| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
"git commit-graph write" learned to limit the number of bloom
filters that are computed from scratch with the --max-new-filters
option.
* tb/bloom-improvements:
commit-graph: introduce 'commitGraph.maxNewFilters'
builtin/commit-graph.c: introduce '--max-new-filters=<n>'
commit-graph: rename 'split_commit_graph_opts'
bloom: encode out-of-bounds filters as non-empty
bloom/diff: properly short-circuit on max_changes
bloom: use provided 'struct bloom_filter_settings'
bloom: split 'get_bloom_filter()' in two
commit-graph.c: store maximum changed paths
commit-graph: respect 'commitGraph.readChangedPaths'
t/helper/test-read-graph.c: prepare repo settings
commit-graph: pass a 'struct repository *' in more places
t4216: use an '&&'-chain
commit-graph: introduce 'get_bloom_filter_settings()'
|
|
Introduce a configuration variable to specify a default value for the
recently-introduce '--max-new-filters' option of 'git commit-graph
write'.
Helped-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Introduce a command-line flag to specify the maximum number of new Bloom
filters that a 'git commit-graph write' is willing to compute from
scratch.
Prior to this patch, a commit-graph write with '--changed-paths' would
compute Bloom filters for all selected commits which haven't already
been computed (i.e., by a previous commit-graph write with '--split'
such that a roll-up or replacement is performed).
This behavior can cause prohibitively-long commit-graph writes for a
variety of reasons:
* There may be lots of filters whose diffs take a long time to
generate (for example, they have close to the maximum number of
changes, diffing itself takes a long time, etc).
* Old-style commit-graphs (which encode filters with too many entries
as not having been computed at all) cause us to waste time
recomputing filters that appear to have not been computed only to
discover that they are too-large.
This can make the upper-bound of the time it takes for 'git commit-graph
write --changed-paths' to be rather unpredictable.
To make this command behave more predictably, introduce
'--max-new-filters=<n>' to allow computing at most '<n>' Bloom filters
from scratch. This lets "computing" already-known filters proceed
quickly, while bounding the number of slow tasks that Git is willing to
do.
Helped-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
When a changed-path Bloom filter has either zero, or more than a
certain number (commonly 512) of entries, the commit-graph machinery
encodes it as "missing". More specifically, it sets the indices adjacent
in the BIDX chunk as equal to each other to indicate a "length 0"
filter; that is, that the filter occupies zero bytes on disk.
This has heretofore been fine, since the commit-graph machinery has no
need to care about these filters with too few or too many changed paths.
Both cases act like no filter has been generated at all, and so there is
no need to store them.
In a subsequent commit, however, the commit-graph machinery will learn
to only compute Bloom filters for some commits in the current
commit-graph layer. This is a change from the current implementation
which computes Bloom filters for all commits that are in the layer being
written. Critically for this patch, only computing some of the Bloom
filters means adding a third state for length 0 Bloom filters: zero
entries, too many entries, or "hasn't been computed".
It will be important for that future patch to distinguish between "not
representable" (i.e., zero or too-many changed paths), and "hasn't been
computed". In particular, we don't want to waste time recomputing
filters that have already been computed.
To that end, change how we store Bloom filters in the "computed but not
representable" category:
- Bloom filters with no entries are stored as a single byte with all
bits low (i.e., all queries to that Bloom filter will return
"definitely not")
- Bloom filters with too many entries are stored as a single byte with
all bits set high (i.e., all queries to that Bloom filter will
return "maybe").
These rules are sufficient to not incur a behavior change by changing
the on-disk representation of these two classes. Likewise, no
specification changes are necessary for the commit-graph format, either:
- Filters that were previously empty will be recomputed and stored
according to the new rules, and
- old clients reading filters generated by new clients will interpret
the filters correctly and be none the wiser to how they were
generated.
Clients will invoke the Bloom machinery in more cases than before, but
this can be addressed by returning a NULL filter when all bits are set
high. This can be addressed in a future patch.
Note that this does increase the size of on-disk commit-graphs, but far
less than other proposals. In particular, this is generally more
efficient than storing a bitmap for which commits haven't computed their
Bloom filters. Storing a bitmap incurs a penalty of one bit per commit,
whereas storing explicit filters as above incurs a penalty of one byte
per too-large or empty commit.
In practice, these boundary commits likely occupy a small proportion of
the overall number of commits, and so the size penalty is likely smaller
than storing a bitmap for all commits.
See, for example, these relative proportions of such boundary commits
(collected by SZEDER Gábor):
| Percentage of | commit-graph | |
| commits modifying | file size | |
├────────┬──────────────┼───────────────────┤ pct. |
| 0 path | >= 512 paths | before | after | change |
┌────────────────┼────────┼──────────────┼─────────┼─────────┼───────────┤
| android-base | 13.20% | 0.13% | 37.468M | 37.534M | +0.1741 % |
| cmssw | 0.15% | 0.23% | 17.118M | 17.119M | +0.0091 % |
| cpython | 3.07% | 0.01% | 7.967M | 7.971M | +0.0423 % |
| elasticsearch | 0.70% | 1.00% | 8.833M | 8.835M | +0.0128 % |
| gcc | 0.00% | 0.08% | 16.073M | 16.074M | +0.0030 % |
| gecko-dev | 0.14% | 0.64% | 59.868M | 59.874M | +0.0105 % |
| git | 0.11% | 0.02% | 3.895M | 3.895M | +0.0020 % |
| glibc | 0.02% | 0.10% | 3.555M | 3.555M | +0.0021 % |
| go | 0.00% | 0.07% | 3.186M | 3.186M | +0.0018 % |
| homebrew-cask | 0.40% | 0.02% | 7.035M | 7.035M | +0.0065 % |
| homebrew-core | 0.01% | 0.01% | 11.611M | 11.611M | +0.0002 % |
| jdk | 0.26% | 5.64% | 5.537M | 5.540M | +0.0590 % |
| linux | 0.01% | 0.51% | 63.735M | 63.740M | +0.0073 % |
| llvm-project | 0.12% | 0.03% | 25.515M | 25.516M | +0.0050 % |
| rails | 0.10% | 0.10% | 6.252M | 6.252M | +0.0027 % |
| rust | 0.07% | 0.17% | 9.364M | 9.364M | +0.0033 % |
| tensorflow | 0.09% | 1.02% | 7.009M | 7.010M | +0.0158 % |
| webkit | 0.05% | 0.31% | 17.405M | 17.406M | +0.0047 % |
(where the above increase is determined by computing a non-split
commit-graph before and after this patch).
Given that these projects are all "large" by commit count, the storage
cost by writing these filters explicitly is negligible. In the most
extreme example, android-base (which has 494,848 commits at the time of
writing) would have its commit-graph increase by a modest 68.4 KB.
Finally, a test to exercise filters which contain too many changed path
entries will be introduced in a subsequent patch.
Suggested-by: SZEDER Gábor <szeder.dev@gmail.com>
Suggested-by: Jakub Narębski <jnareb@gmail.com>
Helped-by: Derrick Stolee <dstolee@microsoft.com>
Helped-by: SZEDER Gábor <szeder.dev@gmail.com>
Helped-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Commit e3696980 (diff: halt tree-diff early after max_changes,
2020-03-30) intended to create a mechanism to short-circuit a diff
calculation after a certain number of paths were modified. By
incrementing a "num_changes" counter throughout the recursive
ll_diff_tree_paths(), this was supposed to match the number of changes
that would be written into the changed-path Bloom filters.
Unfortunately, this was not implemented correctly and instead misses
simple cases like file modifications. This then does not stop very
large changed-path filters from being written (unless they add or remove
many files).
To start, change the implementation in ll_diff_tree_paths() to instead
use the global diff_queue_diff struct's 'nr' member as the count. This
is a way to simplify the logic instead of making more mistakes in the
complicated diff code.
This has a drawback: the diff_queue_diff struct only lists the paths
corresponding to blob changes, not their leading directories. Thus,
get_or_compute_bloom_filter() needs an additional check to see if the
hashmap with the leading directories becomes too large.
One reason why this was not caught by test cases was that the test in
t4216-log-bloom.sh that was supposed to check this "too many changes"
condition only checked this on the initial commit of a repository. The
old logic counted these values correctly. Update this test in a few
ways:
1. Use GIT_TEST_BLOOM_SETTINGS_MAX_CHANGED_PATHS to reduce the limit,
allowing smaller commits to engage with this logic.
2. Create several interesting cases of edits, adds, removes, and mode
changes (in the second commit). By testing both sides of the
inequality with the *_MAX_CHANGED_PATHS variable, we can see that
the count is exactly correct, so none of these changes are missed
or over-counted.
3. Use the trace2 data value filter_found_large to verify that these
commits are on the correct side of the limit.
Another way to verify the behavior is correct is through performance
tests. By testing on my local copies of the Git repository and the Linux
kernel repository, I could measure the effect of these short-circuits
when computing a fresh commit-graph file with changed-path Bloom filters
using the command
GIT_TEST_BLOOM_SETTINGS_MAX_CHANGED_PATHS=N time \
git commit-graph write --reachable --changed-paths
and reporting the wall time and resulting commit-graph size.
For Git, the results are
| | N=1 | N=10 | N=512 |
|--------|----------------|----------------|----------------|
| HEAD~1 | 10.90s 9.18MB | 11.11s 9.34MB | 11.31s 9.35MB |
| HEAD | 9.21s 8.62MB | 11.11s 9.29MB | 11.29s 9.34MB |
For Linux, the results are
| | N=1 | N=20 | N=512 |
|--------|----------------|---------------|---------------|
| HEAD~1 | 61.28s 64.3MB | 76.9s 72.6MB | 77.6s 72.6MB |
| HEAD | 49.44s 56.3MB | 68.7s 65.9MB | 69.2s 65.9MB |
Naturally, the improvement becomes much less as the limit grows, as
fewer commits satisfy the short-circuit.
Reported-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
For now, we assume that there is a fixed constant describing the
maximum number of changed paths we are willing to store in a Bloom
filter.
Prepare for that to (at least partially) not be the case by making it a
member of the 'struct bloom_filter_settings'. This will be helpful in
the subsequent patches by reducing the size of test cases that exercise
storing too many changed paths, as well as preparing for an eventual
future in which this value might change.
This patch alone does not cause newly generated Bloom filters to use
a custom upper-bound on the maximum number of changed paths a single
Bloom filter can hold, that will occur in a later patch.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Git uses the 'core.commitGraph' configuration value to control whether
or not the commit graph is used when parsing commits or performing a
traversal.
Now that commit-graphs can also contain a section for changed-path Bloom
filters, administrators that already have commit-graphs may find it
convenient to use those graphs without relying on their changed-path
Bloom filters. This can happen, for example, during a staged roll-out,
or in the event of an incident.
Introduce 'commitGraph.readChangedPaths' to control whether or not Bloom
filters are read. Note that this configuration is independent from both:
- 'core.commitGraph', to allow flexibility in using all parts of a
commit-graph _except_ for its Bloom filters.
- The '--changed-paths' option for 'git commit-graph write', to allow
reading and writing Bloom filters to be controlled independently.
When the variable is set, pretend as if no Bloom data was specified at
all. This avoids adding additional special-casing outside of the
commit-graph internals.
Suggested-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
In a759bfa9ee (t4216: add end to end tests for git log with Bloom
filters, 2020-04-06), a 'rm' invocation was added without a
corresponding '&&' chain.
When 'trace.perf' already exists, everything works fine. However, the
function can be executed without 'trace.perf' on disk (eg., when the
subset of tests run is altered with '--run'), and so the bare 'rm'
complains about a missing file.
To remove some noise from the test log, invoke 'rm' with '-f', at which
point it is sensible to place the 'rm -f' in an '&&'-chain, which is
both (1) our usual style, and (2) avoids a broken chain in the future if
more commands are added at the beginning of the function.
Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Many places in the code often need a pointer to the commit-graph's
'struct bloom_filter_settings', in which case they often take the value
from the top-most commit-graph.
In the non-split case, this works as expected. In the split case,
however, things get a little tricky. Not all layers in a chain of
incremental commit-graphs are required to themselves have Bloom data,
and so whether or not some part of the code uses Bloom filters depends
entirely on whether or not the top-most level of the commit-graph chain
has Bloom filters.
This has been the behavior since Bloom filters were introduced, and has
been codified into the tests since a759bfa9ee (t4216: add end to end
tests for git log with Bloom filters, 2020-04-06). In fact, t4216.130
requires that Bloom filters are not used in exactly the case described
earlier.
There is no reason that this needs to be the case, since it is perfectly
valid for commits in an earlier layer to have Bloom filters when commits
in a newer layer do not.
Since Bloom settings are guaranteed in practice to be the same for any
layer in a chain that has Bloom data, it is sufficient to traverse the
'->base_graph' pointer until either (1) a non-null 'struct
bloom_filter_settings *' is found, or (2) until we are at the root of
the commit-graph chain.
Introduce a 'get_bloom_filter_settings()' function that does just this,
and use it instead of purely dereferencing the top-most graph's
'->bloom_filter_settings' pointer.
While we're at it, add an additional test in t5324 to guard against code
in the commit-graph writing machinery that doesn't correctly handle a
NULL 'struct bloom_filter *'.
Co-authored-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The commit-graph format reserved a byte among the header of the file to
store a "hash version". During the SHA-256 work, this was not modified
because file formats are not necessarily intended to work across hash
versions. If a repository has SHA-256 as its hash algorithm, it
automatically up-shifts the lengths of object names in all necessary
formats.
However, since we have this byte available for adjusting the version, we
can make the file formats more obviously incompatible instead of relying
on other context from the repository.
Update the oid_version() method in commit-graph.c to add a new value, 2,
for sha-256. This automatically writes the new value in a SHA-256
repository _and_ verifies the value is correct. This is a breaking
change relative to the current 'master' branch since 092b677 (Merge
branch 'bc/sha-256-cvs-svn-updates', 2020-08-13) but it is not breaking
relative to any released version of Git.
The test impact is relatively minor: the output of 'test-tool
read-graph' lists the header information, so those instances of '1' need
to be replaced with a variable determined by GIT_TEST_DEFAULT_HASH. A
more careful test is added that specifically creates a repository of
each type then swaps the commit-graph files. The important value here is
that the "git log" command succeeds while writing a message to stderr.
Helped-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Reviewed-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Updates to the changed-paths bloom filter.
* ds/commit-graph-bloom-updates:
commit-graph: check all leading directories in changed path Bloom filters
revision: empty pathspecs should not use Bloom filters
revision.c: fix whitespace
commit-graph: check chunk sizes after writing
commit-graph: simplify chunk writes into loop
commit-graph: unify the signatures of all write_graph_chunk_*() functions
commit-graph: persist existence of changed-paths
bloom: fix logic in get_bloom_filter()
commit-graph: change test to die on parse, not load
commit-graph: place bloom_settings in context
|
|
The file 'dir/subdir/file' can only be modified if its leading
directories 'dir' and 'dir/subdir' are modified as well.
So when checking modified path Bloom filters looking for commits
modifying a path with multiple path components, then check not only
the full path in the Bloom filters, but all its leading directories as
well. Take care to check these paths in "deepest first" order,
because it's the full path that is least likely to be modified, and
the Bloom filter queries can short circuit sooner.
This can significantly reduce the average false positive rate, by
about an order of magnitude or three(!), and can further speed up
pathspec-limited revision walks. The table below compares the average
false positive rate and runtime of
git rev-list HEAD -- "$path"
before and after this change for 5000+ randomly* selected paths from
each repository:
Average false Average Average
positive rate runtime runtime
before after before after difference
------------------------------------------------------------------
git 3.220% 0.7853% 0.0558s 0.0387s -30.6%
linux 2.453% 0.0296% 0.1046s 0.0766s -26.8%
tensorflow 2.536% 0.6977% 0.0594s 0.0420s -29.2%
*Path selection was done with the following pipeline:
git ls-tree -r --name-only HEAD | sort -R | head -n 5000
The improvements in runtime are much smaller than the improvements in
average false positive rate, as we are clearly reaching diminishing
returns here. However, all these timings depend on that accessing
tree objects is reasonably fast (warm caches). If we had a partial
clone and the tree objects had to be fetched from a promisor remote,
e.g.:
$ git clone --filter=tree:0 --bare file://.../webkit.git webkit.notrees.git
$ git -C webkit.git -c core.modifiedPathBloomFilters=1 \
commit-graph write --reachable
$ cp webkit.git/objects/info/commit-graph webkit.notrees.git/objects/info/
$ git -C webkit.notrees.git -c core.modifiedPathBloomFilters=1 \
rev-list HEAD -- "$path"
then checking all leading path component can reduce the runtime from
over an hour to a few seconds (and this is with the clone and the
promisor on the same machine).
This adjusts the tracing values in t4216-log-bloom.sh, which provides a
concrete way to notice the improvement.
Helped-by: Taylor Blau <me@ttaylorr.com>
Helped-by: René Scharfe <l.s.r@web.de>
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The prepare_to_use_bloom_filter() method was not intended to be called
on an empty pathspec. However, 'git log -- .' and 'git log' are subtly
different: the latter reports all commits while the former will simplify
commits that do not change the root tree.
This means that the path used to construct the bloom_key might be empty,
and that value is not added to the Bloom filter during construction.
That means that the results are likely incorrect!
To resolve the issue, be careful about the length of the path and stop
filling Bloom filters. To be completely sure we do not use them, drop
the pointer to the bloom_filter_settings from the commit-graph. That
allows our test to look at the trace2 logs to verify no Bloom filter
statistics are reported.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The changed-path Bloom filters were released in v2.27.0, but have a
significant drawback. A user can opt-in to writing the changed-path
filters using the "--changed-paths" option to "git commit-graph write"
but the next write will drop the filters unless that option is
specified.
This becomes even more important when considering the interaction with
gc.writeCommitGraph (on by default) or fetch.writeCommitGraph (part of
features.experimental). These config options trigger commit-graph writes
that the user did not signal, and hence there is no --changed-paths
option available.
Allow a user that opts-in to the changed-path filters to persist the
property of "my commit-graph has changed-path filters" automatically. A
user can drop filters using the --no-changed-paths option.
In the process, we need to be extremely careful to match the Bloom
filter settings as specified by the commit-graph. This will allow future
versions of Git to customize these settings, and the version with this
change will persist those settings as commit-graphs are rewritten on
top.
Use the trace2 API to signal the settings used during the write, and
check that output in a test after manually adjusting the correct bytes
in the commit-graph file.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
The get_bloom_filter() method is a bit complicated in some parts where
it does not need to be. In particular, it needs to return a NULL filter
only when compute_if_not_present is zero AND the filter data cannot be
loaded from a commit-graph file. This currently happens by accident
because the commit-graph does not load changed-path Bloom filters from
an existing commit-graph when writing a new one. This will change in a
later patch.
Also clean up some style issues while we are here.
One side-effect of returning a NULL filter is that the filters that are
reported as "too large" will now be reported as NULL insead of length
zero. This case was not properly covered before, so add a test. Further,
remote the counting of the zero-length filters from revision.c and the
trace2 logs.
Helped-by: René Scharfe <l.s.r@web.de>
Helped-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
Test fix.
* gs/commit-graph-path-filter:
t4216: avoid unnecessary subshell in test_bloom_filters_not_used
|
|
Seems to trigger a bug in at least OpenBSD's 6.7 sh where it is
interpreted as a history lookup and therefore fails 125-126, 128,
130.
Remove the subshell and get a space between ! and grep, so tests
pass successfully.
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
* We need a `final_new_line` to make our source code as text file, per
POSIX and C specification.
* `bloom_filters` should be limited to interal linkage only
Signed-off-by: Đoàn Trần Công Danh <congdanhqx@gmail.com>
Signed-off-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
|
These tests exercises writing commit graph with Bloom filters
and exercises 'git log -- path' with all the applicable
options. They check that the output is the same with and
without Bloom filters, confirm Bloom filters were used by
checking if trace2 statistics were logged correctly.
Also confirms cases where Bloom filters are not used:
1. Multiple path specs,
2. --walk-reflogs (see patch titled 'revision.c: use Bloom filters...'
for details,
3. If the latest commit graph does not have Bloom filters
Signed-off-by: Garima Singh <garima.singh@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
|
Junio C Hamano
Taylor Blau
Derrick Stolee
SZEDER Gábor
Carlo Marcelo Arenas Belón
Đoàn Trần Công Danh
Garima Singh