Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Content-Type: text/plain; charset="UTF-8"
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
MIME-Version: 1.0
From:   Ben Hutchings <ben@decadent.org.uk>
To:     linux-kernel@vger.kernel.org, stable@vger.kernel.org
CC:     akpm@linux-foundation.org, "Randy Dunlap" <rdunlap@infradead.org>,
        "Michael Kerrisk" <mtk.manpages@gmail.com>,
        "Jens Axboe" <axboe@kernel.dk>,
        "Mikulas Patocka" <mpatocka@redhat.com>,
        "Joe Lawrence" <joe.lawrence@redhat.com>,
        "Al Viro" <viro@zeniv.linux.org.uk>,
        "Josh Poimboeuf" <jpoimboe@redhat.com>,
        "Linus Torvalds" <torvalds@linux-foundation.org>
Date:   Thu, 07 Jun 2018 15:05:21 +0100
Message-ID: <lsq.1528380321.522642950@decadent.org.uk>
Subject: [PATCH 3.16 209/410] sysctl: check for UINT_MAX before unsigned
 int min/max
In-Reply-To: <lsq.1528380320.647747352@decadent.org.uk>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

3.16.57-rc1 review patch.  If anyone has any objections, please let me know.

------------------

From: Joe Lawrence <joe.lawrence@redhat.com>

commit fb910c42ccebf853c29296185c45c11164a56098 upstream.

Mikulas noticed in the existing do_proc_douintvec_minmax_conv() and
do_proc_dopipe_max_size_conv() introduced in this patchset, that they
inconsistently handle overflow and min/max range inputs:

For example:

  0 ... param->min - 1 ---> ERANGE
  param->min ... param->max ---> the value is accepted
  param->max + 1 ... 0x100000000L + param->min - 1 ---> ERANGE
  0x100000000L + param->min ... 0x100000000L + param->max ---> EINVAL
  0x100000000L + param->max + 1, 0x200000000L + param->min - 1 ---> ERANGE
  0x200000000L + param->min ... 0x200000000L + param->max ---> EINVAL
  0x200000000L + param->max + 1, 0x300000000L + param->min - 1 ---> ERANGE

In do_proc_do*() routines which store values into unsigned int variables
(4 bytes wide for 64-bit builds), first validate that the input unsigned
long value (8 bytes wide for 64-bit builds) will fit inside the smaller
unsigned int variable.  Then check that the unsigned int value falls
inside the specified parameter min, max range.  Otherwise the unsigned
long -> unsigned int conversion drops leading bits from the input value,
leading to the inconsistent pattern Mikulas documented above.

Link: http://lkml.kernel.org/r/1507658689-11669-5-git-send-email-joe.lawrence@redhat.com
Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com>
Reported-by: Mikulas Patocka <mpatocka@redhat.com>
Reviewed-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[bwh: Backported to 3.16:
 - Drop changes in do_proc_douintvec_minmax_conv()
 - Adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
---
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -2233,17 +2233,18 @@ static int do_proc_dopipe_max_size_conv(
 	struct do_proc_dopipe_max_size_conv_param *param = data;
 
 	if (write) {
-		unsigned int val = round_pipe_size(*lvalp);
+		unsigned int val;
 
+		if (*lvalp > UINT_MAX)
+			return -EINVAL;
+
+		val = round_pipe_size(*lvalp);
 		if (*negp || val == 0)
 			return -EINVAL;
 
 		if (param->min && *param->min > val)
 			return -ERANGE;
 
-		if (*lvalp > UINT_MAX)
-			return -EINVAL;
-
 		*valp = val;
 	} else {
 		unsigned int val = *valp;