Date: Sat, 03 Nov 2001 06:23:19 +0900 (JST) From: Mitsuru IWASAKI <iwasaki@jp.FreeBSD.org> To: jhb@FreeBSD.org Cc: cvs-all@FreeBSD.org, cvs-committers@FreeBSD.org Subject: Re: cvs commit: src/sys/conf files src/sys/dev/acpica acpi.c src Message-ID: <20011103.062319.41627011.iwasaki@jp.FreeBSD.org> In-Reply-To: <XFMail.011102010133.jhb@FreeBSD.org> References: <200111011634.fA1GY8M18688@freefall.freebsd.org> <XFMail.011102010133.jhb@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi,
> On 01-Nov-01 Mitsuru IWASAKI wrote:
> > iwasaki 2001/11/01 08:34:08 PST
> >
> > Modified files:
> > sys/conf files
> > sys/dev/acpica acpi.c
> > sys/dev/syscons syscons.c
> > sys/i386/apm apm.c
> > sys/i386/isa clock.c
> > sys/sys kernel.h
> > Added files:
> > sys/kern subr_power.c
> > sys/sys power.h
> > Log:
> > Some fix for the recent apm module changes.
> > - Now that apm loadable module can inform its existence to other kernel
> > components (e.g. i386/isa/clock.c:startrtclock()'s TCS hack).
> > - Exchange priority of SI_SUB_CPU and SI_SUB_KLD for above purpose.
> > - Add simple arbitration mechanism for APM vs. ACPI. This prevents
> > the kernel enables both of them.
> > - Remove obsolete `#ifdef DEV_APM' related code.
> > - Add abstracted interface for Powermanagement operations. Public apm(4)
> > functions, such as apm_suspend(), should be replaced new interfaces.
> > Currently only power_pm_suspend (successor of apm_suspend) is
> > implemented.
> >
> > Reviewed by: peter, arch@ and audit@
>
> Hmm, I actually really don't like swapping those SYSINIT's. Probably what you
> should do is do whwat the alpha does and initalize the timecounters during the
> SI_SUB_CLOCKS sysinit. If SI_SUB_CLOCKS is before SI_SUB_KLD, then you should
> create a SI_SUB_TC after SI_SUB_KLD and initialize timecounters there.
Ok, like this?
Index: i386/isa/clock.c
===================================================================
RCS file: /home/ncvs/src/sys/i386/isa/clock.c,v
retrieving revision 1.178
diff -u -r1.178 clock.c
--- i386/isa/clock.c 1 Nov 2001 16:33:36 -0000 1.178
+++ i386/isa/clock.c 2 Nov 2001 12:07:27 -0000
@@ -788,8 +788,6 @@
}
set_timer_freq(timer_freq, hz);
- i8254_timecounter.tc_frequency = timer_freq;
- tc_init(&i8254_timecounter);
#ifndef CLK_USE_TSC_CALIBRATION
if (tsc_freq != 0) {
@@ -814,37 +812,6 @@
printf("TSC clock: %u Hz (Method B)\n", tsc_freq);
#endif
}
-
-#if !defined(SMP)
- /*
- * We can not use the TSC in SMP mode, until we figure out a
- * cheap (impossible), reliable and precise (yeah right!) way
- * to synchronize the TSCs of all the CPUs.
- * Curse Intel for leaving the counter out of the I/O APIC.
- */
-
- /*
- * We can not use the TSC if we support APM. Precise timekeeping
- * on an APM'ed machine is at best a fools pursuit, since
- * any and all of the time spent in various SMM code can't
- * be reliably accounted for. Reading the RTC is your only
- * source of reliable time info. The i8254 looses too of course
- * but we need to have some kind of time...
- * We don't know at this point whether APM is going to be used
- * or not, nor when it might be activated. Play it safe.
- */
- if (power_pm_get_type() == POWER_PM_TYPE_APM) {
- if (bootverbose)
- printf("TSC initialization skipped: APM enabled.\n");
- return;
- }
-
- if (tsc_present && tsc_freq != 0 && !tsc_is_broken) {
- tsc_timecounter.tc_frequency = tsc_freq;
- tc_init(&tsc_timecounter);
- }
-
-#endif /* !defined(SMP) */
}
/*
@@ -992,6 +959,34 @@
int apic_8254_trial;
void *clkdesc;
#endif /* APIC_IO */
+
+ i8254_timecounter.tc_frequency = timer_freq;
+ tc_init(&i8254_timecounter);
+#if !defined(SMP)
+ /*
+ * We can not use the TSC in SMP mode, until we figure out a
+ * cheap (impossible), reliable and precise (yeah right!) way
+ * to synchronize the TSCs of all the CPUs.
+ * Curse Intel for leaving the counter out of the I/O APIC.
+ */
+
+ /*
+ * We can not use the TSC if we support APM. Precise timekeeping
+ * on an APM'ed machine is at best a fools pursuit, since
+ * any and all of the time spent in various SMM code can't
+ * be reliably accounted for. Reading the RTC is your only
+ * source of reliable time info. The i8254 looses too of course
+ * but we need to have some kind of time...
+ * We don't know at this point whether APM is going to be used
+ * or not, nor when it might be activated. Play it safe.
+ */
+ if (power_pm_get_type() != POWER_PM_TYPE_APM &&
+ tsc_present && tsc_freq != 0 && !tsc_is_broken) {
+ tsc_timecounter.tc_frequency = tsc_freq;
+ tc_init(&tsc_timecounter);
+ }
+
+#endif /* !defined(SMP) */
if (statclock_disable) {
/*
Index: sys/kernel.h
===================================================================
RCS file: /home/ncvs/src/sys/sys/kernel.h,v
retrieving revision 1.96
diff -u -r1.96 kernel.h
--- sys/kernel.h 1 Nov 2001 16:34:06 -0000 1.96
+++ sys/kernel.h 2 Nov 2001 11:19:03 -0000
@@ -119,8 +119,8 @@
SI_SUB_WITNESS = 0x1A80000, /* witness initialization */
SI_SUB_LOCK = 0x1B00000, /* lockmgr locks */
SI_SUB_EVENTHANDLER = 0x1C00000, /* eventhandler init */
- SI_SUB_KLD = 0x2000000, /* KLD and module setup */
- SI_SUB_CPU = 0x2100000, /* CPU resource(s)*/
+ SI_SUB_CPU = 0x2000000, /* CPU resource(s)*/
+ SI_SUB_KLD = 0x2100000, /* KLD and module setup */
SI_SUB_INTRINSIC = 0x2200000, /* proc 0*/
SI_SUB_VM_CONF = 0x2300000, /* config VM, set limits*/
SI_SUB_RUN_QUEUE = 0x2400000, /* set up run queue*/
But I won't commit this unless weird problem is reported :-)
However, if someone commit this I don't complain about it.
> Logically, we need the CPU up and initialized before we start trying to figure
> out what modules we have and how to init them. At least IMO.
Hmmm, really? I checked all of SI_SUB_KLD sysinits and modevent functions
briefly, but I didn't see any dependency on SI_SUB_CPU sysinits.
I believe that kernel linker should not depend on CPU initialization...
Thanks
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011103.062319.41627011.iwasaki>