scripts: handle BrokenPipeError for python scripts
Subsystems (2)
PATCH 1/1 scripts: handle BrokenPipeError for python scripts
In the follow-up of commit fb3041d61f68 ("kbuild: fix SIGPIPE error message for AR=gcc-ar and AR=llvm-ar"), Kees Cook pointed out that tools should _not_ catch their own SIGPIPEs [1] [2].
Based on his feedback, LLVM was fixed [3].
However, Python's default behavior is to show noisy bracktrace when SIGPIPE is sent. So, scripts written in Python are basically in the same situation as the buggy llvm tools.
Example:
$ make -s allnoconfig
$ make -s allmodconfig
$ scripts/diffconfig .config.old .config | head -n1
-ALIX n
Traceback (most recent call last):
File "/home/masahiro/linux/scripts/diffconfig", line 132, in <module>
main()
File "/home/masahiro/linux/scripts/diffconfig", line 130, in main
print_config("+", config, None, b[config])
File "/home/masahiro/linux/scripts/diffconfig", line 64, in print_config
print("+%s %s" % (config, new_value))
BrokenPipeError: [Errno 32] Broken pipePython documentatin [4] notes how to make scripts die immediately and silently:
"""
Piping output of your program to tools like head(1) will cause a SIGPIPE signal to be sent to your process when the receiver of its standard output closes early. This results in an exception like
BrokenPipeError: [Errno 32] Broken pipe. To handle this case, wrap your entry point to catch this exception as follows:
import os
import sys
def main():
try:
# simulate large output (your code replaces this loop) for x in range(10000):
print("y")
# flush output here to force SIGPIPE to be triggered
# while inside this try block.
sys.stdout.flush()
except BrokenPipeError:
# Python flushes standard streams on exit; redirect remaining output # to devnull to avoid another BrokenPipeError at shutdown devnull = os.open(os.devnull, os.O_WRONLY) os.dup2(devnull, sys.stdout.fileno())
sys.exit(1) # Python exits with error code 1 on EPIPE
if __name__ == '__main__':
main()
Do not set SIGPIPE’s disposition to SIG_DFL in order to avoid BrokenPipeError. Doing that would cause your program to exit unexpectedly whenever any socket connection is interrupted while your program is still writing to it.
"""
Currently, tools/perf/scripts/python/intel-pt-events.py seems the only script that fixes the issue that way.
tools/perf/scripts/python/compaction-times.py uses another approach signal.signal(signal.SIGPIPE, signal.SIG_DFL) but the Python documentation clearly says "Don't do it".
I cannot fix all Python scripts since there are so many. I fixed some in the scripts/ directory.
[1]: https://lore.kernel.org/all/202211161056.1B9611A@keescook/
[2]: https://github.com/llvm/llvm-project/issues/59037
[3]: https://github.com/llvm/llvm-project/commit/4787efa38066adb51e2c049499d25b3610c0877b
[4]: https://docs.python.org/3/library/signal.html#note-on-sigpipe
---
scripts/checkkconfigsymbols.py | 13 ++++++++++++- scripts/clang-tools/run-clang-tools.py | 21 ++++++++++++++------- scripts/diffconfig | 16 ++++++++++++++-- 3 files changed, 40 insertions(+), 10 deletions(-)
scripts/checkkconfigsymbols.py +12 -1
@@ -115,7 +115,7 @@ def parse_options(): return args -def main(): +def print_undefined_symbols(): """Main function of this module.""" args = parse_options()
@@ -467,5 +467,16 @@ def parse_kconfig_file(kfile): return defined, references +def main(): + try: + print_undefined_symbols() + except BrokenPipeError: + # Python flushes standard streams on exit; redirect remaining output + # to devnull to avoid another BrokenPipeError at shutdown + devnull = os.open(os.devnull, os.O_WRONLY) + os.dup2(devnull, sys.stdout.fileno()) + sys.exit(1) # Python exits with error code 1 on EPIPE + + if __name__ == "__main__": main()
scripts/clang-tools/run-clang-tools.py +14 -7
@@ -61,14 +61,21 @@ def run_analysis(entry): def main(): - args = parse_arguments() + try: + args = parse_arguments() - lock = multiprocessing.Lock() - pool = multiprocessing.Pool(initializer=init, initargs=(lock, args)) - # Read JSON data into the datastore variable - with open(args.path, "r") as f: - datastore = json.load(f) - pool.map(run_analysis, datastore) + lock = multiprocessing.Lock() + pool = multiprocessing.Pool(initializer=init, initargs=(lock, args)) + # Read JSON data into the datastore variable + with open(args.path, "r") as f: + datastore = json.load(f) + pool.map(run_analysis, datastore) + except BrokenPipeError: + # Python flushes standard streams on exit; redirect remaining output + # to devnull to avoid another BrokenPipeError at shutdown + devnull = os.open(os.devnull, os.O_WRONLY) + os.dup2(devnull, sys.stdout.fileno()) + sys.exit(1) # Python exits with error code 1 on EPIPE if __name__ == "__main__":
scripts/diffconfig +14 -2
@@ -65,7 +65,7 @@ def print_config(op, config, value, new_value): else: print(" %s %s -> %s" % (config, value, new_value)) -def main(): +def show_diff(): global merge_style # parse command line args
@@ -129,4 +129,16 @@ def main(): for config in new: print_config("+", config, None, b[config]) -main() +def main(): + try: + show_diff() + except BrokenPipeError: + # Python flushes standard streams on exit; redirect remaining output + # to devnull to avoid another BrokenPipeError at shutdown + devnull = os.open(os.devnull, os.O_WRONLY) + os.dup2(devnull, sys.stdout.fileno()) + sys.exit(1) # Python exits with error code 1 on EPIPE + + +if __name__ == '__main__': + main()
Signature · 1 line
2.34.1
Quoted 26 lines · Masahiro Yamada
In the follow-up of commit
fb3041d61f68("kbuild: fix SIGPIPE error message for AR=gcc-ar and AR=llvm-ar"), Kees Cook pointed out that tools should _not_ catch their own SIGPIPEs [1] [2].Based on his feedback, LLVM was fixed [3].
However, Python's default behavior is to show noisy bracktrace when SIGPIPE is sent. So, scripts written in Python are basically in the same situation as the buggy llvm tools.
Example:
$ make -s allnoconfig $ make -s allmodconfig $ scripts/diffconfig .config.old .config | head -n1 -ALIX n Traceback (most recent call last): File "/home/masahiro/linux/scripts/diffconfig", line 132, in <module> main() File "/home/masahiro/linux/scripts/diffconfig", line 130, in main print_config("+", config, None, b[config]) File "/home/masahiro/linux/scripts/diffconfig", line 64, in print_config print("+%s %s" % (config, new_value)) BrokenPipeError: [Errno 32] Broken pipe
Quoted 1 line
Python documentatin [4] notes how to make scripts die immediately and
typo: s/documentatin/documentation/
Quoted 39 lines
silently:
"""
Piping output of your program to tools like head(1) will cause a SIGPIPE signal to be sent to your process when the receiver of its standard output closes early. This results in an exception like
BrokenPipeError: [Errno 32] Broken pipe. To handle this case, wrap your entry point to catch this exception as follows:import os
import sysdef
main():
try:
# simulate large output (your code replaces this loop) for x in range(10000):
print("y")
# flush output here to force SIGPIPE to be triggered
# while inside this try block.
sys.stdout.flush()
except BrokenPipeError:
# Python flushes standard streams on exit; redirect remaining output # to devnull to avoid another BrokenPipeError at shutdown devnull =os.open(os.devnull, os.O_WRONLY)os.dup2(devnull, sys.stdout.fileno())
sys.exit(1)# Python exits with error code 1 on EPIPEif
__name__== '__main__':
main()Do not set SIGPIPE’s disposition to
SIG_DFLin order to avoid BrokenPipeError. Doing that would cause your program to exit unexpectedly whenever any socket connection is interrupted while your program is still writing to it.
"""Currently,
tools/perf/scripts/python/intel-pt-events.pyseems the only script that fixes the issue that way.
Quoted 6 lines
tools/perf/scripts/python/compaction-times.pyuses another approachsignal.signal(signal.SIGPIPE, signal.SIG_DFL)but the Python documentation clearly says "Don't do it".I cannot fix all Python scripts since there are so many. I fixed some in the
scripts/directory.
That's ok; "Rome wasn't built in a day." This is a good start! Thank you for the patch!
Quoted 112 lines
[1]: https://lore.kernel.org/all/202211161056.1B9611A@keescook/
[2]: https://github.com/llvm/llvm-project/issues/59037
[3]: https://github.com/llvm/llvm-project/commit/4787efa38066adb51e2c049499d25b3610c0877b
[4]: https://docs.python.org/3/library/signal.html#note-on-sigpipeCc Kees Cook <[email protected]>Signed-off-by Masahiro Yamada <[email protected]>---
scripts/checkkconfigsymbols.py | 13 ++++++++++++- scripts/clang-tools/run-clang-tools.py | 21 ++++++++++++++------- scripts/diffconfig | 16 ++++++++++++++-- 3 files changed, 40 insertions(+), 10 deletions(-)scripts/checkkconfigsymbols.py +12
-1@@ -115,7 +115,7 @@ def parse_options(): return args -def main(): +def print_undefined_symbols(): """Main function of this module.""" args = parse_options()@@ -467,5 +467,16 @@ def parse_kconfig_file(kfile): return defined, references +def main(): + try: + print_undefined_symbols() + except BrokenPipeError: + # Python flushes standard streams on exit; redirect remaining output + # to devnull to avoid another BrokenPipeError at shutdown + devnull = os.open(os.devnull, os.O_WRONLY) + os.dup2(devnull, sys.stdout.fileno()) + sys.exit(1) # Python exits with error code 1 on EPIPE + + if __name__ == "__main__": main()scripts/clang-tools/run-clang-tools.py +14
-7@@ -61,14 +61,21 @@ def run_analysis(entry): def main(): - args = parse_arguments() + try: + args = parse_arguments() - lock = multiprocessing.Lock() - pool = multiprocessing.Pool(initializer=init, initargs=(lock, args)) - # Read JSON data into the datastore variable - with open(args.path, "r") as f: - datastore = json.load(f) - pool.map(run_analysis, datastore) + lock = multiprocessing.Lock() + pool = multiprocessing.Pool(initializer=init, initargs=(lock, args)) + # Read JSON data into the datastore variable + with open(args.path, "r") as f: + datastore = json.load(f) + pool.map(run_analysis, datastore) + except BrokenPipeError: + # Python flushes standard streams on exit; redirect remaining output + # to devnull to avoid another BrokenPipeError at shutdown + devnull = os.open(os.devnull, os.O_WRONLY) + os.dup2(devnull, sys.stdout.fileno()) + sys.exit(1) # Python exits with error code 1 on EPIPE if __name__ == "__main__":scripts/diffconfig +14
-3@@ -65,7 +65,7 @@ def print_config(op, config, value, new_value): else: print(" %s %s -> %s" % (config, value, new_value)) -def main(): +def show_diff(): global merge_style # parse command line args@@ -129,4 +129,16 @@ def main(): for config in new: print_config("+", config, None, b[config]) -main() +def main(): + try: + show_diff() + except BrokenPipeError: + # Python flushes standard streams on exit; redirect remaining output + # to devnull to avoid another BrokenPipeError at shutdown + devnull = os.open(os.devnull, os.O_WRONLY) + os.dup2(devnull, sys.stdout.fileno()) + sys.exit(1) # Python exits with error code 1 on EPIPE + + +if __name__ == '__main__': + main() --2.34.1
Signature · 2 lines
Thanks, ~Nick Desaulniers
Nicolas Schier
Reviewed-byQuoted 27 lines · On Thu, Jan 12, 2023 at 10:55:45AM -0800 Nick Desaulniers wrote:
Quoted 26 lines · Masahiro Yamada
In the follow-up of commit
fb3041d61f68("kbuild: fix SIGPIPE error message for AR=gcc-ar and AR=llvm-ar"), Kees Cook pointed out that tools should _not_ catch their own SIGPIPEs [1] [2].Based on his feedback, LLVM was fixed [3].
However, Python's default behavior is to show noisy bracktrace when SIGPIPE is sent. So, scripts written in Python are basically in the same situation as the buggy llvm tools.
Example:
$ make -s allnoconfig $ make -s allmodconfig $ scripts/diffconfig .config.old .config | head -n1 -ALIX n Traceback (most recent call last): File "/home/masahiro/linux/scripts/diffconfig", line 132, in <module> main() File "/home/masahiro/linux/scripts/diffconfig", line 130, in main print_config("+", config, None, b[config]) File "/home/masahiro/linux/scripts/diffconfig", line 64, in print_config print("+%s %s" % (config, new_value)) BrokenPipeError: [Errno 32] Broken pipe
Quoted 41 lines
Quoted 1 line
Python documentatin [4] notes how to make scripts die immediately and
typo: s/documentatin/documentation/
Quoted 37 lines
silently:
"""
Piping output of your program to tools like head(1) will cause a SIGPIPE signal to be sent to your process when the receiver of its standard output closes early. This results in an exception like
BrokenPipeError: [Errno 32] Broken pipe. To handle this case, wrap your entry point to catch this exception as follows:import os
import sysdef
main():
try:
# simulate large output (your code replaces this loop) for x in range(10000):
print("y")
# flush output here to force SIGPIPE to be triggered
# while inside this try block.
sys.stdout.flush()
except BrokenPipeError:
# Python flushes standard streams on exit; redirect remaining output # to devnull to avoid another BrokenPipeError at shutdown devnull =os.open(os.devnull, os.O_WRONLY)os.dup2(devnull, sys.stdout.fileno())
sys.exit(1)# Python exits with error code 1 on EPIPEif
__name__== '__main__':
main()Do not set SIGPIPE’s disposition to
SIG_DFLin order to avoid BrokenPipeError. Doing that would cause your program to exit unexpectedly whenever any socket connection is interrupted while your program is still writing to it.
"""Currently,
tools/perf/scripts/python/intel-pt-events.pyseems the
Hi Masahiro,
should it be "... seems to be the ..."?
Quoted 130 lines
Quoted 8 lines
only script that fixes the issue that way.
tools/perf/scripts/python/compaction-times.pyuses another approachsignal.signal(signal.SIGPIPE, signal.SIG_DFL)but the Python documentation clearly says "Don't do it".I cannot fix all Python scripts since there are so many. I fixed some in the
scripts/directory.That's ok; "Rome wasn't built in a day." This is a good start! Thank you for the patch!
Reviewed-by Nick Desaulniers <[email protected]>Quoted 112 lines
[1]: https://lore.kernel.org/all/202211161056.1B9611A@keescook/
[2]: https://github.com/llvm/llvm-project/issues/59037
[3]: https://github.com/llvm/llvm-project/commit/4787efa38066adb51e2c049499d25b3610c0877b
[4]: https://docs.python.org/3/library/signal.html#note-on-sigpipeCc Kees Cook <[email protected]>Signed-off-by Masahiro Yamada <[email protected]>---
scripts/checkkconfigsymbols.py | 13 ++++++++++++- scripts/clang-tools/run-clang-tools.py | 21 ++++++++++++++------- scripts/diffconfig | 16 ++++++++++++++-- 3 files changed, 40 insertions(+), 10 deletions(-)scripts/checkkconfigsymbols.py +12
-1@@ -115,7 +115,7 @@ def parse_options(): return args -def main(): +def print_undefined_symbols(): """Main function of this module.""" args = parse_options()@@ -467,5 +467,16 @@ def parse_kconfig_file(kfile): return defined, references +def main(): + try: + print_undefined_symbols() + except BrokenPipeError: + # Python flushes standard streams on exit; redirect remaining output + # to devnull to avoid another BrokenPipeError at shutdown + devnull = os.open(os.devnull, os.O_WRONLY) + os.dup2(devnull, sys.stdout.fileno()) + sys.exit(1) # Python exits with error code 1 on EPIPE + + if __name__ == "__main__": main()scripts/clang-tools/run-clang-tools.py +14
-7@@ -61,14 +61,21 @@ def run_analysis(entry): def main(): - args = parse_arguments() + try: + args = parse_arguments() - lock = multiprocessing.Lock() - pool = multiprocessing.Pool(initializer=init, initargs=(lock, args)) - # Read JSON data into the datastore variable - with open(args.path, "r") as f: - datastore = json.load(f) - pool.map(run_analysis, datastore) + lock = multiprocessing.Lock() + pool = multiprocessing.Pool(initializer=init, initargs=(lock, args)) + # Read JSON data into the datastore variable + with open(args.path, "r") as f: + datastore = json.load(f) + pool.map(run_analysis, datastore) + except BrokenPipeError: + # Python flushes standard streams on exit; redirect remaining output + # to devnull to avoid another BrokenPipeError at shutdown + devnull = os.open(os.devnull, os.O_WRONLY) + os.dup2(devnull, sys.stdout.fileno()) + sys.exit(1) # Python exits with error code 1 on EPIPE if __name__ == "__main__":scripts/diffconfig +14
-3@@ -65,7 +65,7 @@ def print_config(op, config, value, new_value): else: print(" %s %s -> %s" % (config, value, new_value)) -def main(): +def show_diff(): global merge_style # parse command line args@@ -129,4 +129,16 @@ def main(): for config in new: print_config("+", config, None, b[config]) -main() +def main(): + try: + show_diff() + except BrokenPipeError: + # Python flushes standard streams on exit; redirect remaining output + # to devnull to avoid another BrokenPipeError at shutdown + devnull = os.open(os.devnull, os.O_WRONLY) + os.dup2(devnull, sys.stdout.fileno()) + sys.exit(1) # Python exits with error code 1 on EPIPE + + +if __name__ == '__main__': + main() --2.34.1
Signature · 2 lines
Thanks, ~Nick Desaulniers
Quoted 6 lines · Masahiro Yamada
def
main():
try:
# simulate large output (your code replaces this loop) for x in range(10000):
print("y")
# flush output here to force SIGPIPE to be triggered
Quoted 8 lines
# while inside this try block.
sys.stdout.flush()
except BrokenPipeError:
# Python flushes standard streams on exit; redirect remaining output # to devnull to avoid another BrokenPipeError at shutdown devnull =os.open(os.devnull, os.O_WRONLY)os.dup2(devnull, sys.stdout.fileno())
sys.exit(1)# Python exits with error code 1 on EPIPE
I still think this is wrong -- they should not continue piping, and should just die with SIGPIPE. It should simply be:
signal(SIGPIPE, SIG_DFL);
Nothing else needed. No wasted CPU cycles, shell handling continues as per normal.
Quoted 7 lines
if __name__ == '__main__': main()Do not set SIGPIPE’s disposition to
SIG_DFLin order to avoid BrokenPipeError. Doing that would cause your program to exit unexpectedly whenever any socket connection is interrupted while your program is still writing to it.
This advise is for socket programs, not command-line tools.
-Kees
Signature · 1 line
Kees Cook
Quoted 17 lines · Kees Cook
Quoted 6 lines · Masahiro Yamada
def
main():
try:
# simulate large output (your code replaces this loop) for x in range(10000):
print("y")
# flush output here to force SIGPIPE to be triggeredQuoted 8 lines
# while inside this try block.
sys.stdout.flush()
except BrokenPipeError:
# Python flushes standard streams on exit; redirect remaining output # to devnull to avoid another BrokenPipeError at shutdown devnull =os.open(os.devnull, os.O_WRONLY)os.dup2(devnull, sys.stdout.fileno())
sys.exit(1)# Python exits with error code 1 on EPIPE
Quoted 7 lines
I still think this is wrong -- they should not continue piping, and should just die with SIGPIPE. It should simply be:
signal(SIGPIPE, SIG_DFL);Nothing else needed. No wasted CPU cycles, shell handling continues as per normal.
I prefer try-and-except because it is Python's coding style, and we can do something before the exit.
(for example, clean up temporary files)
Quoted 10 lines
Quoted 8 lines
if __name__ == '__main__': main()Do not set SIGPIPE’s disposition to
SIG_DFLin order to avoid BrokenPipeError. Doing that would cause your program to exit unexpectedly whenever any socket connection is interrupted while your program is still writing to it.This advise is for socket programs, not command-line tools.
I still do not understand what is bad
about using this for command-line tools.
Quoted 5 lines
-Kees
--
Kees Cook
Signature · 2 lines
Best Regards Masahiro Yamada
Quoted 52 lines · Masahiro Yamada
Quoted 17 lines · Kees Cook
Quoted 6 lines · Masahiro Yamada
def
main():
try:
# simulate large output (your code replaces this loop) for x in range(10000):
print("y")
# flush output here to force SIGPIPE to be triggeredQuoted 8 lines
# while inside this try block.
sys.stdout.flush()
except BrokenPipeError:
# Python flushes standard streams on exit; redirect remaining output # to devnull to avoid another BrokenPipeError at shutdown devnull =os.open(os.devnull, os.O_WRONLY)os.dup2(devnull, sys.stdout.fileno())
sys.exit(1)# Python exits with error code 1 on EPIPEQuoted 7 lines
I still think this is wrong -- they should not continue piping, and should just die with SIGPIPE. It should simply be:
signal(SIGPIPE, SIG_DFL);Nothing else needed. No wasted CPU cycles, shell handling continues as per normal.
I prefer try-and-except because it is Python's coding style, and we can do something before the exit.
(for example, clean up temporary files)Quoted 10 lines
Quoted 8 lines
if __name__ == '__main__': main()Do not set SIGPIPE’s disposition to
SIG_DFLin order to avoid BrokenPipeError. Doing that would cause your program to exit unexpectedly whenever any socket connection is interrupted while your program is still writing to it.This advise is for socket programs, not command-line tools.
I still do not understand what is bad
about using this for command-line tools.Quoted 3 lines
-Kees
Quoted 8 lines
Quoted 2 lines
--
Kees Cook--
Best Regards
Masahiro Yamada
Applied with the typos fixes.
Signature · 2 lines
Best Regards Masahiro Yamada