linux-kernel - Re: [syzbot] [iomap?] kernel BUG in folio_end

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: <87tsz7iea2.fsf@jogness.linutronix.de>
Date: Thu, 06 Nov 2025 12:42:21 +0106
From: John Ogness <john.ogness@...utronix.de>
To: Petr Mladek <pmladek@...e.com>
Cc: Joanne Koong <joannelkoong@...il.com>, syzbot
 <syzbot+3686758660f980b402dc@...kaller.appspotmail.com>,
 "amurray@...goodpenguin.co.uk" <amurray@...goodpenguin.co.uk>,
 brauner@...nel.org, chao@...nel.org, djwong@...nel.org,
 jaegeuk@...nel.org, linux-f2fs-devel@...ts.sourceforge.net,
 linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
 linux-xfs@...r.kernel.org, syzkaller-bugs@...glegroups.com
Subject: Re: [syzbot] [iomap?] kernel BUG in folio_end_read (2)

On 2025-11-05, John Ogness <john.ogness@...utronix.de> wrote:
>> Another question is whether this is the only problem caused the patch.
>
> This comparison is quite special. It caught my attention while combing
> through the code.

The reason that this comparison is special is because it is the only one
that does not take wrapping into account. I did it that way originally
because it is AND with a wrap check. But this is an ugly special
case. It should use the same wrap check as the other 3 cases in
nbcon.c. If it had, the bug would not have happened.

I always considered these wrap checks to be non-obvious and
error-prone. So what if we create a nice helper function to simplify and
unify the wrap checks? Something like this:

diff --git a/kernel/printk/printk_ringbuffer.c b/kernel/printk/printk_ringbuffer.c
index 839f504db6d30..8499ee642c31d 100644
--- a/kernel/printk/printk_ringbuffer.c
+++ b/kernel/printk/printk_ringbuffer.c
@@ -390,6 +390,17 @@ static unsigned int to_blk_size(unsigned int size)
 	return size;
 }
 
+/*
+ * Check if @lpos1 is before @lpos2. This takes ringbuffer wrapping
+ * into account. If @lpos1 is more than a full wrap before @lpos2,
+ * it is considered to be after @lpos2.
+ */
+static bool lpos1_before_lpos2(struct prb_data_ring *data_ring,
+			       unsigned long lpos1, unsigned long lpos2)
+{
+	return lpos2 - lpos1 - 1 < DATA_SIZE(data_ring);
+}
+
 /*
  * Sanity checker for reserve size. The ringbuffer code assumes that a data
  * block does not exceed the maximum possible size that could fit within the
@@ -577,7 +588,7 @@ static bool data_make_reusable(struct printk_ringbuffer *rb,
 	unsigned long id;
 
 	/* Loop until @lpos_begin has advanced to or beyond @lpos_end. */
-	while ((lpos_end - lpos_begin) - 1 < DATA_SIZE(data_ring)) {
+	while (lpos1_before_lpos2(data_ring, lpos_begin, lpos_end)) {
 		blk = to_block(data_ring, lpos_begin);
 
 		/*
@@ -668,7 +679,7 @@ static bool data_push_tail(struct printk_ringbuffer *rb, unsigned long lpos)
 	 * sees the new tail lpos, any descriptor states that transitioned to
 	 * the reusable state must already be visible.
 	 */
-	while ((lpos - tail_lpos) - 1 < DATA_SIZE(data_ring)) {
+	while (lpos1_before_lpos2(data_ring, tail_lpos, lpos)) {
 		/*
 		 * Make all descriptors reusable that are associated with
 		 * data blocks before @lpos.
@@ -1149,7 +1160,7 @@ static char *data_realloc(struct printk_ringbuffer *rb, unsigned int size,
 	next_lpos = get_next_lpos(data_ring, blk_lpos->begin, size);
 
 	/* If the data block does not increase, there is nothing to do. */
-	if (head_lpos - next_lpos < DATA_SIZE(data_ring)) {
+	if (!lpos1_before_lpos2(data_ring, head_lpos, next_lpos)) {
 		if (wrapped)
 			blk = to_block(data_ring, 0);
 		else
@@ -1262,7 +1273,7 @@ static const char *get_data(struct prb_data_ring *data_ring,
 
 	/* Regular data block: @begin less than @next and in same wrap. */
 	if (!is_blk_wrapped(data_ring, blk_lpos->begin, blk_lpos->next) &&
-	    blk_lpos->begin < blk_lpos->next) {
+	    lpos1_before_lpos2(data_ring, blk_lpos->begin, blk_lpos->next)) {
 		db = to_block(data_ring, blk_lpos->begin);
 		*data_size = blk_lpos->next - blk_lpos->begin;
 
This change also fixes the issue. Thoughts?

John