From c5a5839856119a3644dcc0775a046ed0ee3081c3 Mon Sep 17 00:00:00 2001 From: Ahmed Karaman Date: Fri, 26 Jun 2020 18:45:44 +0200 Subject: scripts/performance: Add topN_perf.py script Syntax: topN_perf.py [-h] [-n] -- \ [] \ [] [-h] - Print the script arguments help message. [-n] - Specify the number of top functions to print. - If this flag is not specified, the tool defaults to 25. Example of usage: topN_perf.py -n 20 -- qemu-arm coulomb_double-arm Example Output: No. Percentage Name Invoked by ---- ---------- ------------------------- ------------------------- 1 16.25% float64_mul qemu-x86_64 2 12.01% float64_sub qemu-x86_64 3 11.99% float64_add qemu-x86_64 4 5.69% helper_mulsd qemu-x86_64 5 4.68% helper_addsd qemu-x86_64 6 4.43% helper_lookup_tb_ptr qemu-x86_64 7 4.28% helper_subsd qemu-x86_64 8 2.71% f64_compare qemu-x86_64 9 2.71% helper_ucomisd qemu-x86_64 10 1.04% helper_pand_xmm qemu-x86_64 11 0.71% float64_div qemu-x86_64 12 0.63% helper_pxor_xmm qemu-x86_64 13 0.50% 0x00007f7b7004ef95 [JIT] tid 491 14 0.50% 0x00007f7b70044e83 [JIT] tid 491 15 0.36% helper_por_xmm qemu-x86_64 16 0.32% helper_cc_compute_all qemu-x86_64 17 0.30% 0x00007f7b700433f0 [JIT] tid 491 18 0.30% float64_compare_quiet qemu-x86_64 19 0.27% soft_f64_addsub qemu-x86_64 20 0.26% round_to_int qemu-x86_64 Signed-off-by: Ahmed Karaman Signed-off-by: Aleksandar Markovic Reviewed-by: Aleksandar Markovic Message-Id: <20200626164546.22102-2-ahmedkhaledkaraman@gmail.com> --- scripts/performance/topN_perf.py | 149 +++++++++++++++++++++++++++++++++++++++ 1 file changed, 149 insertions(+) create mode 100755 scripts/performance/topN_perf.py (limited to 'scripts/performance') diff --git a/scripts/performance/topN_perf.py b/scripts/performance/topN_perf.py new file mode 100755 index 0000000..07be195 --- /dev/null +++ b/scripts/performance/topN_perf.py @@ -0,0 +1,149 @@ +#!/usr/bin/env python3 + +# Print the top N most executed functions in QEMU using perf. +# Syntax: +# topN_perf.py [-h] [-n] -- \ +# [] \ +# [] +# +# [-h] - Print the script arguments help message. +# [-n] - Specify the number of top functions to print. +# - If this flag is not specified, the tool defaults to 25. +# +# Example of usage: +# topN_perf.py -n 20 -- qemu-arm coulomb_double-arm +# +# This file is a part of the project "TCG Continuous Benchmarking". +# +# Copyright (C) 2020 Ahmed Karaman +# Copyright (C) 2020 Aleksandar Markovic +# +# This program is free software: you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation, either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see . + +import argparse +import os +import subprocess +import sys + + +# Parse the command line arguments +parser = argparse.ArgumentParser( + usage='topN_perf.py [-h] [-n] -- ' + ' [] ' + ' []') + +parser.add_argument('-n', dest='top', type=int, default=25, + help='Specify the number of top functions to print.') + +parser.add_argument('command', type=str, nargs='+', help=argparse.SUPPRESS) + +args = parser.parse_args() + +# Extract the needed variables from the args +command = args.command +top = args.top + +# Insure that perf is installed +check_perf_presence = subprocess.run(["which", "perf"], + stdout=subprocess.DEVNULL) +if check_perf_presence.returncode: + sys.exit("Please install perf before running the script!") + +# Insure user has previllage to run perf +check_perf_executability = subprocess.run(["perf", "stat", "ls", "/"], + stdout=subprocess.DEVNULL, + stderr=subprocess.DEVNULL) +if check_perf_executability.returncode: + sys.exit( +""" +Error: +You may not have permission to collect stats. + +Consider tweaking /proc/sys/kernel/perf_event_paranoid, +which controls use of the performance events system by +unprivileged users (without CAP_SYS_ADMIN). + + -1: Allow use of (almost) all events by all users + Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK + 0: Disallow ftrace function tracepoint by users without CAP_SYS_ADMIN + Disallow raw tracepoint access by users without CAP_SYS_ADMIN + 1: Disallow CPU event access by users without CAP_SYS_ADMIN + 2: Disallow kernel profiling by users without CAP_SYS_ADMIN + +To make this setting permanent, edit /etc/sysctl.conf too, e.g.: + kernel.perf_event_paranoid = -1 + +* Alternatively, you can run this script under sudo privileges. +""" +) + +# Run perf record +perf_record = subprocess.run((["perf", "record", "--output=/tmp/perf.data"] + + command), + stdout=subprocess.DEVNULL, + stderr=subprocess.PIPE) +if perf_record.returncode: + os.unlink('/tmp/perf.data') + sys.exit(perf_record.stderr.decode("utf-8")) + +# Save perf report output to /tmp/perf_report.out +with open("/tmp/perf_report.out", "w") as output: + perf_report = subprocess.run( + ["perf", "report", "--input=/tmp/perf.data", "--stdio"], + stdout=output, + stderr=subprocess.PIPE) + if perf_report.returncode: + os.unlink('/tmp/perf.data') + output.close() + os.unlink('/tmp/perf_report.out') + sys.exit(perf_report.stderr.decode("utf-8")) + +# Read the reported data to functions[] +functions = [] +with open("/tmp/perf_report.out", "r") as data: + # Only read lines that are not comments (comments start with #) + # Only read lines that are not empty + functions = [line for line in data.readlines() if line and line[0] + != '#' and line[0] != "\n"] + +# Limit the number of top functions to "top" +number_of_top_functions = top if len(functions) > top else len(functions) + +# Store the data of the top functions in top_functions[] +top_functions = functions[:number_of_top_functions] + +# Print table header +print('{:>4} {:>10} {:<30} {}\n{} {} {} {}'.format('No.', + 'Percentage', + 'Name', + 'Invoked by', + '-' * 4, + '-' * 10, + '-' * 30, + '-' * 25)) + +# Print top N functions +for (index, function) in enumerate(top_functions, start=1): + function_data = function.split() + function_percentage = function_data[0] + function_name = function_data[-1] + function_invoker = ' '.join(function_data[2:-2]) + print('{:>4} {:>10} {:<30} {}'.format(index, + function_percentage, + function_name, + function_invoker)) + +# Remove intermediate files +os.unlink('/tmp/perf.data') +os.unlink('/tmp/perf_report.out') -- cgit v1.1 From 5c362ccfdec5fa8e6e43d0d5e82a600b34b8e6d2 Mon Sep 17 00:00:00 2001 From: Ahmed Karaman Date: Fri, 26 Jun 2020 18:45:45 +0200 Subject: scripts/performance: Add topN_callgrind.py script Python script that prints the top N most executed functions in QEMU using callgrind. Syntax: topN_callgrind.py [-h] [-n] -- \ [] \ [] [-h] - Print the script arguments help message. [-n] - Specify the number of top functions to print. - If this flag is not specified, the tool defaults to 25. Example of usage: topN_callgrind.py -n 20 -- qemu-arm coulomb_double-arm Example Output: No. Percentage Function Name Source File ---- --------- ------------------ ------------------------------ 1 24.577% 0x00000000082db000 ??? 2 20.467% float64_mul /fpu/softfloat.c 3 14.720% float64_sub /fpu/softfloat.c 4 13.864% float64_add /fpu/softfloat.c 5 4.876% helper_mulsd /target/i386/ops_sse.h 6 3.767% helper_subsd /target/i386/ops_sse.h 7 3.549% helper_addsd /target/i386/ops_sse.h 8 2.185% helper_ucomisd /target/i386/ops_sse.h 9 1.667% helper_lookup_tb_ptr /include/exec/tb-lookup.h 10 1.662% f64_compare /fpu/softfloat.c 11 1.509% helper_lookup_tb_ptr /accel/tcg/tcg-runtime.c 12 0.635% helper_lookup_tb_ptr /include/exec/exec-all.h 13 0.616% float64_div /fpu/softfloat.c 14 0.502% helper_pand_xmm /target/i386/ops_sse.h 15 0.502% float64_mul /include/fpu/softfloat.h 16 0.476% helper_lookup_tb_ptr /target/i386/cpu.h 17 0.437% float64_compare_quiet /fpu/softfloat.c 18 0.414% helper_pxor_xmm /target/i386/ops_sse.h 19 0.353% round_to_int /fpu/softfloat.c 20 0.347% helper_cc_compute_all /target/i386/cc_helper.c Signed-off-by: Ahmed Karaman Signed-off-by: Aleksandar Markovic Reviewed-by: Aleksandar Markovic Message-Id: <20200626164546.22102-3-ahmedkhaledkaraman@gmail.com> --- scripts/performance/topN_callgrind.py | 140 ++++++++++++++++++++++++++++++++++ 1 file changed, 140 insertions(+) create mode 100755 scripts/performance/topN_callgrind.py (limited to 'scripts/performance') diff --git a/scripts/performance/topN_callgrind.py b/scripts/performance/topN_callgrind.py new file mode 100755 index 0000000..67c5919 --- /dev/null +++ b/scripts/performance/topN_callgrind.py @@ -0,0 +1,140 @@ +#!/usr/bin/env python3 + +# Print the top N most executed functions in QEMU using callgrind. +# Syntax: +# topN_callgrind.py [-h] [-n] -- \ +# [] \ +# [] +# +# [-h] - Print the script arguments help message. +# [-n] - Specify the number of top functions to print. +# - If this flag is not specified, the tool defaults to 25. +# +# Example of usage: +# topN_callgrind.py -n 20 -- qemu-arm coulomb_double-arm +# +# This file is a part of the project "TCG Continuous Benchmarking". +# +# Copyright (C) 2020 Ahmed Karaman +# Copyright (C) 2020 Aleksandar Markovic +# +# This program is free software: you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation, either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see . + +import argparse +import os +import subprocess +import sys + + +# Parse the command line arguments +parser = argparse.ArgumentParser( + usage='topN_callgrind.py [-h] [-n] -- ' + ' [] ' + ' []') + +parser.add_argument('-n', dest='top', type=int, default=25, + help='Specify the number of top functions to print.') + +parser.add_argument('command', type=str, nargs='+', help=argparse.SUPPRESS) + +args = parser.parse_args() + +# Extract the needed variables from the args +command = args.command +top = args.top + +# Insure that valgrind is installed +check_valgrind_presence = subprocess.run(["which", "valgrind"], + stdout=subprocess.DEVNULL) +if check_valgrind_presence.returncode: + sys.exit("Please install valgrind before running the script!") + +# Run callgrind +callgrind = subprocess.run(( + ["valgrind", "--tool=callgrind", "--callgrind-out-file=/tmp/callgrind.data"] + + command), + stdout=subprocess.DEVNULL, + stderr=subprocess.PIPE) +if callgrind.returncode: + sys.exit(callgrind.stderr.decode("utf-8")) + +# Save callgrind_annotate output to /tmp/callgrind_annotate.out +with open("/tmp/callgrind_annotate.out", "w") as output: + callgrind_annotate = subprocess.run(["callgrind_annotate", + "/tmp/callgrind.data"], + stdout=output, + stderr=subprocess.PIPE) + if callgrind_annotate.returncode: + os.unlink('/tmp/callgrind.data') + output.close() + os.unlink('/tmp/callgrind_annotate.out') + sys.exit(callgrind_annotate.stderr.decode("utf-8")) + +# Read the callgrind_annotate output to callgrind_data[] +callgrind_data = [] +with open('/tmp/callgrind_annotate.out', 'r') as data: + callgrind_data = data.readlines() + +# Line number with the total number of instructions +total_instructions_line_number = 20 + +# Get the total number of instructions +total_instructions_line_data = callgrind_data[total_instructions_line_number] +total_number_of_instructions = total_instructions_line_data.split(' ')[0] +total_number_of_instructions = int( + total_number_of_instructions.replace(',', '')) + +# Line number with the top function +first_func_line = 25 + +# Number of functions recorded by callgrind, last two lines are always empty +number_of_functions = len(callgrind_data) - first_func_line - 2 + +# Limit the number of top functions to "top" +number_of_top_functions = (top if number_of_functions > + top else number_of_functions) + +# Store the data of the top functions in top_functions[] +top_functions = callgrind_data[first_func_line: + first_func_line + number_of_top_functions] + +# Print table header +print('{:>4} {:>10} {:<30} {}\n{} {} {} {}'.format('No.', + 'Percentage', + 'Function Name', + 'Source File', + '-' * 4, + '-' * 10, + '-' * 30, + '-' * 30, + )) + +# Print top N functions +for (index, function) in enumerate(top_functions, start=1): + function_data = function.split() + # Calculate function percentage + function_instructions = float(function_data[0].replace(',', '')) + function_percentage = (function_instructions / + total_number_of_instructions)*100 + # Get function name and source files path + function_source_file, function_name = function_data[1].split(':') + # Print extracted data + print('{:>4} {:>9.3f}% {:<30} {}'.format(index, + round(function_percentage, 3), + function_name, + function_source_file)) + +# Remove intermediate files +os.unlink('/tmp/callgrind.data') +os.unlink('/tmp/callgrind_annotate.out') -- cgit v1.1