1
0
Fork 0
mirror of https://github.com/git/git.git synced 2024-05-09 14:56:07 +02:00
git/t/t0211-trace2-perf.sh
Jeff Hostetler 8ad575646c trace2: add stopwatch timers
Add stopwatch timer mechanism to Trace2.

Timers are an alternative to Trace2 Regions.  Regions are useful for
measuring the time spent in various computation phases, such as the
time to read the index, time to scan for unstaged files, time to scan
for untracked files, and etc.

However, regions are not appropriate in all places.  For example,
during a checkout, it would be very inefficient to use regions to
measure the total time spent inflating objects from the ODB from
across the entire lifetime of the process; a per-unzip() region would
flood the output and significantly slow the command; and some form of
post-processing would be requried to compute the time spent in unzip().

Timers can be used to measure a series of timer intervals and emit
a single summary event (at thread and/or process exit).

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-10-24 12:45:26 -07:00

226 lines
7.6 KiB
Bash
Executable File

#!/bin/sh
test_description='test trace2 facility (perf target)'
TEST_PASSES_SANITIZE_LEAK=true
. ./test-lib.sh
# Turn off any inherited trace2 settings for this test.
sane_unset GIT_TRACE2 GIT_TRACE2_PERF GIT_TRACE2_EVENT
sane_unset GIT_TRACE2_PERF_BRIEF
sane_unset GIT_TRACE2_CONFIG_PARAMS
# Add t/helper directory to PATH so that we can use a relative
# path to run nested instances of test-tool.exe (see 004child).
# This helps with HEREDOC comparisons later.
TTDIR="$GIT_BUILD_DIR/t/helper/" && export TTDIR
PATH="$TTDIR:$PATH" && export PATH
# Warning: use of 'test_cmp' may run test-tool.exe and/or git.exe
# Warning: to do the actual diff/comparison, so the HEREDOCs here
# Warning: only cover our actual calls to test-tool and/or git.
# Warning: So you may see extra lines in artifact files when
# Warning: interactively debugging.
V=$(git version | sed -e 's/^git version //') && export V
# There are multiple trace2 targets: normal, perf, and event.
# Trace2 events will/can be written to each active target (subject
# to whatever filtering that target decides to do).
# Test each target independently.
#
# Defer setting GIT_TRACE2_PERF until the actual command we want to
# test because hidden git and test-tool commands in the test
# harness can contaminate our output.
# Enable "brief" feature which turns off the prefix:
# "<clock> <file>:<line> | <nr_parents> | "
GIT_TRACE2_PERF_BRIEF=1 && export GIT_TRACE2_PERF_BRIEF
# Repeat some of the t0210 tests using the perf target stream instead of
# the normal stream.
#
# Tokens here of the form _FIELD_ have been replaced in the observed output.
# Verb 001return
#
# Implicit return from cmd_<verb> function propagates <code>.
test_expect_success 'perf stream, return code 0' '
test_when_finished "rm trace.perf actual expect" &&
GIT_TRACE2_PERF="$(pwd)/trace.perf" test-tool trace2 001return 0 &&
perl "$TEST_DIRECTORY/t0211/scrub_perf.perl" <trace.perf >actual &&
cat >expect <<-EOF &&
d0|main|version|||||$V
d0|main|start||_T_ABS_|||_EXE_ trace2 001return 0
d0|main|cmd_name|||||trace2 (trace2)
d0|main|exit||_T_ABS_|||code:0
d0|main|atexit||_T_ABS_|||code:0
EOF
test_cmp expect actual
'
test_expect_success 'perf stream, return code 1' '
test_when_finished "rm trace.perf actual expect" &&
test_must_fail env GIT_TRACE2_PERF="$(pwd)/trace.perf" test-tool trace2 001return 1 &&
perl "$TEST_DIRECTORY/t0211/scrub_perf.perl" <trace.perf >actual &&
cat >expect <<-EOF &&
d0|main|version|||||$V
d0|main|start||_T_ABS_|||_EXE_ trace2 001return 1
d0|main|cmd_name|||||trace2 (trace2)
d0|main|exit||_T_ABS_|||code:1
d0|main|atexit||_T_ABS_|||code:1
EOF
test_cmp expect actual
'
# Verb 003error
#
# To the above, add multiple 'error <msg>' events
test_expect_success 'perf stream, error event' '
test_when_finished "rm trace.perf actual expect" &&
GIT_TRACE2_PERF="$(pwd)/trace.perf" test-tool trace2 003error "hello world" "this is a test" &&
perl "$TEST_DIRECTORY/t0211/scrub_perf.perl" <trace.perf >actual &&
cat >expect <<-EOF &&
d0|main|version|||||$V
d0|main|start||_T_ABS_|||_EXE_ trace2 003error '\''hello world'\'' '\''this is a test'\''
d0|main|cmd_name|||||trace2 (trace2)
d0|main|error|||||hello world
d0|main|error|||||this is a test
d0|main|exit||_T_ABS_|||code:0
d0|main|atexit||_T_ABS_|||code:0
EOF
test_cmp expect actual
'
# Verb 004child
#
# Test nested spawning of child processes.
#
# Conceptually, this looks like:
# P1: TT trace2 004child
# P2: |--- TT trace2 004child
# P3: |--- TT trace2 001return 0
#
# Which should generate events:
# P1: version
# P1: start
# P1: cmd_name
# P1: child_start
# P2: version
# P2: start
# P2: cmd_name
# P2: child_start
# P3: version
# P3: start
# P3: cmd_name
# P3: exit
# P3: atexit
# P2: child_exit
# P2: exit
# P2: atexit
# P1: child_exit
# P1: exit
# P1: atexit
test_expect_success 'perf stream, child processes' '
test_when_finished "rm trace.perf actual expect" &&
GIT_TRACE2_PERF="$(pwd)/trace.perf" test-tool trace2 004child test-tool trace2 004child test-tool trace2 001return 0 &&
perl "$TEST_DIRECTORY/t0211/scrub_perf.perl" <trace.perf >actual &&
cat >expect <<-EOF &&
d0|main|version|||||$V
d0|main|start||_T_ABS_|||_EXE_ trace2 004child test-tool trace2 004child test-tool trace2 001return 0
d0|main|cmd_name|||||trace2 (trace2)
d0|main|child_start||_T_ABS_|||[ch0] class:? argv:[test-tool trace2 004child test-tool trace2 001return 0]
d1|main|version|||||$V
d1|main|start||_T_ABS_|||_EXE_ trace2 004child test-tool trace2 001return 0
d1|main|cmd_name|||||trace2 (trace2/trace2)
d1|main|child_start||_T_ABS_|||[ch0] class:? argv:[test-tool trace2 001return 0]
d2|main|version|||||$V
d2|main|start||_T_ABS_|||_EXE_ trace2 001return 0
d2|main|cmd_name|||||trace2 (trace2/trace2/trace2)
d2|main|exit||_T_ABS_|||code:0
d2|main|atexit||_T_ABS_|||code:0
d1|main|child_exit||_T_ABS_|_T_REL_||[ch0] pid:_PID_ code:0
d1|main|exit||_T_ABS_|||code:0
d1|main|atexit||_T_ABS_|||code:0
d0|main|child_exit||_T_ABS_|_T_REL_||[ch0] pid:_PID_ code:0
d0|main|exit||_T_ABS_|||code:0
d0|main|atexit||_T_ABS_|||code:0
EOF
test_cmp expect actual
'
sane_unset GIT_TRACE2_PERF_BRIEF
# Now test without environment variables and get all Trace2 settings
# from the global config.
test_expect_success 'using global config, perf stream, return code 0' '
test_when_finished "rm trace.perf actual expect" &&
test_config_global trace2.perfBrief 1 &&
test_config_global trace2.perfTarget "$(pwd)/trace.perf" &&
test-tool trace2 001return 0 &&
perl "$TEST_DIRECTORY/t0211/scrub_perf.perl" <trace.perf >actual &&
cat >expect <<-EOF &&
d0|main|version|||||$V
d0|main|start||_T_ABS_|||_EXE_ trace2 001return 0
d0|main|cmd_name|||||trace2 (trace2)
d0|main|exit||_T_ABS_|||code:0
d0|main|atexit||_T_ABS_|||code:0
EOF
test_cmp expect actual
'
# Exercise the stopwatch timers in a loop and confirm that we have
# as many start/stop intervals as expected. We cannot really test the
# actual (total, min, max) timer values, so we have to assume that they
# are good, but we can verify the interval count.
#
# The timer "test/test1" should only emit a global summary "timer" event.
# The timer "test/test2" should emit per-thread "th_timer" events and a
# global summary "timer" event.
have_timer_event () {
thread=$1 event=$2 category=$3 name=$4 intervals=$5 file=$6 &&
pattern="d0|${thread}|${event}||||${category}|name:${name} intervals:${intervals}" &&
grep "${pattern}" ${file}
}
test_expect_success 'stopwatch timer test/test1' '
test_when_finished "rm trace.perf actual" &&
test_config_global trace2.perfBrief 1 &&
test_config_global trace2.perfTarget "$(pwd)/trace.perf" &&
# Use the timer "test1" 5 times from "main".
test-tool trace2 100timer 5 10 &&
perl "$TEST_DIRECTORY/t0211/scrub_perf.perl" <trace.perf >actual &&
have_timer_event "main" "timer" "test" "test1" 5 actual
'
test_expect_success 'stopwatch timer test/test2' '
test_when_finished "rm trace.perf actual" &&
test_config_global trace2.perfBrief 1 &&
test_config_global trace2.perfTarget "$(pwd)/trace.perf" &&
# Use the timer "test2" 5 times each in 3 threads.
test-tool trace2 101timer 5 10 3 &&
perl "$TEST_DIRECTORY/t0211/scrub_perf.perl" <trace.perf >actual &&
# So we should have 3 per-thread events of 5 each.
have_timer_event "th01:ut_101" "th_timer" "test" "test2" 5 actual &&
have_timer_event "th02:ut_101" "th_timer" "test" "test2" 5 actual &&
have_timer_event "th03:ut_101" "th_timer" "test" "test2" 5 actual &&
# And we should have 15 total uses.
have_timer_event "main" "timer" "test" "test2" 15 actual
'
test_done