make sure that the FIT is disabled as we dont expect it.

Probably add a comment about how care needs to be taken if the wathchdog is enable be firmware or a previous kernel.

> 
>>> Signed-off-by: Laurentiu Tudor<Laurentiu.Tudor@freescale.com>
>>> ---
>>> arch/powerpc/kernel/time.c |    2 +-
>>> 1 files changed, 1 insertions(+), 1 deletions(-)
>>> 
>>> diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c
>>> index 03b29a6..e8b5cdc 100644
>>> --- a/arch/powerpc/kernel/time.c
>>> +++ b/arch/powerpc/kernel/time.c
>>> @@ -721,7 +721,7 @@ void start_cpu_decrementer(void)
>>> 	mtspr(SPRN_TSR, TSR_ENW | TSR_WIS | TSR_DIS | TSR_FIS);
>>> 
>>> 	/* Enable decrementer interrupt */
>>> -	mtspr(SPRN_TCR, TCR_DIE);
>>> +	mtspr(SPRN_TCR, mfspr(SPRN_TCR) | TCR_DIE);
>>> #endif /* defined(CONFIG_BOOKE) || defined(CONFIG_40x) */
>>> }
>>> 
>>> --
>>> 1.7.1
>>> 
>>> 
>>> _______________________________________________
>>> Linuxppc-dev mailing list
>>> Linuxppc-dev@lists.ozlabs.org
>>> https://lists.ozlabs.org/listinfo/linuxppc-dev
>> 
>> 
> 
>
Laurentiu TUDOR - Sept. 1, 2011, 3:38 p.m.
On 9/1/2011 4:56 PM, Kumar Gala wrote:
>
> On Sep 1, 2011, at 3:31 AM, Tudor Laurentiu wrote:
>
>> On 9/1/2011 9:27 AM, Kumar Gala wrote:
>>>
>>> On Aug 25, 2011, at 7:19 AM, Laurentiu Tudor wrote:
>>>
>>>> Clearing the other TCR bits might break code that sets them (e.g. to setup
>>>> the watchdog or fixed interval timer) before start_cpu_decrementer() gets
>>>> called.
>>>
>>> Can you be more specific about the case you're concerned about.  I'm not sure we want to leave the FIT enabled if it was previously.  Watchdog I could possibly see leaving alone.
>>
>> My particular case was this: I was experimenting with having the watchdog enabled during the boot process, in case something goes wrong before the normal watchdog mechanism kicks in. That didn't work without making this modification.
>> Then, thinking that having a function start_cpu_decrementer() touching the other timers and not only the decrementer as its name implies might not be good. So I submitted this patch to the public mailing list to let the community decide if it's useful.
>
> So I think renaming the function is fine, and I think not touching the watchdog bits is also reasonable.  So if you update the patch along those lines I think we can push in.  I do want to make sure that the FIT is disabled as we dont expect it.

how about:

static void setup_cpu_timers()

?

> Probably add a comment about how care needs to be taken if the wathchdog is enable be firmware or a previous kernel.

OK.
Kumar Gala - Sept. 1, 2011, 6:49 p.m.
On Sep 1, 2011, at 10:38 AM, Tudor Laurentiu wrote:

> On 9/1/2011 4:56 PM, Kumar Gala wrote:
>> 
>> On Sep 1, 2011, at 3:31 AM, Tudor Laurentiu wrote:
>> 
>>> On 9/1/2011 9:27 AM, Kumar Gala wrote:
>>>> 
>>>> On Aug 25, 2011, at 7:19 AM, Laurentiu Tudor wrote:
>>>> 
>>>>> Clearing the other TCR bits might break code that sets them (e.g. to setup
>>>>> the watchdog or fixed interval timer) before start_cpu_decrementer() gets
>>>>> called.
>>>> 
>>>> Can you be more specific about the case you're concerned about.  I'm not sure we want to leave the FIT enabled if it was previously.  Watchdog I could possibly see leaving alone.
>>> 
>>> My particular case was this: I was experimenting with having the watchdog enabled during the boot process, in case something goes wrong before the normal watchdog mechanism kicks in. That didn't work without making this modification.
>>> Then, thinking that having a function start_cpu_decrementer() touching the other timers and not only the decrementer as its name implies might not be good. So I submitted this patch to the public mailing list to let the community decide if it's useful.
>> 
>> So I think renaming the function is fine, and I think not touching the watchdog bits is also reasonable.  So if you update the patch along those lines I think we can push in.  I do want to make sure that the FIT is disabled as we dont expect it.
> 
> how about:
> 
> static void setup_cpu_timers()
> 
> ?

Sounds fine to me

> 
>> Probably add a comment about how care needs to be taken if the wathchdog is enable be firmware or a previous kernel.
> 
> OK.
> 
> 
> 
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev

Patch

diff --git a/arch/powerpc/kernel/time.c b/arch/powerpc/kernel/time.c
index 03b29a6..e8b5cdc 100644
--- a/arch/powerpc/kernel/time.c
+++ b/arch/powerpc/kernel/time.c
@@ -721,7 +721,7 @@  void start_cpu_decrementer(void)
 	mtspr(SPRN_TSR, TSR_ENW | TSR_WIS | TSR_DIS | TSR_FIS);
 
 	/* Enable decrementer interrupt */
-	mtspr(SPRN_TCR, TCR_DIE);
+	mtspr(SPRN_TCR, mfspr(SPRN_TCR) | TCR_DIE);
 #endif /* defined(CONFIG_BOOKE) || defined(CONFIG_40x) */
 }