* [PATCH v2] Remove Intel compiler support
@ 2022-10-11 17:14 Masahiro Yamada
2022-10-13 20:02 ` Nathan Chancellor
2022-10-14 14:39 ` Miguel Ojeda
0 siblings, 2 replies; 5+ messages in thread
From: Masahiro Yamada @ 2022-10-11 17:14 UTC (permalink / raw)
To: Linus Torvalds, Thomas Gleixner, Ingo Molnar, Borislav Petkov,
Dave Hansen, x86, H . Peter Anvin
Cc: Greg Kroah-Hartman, linux-ia64, Nick Desaulniers, Miguel Ojeda,
Masahiro Yamada, Arnd Bergmann, Len Brown, Nathan Chancellor,
Nick Terrell, Rafael J. Wysocki, Robert Moore, Tom Rix, devel,
linux-acpi, linux-kernel, llvm
include/linux/compiler-intel.h had no update in the past 3 years.
We often forget about the third C compiler to build the kernel.
For example, commit a0a12c3ed057 ("asm goto: eradicate CC_HAS_ASM_GOTO")
only mentioned GCC and Clang.
init/Kconfig defines CC_IS_GCC and CC_IS_CLANG but not CC_IS_ICC,
and nobody has reported any issue.
I guess the Intel Compiler support is broken, and nobody is caring
about it.
Harald Arnesen pointed out ICC (classic Intel C/C++ compiler) is
deprecated:
$ icc -v
icc: remark #10441: The Intel(R) C++ Compiler Classic (ICC) is
deprecated and will be removed from product release in the second half
of 2023. The Intel(R) oneAPI DPC++/C++ Compiler (ICX) is the recommended
compiler moving forward. Please transition to use this compiler. Use
'-diag-disable=10441' to disable this message.
icc version 2021.7.0 (gcc version 12.1.0 compatibility)
Arnd Bergmann provided a link to the article, "Intel C/C++ compilers
complete adoption of LLVM".
Link: https://www.intel.com/content/www/us/en/developer/articles/technical/adoption-of-llvm-complete-icx.html
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
---
Changes in v2:
- Remove "not supported by icc." comments from compiler_attributes.h
- Clean up __assume_aligned() macro
- Add more info from Harald Arnesen and Arnd Bergmann
arch/ia64/include/uapi/asm/cmpxchg.h | 6 +-
arch/ia64/include/uapi/asm/intel_intrin.h | 162 ----------------------
arch/ia64/include/uapi/asm/intrinsics.h | 6 +-
include/acpi/platform/acenv.h | 5 +-
include/acpi/platform/acenvex.h | 2 +-
include/acpi/platform/acintel.h | 55 --------
include/linux/compiler-intel.h | 34 -----
include/linux/compiler_attributes.h | 14 +-
include/linux/compiler_types.h | 2 -
lib/zstd/common/compiler.h | 2 +-
lib/zstd/compress/zstd_fast.c | 7 -
scripts/cc-version.sh | 2 -
scripts/min-tool-version.sh | 4 -
13 files changed, 6 insertions(+), 295 deletions(-)
delete mode 100644 arch/ia64/include/uapi/asm/intel_intrin.h
delete mode 100644 include/acpi/platform/acintel.h
delete mode 100644 include/linux/compiler-intel.h
diff --git a/arch/ia64/include/uapi/asm/cmpxchg.h b/arch/ia64/include/uapi/asm/cmpxchg.h
index ca2e02685343..259ae57570bf 100644
--- a/arch/ia64/include/uapi/asm/cmpxchg.h
+++ b/arch/ia64/include/uapi/asm/cmpxchg.h
@@ -15,11 +15,7 @@
#include <linux/types.h>
/* include compiler specific intrinsics */
#include <asm/ia64regs.h>
-#ifdef __INTEL_COMPILER
-# include <asm/intel_intrin.h>
-#else
-# include <asm/gcc_intrin.h>
-#endif
+#include <asm/gcc_intrin.h>
/*
* This function doesn't exist, so you'll get a linker error if
diff --git a/arch/ia64/include/uapi/asm/intel_intrin.h b/arch/ia64/include/uapi/asm/intel_intrin.h
deleted file mode 100644
index dc1884dc54b5..000000000000
--- a/arch/ia64/include/uapi/asm/intel_intrin.h
+++ /dev/null
@@ -1,162 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
-#ifndef _ASM_IA64_INTEL_INTRIN_H
-#define _ASM_IA64_INTEL_INTRIN_H
-/*
- * Intel Compiler Intrinsics
- *
- * Copyright (C) 2002,2003 Jun Nakajima <jun.nakajima@intel.com>
- * Copyright (C) 2002,2003 Suresh Siddha <suresh.b.siddha@intel.com>
- * Copyright (C) 2005,2006 Hongjiu Lu <hongjiu.lu@intel.com>
- *
- */
-#include <ia64intrin.h>
-
-#define ia64_barrier() __memory_barrier()
-
-#define ia64_stop() /* Nothing: As of now stop bit is generated for each
- * intrinsic
- */
-
-#define ia64_getreg __getReg
-#define ia64_setreg __setReg
-
-#define ia64_hint __hint
-#define ia64_hint_pause __hint_pause
-
-#define ia64_mux1_brcst _m64_mux1_brcst
-#define ia64_mux1_mix _m64_mux1_mix
-#define ia64_mux1_shuf _m64_mux1_shuf
-#define ia64_mux1_alt _m64_mux1_alt
-#define ia64_mux1_rev _m64_mux1_rev
-
-#define ia64_mux1(x,v) _m_to_int64(_m64_mux1(_m_from_int64(x), (v)))
-#define ia64_popcnt _m64_popcnt
-#define ia64_getf_exp __getf_exp
-#define ia64_shrp _m64_shrp
-
-#define ia64_tpa __tpa
-#define ia64_invala __invala
-#define ia64_invala_gr __invala_gr
-#define ia64_invala_fr __invala_fr
-#define ia64_nop __nop
-#define ia64_sum __sum
-#define ia64_ssm __ssm
-#define ia64_rum __rum
-#define ia64_rsm __rsm
-#define ia64_fc __fc
-
-#define ia64_ldfs __ldfs
-#define ia64_ldfd __ldfd
-#define ia64_ldfe __ldfe
-#define ia64_ldf8 __ldf8
-#define ia64_ldf_fill __ldf_fill
-
-#define ia64_stfs __stfs
-#define ia64_stfd __stfd
-#define ia64_stfe __stfe
-#define ia64_stf8 __stf8
-#define ia64_stf_spill __stf_spill
-
-#define ia64_mf __mf
-#define ia64_mfa __mfa
-
-#define ia64_fetchadd4_acq __fetchadd4_acq
-#define ia64_fetchadd4_rel __fetchadd4_rel
-#define ia64_fetchadd8_acq __fetchadd8_acq
-#define ia64_fetchadd8_rel __fetchadd8_rel
-
-#define ia64_xchg1 _InterlockedExchange8
-#define ia64_xchg2 _InterlockedExchange16
-#define ia64_xchg4 _InterlockedExchange
-#define ia64_xchg8 _InterlockedExchange64
-
-#define ia64_cmpxchg1_rel _InterlockedCompareExchange8_rel
-#define ia64_cmpxchg1_acq _InterlockedCompareExchange8_acq
-#define ia64_cmpxchg2_rel _InterlockedCompareExchange16_rel
-#define ia64_cmpxchg2_acq _InterlockedCompareExchange16_acq
-#define ia64_cmpxchg4_rel _InterlockedCompareExchange_rel
-#define ia64_cmpxchg4_acq _InterlockedCompareExchange_acq
-#define ia64_cmpxchg8_rel _InterlockedCompareExchange64_rel
-#define ia64_cmpxchg8_acq _InterlockedCompareExchange64_acq
-
-#define __ia64_set_dbr(index, val) \
- __setIndReg(_IA64_REG_INDR_DBR, index, val)
-#define ia64_set_ibr(index, val) \
- __setIndReg(_IA64_REG_INDR_IBR, index, val)
-#define ia64_set_pkr(index, val) \
- __setIndReg(_IA64_REG_INDR_PKR, index, val)
-#define ia64_set_pmc(index, val) \
- __setIndReg(_IA64_REG_INDR_PMC, index, val)
-#define ia64_set_pmd(index, val) \
- __setIndReg(_IA64_REG_INDR_PMD, index, val)
-#define ia64_set_rr(index, val) \
- __setIndReg(_IA64_REG_INDR_RR, index, val)
-
-#define ia64_get_cpuid(index) \
- __getIndReg(_IA64_REG_INDR_CPUID, index)
-#define __ia64_get_dbr(index) __getIndReg(_IA64_REG_INDR_DBR, index)
-#define ia64_get_ibr(index) __getIndReg(_IA64_REG_INDR_IBR, index)
-#define ia64_get_pkr(index) __getIndReg(_IA64_REG_INDR_PKR, index)
-#define ia64_get_pmc(index) __getIndReg(_IA64_REG_INDR_PMC, index)
-#define ia64_get_pmd(index) __getIndReg(_IA64_REG_INDR_PMD, index)
-#define ia64_get_rr(index) __getIndReg(_IA64_REG_INDR_RR, index)
-
-#define ia64_srlz_d __dsrlz
-#define ia64_srlz_i __isrlz
-
-#define ia64_dv_serialize_data()
-#define ia64_dv_serialize_instruction()
-
-#define ia64_st1_rel __st1_rel
-#define ia64_st2_rel __st2_rel
-#define ia64_st4_rel __st4_rel
-#define ia64_st8_rel __st8_rel
-
-/* FIXME: need st4.rel.nta intrinsic */
-#define ia64_st4_rel_nta __st4_rel
-
-#define ia64_ld1_acq __ld1_acq
-#define ia64_ld2_acq __ld2_acq
-#define ia64_ld4_acq __ld4_acq
-#define ia64_ld8_acq __ld8_acq
-
-#define ia64_sync_i __synci
-#define ia64_thash __thash
-#define ia64_ttag __ttag
-#define ia64_itcd __itcd
-#define ia64_itci __itci
-#define ia64_itrd __itrd
-#define ia64_itri __itri
-#define ia64_ptce __ptce
-#define ia64_ptcl __ptcl
-#define ia64_ptcg __ptcg
-#define ia64_ptcga __ptcga
-#define ia64_ptri __ptri
-#define ia64_ptrd __ptrd
-#define ia64_dep_mi _m64_dep_mi
-
-/* Values for lfhint in __lfetch and __lfetch_fault */
-
-#define ia64_lfhint_none __lfhint_none
-#define ia64_lfhint_nt1 __lfhint_nt1
-#define ia64_lfhint_nt2 __lfhint_nt2
-#define ia64_lfhint_nta __lfhint_nta
-
-#define ia64_lfetch __lfetch
-#define ia64_lfetch_excl __lfetch_excl
-#define ia64_lfetch_fault __lfetch_fault
-#define ia64_lfetch_fault_excl __lfetch_fault_excl
-
-#define ia64_intrin_local_irq_restore(x) \
-do { \
- if ((x) != 0) { \
- ia64_ssm(IA64_PSR_I); \
- ia64_srlz_d(); \
- } else { \
- ia64_rsm(IA64_PSR_I); \
- } \
-} while (0)
-
-#define __builtin_trap() __break(0);
-
-#endif /* _ASM_IA64_INTEL_INTRIN_H */
diff --git a/arch/ia64/include/uapi/asm/intrinsics.h b/arch/ia64/include/uapi/asm/intrinsics.h
index a0e0a064f5b1..63f27c4ec739 100644
--- a/arch/ia64/include/uapi/asm/intrinsics.h
+++ b/arch/ia64/include/uapi/asm/intrinsics.h
@@ -14,11 +14,7 @@
#include <linux/types.h>
/* include compiler specific intrinsics */
#include <asm/ia64regs.h>
-#ifdef __INTEL_COMPILER
-# include <asm/intel_intrin.h>
-#else
-# include <asm/gcc_intrin.h>
-#endif
+#include <asm/gcc_intrin.h>
#include <asm/cmpxchg.h>
#define ia64_set_rr0_to_rr4(val0, val1, val2, val3, val4) \
diff --git a/include/acpi/platform/acenv.h b/include/acpi/platform/acenv.h
index 03eb3d977075..9e4f7564201a 100644
--- a/include/acpi/platform/acenv.h
+++ b/include/acpi/platform/acenv.h
@@ -148,15 +148,12 @@
*
*****************************************************************************/
-#if defined(__GNUC__) && !defined(__INTEL_COMPILER)
+#if defined(__GNUC__)
#include <acpi/platform/acgcc.h>
#elif defined(_MSC_VER)
#include "acmsvc.h"
-#elif defined(__INTEL_COMPILER)
-#include <acpi/platform/acintel.h>
-
#endif
#if defined(_LINUX) || defined(__linux__)
diff --git a/include/acpi/platform/acenvex.h b/include/acpi/platform/acenvex.h
index 3a6b1db9a984..72cc7bab469e 100644
--- a/include/acpi/platform/acenvex.h
+++ b/include/acpi/platform/acenvex.h
@@ -35,7 +35,7 @@
#endif
-#if defined(__GNUC__) && !defined(__INTEL_COMPILER)
+#if defined(__GNUC__)
#include "acgccex.h"
#elif defined(_MSC_VER)
diff --git a/include/acpi/platform/acintel.h b/include/acpi/platform/acintel.h
deleted file mode 100644
index 85b1ae86ee63..000000000000
--- a/include/acpi/platform/acintel.h
+++ /dev/null
@@ -1,55 +0,0 @@
-/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
-/******************************************************************************
- *
- * Name: acintel.h - VC specific defines, etc.
- *
- * Copyright (C) 2000 - 2022, Intel Corp.
- *
- *****************************************************************************/
-
-#ifndef __ACINTEL_H__
-#define __ACINTEL_H__
-
-/*
- * Use compiler specific <stdarg.h> is a good practice for even when
- * -nostdinc is specified (i.e., ACPI_USE_STANDARD_HEADERS undefined.
- */
-#ifndef va_arg
-#include <stdarg.h>
-#endif
-
-/* Configuration specific to Intel 64-bit C compiler */
-
-#define COMPILER_DEPENDENT_INT64 __int64
-#define COMPILER_DEPENDENT_UINT64 unsigned __int64
-#define ACPI_INLINE __inline
-
-/*
- * Calling conventions:
- *
- * ACPI_SYSTEM_XFACE - Interfaces to host OS (handlers, threads)
- * ACPI_EXTERNAL_XFACE - External ACPI interfaces
- * ACPI_INTERNAL_XFACE - Internal ACPI interfaces
- * ACPI_INTERNAL_VAR_XFACE - Internal variable-parameter list interfaces
- */
-#define ACPI_SYSTEM_XFACE
-#define ACPI_EXTERNAL_XFACE
-#define ACPI_INTERNAL_XFACE
-#define ACPI_INTERNAL_VAR_XFACE
-
-/* remark 981 - operands evaluated in no particular order */
-#pragma warning(disable:981)
-
-/* warn C4100: unreferenced formal parameter */
-#pragma warning(disable:4100)
-
-/* warn C4127: conditional expression is constant */
-#pragma warning(disable:4127)
-
-/* warn C4706: assignment within conditional expression */
-#pragma warning(disable:4706)
-
-/* warn C4214: bit field types other than int */
-#pragma warning(disable:4214)
-
-#endif /* __ACINTEL_H__ */
diff --git a/include/linux/compiler-intel.h b/include/linux/compiler-intel.h
deleted file mode 100644
index b17f3cd18334..000000000000
--- a/include/linux/compiler-intel.h
+++ /dev/null
@@ -1,34 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-#ifndef __LINUX_COMPILER_TYPES_H
-#error "Please don't include <linux/compiler-intel.h> directly, include <linux/compiler.h> instead."
-#endif
-
-#ifdef __ECC
-
-/* Compiler specific definitions for Intel ECC compiler */
-
-#include <asm/intrinsics.h>
-
-/* Intel ECC compiler doesn't support gcc specific asm stmts.
- * It uses intrinsics to do the equivalent things.
- */
-
-#define barrier() __memory_barrier()
-#define barrier_data(ptr) barrier()
-
-#define RELOC_HIDE(ptr, off) \
- ({ unsigned long __ptr; \
- __ptr = (unsigned long) (ptr); \
- (typeof(ptr)) (__ptr + (off)); })
-
-/* This should act as an optimization barrier on var.
- * Given that this compiler does not have inline assembly, a compiler barrier
- * is the best we can do.
- */
-#define OPTIMIZER_HIDE_VAR(var) barrier()
-
-#endif
-
-/* icc has this, but it's called _bswap16 */
-#define __HAVE_BUILTIN_BSWAP16__
-#define __builtin_bswap16 _bswap16
diff --git a/include/linux/compiler_attributes.h b/include/linux/compiler_attributes.h
index 898b3458b24a..9221302f6ae8 100644
--- a/include/linux/compiler_attributes.h
+++ b/include/linux/compiler_attributes.h
@@ -64,16 +64,10 @@
* compiler should see some alignment anyway, when the return value is
* massaged by 'flags = ptr & 3; ptr &= ~3;').
*
- * Optional: not supported by icc
- *
* gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-assume_005faligned-function-attribute
* clang: https://clang.llvm.org/docs/AttributeReference.html#assume-aligned
*/
-#if __has_attribute(__assume_aligned__)
-# define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
-#else
-# define __assume_aligned(a, ...)
-#endif
+#define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
/*
* gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-cold-function-attribute
@@ -91,7 +85,6 @@
/*
* Optional: only supported since gcc >= 9
* Optional: not supported by clang
- * Optional: not supported by icc
*
* gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-copy-function-attribute
*/
@@ -104,7 +97,6 @@
/*
* Optional: not supported by gcc
* Optional: only supported since clang >= 14.0
- * Optional: not supported by icc
*
* clang: https://clang.llvm.org/docs/AttributeReference.html#diagnose_as_builtin
*/
@@ -128,7 +120,6 @@
/*
* Optional: not supported by clang
- * Optional: not supported by icc
*
* gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Type-Attributes.html#index-designated_005finit-type-attribute
*/
@@ -242,7 +233,6 @@
/*
* Optional: only supported since gcc >= 8
* Optional: not supported by clang
- * Optional: not supported by icc
*
* gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-nonstring-variable-attribute
*/
@@ -273,7 +263,6 @@
/*
* Optional: not supported by gcc.
- * Optional: not supported by icc.
*
* clang: https://clang.llvm.org/docs/AttributeReference.html#overloadable
*/
@@ -293,7 +282,6 @@
* Note: the "type" argument should match any __builtin_object_size(p, type) usage.
*
* Optional: not supported by gcc.
- * Optional: not supported by icc.
*
* clang: https://clang.llvm.org/docs/AttributeReference.html#pass-object-size-pass-dynamic-object-size
*/
diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index 74e04ecd4c89..0e06a1a79401 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h
@@ -92,8 +92,6 @@ static inline void __chk_io_ptr(const volatile void __iomem *ptr) { }
/* Compiler specific macros. */
#ifdef __clang__
#include <linux/compiler-clang.h>
-#elif defined(__INTEL_COMPILER)
-#include <linux/compiler-intel.h>
#elif defined(__GNUC__)
/* The above compilers also define __GNUC__, so order is important here. */
#include <linux/compiler-gcc.h>
diff --git a/lib/zstd/common/compiler.h b/lib/zstd/common/compiler.h
index f5a9c70a228a..c281a6430cd4 100644
--- a/lib/zstd/common/compiler.h
+++ b/lib/zstd/common/compiler.h
@@ -116,7 +116,7 @@
/* vectorization
* older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax */
-#if !defined(__INTEL_COMPILER) && !defined(__clang__) && defined(__GNUC__)
+#if !defined(__clang__) && defined(__GNUC__)
# if (__GNUC__ == 4 && __GNUC_MINOR__ > 3) || (__GNUC__ >= 5)
# define DONT_VECTORIZE __attribute__((optimize("no-tree-vectorize")))
# else
diff --git a/lib/zstd/compress/zstd_fast.c b/lib/zstd/compress/zstd_fast.c
index 96b7d48e2868..800f3865119f 100644
--- a/lib/zstd/compress/zstd_fast.c
+++ b/lib/zstd/compress/zstd_fast.c
@@ -80,13 +80,6 @@ ZSTD_compressBlock_fast_generic(
}
/* Main Search Loop */
-#ifdef __INTEL_COMPILER
- /* From intel 'The vector pragma indicates that the loop should be
- * vectorized if it is legal to do so'. Can be used together with
- * #pragma ivdep (but have opted to exclude that because intel
- * warns against using it).*/
- #pragma vector always
-#endif
while (ip1 < ilimit) { /* < instead of <=, because check at ip0+2 */
size_t mLength;
BYTE const* ip2 = ip0 + 2;
diff --git a/scripts/cc-version.sh b/scripts/cc-version.sh
index 2401c86fcf53..0573c92e841d 100755
--- a/scripts/cc-version.sh
+++ b/scripts/cc-version.sh
@@ -12,8 +12,6 @@ get_c_compiler_info()
cat <<- EOF | "$@" -E -P -x c - 2>/dev/null
#if defined(__clang__)
Clang __clang_major__ __clang_minor__ __clang_patchlevel__
- #elif defined(__INTEL_COMPILER)
- ICC __INTEL_COMPILER __INTEL_COMPILER_UPDATE
#elif defined(__GNUC__)
GCC __GNUC__ __GNUC_MINOR__ __GNUC_PATCHLEVEL__
#else
diff --git a/scripts/min-tool-version.sh b/scripts/min-tool-version.sh
index b6593eac5003..8766e248ffbb 100755
--- a/scripts/min-tool-version.sh
+++ b/scripts/min-tool-version.sh
@@ -19,10 +19,6 @@ binutils)
gcc)
echo 5.1.0
;;
-icc)
- # temporary
- echo 16.0.3
- ;;
llvm)
if [ "$SRCARCH" = s390 ]; then
echo 14.0.0
--
2.34.1
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH v2] Remove Intel compiler support
2022-10-11 17:14 [PATCH v2] Remove Intel compiler support Masahiro Yamada
@ 2022-10-13 20:02 ` Nathan Chancellor
2022-10-14 14:39 ` Miguel Ojeda
1 sibling, 0 replies; 5+ messages in thread
From: Nathan Chancellor @ 2022-10-13 20:02 UTC (permalink / raw)
To: Masahiro Yamada
Cc: Linus Torvalds, Thomas Gleixner, Ingo Molnar, Borislav Petkov,
Dave Hansen, x86, H . Peter Anvin, Greg Kroah-Hartman,
linux-ia64, Nick Desaulniers, Miguel Ojeda, Arnd Bergmann,
Len Brown, Nick Terrell, Rafael J. Wysocki, Robert Moore,
Tom Rix, devel, linux-acpi, linux-kernel, llvm
On Wed, Oct 12, 2022 at 02:14:27AM +0900, Masahiro Yamada wrote:
> include/linux/compiler-intel.h had no update in the past 3 years.
>
> We often forget about the third C compiler to build the kernel.
> For example, commit a0a12c3ed057 ("asm goto: eradicate CC_HAS_ASM_GOTO")
> only mentioned GCC and Clang.
>
> init/Kconfig defines CC_IS_GCC and CC_IS_CLANG but not CC_IS_ICC,
> and nobody has reported any issue.
>
> I guess the Intel Compiler support is broken, and nobody is caring
> about it.
>
> Harald Arnesen pointed out ICC (classic Intel C/C++ compiler) is
> deprecated:
>
> $ icc -v
> icc: remark #10441: The Intel(R) C++ Compiler Classic (ICC) is
> deprecated and will be removed from product release in the second half
> of 2023. The Intel(R) oneAPI DPC++/C++ Compiler (ICX) is the recommended
> compiler moving forward. Please transition to use this compiler. Use
> '-diag-disable=10441' to disable this message.
> icc version 2021.7.0 (gcc version 12.1.0 compatibility)
>
> Arnd Bergmann provided a link to the article, "Intel C/C++ compilers
> complete adoption of LLVM".
>
> Link: https://www.intel.com/content/www/us/en/developer/articles/technical/adoption-of-llvm-complete-icx.html
> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
> Acked-by: Arnd Bergmann <arnd@arndb.de>
> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
> ---
>
> Changes in v2:
> - Remove "not supported by icc." comments from compiler_attributes.h
> - Clean up __assume_aligned() macro
> - Add more info from Harald Arnesen and Arnd Bergmann
>
> arch/ia64/include/uapi/asm/cmpxchg.h | 6 +-
> arch/ia64/include/uapi/asm/intel_intrin.h | 162 ----------------------
> arch/ia64/include/uapi/asm/intrinsics.h | 6 +-
> include/acpi/platform/acenv.h | 5 +-
> include/acpi/platform/acenvex.h | 2 +-
> include/acpi/platform/acintel.h | 55 --------
> include/linux/compiler-intel.h | 34 -----
> include/linux/compiler_attributes.h | 14 +-
> include/linux/compiler_types.h | 2 -
> lib/zstd/common/compiler.h | 2 +-
> lib/zstd/compress/zstd_fast.c | 7 -
> scripts/cc-version.sh | 2 -
> scripts/min-tool-version.sh | 4 -
> 13 files changed, 6 insertions(+), 295 deletions(-)
> delete mode 100644 arch/ia64/include/uapi/asm/intel_intrin.h
> delete mode 100644 include/acpi/platform/acintel.h
> delete mode 100644 include/linux/compiler-intel.h
>
> diff --git a/arch/ia64/include/uapi/asm/cmpxchg.h b/arch/ia64/include/uapi/asm/cmpxchg.h
> index ca2e02685343..259ae57570bf 100644
> --- a/arch/ia64/include/uapi/asm/cmpxchg.h
> +++ b/arch/ia64/include/uapi/asm/cmpxchg.h
> @@ -15,11 +15,7 @@
> #include <linux/types.h>
> /* include compiler specific intrinsics */
> #include <asm/ia64regs.h>
> -#ifdef __INTEL_COMPILER
> -# include <asm/intel_intrin.h>
> -#else
> -# include <asm/gcc_intrin.h>
> -#endif
> +#include <asm/gcc_intrin.h>
>
> /*
> * This function doesn't exist, so you'll get a linker error if
> diff --git a/arch/ia64/include/uapi/asm/intel_intrin.h b/arch/ia64/include/uapi/asm/intel_intrin.h
> deleted file mode 100644
> index dc1884dc54b5..000000000000
> --- a/arch/ia64/include/uapi/asm/intel_intrin.h
> +++ /dev/null
> @@ -1,162 +0,0 @@
> -/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
> -#ifndef _ASM_IA64_INTEL_INTRIN_H
> -#define _ASM_IA64_INTEL_INTRIN_H
> -/*
> - * Intel Compiler Intrinsics
> - *
> - * Copyright (C) 2002,2003 Jun Nakajima <jun.nakajima@intel.com>
> - * Copyright (C) 2002,2003 Suresh Siddha <suresh.b.siddha@intel.com>
> - * Copyright (C) 2005,2006 Hongjiu Lu <hongjiu.lu@intel.com>
> - *
> - */
> -#include <ia64intrin.h>
> -
> -#define ia64_barrier() __memory_barrier()
> -
> -#define ia64_stop() /* Nothing: As of now stop bit is generated for each
> - * intrinsic
> - */
> -
> -#define ia64_getreg __getReg
> -#define ia64_setreg __setReg
> -
> -#define ia64_hint __hint
> -#define ia64_hint_pause __hint_pause
> -
> -#define ia64_mux1_brcst _m64_mux1_brcst
> -#define ia64_mux1_mix _m64_mux1_mix
> -#define ia64_mux1_shuf _m64_mux1_shuf
> -#define ia64_mux1_alt _m64_mux1_alt
> -#define ia64_mux1_rev _m64_mux1_rev
> -
> -#define ia64_mux1(x,v) _m_to_int64(_m64_mux1(_m_from_int64(x), (v)))
> -#define ia64_popcnt _m64_popcnt
> -#define ia64_getf_exp __getf_exp
> -#define ia64_shrp _m64_shrp
> -
> -#define ia64_tpa __tpa
> -#define ia64_invala __invala
> -#define ia64_invala_gr __invala_gr
> -#define ia64_invala_fr __invala_fr
> -#define ia64_nop __nop
> -#define ia64_sum __sum
> -#define ia64_ssm __ssm
> -#define ia64_rum __rum
> -#define ia64_rsm __rsm
> -#define ia64_fc __fc
> -
> -#define ia64_ldfs __ldfs
> -#define ia64_ldfd __ldfd
> -#define ia64_ldfe __ldfe
> -#define ia64_ldf8 __ldf8
> -#define ia64_ldf_fill __ldf_fill
> -
> -#define ia64_stfs __stfs
> -#define ia64_stfd __stfd
> -#define ia64_stfe __stfe
> -#define ia64_stf8 __stf8
> -#define ia64_stf_spill __stf_spill
> -
> -#define ia64_mf __mf
> -#define ia64_mfa __mfa
> -
> -#define ia64_fetchadd4_acq __fetchadd4_acq
> -#define ia64_fetchadd4_rel __fetchadd4_rel
> -#define ia64_fetchadd8_acq __fetchadd8_acq
> -#define ia64_fetchadd8_rel __fetchadd8_rel
> -
> -#define ia64_xchg1 _InterlockedExchange8
> -#define ia64_xchg2 _InterlockedExchange16
> -#define ia64_xchg4 _InterlockedExchange
> -#define ia64_xchg8 _InterlockedExchange64
> -
> -#define ia64_cmpxchg1_rel _InterlockedCompareExchange8_rel
> -#define ia64_cmpxchg1_acq _InterlockedCompareExchange8_acq
> -#define ia64_cmpxchg2_rel _InterlockedCompareExchange16_rel
> -#define ia64_cmpxchg2_acq _InterlockedCompareExchange16_acq
> -#define ia64_cmpxchg4_rel _InterlockedCompareExchange_rel
> -#define ia64_cmpxchg4_acq _InterlockedCompareExchange_acq
> -#define ia64_cmpxchg8_rel _InterlockedCompareExchange64_rel
> -#define ia64_cmpxchg8_acq _InterlockedCompareExchange64_acq
> -
> -#define __ia64_set_dbr(index, val) \
> - __setIndReg(_IA64_REG_INDR_DBR, index, val)
> -#define ia64_set_ibr(index, val) \
> - __setIndReg(_IA64_REG_INDR_IBR, index, val)
> -#define ia64_set_pkr(index, val) \
> - __setIndReg(_IA64_REG_INDR_PKR, index, val)
> -#define ia64_set_pmc(index, val) \
> - __setIndReg(_IA64_REG_INDR_PMC, index, val)
> -#define ia64_set_pmd(index, val) \
> - __setIndReg(_IA64_REG_INDR_PMD, index, val)
> -#define ia64_set_rr(index, val) \
> - __setIndReg(_IA64_REG_INDR_RR, index, val)
> -
> -#define ia64_get_cpuid(index) \
> - __getIndReg(_IA64_REG_INDR_CPUID, index)
> -#define __ia64_get_dbr(index) __getIndReg(_IA64_REG_INDR_DBR, index)
> -#define ia64_get_ibr(index) __getIndReg(_IA64_REG_INDR_IBR, index)
> -#define ia64_get_pkr(index) __getIndReg(_IA64_REG_INDR_PKR, index)
> -#define ia64_get_pmc(index) __getIndReg(_IA64_REG_INDR_PMC, index)
> -#define ia64_get_pmd(index) __getIndReg(_IA64_REG_INDR_PMD, index)
> -#define ia64_get_rr(index) __getIndReg(_IA64_REG_INDR_RR, index)
> -
> -#define ia64_srlz_d __dsrlz
> -#define ia64_srlz_i __isrlz
> -
> -#define ia64_dv_serialize_data()
> -#define ia64_dv_serialize_instruction()
> -
> -#define ia64_st1_rel __st1_rel
> -#define ia64_st2_rel __st2_rel
> -#define ia64_st4_rel __st4_rel
> -#define ia64_st8_rel __st8_rel
> -
> -/* FIXME: need st4.rel.nta intrinsic */
> -#define ia64_st4_rel_nta __st4_rel
> -
> -#define ia64_ld1_acq __ld1_acq
> -#define ia64_ld2_acq __ld2_acq
> -#define ia64_ld4_acq __ld4_acq
> -#define ia64_ld8_acq __ld8_acq
> -
> -#define ia64_sync_i __synci
> -#define ia64_thash __thash
> -#define ia64_ttag __ttag
> -#define ia64_itcd __itcd
> -#define ia64_itci __itci
> -#define ia64_itrd __itrd
> -#define ia64_itri __itri
> -#define ia64_ptce __ptce
> -#define ia64_ptcl __ptcl
> -#define ia64_ptcg __ptcg
> -#define ia64_ptcga __ptcga
> -#define ia64_ptri __ptri
> -#define ia64_ptrd __ptrd
> -#define ia64_dep_mi _m64_dep_mi
> -
> -/* Values for lfhint in __lfetch and __lfetch_fault */
> -
> -#define ia64_lfhint_none __lfhint_none
> -#define ia64_lfhint_nt1 __lfhint_nt1
> -#define ia64_lfhint_nt2 __lfhint_nt2
> -#define ia64_lfhint_nta __lfhint_nta
> -
> -#define ia64_lfetch __lfetch
> -#define ia64_lfetch_excl __lfetch_excl
> -#define ia64_lfetch_fault __lfetch_fault
> -#define ia64_lfetch_fault_excl __lfetch_fault_excl
> -
> -#define ia64_intrin_local_irq_restore(x) \
> -do { \
> - if ((x) != 0) { \
> - ia64_ssm(IA64_PSR_I); \
> - ia64_srlz_d(); \
> - } else { \
> - ia64_rsm(IA64_PSR_I); \
> - } \
> -} while (0)
> -
> -#define __builtin_trap() __break(0);
> -
> -#endif /* _ASM_IA64_INTEL_INTRIN_H */
> diff --git a/arch/ia64/include/uapi/asm/intrinsics.h b/arch/ia64/include/uapi/asm/intrinsics.h
> index a0e0a064f5b1..63f27c4ec739 100644
> --- a/arch/ia64/include/uapi/asm/intrinsics.h
> +++ b/arch/ia64/include/uapi/asm/intrinsics.h
> @@ -14,11 +14,7 @@
> #include <linux/types.h>
> /* include compiler specific intrinsics */
> #include <asm/ia64regs.h>
> -#ifdef __INTEL_COMPILER
> -# include <asm/intel_intrin.h>
> -#else
> -# include <asm/gcc_intrin.h>
> -#endif
> +#include <asm/gcc_intrin.h>
> #include <asm/cmpxchg.h>
>
> #define ia64_set_rr0_to_rr4(val0, val1, val2, val3, val4) \
> diff --git a/include/acpi/platform/acenv.h b/include/acpi/platform/acenv.h
> index 03eb3d977075..9e4f7564201a 100644
> --- a/include/acpi/platform/acenv.h
> +++ b/include/acpi/platform/acenv.h
> @@ -148,15 +148,12 @@
> *
> *****************************************************************************/
>
> -#if defined(__GNUC__) && !defined(__INTEL_COMPILER)
> +#if defined(__GNUC__)
> #include <acpi/platform/acgcc.h>
>
> #elif defined(_MSC_VER)
> #include "acmsvc.h"
>
> -#elif defined(__INTEL_COMPILER)
> -#include <acpi/platform/acintel.h>
> -
> #endif
>
> #if defined(_LINUX) || defined(__linux__)
> diff --git a/include/acpi/platform/acenvex.h b/include/acpi/platform/acenvex.h
> index 3a6b1db9a984..72cc7bab469e 100644
> --- a/include/acpi/platform/acenvex.h
> +++ b/include/acpi/platform/acenvex.h
> @@ -35,7 +35,7 @@
>
> #endif
>
> -#if defined(__GNUC__) && !defined(__INTEL_COMPILER)
> +#if defined(__GNUC__)
> #include "acgccex.h"
>
> #elif defined(_MSC_VER)
> diff --git a/include/acpi/platform/acintel.h b/include/acpi/platform/acintel.h
> deleted file mode 100644
> index 85b1ae86ee63..000000000000
> --- a/include/acpi/platform/acintel.h
> +++ /dev/null
> @@ -1,55 +0,0 @@
> -/* SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0 */
> -/******************************************************************************
> - *
> - * Name: acintel.h - VC specific defines, etc.
> - *
> - * Copyright (C) 2000 - 2022, Intel Corp.
> - *
> - *****************************************************************************/
> -
> -#ifndef __ACINTEL_H__
> -#define __ACINTEL_H__
> -
> -/*
> - * Use compiler specific <stdarg.h> is a good practice for even when
> - * -nostdinc is specified (i.e., ACPI_USE_STANDARD_HEADERS undefined.
> - */
> -#ifndef va_arg
> -#include <stdarg.h>
> -#endif
> -
> -/* Configuration specific to Intel 64-bit C compiler */
> -
> -#define COMPILER_DEPENDENT_INT64 __int64
> -#define COMPILER_DEPENDENT_UINT64 unsigned __int64
> -#define ACPI_INLINE __inline
> -
> -/*
> - * Calling conventions:
> - *
> - * ACPI_SYSTEM_XFACE - Interfaces to host OS (handlers, threads)
> - * ACPI_EXTERNAL_XFACE - External ACPI interfaces
> - * ACPI_INTERNAL_XFACE - Internal ACPI interfaces
> - * ACPI_INTERNAL_VAR_XFACE - Internal variable-parameter list interfaces
> - */
> -#define ACPI_SYSTEM_XFACE
> -#define ACPI_EXTERNAL_XFACE
> -#define ACPI_INTERNAL_XFACE
> -#define ACPI_INTERNAL_VAR_XFACE
> -
> -/* remark 981 - operands evaluated in no particular order */
> -#pragma warning(disable:981)
> -
> -/* warn C4100: unreferenced formal parameter */
> -#pragma warning(disable:4100)
> -
> -/* warn C4127: conditional expression is constant */
> -#pragma warning(disable:4127)
> -
> -/* warn C4706: assignment within conditional expression */
> -#pragma warning(disable:4706)
> -
> -/* warn C4214: bit field types other than int */
> -#pragma warning(disable:4214)
> -
> -#endif /* __ACINTEL_H__ */
> diff --git a/include/linux/compiler-intel.h b/include/linux/compiler-intel.h
> deleted file mode 100644
> index b17f3cd18334..000000000000
> --- a/include/linux/compiler-intel.h
> +++ /dev/null
> @@ -1,34 +0,0 @@
> -/* SPDX-License-Identifier: GPL-2.0 */
> -#ifndef __LINUX_COMPILER_TYPES_H
> -#error "Please don't include <linux/compiler-intel.h> directly, include <linux/compiler.h> instead."
> -#endif
> -
> -#ifdef __ECC
> -
> -/* Compiler specific definitions for Intel ECC compiler */
> -
> -#include <asm/intrinsics.h>
> -
> -/* Intel ECC compiler doesn't support gcc specific asm stmts.
> - * It uses intrinsics to do the equivalent things.
> - */
> -
> -#define barrier() __memory_barrier()
> -#define barrier_data(ptr) barrier()
> -
> -#define RELOC_HIDE(ptr, off) \
> - ({ unsigned long __ptr; \
> - __ptr = (unsigned long) (ptr); \
> - (typeof(ptr)) (__ptr + (off)); })
> -
> -/* This should act as an optimization barrier on var.
> - * Given that this compiler does not have inline assembly, a compiler barrier
> - * is the best we can do.
> - */
> -#define OPTIMIZER_HIDE_VAR(var) barrier()
> -
> -#endif
> -
> -/* icc has this, but it's called _bswap16 */
> -#define __HAVE_BUILTIN_BSWAP16__
> -#define __builtin_bswap16 _bswap16
> diff --git a/include/linux/compiler_attributes.h b/include/linux/compiler_attributes.h
> index 898b3458b24a..9221302f6ae8 100644
> --- a/include/linux/compiler_attributes.h
> +++ b/include/linux/compiler_attributes.h
> @@ -64,16 +64,10 @@
> * compiler should see some alignment anyway, when the return value is
> * massaged by 'flags = ptr & 3; ptr &= ~3;').
> *
> - * Optional: not supported by icc
> - *
> * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-assume_005faligned-function-attribute
> * clang: https://clang.llvm.org/docs/AttributeReference.html#assume-aligned
> */
> -#if __has_attribute(__assume_aligned__)
> -# define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
> -#else
> -# define __assume_aligned(a, ...)
> -#endif
> +#define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
>
> /*
> * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-cold-function-attribute
> @@ -91,7 +85,6 @@
> /*
> * Optional: only supported since gcc >= 9
> * Optional: not supported by clang
> - * Optional: not supported by icc
> *
> * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-copy-function-attribute
> */
> @@ -104,7 +97,6 @@
> /*
> * Optional: not supported by gcc
> * Optional: only supported since clang >= 14.0
> - * Optional: not supported by icc
> *
> * clang: https://clang.llvm.org/docs/AttributeReference.html#diagnose_as_builtin
> */
> @@ -128,7 +120,6 @@
>
> /*
> * Optional: not supported by clang
> - * Optional: not supported by icc
> *
> * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Type-Attributes.html#index-designated_005finit-type-attribute
> */
> @@ -242,7 +233,6 @@
> /*
> * Optional: only supported since gcc >= 8
> * Optional: not supported by clang
> - * Optional: not supported by icc
> *
> * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-nonstring-variable-attribute
> */
> @@ -273,7 +263,6 @@
>
> /*
> * Optional: not supported by gcc.
> - * Optional: not supported by icc.
> *
> * clang: https://clang.llvm.org/docs/AttributeReference.html#overloadable
> */
> @@ -293,7 +282,6 @@
> * Note: the "type" argument should match any __builtin_object_size(p, type) usage.
> *
> * Optional: not supported by gcc.
> - * Optional: not supported by icc.
> *
> * clang: https://clang.llvm.org/docs/AttributeReference.html#pass-object-size-pass-dynamic-object-size
> */
> diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
> index 74e04ecd4c89..0e06a1a79401 100644
> --- a/include/linux/compiler_types.h
> +++ b/include/linux/compiler_types.h
> @@ -92,8 +92,6 @@ static inline void __chk_io_ptr(const volatile void __iomem *ptr) { }
> /* Compiler specific macros. */
> #ifdef __clang__
> #include <linux/compiler-clang.h>
> -#elif defined(__INTEL_COMPILER)
> -#include <linux/compiler-intel.h>
> #elif defined(__GNUC__)
> /* The above compilers also define __GNUC__, so order is important here. */
> #include <linux/compiler-gcc.h>
> diff --git a/lib/zstd/common/compiler.h b/lib/zstd/common/compiler.h
> index f5a9c70a228a..c281a6430cd4 100644
> --- a/lib/zstd/common/compiler.h
> +++ b/lib/zstd/common/compiler.h
> @@ -116,7 +116,7 @@
>
> /* vectorization
> * older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax */
> -#if !defined(__INTEL_COMPILER) && !defined(__clang__) && defined(__GNUC__)
> +#if !defined(__clang__) && defined(__GNUC__)
> # if (__GNUC__ == 4 && __GNUC_MINOR__ > 3) || (__GNUC__ >= 5)
> # define DONT_VECTORIZE __attribute__((optimize("no-tree-vectorize")))
> # else
> diff --git a/lib/zstd/compress/zstd_fast.c b/lib/zstd/compress/zstd_fast.c
> index 96b7d48e2868..800f3865119f 100644
> --- a/lib/zstd/compress/zstd_fast.c
> +++ b/lib/zstd/compress/zstd_fast.c
> @@ -80,13 +80,6 @@ ZSTD_compressBlock_fast_generic(
> }
>
> /* Main Search Loop */
> -#ifdef __INTEL_COMPILER
> - /* From intel 'The vector pragma indicates that the loop should be
> - * vectorized if it is legal to do so'. Can be used together with
> - * #pragma ivdep (but have opted to exclude that because intel
> - * warns against using it).*/
> - #pragma vector always
> -#endif
> while (ip1 < ilimit) { /* < instead of <=, because check at ip0+2 */
> size_t mLength;
> BYTE const* ip2 = ip0 + 2;
> diff --git a/scripts/cc-version.sh b/scripts/cc-version.sh
> index 2401c86fcf53..0573c92e841d 100755
> --- a/scripts/cc-version.sh
> +++ b/scripts/cc-version.sh
> @@ -12,8 +12,6 @@ get_c_compiler_info()
> cat <<- EOF | "$@" -E -P -x c - 2>/dev/null
> #if defined(__clang__)
> Clang __clang_major__ __clang_minor__ __clang_patchlevel__
> - #elif defined(__INTEL_COMPILER)
> - ICC __INTEL_COMPILER __INTEL_COMPILER_UPDATE
> #elif defined(__GNUC__)
> GCC __GNUC__ __GNUC_MINOR__ __GNUC_PATCHLEVEL__
> #else
> diff --git a/scripts/min-tool-version.sh b/scripts/min-tool-version.sh
> index b6593eac5003..8766e248ffbb 100755
> --- a/scripts/min-tool-version.sh
> +++ b/scripts/min-tool-version.sh
> @@ -19,10 +19,6 @@ binutils)
> gcc)
> echo 5.1.0
> ;;
> -icc)
> - # temporary
> - echo 16.0.3
> - ;;
> llvm)
> if [ "$SRCARCH" = s390 ]; then
> echo 14.0.0
> --
> 2.34.1
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH v2] Remove Intel compiler support
2022-10-11 17:14 [PATCH v2] Remove Intel compiler support Masahiro Yamada
2022-10-13 20:02 ` Nathan Chancellor
@ 2022-10-14 14:39 ` Miguel Ojeda
2022-10-16 18:22 ` Masahiro Yamada
1 sibling, 1 reply; 5+ messages in thread
From: Miguel Ojeda @ 2022-10-14 14:39 UTC (permalink / raw)
To: Masahiro Yamada, Nick Terrell
Cc: Linus Torvalds, Thomas Gleixner, Ingo Molnar, Borislav Petkov,
Dave Hansen, x86, H . Peter Anvin, Greg Kroah-Hartman,
linux-ia64, Nick Desaulniers, Miguel Ojeda, Arnd Bergmann,
Len Brown, Nathan Chancellor, Rafael J. Wysocki, Robert Moore,
Tom Rix, devel, linux-acpi, linux-kernel, llvm
On Tue, Oct 11, 2022 at 7:16 PM Masahiro Yamada <masahiroy@kernel.org> wrote:
>
> diff --git a/include/linux/compiler_attributes.h b/include/linux/compiler_attributes.h
> index 898b3458b24a..9221302f6ae8 100644
> --- a/include/linux/compiler_attributes.h
> +++ b/include/linux/compiler_attributes.h
> @@ -64,16 +64,10 @@
> * compiler should see some alignment anyway, when the return value is
> * massaged by 'flags = ptr & 3; ptr &= ~3;').
> *
> - * Optional: not supported by icc
> - *
> * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-assume_005faligned-function-attribute
> * clang: https://clang.llvm.org/docs/AttributeReference.html#assume-aligned
> */
> -#if __has_attribute(__assume_aligned__)
> -# define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
> -#else
> -# define __assume_aligned(a, ...)
> -#endif
> +#define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
Thanks for cleaning the conditional inclusion here. I double-checked
it is indeed available for both GCC and Clang current minimum versions
just in case: https://godbolt.org/z/PxaqeEdcE.
> diff --git a/lib/zstd/common/compiler.h b/lib/zstd/common/compiler.h
> index f5a9c70a228a..c281a6430cd4 100644
> --- a/lib/zstd/common/compiler.h
> +++ b/lib/zstd/common/compiler.h
> @@ -116,7 +116,7 @@
>
> /* vectorization
> * older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax */
> -#if !defined(__INTEL_COMPILER) && !defined(__clang__) && defined(__GNUC__)
> +#if !defined(__clang__) && defined(__GNUC__)
> # if (__GNUC__ == 4 && __GNUC_MINOR__ > 3) || (__GNUC__ >= 5)
> # define DONT_VECTORIZE __attribute__((optimize("no-tree-vectorize")))
> # else
These files come from upstream Zstandard -- should we keep those lines
to minimize divergence?
https://github.com/facebook/zstd/blob/v1.4.10/lib/common/compiler.h#L154.
Commit e0c1b49f5b67 ("lib: zstd: Upgrade to latest upstream zstd
version 1.4.10") is the latest upgrade, and says:
This patch is 100% generated from upstream zstd commit 20821a46f412 [0].
This patch is very large because it is transitioning from the custom
kernel zstd to using upstream directly. The new zstd follows upstreams
file structure which is different. Future update patches will be much
smaller because they will only contain the changes from one upstream
zstd release.
So I think Nick would prefer to keep the changes as minimal as
possible with respect to upstream.
Further reading seems to suggest this is the case, e.g. see this
commit upstream that introduces a space to match the kernel:
https://github.com/facebook/zstd/commit/b53da1f6f499f0d44c5f40795b080d967b24e5fa.
> diff --git a/lib/zstd/compress/zstd_fast.c b/lib/zstd/compress/zstd_fast.c
> index 96b7d48e2868..800f3865119f 100644
> --- a/lib/zstd/compress/zstd_fast.c
> +++ b/lib/zstd/compress/zstd_fast.c
> @@ -80,13 +80,6 @@ ZSTD_compressBlock_fast_generic(
> }
>
> /* Main Search Loop */
> -#ifdef __INTEL_COMPILER
> - /* From intel 'The vector pragma indicates that the loop should be
> - * vectorized if it is legal to do so'. Can be used together with
> - * #pragma ivdep (but have opted to exclude that because intel
> - * warns against using it).*/
> - #pragma vector always
> -#endif
> while (ip1 < ilimit) { /* < instead of <=, because check at ip0+2 */
> size_t mLength;
> BYTE const* ip2 = ip0 + 2;
Ditto: https://github.com/facebook/zstd/blob/v1.4.10/lib/compress/zstd_fast.c#L83.
Apart from the zstd divergence which I am not sure about, everything
looks good to me!
Reviewed-by: Miguel Ojeda <ojeda@kernel.org>
Cheers,
Miguel
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH v2] Remove Intel compiler support
2022-10-14 14:39 ` Miguel Ojeda
@ 2022-10-16 18:22 ` Masahiro Yamada
2022-10-17 20:02 ` Nick Terrell
0 siblings, 1 reply; 5+ messages in thread
From: Masahiro Yamada @ 2022-10-16 18:22 UTC (permalink / raw)
To: Miguel Ojeda
Cc: Nick Terrell, Linus Torvalds, Thomas Gleixner, Ingo Molnar,
Borislav Petkov, Dave Hansen, x86, H . Peter Anvin,
Greg Kroah-Hartman, linux-ia64, Nick Desaulniers, Miguel Ojeda,
Arnd Bergmann, Len Brown, Nathan Chancellor, Rafael J. Wysocki,
Robert Moore, Tom Rix, devel, linux-acpi, linux-kernel, llvm
On Fri, Oct 14, 2022 at 11:40 PM Miguel Ojeda
<miguel.ojeda.sandonis@gmail.com> wrote:
>
> On Tue, Oct 11, 2022 at 7:16 PM Masahiro Yamada <masahiroy@kernel.org> wrote:
> >
> > diff --git a/include/linux/compiler_attributes.h b/include/linux/compiler_attributes.h
> > index 898b3458b24a..9221302f6ae8 100644
> > --- a/include/linux/compiler_attributes.h
> > +++ b/include/linux/compiler_attributes.h
> > @@ -64,16 +64,10 @@
> > * compiler should see some alignment anyway, when the return value is
> > * massaged by 'flags = ptr & 3; ptr &= ~3;').
> > *
> > - * Optional: not supported by icc
> > - *
> > * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-assume_005faligned-function-attribute
> > * clang: https://clang.llvm.org/docs/AttributeReference.html#assume-aligned
> > */
> > -#if __has_attribute(__assume_aligned__)
> > -# define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
> > -#else
> > -# define __assume_aligned(a, ...)
> > -#endif
> > +#define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
>
> Thanks for cleaning the conditional inclusion here. I double-checked
> it is indeed available for both GCC and Clang current minimum versions
> just in case: https://godbolt.org/z/PxaqeEdcE.
>
> > diff --git a/lib/zstd/common/compiler.h b/lib/zstd/common/compiler.h
> > index f5a9c70a228a..c281a6430cd4 100644
> > --- a/lib/zstd/common/compiler.h
> > +++ b/lib/zstd/common/compiler.h
> > @@ -116,7 +116,7 @@
> >
> > /* vectorization
> > * older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax */
> > -#if !defined(__INTEL_COMPILER) && !defined(__clang__) && defined(__GNUC__)
> > +#if !defined(__clang__) && defined(__GNUC__)
> > # if (__GNUC__ == 4 && __GNUC_MINOR__ > 3) || (__GNUC__ >= 5)
> > # define DONT_VECTORIZE __attribute__((optimize("no-tree-vectorize")))
> > # else
>
> These files come from upstream Zstandard -- should we keep those lines
> to minimize divergence?
> https://github.com/facebook/zstd/blob/v1.4.10/lib/common/compiler.h#L154.
>
> Commit e0c1b49f5b67 ("lib: zstd: Upgrade to latest upstream zstd
> version 1.4.10") is the latest upgrade, and says:
>
> This patch is 100% generated from upstream zstd commit 20821a46f412 [0].
>
> This patch is very large because it is transitioning from the custom
> kernel zstd to using upstream directly. The new zstd follows upstreams
> file structure which is different. Future update patches will be much
> smaller because they will only contain the changes from one upstream
> zstd release.
>
> So I think Nick would prefer to keep the changes as minimal as
> possible with respect to upstream.
>
> Further reading seems to suggest this is the case, e.g. see this
> commit upstream that introduces a space to match the kernel:
> https://github.com/facebook/zstd/commit/b53da1f6f499f0d44c5f40795b080d967b24e5fa.
>
> > diff --git a/lib/zstd/compress/zstd_fast.c b/lib/zstd/compress/zstd_fast.c
> > index 96b7d48e2868..800f3865119f 100644
> > --- a/lib/zstd/compress/zstd_fast.c
> > +++ b/lib/zstd/compress/zstd_fast.c
> > @@ -80,13 +80,6 @@ ZSTD_compressBlock_fast_generic(
> > }
> >
> > /* Main Search Loop */
> > -#ifdef __INTEL_COMPILER
> > - /* From intel 'The vector pragma indicates that the loop should be
> > - * vectorized if it is legal to do so'. Can be used together with
> > - * #pragma ivdep (but have opted to exclude that because intel
> > - * warns against using it).*/
> > - #pragma vector always
> > -#endif
> > while (ip1 < ilimit) { /* < instead of <=, because check at ip0+2 */
> > size_t mLength;
> > BYTE const* ip2 = ip0 + 2;
>
> Ditto: https://github.com/facebook/zstd/blob/v1.4.10/lib/compress/zstd_fast.c#L83.
>
> Apart from the zstd divergence which I am not sure about, everything
> looks good to me!
>
> Reviewed-by: Miguel Ojeda <ojeda@kernel.org>
>
> Cheers,
> Miguel
Thanks for your close review.
I will drop zstd changes and send v3.
--
Best Regards
Masahiro Yamada
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH v2] Remove Intel compiler support
2022-10-16 18:22 ` Masahiro Yamada
@ 2022-10-17 20:02 ` Nick Terrell
0 siblings, 0 replies; 5+ messages in thread
From: Nick Terrell @ 2022-10-17 20:02 UTC (permalink / raw)
To: Masahiro Yamada
Cc: Miguel Ojeda, Linus Torvalds, Thomas Gleixner, Ingo Molnar,
Borislav Petkov, Dave Hansen, X86 ML, H . Peter Anvin,
Greg Kroah-Hartman, linux-ia64, Nick Desaulniers, Miguel Ojeda,
Arnd Bergmann, Len Brown, Nathan Chancellor, Rafael J. Wysocki,
Robert Moore, Tom Rix, devel, linux-acpi,
Linux Kernel Mailing List, llvm
> On Oct 16, 2022, at 11:22 AM, Masahiro Yamada <masahiroy@kernel.org> wrote:
>
> !-------------------------------------------------------------------|
> This Message Is From an External Sender
>
> |-------------------------------------------------------------------!
>
> On Fri, Oct 14, 2022 at 11:40 PM Miguel Ojeda
> <miguel.ojeda.sandonis@gmail.com> wrote:
>>
>> On Tue, Oct 11, 2022 at 7:16 PM Masahiro Yamada <masahiroy@kernel.org> wrote:
>>>
>>> diff --git a/include/linux/compiler_attributes.h b/include/linux/compiler_attributes.h
>>> index 898b3458b24a..9221302f6ae8 100644
>>> --- a/include/linux/compiler_attributes.h
>>> +++ b/include/linux/compiler_attributes.h
>>> @@ -64,16 +64,10 @@
>>> * compiler should see some alignment anyway, when the return value is
>>> * massaged by 'flags = ptr & 3; ptr &= ~3;').
>>> *
>>> - * Optional: not supported by icc
>>> - *
>>> * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-assume_005faligned-function-attribute
>>> * clang: https://clang.llvm.org/docs/AttributeReference.html#assume-aligned
>>> */
>>> -#if __has_attribute(__assume_aligned__)
>>> -# define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
>>> -#else
>>> -# define __assume_aligned(a, ...)
>>> -#endif
>>> +#define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
>>
>> Thanks for cleaning the conditional inclusion here. I double-checked
>> it is indeed available for both GCC and Clang current minimum versions
>> just in case: https://godbolt.org/z/PxaqeEdcE.
>>
>>> diff --git a/lib/zstd/common/compiler.h b/lib/zstd/common/compiler.h
>>> index f5a9c70a228a..c281a6430cd4 100644
>>> --- a/lib/zstd/common/compiler.h
>>> +++ b/lib/zstd/common/compiler.h
>>> @@ -116,7 +116,7 @@
>>>
>>> /* vectorization
>>> * older GCC (pre gcc-4.3 picked as the cutoff) uses a different syntax */
>>> -#if !defined(__INTEL_COMPILER) && !defined(__clang__) && defined(__GNUC__)
>>> +#if !defined(__clang__) && defined(__GNUC__)
>>> # if (__GNUC__ == 4 && __GNUC_MINOR__ > 3) || (__GNUC__ >= 5)
>>> # define DONT_VECTORIZE __attribute__((optimize("no-tree-vectorize")))
>>> # else
>>
>> These files come from upstream Zstandard -- should we keep those lines
>> to minimize divergence?
>> https://github.com/facebook/zstd/blob/v1.4.10/lib/common/compiler.h#L154.
>>
>> Commit e0c1b49f5b67 ("lib: zstd: Upgrade to latest upstream zstd
>> version 1.4.10") is the latest upgrade, and says:
>>
>> This patch is 100% generated from upstream zstd commit 20821a46f412 [0].
>>
>> This patch is very large because it is transitioning from the custom
>> kernel zstd to using upstream directly. The new zstd follows upstreams
>> file structure which is different. Future update patches will be much
>> smaller because they will only contain the changes from one upstream
>> zstd release.
>>
>> So I think Nick would prefer to keep the changes as minimal as
>> possible with respect to upstream.
>>
>> Further reading seems to suggest this is the case, e.g. see this
>> commit upstream that introduces a space to match the kernel:
>> https://github.com/facebook/zstd/commit/b53da1f6f499f0d44c5f40795b080d967b24e5fa.
>>
>>> diff --git a/lib/zstd/compress/zstd_fast.c b/lib/zstd/compress/zstd_fast.c
>>> index 96b7d48e2868..800f3865119f 100644
>>> --- a/lib/zstd/compress/zstd_fast.c
>>> +++ b/lib/zstd/compress/zstd_fast.c
>>> @@ -80,13 +80,6 @@ ZSTD_compressBlock_fast_generic(
>>> }
>>>
>>> /* Main Search Loop */
>>> -#ifdef __INTEL_COMPILER
>>> - /* From intel 'The vector pragma indicates that the loop should be
>>> - * vectorized if it is legal to do so'. Can be used together with
>>> - * #pragma ivdep (but have opted to exclude that because intel
>>> - * warns against using it).*/
>>> - #pragma vector always
>>> -#endif
>>> while (ip1 < ilimit) { /* < instead of <=, because check at ip0+2 */
>>> size_t mLength;
>>> BYTE const* ip2 = ip0 + 2;
>>
>> Ditto: https://github.com/facebook/zstd/blob/v1.4.10/lib/compress/zstd_fast.c#L83.
>>
>> Apart from the zstd divergence which I am not sure about, everything
>> looks good to me!
>>
>> Reviewed-by: Miguel Ojeda <ojeda@kernel.org>
>>
>> Cheers,
>> Miguel
>
>
> Thanks for your close review.
>
> I will drop zstd changes and send v3.
Thanks!
As Miguel pointed out, we sync zstd in from upstream, so any changes here would
get overwritten next sync. If the idea is to just remove icc specific code for code
hygiene, I'd leave zstd as-is to keep it more in line with upstream. However, if there
is a very strong reason to delete it, we could remove it as part of our import process.
And if anything is broken, please open an issue on zstd's GitHub or email me.
Best,
Nick Terrell
> --
> Best Regards
> Masahiro Yamada
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2022-10-17 20:02 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-10-11 17:14 [PATCH v2] Remove Intel compiler support Masahiro Yamada
2022-10-13 20:02 ` Nathan Chancellor
2022-10-14 14:39 ` Miguel Ojeda
2022-10-16 18:22 ` Masahiro Yamada
2022-10-17 20:02 ` Nick Terrell
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).