[<prev] [next>] [day] [month] [year] [list]
Message-ID: <20190216194423.3727d498@draws-f6bvp>
Date: Sat, 16 Feb 2019 19:44:23 +0100
From: f6bvp <f6bvp@...e.fr>
To: netdev@...r.kernel.org, linux-hams@...r.kernel.org
Cc: dvyukov@...gle.com, richard@...-stearn.co.uk, thomas@...erried.de,
f6bvp@...at.org
Subject: [ROSE] Bug : waiting for rose module to become free
Hi,
A nastly bug is still pending in ROSE module that prevents it to be removed.
When trying to remove rose module we enter into an endless loop with displayed message :
"waiting for rose module to become free"
A patch from Richard Stern had been committed a while ago but it did not get a chance to be applied.
https://marc.info/?l=linux-hams&m=150870350516064&w=2
Since then, there have been some changes in af_rose.c and net code architecture that prevents simply adding missing dev_put(dev) in present kernel.
>From time to time we can observe a negative count (-1 for integer value is displayed as 65535 in /proc/net/rose_neigh) that probably means there is a bug in rose module refcount (??).
/proc/net/rose_neigh
00015 SV1HCC-9 ax0 2 1 DCE yes 0 0
00013 PE1OWG-11 ax0 2 65535 DCE yes 0 0
00011 F5KTR-11 ax0 1 1 DCE yes 0 0
00007 F3KT-11 ax0 2 1 DCE yes 0 0
00003 SK7ES-9 ax0 2 1 DTE yes 0 0
00002 K4GBB-9 ax0 4 0 DCE yes 0 0
00001 RSLOOP-0 ??? 1 2 DCE yes 0 0
Help is kindly requested to get a valid patch.
Bernard,
f6bvp
Powered by blists - more mailing lists