[PATCH][2.5] wait for devices to wake up before mounting root, but don't hang

[PATCH][2.5] wait for devices to wake up before mounting root, but don't hang

[Lightweight patch manager]

Don't hang while waiting for root device to come up

Someone mentioned that USB devices etc. can't be used for root fs since 
it's probably not yet available at boot time. This approach is broken, for 
sure, but it's at least supposed to rescue a small part of the whole (if 
you see a panic, you know something went atree).

--- linux-2.5.20/init/do_mounts.c	Sun Jun  2 19:44:47 2002
+++ thunder-2.5.20/init/do_mounts.c	Mon Jun  3 17:06:25 2002
@@ -298,6 +298,7 @@
 }
 static void __init mount_block_root(char *name, int flags)
 {
+	long l = 0;
 	char *fs_names = __getname();
 	char *p;
 
@@ -318,7 +319,18 @@
 		 * Allow the user to distinguish between failed open
 		 * and bad superblock on root device.
 		 */
-		printk ("VFS: Cannot open root device \"%s\" or %s\n",
+		l++;
+		if (l < 60) {
+		    printk ("VFS: Cannot open root device \"%s\" or %s, %li retries left...\n",
+			    root_device_name, kdevname (ROOT_DEV), 60-l);
+	
+		    /* Wait a second and retry */
+		    current->state = TASK_UNINTERRUPTIBLE;
+		    schedule_timeout(HZ);
+	
+		    goto retry;
+		}
+		printk ("VFS: Cannot open root device \"%s\" or %s ultimately.\n",
 			root_device_name, kdevname (ROOT_DEV));
 		printk ("Please append a correct \"root=\" boot option\n");
 		panic("VFS: Unable to mount root fs on %s",
-- 
Lightweight patch manager using pine. If you have any objections, tell me.

Re: [PATCH][2.5] wait for devices to wake up before mounting root, but don't hang

[Greg KH]

On Mon, Jun 03, 2002 at 05:31:03PM -0600, Lightweight patch manager wrote:
> Don't hang while waiting for root device to come up
> 
> Someone mentioned that USB devices etc. can't be used for root fs since 
> it's probably not yet available at boot time. This approach is broken, for 
> sure, but it's at least supposed to rescue a small part of the whole (if 
> you see a panic, you know something went atree).

Why only loop for 60 seconds?

I prefer the patch posted previously by Eric Lammerts that just loops
forever.  But that's just me :)

thanks,

greg k-h