linux-kernel - [PATCH 000 of 4] Introduction

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Wed, 11 Oct 2006 16:09:01 +1000
From:	NeilBrown <neilb@...e.de>
To:	Andrew Morton <akpm@...l.org>
Cc:	Al Viro <viro@....linux.org.uk>
Subject: [PATCH 000 of 4] Introduction

Following 4 patches address issues with lockdep, particularly around bd_mutex.

They are against 2.6.18-mm3 and do *not* apply against -linus as -mm already has
some changes to the handling of bd_mutex nesting.  2-4 probably apply on top of -linus plus
-mm/broken-out/remove-the-old-bd_mutex-lockdep-annotation.patch  

I believe they are probably ok for 2.6.19.

The core issue is that blkdev_get when called on a partition needs to
lock (bd_mutex) the partition and the whole device.  lockdep would
normally see this as a possible deadlock and needs to be told that
this particular nesting is known to be safe.

The code to do this is in -linus is rather messy, largely because the
locking itself is messy.  The bd_mutex for the whole is taken several
times while bd_mutex for the partition is held, and it is taken at
both levels of the recursion (blkdev_get calls blkdev_get - only to
one level).

As key observation to simplifying the locking is to observer that a
lot of the locking is there to protect the updating of bd_part_count.
If those updates are moved, the locking can become simpler.

The first patch removes the current approach in -mm to handling this
nesting and explains why it is not ideal.
This reverts new-bd_mutex-lockdep-annotation.patch

The second simplifies the locking as explained above.

The third adds the mutex_lock_nested annotations, which are now trivial.

The last fixes a tangentially related lockdep problem in md - there is
a false relationship between bd_mutex and md->reconfig_mutex which
needs to be clarified.

 [PATCH 000 of 4] Introduction
 [PATCH 001 of 4] Remove lock_key approach to managing nested bd_mutex locks.
 [PATCH 002 of 4] Simplify some aspects of bd_mutex nesting.
 [PATCH 003 of 4] Use mutex_lock_nested for bd_mutex to avoid  lockdep warning.
 [PATCH 004 of 4] Avoid lockdep warning in md.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/