failsafe mechanism to HPET clock calibration
authorRavikiran G Thirumalai <kiran@scalex86.org>
Fri, 13 Apr 2007 23:28:20 +0000 (16:28 -0700)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Sun, 15 Apr 2007 04:44:03 +0000 (21:44 -0700)
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 <kiran@scalex86.org>
Signed-off-by: Shai Fultheim <shai@scalex86.org>
Cc: Jack Steiner <steiner@sgi.com>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
arch/x86_64/kernel/hpet.c

index 8cf0b8a13778ae0e497ffd228fb24f9eef8d163d..b8286968662d0d19d21a99443f604016e0883ce2 100644 (file)
@@ -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;
 }