From: Matthias Kaehlcke <mka@chromium.org>
To: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
Sasha Levin <sashal@kernel.org>
Cc: stable@vger.kernel.org, clang-built-linux@googlegroups.com,
Nick Desaulniers <ndesaulniers@google.com>,
Rasmus Villemoes <linux@rasmusvillemoes.dk>,
Ido Schimmel <idosch@mellanox.com>,
Will Deacon <will.deacon@arm.com>,
Vadim Pasternak <vadimp@mellanox.com>,
Andrey Ryabinin <aryabinin@virtuozzo.com>,
Jacek Anaszewski <jacek.anaszewski@gmail.com>,
Pavel Machek <pavel@ucw.cz>,
Andrew Morton <akpm@linux-foundation.org>,
Linus Torvalds <torvalds@linux-foundation.org>,
Matthias Kaehlcke <mka@chromium.org>
Subject: [PATCH 5.1] include/linux/bitops.h: sanitize rotate primitives
Date: Tue, 4 Jun 2019 08:25:31 -0700 [thread overview]
Message-ID: <20190604152530.106715-1-mka@chromium.org> (raw)
From: Rasmus Villemoes <linux@rasmusvillemoes.dk>
commit ef4d6f6b275c498f8e5626c99dbeefdc5027f843 upstream
The ror32 implementation (word >> shift) | (word << (32 - shift) has
undefined behaviour if shift is outside the [1, 31] range. Similarly
for the 64 bit variants. Most callers pass a compile-time constant
(naturally in that range), but there's an UBSAN report that these may
actually be called with a shift count of 0.
Instead of special-casing that, we can make them DTRT for all values of
shift while also avoiding UB. For some reason, this was already partly
done for rol32 (which was well-defined for [0, 31]). gcc 8 recognizes
these patterns as rotates, so for example
__u32 rol32(__u32 word, unsigned int shift)
{
return (word << (shift & 31)) | (word >> ((-shift) & 31));
}
compiles to
0000000000000020 <rol32>:
20: 89 f8 mov %edi,%eax
22: 89 f1 mov %esi,%ecx
24: d3 c0 rol %cl,%eax
26: c3 retq
Older compilers unfortunately do not do as well, but this only affects
the small minority of users that don't pass constants.
Due to integer promotions, ro[lr]8 were already well-defined for shifts
in [0, 8], and ro[lr]16 were mostly well-defined for shifts in [0, 16]
(only mostly - u16 gets promoted to _signed_ int, so if bit 15 is set,
word << 16 is undefined). For consistency, update those as well.
Link: http://lkml.kernel.org/r/20190410211906.2190-1-linux@rasmusvillemoes.dk
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Reported-by: Ido Schimmel <idosch@mellanox.com>
Tested-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Will Deacon <will.deacon@arm.com>
Cc: Vadim Pasternak <vadimp@mellanox.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Jacek Anaszewski <jacek.anaszewski@gmail.com>
Cc: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Matthias Kaehlcke <mka@chromium.org>
---
Hi Greg and Sasha,
Please pick this patch for 5.1. It fixes (at least) crashes due
to undefined instructions in BPF code on arm32 when building with
clang:
[ 4.271668] Internal error: Oops - undefined instruction: 0 [#1] PREEMPT SMP ARM
[ 4.271675] Modules linked in:
[ 4.284143] joydev
[ 4.284155] CPU: 2 PID: 279 Comm: udevd Not tainted 4.19.44 #95
[ 4.284157] Hardware name: Rockchip (Device Tree)
[ 4.284167] PC is at 0xbf017de8
[ 4.284176] LR is at sk_filter_trim_cap+0xfc/0x244
[ 4.284181] pc : [<bf017de8>] lr : [<c0890c34>] psr: 60070013
[ 4.323863] sp : ed36fcd8 ip : 00000000 fp : ed36fd28
[ 4.329698] r10: c100b608 r9 : 00000000 r8 : ed2d6600
[ 4.335531] r7 : 00000000 r6 : 00000071 r5 : 00000000 r4 : ed59c180
[ 4.342812] r3 : 00000000 r2 : ed59c180 r1 : 00000000 r0 : 2a1884a1
[ 4.350105] Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment none
[ 4.358075] Control: 30c5387d Table: 2d22b880 DAC: 44541fd3
[ 4.364495] Process udevd (pid: 279, stack limit = 0xb5b8ffdf)
[ 4.371009] Stack: (0xed36fcd8 to 0xed370000)
[ 4.375864] fcc0: bf017a9c 60070013
[ 4.385001] fce0: 00000000 00000000 ed36fcd8 00000000 00000089 00000000 ed2d6600 00000000
[ 4.394137] fd00: 00000000 00000000 ed59c180 f0c12000 00000000 ed289000 ed59c1a0 00000001
[ 4.403274] fd20: 00000071 00000000 ed59c180 f0c12000 00000000 ed289000 ed59c1a0 00000001
...
[ 4.613384] [<c0890c34>] (sk_filter_trim_cap) from [<00000001>] (0x1)
[ 4.620580] Code: aa000002 e0200000 e3a01000 ea000045 (f7e6c179)
Thanks
Matthias
---
include/linux/bitops.h | 16 ++++++++--------
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/include/linux/bitops.h b/include/linux/bitops.h
index 7ddb1349394d..7ac2e46112b7 100644
--- a/include/linux/bitops.h
+++ b/include/linux/bitops.h
@@ -60,7 +60,7 @@ static __always_inline unsigned long hweight_long(unsigned long w)
*/
static inline __u64 rol64(__u64 word, unsigned int shift)
{
- return (word << shift) | (word >> (64 - shift));
+ return (word << (shift & 63)) | (word >> ((-shift) & 63));
}
/**
@@ -70,7 +70,7 @@ static inline __u64 rol64(__u64 word, unsigned int shift)
*/
static inline __u64 ror64(__u64 word, unsigned int shift)
{
- return (word >> shift) | (word << (64 - shift));
+ return (word >> (shift & 63)) | (word << ((-shift) & 63));
}
/**
@@ -80,7 +80,7 @@ static inline __u64 ror64(__u64 word, unsigned int shift)
*/
static inline __u32 rol32(__u32 word, unsigned int shift)
{
- return (word << shift) | (word >> ((-shift) & 31));
+ return (word << (shift & 31)) | (word >> ((-shift) & 31));
}
/**
@@ -90,7 +90,7 @@ static inline __u32 rol32(__u32 word, unsigned int shift)
*/
static inline __u32 ror32(__u32 word, unsigned int shift)
{
- return (word >> shift) | (word << (32 - shift));
+ return (word >> (shift & 31)) | (word << ((-shift) & 31));
}
/**
@@ -100,7 +100,7 @@ static inline __u32 ror32(__u32 word, unsigned int shift)
*/
static inline __u16 rol16(__u16 word, unsigned int shift)
{
- return (word << shift) | (word >> (16 - shift));
+ return (word << (shift & 15)) | (word >> ((-shift) & 15));
}
/**
@@ -110,7 +110,7 @@ static inline __u16 rol16(__u16 word, unsigned int shift)
*/
static inline __u16 ror16(__u16 word, unsigned int shift)
{
- return (word >> shift) | (word << (16 - shift));
+ return (word >> (shift & 15)) | (word << ((-shift) & 15));
}
/**
@@ -120,7 +120,7 @@ static inline __u16 ror16(__u16 word, unsigned int shift)
*/
static inline __u8 rol8(__u8 word, unsigned int shift)
{
- return (word << shift) | (word >> (8 - shift));
+ return (word << (shift & 7)) | (word >> ((-shift) & 7));
}
/**
@@ -130,7 +130,7 @@ static inline __u8 rol8(__u8 word, unsigned int shift)
*/
static inline __u8 ror8(__u8 word, unsigned int shift)
{
- return (word >> shift) | (word << (8 - shift));
+ return (word >> (shift & 7)) | (word << ((-shift) & 7));
}
/**
--
2.22.0.rc1.311.g5d7573a151-goog
next reply other threads:[~2019-06-04 15:25 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-06-04 15:25 Matthias Kaehlcke [this message]
2019-06-06 1:25 ` [PATCH 5.1] include/linux/bitops.h: sanitize rotate primitives Sasha Levin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190604152530.106715-1-mka@chromium.org \
--to=mka@chromium.org \
--cc=akpm@linux-foundation.org \
--cc=aryabinin@virtuozzo.com \
--cc=clang-built-linux@googlegroups.com \
--cc=gregkh@linuxfoundation.org \
--cc=idosch@mellanox.com \
--cc=jacek.anaszewski@gmail.com \
--cc=linux@rasmusvillemoes.dk \
--cc=ndesaulniers@google.com \
--cc=pavel@ucw.cz \
--cc=sashal@kernel.org \
--cc=stable@vger.kernel.org \
--cc=torvalds@linux-foundation.org \
--cc=vadimp@mellanox.com \
--cc=will.deacon@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox