[PATCH 2.6][PPC32] RESEND: don't recursively crash in die() on CHRP/PReP machines

This patch avoids recursive crash (leading to kernel stack overflow) in
die() on CHRP/PReP machines when CONFIG_PMAC_BACKLIGHT=y.
set_backlight_* functions are placed in pmac section, which is discarded
when _machine != _MACH_Pmac.

I already posted this patch to LKML few months ago:
http://www.ussg.iu.edu/hypermail/linux/kernel/0412.0/0300.html
and it has been applied to linux-2.4 tree, but still not 2.6.
(patch was made against 2.4.27, but still applies cleanly against
kernels up to 2.6.11.12)

Signed-off-by: Jakub Bogusz <redacted>

--- linux-2.4.27/arch/ppc/kernel/traps.c.orig	Wed Apr 14 15:05:27 2004
+++ linux-2.4.27/arch/ppc/kernel/traps.c	Mon Nov 29 19:05:28 2004

@@ -88,8 +88,10 @@
 	console_verbose();
 	spin_lock_irq(&die_lock);
 #ifdef CONFIG_PMAC_BACKLIGHT
-	set_backlight_enable(1);
-	set_backlight_level(BACKLIGHT_MAX);
+	if (_machine == _MACH_Pmac) {
+		set_backlight_enable(1);
+		set_backlight_level(BACKLIGHT_MAX);
+	}
 #endif
 	printk("Oops: %s, sig: %ld\n", str, err);
 	show_regs(fp);

-- 
Jakub Bogusz    http://qboosh.cs.net.pl/

On Tue, 2005-06-14 at 21:38 +0200, Jakub Bogusz wrote:

This patch avoids recursive crash (leading to kernel stack overflow) in
die() on CHRP/PReP machines when CONFIG_PMAC_BACKLIGHT=y.
set_backlight_* functions are placed in pmac section, which is discarded
when _machine != _MACH_Pmac.

I already posted this patch to LKML few months ago:
http://www.ussg.iu.edu/hypermail/linux/kernel/0412.0/0300.html
and it has been applied to linux-2.4 tree, but still not 2.6.
(patch was made against 2.4.27, but still applies cleanly against
kernels up to 2.6.11.12)

Signed-off-by: Jakub Bogusz <redacted>

Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>