Message ID | 27278b212848db65313d91d3c39407212bc21fe8.1524146802.git.g.nault@alphalink.fr |
---|---|
State | Accepted, archived |
Delegated to: | David Miller |
Headers | show |
Series | [net] l2tp: fix {pppol2tp,l2tp_dfs}_seq_stop() in case of seq_file overflow | expand |
From: Guillaume Nault <g.nault@alphalink.fr> Date: Thu, 19 Apr 2018 16:20:48 +0200 > Commit 0e0c3fee3a59 ("l2tp: hold reference on tunnels printed in pppol2tp proc file") > assumed that if pppol2tp_seq_stop() was called with non-NULL private > data (the 'v' pointer), then pppol2tp_seq_start() would not be called > again. It turns out that this isn't guaranteed, and overflowing the > seq_file's buffer in pppol2tp_seq_show() is a way to get into this > situation. > > Therefore, pppol2tp_seq_stop() needs to reset pd->tunnel, so that > pppol2tp_seq_start() won't drop a reference again if it gets called. > We also have to clear pd->session, because the rest of the code expects > a non-NULL tunnel when pd->session is set. > > The l2tp_debugfs module has the same issue. Fix it in the same way. > > Fixes: 0e0c3fee3a59 ("l2tp: hold reference on tunnels printed in pppol2tp proc file") > Fixes: f726214d9b23 ("l2tp: hold reference on tunnels printed in l2tp/tunnels debugfs file") > Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> Applied and queued up for -stable, thank you.
diff --git a/net/l2tp/l2tp_debugfs.c b/net/l2tp/l2tp_debugfs.c index b8f9d45bfeb1..7f1e842ef05a 100644 --- a/net/l2tp/l2tp_debugfs.c +++ b/net/l2tp/l2tp_debugfs.c @@ -106,8 +106,11 @@ static void l2tp_dfs_seq_stop(struct seq_file *p, void *v) return; /* Drop reference taken by last invocation of l2tp_dfs_next_tunnel() */ - if (pd->tunnel) + if (pd->tunnel) { l2tp_tunnel_dec_refcount(pd->tunnel); + pd->tunnel = NULL; + pd->session = NULL; + } } static void l2tp_dfs_seq_tunnel_show(struct seq_file *m, void *v) diff --git a/net/l2tp/l2tp_ppp.c b/net/l2tp/l2tp_ppp.c index 7d0c963680e6..1404bc1c1bb7 100644 --- a/net/l2tp/l2tp_ppp.c +++ b/net/l2tp/l2tp_ppp.c @@ -1618,8 +1618,11 @@ static void pppol2tp_seq_stop(struct seq_file *p, void *v) return; /* Drop reference taken by last invocation of pppol2tp_next_tunnel() */ - if (pd->tunnel) + if (pd->tunnel) { l2tp_tunnel_dec_refcount(pd->tunnel); + pd->tunnel = NULL; + pd->session = NULL; + } } static void pppol2tp_seq_tunnel_show(struct seq_file *m, void *v)
Commit 0e0c3fee3a59 ("l2tp: hold reference on tunnels printed in pppol2tp proc file") assumed that if pppol2tp_seq_stop() was called with non-NULL private data (the 'v' pointer), then pppol2tp_seq_start() would not be called again. It turns out that this isn't guaranteed, and overflowing the seq_file's buffer in pppol2tp_seq_show() is a way to get into this situation. Therefore, pppol2tp_seq_stop() needs to reset pd->tunnel, so that pppol2tp_seq_start() won't drop a reference again if it gets called. We also have to clear pd->session, because the rest of the code expects a non-NULL tunnel when pd->session is set. The l2tp_debugfs module has the same issue. Fix it in the same way. Fixes: 0e0c3fee3a59 ("l2tp: hold reference on tunnels printed in pppol2tp proc file") Fixes: f726214d9b23 ("l2tp: hold reference on tunnels printed in l2tp/tunnels debugfs file") Signed-off-by: Guillaume Nault <g.nault@alphalink.fr> --- net/l2tp/l2tp_debugfs.c | 5 ++++- net/l2tp/l2tp_ppp.c | 5 ++++- 2 files changed, 8 insertions(+), 2 deletions(-) Reproduced by tweaking .show() to write overly long data. I don't believe the bug can be triggered in practice, but this behaviour is part of the seq_file API and the fix doesn't increase complexity.