From: Arnd Bergmann <arnd@arndb.de>
To: kasan-dev@googlegroups.com
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>,
Alexander Potapenko <glider@google.com>,
Dmitry Vyukov <dvyukov@google.com>,
netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
linux-media@vger.kernel.org, linux-wireless@vger.kernel.org,
kernel-build-reports@lists.linaro.org,
"David S . Miller" <davem@davemloft.net>,
Arnd Bergmann <arnd@arndb.de>,
Christian Borntraeger <borntraeger@de.ibm.com>,
Paul McKenney <paulmck@linux.vnet.ibm.com>
Subject: [PATCH 02/26] rewrite READ_ONCE/WRITE_ONCE
Date: Thu, 2 Mar 2017 17:38:10 +0100 [thread overview]
Message-ID: <20170302163834.2273519-3-arnd@arndb.de> (raw)
In-Reply-To: <20170302163834.2273519-1-arnd@arndb.de>
When CONFIG_KASAN is enabled, the READ_ONCE/WRITE_ONCE macros cause
rather large kernel stacks, e.g.:
mm/vmscan.c: In function 'shrink_page_list':
mm/vmscan.c:1333:1: error: the frame size of 3456 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
block/cfq-iosched.c: In function 'cfqg_stats_add_aux':
block/cfq-iosched.c:750:1: error: the frame size of 4048 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
fs/btrfs/disk-io.c: In function 'open_ctree':
fs/btrfs/disk-io.c:3314:1: error: the frame size of 3136 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
fs/btrfs/relocation.c: In function 'build_backref_tree':
fs/btrfs/relocation.c:1193:1: error: the frame size of 4336 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
fs/fscache/stats.c: In function 'fscache_stats_show':
fs/fscache/stats.c:287:1: error: the frame size of 6512 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
fs/jbd2/commit.c: In function 'jbd2_journal_commit_transaction':
fs/jbd2/commit.c:1139:1: error: the frame size of 3760 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
This attempts a rewrite of the two macros, using a simpler implementation
for the most common case of having a naturally aligned 1, 2, 4, or (on
64-bit architectures) 8 byte object that can be accessed with a single
instruction. For these, we go back to a volatile pointer dereference
that we had with the ACCESS_ONCE macro.
READ_ONCE/WRITE_ONCE also try to handle unaligned objects and objects
of other sizes by forcing either a word-size access (which may trap
on some architectures) or doing a non-atomic memcpy. I could not figure
out what these are actually used for, but they appear to be done
intentionally, so I'm leaving that code untouched.
I had to fix up a couple of files that either use WRITE_ONCE() as an
implicit typecast, or ignore the result of READ_ONCE(). In all cases,
the modified code seems no worse to me than the original.
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Paul McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
arch/x86/include/asm/switch_to.h | 2 +-
fs/overlayfs/util.c | 6 ++---
include/linux/compiler.h | 47 ++++++++++++++++++++++++++++++++--------
3 files changed, 42 insertions(+), 13 deletions(-)
diff --git a/arch/x86/include/asm/switch_to.h b/arch/x86/include/asm/switch_to.h
index fcc5cd387fd1..0c243dd569fe 100644
--- a/arch/x86/include/asm/switch_to.h
+++ b/arch/x86/include/asm/switch_to.h
@@ -30,7 +30,7 @@ static inline void prepare_switch_to(struct task_struct *prev,
*
* To minimize cache pollution, just follow the stack pointer.
*/
- READ_ONCE(*(unsigned char *)next->thread.sp);
+ (void)READ_ONCE(*(unsigned char *)next->thread.sp);
#endif
}
diff --git a/fs/overlayfs/util.c b/fs/overlayfs/util.c
index 952286f4826c..1c10632a48bb 100644
--- a/fs/overlayfs/util.c
+++ b/fs/overlayfs/util.c
@@ -222,8 +222,8 @@ void ovl_dentry_update(struct dentry *dentry, struct dentry *upperdentry)
void ovl_inode_init(struct inode *inode, struct inode *realinode, bool is_upper)
{
- WRITE_ONCE(inode->i_private, (unsigned long) realinode |
- (is_upper ? OVL_ISUPPER_MASK : 0));
+ WRITE_ONCE(inode->i_private, (void *)((unsigned long) realinode |
+ (is_upper ? OVL_ISUPPER_MASK : 0)));
}
void ovl_inode_update(struct inode *inode, struct inode *upperinode)
@@ -231,7 +231,7 @@ void ovl_inode_update(struct inode *inode, struct inode *upperinode)
WARN_ON(!upperinode);
WARN_ON(!inode_unhashed(inode));
WRITE_ONCE(inode->i_private,
- (unsigned long) upperinode | OVL_ISUPPER_MASK);
+ (void *)((unsigned long) upperinode | OVL_ISUPPER_MASK));
if (!S_ISDIR(upperinode->i_mode))
__insert_inode_hash(inode, (unsigned long) upperinode);
}
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index 56b90897a459..b619f5853af8 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -288,6 +288,10 @@ static __always_inline void __write_once_size(volatile void *p, void *res, int s
}
}
+#define __ALIGNED_WORD(x) \
+ ((sizeof(x) == 1 || sizeof(x) == 2 || sizeof(x) == 4 || \
+ sizeof(x) == sizeof(long)) && (sizeof(x) == __alignof__(x))) \
+
/*
* Prevent the compiler from merging or refetching reads or writes. The
* compiler is also forbidden from reordering successive instances of
@@ -309,8 +313,13 @@ static __always_inline void __write_once_size(volatile void *p, void *res, int s
* mutilate accesses that either do not require ordering or that interact
* with an explicit memory barrier or atomic instruction that provides the
* required ordering.
+ *
+ * Unaligned data is particularly tricky here: if the type that gets
+ * passed in is not naturally aligned, we cast to a type of higher
+ * alignment, which is not well-defined in C. This is fine as long
+ * as the actual data is aligned, but otherwise might require a trap
+ * to satisfy the load.
*/
-
#define __READ_ONCE(x, check) \
({ \
union { typeof(x) __val; char __c[1]; } __u; \
@@ -320,7 +329,32 @@ static __always_inline void __write_once_size(volatile void *p, void *res, int s
__read_once_size_nocheck(&(x), __u.__c, sizeof(x)); \
__u.__val; \
})
-#define READ_ONCE(x) __READ_ONCE(x, 1)
+
+#define __WRITE_ONCE(x, val) \
+({ \
+ union { typeof(x) __val; char __c[1]; } __u = \
+ { .__val = (__force typeof(x)) (val) }; \
+ __write_once_size(&(x), __u.__c, sizeof(x)); \
+ __u.__val; \
+})
+
+
+/*
+ * the common case is simple: x is naturally aligned, not an array,
+ * and accessible with a single load, avoiding the need for local
+ * variables. With KASAN, this is important as any call to
+ *__write_once_size(),__read_once_size_nocheck() or __read_once_size()
+ * uses significant amounts of stack space for checking that we don't
+ * overflow the union.
+ */
+#define __READ_ONCE_SIMPLE(x) \
+ (typeof(x))(*(volatile typeof(&(x)))&(x))
+
+#define __WRITE_ONCE_SIMPLE(x, val) \
+ ({*(volatile typeof(&(x)))&(x) = (val); })
+
+#define READ_ONCE(x) __builtin_choose_expr(__ALIGNED_WORD(x), \
+ __READ_ONCE_SIMPLE(x), __READ_ONCE(x, 1))
/*
* Use READ_ONCE_NOCHECK() instead of READ_ONCE() if you need
@@ -328,13 +362,8 @@ static __always_inline void __write_once_size(volatile void *p, void *res, int s
*/
#define READ_ONCE_NOCHECK(x) __READ_ONCE(x, 0)
-#define WRITE_ONCE(x, val) \
-({ \
- union { typeof(x) __val; char __c[1]; } __u = \
- { .__val = (__force typeof(x)) (val) }; \
- __write_once_size(&(x), __u.__c, sizeof(x)); \
- __u.__val; \
-})
+#define WRITE_ONCE(x, val) do { __builtin_choose_expr(__ALIGNED_WORD(x), \
+ __WRITE_ONCE_SIMPLE(x, val), __WRITE_ONCE(x, val)); } while (0)
#endif /* __KERNEL__ */
--
2.9.0
next prev parent reply other threads:[~2017-03-02 16:48 UTC|newest]
Thread overview: 68+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-03-02 16:38 [PATCH 00/26] bring back stack frame warning with KASAN Arnd Bergmann
2017-03-02 16:38 ` [PATCH 01/26] compiler: introduce noinline_for_kasan annotation Arnd Bergmann
2017-03-03 13:50 ` Andrey Ryabinin
2017-03-03 13:55 ` Alexander Potapenko
2017-03-03 14:30 ` Arnd Bergmann
2017-03-03 14:33 ` Alexander Potapenko
2017-03-03 14:51 ` Arnd Bergmann
2017-03-03 16:34 ` David Laight
2017-03-02 16:38 ` Arnd Bergmann [this message]
2017-03-02 16:51 ` [PATCH 02/26] rewrite READ_ONCE/WRITE_ONCE Christian Borntraeger
2017-03-02 17:55 ` Arnd Bergmann
2017-03-02 19:00 ` Christian Borntraeger
2017-03-02 21:45 ` Arnd Bergmann
2017-03-03 8:26 ` Christian Borntraeger
2017-03-03 9:54 ` Arnd Bergmann
2017-03-03 14:49 ` Peter Zijlstra
2017-03-03 14:57 ` Peter Zijlstra
2017-03-02 16:38 ` [PATCH 03/26] typecheck.h: avoid local variables in typecheck() macro Arnd Bergmann
2017-03-02 16:38 ` [PATCH 04/26] tty: kbd: reduce stack size with KASAN Arnd Bergmann
2017-03-02 16:38 ` [PATCH 05/26] netlink: mark nla_put_{u8,u16,u32} noinline_for_kasan Arnd Bergmann
2017-03-02 16:38 ` [PATCH 06/26] rocker: mark rocker_tlv_put_* functions as noinline_for_kasan Arnd Bergmann
2017-03-02 16:38 ` [PATCH 07/26] brcmsmac: reduce stack size with KASAN Arnd Bergmann
2017-03-06 9:16 ` Arend Van Spriel
2017-03-06 10:38 ` Arnd Bergmann
2017-03-06 11:02 ` Arend Van Spriel
2017-03-06 11:16 ` Arnd Bergmann
2017-03-06 11:18 ` Arnd Bergmann
2017-03-02 16:38 ` [PATCH 08/26] brcmsmac: make some local variables 'static const' to reduce stack size Arnd Bergmann
2017-03-06 9:30 ` Arend Van Spriel
2017-03-06 16:19 ` Kalle Valo
2017-03-06 21:34 ` Arnd Bergmann
2017-03-07 9:44 ` Kalle Valo
2017-03-07 9:55 ` Arend Van Spriel
2017-03-02 16:38 ` [PATCH 09/26] brcmsmac: split up wlc_phy_workarounds_nphy Arnd Bergmann
2017-03-06 9:31 ` Arend Van Spriel
2017-03-02 16:38 ` [PATCH 10/26] brcmsmac: reindent split functions Arnd Bergmann
2017-03-06 9:33 ` Arend Van Spriel
2017-03-06 16:24 ` Kalle Valo
2017-03-02 16:38 ` [PATCH 11/26] rtlwifi: reduce stack usage for KASAN Arnd Bergmann
2017-03-02 16:38 ` [PATCH 12/26] wl3501_cs: reduce stack size " Arnd Bergmann
2017-03-02 16:38 ` [PATCH 13/26] rtl8180: " Arnd Bergmann
2017-03-02 16:38 ` [PATCH 14/26] [media] dvb-frontends: reduce stack size in i2c access Arnd Bergmann
2017-03-02 16:38 ` [PATCH 15/26] [media] tuners: i2c: reduce stack usage for tuner_i2c_xfer_* Arnd Bergmann
2017-03-02 16:38 ` [PATCH 16/26] [media] i2c: adv7604: mark register access as noinline_for_kasan Arnd Bergmann
2017-03-02 16:38 ` [PATCH 17/26] [media] i2c: ks0127: reduce stack frame size for KASAN Arnd Bergmann
2017-03-02 16:38 ` [PATCH 18/26] [media] i2c: cx25840: avoid stack overflow with KASAN Arnd Bergmann
2017-03-02 16:38 ` [PATCH 19/26] [media] r820t: mark register functions as noinline_for_kasan Arnd Bergmann
2017-03-02 16:38 ` [PATCH 20/26] [media] em28xx: split up em28xx_dvb_init to reduce stack size Arnd Bergmann
2017-04-17 17:07 ` Mauro Carvalho Chehab
2017-03-02 16:38 ` [PATCH 21/26] drm/bridge: ps8622: reduce stack size for KASAN Arnd Bergmann
2017-03-02 16:38 ` [PATCH 22/26] drm/i915/gvt: don't overflow the kernel stack with KASAN Arnd Bergmann
2017-03-02 16:38 ` [PATCH 23/26] mtd: cfi: reduce stack size " Arnd Bergmann
2017-03-02 16:38 ` [PATCH 24/26] ocfs2: " Arnd Bergmann
2017-03-02 17:46 ` Joe Perches
2017-03-02 22:22 ` Arnd Bergmann
2017-03-02 22:40 ` Joe Perches
2017-03-02 22:59 ` Arnd Bergmann
2017-03-02 23:58 ` Joe Perches
2017-03-02 16:38 ` [PATCH 25/26] isdn: eicon: mark divascapi incompatible with kasan Arnd Bergmann
2017-03-03 14:20 ` Andrey Ryabinin
2017-03-03 14:54 ` Arnd Bergmann
2017-03-03 15:22 ` Andrey Ryabinin
2017-03-03 15:37 ` Arnd Bergmann
2017-03-02 16:38 ` [PATCH 26/26] kasan: rework Kconfig settings Arnd Bergmann
2017-03-03 14:51 ` Andrey Ryabinin
2017-03-03 15:03 ` Arnd Bergmann
2017-03-03 12:25 ` [PATCH 00/26] bring back stack frame warning with KASAN Alexander Potapenko
2017-03-03 12:54 ` Arnd Bergmann
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=20170302163834.2273519-3-arnd@arndb.de \
--to=arnd@arndb.de \
--cc=aryabinin@virtuozzo.com \
--cc=borntraeger@de.ibm.com \
--cc=davem@davemloft.net \
--cc=dvyukov@google.com \
--cc=glider@google.com \
--cc=kasan-dev@googlegroups.com \
--cc=kernel-build-reports@lists.linaro.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-media@vger.kernel.org \
--cc=linux-wireless@vger.kernel.org \
--cc=netdev@vger.kernel.org \
--cc=paulmck@linux.vnet.ibm.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