Re: backlight: oops in __mutex_lock_slowpath during head /sys/class/graphics/fb0/* in 2.6.18

[Michael Hanselmann <linux-kernel@hansmi.ch>]

On Thu, Sep 21, 2006 at 12:37:42PM -0700, Andrew Morton wrote:
> On Thu, 21 Sep 2006 14:19:52 +0200
> Olaf Hering <olaf@aepfle.de> wrote:
> > The bl_curve code has some room for improvement.

Agreed.

> > inst-sys:~ # dmesg
> > valkyriefb: vmode 13 does not support cmode 1.
> > Unable to handle kernel paging request for data at address 0x00000000
> > Faulting instruction address: 0xc02c7bd8

Seems like not all drivers use the framebuffer_alloc() function and
won't have an initialized mutex. But those don't have a backlight,
anyway.

What about the patch below? Does it work for you?

Signed-off-by: Michael Hanselmann <linux-kernel@hansmi.ch>

---
diff -Nrup --exclude-from linux-exclude-from linux-2.6.18.orig/drivers/video/fbsysfs.c linux-2.6.18/drivers/video/fbsysfs.c
--- linux-2.6.18.orig/drivers/video/fbsysfs.c	2006-09-21 23:40:06.000000000 +0200
+++ linux-2.6.18/drivers/video/fbsysfs.c	2006-09-21 23:50:00.000000000 +0200
@@ -397,6 +397,12 @@ static ssize_t store_bl_curve(struct cla
 	u8 tmp_curve[FB_BACKLIGHT_LEVELS];
 	unsigned int i;
 
+	/* Some drivers don't use framebuffer_alloc(), but those also
+	 * don't have backlights.
+	 */
+	if (!fb_info || !fb_info->bl_dev)
+		return -ENODEV;
+
 	if (count != (FB_BACKLIGHT_LEVELS / 8 * 24))
 		return -EINVAL;
 
@@ -430,6 +436,12 @@ static ssize_t show_bl_curve(struct clas
 	ssize_t len = 0;
 	unsigned int i;
 
+	/* Some drivers don't use framebuffer_alloc(), but those also
+	 * don't have backlights.
+	 */
+	if (!fb_info || !fb_info->bl_dev)
+		return -ENODEV;
+
 	mutex_lock(&fb_info->bl_mutex);
 	for (i = 0; i < FB_BACKLIGHT_LEVELS; i += 8)
 		len += snprintf(&buf[len], PAGE_SIZE,