When this code was migrated to the string_list API in
d88b14b3fd (commit-graph: use string-list API for input, 2018-06-27)
it was made to use used both STRING_LIST_INIT_NODUP and a
strbuf_detach() pattern.
Those should not be used together if string_list_clear() is expected
to free the memory, instead we need to either use STRING_LIST_INIT_DUP
with a string_list_append_nodup(), or a STRING_LIST_INIT_NODUP and
manually fiddle with the "strdup_strings" member before calling
string_list_clear(). Let's do the former.
Since "strdup_strings = 1" is set now other code might be broken by
relying on "pack_indexes" not to duplicate it strings, but that
doesn't happen. When we pass this down to write_commit_graph() that
code uses the "struct string_list" without modifying it. Let's add a
"const" to the variable to have the compiler enforce that assumption.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When queueing references in git-rev-list(1), we try to optimize parsing
of commits via the commit-graph. To do so, we first look up the object's
type, and if it is a commit we call `repo_parse_commit()` instead of
`parse_object()`. This is quite inefficient though given that we're
always uncompressing the object header in order to determine the type.
Instead, we can opportunistically search the commit-graph for the object
ID: in case it's found, we know it's a commit and can directly fill in
the commit object without having to uncompress the object header.
Expose a new function `lookup_commit_in_graph()`, which tries to find a
commit in the commit-graph by ID, and convert `get_reference()` to use
this function. This provides a big performance win in cases where we
load references in a repository with lots of references pointing to
commits. The following has been executed in a real-world repository with
about 2.2 million refs:
Benchmark #1: HEAD~: rev-list --unsorted-input --objects --quiet --not --all --not $newrev
Time (mean ± σ): 4.458 s ± 0.044 s [User: 4.115 s, System: 0.342 s]
Range (min … max): 4.409 s … 4.534 s 10 runs
Benchmark #2: HEAD: rev-list --unsorted-input --objects --quiet --not --all --not $newrev
Time (mean ± σ): 3.089 s ± 0.015 s [User: 2.768 s, System: 0.321 s]
Range (min … max): 3.061 s … 3.105 s 10 runs
Summary
'HEAD: rev-list --unsorted-input --objects --quiet --not --all --not $newrev' ran
1.44 ± 0.02 times faster than 'HEAD~: rev-list --unsorted-input --objects --quiet --not --all --not $newrev'
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We have two established generation number versions:
1: topological levels
2: corrected commit dates
The corrected commit dates are enabled by default, but they also write
extra data in the GDAT and GDOV chunks. Services that host Git data
might want to have more control over when this feature rolls out than
just updating the Git binaries.
Add a new "commitGraph.generationVersion" config option that specifies
the intended generation number version. If this value is less than 2,
then the GDAT chunk is never written _or read_ from an existing file.
This can replace our use of the GIT_TEST_COMMIT_GRAPH_NO_GDAT
environment variable in the test suite. Remove it.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
091f4cf (commit: don't use generation numbers if not needed,
2018-08-30) changed paint_down_to_common() to use commit dates instead
of generation numbers v1 (topological levels) as the performance
regressed on certain topologies. With generation number v2 (corrected
commit dates) implemented, we no longer have to rely on commit dates and
can use generation numbers.
For example, the command `git merge-base v4.8 v4.9` on the Linux
repository walks 167468 commits, taking 0.135s for committer date and
167496 commits, taking 0.157s for corrected committer date respectively.
While using corrected commit dates, Git walks nearly the same number of
commits as commit date, the process is slower as for each comparision we
have to access a commit-slab (for corrected committer date) instead of
accessing struct member (for committer date).
This change incidentally broke the fragile t6404-recursive-merge test.
t6404-recursive-merge sets up a unique repository where all commits have
the same committer date without a well-defined merge-base.
While running tests with GIT_TEST_COMMIT_GRAPH unset, we use committer
date as a heuristic in paint_down_to_common(). 6404.1 'combined merge
conflicts' merges commits in the order:
- Merge C with B to form an intermediate commit.
- Merge the intermediate commit with A.
With GIT_TEST_COMMIT_GRAPH=1, we write a commit-graph and subsequently
use the corrected committer date, which changes the order in which
commits are merged:
- Merge A with B to form an intermediate commit.
- Merge the intermediate commit with C.
While resulting repositories are equivalent, 6404.4 'virtual trees were
processed' fails with GIT_TEST_COMMIT_GRAPH=1 as we are selecting
different merge-bases and thus have different object ids for the
intermediate commits.
As this has already causes problems (as noted in 859fdc0 (commit-graph:
define GIT_TEST_COMMIT_GRAPH, 2018-08-29)), we disable commit graph
within t6404-recursive-merge.
Signed-off-by: Abhishek Kumar <abhishekkumar8222@gmail.com>
Reviewed-by: Taylor Blau <me@ttaylorr.com>
Reviewed-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since there are released versions of Git that understand generation
numbers in the commit-graph's CDAT chunk but do not understand the GDAT
chunk, the following scenario is possible:
1. "New" Git writes a commit-graph with the GDAT chunk.
2. "Old" Git writes a split commit-graph on top without a GDAT chunk.
If each layer of split commit-graph is treated independently, as it was
the case before this commit, with Git inspecting only the current layer
for chunk_generation_data pointer, commits in the lower layer (one with
GDAT) whould have corrected commit date as their generation number,
while commits in the upper layer would have topological levels as their
generation. Corrected commit dates usually have much larger values than
topological levels. This means that if we take two commits, one from the
upper layer, and one reachable from it in the lower layer, then the
expectation that the generation of a parent is smaller than the
generation of a child would be violated.
It is difficult to expose this issue in a test. Since we _start_ with
artificially low generation numbers, any commit walk that prioritizes
generation numbers will walk all of the commits with high generation
number before walking the commits with low generation number. In all the
cases I tried, the commit-graph layers themselves "protect" any
incorrect behavior since none of the commits in the lower layer can
reach the commits in the upper layer.
This issue would manifest itself as a performance problem in this case,
especially with something like "git log --graph" since the low
generation numbers would cause the in-degree queue to walk all of the
commits in the lower layer before allowing the topo-order queue to write
anything to output (depending on the size of the upper layer).
Therefore, When writing the new layer in split commit-graph, we write a
GDAT chunk only if the topmost layer has a GDAT chunk. This guarantees
that if a layer has GDAT chunk, all lower layers must have a GDAT chunk
as well.
Rewriting layers follows similar approach: if the topmost layer below
the set of layers being rewritten (in the split commit-graph chain)
exists, and it does not contain GDAT chunk, then the result of rewrite
does not have GDAT chunks either.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Abhishek Kumar <abhishekkumar8222@gmail.com>
Reviewed-by: Taylor Blau <me@ttaylorr.com>
Reviewed-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As discovered by Ævar, we cannot increment graph version to
distinguish between generation numbers v1 and v2 [1]. Thus, one of
pre-requistes before implementing generation number v2 was to
distinguish between graph versions in a backwards compatible manner.
We are going to introduce a new chunk called Generation DATa chunk (or
GDAT). GDAT will store corrected committer date offsets whereas CDAT
will still store topological level.
Old Git does not understand GDAT chunk and would ignore it, reading
topological levels from CDAT. New Git can parse GDAT and take advantage
of newer generation numbers, falling back to topological levels when
GDAT chunk is missing (as it would happen with a commit-graph written
by old Git).
We introduce a test environment variable 'GIT_TEST_COMMIT_GRAPH_NO_GDAT'
which forces commit-graph file to be written without generation data
chunk to emulate a commit-graph file written by old Git.
To minimize the space required to store corrrected commit date, Git
stores corrected commit date offsets into the commit-graph file, instea
of corrected commit dates. This saves us 4 bytes per commit, decreasing
the GDAT chunk size by half, but it's possible for the offset to
overflow the 4-bytes allocated for storage. As such overflows are and
should be exceedingly rare, we use the following overflow management
scheme:
We introduce a new commit-graph chunk, Generation Data OVerflow ('GDOV')
to store corrected commit dates for commits with offsets greater than
GENERATION_NUMBER_V2_OFFSET_MAX.
If the offset is greater than GENERATION_NUMBER_V2_OFFSET_MAX, we set
the MSB of the offset and the other bits store the position of corrected
commit date in GDOV chunk, similar to how Extra Edge List is maintained.
We test the overflow-related code with the following repo history:
F - N - U
/ \
U - N - U N
\ /
N - F - N
Where the commits denoted by U have committer date of zero seconds
since Unix epoch, the commits denoted by N have committer date of
1112354055 (default committer date for the test suite) seconds since
Unix epoch and the commits denoted by F have committer date of
(2 ^ 31 - 2) seconds since Unix epoch.
The largest offset observed is 2 ^ 31, just large enough to overflow.
[1]: https://lore.kernel.org/git/87a7gdspo4.fsf@evledraar.gmail.com/
Signed-off-by: Abhishek Kumar <abhishekkumar8222@gmail.com>
Reviewed-by: Taylor Blau <me@ttaylorr.com>
Reviewed-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a preparatory step for introducing corrected commit dates, let's
return timestamp_t values from commit_graph_generation(), use
timestamp_t for local variables and define GENERATION_NUMBER_INFINITY
as (2 ^ 63 - 1) instead.
We rename GENERATION_NUMBER_MAX to GENERATION_NUMBER_V1_MAX to
represent the largest topological level we can store in the commit data
chunk.
With corrected commit dates implemented, we will have two such *_MAX
variables to denote the largest offset and largest topological level
that can be stored.
Signed-off-by: Abhishek Kumar <abhishekkumar8222@gmail.com>
Reviewed-by: Taylor Blau <me@ttaylorr.com>
Reviewed-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a later commit we will introduce corrected commit date as the
generation number v2. Corrected commit dates will be stored in the new
seperate Generation Data chunk. However, to ensure backwards
compatibility with "Old" Git we need to continue to write generation
number v1 (topological levels) to the commit data chunk. Thus, we need
to compute and store both versions of generation numbers to write the
commit-graph file.
Therefore, let's introduce a commit-slab `topo_level_slab` to store
topological levels; corrected commit date will be stored in the member
`generation` of struct commit_graph_data.
The macros `GENERATION_NUMBER_INFINITY` and `GENERATION_NUMBER_ZERO`
mark commits not in the commit-graph file and commits written by a
version of Git that did not compute generation numbers respectively.
Generation numbers are computed identically for both kinds of commits.
A "slab-miss" should return `GENERATION_NUMBER_INFINITY` as the commit
is not in the commit-graph file. However, since the slab is
zero-initialized, it returns 0 (or rather `GENERATION_NUMBER_ZERO`).
Thus, we no longer need to check if the topological level of a commit is
`GENERATION_NUMBER_INFINITY`.
We will add a pointer to the slab in `struct write_commit_graph_context`
and `struct commit_graph` to populate the slab in
`fill_commit_graph_info` if the commit has a pre-computed topological
level as in case of split commit-graphs.
Signed-off-by: Abhishek Kumar <abhishekkumar8222@gmail.com>
Reviewed-by: Taylor Blau <me@ttaylorr.com>
Reviewed-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"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 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>
In the subsequent commit, additional options will be added to the
commit-graph API which have nothing to do with splitting.
Rename the 'split_commit_graph_opts' structure to the more-generic
'commit_graph_opts' to encompass both. Likewise, rename the 'flags'
member to instead be 'split_flags' to clarify that it only has to do
with the behavior implied by '--split'.
Suggested-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The first new task in the 'git maintenance' builtin is the
'commit-graph' task. This updates the commit-graph file
incrementally with the command
git commit-graph write --reachable --split
By writing an incremental commit-graph file using the "--split"
option we minimize the disruption from this operation. The default
behavior is to merge layers until the new "top" layer is less than
half the size of the layer below. This provides quick writes most
of the time, with the longer writes following a power law
distribution.
Most importantly, concurrent Git processes only look at the
commit-graph-chain file for a very short amount of time, so they
will verly likely not be holding a handle to the file when we try
to replace it. (This only matters on Windows.)
If a concurrent process reads the old commit-graph-chain file, but
our job expires some of the .graph files before they can be read,
then those processes will see a warning message (but not fail).
This could be avoided by a future update to use the --expire-time
argument when writing the commit-graph.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a future commit, some commit-graph internals will want access to
'r->settings', but we only have the 'struct object_directory *'
corresponding to that repository.
Add an additional parameter to pass the repository around in more
places.
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>
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 changed-path Bloom filter is improved using ideas from an
independent implementation.
* sg/commit-graph-cleanups:
commit-graph: simplify write_commit_graph_file() #2
commit-graph: simplify write_commit_graph_file() #1
commit-graph: simplify parse_commit_graph() #2
commit-graph: simplify parse_commit_graph() #1
commit-graph: clean up #includes
diff.h: drop diff_tree_oid() & friends' return value
commit-slab: add a function to deep free entries on the slab
commit-graph-format.txt: all multi-byte numbers are in network byte order
commit-graph: fix parsing the Chunk Lookup table
tree-walk.c: don't match submodule entries for 'submod/anything'
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>
43d3561 (commit-graph write: don't die if the existing graph is corrupt,
2019-03-25) introduced the GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD environment
variable. This was created to verify that commit-graph was not loaded
when writing a new non-incremental commit-graph.
An upcoming change wants to load a commit-graph in some valuable cases,
but we want to maintain that we don't trust the commit-graph data when
writing our new file. Instead of dying on load, instead die if we ever
try to parse a commit from the commit-graph. This functionally verifies
the same intended behavior, but allows a more advanced feature in the
next change.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The struct commit is used in many contexts. However, members
`generation` and `graph_pos` are only used for commit-graph related
operations and otherwise waste memory.
This wastage would have been more pronounced as we transition to
generation number v2, which uses 64-bit generation number instead of
current 32-bits.
As they are often accessed together, let's introduce struct
commit_graph_data and move them to a commit_graph_data slab.
While the overall test suite runs just as fast as master,
(series: 26m48s, master: 27m34s, faster by 2.87%), certain commands
like `git merge-base --is-ancestor` were slowed by 40% as discovered
by Szeder Gábor [1]. After minimizing commit-slab access, the slow down
persists but is closer to 20%.
Derrick Stolee believes the slow down is attributable to the underlying
algorithm rather than the slowness of commit-slab access [2] and we will
follow-up in a later series.
[1]: https://lore.kernel.org/git/20200607195347.GA8232@szeder.dev/
[2]: https://lore.kernel.org/git/13db757a-9412-7f1e-805c-8a028c4ab2b1@gmail.com/
Signed-off-by: Abhishek Kumar <abhishekkumar8222@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Our CodingGuidelines says that it's sufficient to include one of
'git-compat-util.h' and 'cache.h', but both 'commit-graph.c' and
'commit-graph.h' include both. Let's include only 'git-compat-util.h'
to loose a bunch of unnecessary dependencies; but include 'hash.h',
because 'commit-graph.h' does require the definition of 'struct
object_id'.
'commit-graph.h' explicitly includes 'repository.h' and
'string-list.h', but only needs the declaration of a few structs from
them. Drop these includes and forward-declare the necessary structs
instead.
'commit-graph.c' includes 'dir.h', but doesn't actually use anything
from there, so let's drop that #include as well.
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>
Since 7c5c9b9c57 (commit-graph: error out on invalid commit oids in
'write --stdin-commits', 2019-08-05), the commit-graph builtin dies on
receiving non-commit OIDs as input to '--stdin-commits'.
This behavior can be cumbersome to work around in, say, the case of
piping 'git for-each-ref' to 'git commit-graph write --stdin-commits' if
the caller does not want to cull out non-commits themselves. In this
situation, it would be ideal if 'git commit-graph write' wrote the graph
containing the inputs that did pertain to commits, and silently ignored
the remainder of the input.
Some options have been proposed to the effect of '--[no-]check-oids'
which would allow callers to have the commit-graph builtin do just that.
After some discussion, it is difficult to imagine a caller who wouldn't
want to pass '--no-check-oids', suggesting that we should get rid of the
behavior of complaining about non-commit inputs altogether.
If callers do wish to retain this behavior, they can easily work around
this change by doing the following:
git for-each-ref --format='%(objectname) %(objecttype) %(*objecttype)' |
awk '
!/commit/ { print "not-a-commit:"$1 }
/commit/ { print $1 }
' |
git commit-graph write --stdin-commits
To make it so that valid OIDs that refer to non-existent objects are
indeed an error after loosening the error handling, perform an extra
lookup to make sure that object indeed exists before sending it to the
commit-graph internals.
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git blame" learns to take advantage of the "changed-paths" Bloom
filter stored in the commit-graph file.
* ds/blame-on-bloom:
test-bloom: check that we have expected arguments
test-bloom: fix some whitespace issues
blame: drop unused parameter from maybe_changed_path
blame: use changed-path Bloom filters
tests: write commit-graph with Bloom filters
revision: complicated pathspecs disable filters
Introduce an extension to the commit-graph to make it efficient to
check for the paths that were modified at each commit using Bloom
filters.
* gs/commit-graph-path-filter:
bloom: ignore renames when computing changed paths
commit-graph: add GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS test flag
t4216: add end to end tests for git log with Bloom filters
revision.c: add trace2 stats around Bloom filter usage
revision.c: use Bloom filters to speed up path based revision walks
commit-graph: add --changed-paths option to write subcommand
commit-graph: reuse existing Bloom filters during write
commit-graph: write Bloom filters to commit graph file
commit-graph: examine commits by generation number
commit-graph: examine changed-path objects in pack order
commit-graph: compute Bloom filters for changed paths
diff: halt tree-diff early after max_changes
bloom.c: core Bloom filter implementation for changed paths.
bloom.c: introduce core Bloom filter constructs
bloom.c: add the murmur3 hash implementation
commit-graph: define and use MAX_NUM_CHUNKS
We don't ever refer to the descriptor after mmap-ing it. And keeping it
open means we can run out of descriptors in degenerate cases (e.g.,
thousands of split chain files). Let's close it as soon as possible.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The GIT_TEST_COMMIT_GRAPH environment variable updates the commit-
graph file whenever "git commit" is run, ensuring that we always
have an updated commit-graph throughout the test suite. The
GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS environment variable was
introduced to write the changed-path Bloom filters whenever "git
commit-graph write" is run. However, the GIT_TEST_COMMIT_GRAPH
trick doesn't launch a separate process and instead writes it
directly.
To expand the number of tests that have commits in the commit-graph
file, add a helper method that computes the commit-graph and place
that helper inside "git commit" and "git merge".
In the helper method, check GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS
to ensure we are writing changed-path Bloom filters whenever
possible.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The 'write_commit_graph()' function takes in either a string list of
pack indices, or a string list of hexadecimal commit OIDs. These
correspond to the '--stdin-packs' and '--stdin-commits' mode(s) from
'git commit-graph write'.
Using a string_list of hexadecimal commit IDs is not the most efficient
use of memory, since we can instead use the 'struct oidset', which is
more well-suited for this case.
This has another benefit which will become apparent in the following
commit. This is that we are about to disambiguate the kinds of errors we
produce with '--stdin-commits' into "non-hex input" and "hex-input, but
referring to a non-commit object". By having 'write_commit_graph' take
in a 'struct oidset *' of commits, we place the burden on the caller (in
this case, the builtin) to handle the first case, and the commit-graph
machinery can handle the second case.
Suggested-by: Jeff King <peff@peff.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When using split commit-graphs, it is sometimes useful to completely
replace the commit-graph chain with a new base.
For example, consider a scenario in which a repository builds a new
commit-graph incremental for each push. Occasionally (say, after some
fixed number of pushes), they may wish to rebuild the commit-graph chain
with all reachable commits.
They can do so with
$ git commit-graph write --reachable
but this removes the chain entirely and replaces it with a single
commit-graph in 'objects/info/commit-graph'. Unfortunately, this means
that the next push will have to move this commit-graph into the first
layer of a new chain, and then write its new commits on top.
Avoid such copying entirely by allowing the caller to specify that they
wish to replace the entirety of their commit-graph chain, while also
specifying that the new commit-graph should become the basis of a fresh,
length-one chain.
This addresses the above situation by making it possible for the caller
to instead write:
$ git commit-graph write --reachable --split=replace
which writes a new length-one chain to 'objects/info/commit-graphs',
making the commit-graph incremental generated by the subsequent push
relatively cheap by avoiding the aforementioned copy.
In order to do this, remove an assumption in 'write_commit_graph_file'
that chains are always at least two incrementals long.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In the previous commit, we laid the groundwork for supporting different
splitting strategies. In this commit, we introduce the first splitting
strategy: 'no-merge'.
Passing '--split=no-merge' is useful for callers which wish to write a
new incremental commit-graph, but do not want to spend effort condensing
the incremental chain [1]. Previously, this was possible by passing
'--size-multiple=0', but this no longer the case following 63020f175f
(commit-graph: prefer default size_mult when given zero, 2020-01-02).
When '--split=no-merge' is given, the commit-graph machinery will never
condense an existing chain, and it will always write a new incremental.
[1]: This might occur when, for example, a server administrator running
some program after each push may want to ensure that each job runs
proportional in time to the size of the push, and does not "jump" when
the commit-graph machinery decides to trigger a merge.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
With '--split', the commit-graph machinery writes new commits in another
incremental commit-graph which is part of the existing chain, and
optionally decides to condense the chain into a single commit-graph.
This is done to ensure that the asymptotic behavior of looking up a
commit in an incremental chain is not dominated by the number of
incrementals in that chain. It can be controlled by the '--max-commits'
and '--size-multiple' options.
In the next two commits, we will introduce additional splitting
strategies that can exert additional control over:
- when a split commit-graph is and isn't written, and
- when the existing commit-graph chain is discarded completely and
replaced with another graph
To prepare for this, make '--split' take an optional strategy (as in
'--split[=<strategy>]'), and add a new enum to describe which strategy
is being used. For now, no strategies are given, and the only enumerated
value is 'COMMIT_GRAPH_SPLIT_UNSPECIFIED', indicating the absence of a
strategy.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS test flag to the test setup suite
in order to toggle writing Bloom filters when running any of the git tests.
If set to true, we will compute and write Bloom filters every time a test
calls `git commit-graph write`, as if the `--changed-paths` option was
passed in.
The test suite passes when GIT_TEST_COMMIT_GRAPH and
GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS are enabled.
Helped-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Garima Singh <garima.singh@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Update the technical documentation for commit-graph-format with
the formats for the Bloom filter index (BIDX) and Bloom filter
data (BDAT) chunks. Write the computed Bloom filters information
to the commit graph file using this format.
Helped-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Garima Singh <garima.singh@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add new COMMIT_GRAPH_WRITE_CHANGED_PATHS flag that makes Git compute
Bloom filters for the paths that changed between a commit and it's
first parent, for each commit in the commit-graph. This computation
is done on a commit-by-commit basis.
We will write these Bloom filters to the commit-graph file, to store
this data on disk, in the next change in this series.
Helped-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Garima Singh <garima.singh@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Apply a similar treatment as in the previous patch to pass a 'struct
object_directory *' through the 'load_commit_graph_one_fd_st'
initializer, too.
This prevents a potential bug where a pointer comparison is made to a
NULL 'g->odb', which would cause the commit-graph machinery to think
that a pair of commit-graphs belonged to different alternates when in
fact they do not (i.e., in the case of no '--object-dir').
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As of the previous patch, all calls to 'commit-graph.c' functions which
perform path normalization (for e.g., 'get_commit_graph_filename()') are
of the form 'ctx->odb->path', which is always in normalized form.
Now that there are no callers passing non-normalized paths to these
functions, ensure that future callers are bound by the same restrictions
by making these functions take a 'struct object_directory *' instead of
a 'const char *'. To match, replace all calls with arguments of the form
'ctx->odb->path' with 'ctx->odb' To recover the path, functions that
perform path manipulation simply use 'odb->path'.
Further, avoid string comparisons with arguments of the form
'odb->path', and instead prefer raw pointer comparisons, which
accomplish the same effect, but are far less brittle.
This has a pleasant side-effect of making these functions much more
robust to paths that cannot be normalized by 'normalize_path_copy()',
i.e., because they are outside of the current working directory.
For example, prior to this patch, Valgrind reports that the following
uninitialized memory read [1]:
$ ( cd t && GIT_DIR=../.git valgrind git rev-parse HEAD^ )
because 'normalize_path_copy()' can't normalize '../.git' (since it's
relative to but above of the current working directory) [2].
By using a 'struct object_directory *' directly,
'get_commit_graph_filename()' does not need to normalize, because all
paths are relative to the current working directory since they are
always read from the '->path' of an object directory.
[1]: https://lore.kernel.org/git/20191027042116.GA5801@sigill.intra.peff.net.
[2]: The bug here is that 'get_commit_graph_filename()' returns the
result of 'normalize_path_copy()' without checking the return
value.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a previous patch, the 'char *object_dir' in 'struct commit_graph' was
replaced with a 'struct object_directory'. This patch applies the same
treatment to 'struct commit_graph', which is another intermediate step
towards getting rid of all path normalization in 'commit-graph.c'.
Instead of taking a 'char *object_dir', functions that construct a
'struct commit_graph' now take a 'struct object_directory *'. Any code
that needs an object directory path use '->path' instead.
This ensures that all calls to functions that perform path normalization
are given arguments which do not themselves require normalization. This
prepares those functions to drop their normalization entirely, which
will occur in the subsequent patch.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There are lots of places in 'commit-graph.h' where a function either has
(or almost has) a full 'struct object_directory *', accesses '->path',
and then throws away the rest of the struct.
This can cause headaches when comparing the locations of object
directories across alternates (e.g., in the case of deciding if two
commit-graph layers can be merged). These paths are normalized with
'normalize_path_copy()' which mitigates some comparison issues, but not
all [1].
Replace usage of 'char *object_dir' with 'odb->path' by storing a
'struct object_directory *' in the 'write_commit_graph_context'
structure. This is an intermediate step towards getting rid of all path
normalization in 'commit-graph.c'.
Resolving a user-provided '--object-dir' argument now requires that we
compare it to the known alternates for equality. Prior to this patch,
an unknown '--object-dir' argument would silently exit with status zero.
This can clearly lead to unintended behavior, such as verifying
commit-graphs that aren't in a repository's own object store (or one of
its alternates), or causing a typo to mask a legitimate commit-graph
verification failure. Make this error non-silent by 'die()'-ing when the
given '--object-dir' does not match any known alternate object store.
[1]: In my testing, for example, I can get one side of the commit-graph
code to fill object_dir with "./objects" and the other with just
"objects".
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When the client has asked for certain shallow options like
"deepen-since", we do a custom rev-list walk that pretends to be
shallow. Before doing so, we have to disable the commit-graph, since it
is not compatible with the shallow view of the repository. That's
handled by 829a321569 (commit-graph: close_commit_graph before shallow
walk, 2018-08-20). That commit literally closes and frees our
repo->objects->commit_graph struct.
That creates an interesting problem for commits that have _already_ been
parsed using the commit graph. Their commit->object.parsed flag is set,
their commit->graph_pos is set, but their commit->maybe_tree may still
be NULL. When somebody later calls repo_get_commit_tree(), we see that
we haven't loaded the tree oid yet and try to get it from the commit
graph. But since it has been freed, we segfault!
So the root of the issue is a data dependency between the commit's
lazy-load of the tree oid and the fact that the commit graph can go
away mid-process. How can we resolve it?
There are a couple of general approaches:
1. The obvious answer is to avoid loading the tree from the graph when
we see that it's NULL. But then what do we return for the tree oid?
If we return NULL, our caller in do_traverse() will rightly
complain that we have no tree. We'd have to fallback to loading the
actual commit object and re-parsing it. That requires teaching
parse_commit_buffer() to understand re-parsing (i.e., not starting
from a clean slate and not leaking any allocated bits like parent
list pointers).
2. When we close the commit graph, walk through the set of in-memory
objects and clear any graph_pos pointers. But this means we also
have to "unparse" any such commits so that we know they still need
to open the commit object to fill in their trees. So it's no less
complicated than (1), and is more expensive (since we clear objects
we might not later need).
3. Stop freeing the commit-graph struct. Continue to let it be used
for lazy-loads of tree oids, but let upload-pack specify that it
shouldn't be used for further commit parsing.
4. Push the whole shallow rev-list out to its own sub-process, with
the commit-graph disabled from the start, giving it a clean memory
space to work from.
I've chosen (3) here. Options (1) and (2) would work, but are
non-trivial to implement. Option (4) is more expensive, and I'm not sure
how complicated it is (shelling out for the actual rev-list part is
easy, but we do then parse the resulting commits internally, and I'm not
clear which parts need to be handling shallow-ness).
The new test in t5500 triggers this segfault, but see the comments there
for how horribly intimate it has to be with how both upload-pack and
commit graphs work.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
While 'git commit-graph write --stdin-commits' expects commit object
ids as input, it accepts and silently skips over any invalid commit
object ids, and still exits with success:
# nonsense
$ echo not-a-commit-oid | git commit-graph write --stdin-commits
$ echo $?
0
# sometimes I forgot that refs are not good...
$ echo HEAD | git commit-graph write --stdin-commits
$ echo $?
0
# valid tree OID, but not a commit OID
$ git rev-parse HEAD^{tree} | git commit-graph write --stdin-commits
$ echo $?
0
$ ls -l .git/objects/info/commit-graph
ls: cannot access '.git/objects/info/commit-graph': No such file or directory
Check that all input records are indeed valid commit object ids and
return with error otherwise, the same way '--stdin-packs' handles
invalid input; see e103f7276f (commit-graph: return with errors during
write, 2019-06-12).
Note that it should only return with error when encountering an
invalid commit object id coming from standard input. However,
'--reachable' uses the same code path to process object ids pointed to
by all refs, and that includes tag object ids as well, which should
still be skipped over. Therefore add a new flag to 'enum
commit_graph_write_flags' and a corresponding field to 'struct
write_commit_graph_context', so we can differentiate between those two
cases.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Acked-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If we wrote a commit-graph chain, we only modified the tip file in
the chain. It is valuable to verify what we wrote, but not waste
time checking files we did not write.
Add a '--shallow' option to the 'git commit-graph verify' subcommand
and check that it does not read the base graph in a two-file chain.
Making the verify subcommand read from a chain of commit-graphs takes
some rearranging of the builtin code.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The split commit-graph feature is now fully implemented, but needs
some more run-time configurability. Allow direct callers to 'git
commit-graph write --split' to specify the values used in the
merge strategy and the expire time.
Update the documentation to specify these values.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In an environment like a fork network, it is helpful to have a
commit-graph chain that spans both the base repo and the fork repo. The
fork is usually a small set of data on top of the large repo, but
sometimes the fork is much larger. For example, git-for-windows/git has
almost double the number of commits as git/git because it rebases its
commits on every major version update.
To allow cross-alternate commit-graph chains, we need a few pieces:
1. When looking for a graph-{hash}.graph file, check all alternates.
2. When merging commit-graph chains, do not merge across alternates.
3. When writing a new commit-graph chain based on a commit-graph file
in another object directory, do not allow success if the base file
has of the name "commit-graph" instead of
"commit-graphs/graph-{hash}.graph".
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Extend write_commit_graph() to write a commit-graph chain when given the
COMMIT_GRAPH_SPLIT flag.
This implementation is purposefully simplistic in how it creates a new
chain. The commits not already in the chain are added to a new tip
commit-graph file.
Much of the logic around writing a graph-{hash}.graph file and updating
the commit-graph-chain file is the same as the commit-graph file case.
However, there are several places where we need to do some extra logic
in the split case.
Track the list of graph filenames before and after the planned write.
This will be more important when we start merging graph files, but it
also allows us to upgrade our commit-graph file to the appropriate
graph-{hash}.graph file when we upgrade to a chain of commit-graphs.
Note that we use the eighth byte of the commit-graph header to store the
number of base graph files. This determines the length of the base
graphs chunk.
A subtle change of behavior with the new logic is that we do not write a
commit-graph if we our commit list is empty. This extends to the typical
case, which is reflected in t5318-commit-graph.sh.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
To quickly verify a commit-graph chain is valid on load, we will
read from the new "Base Graphs Chunk" of each file in the chain.
This will prevent accidentally loading incorrect data from manually
editing the commit-graph-chain file or renaming graph-{hash}.graph
files.
The commit_graph struct already had an object_id struct "oid", but
it was never initialized or used. Add a line to read the hash from
the end of the commit-graph file and into the oid member.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
To prepare for a chain of commit-graph files, augment the
commit_graph struct to point to a base commit_graph. As we load
commits from the graph, we may actually want to read from a base
file according to the graph position.
The "graph position" of a commit is given by concatenating the
lexicographic commit orders from each of the commit-graph files in
the chain. This means that we must distinguish two values:
* lexicographic index : the position within the lexicographic
order in a single commit-graph file.
* graph position: the position within the concatenated order
of multiple commit-graph files
Given the lexicographic index of a commit in a graph, we can
compute the graph position by adding the number of commits in
the lower-level graphs. To find the lexicographic index of
a commit, we subtract the number of commits in lower-level graphs.
While here, change insert_parent_or_die() to take a uint32_t
position, as that is the type used by its only caller and that
makes more sense with the limits in the commit-graph format.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The close_commit_graph() method took a repository struct, but then
only uses the raw_object_store within. Change the function prototype
to make the method more flexible.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The write_commit_graph() and write_commit_graph_reachable() methods
currently take two boolean parameters: 'append' and 'report_progress'.
As we update these methods, adding more parameters this way becomes
cluttered and hard to maintain.
Collapse these parameters into a 'flags' parameter, and adjust the
callers to provide flags as necessary.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The write_commit_graph() method uses die() to report failure and
exit when confronted with an unexpected condition. This use of
die() in a library function is incorrect and is now replaced by
error() statements and an int return type. Return zero on success
and a negative value on failure.
Now that we use 'goto cleanup' to jump to the terminal condition
on an error, we have new paths that could lead to uninitialized
values. New initializers are added to correct for this.
The builtins 'commit-graph', 'gc', and 'commit' call these methods,
so update them to check the return value. Test that 'git commit-graph
write' returns a proper error code when hitting a failure condition
in write_commit_graph().
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When the commit-graph is written we end up calling
parse_commit(). This will in turn invoke code that'll consult the
existing commit-graph about the commit, if the graph is corrupted we
die.
We thus get into a state where a failing "commit-graph verify" can't
be followed-up with a "commit-graph write" if core.commitGraph=true is
set, the graph either needs to be manually removed to proceed, or
core.commitGraph needs to be set to "false".
Change the "commit-graph write" codepath to use a new
parse_commit_no_graph() helper instead of parse_commit() to avoid
this. The latter will call repo_parse_commit_internal() with
use_commit_graph=1 as seen in 177722b344 ("commit: integrate commit
graph with commit parsing", 2018-04-10).
Not using the old graph at all slows down the writing of the new graph
by some small amount, but is a sensible way to prevent an error in the
existing commit-graph from spreading.
Just fixing the current issue would be likely to result in code that's
inadvertently broken in the future. New code might use the
commit-graph at a distance. To detect such cases introduce a
"GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD" setting used when we do our
corruption tests, and test that a "write/verify" combo works after
every one of our current test cases where we now detect commit-graph
corruption.
Some of the code changes here might be strictly unnecessary, e.g. I
was unable to find cases where the parse_commit() called from
write_graph_chunk_data() didn't exit early due to
"item->object.parsed" being true in
repo_parse_commit_internal() (before the use_commit_graph=1 has any
effect). But let's also convert those cases for good measure, we do
not have exhaustive tests for all possible types of commit-graph
corruption.
This might need to be re-visited if we learn to write the commit-graph
incrementally, but probably not. Hopefully we'll just start by finding
out what commits we have in total, then read the old graph(s) to see
what they cover, and finally write a new graph file with everything
that's missing. In that case the new graph writing code just needs to
continue to use e.g. a parse_commit() that doesn't consult the
existing commit-graphs.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
An earlier change implemented load_commit_graph_one_fd_st() in a way
that was bug-compatible with earlier code in terms of the "graph file
%s is too small" error message printing out the path to the
commit-graph (".git/objects/info/commit-graph").
But change that, because:
* A function that takes an already-open file descriptor also needing
the filename isn't very intuitive.
* The vast majority of errors we might emit when loading the graph
come from parse_commit_graph(), which doesn't report the
filename. Let's not do that either in this case for consistency.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
