linux-kernel - [PATCH 2/4] printk: honor LOG_PREFIX in devkmsg

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:	Wed, 16 Jul 2014 12:26:58 -0500
From:	Alex Elder <elder@...aro.org>
To:	akpm@...ux-foundation.org
Cc:	pmladek@...e.cz, bp@...e.de, john.stultz@...aro.org, jack@...e.cz,
	linux-kernel@...r.kernel.org
Subject: [PATCH 2/4] printk: honor LOG_PREFIX in devkmsg_read()

In devkmsg_read(), a variable "cont" holds a character that's used
to indicate whether a given log line is a "continuation", that is,
whether a log record should be merged with the one before or after
it.  If a record should be merged with its successor (but not its
predecessor) that character is 'c'.  And the line following such a
'c' log record is normally marked with a '+' to show it is continues
its predecessor.  Any other cases are marked '-', indicating the
log record stands on its own.

There is an exception.  If a log record is marked LOG_PREFIX, it
indicates that this record represents a new log entry, implicitly
terminating the predecessor--even if the predecessor would otherwise
have been continued.  So a record marked LOG_PREFIX (that is not
also marked LOG_CONT) should have '-' for its "cont" variable.

The logic that determines this "continuation" character has a bug
that gets that exceptional case wrong.

The specific case that produces the wrong result is when all of
these conditions are non-zero:
    user->prev & LOG_CONT
    msg->flags & LOG_PREFIX
    msg->flags & LOG_CONT
The bug is that despite the message's LOG_PREFIX flag, the
"cont" character is getting set to '+' rather than 'c'.

The problem is that the message's LOG_PREFIX flag is getting
ignored if its LOG_CONT flag is also set.  Rearrange the logic
here to produce the correct result.

The following table concisely defines the problem:

     prev | msg  | msg  ||"cont"
     CONT |PREFIX| CONT || char
    ------+------+------++------
     clear| clear| clear||  '-'
     clear| clear|  set ||  'c'
     clear|  set | clear||  '-'
     clear|  set |  set ||  'c'
      set | clear| clear||  '+'
      set |  set |  set ||  '+'
      set |  set | clear||  '-'
      set |  set |  set ||  '+'      <-- should be 'c'

Signed-off-by: Alex Elder <elder@...aro.org>
---
 kernel/printk/printk.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index 301ade3..9e9cf93 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -572,7 +572,7 @@ static ssize_t devkmsg_read(struct file *file, char __user *buf,
 	struct printk_log *msg;
 	u64 ts_usec;
 	size_t i;
-	char cont = '-';
+	char cont;
 	size_t len;
 	ssize_t ret;

@@ -619,11 +619,12 @@ static ssize_t devkmsg_read(struct file *file, char __user *buf,
 	 * better readable output. 'c' in the record flags mark the first
 	 * fragment of a line, '+' the following.
 	 */
-	if (msg->flags & LOG_CONT && !(user->prev & LOG_CONT))
-		cont = 'c';
-	else if ((msg->flags & LOG_CONT) ||
-		 ((user->prev & LOG_CONT) && !(msg->flags & LOG_PREFIX)))
+	if (user->prev & LOG_CONT && !(msg->flags & LOG_PREFIX))
 		cont = '+';
+	else if (msg->flags & LOG_CONT)
+		cont = 'c';
+	else
+		cont = '-';

 	len = sprintf(user->buf, "%u,%llu,%llu,%c;",
 		      (msg->facility << 3) | msg->level,
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/