linux-kernel - [PATCH 5.15 105/136] writeback: avoid use-after-free after removing device

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]

Date:   Mon, 29 Aug 2022 12:59:32 +0200
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     linux-kernel@...r.kernel.org
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        stable@...r.kernel.org, Khazhismel Kumykov <khazhy@...gle.com>,
        Jan Kara <jack@...e.cz>,
        Michael Stapelberg <stapelberg+linux@...gle.com>,
        Wu Fengguang <fengguang.wu@...el.com>,
        Alexander Viro <viro@...iv.linux.org.uk>,
        Andrew Morton <akpm@...ux-foundation.org>
Subject: [PATCH 5.15 105/136] writeback: avoid use-after-free after removing device

From: Khazhismel Kumykov <khazhy@...omium.org>

commit f87904c075515f3e1d8f4a7115869d3b914674fd upstream.

When a disk is removed, bdi_unregister gets called to stop further
writeback and wait for associated delayed work to complete.  However,
wb_inode_writeback_end() may schedule bandwidth estimation dwork after
this has completed, which can result in the timer attempting to access the
just freed bdi_writeback.

Fix this by checking if the bdi_writeback is alive, similar to when
scheduling writeback work.

Since this requires wb->work_lock, and wb_inode_writeback_end() may get
called from interrupt, switch wb->work_lock to an irqsafe lock.

Link: https://lkml.kernel.org/r/20220801155034.3772543-1-khazhy@google.com
Fixes: 45a2966fd641 ("writeback: fix bandwidth estimate for spiky workload")
Signed-off-by: Khazhismel Kumykov <khazhy@...gle.com>
Reviewed-by: Jan Kara <jack@...e.cz>
Cc: Michael Stapelberg <stapelberg+linux@...gle.com>
Cc: Wu Fengguang <fengguang.wu@...el.com>
Cc: Alexander Viro <viro@...iv.linux.org.uk>
Cc: <stable@...r.kernel.org>
Signed-off-by: Andrew Morton <akpm@...ux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
---
 fs/fs-writeback.c   |   12 ++++++------
 mm/backing-dev.c    |   10 +++++-----
 mm/page-writeback.c |    6 +++++-
 3 files changed, 16 insertions(+), 12 deletions(-)

--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -134,10 +134,10 @@ static bool inode_io_list_move_locked(st
 
 static void wb_wakeup(struct bdi_writeback *wb)
 {
-	spin_lock_bh(&wb->work_lock);
+	spin_lock_irq(&wb->work_lock);
 	if (test_bit(WB_registered, &wb->state))
 		mod_delayed_work(bdi_wq, &wb->dwork, 0);
-	spin_unlock_bh(&wb->work_lock);
+	spin_unlock_irq(&wb->work_lock);
 }
 
 static void finish_writeback_work(struct bdi_writeback *wb,
@@ -164,7 +164,7 @@ static void wb_queue_work(struct bdi_wri
 	if (work->done)
 		atomic_inc(&work->done->cnt);
 
-	spin_lock_bh(&wb->work_lock);
+	spin_lock_irq(&wb->work_lock);
 
 	if (test_bit(WB_registered, &wb->state)) {
 		list_add_tail(&work->list, &wb->work_list);
@@ -172,7 +172,7 @@ static void wb_queue_work(struct bdi_wri
 	} else
 		finish_writeback_work(wb, work);
 
-	spin_unlock_bh(&wb->work_lock);
+	spin_unlock_irq(&wb->work_lock);
 }
 
 /**
@@ -2109,13 +2109,13 @@ static struct wb_writeback_work *get_nex
 {
 	struct wb_writeback_work *work = NULL;
 
-	spin_lock_bh(&wb->work_lock);
+	spin_lock_irq(&wb->work_lock);
 	if (!list_empty(&wb->work_list)) {
 		work = list_entry(wb->work_list.next,
 				  struct wb_writeback_work, list);
 		list_del_init(&work->list);
 	}
-	spin_unlock_bh(&wb->work_lock);
+	spin_unlock_irq(&wb->work_lock);
 	return work;
 }
 
--- a/mm/backing-dev.c
+++ b/mm/backing-dev.c
@@ -258,10 +258,10 @@ void wb_wakeup_delayed(struct bdi_writeb
 	unsigned long timeout;
 
 	timeout = msecs_to_jiffies(dirty_writeback_interval * 10);
-	spin_lock_bh(&wb->work_lock);
+	spin_lock_irq(&wb->work_lock);
 	if (test_bit(WB_registered, &wb->state))
 		queue_delayed_work(bdi_wq, &wb->dwork, timeout);
-	spin_unlock_bh(&wb->work_lock);
+	spin_unlock_irq(&wb->work_lock);
 }
 
 static void wb_update_bandwidth_workfn(struct work_struct *work)
@@ -337,12 +337,12 @@ static void cgwb_remove_from_bdi_list(st
 static void wb_shutdown(struct bdi_writeback *wb)
 {
 	/* Make sure nobody queues further work */
-	spin_lock_bh(&wb->work_lock);
+	spin_lock_irq(&wb->work_lock);
 	if (!test_and_clear_bit(WB_registered, &wb->state)) {
-		spin_unlock_bh(&wb->work_lock);
+		spin_unlock_irq(&wb->work_lock);
 		return;
 	}
-	spin_unlock_bh(&wb->work_lock);
+	spin_unlock_irq(&wb->work_lock);
 
 	cgwb_remove_from_bdi_list(wb);
 	/*
--- a/mm/page-writeback.c
+++ b/mm/page-writeback.c
@@ -2755,6 +2755,7 @@ static void wb_inode_writeback_start(str
 
 static void wb_inode_writeback_end(struct bdi_writeback *wb)
 {
+	unsigned long flags;
 	atomic_dec(&wb->writeback_inodes);
 	/*
 	 * Make sure estimate of writeback throughput gets updated after
@@ -2763,7 +2764,10 @@ static void wb_inode_writeback_end(struc
 	 * that if multiple inodes end writeback at a similar time, they get
 	 * batched into one bandwidth update.
 	 */
-	queue_delayed_work(bdi_wq, &wb->bw_dwork, BANDWIDTH_INTERVAL);
+	spin_lock_irqsave(&wb->work_lock, flags);
+	if (test_bit(WB_registered, &wb->state))
+		queue_delayed_work(bdi_wq, &wb->bw_dwork, BANDWIDTH_INTERVAL);
+	spin_unlock_irqrestore(&wb->work_lock, flags);
 }
 
 int test_clear_page_writeback(struct page *page)