Re: 2.5.70-bk[56] breaks disk partitioning with multiple IDE disks

Re: 2.5.70-bk[56] breaks disk partitioning with multiple IDE disks

[Adam J. Richter]

	To add to my own posting, here is a patch to 2.5.70-bk6
that reverts only the changes that replaced ata_unused with
idedeault_driver.drives.  This change makes bk6 work for me with
multiple IDE disk drives present.

	I have tried a number of smaller changes to stock
2.5.70-bk6/drivers/ide/ide.c, but, I have not yet found a change
simpler than this that works.

Adam J. Richter     __     ______________   575 Oroville Road
adam@yggdrasil.com     \ /                  Miplitas, California 95035
+1 408 309-6081         | g g d r a s i l   United States of America
                         "Free Software For The Rest Of Us."


--- linux-2.5.70-bk6/drivers/ide/ide.c	2003-06-01 12:01:28.000000000 -0700
+++ linux/drivers/ide/ide.c	2003-06-01 16:52:28.000000000 -0700
@@ -462,6 +462,7 @@
 	return -ENXIO;
 }
 
+static LIST_HEAD(ata_unused);
 static spinlock_t drives_lock = SPIN_LOCK_UNLOCKED;
 static spinlock_t drivers_lock = SPIN_LOCK_UNLOCKED;
 static LIST_HEAD(drivers);
@@ -1436,6 +1437,9 @@
 	spin_unlock(&drivers_lock);
 	if(idedefault_driver.attach(drive) != 0)
 		panic("ide: default attach failed");
+	spin_lock(&drives_lock);
+	list_add_tail(&drive->list, &ata_unused);
+	spin_unlock(&drives_lock);
 	return 1;
 }
 
@@ -2379,8 +2383,8 @@
 	spin_unlock_irqrestore(&ide_lock, flags);
 	spin_lock(&drives_lock);
 	list_del_init(&drive->list);
+	list_add(&drive->list, &drive->driver->drives);
 	spin_unlock(&drives_lock);
-	/* drive will be added to &idedefault_driver->drives in ata_attach() */
 	return 0;
 }
 
@@ -2403,9 +2407,9 @@
 	list_add(&driver->drivers, &drivers);
 	spin_unlock(&drivers_lock);
 
-	INIT_LIST_HEAD(&list);
 	spin_lock(&drives_lock);
-	list_splice_init(&idedefault_driver.drives, &list);
+	INIT_LIST_HEAD(&list);
+	list_splice_init(&ata_unused, &list);
 	spin_unlock(&drives_lock);
 
 	list_for_each_safe(list_loop, tmp_storage, &list) {

Re: 2.5.70-bk[56] breaks disk partitioning with multiple IDE disks

[William Lee Irwin III]

On Sun, Jun 01, 2003 at 05:45:09PM -0700, Adam J. Richter wrote:
> @@ -1436,6 +1437,9 @@
>  	spin_unlock(&drivers_lock);
>  	if(idedefault_driver.attach(drive) != 0)
>  		panic("ide: default attach failed");
> +	spin_lock(&drives_lock);
> +	list_add_tail(&drive->list, &ata_unused);
> +	spin_unlock(&drives_lock);
>  	return 1;
>  }

This looks dubious; check idedefault_driver.attach(). It was just put on
another list. I don't know what you're trying to do but list_move_tail()
would at least not be as oopsable.


-- wli

2.5.70-bk[56] breaks disk partitioning with multiple IDE disks

[Adam J. Richter]

	Disk partition under linux-2.5.70-bk[56] systems with two
IDE disks is broken.  Under these kernels, the first disk gets
the partitioning of the second disk.  Reverting drivers/ide/ide.c
to the 2.5.70-bk4 version makes the problem go away.

	However, there were other changes made to ide.c that are
probably not related to this problem, and I haven't yet analyzed the
problem enough to suggest a potentially correct patch.

	Note that I use user-level disk partition table parsing via
partx from util-linux and run ide as a loadable module.  It is
possible that other configuations might not experience this problem.

Adam J. Richter     __     ______________   575 Oroville Road
adam@yggdrasil.com     \ /                  Miplitas, California 95035
+1 408 309-6081         | g g d r a s i l   United States of America
                         "Free Software For The Rest Of Us."