All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH v2] mac80211: mesh: fix crash in mesh_path_timer
@ 2016-03-19  2:03 Bob Copeland
  2016-04-05 10:19 ` Johannes Berg
  0 siblings, 1 reply; 2+ messages in thread
From: Bob Copeland @ 2016-03-19  2:03 UTC (permalink / raw)
  To: Johannes Berg; +Cc: linux-wireless, Bob Copeland

The mesh_path_reclaim() function, called from an rcu callback, cancels
the mesh_path_timer associated with a mesh path.  Unfortunately, this
call can happen much later, perhaps after the hash table itself is
destroyed.

Such a situation led to the following crash in mesh_path_send_to_gates()
when dereferencing the tbl pointer:

[   23.901661] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008
[   23.905516] IP: [<ffffffff814c910b>] mesh_path_send_to_gates+0x2b/0x740
[   23.908757] PGD 99ca067 PUD 99c4067 PMD 0
[   23.910789] Oops: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC
[   23.913485] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.5.0-rc6-wt+ #43
[   23.916675] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Debian-1.8.2-1 04/01/2014
[   23.920471] task: ffffffff81685500 ti: ffffffff81678000 task.ti: ffffffff81678000
[   23.922619] RIP: 0010:[<ffffffff814c910b>]  [<ffffffff814c910b>] mesh_path_send_to_gates+0x2b/0x740
[   23.925237] RSP: 0018:ffff88000b403d30  EFLAGS: 00010286
[   23.926739] RAX: 0000000000000000 RBX: ffff880009bc0d20 RCX: 0000000000000102
[   23.928796] RDX: 000000000000002e RSI: 0000000000000001 RDI: ffff880009bc0d20
[   23.930895] RBP: ffff88000b403e18 R08: 0000000000000001 R09: 0000000000000001
[   23.932917] R10: 0000000000000000 R11: 0000000000000001 R12: ffff880009c20940
[   23.936370] R13: ffff880009bc0e70 R14: ffff880009c21c40 R15: ffff880009bc0d20
[   23.939823] FS:  0000000000000000(0000) GS:ffff88000b400000(0000) knlGS:0000000000000000
[   23.943688] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[   23.946429] CR2: 0000000000000008 CR3: 00000000099c5000 CR4: 00000000000006b0
[   23.949861] Stack:
[   23.950840]  000000000000002e ffff880009c20940 ffff88000b403da8 ffffffff8109e551
[   23.954467]  ffffffff82711be2 000000000000002e 0000000000000000 ffffffff8166a5f5
[   23.958141]  0000000000685ce8 0000000000000246 ffff880009bc0d20 ffff880009c20940
[   23.961801] Call Trace:
[   23.962987]  <IRQ>
[   23.963963]  [<ffffffff8109e551>] ? vprintk_emit+0x351/0x5e0
[   23.966782]  [<ffffffff8109e8ff>] ? vprintk_default+0x1f/0x30
[   23.969529]  [<ffffffff810ffa41>] ? printk+0x48/0x50
[   23.971956]  [<ffffffff814ceef3>] mesh_path_timer+0x133/0x160
[   23.974707]  [<ffffffff814cedc0>] ? mesh_nexthop_resolve+0x230/0x230
[   23.977775]  [<ffffffff810b04ee>] call_timer_fn+0xce/0x330
[   23.980448]  [<ffffffff810b0425>] ? call_timer_fn+0x5/0x330
[   23.983126]  [<ffffffff814cedc0>] ? mesh_nexthop_resolve+0x230/0x230
[   23.986091]  [<ffffffff810b097c>] run_timer_softirq+0x22c/0x390

Instead of cancelling in the RCU callback, set a new flag to prevent the
timer from being rearmed, and then cancel the timer synchronously when
freeing the mesh path.  This leaves mesh_path_reclaim() doing nothing
but kfree, so switch to kfree_rcu().

Fixes: 3b302ada7f0a ("mac80211: mesh: move path tables into if_mesh")
Signed-off-by: Bob Copeland <me@bobcopeland.com>

---
v2: fix lock imbalance and remove extra forward decl

 net/mac80211/mesh.h         |  3 +++
 net/mac80211/mesh_hwmp.c    |  4 ++++
 net/mac80211/mesh_pathtbl.c | 33 ++++++++++++++++++---------------
 3 files changed, 25 insertions(+), 15 deletions(-)

diff --git a/net/mac80211/mesh.h b/net/mac80211/mesh.h
index cc6854db156e..e1415c952e9c 100644
--- a/net/mac80211/mesh.h
+++ b/net/mac80211/mesh.h
@@ -32,6 +32,8 @@
  * @MESH_PATH_RESOLVED: the mesh path can has been resolved
  * @MESH_PATH_REQ_QUEUED: there is an unsent path request for this destination
  *	already queued up, waiting for the discovery process to start.
+ * @MESH_PATH_DELETED: the mesh path has been deleted and should no longer
+ *	be used
  *
  * MESH_PATH_RESOLVED is used by the mesh path timer to
  * decide when to stop or cancel the mesh path discovery.
@@ -43,6 +45,7 @@ enum mesh_path_flags {
 	MESH_PATH_FIXED	=	BIT(3),
 	MESH_PATH_RESOLVED =	BIT(4),
 	MESH_PATH_REQ_QUEUED =	BIT(5),
+	MESH_PATH_DELETED =	BIT(6),
 };
 
 /**
diff --git a/net/mac80211/mesh_hwmp.c b/net/mac80211/mesh_hwmp.c
index 5b6aec1a0630..2748cf627ee3 100644
--- a/net/mac80211/mesh_hwmp.c
+++ b/net/mac80211/mesh_hwmp.c
@@ -1012,6 +1012,10 @@ void mesh_path_start_discovery(struct ieee80211_sub_if_data *sdata)
 		goto enddiscovery;
 
 	spin_lock_bh(&mpath->state_lock);
+	if (mpath->flags & MESH_PATH_DELETED) {
+		spin_unlock_bh(&mpath->state_lock);
+		goto enddiscovery;
+	}
 	mpath->flags &= ~MESH_PATH_REQ_QUEUED;
 	if (preq_node->flags & PREQ_Q_F_START) {
 		if (mpath->flags & MESH_PATH_RESOLVING) {
diff --git a/net/mac80211/mesh_pathtbl.c b/net/mac80211/mesh_pathtbl.c
index 7455397f8c3b..1c9412a29ca3 100644
--- a/net/mac80211/mesh_pathtbl.c
+++ b/net/mac80211/mesh_pathtbl.c
@@ -18,6 +18,8 @@
 #include "ieee80211_i.h"
 #include "mesh.h"
 
+static void mesh_path_free_rcu(struct mesh_table *tbl, struct mesh_path *mpath);
+
 static u32 mesh_table_hash(const void *addr, u32 len, u32 seed)
 {
 	/* Use last four bytes of hw addr as hash index */
@@ -40,18 +42,12 @@ static inline bool mpath_expired(struct mesh_path *mpath)
 	       !(mpath->flags & MESH_PATH_FIXED);
 }
 
-static void mesh_path_reclaim(struct rcu_head *rp)
-{
-	struct mesh_path *mpath = container_of(rp, struct mesh_path, rcu);
-
-	del_timer_sync(&mpath->timer);
-	kfree(mpath);
-}
-
-static void mesh_path_rht_free(void *ptr, void *unused_arg)
+static void mesh_path_rht_free(void *ptr, void *tblptr)
 {
 	struct mesh_path *mpath = ptr;
-	call_rcu(&mpath->rcu, mesh_path_reclaim);
+	struct mesh_table *tbl = tblptr;
+
+	mesh_path_free_rcu(tbl, mpath);
 }
 
 static struct mesh_table *mesh_table_alloc(void)
@@ -77,7 +73,7 @@ static struct mesh_table *mesh_table_alloc(void)
 static void mesh_table_free(struct mesh_table *tbl)
 {
 	rhashtable_free_and_destroy(&tbl->rhead,
-				    mesh_path_rht_free, NULL);
+				    mesh_path_rht_free, tbl);
 	kfree(tbl);
 }
 
@@ -551,18 +547,25 @@ out:
 	rhashtable_walk_exit(&iter);
 }
 
-static void __mesh_path_del(struct mesh_table *tbl, struct mesh_path *mpath)
+static void mesh_path_free_rcu(struct mesh_table *tbl,
+			       struct mesh_path *mpath)
 {
 	struct ieee80211_sub_if_data *sdata = mpath->sdata;
 
-	rhashtable_remove_fast(&tbl->rhead, &mpath->rhash, mesh_rht_params);
 	spin_lock_bh(&mpath->state_lock);
-	mpath->flags |= MESH_PATH_RESOLVING;
+	mpath->flags |= MESH_PATH_RESOLVING | MESH_PATH_DELETED;
 	mesh_gate_del(tbl, mpath);
-	call_rcu(&mpath->rcu, mesh_path_reclaim);
 	spin_unlock_bh(&mpath->state_lock);
+	del_timer_sync(&mpath->timer);
 	atomic_dec(&sdata->u.mesh.mpaths);
 	atomic_dec(&tbl->entries);
+	kfree_rcu(mpath, rcu);
+}
+
+static void __mesh_path_del(struct mesh_table *tbl, struct mesh_path *mpath)
+{
+	rhashtable_remove_fast(&tbl->rhead, &mpath->rhash, mesh_rht_params);
+	mesh_path_free_rcu(tbl, mpath);
 }
 
 /**
-- 
2.6.1


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH v2] mac80211: mesh: fix crash in mesh_path_timer
  2016-03-19  2:03 [PATCH v2] mac80211: mesh: fix crash in mesh_path_timer Bob Copeland
@ 2016-04-05 10:19 ` Johannes Berg
  0 siblings, 0 replies; 2+ messages in thread
From: Johannes Berg @ 2016-04-05 10:19 UTC (permalink / raw)
  To: Bob Copeland; +Cc: linux-wireless

On Fri, 2016-03-18 at 22:03 -0400, Bob Copeland wrote:
> The mesh_path_reclaim() function, called from an rcu callback,
> cancels
> the mesh_path_timer associated with a mesh path.  Unfortunately, this
> call can happen much later, perhaps after the hash table itself is
> destroyed.
> 
[...]

applied.

johannes

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2016-04-05 10:19 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-03-19  2:03 [PATCH v2] mac80211: mesh: fix crash in mesh_path_timer Bob Copeland
2016-04-05 10:19 ` Johannes Berg

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.