build: check functionality rather than binutils version
PATCH 1/1 build: check functionality rather than binutils version
Rather than checking the binutils version number, which can lead to unnecessary disabling of AVX512 if fixes have been backported to distro versions, we can instead check the output of "as" from binutils to see if it is correct.
These checks use the minimal assembly reproduction code posted to the public bug tracker for gcc/binutils for those issues [1][2]. If the binutils bug is present, the instruction parameters - specifically the displacement parameter - will be different in the disassembled output compared to the input. Therefore each check involves assembling a single instruction and disassembling it again, checking that the two match.
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028
[2] https://sourceware.org/bugzilla/show_bug.cgi?id=23465
--- buildtools/binutils-avx512-check.sh | 26 ++++++++++++++++++++++++++ buildtools/meson.build | 3 +-- config/x86/meson.build | 10 +++------- meson.build | 5 ++++- 4 files changed, 34 insertions(+), 10 deletions(-) create mode 100755 buildtools/binutils-avx512-check.sh
buildtools/binutils-avx512-check.sh new file +26 -0
@@ -0,0 +1,26 @@ +#! /bin/sh +# SPDX-License-Identifier: BSD-3-Clause +# Copyright(c) 2020 Intel Corporation + +AS=${AS:-as} +MESON_BUILD_ROOT=${MESON_BUILD_ROOT:-/tmp} +OBJFILE=$MESON_BUILD_ROOT/binutils-avx512-check.o +# from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028 +GATHER_PARAMS='0x8(,%ymm1,1),%ymm0{%k2}' +# from https://sourceware.org/bugzilla/show_bug.cgi?id=23465 +MOVAPS_PARAMS='0x40(,%rax,1),%zmm0' + +# assemble vmovaps statement to file +echo "vmovaps $MOVAPS_PARAMS" | $AS --64 -o $OBJFILE - +# check the parameters are correct in the output, using objdump +objdump -d --no-show-raw-insn $OBJFILE | grep -q $MOVAPS_PARAMS || { + echo "vmovaps displacement error with as" + exit 1 +} + +# assemble vpgather to file and similarly check +echo "vpgatherqq $GATHER_PARAMS" | $AS --64 -o $OBJFILE - +objdump -d --no-show-raw-insn $OBJFILE | grep -q $GATHER_PARAMS || { + echo "vpgatherqq displacement error with as" + exit 1 +}
buildtools/meson.build +1 -2
@@ -1,13 +1,12 @@ # SPDX-License-Identifier: BSD-3-Clause # Copyright(c) 2017-2019 Intel Corporation -subdir('pmdinfogen') - pkgconf = find_program('pkg-config', 'pkgconf', required: false) pmdinfo = find_program('gen-pmdinfo-cfile.sh') list_dir_globs = find_program('list-dir-globs.py') check_symbols = find_program('check-symbols.sh') ldflags_ibverbs_static = find_program('options-ibverbs-static.sh') +binutils_avx512_check = find_program('binutils-avx512-check.sh') # set up map-to-def script using python, either built-in or external python3 = import('python').find_installation(required: false)
config/x86/meson.build +3 -7
@@ -3,14 +3,10 @@ # get binutils version for the workaround of Bug 97 if not is_windows - ldver = run_command('ld', '-v').stdout().strip() - if ldver.contains('2.30') and cc.has_argument('-mno-avx512f') + as_ok = run_command(binutils_avx512_check) + if as_ok.returncode() != 0 and cc.has_argument('-mno-avx512f') machine_args += '-mno-avx512f' - message('Binutils 2.30 detected, disabling AVX512 support as workaround for bug #97') - endif - if ldver.contains('2.31') and cc.has_argument('-mno-avx512f') - machine_args += '-mno-avx512f' - message('Binutils 2.31 detected, disabling AVX512 support as workaround for bug #249') + message('Binutils error with AVX512 assembly, disabling AVX512 support') endif endif
meson.build +4 -1
@@ -41,10 +41,13 @@ global_inc = include_directories('.', 'config', 'lib/librte_eal/@0@/include'.format(host_machine.system()), 'lib/librte_eal/@0@/include'.format(arch_subdir), ) + +# do configuration and get tool paths +subdir('buildtools') subdir('config') # build libs and drivers -subdir('buildtools') +subdir('buildtools/pmdinfogen') subdir('lib') subdir('drivers')
Signature · 1 line
2.25.1
17/06/2020 12:40, Bruce Richardson:
Quoted 11 lines
Rather than checking the binutils version number, which can lead to unnecessary disabling of AVX512 if fixes have been backported to distro versions, we can instead check the output of "as" from binutils to see if it is correct.
These checks use the minimal assembly reproduction code posted to the public bug tracker for gcc/binutils for those issues [1][2]. If the binutils bug is present, the instruction parameters - specifically the displacement parameter - will be different in the disassembled output compared to the input. Therefore each check involves assembling a single instruction and disassembling it again, checking that the two match.
Whaoh, that's a very specific optimization :) Which distro is affected (backported binutils fix)?
buildtools/meson.build line 4
Quoted 8 lines
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028
[2] https://sourceware.org/bugzilla/show_bug.cgi?id=23465Signed-off-by Bruce Richardson <[email protected]>---
buildtools/meson.build +0
-1-subdir('pmdinfogen')
Why removing pmdinfogen?
meson.build global_inc = include_directories('.', 'config', · line 50
Quoted 9 lines
meson.build +3
-1+# do configuration and get tool paths +subdir('buildtools') subdir('config') # build libs and drivers -subdir('buildtools') +subdir('buildtools/pmdinfogen')
Is it related to binutils check?
[...]
meson.build
Quoted 4 lines
meson.build +2
-2- ldver = run_command('ld', '-v').stdout().strip() - if ldver.contains('2.30') and cc.has_argument('-mno-avx512f') + as_ok = run_command(binutils_avx512_check) + if as_ok.returncode() != 0 and cc.has_argument('-mno-avx512f')
"as_ok" is difficult to understand.
I would suggest "binutils_avx512_ok".
meson.build
Quoted 1 line
meson.build +1
-0+ message('Binutils error with AVX512 assembly, disabling AVX512 support')
It looks like something which should be part of meson itself.
Quoted 13 lines · Thomas Monjalon
17/06/2020 12:40, Bruce Richardson:
Quoted 11 lines
Rather than checking the binutils version number, which can lead to unnecessary disabling of AVX512 if fixes have been backported to distro versions, we can instead check the output of "as" from binutils to see if it is correct.
These checks use the minimal assembly reproduction code posted to the public bug tracker for gcc/binutils for those issues [1][2]. If the binutils bug is present, the instruction parameters - specifically the displacement parameter - will be different in the disassembled output compared to the input. Therefore each check involves assembling a single instruction and disassembling it again, checking that the two match.
Quoted 3 lines
Whaoh, that's a very specific optimization :) Which distro is affected (backported binutils fix)?
Not an optimization, it's a bug, which necessitates disabling a whole instruction set, so we should really check for the bug rather than assuming all versions are affected. From [2] below, we have confirmation that the fix has been backported to both 2.30 and 2.31 binutils branches.
Quoted 10 lines
buildtools/meson.build line 4
Quoted 8 lines
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028
[2] https://sourceware.org/bugzilla/show_bug.cgi?id=23465Signed-off-by Bruce Richardson <[email protected]>---
buildtools/meson.build +0
-1-subdir('pmdinfogen')Why removing pmdinfogen?
The pmdinfogen requires values that are set in the config folder, but the config folder processing now uses scripts that are in the buildtools folder. While we could refer to them using relative paths, I felt it better to separate out the compilation of pmdinfogen to the rest of the buildutils processing, which is simply assigning variables to scripts to make them easier to call later in the processing.
Quoted 7 lines
Quoted 7 lines
meson.build +2
-0+# do configuration and get tool paths +subdir('buildtools') subdir('config')
Quoted 6 lines
meson.build global_inc = include_directories('.', 'config', · line 50
Quoted 3 lines
# build libs and driversmeson.build +1
-1-subdir('buildtools') +subdir('buildtools/pmdinfogen')Is it related to binutils check?
Pretty much. It just makes it easier to have the config processing use scripts in the buildtools folder.
Quoted 9 lines
[...]
meson.build
Quoted 4 lines
meson.build +2
-2- ldver = run_command('ld', '-v').stdout().strip() - if ldver.contains('2.30') and cc.has_argument('-mno-avx512f') + as_ok = run_command(binutils_avx512_check) + if as_ok.returncode() != 0 and cc.has_argument('-mno-avx512f')"
as_ok" is difficult to understand.
I would suggest "binutils_avx512_ok".
Sure, can change for v2 - though I'd suggest the slightly shorter binutils_ok.
Quoted 4 lines
meson.build
Quoted 1 line
meson.build +1
-0+ message('Binutils error with AVX512 assembly, disabling AVX512 support')It looks like something which should be part of meson itself.
That would certainly be nice to have, but I consider it infeasible, since there will be, even in the best case, a lag between a bug being discovered and workarounds making its way into tools. Even then you still have an update problem to work around it - update binutils and bug goes away, or update meson and you can work around it. In both cases you are requiring the user to update something, so fixing in DPDK is best.
/Bruce
PATCH v2 1/1 build: check functionality rather than binutils version
In reply to Bruce RichardsonRather than checking the binutils version number, which can lead to unnecessary disabling of AVX512 if fixes have been backported to distro versions, we can instead check the output of "as" from binutils to see if it is correct.
The check in the script uses the minimal assembly reproduction code posted to the public bug tracker for gcc/binutils for those issues [1]. If the binutils bug is present, the instruction parameters - specifically the displacement parameter - will be different in the disassembled output compared to the input. Therefore the check involves assembling a single instruction and disassembling it again, checking that the two match.
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028
---
V2:
- Renamed "as_ok" variable to "binutils_ok" for readability
- Removed one test case from the script because even though two DPDK bugs were filed, the one binutils bug is the root cause, and a single commit fixes them both
- Changed message() to warning() in the printout
--- buildtools/binutils-avx512-check.sh | 16 ++++++++++++++++ buildtools/meson.build | 3 +-- config/x86/meson.build | 10 +++------- meson.build | 5 ++++- 4 files changed, 24 insertions(+), 10 deletions(-) create mode 100755 buildtools/binutils-avx512-check.sh
buildtools/binutils-avx512-check.sh new file +16 -0
@@ -0,0 +1,16 @@ +#! /bin/sh +# SPDX-License-Identifier: BSD-3-Clause +# Copyright(c) 2020 Intel Corporation + +AS=${AS:-as} +MESON_BUILD_ROOT=${MESON_BUILD_ROOT:-/tmp} +OBJFILE=$MESON_BUILD_ROOT/binutils-avx512-check.o +# from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028 +GATHER_PARAMS='0x8(,%ymm1,1),%ymm0{%k2}' + +# assemble vpgather to file and similarly check +echo "vpgatherqq $GATHER_PARAMS" | $AS --64 -o $OBJFILE - +objdump -d --no-show-raw-insn $OBJFILE | grep -q $GATHER_PARAMS || { + echo "vpgatherqq displacement error with as" + exit 1 +}
buildtools/meson.build +1 -2
@@ -1,13 +1,12 @@ # SPDX-License-Identifier: BSD-3-Clause # Copyright(c) 2017-2019 Intel Corporation -subdir('pmdinfogen') - pkgconf = find_program('pkg-config', 'pkgconf', required: false) pmdinfo = find_program('gen-pmdinfo-cfile.sh') list_dir_globs = find_program('list-dir-globs.py') check_symbols = find_program('check-symbols.sh') ldflags_ibverbs_static = find_program('options-ibverbs-static.sh') +binutils_avx512_check = find_program('binutils-avx512-check.sh') # set up map-to-def script using python, either built-in or external python3 = import('python').find_installation(required: false)
config/x86/meson.build +3 -7
@@ -3,14 +3,10 @@ # get binutils version for the workaround of Bug 97 if not is_windows - ldver = run_command('ld', '-v').stdout().strip() - if ldver.contains('2.30') and cc.has_argument('-mno-avx512f') + binutils_ok = run_command(binutils_avx512_check) + if binutils_ok.returncode() != 0 and cc.has_argument('-mno-avx512f') machine_args += '-mno-avx512f' - message('Binutils 2.30 detected, disabling AVX512 support as workaround for bug #97') - endif - if ldver.contains('2.31') and cc.has_argument('-mno-avx512f') - machine_args += '-mno-avx512f' - message('Binutils 2.31 detected, disabling AVX512 support as workaround for bug #249') + warning('Binutils error with AVX512 assembly, disabling AVX512 support') endif endif
meson.build +4 -1
@@ -41,10 +41,13 @@ global_inc = include_directories('.', 'config', 'lib/librte_eal/@0@/include'.format(host_machine.system()), 'lib/librte_eal/@0@/include'.format(arch_subdir), ) + +# do configuration and get tool paths +subdir('buildtools') subdir('config') # build libs and drivers -subdir('buildtools') +subdir('buildtools/pmdinfogen') subdir('lib') subdir('drivers')
Signature · 1 line
2.25.1
Quoted 21 lines
-----Original Message-----
From: dev <[email protected]> On Behalf Of Bruce Richardson
Sent: Thursday, June 18, 2020 12:57 PM
To: [email protected]
Cc: Yigit, Ferruh <[email protected]>; [email protected]; Richardson, Bruce <[email protected]>
Subject: [dpdk-dev] [PATCH v2] build: check functionality rather than binutils versionRather than checking the binutils version number, which can lead to unnecessary disabling of AVX512 if fixes have been backported to distro versions, we can instead check the output of "as" from binutils to see if it is correct.
The check in the script uses the minimal assembly reproduction code posted to the public bug tracker for gcc/binutils for those issues [1]. If the binutils bug is present, the instruction parameters - specifically the displacement parameter - will be different in the disassembled output compared to the input. Therefore the check involves assembling a single instruction and disassembling it again, checking that the two match.
Quoted 3 lines
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028
Signed-off-by Bruce Richardson <[email protected]>
Tested on binutils 2.30 without backported fixes, can confirm that __AVX512F__ define is not present at meson configure time.
18/06/2020 13:56, Bruce Richardson:
Quoted 11 lines
Rather than checking the binutils version number, which can lead to unnecessary disabling of AVX512 if fixes have been backported to distro versions, we can instead check the output of "as" from binutils to see if it is correct.
The check in the script uses the minimal assembly reproduction code posted to the public bug tracker for gcc/binutils for those issues [1]. If the binutils bug is present, the instruction parameters - specifically the displacement parameter - will be different in the disassembled output compared to the input. Therefore the check involves assembling a single instruction and disassembling it again, checking that the two match.
[...]
buildtools/binutils-avx512-check.sh line 16
Quoted 13 lines
buildtools/binutils-avx512-check.sh new file +11
-0+MESON_BUILD_ROOT=${MESON_BUILD_ROOT:-/tmp} +OBJFILE=$MESON_BUILD_ROOT/binutils-avx512-check.o +# from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028 +GATHER_PARAMS='0x8(,%ymm1,1),%ymm0{%k2}' + +# assemble vpgather to file and similarly check +echo "vpgatherqq $GATHER_PARAMS" | $AS --64 -o $OBJFILE - +objdump -d --no-show-raw-insn $OBJFILE | grep -q $GATHER_PARAMS || { + echo "vpgatherqq displacement error with as" + exit 1 +}
For the temporary OBJFILE, please use mktemp and trap for cleanup. If you grep "mktemp" in DPDK, you will see the filename is pretty well standardized with "dpdk." as prefix.
Thanks
Quoted 23 lines · Thomas Monjalon
18/06/2020 13:56, Bruce Richardson:
Quoted 11 lines
Rather than checking the binutils version number, which can lead to unnecessary disabling of AVX512 if fixes have been backported to distro versions, we can instead check the output of "as" from binutils to see if it is correct.
The check in the script uses the minimal assembly reproduction code posted to the public bug tracker for gcc/binutils for those issues [1]. If the binutils bug is present, the instruction parameters - specifically the displacement parameter - will be different in the disassembled output compared to the input. Therefore the check involves assembling a single instruction and disassembling it again, checking that the two match.
[...]
Quoted 10 lines
buildtools/binutils-avx512-check.sh new file +8
-0+MESON_BUILD_ROOT=${MESON_BUILD_ROOT:-/tmp} +OBJFILE=$MESON_BUILD_ROOT/binutils-avx512-check.o +# from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028 +GATHER_PARAMS='0x8(,%ymm1,1),%ymm0{%k2}' + +# assemble vpgather to file and similarly check +echo "vpgatherqq $GATHER_PARAMS" | $AS --64 -o $OBJFILE - +objdump -d --no-show-raw-insn $OBJFILE | grep -q $GATHER_PARAMS || {
Quoted 9 lines
buildtools/binutils-avx512-check.sh line 16
Quoted 3 lines
buildtools/binutils-avx512-check.sh new file +3
-0+ echo "vpgatherqq displacement error with as" + exit 1 +}For the temporary OBJFILE, please use mktemp and trap for cleanup. If you grep "mktemp" in DPDK, you will see the filename is pretty well standardized with "dpdk." as prefix.
Thanks
Will change for v3
PATCH v3 1/1 build: check functionality rather than binutils version
In reply to Bruce RichardsonRather than checking the binutils version number, which can lead to unnecessary disabling of AVX512 if fixes have been backported to distro versions, we can instead check the output of "as" from binutils to see if it is correct.
The check in the script uses the minimal assembly reproduction code posted to the public bug tracker for gcc/binutils for those issues [1]. If the binutils bug is present, the instruction parameters - specifically the displacement parameter - will be different in the disassembled output compared to the input. Therefore the check involves assembling a single instruction and disassembling it again, checking that the two match.
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028
---
V3:
- Use mktemp to create a temporary file rather than place object file in build folder. Use trap to ensure temp file deletion on exit.
V2:
- Renamed "as_ok" variable to "binutils_ok" for readability
- Removed one test case from the script because even though two DPDK bugs were filed, the one binutils bug is the root cause, and a single commit fixes them both
- Changed message() to warning() in the printout
--- buildtools/binutils-avx512-check.sh | 16 ++++++++++++++++ buildtools/meson.build | 3 +-- config/x86/meson.build | 10 +++------- meson.build | 5 ++++- 4 files changed, 24 insertions(+), 10 deletions(-) create mode 100755 buildtools/binutils-avx512-check.sh
buildtools/binutils-avx512-check.sh new file +16 -0
@@ -0,0 +1,16 @@ +#! /bin/sh +# SPDX-License-Identifier: BSD-3-Clause +# Copyright(c) 2020 Intel Corporation + +AS=${AS:-as} +OBJFILE=$(mktemp -t dpdk.binutils-check.XXXXXX.o) +trap 'rm -f "$OBJFILE"' EXIT +# from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028 +GATHER_PARAMS='0x8(,%ymm1,1),%ymm0{%k2}' + +# assemble vpgather to file and similarly check +echo "vpgatherqq $GATHER_PARAMS" | $AS --64 -o $OBJFILE - +objdump -d --no-show-raw-insn $OBJFILE | grep -q $GATHER_PARAMS || { + echo "vpgatherqq displacement error with as" + exit 1 +}
buildtools/meson.build +1 -2
@@ -1,13 +1,12 @@ # SPDX-License-Identifier: BSD-3-Clause # Copyright(c) 2017-2019 Intel Corporation -subdir('pmdinfogen') - pkgconf = find_program('pkg-config', 'pkgconf', required: false) pmdinfo = find_program('gen-pmdinfo-cfile.sh') list_dir_globs = find_program('list-dir-globs.py') check_symbols = find_program('check-symbols.sh') ldflags_ibverbs_static = find_program('options-ibverbs-static.sh') +binutils_avx512_check = find_program('binutils-avx512-check.sh') # set up map-to-win script using python, either built-in or external python3 = import('python').find_installation(required: false)
config/x86/meson.build +3 -7
@@ -3,14 +3,10 @@ # get binutils version for the workaround of Bug 97 if not is_windows - ldver = run_command('ld', '-v').stdout().strip() - if ldver.contains('2.30') and cc.has_argument('-mno-avx512f') + binutils_ok = run_command(binutils_avx512_check) + if binutils_ok.returncode() != 0 and cc.has_argument('-mno-avx512f') machine_args += '-mno-avx512f' - message('Binutils 2.30 detected, disabling AVX512 support as workaround for bug #97') - endif - if ldver.contains('2.31') and cc.has_argument('-mno-avx512f') - machine_args += '-mno-avx512f' - message('Binutils 2.31 detected, disabling AVX512 support as workaround for bug #249') + warning('Binutils error with AVX512 assembly, disabling AVX512 support') endif endif
meson.build +4 -1
@@ -40,10 +40,13 @@ global_inc = include_directories('.', 'config', 'lib/librte_eal/@0@/include'.format(host_machine.system()), 'lib/librte_eal/@0@/include'.format(arch_subdir), ) + +# do configuration and get tool paths +subdir('buildtools') subdir('config') # build libs and drivers -subdir('buildtools') +subdir('buildtools/pmdinfogen') subdir('lib') subdir('drivers')
Signature · 1 line
2.25.1
Thomas Monjalon
Applied04/07/2020 13:48, Bruce Richardson:
Quoted 22 lines
Rather than checking the binutils version number, which can lead to unnecessary disabling of AVX512 if fixes have been backported to distro versions, we can instead check the output of "as" from binutils to see if it is correct.
The check in the script uses the minimal assembly reproduction code posted to the public bug tracker for gcc/binutils for those issues [1]. If the binutils bug is present, the instruction parameters - specifically the displacement parameter - will be different in the disassembled output compared to the input. Therefore the check involves assembling a single instruction and disassembling it again, checking that the two match.
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90028
Signed-off-by Bruce Richardson <[email protected]>Tested-by Harry van Haaren <[email protected]>---
V3:
- Use mktemp to create a temporary file rather than place object file in build folder. Use trap to ensure temp file deletion on exit.
Quoted 6 lines
V2:
- Renamed "as_ok" variable to "binutils_ok" for readability
- Removed one test case from the script because even though two DPDK bugs were filed, the one binutils bug is the root cause, and a single commit fixes them both
- Changedmessage()towarning()in the printout
Applied, thanks