From c9c57929d23e44f258d1b6e7f089e72c85f0bd1c Mon Sep 17 00:00:00 2001 From: Ravikiran G Thirumalai Date: Fri, 13 Apr 2007 16:28:20 -0700 Subject: [PATCH] failsafe mechanism to HPET clock calibration Provide a failsafe mechanism to avoid kernel spinning forever at read_hpet_tsc during early kernel bootup. This failsafe mechanism was originally introduced in commit 2f7a2a79c3ebb44f8b1b7d9b4fd3a650eb69e544, but looks like the hpet split from time.c lost it again. This reintroduces the failsafe mechanism Signed-off-by: Ravikiran Thirumalai Signed-off-by: Shai Fultheim Cc: Jack Steiner Cc: john stultz Cc: Andi Kleen Signed-off-by: Linus Torvalds --- arch/x86_64/kernel/hpet.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/arch/x86_64/kernel/hpet.c b/arch/x86_64/kernel/hpet.c index 8cf0b8a13778ae..b8286968662d0d 100644 --- a/arch/x86_64/kernel/hpet.c +++ b/arch/x86_64/kernel/hpet.c @@ -191,6 +191,7 @@ int hpet_reenable(void) #define TICK_COUNT 100000000 #define TICK_MIN 5000 +#define MAX_TRIES 5 /* * Some platforms take periodic SMI interrupts with 5ms duration. Make sure none @@ -198,13 +199,15 @@ int hpet_reenable(void) */ static void __init read_hpet_tsc(int *hpet, int *tsc) { - int tsc1, tsc2, hpet1; + int tsc1, tsc2, hpet1, i; - do { + for (i = 0; i < MAX_TRIES; i++) { tsc1 = get_cycles_sync(); hpet1 = hpet_readl(HPET_COUNTER); tsc2 = get_cycles_sync(); - } while (tsc2 - tsc1 > TICK_MIN); + if (tsc2 - tsc1 > TICK_MIN) + break; + } *hpet = hpet1; *tsc = tsc2; }