[PATCH] DEC init_cycle_counter

[PATCH] DEC init_cycle_counter

[Florian Lohoff]

Comments ?

Index: time.c
===================================================================
RCS file: /cvs/linux/arch/mips/dec/time.c,v
retrieving revision 1.11
diff -u -r1.11 arch/mips/dec/time.c
--- arch/mips/dec/time.c	2000/11/23 02:00:49	1.11
+++ arch/mips/dec/time.c	2000/12/02 12:20:09
@@ -423,46 +423,6 @@
 	timer_interrupt(irq, dev_id, regs);
 }
 
-char cyclecounter_available;
-
-static inline void init_cycle_counter(void)
-{
-	switch (mips_cputype) {
-	case CPU_UNKNOWN:
-	case CPU_R2000:
-	case CPU_R3000:
-	case CPU_R3000A:
-	case CPU_R3041:
-	case CPU_R3051:
-	case CPU_R3052:
-	case CPU_R3081:
-	case CPU_R3081E:
-	case CPU_R6000:
-	case CPU_R6000A:
-	case CPU_R8000:	/* Not shure about that one, play safe */
-		cyclecounter_available = 0;
-		break;
-	case CPU_R4000PC:
-	case CPU_R4000SC:
-	case CPU_R4000MC:
-	case CPU_R4200:
-	case CPU_R4400PC:
-	case CPU_R4400SC:
-	case CPU_R4400MC:
-	case CPU_R4600:
-	case CPU_R10000:
-	case CPU_R4300:
-	case CPU_R4650:
-	case CPU_R4700:
-	case CPU_R5000:
-	case CPU_R5000A:
-	case CPU_R4640:
-	case CPU_NEVADA:
-		cyclecounter_available = 1;
-		break;
-	}
-}
-
 struct irqaction irq0 = {timer_interrupt, SA_INTERRUPT, 0,
 			 "timer", NULL, NULL};
 
@@ -513,8 +473,6 @@
 	xtime.tv_sec = mktime(year, mon, day, hour, min, sec);
 	xtime.tv_usec = 0;
 	write_unlock_irq(&xtime_lock);
-
-	init_cycle_counter();
 
 	if (cyclecounter_available) {
 		write_32bit_cp0_register(CP0_COUNT, 0);


-- 
Florian Lohoff                  flo@rfc822.org             +49-5201-669912
     Why is it called "common sense" when nobody seems to have any?

Re: [PATCH] DEC init_cycle_counter

[Juan J. Quintela]

>>>>> "florian" == Florian Lohoff <flo@rfc822.org> writes:

florian> Comments ?

[snip]
 
florian> -char cyclecounter_available;
florian> -

[snip]
 
florian>  	if (cyclecounter_available) {
florian>  		write_32bit_cp0_register(CP0_COUNT, 0);

You delete the definition of cyclecounter_available and you mantain
one of its uses ..... that is a no-no, I will bet that with that
patch, this don't compile (and less it works .....).

Later, Juan.

-- 
In theory, practice and theory are the same, but in practice they 
are different -- Larry McVoy

Re: [PATCH] DEC init_cycle_counter

[Florian Lohoff]

On Sat, Dec 02, 2000 at 02:02:51PM +0100, Juan J. Quintela wrote:
> You delete the definition of cyclecounter_available and you mantain
> one of its uses ..... that is a no-no, I will bet that with that
> patch, this don't compile (and less it works .....).
> 

It actually compiles but i cant test due to the TLB and swapping stuff.

This whole cyle_counter stuff has moved to the generic mips time.c
in arch/mips/kernel/time.c and seemed to be a complete code duplication.

Flo
-- 
Florian Lohoff                  flo@rfc822.org             +49-5201-669912
     Why is it called "common sense" when nobody seems to have any?

Re: [PATCH] DEC init_cycle_counter

[Kevin D. Kissell]

Only that if you adopted the CPU probe and configuration
scheme that we've been promulgating at MIPS, it would
already have been taken care of in kernel/cpu_probe.c, and
could be tested as (mips_cpu.options  & MIPS_CPU_COUNTER).
I believe I even checked in the headers and C file in the 2.2
repository, though I don't think I "wired it up" to replace the old
assembler spaghetti.

            Kevin K.

----- Original Message -----
From: "Florian Lohoff" <flo@rfc822.org>
To: <linux-mips@oss.sgi.com>
Sent: Saturday, December 02, 2000 1:27 PM
Subject: [PATCH] DEC init_cycle_counter


>
> Comments ?
>
> Index: time.c
> ===================================================================
> RCS file: /cvs/linux/arch/mips/dec/time.c,v
> retrieving revision 1.11
> diff -u -r1.11 arch/mips/dec/time.c
> --- arch/mips/dec/time.c 2000/11/23 02:00:49 1.11
> +++ arch/mips/dec/time.c 2000/12/02 12:20:09
> @@ -423,46 +423,6 @@
>   timer_interrupt(irq, dev_id, regs);
>  }
>
> -char cyclecounter_available;
> -
> -static inline void init_cycle_counter(void)
> -{
> - switch (mips_cputype) {
> - case CPU_UNKNOWN:
> - case CPU_R2000:
> - case CPU_R3000:
> - case CPU_R3000A:
> - case CPU_R3041:
> - case CPU_R3051:
> - case CPU_R3052:
> - case CPU_R3081:
> - case CPU_R3081E:
> - case CPU_R6000:
> - case CPU_R6000A:
> - case CPU_R8000: /* Not shure about that one, play safe */
> - cyclecounter_available = 0;
> - break;
> - case CPU_R4000PC:
> - case CPU_R4000SC:
> - case CPU_R4000MC:
> - case CPU_R4200:
> - case CPU_R4400PC:
> - case CPU_R4400SC:
> - case CPU_R4400MC:
> - case CPU_R4600:
> - case CPU_R10000:
> - case CPU_R4300:
> - case CPU_R4650:
> - case CPU_R4700:
> - case CPU_R5000:
> - case CPU_R5000A:
> - case CPU_R4640:
> - case CPU_NEVADA:
> - cyclecounter_available = 1;
> - break;
> - }
> -}
> -
>  struct irqaction irq0 = {timer_interrupt, SA_INTERRUPT, 0,
>   "timer", NULL, NULL};
>
> @@ -513,8 +473,6 @@
>   xtime.tv_sec = mktime(year, mon, day, hour, min, sec);
>   xtime.tv_usec = 0;
>   write_unlock_irq(&xtime_lock);
> -
> - init_cycle_counter();
>
>   if (cyclecounter_available) {
>   write_32bit_cp0_register(CP0_COUNT, 0);
>
>
> --
> Florian Lohoff                  flo@rfc822.org             +49-5201-669912
>      Why is it called "common sense" when nobody seems to have any?
>

Re: [PATCH] DEC init_cycle_counter

[Kevin D. Kissell]

Only that if you adopted the CPU probe and configuration
scheme that we've been promulgating at MIPS, it would
already have been taken care of in kernel/cpu_probe.c, and
could be tested as (mips_cpu.options  & MIPS_CPU_COUNTER).
I believe I even checked in the headers and C file in the 2.2
repository, though I don't think I "wired it up" to replace the old
assembler spaghetti.

            Kevin K.

----- Original Message -----
From: "Florian Lohoff" <flo@rfc822.org>
To: <linux-mips@oss.sgi.com>
Sent: Saturday, December 02, 2000 1:27 PM
Subject: [PATCH] DEC init_cycle_counter


>
> Comments ?
>
> Index: time.c
> ===================================================================
> RCS file: /cvs/linux/arch/mips/dec/time.c,v
> retrieving revision 1.11
> diff -u -r1.11 arch/mips/dec/time.c
> --- arch/mips/dec/time.c 2000/11/23 02:00:49 1.11
> +++ arch/mips/dec/time.c 2000/12/02 12:20:09
> @@ -423,46 +423,6 @@
>   timer_interrupt(irq, dev_id, regs);
>  }
>
> -char cyclecounter_available;
> -
> -static inline void init_cycle_counter(void)
> -{
> - switch (mips_cputype) {
> - case CPU_UNKNOWN:
> - case CPU_R2000:
> - case CPU_R3000:
> - case CPU_R3000A:
> - case CPU_R3041:
> - case CPU_R3051:
> - case CPU_R3052:
> - case CPU_R3081:
> - case CPU_R3081E:
> - case CPU_R6000:
> - case CPU_R6000A:
> - case CPU_R8000: /* Not shure about that one, play safe */
> - cyclecounter_available = 0;
> - break;
> - case CPU_R4000PC:
> - case CPU_R4000SC:
> - case CPU_R4000MC:
> - case CPU_R4200:
> - case CPU_R4400PC:
> - case CPU_R4400SC:
> - case CPU_R4400MC:
> - case CPU_R4600:
> - case CPU_R10000:
> - case CPU_R4300:
> - case CPU_R4650:
> - case CPU_R4700:
> - case CPU_R5000:
> - case CPU_R5000A:
> - case CPU_R4640:
> - case CPU_NEVADA:
> - cyclecounter_available = 1;
> - break;
> - }
> -}
> -
>  struct irqaction irq0 = {timer_interrupt, SA_INTERRUPT, 0,
>   "timer", NULL, NULL};
>
> @@ -513,8 +473,6 @@
>   xtime.tv_sec = mktime(year, mon, day, hour, min, sec);
>   xtime.tv_usec = 0;
>   write_unlock_irq(&xtime_lock);
> -
> - init_cycle_counter();
>
>   if (cyclecounter_available) {
>   write_32bit_cp0_register(CP0_COUNT, 0);
>
>
> --
> Florian Lohoff                  flo@rfc822.org             +49-5201-669912
>      Why is it called "common sense" when nobody seems to have any?
>

Re: [PATCH] DEC init_cycle_counter

[Ralf Baechle]

On Sat, Dec 02, 2000 at 01:27:48PM +0100, Florian Lohoff wrote:

> Comments ?

I got a better an more complete from MIPS in the patch queue.

  Ralf