Attachment 155283 Details for Bug 240782 – RHEL3 fix for this issue

[patch] RHEL3 fix for this issue

240782.patch (text/plain), 1.27 KB, created by Prarit Bhargava on 2007-05-23 18:06:44 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Prarit Bhargava

Created: 2007-05-23 18:06:44 UTC

Size: 1.27 KB

patch

obsolete

>--- linux-2.4.21.orig/arch/ia64/kernel/time.c	2007-05-23 11:46:20.000000000 -0400
>+++ linux-2.4.21/arch/ia64/kernel/time.c	2007-05-23 11:49:18.000000000 -0400
>@@ -35,8 +35,12 @@ unsigned long last_cli_ip;
> #endif
> 
> /*
>- * Return the number of micro-seconds that elapsed since the last update to jiffy.  The
>- * BR_XTIME_LOCK must be at least read-locked when calling this routine.
>+ * Return the number of nano-seconds that elapsed since the last
>+ * update to jiffy.  It is quite possible that the timer interrupt
>+ * will interrupt this and result in a race for any of jiffies,
>+ * wall_jiffies or itm_next.  Thus, the xtime_lock must be at least
>+ * read synchronised when calling this routine (see do_gettimeofday()
>+ * below for an example).
>  */
> static inline unsigned long
> gettimeoffset (void)
>@@ -50,15 +54,6 @@ gettimeoffset (void)
> 
> 	last_tick = (cpu_data(time_keeper_id)->itm_next
> 		     - (lost + 1)*cpu_data(time_keeper_id)->itm_delta);
>-
>-	now = ia64_get_itc();
>-	if ((long) (now - last_tick) < 0) {
>-# if 1
>-		printk("CPU %d: now < last_tick (now=0x%lx,last_tick=0x%lx)!\n",
>-		       smp_processor_id(), now, last_tick);
>-# endif
>-		return last_time_offset;
>-	}
> 	elapsed_cycles = now - last_tick;
> 	return (elapsed_cycles*local_cpu_data->usec_per_cyc) >> IA64_USEC_PER_CYC_SHIFT;
> }

Actions: View | Diff

Attachments on bug 240782: 155120 | 155255 | 155283