The patch titled
Subject: ocfs2: cluster: fix potential deadlock on &qs->qs_lock
has been added to the -mm mm-nonmm-unstable branch. Its filename is
ocfs2-cluster-fix-potential-deadlock-on-qs-qs_lock.patch
This patch will shortly appear at
https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/ocfs2-cluster-fix-potential-deadlock-on-qs-qs_lock.patch
This patch will later appear in the mm-nonmm-unstable branch at
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next via the mm-everything
branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
and is updated there every 2-3 working days
------------------------------------------------------
From: Chengfeng Ye <dg573847474@xxxxxxxxx>
Subject: ocfs2: cluster: fix potential deadlock on &qs->qs_lock
Date: Wed, 2 Aug 2023 12:38:24 +0000
&qs->qs_lock is acquired by timer o2net_idle_timer() along the following
call chain. Thus the acquisition of the lock under process context should
disable bottom half, otherwise deadlock could happen if the timer happens
to preempt the execution while the lock is held in process context on the
same CPU.
<timer interrupt>
-> o2net_idle_timer()
-> o2quo_conn_err()
-> spin_lock(&qs->qs_lock)
Several lock acquisition of &qs->qs_lock under process contex do not
disable irq or bottom half. The patch fixes these potential deadlocks
scenerio by using spin_lock_bh() on &qs->qs_lock.
This flaw was found by an experimental static analysis tool I am
developing for irq-related deadlock. x86_64 allmodconfig using gcc shows
no new warning.
Link: https://lkml.kernel.org/r/20230802123824.15301-1-dg573847474@xxxxxxxxx
Signed-off-by: Chengfeng Ye <dg573847474@xxxxxxxxx>
Cc: Mark Fasheh <mark@xxxxxxxxxx>
Cc: Joel Becker <jlbec@xxxxxxxxxxxx>
Cc: Junxiao Bi <junxiao.bi@xxxxxxxxxx>
Cc: Joseph Qi <jiangqi903@xxxxxxxxx>
Cc: Gang He <ghe@xxxxxxxx>
Cc: Jun Piao <piaojun@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
fs/ocfs2/cluster/quorum.c | 26 +++++++++++++-------------
1 file changed, 13 insertions(+), 13 deletions(-)
--- a/fs/ocfs2/cluster/quorum.c~ocfs2-cluster-fix-potential-deadlock-on-qs-qs_lock
+++ a/fs/ocfs2/cluster/quorum.c
@@ -93,7 +93,7 @@ static void o2quo_make_decision(struct w
int lowest_hb, lowest_reachable = 0, fence = 0;
struct o2quo_state *qs = &o2quo_state;
- spin_lock(&qs->qs_lock);
+ spin_lock_bh(&qs->qs_lock);
lowest_hb = find_first_bit(qs->qs_hb_bm, O2NM_MAX_NODES);
if (lowest_hb != O2NM_MAX_NODES)
@@ -146,14 +146,14 @@ static void o2quo_make_decision(struct w
out:
if (fence) {
- spin_unlock(&qs->qs_lock);
+ spin_unlock_bh(&qs->qs_lock);
o2quo_fence_self();
} else {
mlog(ML_NOTICE, "not fencing this node, heartbeating: %d, "
"connected: %d, lowest: %d (%sreachable)\n",
qs->qs_heartbeating, qs->qs_connected, lowest_hb,
lowest_reachable ? "" : "un");
- spin_unlock(&qs->qs_lock);
+ spin_unlock_bh(&qs->qs_lock);
}
@@ -196,7 +196,7 @@ void o2quo_hb_up(u8 node)
{
struct o2quo_state *qs = &o2quo_state;
- spin_lock(&qs->qs_lock);
+ spin_lock_bh(&qs->qs_lock);
qs->qs_heartbeating++;
mlog_bug_on_msg(qs->qs_heartbeating == O2NM_MAX_NODES,
@@ -211,7 +211,7 @@ void o2quo_hb_up(u8 node)
else
o2quo_clear_hold(qs, node);
- spin_unlock(&qs->qs_lock);
+ spin_unlock_bh(&qs->qs_lock);
}
/* hb going down releases any holds we might have had due to this node from
@@ -220,7 +220,7 @@ void o2quo_hb_down(u8 node)
{
struct o2quo_state *qs = &o2quo_state;
- spin_lock(&qs->qs_lock);
+ spin_lock_bh(&qs->qs_lock);
qs->qs_heartbeating--;
mlog_bug_on_msg(qs->qs_heartbeating < 0,
@@ -233,7 +233,7 @@ void o2quo_hb_down(u8 node)
o2quo_clear_hold(qs, node);
- spin_unlock(&qs->qs_lock);
+ spin_unlock_bh(&qs->qs_lock);
}
/* this tells us that we've decided that the node is still heartbeating
@@ -245,14 +245,14 @@ void o2quo_hb_still_up(u8 node)
{
struct o2quo_state *qs = &o2quo_state;
- spin_lock(&qs->qs_lock);
+ spin_lock_bh(&qs->qs_lock);
mlog(0, "node %u\n", node);
qs->qs_pending = 1;
o2quo_clear_hold(qs, node);
- spin_unlock(&qs->qs_lock);
+ spin_unlock_bh(&qs->qs_lock);
}
/* This is analogous to hb_up. as a node's connection comes up we delay the
@@ -264,7 +264,7 @@ void o2quo_conn_up(u8 node)
{
struct o2quo_state *qs = &o2quo_state;
- spin_lock(&qs->qs_lock);
+ spin_lock_bh(&qs->qs_lock);
qs->qs_connected++;
mlog_bug_on_msg(qs->qs_connected == O2NM_MAX_NODES,
@@ -279,7 +279,7 @@ void o2quo_conn_up(u8 node)
else
o2quo_clear_hold(qs, node);
- spin_unlock(&qs->qs_lock);
+ spin_unlock_bh(&qs->qs_lock);
}
/* we've decided that we won't ever be connecting to the node again. if it's
@@ -290,7 +290,7 @@ void o2quo_conn_err(u8 node)
{
struct o2quo_state *qs = &o2quo_state;
- spin_lock(&qs->qs_lock);
+ spin_lock_bh(&qs->qs_lock);
if (test_bit(node, qs->qs_conn_bm)) {
qs->qs_connected--;
@@ -307,7 +307,7 @@ void o2quo_conn_err(u8 node)
mlog(0, "node %u, %d total\n", node, qs->qs_connected);
- spin_unlock(&qs->qs_lock);
+ spin_unlock_bh(&qs->qs_lock);
}
void o2quo_init(void)
_
Patches currently in -mm which might be from dg573847474@xxxxxxxxx are
ocfs2-cluster-fix-potential-deadlock-on-qs-qs_lock.patch
ocfs2-cluster-fix-potential-deadlock-on-o2net_debug_lock.patch
