localrepo: experimental support for non-zlib revlog compression
The final part of integrating the compression manager APIs into
revlog storage is the plumbing for repositories to advertise they
are using non-zlib storage and for revlogs to instantiate a non-zlib
compression engine.
The main intent of the compression manager work was to zstd all
of the things. Adding zstd to revlogs has proved to be more involved
than other places because revlogs are... special. Very small inputs
and the use of delta chains (which are themselves a form of
compression) are a completely different use case from streaming
compression, which bundles and the wire protocol employ. I've
conducted numerous experiments with zstd in revlogs and have yet
to formalize compression settings and a storage architecture that
I'm confident I won't regret later. In other words, I'm not yet
ready to commit to a new mechanism for using zstd - or any other
compression format - in revlogs.
That being said, having some support for zstd (and other compression
formats) in revlogs in core is beneficial. It can allow others to
conduct experiments.
This patch introduces *highly experimental* support for non-zlib
compression formats in revlogs. Introduced is a config option to
control which compression engine to use. Also introduced is a namespace
of "exp-compression-*" requirements to denote support for non-zlib
compression in revlogs. I've prefixed the namespace with "exp-"
(short for "experimental") because I'm not confident of the
requirements "schema" and in no way want to give the illusion of
supporting these requirements in the future. I fully intend to drop
support for these requirements once we figure out what we're doing
with zstd in revlogs.
A good portion of the patch is teaching the requirements system
about registered compression engines and passing the requested
compression engine as an opener option so revlogs can instantiate
the proper compression engine for new operations.
That's a verbose way of saying "we can now use zstd in revlogs!"
On an `hg pull` conversion of the mozilla-unified repo with no extra
redelta settings (like aggressivemergedeltas), we can see the impact
of zstd vs zlib in revlogs:
$ hg perfrevlogchunks -c
! chunk
! wall 2.032052 comb 2.040000 user 1.990000 sys 0.050000 (best of 5)
! wall 1.866360 comb 1.860000 user 1.820000 sys 0.040000 (best of 6)
! chunk batch
! wall 1.877261 comb 1.870000 user 1.860000 sys 0.010000 (best of 6)
! wall 1.705410 comb 1.710000 user 1.690000 sys 0.020000 (best of 6)
$ hg perfrevlogchunks -m
! chunk
! wall 2.721427 comb 2.720000 user 2.640000 sys 0.080000 (best of 4)
! wall 2.035076 comb 2.030000 user 1.950000 sys 0.080000 (best of 5)
! chunk batch
! wall 2.614561 comb 2.620000 user 2.580000 sys 0.040000 (best of 4)
! wall 1.910252 comb 1.910000 user 1.880000 sys 0.030000 (best of 6)
$ hg perfrevlog -c -d 1
! wall 4.812885 comb 4.820000 user 4.800000 sys 0.020000 (best of 3)
! wall 4.699621 comb 4.710000 user 4.700000 sys 0.010000 (best of 3)
$ hg perfrevlog -m -d 1000
! wall 34.252800 comb 34.250000 user 33.730000 sys 0.520000 (best of 3)
! wall 24.094999 comb 24.090000 user 23.320000 sys 0.770000 (best of 3)
Only modest wins for the changelog. But manifest reading is
significantly faster. What's going on?
One reason might be data volume. zstd decompresses faster. So given
more bytes, it will put more distance between it and zlib.
Another reason is size. In the current design, zstd revlogs are
*larger*:
debugcreatestreamclonebundle (size in bytes)
zlib: 1,638,852,492
zstd: 1,680,601,332
I haven't investigated this fully, but I reckon a significant cause of
larger revlogs is that the zstd frame/header has more bytes than
zlib's. For very small inputs or data that doesn't compress well, we'll
tend to store more uncompressed chunks than with zlib (because the
compressed size isn't smaller than original). This will make revlog
reading faster because it is doing less decompression.
Moving on to bundle performance:
$ hg bundle -a -t none-v2 (total CPU time)
zlib: 102.79s
zstd: 97.75s
So, marginal CPU decrease for reading all chunks in all revlogs
(this is somewhat disappointing).
$ hg bundle -a -t <engine>-v2 (total CPU time)
zlib: 191.59s
zstd: 115.36s
This last test effectively measures the difference between zlib->zlib
and zstd->zstd for revlogs to bundle. This is a rough approximation of
what a server does during `hg clone`.
There are some promising results for zstd. But not enough for me to
feel comfortable advertising it to users. We'll get there...
"""This unit test primarily tests parsers.parse_index2().
It also checks certain aspects of the parsers module as a whole.
"""
from __future__ import absolute_import, print_function
import struct
import subprocess
import sys
from mercurial.node import (
nullid,
nullrev,
)
from mercurial import (
parsers,
)
# original python implementation
def gettype(q):
return int(q & 0xFFFF)
def offset_type(offset, type):
return long(long(offset) << 16 | type)
indexformatng = ">Qiiiiii20s12x"
def py_parseindex(data, inline) :
s = 64
cache = None
index = []
nodemap = {nullid: nullrev}
n = off = 0
l = len(data) - s
append = index.append
if inline:
cache = (0, data)
while off <= l:
e = struct.unpack(indexformatng, data[off:off + s])
nodemap[e[7]] = n
append(e)
n += 1
if e[1] < 0:
break
off += e[1] + s
else:
while off <= l:
e = struct.unpack(indexformatng, data[off:off + s])
nodemap[e[7]] = n
append(e)
n += 1
off += s
e = list(index[0])
type = gettype(e[0])
e[0] = offset_type(0, type)
index[0] = tuple(e)
# add the magic null revision at -1
index.append((0, 0, 0, -1, -1, -1, -1, nullid))
return index, cache
data_inlined = '\x00\x01\x00\x01\x00\x00\x00\x00\x00\x00\x01\x8c' \
'\x00\x00\x04\x07\x00\x00\x00\x00\x00\x00\x15\x15\xff\xff\xff' \
'\xff\xff\xff\xff\xff\xebG\x97\xb7\x1fB\x04\xcf\x13V\x81\tw\x1b' \
'w\xdduR\xda\xc6\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00' \
'x\x9c\x9d\x93?O\xc30\x10\xc5\xf7|\x8a\xdb\x9a\xa8m\x06\xd8*\x95' \
'\x81B\xa1\xa2\xa2R\xcb\x86Pd\x9a\x0b5$vd_\x04\xfd\xf6\x9c\xff@' \
'\x11!\x0b\xd9\xec\xf7\xbbw\xe7gG6\xad6\x04\xdaN\xc0\x92\xa0$)' \
'\xb1\x82\xa2\xd1%\x16\xa4\x8b7\xa9\xca\xd4-\xb2Y\x02\xfc\xc9' \
'\xcaS\xf9\xaeX\xed\xb6\xd77Q\x02\x83\xd4\x19\xf5--Y\xea\xe1W' \
'\xab\xed\x10\xceR\x0f_\xdf\xdf\r\xe1,\xf5\xf0\xcb\xf5 \xceR\x0f' \
'_\xdc\x0e\x0e\xc3R\x0f_\xae\x96\x9b!\x9e\xa5\x1e\xbf\xdb,\x06' \
'\xc7q\x9a/\x88\x82\xc3B\xea\xb5\xb4TJ\x93\xb6\x82\x0e\xe16\xe6' \
'KQ\xdb\xaf\xecG\xa3\xd1 \x01\xd3\x0b_^\xe8\xaa\xa0\xae\xad\xd1' \
'&\xbef\x1bz\x08\xb0|\xc9Xz\x06\xf6Z\x91\x90J\xaa\x17\x90\xaa' \
'\xd2\xa6\x11$5C\xcf\xba#\xa0\x03\x02*2\x92-\xfc\xb1\x94\xdf\xe2' \
'\xae\xb8\'m\x8ey0^\x85\xd3\x82\xb4\xf0`:\x9c\x00\x8a\xfd\x01' \
'\xb0\xc6\x86\x8b\xdd\xae\x80\xf3\xa9\x9fd\x16\n\x00R%\x1a\x06' \
'\xe9\xd8b\x98\x1d\xf4\xf3+\x9bf\x01\xd8p\x1b\xf3.\xed\x9f^g\xc3' \
'^\xd9W81T\xdb\xd5\x04sx|\xf2\xeb\xd6`%?x\xed"\x831\xbf\xf3\xdc' \
'b\xeb%gaY\xe1\xad\x9f\xb9f\'1w\xa9\xa5a\x83s\x82J\xb98\xbc4\x8b' \
'\x83\x00\x9f$z\xb8#\xa5\xb1\xdf\x98\xd9\xec\x1b\x89O\xe3Ts\x9a4' \
'\x17m\x8b\xfc\x8f\xa5\x95\x9a\xfc\xfa\xed,\xe5|\xa1\xfe\x15\xb9' \
'\xbc\xb2\x93\x1f\xf2\x95\xff\xdf,\x1a\xc5\xe7\x17*\x93Oz:>\x0e'
data_non_inlined = '\x00\x00\x00\x01\x00\x00\x00\x00\x00\x01D\x19' \
'\x00\x07e\x12\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff' \
'\xff\xff\xff\xff\xd1\xf4\xbb\xb0\xbe\xfc\x13\xbd\x8c\xd3\x9d' \
'\x0f\xcd\xd9;\x8c\x07\x8cJ/\x00\x00\x00\x00\x00\x00\x00\x00\x00' \
'\x00\x00\x00\x00\x00\x00\x01D\x19\x00\x00\x00\x00\x00\xdf\x00' \
'\x00\x01q\x00\x00\x00\x01\x00\x00\x00\x01\x00\x00\x00\x00\xff' \
'\xff\xff\xff\xc1\x12\xb9\x04\x96\xa4Z1t\x91\xdfsJ\x90\xf0\x9bh' \
'\x07l&\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00' \
'\x00\x01D\xf8\x00\x00\x00\x00\x01\x1b\x00\x00\x01\xb8\x00\x00' \
'\x00\x01\x00\x00\x00\x02\x00\x00\x00\x01\xff\xff\xff\xff\x02\n' \
'\x0e\xc6&\xa1\x92\xae6\x0b\x02i\xfe-\xe5\xbao\x05\xd1\xe7\x00' \
'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01F' \
'\x13\x00\x00\x00\x00\x01\xec\x00\x00\x03\x06\x00\x00\x00\x01' \
'\x00\x00\x00\x03\x00\x00\x00\x02\xff\xff\xff\xff\x12\xcb\xeby1' \
'\xb6\r\x98B\xcb\x07\xbd`\x8f\x92\xd9\xc4\x84\xbdK\x00\x00\x00' \
'\x00\x00\x00\x00\x00\x00\x00\x00\x00'
def parse_index2(data, inline):
index, chunkcache = parsers.parse_index2(data, inline)
return list(index), chunkcache
def importparsers(hexversion):
"""Import mercurial.parsers with the given sys.hexversion."""
# The file parsers.c inspects sys.hexversion to determine the version
# of the currently-running Python interpreter, so we monkey-patch
# sys.hexversion to simulate using different versions.
code = ("import sys; sys.hexversion=%s; "
"import mercurial.parsers" % hexversion)
cmd = "python -c \"%s\"" % code
# We need to do these tests inside a subprocess because parser.c's
# version-checking code happens inside the module init function, and
# when using reload() to reimport an extension module, "The init function
# of extension modules is not called a second time"
# (from http://docs.python.org/2/library/functions.html?#reload).
p = subprocess.Popen(cmd, shell=True,
stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
return p.communicate() # returns stdout, stderr
def printhexfail(testnumber, hexversion, stdout, expected):
try:
hexstring = hex(hexversion)
except TypeError:
hexstring = None
print("FAILED: version test #%s with Python %s and patched "
"sys.hexversion %r (%r):\n Expected %s but got:\n-->'%s'\n" %
(testnumber, sys.version_info, hexversion, hexstring, expected,
stdout))
def testversionokay(testnumber, hexversion):
stdout, stderr = importparsers(hexversion)
if stdout:
printhexfail(testnumber, hexversion, stdout, expected="no stdout")
def testversionfail(testnumber, hexversion):
stdout, stderr = importparsers(hexversion)
# We include versionerrortext to distinguish from other ImportErrors.
errtext = "ImportError: %s" % parsers.versionerrortext
if errtext not in stdout:
printhexfail(testnumber, hexversion, stdout,
expected="stdout to contain %r" % errtext)
def makehex(major, minor, micro):
return int("%x%02x%02x00" % (major, minor, micro), 16)
def runversiontests():
"""Check the version-detection logic when importing parsers."""
info = sys.version_info
major, minor, micro = info[0], info[1], info[2]
# Test same major-minor versions.
testversionokay(1, makehex(major, minor, micro))
testversionokay(2, makehex(major, minor, micro + 1))
# Test different major-minor versions.
testversionfail(3, makehex(major + 1, minor, micro))
testversionfail(4, makehex(major, minor + 1, micro))
testversionfail(5, "'foo'")
def runtest() :
# Only test the version-detection logic if it is present.
try:
parsers.versionerrortext
except AttributeError:
pass
else:
runversiontests()
# Check that parse_index2() raises TypeError on bad arguments.
try:
parse_index2(0, True)
except TypeError:
pass
else:
print("Expected to get TypeError.")
# Check parsers.parse_index2() on an index file against the original
# Python implementation of parseindex, both with and without inlined data.
py_res_1 = py_parseindex(data_inlined, True)
c_res_1 = parse_index2(data_inlined, True)
py_res_2 = py_parseindex(data_non_inlined, False)
c_res_2 = parse_index2(data_non_inlined, False)
if py_res_1 != c_res_1:
print("Parse index result (with inlined data) differs!")
if py_res_2 != c_res_2:
print("Parse index result (no inlined data) differs!")
ix = parsers.parse_index2(data_inlined, True)[0]
for i, r in enumerate(ix):
if r[7] == nullid:
i = -1
try:
if ix[r[7]] != i:
print('Reverse lookup inconsistent for %r'
% r[7].encode('hex'))
except TypeError:
# pure version doesn't support this
break
print("done")
runtest()