sched_clock(), used by printk(), calls __USE_RTC() to know
whether to use realtime clock or timebase.
__USE_RTC() uses cpu_has_feature() which is initialised by
machine_init(). Before machine_init(), __USE_RTC() returns true,
leading to a program check exception on CPUs not having realtime
clock.
In order to be able to use printk() earlier, use feature fixup.
Feature fixups are applies in early_init(), enabling the use of
printk() earlier.
Signed-off-by: Christophe Leroy <[email protected]>
---
arch/powerpc/include/asm/time.h | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/arch/powerpc/include/asm/time.h b/arch/powerpc/include/asm/time.h
index 54f4ec1f9fab..3455cb54c333 100644
--- a/arch/powerpc/include/asm/time.h
+++ b/arch/powerpc/include/asm/time.h
@@ -42,7 +42,14 @@ struct div_result {
/* Accessor functions for the timebase (RTC on 601) registers. */
/* If one day CONFIG_POWER is added just define __USE_RTC as 1 */
#ifdef CONFIG_PPC_BOOK3S_32
-#define __USE_RTC() (cpu_has_feature(CPU_FTR_USE_RTC))
+static inline bool __USE_RTC(void)
+{
+ asm_volatile_goto(ASM_FTR_IFCLR("nop;", "b %1;", %0) ::
+ "i" (CPU_FTR_USE_RTC) :: l_use_rtc);
+ return false;
+l_use_rtc:
+ return true;
+}
#else
#define __USE_RTC() 0
#endif
--
2.13.3
Christophe Leroy <[email protected]> writes:
> sched_clock(), used by printk(), calls __USE_RTC() to know
> whether to use realtime clock or timebase.
>
> __USE_RTC() uses cpu_has_feature() which is initialised by
> machine_init(). Before machine_init(), __USE_RTC() returns true,
> leading to a program check exception on CPUs not having realtime
> clock.
>
> In order to be able to use printk() earlier, use feature fixup.
> Feature fixups are applies in early_init(), enabling the use of
> printk() earlier.
>
> Signed-off-by: Christophe Leroy <[email protected]>
> ---
> arch/powerpc/include/asm/time.h | 9 ++++++++-
> 1 file changed, 8 insertions(+), 1 deletion(-)
The other option would be just to make this a compile time decision, eg.
add CONFIG_PPC_601 and use that to gate whether we use RTC.
Given how many 601 users there are, maybe 1?, I think that would be a
simpler option and avoids complicating the code / binary for everyone
else.
cheers
> diff --git a/arch/powerpc/include/asm/time.h b/arch/powerpc/include/asm/time.h
> index 54f4ec1f9fab..3455cb54c333 100644
> --- a/arch/powerpc/include/asm/time.h
> +++ b/arch/powerpc/include/asm/time.h
> @@ -42,7 +42,14 @@ struct div_result {
> /* Accessor functions for the timebase (RTC on 601) registers. */
> /* If one day CONFIG_POWER is added just define __USE_RTC as 1 */
> #ifdef CONFIG_PPC_BOOK3S_32
> -#define __USE_RTC() (cpu_has_feature(CPU_FTR_USE_RTC))
> +static inline bool __USE_RTC(void)
> +{
> + asm_volatile_goto(ASM_FTR_IFCLR("nop;", "b %1;", %0) ::
> + "i" (CPU_FTR_USE_RTC) :: l_use_rtc);
> + return false;
> +l_use_rtc:
> + return true;
> +}
> #else
> #define __USE_RTC() 0
> #endif
> --
> 2.13.3
Le 26/08/2019 à 13:41, Michael Ellerman a écrit :
> Christophe Leroy <[email protected]> writes:
>> sched_clock(), used by printk(), calls __USE_RTC() to know
>> whether to use realtime clock or timebase.
>>
>> __USE_RTC() uses cpu_has_feature() which is initialised by
>> machine_init(). Before machine_init(), __USE_RTC() returns true,
>> leading to a program check exception on CPUs not having realtime
>> clock.
>>
>> In order to be able to use printk() earlier, use feature fixup.
>> Feature fixups are applies in early_init(), enabling the use of
>> printk() earlier.
>>
>> Signed-off-by: Christophe Leroy <[email protected]>
>> ---
>> arch/powerpc/include/asm/time.h | 9 ++++++++-
>> 1 file changed, 8 insertions(+), 1 deletion(-)
>
> The other option would be just to make this a compile time decision, eg.
> add CONFIG_PPC_601 and use that to gate whether we use RTC.
Euh ... yes OK, why not. And that would help simplify many places in the
code. I can propose something in that direction, but it will be a bigger
change.
>
> Given how many 601 users there are, maybe 1?, I think that would be a
> simpler option and avoids complicating the code / binary for everyone
> else.
However this patch doesn't complicate things more than it was with
cpu_has_feature() which does exactly the same but using static keys,
does it ?
Christophe
>
> cheers
>
>> diff --git a/arch/powerpc/include/asm/time.h b/arch/powerpc/include/asm/time.h
>> index 54f4ec1f9fab..3455cb54c333 100644
>> --- a/arch/powerpc/include/asm/time.h
>> +++ b/arch/powerpc/include/asm/time.h
>> @@ -42,7 +42,14 @@ struct div_result {
>> /* Accessor functions for the timebase (RTC on 601) registers. */
>> /* If one day CONFIG_POWER is added just define __USE_RTC as 1 */
>> #ifdef CONFIG_PPC_BOOK3S_32
>> -#define __USE_RTC() (cpu_has_feature(CPU_FTR_USE_RTC))
>> +static inline bool __USE_RTC(void)
>> +{
>> + asm_volatile_goto(ASM_FTR_IFCLR("nop;", "b %1;", %0) ::
>> + "i" (CPU_FTR_USE_RTC) :: l_use_rtc);
>> + return false;
>> +l_use_rtc:
>> + return true;
>> +}
>> #else
>> #define __USE_RTC() 0
>> #endif
>> --
>> 2.13.3
On Mon, Aug 26, 2019 at 09:41:39PM +1000, Michael Ellerman wrote:
> Given how many 601 users there are, maybe 1?, I think that would be a
> simpler option and avoids complicating the code / binary for everyone
> else.
Or you could remove 601 support altogether?
Segher
Le 26/08/2019 à 15:25, Benjamin Herrenschmidt a écrit :
> On Mon, 2019-08-26 at 21:41 +1000, Michael Ellerman wrote:
>> Christophe Leroy <[email protected]> writes:
>>> sched_clock(), used by printk(), calls __USE_RTC() to know
>>> whether to use realtime clock or timebase.
>>>
>>> __USE_RTC() uses cpu_has_feature() which is initialised by
>>> machine_init(). Before machine_init(), __USE_RTC() returns true,
>>> leading to a program check exception on CPUs not having realtime
>>> clock.
>>>
>>> In order to be able to use printk() earlier, use feature fixup.
>>> Feature fixups are applies in early_init(), enabling the use of
>>> printk() earlier.
>>>
>>> Signed-off-by: Christophe Leroy <[email protected]>
>>> ---
>>> arch/powerpc/include/asm/time.h | 9 ++++++++-
>>> 1 file changed, 8 insertions(+), 1 deletion(-)
>>
>> The other option would be just to make this a compile time decision, eg.
>> add CONFIG_PPC_601 and use that to gate whether we use RTC.
>>
>> Given how many 601 users there are, maybe 1?, I think that would be a
>> simpler option and avoids complicating the code / binary for everyone
>> else.
>
> Didn't we ditch 601 support years ago anyway ? We had workaround we
> threw out I think...
There are still workarounds for 601 in the kernel, for exemple (in
arch/powerpc/include/asm/ppc_asm.h)
/* various errata or part fixups */
#ifdef CONFIG_PPC601_SYNC_FIX
#define SYNC \
BEGIN_FTR_SECTION \
sync; \
isync; \
END_FTR_SECTION_IFSET(CPU_FTR_601)
#define SYNC_601 \
BEGIN_FTR_SECTION \
sync; \
END_FTR_SECTION_IFSET(CPU_FTR_601)
#define ISYNC_601 \
BEGIN_FTR_SECTION \
isync; \
END_FTR_SECTION_IFSET(CPU_FTR_601)
#else
#define SYNC
#define SYNC_601
#define ISYNC_601
#endif
But if you think we can get rid of 601 completely, I'm happy with that.
Christophe
>
> Cheers,
> Ben.
>
>> cheers
>>
>>> diff --git a/arch/powerpc/include/asm/time.h b/arch/powerpc/include/asm/time.h
>>> index 54f4ec1f9fab..3455cb54c333 100644
>>> --- a/arch/powerpc/include/asm/time.h
>>> +++ b/arch/powerpc/include/asm/time.h
>>> @@ -42,7 +42,14 @@ struct div_result {
>>> /* Accessor functions for the timebase (RTC on 601) registers. */
>>> /* If one day CONFIG_POWER is added just define __USE_RTC as 1 */
>>> #ifdef CONFIG_PPC_BOOK3S_32
>>> -#define __USE_RTC() (cpu_has_feature(CPU_FTR_USE_RTC))
>>> +static inline bool __USE_RTC(void)
>>> +{
>>> + asm_volatile_goto(ASM_FTR_IFCLR("nop;", "b %1;", %0) ::
>>> + "i" (CPU_FTR_USE_RTC) :: l_use_rtc);
>>> + return false;
>>> +l_use_rtc:
>>> + return true;
>>> +}
>>> #else
>>> #define __USE_RTC() 0
>>> #endif
>>> --
>>> 2.13.3
On Mon, 2019-08-26 at 21:41 +1000, Michael Ellerman wrote:
> Christophe Leroy <[email protected]> writes:
> > sched_clock(), used by printk(), calls __USE_RTC() to know
> > whether to use realtime clock or timebase.
> >
> > __USE_RTC() uses cpu_has_feature() which is initialised by
> > machine_init(). Before machine_init(), __USE_RTC() returns true,
> > leading to a program check exception on CPUs not having realtime
> > clock.
> >
> > In order to be able to use printk() earlier, use feature fixup.
> > Feature fixups are applies in early_init(), enabling the use of
> > printk() earlier.
> >
> > Signed-off-by: Christophe Leroy <[email protected]>
> > ---
> > arch/powerpc/include/asm/time.h | 9 ++++++++-
> > 1 file changed, 8 insertions(+), 1 deletion(-)
>
> The other option would be just to make this a compile time decision, eg.
> add CONFIG_PPC_601 and use that to gate whether we use RTC.
>
> Given how many 601 users there are, maybe 1?, I think that would be a
> simpler option and avoids complicating the code / binary for everyone
> else.
Didn't we ditch 601 support years ago anyway ? We had workaround we
threw out I think...
Cheers,
Ben.
> cheers
>
> > diff --git a/arch/powerpc/include/asm/time.h b/arch/powerpc/include/asm/time.h
> > index 54f4ec1f9fab..3455cb54c333 100644
> > --- a/arch/powerpc/include/asm/time.h
> > +++ b/arch/powerpc/include/asm/time.h
> > @@ -42,7 +42,14 @@ struct div_result {
> > /* Accessor functions for the timebase (RTC on 601) registers. */
> > /* If one day CONFIG_POWER is added just define __USE_RTC as 1 */
> > #ifdef CONFIG_PPC_BOOK3S_32
> > -#define __USE_RTC() (cpu_has_feature(CPU_FTR_USE_RTC))
> > +static inline bool __USE_RTC(void)
> > +{
> > + asm_volatile_goto(ASM_FTR_IFCLR("nop;", "b %1;", %0) ::
> > + "i" (CPU_FTR_USE_RTC) :: l_use_rtc);
> > + return false;
> > +l_use_rtc:
> > + return true;
> > +}
> > #else
> > #define __USE_RTC() 0
> > #endif
> > --
> > 2.13.3
Benjamin Herrenschmidt <[email protected]> writes:
> On Mon, 2019-08-26 at 21:41 +1000, Michael Ellerman wrote:
>> Christophe Leroy <[email protected]> writes:
>> > sched_clock(), used by printk(), calls __USE_RTC() to know
>> > whether to use realtime clock or timebase.
>> >
>> > __USE_RTC() uses cpu_has_feature() which is initialised by
>> > machine_init(). Before machine_init(), __USE_RTC() returns true,
>> > leading to a program check exception on CPUs not having realtime
>> > clock.
>> >
>> > In order to be able to use printk() earlier, use feature fixup.
>> > Feature fixups are applies in early_init(), enabling the use of
>> > printk() earlier.
>> >
>> > Signed-off-by: Christophe Leroy <[email protected]>
>> > ---
>> > arch/powerpc/include/asm/time.h | 9 ++++++++-
>> > 1 file changed, 8 insertions(+), 1 deletion(-)
>>
>> The other option would be just to make this a compile time decision, eg.
>> add CONFIG_PPC_601 and use that to gate whether we use RTC.
>>
>> Given how many 601 users there are, maybe 1?, I think that would be a
>> simpler option and avoids complicating the code / binary for everyone
>> else.
>
> Didn't we ditch 601 support years ago anyway ? We had workaround we
> threw out I think...
Paul said his still booted recently.
cheers
Christophe Leroy <[email protected]> writes:
> Le 26/08/2019 à 13:41, Michael Ellerman a écrit :
>> Christophe Leroy <[email protected]> writes:
>>> sched_clock(), used by printk(), calls __USE_RTC() to know
>>> whether to use realtime clock or timebase.
>>>
>>> __USE_RTC() uses cpu_has_feature() which is initialised by
>>> machine_init(). Before machine_init(), __USE_RTC() returns true,
>>> leading to a program check exception on CPUs not having realtime
>>> clock.
>>>
>>> In order to be able to use printk() earlier, use feature fixup.
>>> Feature fixups are applies in early_init(), enabling the use of
>>> printk() earlier.
>>>
>>> Signed-off-by: Christophe Leroy <[email protected]>
>>> ---
>>> arch/powerpc/include/asm/time.h | 9 ++++++++-
>>> 1 file changed, 8 insertions(+), 1 deletion(-)
>>
>> The other option would be just to make this a compile time decision, eg.
>> add CONFIG_PPC_601 and use that to gate whether we use RTC.
>
> Euh ... yes OK, why not. And that would help simplify many places in the
> code. I can propose something in that direction, but it will be a bigger
> change.
Thanks.
>> Given how many 601 users there are, maybe 1?, I think that would be a
>> simpler option and avoids complicating the code / binary for everyone
>> else.
>
> However this patch doesn't complicate things more than it was with
> cpu_has_feature() which does exactly the same but using static keys,
> does it ?
It's more complicated in that it's not using cpu_has_feature() it's
doing some custom thing that is not used anywhere else. But yeah I guess
it's not much extra complication.
cheers