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]
Message-ID: <20231221150444.1e47a0377f80.If7e8ba721ba2956f12c6e8405e7d61e154aa7ae7@changeid>
Date: Thu, 21 Dec 2023 15:04:45 +0100
From: Johannes Berg <johannes@...solutions.net>
To: linux-kernel@...r.kernel.org
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Johannes Berg <johannes.berg@...el.com>,
	Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
Subject: [PATCH] debugfs: initialize cancellations earlier

From: Johannes Berg <johannes.berg@...el.com>

Tetsuo Handa pointed out that in the (now reverted)
lockdep commit I initialized the data too late. The
same is true for the cancellation data, it must be
initialized before the cmpxchg(), otherwise it may
be done twice and possibly even overwriting data in
there already when there's a race. Fix that, which
also requires destroying the mutex in case we lost
the race.

Fixes: 8c88a474357e ("debugfs: add API to allow debugfs operations cancellation")
Reported-by: Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
Signed-off-by: Johannes Berg <johannes.berg@...el.com>
---
 fs/debugfs/file.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/fs/debugfs/file.c b/fs/debugfs/file.c
index 5063434be0fc..6d7c1a49581f 100644
--- a/fs/debugfs/file.c
+++ b/fs/debugfs/file.c
@@ -104,12 +104,14 @@ int debugfs_file_get(struct dentry *dentry)
 					~DEBUGFS_FSDATA_IS_REAL_FOPS_BIT);
 		refcount_set(&fsd->active_users, 1);
 		init_completion(&fsd->active_users_drained);
+		INIT_LIST_HEAD(&fsd->cancellations);
+		mutex_init(&fsd->cancellations_mtx);
+
 		if (cmpxchg(&dentry->d_fsdata, d_fsd, fsd) != d_fsd) {
+			mutex_destroy(&fsd->cancellations_mtx);
 			kfree(fsd);
 			fsd = READ_ONCE(dentry->d_fsdata);
 		}
-		INIT_LIST_HEAD(&fsd->cancellations);
-		mutex_init(&fsd->cancellations_mtx);
 	}
 
 	/*
-- 
2.43.0


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ