lkml.org 
[lkml]   [2019]   [Sep]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] arm64: fix unreachable code issue with cmpxchg
On Mon, Sep 09, 2019 at 10:21:35PM +0200, Arnd Bergmann wrote:
> On arm64 build with clang, sometimes the __cmpxchg_mb is not inlined
> when CONFIG_OPTIMIZE_INLINING is set.
> Clang then fails a compile-time assertion, because it cannot tell at
> compile time what the size of the argument is:
>
> mm/memcontrol.o: In function `__cmpxchg_mb':
> memcontrol.c:(.text+0x1a4c): undefined reference to `__compiletime_assert_175'
> memcontrol.c:(.text+0x1a4c): relocation truncated to fit: R_AARCH64_CALL26 against undefined symbol `__compiletime_assert_175'
>
> Mark all of the cmpxchg() style functions as __always_inline to
> ensure that the compiler can see the result.
>
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---

I was able to reproduce this with the following:

$ git describe HEAD
next-20190904

$ clang --version
Android (5821526 based on r365631) clang version 9.0.6 (https://android.googlesource.com/toolchain/llvm-project 85305eaf1e90ff529d304abac8a979e1d967f0a2) (based on LLVM 9.0.6svn)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /home/andrewm/android-clang/clang-r365631/bin

$ make O=~/linux-build/ ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- CC=clang -j56 allyesconfig Image

(I was unable to reproduce with defconfig+OPTIMIZE_INLINING).

However...

> arch/arm64/include/asm/cmpxchg.h | 15 ++++++++-------
> 1 file changed, 8 insertions(+), 7 deletions(-)
>
> diff --git a/arch/arm64/include/asm/cmpxchg.h b/arch/arm64/include/asm/cmpxchg.h
> index a1398f2f9994..fd64dc8a235f 100644
> --- a/arch/arm64/include/asm/cmpxchg.h
> +++ b/arch/arm64/include/asm/cmpxchg.h
> @@ -19,7 +19,7 @@
> * acquire+release for the latter.
> */
> #define __XCHG_CASE(w, sfx, name, sz, mb, nop_lse, acq, acq_lse, rel, cl) \
> -static inline u##sz __xchg_case_##name##sz(u##sz x, volatile void *ptr) \
> +static __always_inline u##sz __xchg_case_##name##sz(u##sz x, volatile void *ptr)\

This hunk isn't needed, there is no BUILD_BUG here.


> { \
> u##sz ret; \
> unsigned long tmp; \
> @@ -62,7 +62,7 @@ __XCHG_CASE( , , mb_, 64, dmb ish, nop, , a, l, "memory")
> #undef __XCHG_CASE
>
> #define __XCHG_GEN(sfx) \
> -static inline unsigned long __xchg##sfx(unsigned long x, \
> +static __always_inline unsigned long __xchg##sfx(unsigned long x, \
> volatile void *ptr, \
> int size) \
> { \
> @@ -103,8 +103,9 @@ __XCHG_GEN(_mb)
> #define arch_xchg_release(...) __xchg_wrapper(_rel, __VA_ARGS__)
> #define arch_xchg(...) __xchg_wrapper( _mb, __VA_ARGS__)
>
> -#define __CMPXCHG_CASE(name, sz) \
> -static inline u##sz __cmpxchg_case_##name##sz(volatile void *ptr, \
> +#define __CMPXCHG_CASE(name, sz) \
> +static __always_inline u##sz \
> +__cmpxchg_case_##name##sz(volatile void *ptr, \

This hunk isn't needed, there is no BUILD_BUG here.

> u##sz old, \
> u##sz new) \
> { \
> @@ -148,7 +149,7 @@ __CMPXCHG_DBL(_mb)
> #undef __CMPXCHG_DBL
>
> #define __CMPXCHG_GEN(sfx) \
> -static inline unsigned long __cmpxchg##sfx(volatile void *ptr, \
> +static __always_inline unsigned long __cmpxchg##sfx(volatile void *ptr, \
> unsigned long old, \
> unsigned long new, \
> int size) \
> @@ -230,7 +231,7 @@ __CMPXCHG_GEN(_mb)
> })
>
> #define __CMPWAIT_CASE(w, sfx, sz) \
> -static inline void __cmpwait_case_##sz(volatile void *ptr, \
> +static __always_inline void __cmpwait_case_##sz(volatile void *ptr, \
> unsigned long val) \

This hunk isn't needed, there is no BUILD_BUG here.

> { \
> unsigned long tmp; \
> @@ -255,7 +256,7 @@ __CMPWAIT_CASE( , , 64);
> #undef __CMPWAIT_CASE
>
> #define __CMPWAIT_GEN(sfx) \
> -static inline void __cmpwait##sfx(volatile void *ptr, \
> +static __always_inline void __cmpwait##sfx(volatile void *ptr, \
> unsigned long val, \
> int size) \
> { \

Alternatively is it possible to replace the BUILD_BUG's with something else?

I think because we use BUILD_BUG at the end of a switch statement, we make
the assumption that size is known at compile time, for this reason we should
ensure the function containing the BUILD_BUG is __always_inline.

Looking across the kernel where BUILD_BUG is used as a default in a switch
statment ($ git grep -B 3 BUILD_BUG\( | grep default), most instances are
within macros, but many are found in an __always_inline function:

arch/x86/kvm/cpuid.h
mm/kasan/generic.c

Though some are not:

include/linux/signal.h
arch/arm64/include/asm/arm_dsu/pmu.h

I wonder if there may be a latent mole ready to whack with pmu.h?

Anyway with just the three remaining hunks:

Reviewed-by: Andrew Murray <andrew.murray@arm.com>
Tested-by: Andrew Murray <andrew.murray@arm.com>

> --
> 2.20.0
>

\
 
 \ /
  Last update: 2019-09-10 11:23    [W:0.135 / U:22.144 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site