Mercurial > hg
view tests/test-lock-badness.t @ 38732:be4984261611
merge: mark file gets as not thread safe (issue5933)
In default installs, this has the effect of disabling the thread-based
worker on Windows when manifesting files in the working directory. My
measurements have shown that with revlog-based repositories, Mercurial
spends a lot of CPU time in revlog code resolving file data. This ends
up incurring a lot of context switching across threads and slows down
`hg update` operations when going from an empty working directory to
the tip of the repo.
On mozilla-unified (246,351 files) on an i7-6700K (4+4 CPUs):
before: 487s wall
after: 360s wall (equivalent to worker.enabled=false)
cpus=2: 379s wall
Even with only 2 threads, the thread pool is still slower.
The introduction of the thread-based worker (02b36e860e0b) states that
it resulted in a "~50%" speedup for `hg sparse --enable-profile` and
`hg sparse --disable-profile`. This disagrees with my measurement
above. I theorize a few reasons for this:
1) Removal of files from the working directory is I/O - not CPU - bound
and should benefit from a thread pool (unless I/O is insanely fast
and the GIL release is near instantaneous). So tests like `hg sparse
--enable-profile` may exercise deletion throughput and aren't good
benchmarks for worker tasks that are CPU heavy.
2) The patch was authored by someone at Facebook. The results were
likely measured against a repository using remotefilelog. And I
believe that revision retrieval during working directory updates with
remotefilelog will often use a remote store, thus being I/O and not
CPU bound. This probably resulted in an overstated performance gain.
Since there appears to be a need to enable the thread-based worker with
some stores, I've made the flagging of file gets as thread safe
configurable. I've made it experimental because I don't want to formalize
a boolean flag for this option and because this attribute is best
captured against the store implementation. But we don't have a proper
store API for this yet. I'd rather cross this bridge later.
It is possible there are revlog-based repositories that do benefit from
a thread-based worker. I didn't do very comprehensive testing. If there
are, we may want to devise a more proper algorithm for whether to use
the thread-based worker, including possibly config options to limit the
number of threads to use. But until I see evidence that justifies
complexity, simplicity wins.
Differential Revision: https://phab.mercurial-scm.org/D3963
author | Gregory Szorc <gregory.szorc@gmail.com> |
---|---|
date | Wed, 18 Jul 2018 09:49:34 -0700 |
parents | 1e1c1bfb0be4 |
children | e0dbfbd4062c |
line wrap: on
line source
#require unix-permissions no-root no-windows Prepare $ hg init a $ echo a > a/a $ hg -R a ci -A -m a adding a $ hg clone a b updating to branch default 1 files updated, 0 files merged, 0 files removed, 0 files unresolved Test that raising an exception in the release function doesn't cause the lock to choke $ cat > testlock.py << EOF > from mercurial import error, registrar > > cmdtable = {} > command = registrar.command(cmdtable) > > def acquiretestlock(repo, releaseexc): > def unlock(): > if releaseexc: > raise error.Abort(b'expected release exception') > l = repo._lock(repo.vfs, b'testlock', False, unlock, None, b'test lock') > return l > > @command(b'testlockexc') > def testlockexc(ui, repo): > testlock = acquiretestlock(repo, True) > try: > testlock.release() > finally: > try: > testlock = acquiretestlock(repo, False) > except error.LockHeld: > raise error.Abort(b'lockfile on disk even after releasing!') > testlock.release() > EOF $ cat >> $HGRCPATH << EOF > [extensions] > testlock=$TESTTMP/testlock.py > EOF $ hg -R b testlockexc abort: expected release exception [255] One process waiting for another $ cat > hooks.py << EOF > import time > def sleepone(**x): time.sleep(1) > def sleephalf(**x): time.sleep(0.5) > EOF $ echo b > b/b $ hg -R b ci -A -m b --config hooks.precommit="python:`pwd`/hooks.py:sleepone" > stdout & $ hg -R b up -q --config hooks.pre-update="python:`pwd`/hooks.py:sleephalf" \ > > preup-stdout 2>preup-stderr $ wait $ cat preup-stdout $ cat preup-stderr waiting for lock on working directory of b held by process '*' on host '*' (glob) got lock after * seconds (glob) $ cat stdout adding b On processs waiting on another, warning after a long time. $ echo b > b/c $ hg -R b ci -A -m b --config hooks.precommit="python:`pwd`/hooks.py:sleepone" > stdout & $ hg -R b up -q --config hooks.pre-update="python:`pwd`/hooks.py:sleephalf" \ > --config ui.timeout.warn=250 \ > > preup-stdout 2>preup-stderr $ wait $ cat preup-stdout $ cat preup-stderr $ cat stdout adding c On processs waiting on another, warning disabled. $ echo b > b/d $ hg -R b ci -A -m b --config hooks.precommit="python:`pwd`/hooks.py:sleepone" > stdout & $ hg -R b up -q --config hooks.pre-update="python:`pwd`/hooks.py:sleephalf" \ > --config ui.timeout.warn=-1 \ > > preup-stdout 2>preup-stderr $ wait $ cat preup-stdout $ cat preup-stderr $ cat stdout adding d check we still print debug output On processs waiting on another, warning after a long time (debug output on) $ echo b > b/e $ hg -R b ci -A -m b --config hooks.precommit="python:`pwd`/hooks.py:sleepone" > stdout & $ hg -R b up --config hooks.pre-update="python:`pwd`/hooks.py:sleephalf" \ > --config ui.timeout.warn=250 --debug\ > > preup-stdout 2>preup-stderr $ wait $ cat preup-stdout calling hook pre-update: hghook_pre-update.sleephalf waiting for lock on working directory of b held by process '*' on host '*' (glob) got lock after * seconds (glob) 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ cat preup-stderr $ cat stdout adding e On processs waiting on another, warning disabled, (debug output on) $ echo b > b/f $ hg -R b ci -A -m b --config hooks.precommit="python:`pwd`/hooks.py:sleepone" > stdout & $ hg -R b up --config hooks.pre-update="python:`pwd`/hooks.py:sleephalf" \ > --config ui.timeout.warn=-1 --debug\ > > preup-stdout 2>preup-stderr $ wait $ cat preup-stdout calling hook pre-update: hghook_pre-update.sleephalf waiting for lock on working directory of b held by process '*' on host '*' (glob) got lock after * seconds (glob) 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ cat preup-stderr $ cat stdout adding f Pushing to a local read-only repo that can't be locked $ chmod 100 a/.hg/store $ hg -R b push a pushing to a searching for changes abort: could not lock repository a: Permission denied [255] $ chmod 700 a/.hg/store