linux-kernel - [PATCH 5.4 13/43] net/mlx5: Fix a race on command flush flow

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20220314112734.791652369@linuxfoundation.org>
Date:   Mon, 14 Mar 2022 12:53:24 +0100
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     linux-kernel@...r.kernel.org
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        stable@...r.kernel.org, Moshe Shemesh <moshe@...dia.com>,
        Eran Ben Elisha <eranbe@...dia.com>,
        Saeed Mahameed <saeedm@...dia.com>,
        Sasha Levin <sashal@...nel.org>
Subject: [PATCH 5.4 13/43] net/mlx5: Fix a race on command flush flow

From: Moshe Shemesh <moshe@...dia.com>

[ Upstream commit 063bd355595428750803d8736a9bb7c8db67d42d ]

Fix a refcount use after free warning due to a race on command entry.
Such race occurs when one of the commands releases its last refcount and
frees its index and entry while another process running command flush
flow takes refcount to this command entry. The process which handles
commands flush may see this command as needed to be flushed if the other
process released its refcount but didn't release the index yet. Fix it
by adding the needed spin lock.

It fixes the following warning trace:

refcount_t: addition on 0; use-after-free.
WARNING: CPU: 11 PID: 540311 at lib/refcount.c:25 refcount_warn_saturate+0x80/0xe0
...
RIP: 0010:refcount_warn_saturate+0x80/0xe0
...
Call Trace:
 <TASK>
 mlx5_cmd_trigger_completions+0x293/0x340 [mlx5_core]
 mlx5_cmd_flush+0x3a/0xf0 [mlx5_core]
 enter_error_state+0x44/0x80 [mlx5_core]
 mlx5_fw_fatal_reporter_err_work+0x37/0xe0 [mlx5_core]
 process_one_work+0x1be/0x390
 worker_thread+0x4d/0x3d0
 ? rescuer_thread+0x350/0x350
 kthread+0x141/0x160
 ? set_kthread_struct+0x40/0x40
 ret_from_fork+0x1f/0x30
 </TASK>

Fixes: 50b2412b7e78 ("net/mlx5: Avoid possible free of command entry while timeout comp handler")
Signed-off-by: Moshe Shemesh <moshe@...dia.com>
Reviewed-by: Eran Ben Elisha <eranbe@...dia.com>
Signed-off-by: Saeed Mahameed <saeedm@...dia.com>
Signed-off-by: Sasha Levin <sashal@...nel.org>
---
 drivers/net/ethernet/mellanox/mlx5/core/cmd.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
index 1a7aa078f351..6c7b364d0bf0 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
@@ -130,11 +130,8 @@ static int cmd_alloc_index(struct mlx5_cmd *cmd)

 static void cmd_free_index(struct mlx5_cmd *cmd, int idx)
 {
-	unsigned long flags;
-
-	spin_lock_irqsave(&cmd->alloc_lock, flags);
+	lockdep_assert_held(&cmd->alloc_lock);
 	set_bit(idx, &cmd->bitmask);
-	spin_unlock_irqrestore(&cmd->alloc_lock, flags);
 }

 static void cmd_ent_get(struct mlx5_cmd_work_ent *ent)
@@ -144,17 +141,21 @@ static void cmd_ent_get(struct mlx5_cmd_work_ent *ent)

 static void cmd_ent_put(struct mlx5_cmd_work_ent *ent)
 {
+	struct mlx5_cmd *cmd = ent->cmd;
+	unsigned long flags;
+
+	spin_lock_irqsave(&cmd->alloc_lock, flags);
 	if (!refcount_dec_and_test(&ent->refcnt))
-		return;
+		goto out;

 	if (ent->idx >= 0) {
-		struct mlx5_cmd *cmd = ent->cmd;
-
 		cmd_free_index(cmd, ent->idx);
 		up(ent->page_queue ? &cmd->pages_sem : &cmd->sem);
 	}

 	cmd_free_ent(ent);
+out:
+	spin_unlock_irqrestore(&cmd->alloc_lock, flags);
 }

 static struct mlx5_cmd_layout *get_inst(struct mlx5_cmd *cmd, int idx)
-- 
2.34.1