linux/arch/mips/kernel/csrc-ioasic.c

75 lines
1.8 KiB
C
Raw Normal View History

/*
* DEC I/O ASIC's counter clocksource
*
* Copyright (C) 2008 Yoichi Yuasa <yuasa@linux-mips.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*/
#include <linux/clocksource.h>
#include <linux/sched_clock.h>
#include <linux/init.h>
#include <asm/ds1287.h>
#include <asm/time.h>
#include <asm/dec/ioasic.h>
#include <asm/dec/ioasic_addrs.h>
static cycle_t dec_ioasic_hpt_read(struct clocksource *cs)
{
return ioasic_read(IO_REG_FCTR);
}
static struct clocksource clocksource_dec = {
.name = "dec-ioasic",
.read = dec_ioasic_hpt_read,
.mask = CLOCKSOURCE_MASK(32),
.flags = CLOCK_SOURCE_IS_CONTINUOUS,
};
static u64 notrace dec_ioasic_read_sched_clock(void)
{
return ioasic_read(IO_REG_FCTR);
}
MIPS: DECstation HRT initialization rearrangement Not all I/O ASIC versions have the free-running counter implemented, an early revision used in the 5000/1xx models aka 3MIN and 4MIN did not have it. Therefore we cannot unconditionally use it as a clock source. Fortunately if not implemented its register slot has a fixed value so it is enough if we check for the value at the end of the calibration period being the same as at the beginning. This also means we need to look for another high-precision clock source on the systems affected. The 5000/1xx can have an R4000SC processor installed where the CP0 Count register can be used as a clock source. Unfortunately all the R4k DECstations suffer from the missed timer interrupt on CP0 Count reads erratum, so we cannot use the CP0 timer as a clock source and a clock event both at a time. However we never need an R4k clock event device because all DECstations have a DS1287A RTC chip whose periodic interrupt can be used as a clock source. This gives us the following four configuration possibilities for I/O ASIC DECstations: 1. No I/O ASIC counter and no CP0 timer, e.g. R3k 5000/1xx (3MIN). 2. No I/O ASIC counter but the CP0 timer, i.e. R4k 5000/150 (4MIN). 3. The I/O ASIC counter but no CP0 timer, e.g. R3k 5000/240 (3MAX+). 4. The I/O ASIC counter and the CP0 timer, e.g. R4k 5000/260 (4MAX+). For #1 and #2 this change stops the I/O ASIC free-running counter from being installed as a clock source of a 0Hz frequency. For #2 it also arranges for the CP0 timer to be used as a clock source rather than a clock event device, because having an accurate wall clock is more important than a high-precision interval timer. For #3 there is no change. For #4 the change makes the I/O ASIC free-running counter installed as a clock source so that the CP0 timer can be used as a clock event device. Unfortunately the use of the CP0 timer as a clock event device relies on a succesful completion of c0_compare_interrupt. That never happens, because while waiting for a CP0 Compare interrupt to happen the function spins in a loop reading the CP0 Count register. This makes the CP0 Count erratum trigger reliably causing the interrupt waited for to be lost in all cases. As a result #4 resorts to using the CP0 timer as a clock source as well, just as #2. However we want to keep this separate arrangement in case (hope) c0_compare_interrupt is eventually rewritten such that it avoids the erratum. Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/5825/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2013-09-12 19:01:53 +08:00
int __init dec_ioasic_clocksource_init(void)
{
unsigned int freq;
u32 start, end;
MIPS: DECstation HRT calibration bug fixes This change corrects DECstation HRT calibration, by removing the following bugs: 1. Calibration period selection -- HZ / 10 has been chosen, however on DECstation computers, HZ never divides by 10, as the choice for HZ is among 128, 256 and 1024. The choice therefore results in a systematic calibration error, e.g. 6.25% for the usual choice of 128 for HZ: 128 / 10 * 10 = 120 (128 - 120) / 128 -> 6.25% The change therefore makes calibration use HZ / 8 that is always accurate for the HZ values available, getting rid of the systematic error. 2. Calibration starting point synchronisation -- the duration of a number of intervals between DS1287A periodic interrupt assertions is measured, however code does not ensure at the beginning that the interrupt has not been previously asserted. This results in a variable error of e.g. up to another 6.25% for the period of HZ / 8 (8.(3)% with the original HZ / 10 period) and the usual choice of 128 for HZ: 1 / 16 -> 6.25% 1 / 12 -> 8.(3)% The change therefore adds an initial call to ds1287_timer_state that clears any previous periodic interrupt pending. The same issue applies to both I/O ASIC counter and R4k CP0 timer calibration on DECstation systems as similar code is used in both cases and both pieces of code are covered by this fix. On an R3400 test system used this fix results in a change of the I/O ASIC clock frequency reported from values like: I/O ASIC clock frequency 23185830Hz to: I/O ASIC clock frequency 24999288Hz removing the miscalculation by 6.25% from the systematic error and (for the individual sample provided) a further 1.00% from the variable error, accordingly. The nominal I/O ASIC clock frequency is 25MHz on this system. Here's another result, with the fix applied, from a system that has both HRTs available (using an R4400 at 60MHz nominal): MIPS counter frequency 59999328Hz I/O ASIC clock frequency 24999432Hz Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/5807/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2013-09-05 06:47:45 +08:00
int i = HZ / 8;
MIPS: DECstation HRT calibration bug fixes This change corrects DECstation HRT calibration, by removing the following bugs: 1. Calibration period selection -- HZ / 10 has been chosen, however on DECstation computers, HZ never divides by 10, as the choice for HZ is among 128, 256 and 1024. The choice therefore results in a systematic calibration error, e.g. 6.25% for the usual choice of 128 for HZ: 128 / 10 * 10 = 120 (128 - 120) / 128 -> 6.25% The change therefore makes calibration use HZ / 8 that is always accurate for the HZ values available, getting rid of the systematic error. 2. Calibration starting point synchronisation -- the duration of a number of intervals between DS1287A periodic interrupt assertions is measured, however code does not ensure at the beginning that the interrupt has not been previously asserted. This results in a variable error of e.g. up to another 6.25% for the period of HZ / 8 (8.(3)% with the original HZ / 10 period) and the usual choice of 128 for HZ: 1 / 16 -> 6.25% 1 / 12 -> 8.(3)% The change therefore adds an initial call to ds1287_timer_state that clears any previous periodic interrupt pending. The same issue applies to both I/O ASIC counter and R4k CP0 timer calibration on DECstation systems as similar code is used in both cases and both pieces of code are covered by this fix. On an R3400 test system used this fix results in a change of the I/O ASIC clock frequency reported from values like: I/O ASIC clock frequency 23185830Hz to: I/O ASIC clock frequency 24999288Hz removing the miscalculation by 6.25% from the systematic error and (for the individual sample provided) a further 1.00% from the variable error, accordingly. The nominal I/O ASIC clock frequency is 25MHz on this system. Here's another result, with the fix applied, from a system that has both HRTs available (using an R4400 at 60MHz nominal): MIPS counter frequency 59999328Hz I/O ASIC clock frequency 24999432Hz Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/5807/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2013-09-05 06:47:45 +08:00
ds1287_timer_state();
while (!ds1287_timer_state())
;
start = dec_ioasic_hpt_read(&clocksource_dec);
while (i--)
while (!ds1287_timer_state())
;
end = dec_ioasic_hpt_read(&clocksource_dec);
MIPS: DECstation HRT calibration bug fixes This change corrects DECstation HRT calibration, by removing the following bugs: 1. Calibration period selection -- HZ / 10 has been chosen, however on DECstation computers, HZ never divides by 10, as the choice for HZ is among 128, 256 and 1024. The choice therefore results in a systematic calibration error, e.g. 6.25% for the usual choice of 128 for HZ: 128 / 10 * 10 = 120 (128 - 120) / 128 -> 6.25% The change therefore makes calibration use HZ / 8 that is always accurate for the HZ values available, getting rid of the systematic error. 2. Calibration starting point synchronisation -- the duration of a number of intervals between DS1287A periodic interrupt assertions is measured, however code does not ensure at the beginning that the interrupt has not been previously asserted. This results in a variable error of e.g. up to another 6.25% for the period of HZ / 8 (8.(3)% with the original HZ / 10 period) and the usual choice of 128 for HZ: 1 / 16 -> 6.25% 1 / 12 -> 8.(3)% The change therefore adds an initial call to ds1287_timer_state that clears any previous periodic interrupt pending. The same issue applies to both I/O ASIC counter and R4k CP0 timer calibration on DECstation systems as similar code is used in both cases and both pieces of code are covered by this fix. On an R3400 test system used this fix results in a change of the I/O ASIC clock frequency reported from values like: I/O ASIC clock frequency 23185830Hz to: I/O ASIC clock frequency 24999288Hz removing the miscalculation by 6.25% from the systematic error and (for the individual sample provided) a further 1.00% from the variable error, accordingly. The nominal I/O ASIC clock frequency is 25MHz on this system. Here's another result, with the fix applied, from a system that has both HRTs available (using an R4400 at 60MHz nominal): MIPS counter frequency 59999328Hz I/O ASIC clock frequency 24999432Hz Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/5807/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2013-09-05 06:47:45 +08:00
freq = (end - start) * 8;
MIPS: DECstation HRT initialization rearrangement Not all I/O ASIC versions have the free-running counter implemented, an early revision used in the 5000/1xx models aka 3MIN and 4MIN did not have it. Therefore we cannot unconditionally use it as a clock source. Fortunately if not implemented its register slot has a fixed value so it is enough if we check for the value at the end of the calibration period being the same as at the beginning. This also means we need to look for another high-precision clock source on the systems affected. The 5000/1xx can have an R4000SC processor installed where the CP0 Count register can be used as a clock source. Unfortunately all the R4k DECstations suffer from the missed timer interrupt on CP0 Count reads erratum, so we cannot use the CP0 timer as a clock source and a clock event both at a time. However we never need an R4k clock event device because all DECstations have a DS1287A RTC chip whose periodic interrupt can be used as a clock source. This gives us the following four configuration possibilities for I/O ASIC DECstations: 1. No I/O ASIC counter and no CP0 timer, e.g. R3k 5000/1xx (3MIN). 2. No I/O ASIC counter but the CP0 timer, i.e. R4k 5000/150 (4MIN). 3. The I/O ASIC counter but no CP0 timer, e.g. R3k 5000/240 (3MAX+). 4. The I/O ASIC counter and the CP0 timer, e.g. R4k 5000/260 (4MAX+). For #1 and #2 this change stops the I/O ASIC free-running counter from being installed as a clock source of a 0Hz frequency. For #2 it also arranges for the CP0 timer to be used as a clock source rather than a clock event device, because having an accurate wall clock is more important than a high-precision interval timer. For #3 there is no change. For #4 the change makes the I/O ASIC free-running counter installed as a clock source so that the CP0 timer can be used as a clock event device. Unfortunately the use of the CP0 timer as a clock event device relies on a succesful completion of c0_compare_interrupt. That never happens, because while waiting for a CP0 Compare interrupt to happen the function spins in a loop reading the CP0 Count register. This makes the CP0 Count erratum trigger reliably causing the interrupt waited for to be lost in all cases. As a result #4 resorts to using the CP0 timer as a clock source as well, just as #2. However we want to keep this separate arrangement in case (hope) c0_compare_interrupt is eventually rewritten such that it avoids the erratum. Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/5825/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2013-09-12 19:01:53 +08:00
/* An early revision of the I/O ASIC didn't have the counter. */
if (!freq)
return -ENXIO;
printk(KERN_INFO "I/O ASIC clock frequency %dHz\n", freq);
clocksource_dec.rating = 200 + freq / 10000000;
clocksource_register_hz(&clocksource_dec, freq);
sched_clock_register(dec_ioasic_read_sched_clock, 32, freq);
MIPS: DECstation HRT initialization rearrangement Not all I/O ASIC versions have the free-running counter implemented, an early revision used in the 5000/1xx models aka 3MIN and 4MIN did not have it. Therefore we cannot unconditionally use it as a clock source. Fortunately if not implemented its register slot has a fixed value so it is enough if we check for the value at the end of the calibration period being the same as at the beginning. This also means we need to look for another high-precision clock source on the systems affected. The 5000/1xx can have an R4000SC processor installed where the CP0 Count register can be used as a clock source. Unfortunately all the R4k DECstations suffer from the missed timer interrupt on CP0 Count reads erratum, so we cannot use the CP0 timer as a clock source and a clock event both at a time. However we never need an R4k clock event device because all DECstations have a DS1287A RTC chip whose periodic interrupt can be used as a clock source. This gives us the following four configuration possibilities for I/O ASIC DECstations: 1. No I/O ASIC counter and no CP0 timer, e.g. R3k 5000/1xx (3MIN). 2. No I/O ASIC counter but the CP0 timer, i.e. R4k 5000/150 (4MIN). 3. The I/O ASIC counter but no CP0 timer, e.g. R3k 5000/240 (3MAX+). 4. The I/O ASIC counter and the CP0 timer, e.g. R4k 5000/260 (4MAX+). For #1 and #2 this change stops the I/O ASIC free-running counter from being installed as a clock source of a 0Hz frequency. For #2 it also arranges for the CP0 timer to be used as a clock source rather than a clock event device, because having an accurate wall clock is more important than a high-precision interval timer. For #3 there is no change. For #4 the change makes the I/O ASIC free-running counter installed as a clock source so that the CP0 timer can be used as a clock event device. Unfortunately the use of the CP0 timer as a clock event device relies on a succesful completion of c0_compare_interrupt. That never happens, because while waiting for a CP0 Compare interrupt to happen the function spins in a loop reading the CP0 Count register. This makes the CP0 Count erratum trigger reliably causing the interrupt waited for to be lost in all cases. As a result #4 resorts to using the CP0 timer as a clock source as well, just as #2. However we want to keep this separate arrangement in case (hope) c0_compare_interrupt is eventually rewritten such that it avoids the erratum. Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/5825/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2013-09-12 19:01:53 +08:00
return 0;
}