[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <159171920664.3038039.18059422273265286162.stgit@warthog.procyon.org.uk>
Date: Tue, 09 Jun 2020 17:13:26 +0100
From: David Howells <dhowells@...hat.com>
To: linux-afs@...ts.infradead.org
Cc: Kees Cook <keescook@...omium.org>, dhowells@...hat.com,
linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: [PATCH 3/6] afs: Fix use of BUG()
Fix afs_compare_addrs() to use WARN_ON(1) instead of BUG() and return 1
(ie. srx_a > srx_b).
There's no point trying to put actual error handling in as this should not
occur unless a new transport address type is allowed by AFS. And even if
it does, in this particular case, it'll just never match unknown types of
addresses. This BUG() was more of a 'you need to add a case here'
indicator.
Reported-by: Kees Cook <keescook@...omium.org>
Signed-off-by: David Howells <dhowells@...hat.com>
---
fs/afs/vl_alias.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/fs/afs/vl_alias.c b/fs/afs/vl_alias.c
index 093895c49c21..136fc6164e00 100644
--- a/fs/afs/vl_alias.c
+++ b/fs/afs/vl_alias.c
@@ -73,7 +73,8 @@ static int afs_compare_addrs(const struct sockaddr_rxrpc *srx_a,
}
default:
- BUG();
+ WARN_ON(1);
+ diff = 1;
}
out:
Powered by blists - more mailing lists