From 080b6f40763565f65ebb9540219c71ce885cf568 Mon Sep 17 00:00:00 2001 From: Ard Biesheuvel Date: Wed, 28 Oct 2020 18:15:05 +0100 Subject: bpf: Don't rely on GCC __attribute__((optimize)) to disable GCSE Commit 3193c0836 ("bpf: Disable GCC -fgcse optimization for ___bpf_prog_run()") introduced a __no_fgcse macro that expands to a function scope __attribute__((optimize("-fno-gcse"))), to disable a GCC specific optimization that was causing trouble on x86 builds, and was not expected to have any positive effect in the first place. However, as the GCC manual documents, __attribute__((optimize)) is not for production use, and results in all other optimization options to be forgotten for the function in question. This can cause all kinds of trouble, but in one particular reported case, it causes -fno-asynchronous-unwind-tables to be disregarded, resulting in .eh_frame info to be emitted for the function. This reverts commit 3193c0836, and instead, it disables the -fgcse optimization for the entire source file, but only when building for X86 using GCC with CONFIG_BPF_JIT_ALWAYS_ON disabled. Note that the original commit states that CONFIG_RETPOLINE=n triggers the issue, whereas CONFIG_RETPOLINE=y performs better without the optimization, so it is kept disabled in both cases. Fixes: 3193c0836f20 ("bpf: Disable GCC -fgcse optimization for ___bpf_prog_run()") Signed-off-by: Ard Biesheuvel Signed-off-by: Alexei Starovoitov Tested-by: Geert Uytterhoeven Reviewed-by: Nick Desaulniers Link: https://lore.kernel.org/lkml/CAMuHMdUg0WJHEcq6to0-eODpXPOywLot6UD2=GFHpzoj_hCoBQ@mail.gmail.com/ Link: https://lore.kernel.org/bpf/20201028171506.15682-2-ardb@kernel.org --- include/linux/compiler-gcc.h | 2 -- 1 file changed, 2 deletions(-) (limited to 'include/linux/compiler-gcc.h') diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h index d1e3c6896b71..5deb37024574 100644 --- a/include/linux/compiler-gcc.h +++ b/include/linux/compiler-gcc.h @@ -175,5 +175,3 @@ #else #define __diag_GCC_8(s) #endif - -#define __no_fgcse __attribute__((optimize("-fno-gcse"))) -- cgit v1.2.3 From 3347acc6fcd4ee71ad18a9ff9d9dac176b517329 Mon Sep 17 00:00:00 2001 From: Arvind Sankar Date: Fri, 13 Nov 2020 22:51:59 -0800 Subject: compiler.h: fix barrier_data() on clang Commit 815f0ddb346c ("include/linux/compiler*.h: make compiler-*.h mutually exclusive") neglected to copy barrier_data() from compiler-gcc.h into compiler-clang.h. The definition in compiler-gcc.h was really to work around clang's more aggressive optimization, so this broke barrier_data() on clang, and consequently memzero_explicit() as well. For example, this results in at least the memzero_explicit() call in lib/crypto/sha256.c:sha256_transform() being optimized away by clang. Fix this by moving the definition of barrier_data() into compiler.h. Also move the gcc/clang definition of barrier() into compiler.h, __memory_barrier() is icc-specific (and barrier() is already defined using it in compiler-intel.h) and doesn't belong in compiler.h. [rdunlap@infradead.org: fix ALPHA builds when SMP is not enabled] Link: https://lkml.kernel.org/r/20201101231835.4589-1-rdunlap@infradead.org Fixes: 815f0ddb346c ("include/linux/compiler*.h: make compiler-*.h mutually exclusive") Signed-off-by: Arvind Sankar Signed-off-by: Randy Dunlap Signed-off-by: Andrew Morton Tested-by: Nick Desaulniers Reviewed-by: Nick Desaulniers Reviewed-by: Kees Cook Cc: Link: https://lkml.kernel.org/r/20201014212631.207844-1-nivedita@alum.mit.edu Signed-off-by: Linus Torvalds --- include/asm-generic/barrier.h | 1 + include/linux/compiler-clang.h | 6 ------ include/linux/compiler-gcc.h | 19 ------------------- include/linux/compiler.h | 18 ++++++++++++++++-- 4 files changed, 17 insertions(+), 27 deletions(-) (limited to 'include/linux/compiler-gcc.h') diff --git a/include/asm-generic/barrier.h b/include/asm-generic/barrier.h index 798027bb89be..640f09479bdf 100644 --- a/include/asm-generic/barrier.h +++ b/include/asm-generic/barrier.h @@ -13,6 +13,7 @@ #ifndef __ASSEMBLY__ +#include #include #ifndef nop diff --git a/include/linux/compiler-clang.h b/include/linux/compiler-clang.h index 230604e7f057..dd7233c48bf3 100644 --- a/include/linux/compiler-clang.h +++ b/include/linux/compiler-clang.h @@ -60,12 +60,6 @@ #define COMPILER_HAS_GENERIC_BUILTIN_OVERFLOW 1 #endif -/* The following are for compatibility with GCC, from compiler-gcc.h, - * and may be redefined here because they should not be shared with other - * compilers, like ICC. - */ -#define barrier() __asm__ __volatile__("" : : : "memory") - #if __has_feature(shadow_call_stack) # define __noscs __attribute__((__no_sanitize__("shadow-call-stack"))) #endif diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h index 5deb37024574..74c6c0486eed 100644 --- a/include/linux/compiler-gcc.h +++ b/include/linux/compiler-gcc.h @@ -15,25 +15,6 @@ # error Sorry, your version of GCC is too old - please use 4.9 or newer. #endif -/* Optimization barrier */ - -/* The "volatile" is due to gcc bugs */ -#define barrier() __asm__ __volatile__("": : :"memory") -/* - * This version is i.e. to prevent dead stores elimination on @ptr - * where gcc and llvm may behave differently when otherwise using - * normal barrier(): while gcc behavior gets along with a normal - * barrier(), llvm needs an explicit input variable to be assumed - * clobbered. The issue is as follows: while the inline asm might - * access any memory it wants, the compiler could have fit all of - * @ptr into memory registers instead, and since @ptr never escaped - * from that, it proved that the inline asm wasn't touching any of - * it. This version works well with both compilers, i.e. we're telling - * the compiler that the inline asm absolutely may see the contents - * of @ptr. See also: https://llvm.org/bugs/show_bug.cgi?id=15495 - */ -#define barrier_data(ptr) __asm__ __volatile__("": :"r"(ptr) :"memory") - /* * This macro obfuscates arithmetic on a variable address so that gcc * shouldn't recognize the original var, and make assumptions about it. diff --git a/include/linux/compiler.h b/include/linux/compiler.h index e512f5505dad..b8fe0c23cfff 100644 --- a/include/linux/compiler.h +++ b/include/linux/compiler.h @@ -80,11 +80,25 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val, /* Optimization barrier */ #ifndef barrier -# define barrier() __memory_barrier() +/* The "volatile" is due to gcc bugs */ +# define barrier() __asm__ __volatile__("": : :"memory") #endif #ifndef barrier_data -# define barrier_data(ptr) barrier() +/* + * This version is i.e. to prevent dead stores elimination on @ptr + * where gcc and llvm may behave differently when otherwise using + * normal barrier(): while gcc behavior gets along with a normal + * barrier(), llvm needs an explicit input variable to be assumed + * clobbered. The issue is as follows: while the inline asm might + * access any memory it wants, the compiler could have fit all of + * @ptr into memory registers instead, and since @ptr never escaped + * from that, it proved that the inline asm wasn't touching any of + * it. This version works well with both compilers, i.e. we're telling + * the compiler that the inline asm absolutely may see the contents + * of @ptr. See also: https://llvm.org/bugs/show_bug.cgi?id=15495 + */ +# define barrier_data(ptr) __asm__ __volatile__("": :"r"(ptr) :"memory") #endif /* workaround for GCC PR82365 if needed */ -- cgit v1.2.3