From: Paul Mackerras <paulus@samba.org>
Date: Sat, 19 Nov 2005 10:24:55 +0000 (+1100)
Subject: powerpc: Fix bug in timebase synchronization on 32-bit SMP powermac
X-Git-Tag: v2.6.15-rc2~5
X-Git-Url: https://err.no/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=54c4e6b5fee0347ab81e2c2fe4239c455c3422ae;p=linux-2.6

powerpc: Fix bug in timebase synchronization on 32-bit SMP powermac

We were using udelay in the loop on the primary cpu waiting for the
secondary cpu to take the timebase value.  Unfortunately now that
udelay uses the timebase, and the timebase is stopped at this point,
the udelay never terminated.  This fixes it by not using udelay, and
increases the number of loops before we time out to compensate.

Signed-off-by: Paul Mackerras <paulus@samba.org>
---

diff --git a/arch/powerpc/platforms/powermac/smp.c b/arch/powerpc/platforms/powermac/smp.c
index 957b091034..fb2a7c798e 100644
--- a/arch/powerpc/platforms/powermac/smp.c
+++ b/arch/powerpc/platforms/powermac/smp.c
@@ -34,6 +34,7 @@
 #include <linux/errno.h>
 #include <linux/hardirq.h>
 #include <linux/cpu.h>
+#include <linux/compiler.h>
 
 #include <asm/ptrace.h>
 #include <asm/atomic.h>
@@ -631,8 +632,9 @@ void smp_core99_give_timebase(void)
 	mb();
 
 	/* wait for the secondary to have taken it */
-	for (t = 100000; t > 0 && sec_tb_reset; --t)
-		udelay(10);
+	/* note: can't use udelay here, since it needs the timebase running */
+	for (t = 10000000; t > 0 && sec_tb_reset; --t)
+		barrier();
 	if (sec_tb_reset)
 		/* XXX BUG_ON here? */
 		printk(KERN_WARNING "Timeout waiting sync(2) on second CPU\n");