[PATCH -tip] x86: apic LVTT - use APIC_DIVISOR on 64bit mode

[PATCH -tip] x86: apic LVTT - use APIC_DIVISOR on 64bit mode

[Cyrill Gorcunov]

We use APIC_TDR_DIV_16 while setting APIC_TDCR (divisor)
it happened this moment is hidden from __setup_APIC_LVTT
caller. So we better increment caller 'clocks' value to
not change current behaviour and use APIC_DIVISOR
(as already done in 32bit version).

The main benefit - unified procedure code for 32/64bit modes.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
CC: "Maciej W. Rozycki" <macro@linux-mips.org>
---

Please review carefully. Any comments are welcome.
I'm _not_ 100% sure if this patch is safe.

Index: linux-2.6.git/arch/x86/kernel/apic_64.c
===================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic_64.c	2008-07-08 22:52:24.000000000 +0400
+++ linux-2.6.git/arch/x86/kernel/apic_64.c	2008-07-08 23:00:24.000000000 +0400
@@ -165,6 +165,9 @@ int lapic_get_maxlvt(void)
 	return maxlvt;
 }
 
+/* Clock divisor is set to 16 */
+#define APIC_DIVISOR 16
+
 /*
  * This function sets up the local APIC timer, with a timeout of
  * 'clocks' APIC bus clock. During calibration we actually call
@@ -197,7 +200,7 @@ static void __setup_APIC_LVTT(unsigned i
 				| APIC_TDR_DIV_16);
 
 	if (!oneshot)
-		apic_write(APIC_TMICT, clocks);
+		apic_write(APIC_TMICT, clocks/APIC_DIVISOR);
 }
 
 /*
@@ -329,7 +332,7 @@ static void __init calibrate_APIC_clock(
 	 *
 	 * No interrupt enable !
 	 */
-	__setup_APIC_LVTT(250000000, 0, 0);
+	__setup_APIC_LVTT(4000000000, 0, 0);
 
 	apic_start = apic_read(APIC_TMCCT);
 #ifdef CONFIG_X86_PM_TIMER

Re: [PATCH -tip] x86: apic LVTT - use APIC_DIVISOR on 64bit mode

[Ingo Molnar]

* Cyrill Gorcunov <gorcunov@gmail.com> wrote:

> We use APIC_TDR_DIV_16 while setting APIC_TDCR (divisor) it happened 
> this moment is hidden from __setup_APIC_LVTT caller. So we better 
> increment caller 'clocks' value to not change current behaviour and 
> use APIC_DIVISOR (as already done in 32bit version).
> 
> The main benefit - unified procedure code for 32/64bit modes.

> Please review carefully. Any comments are welcome.
> I'm _not_ 100% sure if this patch is safe.

hm, how about the other caller to __setup_APIC_LVTT(), 
lapic_timer_setup():

        case CLOCK_EVT_MODE_ONESHOT:
                __setup_APIC_LVTT(calibration_result,
                                  mode != CLOCK_EVT_MODE_PERIODIC, 1);
                break;

that will affect high-res timers. Wont calibration_result have the same 
value after your patch as well, but only a 16th of it will be used in 
__setup_APIC_LVTT(), causing 16 times shorter timeouts than intended?

I.e. are we fixing a bug, are we changing nothing, or are we introducing 
a bug? :-)

	Ingo

Re: [PATCH -tip] x86: apic LVTT - use APIC_DIVISOR on 64bit mode

[Cyrill Gorcunov]

[Ingo Molnar - Sat, Jul 12, 2008 at 08:05:04AM +0200]
| 
| * Cyrill Gorcunov <gorcunov@gmail.com> wrote:
| 
| > We use APIC_TDR_DIV_16 while setting APIC_TDCR (divisor) it happened 
| > this moment is hidden from __setup_APIC_LVTT caller. So we better 
| > increment caller 'clocks' value to not change current behaviour and 
| > use APIC_DIVISOR (as already done in 32bit version).
| > 
| > The main benefit - unified procedure code for 32/64bit modes.
| 
| > Please review carefully. Any comments are welcome.
| > I'm _not_ 100% sure if this patch is safe.
| 
| hm, how about the other caller to __setup_APIC_LVTT(), 
| lapic_timer_setup():
| 
|         case CLOCK_EVT_MODE_ONESHOT:
|                 __setup_APIC_LVTT(calibration_result,
|                                   mode != CLOCK_EVT_MODE_PERIODIC, 1);
|                 break;
| 
| that will affect high-res timers. Wont calibration_result have the same 
| value after your patch as well, but only a 16th of it will be used in 
| __setup_APIC_LVTT(), causing 16 times shorter timeouts than intended?
| 
| I.e. are we fixing a bug, are we changing nothing, or are we introducing 
| a bug? :-)
| 
| 	Ingo
| 

well, the last I would say :). That happened I sent first version of
patch instead of modified second version. Thanks a lot Ingo for catching
it! Oh my, sorry... Here is an updated version.
---

We use APIC_TDR_DIV_16 while setting APIC_TDCR (divisor).
It happened this moment is hidden from __setup_APIC_LVTT
caller. So we better increment caller 'clocks' value to
not change current behaviour and use APIC_DIVISOR
(as already done in 32bit version).

The main benefit - unified procedure code for 32/64bit modes.

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
CC: "Maciej W. Rozycki" <macro@linux-mips.org>
---

Index: linux-2.6.git/arch/x86/kernel/apic_64.c
===================================================================
--- linux-2.6.git.orig/arch/x86/kernel/apic_64.c	2008-07-12 11:11:33.000000000 +0400
+++ linux-2.6.git/arch/x86/kernel/apic_64.c	2008-07-12 11:14:22.000000000 +0400
@@ -165,6 +165,9 @@ int lapic_get_maxlvt(void)
 	return maxlvt;
 }
 
+/* Clock divisor is set to 16 */
+#define APIC_DIVISOR 16
+
 /*
  * This function sets up the local APIC timer, with a timeout of
  * 'clocks' APIC bus clock. During calibration we actually call
@@ -197,7 +200,7 @@ static void __setup_APIC_LVTT(unsigned i
 				| APIC_TDR_DIV_16);
 
 	if (!oneshot)
-		apic_write(APIC_TMICT, clocks);
+		apic_write(APIC_TMICT, clocks/APIC_DIVISOR);
 }
 
 /*
@@ -329,7 +332,7 @@ static void __init calibrate_APIC_clock(
 	 *
 	 * No interrupt enable !
 	 */
-	__setup_APIC_LVTT(250000000, 0, 0);
+	__setup_APIC_LVTT(4000000000, 0, 0);
 
 	apic_start = apic_read(APIC_TMCCT);
 #ifdef CONFIG_X86_PM_TIMER
@@ -367,7 +370,7 @@ static void __init calibrate_APIC_clock(
 	lapic_clockevent.min_delta_ns =
 		clockevent_delta2ns(0xF, &lapic_clockevent);
 
-	calibration_result = result / HZ;
+	calibration_result = result * APIC_DIVISOR / HZ;
 }
 
 /*

Re: [PATCH -tip] x86: apic LVTT - use APIC_DIVISOR on 64bit mode

[Ingo Molnar]

* Cyrill Gorcunov <gorcunov@gmail.com> wrote:

> @@ -329,7 +332,7 @@ static void __init calibrate_APIC_clock(
>  	 *
>  	 * No interrupt enable !
>  	 */
> -	__setup_APIC_LVTT(250000000, 0, 0);
> +	__setup_APIC_LVTT(4000000000, 0, 0);

note how close it is to 2^32. For this to be unifiable later on this 
needs to be UL i guess, and this:

> -	calibration_result = result / HZ;
> +	calibration_result = result * APIC_DIVISOR / HZ;

might overflow 32 bits.

	Ingo

Re: [PATCH -tip] x86: apic LVTT - use APIC_DIVISOR on 64bit mode

[Cyrill Gorcunov]

[Ingo Molnar - Sat, Jul 12, 2008 at 09:39:10AM +0200]
| 
| * Cyrill Gorcunov <gorcunov@gmail.com> wrote:
| 
| > @@ -329,7 +332,7 @@ static void __init calibrate_APIC_clock(
| >  	 *
| >  	 * No interrupt enable !
| >  	 */
| > -	__setup_APIC_LVTT(250000000, 0, 0);
| > +	__setup_APIC_LVTT(4000000000, 0, 0);
| 
| note how close it is to 2^32. For this to be unifiable later on this 
| needs to be UL i guess, and this:
| 
| > -	calibration_result = result / HZ;
| > +	calibration_result = result * APIC_DIVISOR / HZ;
| 
| might overflow 32 bits.
| 
| 	Ingo
|

hmm... I need more time for analisys, thanks. Drop this patch
please
 
		- Cyrill -

Re: [PATCH -tip] x86: apic LVTT - use APIC_DIVISOR on 64bit mode

[Ingo Molnar]

* Ingo Molnar <mingo@elte.hu> wrote:

> 
> * Cyrill Gorcunov <gorcunov@gmail.com> wrote:
> 
> > We use APIC_TDR_DIV_16 while setting APIC_TDCR (divisor) it happened 
> > this moment is hidden from __setup_APIC_LVTT caller. So we better 
> > increment caller 'clocks' value to not change current behaviour and 
> > use APIC_DIVISOR (as already done in 32bit version).
> > 
> > The main benefit - unified procedure code for 32/64bit modes.
> 
> > Please review carefully. Any comments are welcome.
> > I'm _not_ 100% sure if this patch is safe.
> 
> hm, how about the other caller to __setup_APIC_LVTT(), 
> lapic_timer_setup():
> 
>         case CLOCK_EVT_MODE_ONESHOT:
>                 __setup_APIC_LVTT(calibration_result,
>                                   mode != CLOCK_EVT_MODE_PERIODIC, 1);
>                 break;
> 
> that will affect high-res timers. Wont calibration_result have the same 
> value after your patch as well, but only a 16th of it will be used in 
> __setup_APIC_LVTT(), causing 16 times shorter timeouts than intended?
> 
> I.e. are we fixing a bug, are we changing nothing, or are we 
> introducing a bug? :-)

i think we are introducing a bug :)

	Ingo