[src/netbsd-3]: src/sys/arch/i386/include Pull up revision 1.2 (requested by ...

[tron <tron%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/9d20d4429889
branches:  netbsd-3
changeset: 576595:9d20d4429889
user:      tron <tron%NetBSD.org@localhost>
date:      Tue Jul 12 11:36:54 2005 +0000

description:
Pull up revision 1.2 (requested by fair in ticket #554):
document more precisely the Cyrix TSC lossage: if you execute "hlt"
when in powersave mode, the TSC stops counting!

diffstat:

 sys/arch/i386/include/cpu_counter.h |  6 ++++--
 1 files changed, 4 insertions(+), 2 deletions(-)

diffs (20 lines):

diff -r 987dddb20bda -r 9d20d4429889 sys/arch/i386/include/cpu_counter.h
--- a/sys/arch/i386/include/cpu_counter.h       Tue Jul 12 11:35:03 2005 +0000
+++ b/sys/arch/i386/include/cpu_counter.h       Tue Jul 12 11:36:54 2005 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: cpu_counter.h,v 1.1 2003/02/05 12:18:04 nakayama Exp $ */
+/*     $NetBSD: cpu_counter.h,v 1.1.16.1 2005/07/12 11:36:54 tron Exp $        */
 
 /*-
  * Copyright (c) 2000 The NetBSD Foundation, Inc.
@@ -61,7 +61,9 @@
         * 1) Intel documentation is very specific that code *must* test
         * the CPU feature flag, even if you "know" that a particular
         * rev of the hardware supports it.
-        * 2) We know that the TSC is busted on some Cyrix CPU's..
+        * 2) We know that the TSC is busted on some Cyrix CPU in that if
+        * you execute "hlt" when in powersave mode, TSC stops counting,
+        * even though the CPU clock crystal is still ticking (it always has to).
         */
        return (cpu_feature & CPUID_TSC) != 0;
 #else