linux-kernel - [PATCH] printk: Do not lose last line in kmsg dump

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-next>] [day] [month] [year] [list]

Message-Id: <20190709081042.31551-1-vincent.whitchurch@axis.com>
Date:   Tue,  9 Jul 2019 10:10:42 +0200
From:   Vincent Whitchurch <vincent.whitchurch@...s.com>
To:     pmladek@...e.com, sergey.senozhatsky@...il.com, rostedt@...dmis.org
Cc:     linux-kernel@...r.kernel.org, Vincent Whitchurch <rabinv@...s.com>
Subject: [PATCH] printk: Do not lose last line in kmsg dump

kmsg_dump_get_buffer() is supposed to select the youngest log messages
which fit into the provided buffer.  When that function determines the
correct start index, by looping and calling msg_print_text() with a NULL
buffer, msg_print_text() calculates a length which would allow the
youngest log messages to completely fill the provided buffer.

However, when doing the actual printing, an off-by-one error in
msg_print_text() leads to that function allowing the provided buffer to
only be filled to (size - 1).

So if the lengths of the selected youngest log messages happen to
precisely fill up the provided buffer, the last log message is not
included.

For example, with the following two final prints:

[    6.427502] AAAAAAAAAAAAA
[    6.427769] BBBBBBBB12345

A dump of a 64-byte buffer filled by kmsg_dump_get_buffer(), before this
patch:

 00000000: 3c 30 3e 5b 20 20 20 20 36 2e 35 32 32 31 39 37  <0>[    6.522197
 00000010: 5d 20 41 41 41 41 41 41 41 41 41 41 41 41 41 0a  ] AAAAAAAAAAAAA.
 00000020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
 00000030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................

After this patch:

 00000000: 3c 30 3e 5b 20 20 20 20 36 2e 34 32 37 35 30 32  <0>[    6.427502
 00000010: 5d 20 41 41 41 41 41 41 41 41 41 41 41 41 41 0a  ] AAAAAAAAAAAAA.
 00000020: 3c 30 3e 5b 20 20 20 20 36 2e 34 32 37 37 36 39  <0>[    6.427769
 00000030: 5d 20 42 42 42 42 42 42 42 42 31 32 33 34 35 0a  ] BBBBBBBB12345.

Note that this bug only affects the kmsg dump code.  msg_print_text() is
also used from the syslog code and console_unlock() but this bug does
trigger there since the buffers used there are never filled up
completely (since they are only used to print individual lines, and
their size is always LOG_LINE_MAX + PREFIX_MAX, and PREFIX_MAX has a
value which is larger than the largest possible prefix).

Signed-off-by: Vincent Whitchurch <vincent.whitchurch@...s.com>
---
I posted this patch two years ago and received no replies.  This problem is
still present in mainline.  https://lore.kernel.org/patchwork/patch/781106/

 kernel/printk/printk.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index 1888f6a3b694..7679d779d5cc 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -1318,7 +1318,7 @@ static size_t msg_print_text(const struct printk_log *msg, bool syslog,
 		}

 		if (buf) {
-			if (prefix_len + text_len + 1 >= size - len)
+			if (prefix_len + text_len + 1 > size - len)
 				break;

 			memcpy(buf + len, prefix, prefix_len);
-- 
2.20.0