cf2dc1c238 (speed up alt_odb_usable() with many alternates, 2021-07-07)
introduces a KHASH_INIT invocation with a trailing ';', which while
commonly expected will trigger warnings with pedantic on both
clang[-Wextra-semi] and gcc[-Wpedantic], because that macro has already
a semicolon and is meant to be invoked without one.
while fixing the macro would be a worthy solution (specially considering
this is a common recurring problem), remove the extra ';' for now to
minimize churn.
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This saves 8K per `struct object_directory', meaning it saves
around 800MB in my case involving 100K alternates (half or more
of those alternates are unlikely to hold loose objects).
This is implemented in two parts: a generic, allocation-free
`cbtree' and the `oidtree' wrapper on top of it. The latter
provides allocation using alloc_state as a memory pool to
improve locality and reduce free(3) overhead.
Unlike oid-array, the crit-bit tree does not require sorting.
Performance is bound by the key length, for oidtree that is
fixed at sizeof(struct object_id). There's no need to have
256 oidtrees to mitigate the O(n log n) overhead like we did
with oid-array.
Being a prefix trie, it is natively suited for expanding short
object IDs via prefix-limited iteration in
`find_short_object_filename'.
On my busy workstation, p4205 performance seems to be roughly
unchanged (+/-8%). Startup with 100K total alternates with no
loose objects seems around 10-20% faster on a hot cache.
(800MB in memory savings means more memory for the kernel FS
cache).
The generic cbtree implementation does impose some extra
overhead for oidtree in that it uses memcmp(3) on
"struct object_id" so it wastes cycles comparing 12 extra bytes
on SHA-1 repositories. I've not yet explored reducing this
overhead, but I expect there are many places in our code base
where we'd want to investigate this.
More information on crit-bit trees: https://cr.yp.to/critbit.html
Signed-off-by: Eric Wong <e@80x24.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There's no point in using 8 bits per-directory when 1 bit
will do. This saves us 224 bytes per object directory, which
ends up being 22MB when dealing with 100K alternates.
Signed-off-by: Eric Wong <e@80x24.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
With many alternates, the duplicate check in alt_odb_usable()
wastes many cycles doing repeated fspathcmp() on every existing
alternate. Use a khash to speed up lookups by odb->path.
Since the kh_put_* API uses the supplied key without
duplicating it, we also take advantage of it to replace both
xstrdup() and strbuf_release() in link_alt_odb_entry() with
strbuf_detach() to avoid the allocation and copy.
In a test repository with 50K alternates and each of those 50K
alternates having one alternate each (for a total of 100K total
alternates); this speeds up lookup of a non-existent blob from
over 16 minutes to roughly 2.7 seconds on my busy workstation.
Note: all underlying git object directories were small and
unpacked with only loose objects and no packs. Having to load
packs increases times significantly.
Signed-off-by: Eric Wong <e@80x24.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a recent patch we added a function 'find_kept_pack_entry()' to look
for an object only among kept packs.
While this function avoids doing any lookup work in non-kept packs, it
is still linear in the number of packs, since we have to traverse the
linked list of packs once per object. Let's cache a reduced version of
that list to save us time.
Note that this cache will last the lifetime of the program. We could
invalidate it on reprepare_packed_git(), but there's not much point in
being rigorous here:
- we might already fail to notice new .keep packs showing up after the
program starts. We only reprepare_packed_git() when we fail to find
an object. But adding a new pack won't cause that to happen.
Somebody repacking could add a new pack and delete an old one, but
most of the time we'd have a descriptor or mmap open to the old
pack anyway, so we might not even notice.
- in pack-objects we already cache the .keep state at startup, since
56dfeb6263 (pack-objects: compute local/ignore_pack_keep early,
2016-07-29). So this is just extending that concept further.
- we don't have to worry about any packed_git being removed; we always
keep the old structs around, even after reprepare_packed_git()
We do defensively invalidate the cache in case the set of kept packs
being asked for changes (e.g., only in-core kept packs were cached, but
suddenly the caller also wants on-disk kept packs, too). In theory we
could build all three caches and switch between them, but it's not
necessary, since this patch (and series) never changes the set of kept
packs that it wants to inspect from the cache.
So that "optimization" is more about being defensive in the face of
future changes than it is about asking for multiple kinds of kept packs
in this patch.
Here are p5303 results (as always, measured against the kernel):
Test HEAD^ HEAD
-----------------------------------------------------------------------------------------------
5303.5: repack (1) 57.34(54.66+10.88) 56.98(54.36+10.98) -0.6%
5303.6: repack with kept (1) 57.38(54.83+10.49) 57.17(54.97+10.26) -0.4%
5303.11: repack (50) 71.70(88.99+4.74) 71.62(88.48+5.08) -0.1%
5303.12: repack with kept (50) 72.58(89.61+4.78) 71.56(88.80+4.59) -1.4%
5303.17: repack (1000) 217.19(491.72+14.25) 217.31(490.82+14.53) +0.1%
5303.18: repack with kept (1000) 246.12(520.07+14.93) 217.08(490.37+15.10) -11.8%
and the --stdin-packs case, which scales a little bit better (although
not by that much even at 1,000 packs):
5303.7: repack with --stdin-packs (1) 0.00(0.00+0.00) 0.00(0.00+0.00) =
5303.13: repack with --stdin-packs (50) 3.43(11.75+0.24) 3.43(11.69+0.30) +0.0%
5303.19: repack with --stdin-packs (1000) 130.50(307.15+7.66) 125.13(301.36+8.04) -4.1%
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>
Specify the format of the on-disk reverse index 'pack-*.rev' file, as
well as prepare the code for the existence of such files.
The reverse index maps from pack relative positions (i.e., an index into
the array of object which is sorted by their offsets within the
packfile) to their position within the 'pack-*.idx' file. Today, this is
done by building up a list of (off_t, uint32_t) tuples for each object
(the off_t corresponding to that object's offset, and the uint32_t
corresponding to its position in the index). To convert between pack and
index position quickly, this array of tuples is radix sorted based on
its offset.
This has two major drawbacks:
First, the in-memory cost scales linearly with the number of objects in
a pack. Each 'struct revindex_entry' is sizeof(off_t) +
sizeof(uint32_t) + padding bytes for a total of 16.
To observe this, force Git to load the reverse index by, for e.g.,
running 'git cat-file --batch-check="%(objectsize:disk)"'. When asking
for a single object in a fresh clone of the kernel, Git needs to
allocate 120+ MB of memory in order to hold the reverse index in memory.
Second, the cost to sort also scales with the size of the pack.
Luckily, this is a linear function since 'load_pack_revindex()' uses a
radix sort, but this cost still must be paid once per pack per process.
As an example, it takes ~60x longer to print the _size_ of an object as
it does to print that entire object's _contents_:
Benchmark #1: git.compile cat-file --batch <obj
Time (mean ± σ): 3.4 ms ± 0.1 ms [User: 3.3 ms, System: 2.1 ms]
Range (min … max): 3.2 ms … 3.7 ms 726 runs
Benchmark #2: git.compile cat-file --batch-check="%(objectsize:disk)" <obj
Time (mean ± σ): 210.3 ms ± 8.9 ms [User: 188.2 ms, System: 23.2 ms]
Range (min … max): 193.7 ms … 224.4 ms 13 runs
Instead, avoid computing and sorting the revindex once per process by
writing it to a file when the pack itself is generated.
The format is relatively straightforward. It contains an array of
uint32_t's, the length of which is equal to the number of objects in the
pack. The ith entry in this table contains the index position of the
ith object in the pack, where "ith object in the pack" is determined by
pack offset.
One thing that the on-disk format does _not_ contain is the full (up to)
eight-byte offset corresponding to each object. This is something that
the in-memory revindex contains (it stores an off_t in 'struct
revindex_entry' along with the same uint32_t that the on-disk format
has). Omit it in the on-disk format, since knowing the index position
for some object is sufficient to get a constant-time lookup in the
pack-*.idx file to ask for an object's offset within the pack.
This trades off between the on-disk size of the 'pack-*.rev' file for
runtime to chase down the offset for some object. Even though the lookup
is constant time, the constant is heavier, since it can potentially
involve two pointer walks in v2 indexes (one to access the 4-byte offset
table, and potentially a second to access the double wide offset table).
Consider trying to map an object's pack offset to a relative position
within that pack. In a cold-cache scenario, more page faults occur while
switching between binary searching through the reverse index and
searching through the *.idx file for an object's offset. Sure enough,
with a cold cache (writing '3' into '/proc/sys/vm/drop_caches' after
'sync'ing), printing out the entire object's contents is still
marginally faster than printing its size:
Benchmark #1: git.compile cat-file --batch-check="%(objectsize:disk)" <obj >/dev/null
Time (mean ± σ): 22.6 ms ± 0.5 ms [User: 2.4 ms, System: 7.9 ms]
Range (min … max): 21.4 ms … 23.5 ms 41 runs
Benchmark #2: git.compile cat-file --batch <obj >/dev/null
Time (mean ± σ): 17.2 ms ± 0.7 ms [User: 2.8 ms, System: 5.5 ms]
Range (min … max): 15.6 ms … 18.2 ms 45 runs
(Numbers taken in the kernel after cheating and using the next patch to
generate a reverse index). There are a couple of approaches to improve
cold cache performance not pursued here:
- We could include the object offsets in the reverse index format.
Predictably, this does result in fewer page faults, but it triples
the size of the file, while simultaneously duplicating a ton of data
already available in the .idx file. (This was the original way I
implemented the format, and it did show
`--batch-check='%(objectsize:disk)'` winning out against `--batch`.)
On the other hand, this increase in size also results in a large
block-cache footprint, which could potentially hurt other workloads.
- We could store the mapping from pack to index position in more
cache-friendly way, like constructing a binary search tree from the
table and writing the values in breadth-first order. This would
result in much better locality, but the price you pay is trading
O(1) lookup in 'pack_pos_to_index()' for an O(log n) one (since you
can no longer directly index the table).
So, neither of these approaches are taken here. (Thankfully, the format
is versioned, so we are free to pursue these in the future.) But, cold
cache performance likely isn't interesting outside of one-off cases like
asking for the size of an object directly. In real-world usage, Git is
often performing many operations in the revindex (i.e., asking about
many objects rather than a single one).
The trade-off is worth it, since we will avoid the vast majority of the
cost of generating the revindex that the extra pointer chase will look
like noise in the following patch's benchmarks.
This patch describes the format and prepares callers (like in
pack-revindex.c) to be able to read *.rev files once they exist. An
implementation of the writer will appear in the next patch, and callers
will gradually begin to start using the writer in the patches that
follow after that.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There have been a few bugs wherein Git fetches missing objects whenever
the existence of an object is checked, even though it does not need to
perform such a fetch. To resolve these bugs, we could look at all the
places that has_object_file() (or a similar function) is used. As a
first step, introduce a new function has_object() that checks for the
existence of an object, with a default behavior of not fetching if the
object is missing and the repository is a partial clone. As we verify
each has_object_file() (or similar) usage, we can replace it with
has_object(), and we will know that we are done when we can delete
has_object_file() (and the other similar functions).
Also, the new function has_object() has more appropriate defaults:
besides not fetching, it also does not recheck packed storage.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Both v2 pack index files and the v3 format specified as part of the
NewHash work have similar data starting at the CRC table. Much of the
existing code wants to read either this table or the offset entries
following it, and in doing so computes the offset each time.
In order to share as much code between v2 and v3, compute the offset of
the CRC table and store it when the pack is opened. Use this value to
compute offsets to not only the CRC table, but to the offset entries
beyond it.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We renamed the actual data structure in 910650d2f8 (Rename sha1_array to
oid_array, 2017-03-31), but the file is still called sha1-array. Besides
being slightly confusing, it makes it more annoying to grep for leftover
occurrences of "sha1" in various files, because the header is included
in so many places.
Let's complete the transition by renaming the source and header files
(and fixing up a few comment references).
I kept the "-" in the name, as that seems to be our style; cf.
fc1395f4a4 (sha1_file.c: rename to use dash in file name, 2018-04-10).
We also have oidmap.h and oidset.h without any punctuation, but those
are "struct oidmap" and "struct oidset" in the code. We _could_ make
this "oidarray" to match, but somehow it looks uglier to me because of
the length of "array" (plus it would be a very invasive patch for little
gain).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If a caller sets the object_info.delta_base_sha1 to a non-NULL pointer,
we'll write the oid of the object's delta base to it. But we can
increase our type safety by switching this to a real object_id struct.
All of our callers are just pointing into the hash member of an
object_id anyway, so there's no inconvenience.
Note that we do still keep it as a pointer-to-struct, because the NULL
sentinel value tells us whether the caller is even interested in the
information.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Some codepaths were given a repository instance as a parameter to
work in the repository, but passed the_repository instance to its
callees, which has been cleaned up (somewhat).
* mt/use-passed-repo-more-in-funcs:
sha1-file: allow check_object_signature() to handle any repo
sha1-file: pass git_hash_algo to hash_object_file()
sha1-file: pass git_hash_algo to write_object_file_prepare()
streaming: allow open_istream() to handle any repo
pack-check: use given repo's hash_algo at verify_packfile()
cache-tree: use given repo's hash_algo at verify_one()
diff: make diff_populate_filespec() honor its repo argument
Traditionally, we avoided threaded grep while searching in objects
(as opposed to files in the working tree) as accesses to the object
layer is not thread-safe. This limitation is getting lifted.
* mt/threaded-grep-in-object-store:
grep: use no. of cores as the default no. of threads
grep: move driver pre-load out of critical section
grep: re-enable threads in non-worktree case
grep: protect packed_git [re-]initialization
grep: allow submodule functions to run in parallel
submodule-config: add skip_if_read option to repo_read_gitmodules()
grep: replace grep_read_mutex by internal obj read lock
object-store: allow threaded access to object reading
replace-object: make replace operations thread-safe
grep: fix racy calls in grep_objects()
grep: fix race conditions at grep_submodule()
grep: fix race conditions on userdiff calls
Warn programmers about pretend_object_file() that allows the code
to tentatively use in-core objects.
* jn/pretend-object-doc:
sha1-file: document how to use pretend_object_file
Allow hash_object_file() to work on arbitrary repos by introducing a
git_hash_algo parameter. Change callers which have a struct repository
pointer in their scope to pass on the git_hash_algo from the said repo.
For all other callers, pass on the_hash_algo, which was already being
used internally at hash_object_file(). This functionality will be used
in the following patch to make check_object_signature() be able to work
on arbitrary repos (which, in turn, will be used to fix an
inconsistency at object.c:parse_object()).
Signed-off-by: Matheus Tavares <matheus.bernardino@usp.br>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
has_object_file() said "no" given an object registered to the
system via pretend_object_file(), making it inconsistent with
read_object_file(), causing lazy fetch to attempt fetching an
empty tree from promisor remotes.
* jt/sha1-file-remove-oi-skip-cached:
sha1-file: remove OBJECT_INFO_SKIP_CACHED
Allow object reading to be performed by multiple threads protecting it
with an internal lock, the obj_read_mutex. The lock usage can be toggled
with enable_obj_read_lock() and disable_obj_read_lock(). Currently, the
functions which can be safely called in parallel are:
read_object_file_extended(), repo_read_object_file(),
read_object_file(), read_object_with_reference(), read_object(),
oid_object_info() and oid_object_info_extended(). It's also possible
to use obj_read_lock() and obj_read_unlock() to protect other sections
that cannot execute in parallel with object reading.
Probably there are many spots in the functions listed above that could
be executed unlocked (and thus, in parallel). But, for now, we are most
interested in allowing parallel access to zlib inflation. This is one of
the sections where object reading spends most of the time in (e.g. up to
one-third of git-grep's execution time in the chromium repo corresponds
to inflation) and it's already thread-safe. So, to take advantage of
that, the obj_read_mutex is released when calling git_inflate() and
re-acquired right after, for every calling spot in
oid_object_info_extended()'s call chain. We may refine this lock to also
exploit other possible parallel spots in the future, but for now,
threaded zlib inflation should already give great speedups for threaded
object reading callers.
Note that add_delta_base_cache() was also modified to skip adding
already present entries to the cache. This wasn't possible before, but
it would be now, with the parallel inflation. Take for example the
following situation, where two threads - A and B - are executing the
code at unpack_entry():
1. Thread A is performing the decompression of a base O (which is not
yet in the cache) at PHASE II. Thread B is simultaneously trying to
unpack O, but just starting at PHASE I.
2. Since O is not yet in the cache, B will go to PHASE II to also
perform the decompression.
3. When they finish decompressing, one of them will get the object
reading mutex and go to PHASE III while the other waits for the
mutex. Let’s say A got the mutex first.
4. Thread A will add O to the cache, go throughout the rest of PHASE III
and return.
5. Thread B gets the mutex, also add O to the cache (if the check wasn't
there) and returns.
Finally, it is also important to highlight that the object reading lock
can only ensure thread-safety in the mentioned functions thanks to two
complementary mechanisms: the use of 'struct raw_object_store's
replace_mutex, which guards sections in the object reading machinery
that would otherwise be thread-unsafe; and the 'struct pack_window's
inuse_cnt, which protects window reading operations (such as the one
performed during the inflation of a packed object), allowing them to
execute without the acquisition of the obj_read_mutex.
Signed-off-by: Matheus Tavares <matheus.bernardino@usp.br>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
replace-object functions are very close to being thread-safe: the only
current racy section is the lazy initialization at
prepare_replace_object(). The following patches will protect some object
reading operations to be called threaded, but before that, replace
functions must be protected. To do so, add a mutex to struct
raw_object_store and acquire it before lazy initializing the
replace_map. This won't cause any noticeable performance drop as the
mutex will no longer be used after the replace_map is initialized.
Later, when the replace functions are called in parallel, thread
debuggers might point our use of the added replace_map_initialized flag
as a data race. However, as this boolean variable is initialized as
false and it's only updated once, there's no real harm. It's perfectly
fine if the value is updated right after a thread read it in
replace-map.h:lookup_replace_object() (there'll only be a performance
penalty for the affected threads at that moment). We could cease the
debugger warning protecting the variable reading at the said function.
However, this would negatively affect performance for all threads
calling it, at any time, so it's not really worthy since the warning
doesn't represent a real problem. Instead, to make sure we don't get
false positives (at ThreadSanitizer, at least) an entry for the
respective function is added to .tsan-suppressions.
Signed-off-by: Matheus Tavares <matheus.bernardino@usp.br>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Like in-memory alternates, pretend_object_file contains a trap for the
unwary: careless callers can use it to create references to an object
that does not exist in the on-disk object store.
Add a comment documenting how to use the function without risking such
problems.
The only current caller is blame, which uses pretend_object_file to
create an in-memory commit representing the working tree state.
Noticed during a discussion of how to safely use this function in
operations like "git merge" which, unlike blame, are not read-only.
Inspired-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a partial clone, if a user provides the hash of the empty tree ("git
mktree </dev/null" - for SHA-1, this is 4b825d...) to a command which
requires that that object be parsed, for example:
git diff-tree 4b825d <a non-empty tree>
then Git will lazily fetch the empty tree, unnecessarily, because
parsing of that object invokes repo_has_object_file(), which does not
special-case the empty tree.
Instead, teach repo_has_object_file() to consult find_cached_object()
(which handles the empty tree), thus bringing it in line with the rest
of the object-store-accessing functions. A cost is that
repo_has_object_file() will now need to oideq upon each invocation, but
that is trivial compared to the filesystem lookup or the pack index
search required anyway. (And if find_cached_object() needs to do more
because of previous invocations to pretend_object_file(), all the more
reason to be consistent in whether we present cached objects.)
As a historical note, the function now known as repo_read_object_file()
was taught the empty tree in 346245a1bb ("hard-code the empty tree
object", 2008-02-13), and the function now known as oid_object_info()
was taught the empty tree in c4d9986f5f ("sha1_object_info: examine
cached_object store too", 2011-02-07). repo_has_object_file() was never
updated, perhaps due to oversight. The flag OBJECT_INFO_SKIP_CACHED,
introduced later in dfdd4afcf9 ("sha1_file: teach
sha1_object_info_extended more flags", 2017-06-26) and used in
e83e71c5e1 ("sha1_file: refactor has_sha1_file_with_flags", 2017-06-26),
was introduced to preserve this difference in empty-tree handling, but
now it can be removed.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When loading packfiles on start-up, we traverse the internal packfile
list once per file to avoid reloading packfiles that have already
been loaded. This check runs in quadratic time, so for poorly
maintained repos with a large number of packfiles, it can be pretty
slow.
Add a hashmap containing the packfile names as we load them so that
the average runtime cost of checking for already-loaded packs becomes
constant.
Add a perf test to p5303 to show speed-up.
The existing p5303 test runtimes are dominated by other factors and do
not show an appreciable speed-up. The new test in p5303 clearly exposes
a speed-up in bad cases. In this test we create 10,000 packfiles and
measure the start-up time of git rev-parse, which does little else
besides load in the packs.
Here are the numbers for the new p5303 test:
Test HEAD^ HEAD
---------------------------------------------------------------------
5303.12: load 10,000 packs 1.03(0.92+0.10) 0.12(0.02+0.09) -88.3%
Signed-off-by: Colin Stolley <cstolley@runbox.com>
Helped-by: Jeff King <peff@peff.net>
[jc: squashed the change to call hashmap in install_packed_git() by peff]
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The tips of refs from the alternate object store can be used as
starting point for reachability computation now.
* jk/check-connected-with-alternates:
check_everything_connected: assume alternate ref tips are valid
object-store.h: move for_each_alternate_ref() from transport.h
There's nothing inherently transport-related about enumerating the
alternate ref tips. The code has lived in transport.[ch] because the
only use so far had been advertising available tips during transport.
But it could be used for more, and a future patch will teach rev-list to
access these refs.
Let's move it alongside the other alt-odb code, declaring it in
object-store.h with the implementation in sha1-file.c.
This lets us drop the inclusion of transport.h from receive-pack, which
perhaps shows how it was misplaced (though receive-pack is about
transporting objects, transport.h is mostly about the client side).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The OBJECT_INFO_FOR_PREFETCH bitflag was added to sha1-file.c in 0f4a4fb1
(sha1-file: support OBJECT_INFO_FOR_PREFETCH, 2019-03-29) and is used to
prevent the fetch_objects() method when enabled.
However, there is a problem with the current use. The definition of
OBJECT_INFO_FOR_PREFETCH is given by adding 32 to OBJECT_INFO_QUICK. This is
clearly stated above the definition (in a comment) that this is so
OBJECT_INFO_FOR_PREFETCH implies OBJECT_INFO_QUICK. The problem is that using
"flag & OBJECT_INFO_FOR_PREFETCH" means that OBJECT_INFO_QUICK also implies
OBJECT_INFO_FOR_PREFETCH.
Split out the single bit from OBJECT_INFO_FOR_PREFETCH into a new
OBJECT_INFO_SKIP_FETCH_OBJECT as the single bit and keep
OBJECT_INFO_FOR_PREFETCH as the union of two flags. This allows a clearer use
of flag checking while also keeping the implication of OBJECT_INFO_QUICK.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The code to generate the multi-pack idx file was not prepared to
see too many packfiles and ran out of open file descriptor, which
has been corrected.
* ds/midx-too-many-packs:
midx: add packs to packed_git linked list
midx: pass a repository pointer
Mechanically and systematically drop "extern" from function
declarlation.
* dl/no-extern-in-func-decl:
*.[ch]: manually align parameter lists
*.[ch]: remove extern from function declarations using sed
*.[ch]: remove extern from function declarations using spatch
The multi-pack-index allows searching for objects across multiple
packs using one object list. The original design gains many of
these performance benefits by keeping the packs in the
multi-pack-index out of the packed_git list.
Unfortunately, this has one major drawback. If the multi-pack-index
covers thousands of packs, and a command loads many of those packs,
then we can hit the limit for open file descriptors. The
close_one_pack() method is used to limit this resource, but it
only looks at the packed_git list, and uses an LRU cache to prevent
thrashing.
Instead of complicating this close_one_pack() logic to include
direct references to the multi-pack-index, simply add the packs
opened by the multi-pack-index to the packed_git list. This
immediately solves the file-descriptor limit problem, but requires
some extra steps to avoid performance issues or other problems:
1. Create a multi_pack_index bit in the packed_git struct that is
one if and only if the pack was loaded from a multi-pack-index.
2. Skip packs with the multi_pack_index bit when doing object
lookups and abbreviations. These algorithms already check the
multi-pack-index before the packed_git struct. This has a very
small performance hit, as we need to walk more packed_git
structs. This is acceptable, since these operations run binary
search on the other packs, so this walk-and-ignore logic is
very fast by comparison.
3. When closing a multi-pack-index file, do not close its packs,
as those packs will be closed using close_all_packs(). In some
cases, such as 'git repack', we run 'close_midx()' without also
closing the packs, so we need to un-set the multi_pack_index bit
in those packs. This is necessary, and caught by running
t6501-freshen-objects.sh with GIT_TEST_MULTI_PACK_INDEX=1.
To manually test this change, I inserted trace2 logging into
close_pack_fd() and set pack_max_fds to 10, then ran 'git rev-list
--all --objects' on a copy of the Git repo with 300+ pack-files and
a multi-pack-index. The logs verified the packs are closed as
we read them beyond the file descriptor limit.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In previous patches, extern was mechanically removed from function
declarations without care to formatting, causing parameter lists to be
misaligned. Manually format changed sections such that the parameter
lists should be realigned.
Viewing this patch with 'git diff -w' should produce no output.
Signed-off-by: Denton Liu <liu.denton@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There has been a push to remove extern from function declarations.
Remove some instances of "extern" for function declarations which are
caught by Coccinelle. Note that Coccinelle has some difficulty with
processing functions with `__attribute__` or varargs so some `extern`
declarations are left behind to be dealt with in a future patch.
This was the Coccinelle patch used:
@@
type T;
identifier f;
@@
- extern
T f(...);
and it was run with:
$ git ls-files \*.{c,h} |
grep -v ^compat/ |
xargs spatch --sp-file contrib/coccinelle/noextern.cocci --in-place
Files under `compat/` are intentionally excluded as some are directly
copied from external sources and we should avoid churning them as much
as possible.
Signed-off-by: Denton Liu <liu.denton@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
While running "git diff" in a lazy clone, we can upfront know which
missing blobs we will need, instead of waiting for the on-demand
machinery to discover them one by one. Aim to achieve better
performance by batching the request for these promised blobs.
* jt/batch-fetch-blobs-in-diff:
diff: batch fetching of missing blobs
sha1-file: support OBJECT_INFO_FOR_PREFETCH
Teach oid_object_info_extended() to support a new flag that inhibits
fetching of missing objects. This is equivalent to setting
fetch_is_missing to 0, calling oid_object_info_extended(), then setting
fetch_if_missing to whatever it was before. Update unpack-trees.c to use
this new flag instead of repeatedly setting fetch_if_missing.
This new flag complicates things slightly in that there are now 2 ways
to do the same thing. But this eliminates the need to repeatedly set a
global variable, and more importantly, allows prefetching to be done in
parallel (in the future); hence, this patch.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This member is used to represent the pack checksum of the pack in
question. Expand this member to be GIT_MAX_RAWSZ bytes in length so it
works with longer hashes and rename it to be "hash" instead of "sha1".
This transformation was made with a change to the definition and the
following semantic patch:
@@
struct packed_git *E1;
@@
- E1->sha1
+ E1->hash
@@
struct packed_git E1;
@@
- E1.sha1
+ E1.hash
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The in-core repository instances are passed through more codepaths.
* sb/more-repo-in-api: (23 commits)
t/helper/test-repository: celebrate independence from the_repository
path.h: make REPO_GIT_PATH_FUNC repository agnostic
commit: prepare free_commit_buffer and release_commit_memory for any repo
commit-graph: convert remaining functions to handle any repo
submodule: don't add submodule as odb for push
submodule: use submodule repos for object lookup
pretty: prepare format_commit_message to handle arbitrary repositories
commit: prepare logmsg_reencode to handle arbitrary repositories
commit: prepare repo_unuse_commit_buffer to handle any repo
commit: prepare get_commit_buffer to handle any repo
commit-reach: prepare in_merge_bases[_many] to handle any repo
commit-reach: prepare get_merge_bases to handle any repo
commit-reach.c: allow get_merge_bases_many_0 to handle any repo
commit-reach.c: allow remove_redundant to handle any repo
commit-reach.c: allow merge_bases_many to handle any repo
commit-reach.c: allow paint_down_to_common to handle any repo
commit: allow parse_commit* to handle any repo
object: parse_object to honor its repository argument
object-store: prepare has_{sha1, object}_file to handle any repo
object-store: prepare read_object_file to deal with any repo
...
There are no callers left of has_sha1_file() or its with_flags()
variant. Let's drop them, and convert has_object_file() from a wrapper
into the "real" function. Ironically, the sha1 variant was just copying
into an object_id internally, so the resulting code is actually shorter!
We can also drop the coccinelle rules for catching has_sha1_file()
callers. Since the function no longer exists, the compiler will do that
for us.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Inline odb_load_loose_cache() into its only remaining caller,
odb_loose_cache(). The latter offers a nicer interface for loading the
cache, as it doesn't require callers to deal with fanout directory
numbers directly.
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The loose objects cache is filled one subdirectory at a time as needed.
It is stored in an oid_array, which has to be resorted after each add
operation. So when querying a wide range of objects, the partially
filled array needs to be resorted up to 255 times, which takes over 100
times longer than sorting once.
Use one oid_array for each subdirectory. This ensures that entries have
to only be sorted a single time. It also avoids eight binary search
steps for each cache lookup as a small bonus.
The cache is used for collision checks for the log placeholders %h, %t
and %p, and we can see the change speeding them up in a repository with
ca. 100 objects per subdirectory:
$ git count-objects
26733 objects, 68808 kilobytes
Test HEAD^ HEAD
--------------------------------------------------------------------
4205.1: log with %H 0.51(0.47+0.04) 0.51(0.49+0.02) +0.0%
4205.2: log with %h 0.84(0.82+0.02) 0.60(0.57+0.03) -28.6%
4205.3: log with %T 0.53(0.49+0.04) 0.52(0.48+0.03) -1.9%
4205.4: log with %t 0.84(0.80+0.04) 0.60(0.59+0.01) -28.6%
4205.5: log with %P 0.52(0.48+0.03) 0.51(0.50+0.01) -1.9%
4205.6: log with %p 0.85(0.78+0.06) 0.61(0.56+0.05) -28.2%
4205.7: log with %h-%h-%h 0.96(0.92+0.03) 0.69(0.64+0.04) -28.1%
Reported-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The loose object access code in sha1-file.c is some of the oldest in
Git, and could use some modernizing. It mostly uses "unsigned char *"
for object ids, which these days should be "struct object_id".
It also uses the term "sha1_file" in many functions, which is confusing.
The term "loose_objects" is much better. It clearly distinguishes
them from packed objects (which didn't even exist back when the name
"sha1_file" came into being). And it also distinguishes it from the
checksummed-file concept in csum-file.c (which until recently was
actually called "struct sha1file"!).
This patch converts the functions {open,close,map,stat}_sha1_file() into
open_loose_object(), etc, and switches their sha1 arguments for
object_id structs. Similarly, path functions like fill_sha1_path()
become fill_loose_path() and use object_ids.
The function sha1_loose_object_info() already says "loose", so we can
just drop the "sha1" (and teach it to use object_id).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add and use a function for emptying the loose object cache, so callers
don't have to know any of its implementation details.
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add and use a function for loading the entries of a loose object
subdirectory for a given object ID. It frees callers from deriving the
fanout key; they can use the returned oid_array reference for lookups or
forward range scans.
Suggested-by: Jeff King <peff@peff.net>
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As read_object_file is a widely used function (which is also regularly used
in new code in flight between master..pu), changing its signature is painful
is hard, as other series in flight rely on the original signature. It would
burden the maintainer if we'd just change the signature.
Introduce repo_read_object_file which takes the repository argument, and
hide the original read_object_file as a macro behind
NO_THE_REPOSITORY_COMPATIBILITY_MACROS, similar to
e675765235 (diff.c: remove implicit dependency on the_index, 2018-09-21)
Add a coccinelle patch to convert existing callers, but do not apply
the resulting patch to keep the diff of this patch small.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
read_object_file_extended is not widely used, so migrate it all at once.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In cases where we expect to ask has_sha1_file() about a lot of objects
that we are not likely to have (e.g., during fetch negotiation), we
already use OBJECT_INFO_QUICK to sacrifice accuracy (due to racing with
a simultaneous write or repack) for speed (we avoid re-scanning the pack
directory).
However, even checking for loose objects can be expensive, as we will
stat() each one. On many systems this cost isn't too noticeable, but
stat() can be particularly slow on some operating systems, or due to
network filesystems.
Since the QUICK flag already tells us that we're OK with a slightly
stale answer, we can use that as a cue to look in our in-memory cache of
each object directory. That basically trades an in-memory binary search
for a stat() call.
Note that it is possible for this to actually be _slower_. We'll do a
full readdir() to fill the cache, so if you have a very large number of
loose objects and a very small number of lookups, that readdir() may end
up more expensive.
This shouldn't be a big deal in practice. If you have a large number of
reachable loose objects, you'll already run into performance problems
(which you should remedy by repacking). You may have unreachable objects
which wouldn't otherwise impact performance. Usually these would go away
with the prune step of "git gc", but they may be held for up to 2 weeks
in the default configuration.
So it comes down to how many such objects you might reasonably expect to
have, how much slower is readdir() on N entries versus M stat() calls
(and here we really care about the syscall backing readdir(), like
getdents() on Linux, but I'll just call this readdir() below).
If N is much smaller than M (a typical packed repo), we know this is a
big win (few readdirs() followed by many uses of the resulting cache).
When N and M are similar in size, it's also a win. We care about the
latency of making a syscall, and readdir() should be giving us many
values in a single call. How many?
On Linux, running "strace -e getdents ls" shows a 32k buffer getting 512
entries per call (which is 64 bytes per entry; the name itself is 38
bytes, plus there are some other fields). So we can imagine that this is
always a win as long as the number of loose objects in the repository is
a factor of 500 less than the number of lookups you make. It's hard to
auto-tune this because we don't generally know up front how many lookups
we're going to do. But it's unlikely for this to perform significantly
worse.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Our object_directory struct has a loose objects cache that all users of
the struct can see. But the only one that knows how to load the cache is
find_short_object_filename(). Let's extract that logic in to a reusable
function.
While we're at it, let's also reset the cache when we re-read the object
directories. This shouldn't have an impact on performance, as re-reads
are meant to be rare (and are already expensive, so we avoid them with
things like OBJECT_INFO_QUICK).
Since the cache is already meant to be an approximation, it's tempting
to skip even this bit of safety. But it's necessary to allow more code
to use it. For instance, fetch-pack explicitly re-reads the object
directory after performing its fetch, and would be confused if we didn't
clear the cache.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Our handling of alternate object directories is needlessly different
from the main object directory. As a result, many places in the code
basically look like this:
do_something(r->objects->objdir);
for (odb = r->objects->alt_odb_list; odb; odb = odb->next)
do_something(odb->path);
That gets annoying when do_something() is non-trivial, and we've
resorted to gross hacks like creating fake alternates (see
find_short_object_filename()).
Instead, let's give each raw_object_store a unified list of
object_directory structs. The first will be the main store, and
everything after is an alternate. Very few callers even care about the
distinction, and can just loop over the whole list (and those who care
can just treat the first element differently).
A few observations:
- we don't need r->objects->objectdir anymore, and can just
mechanically convert that to r->objects->odb->path
- object_directory's path field needs to become a real pointer rather
than a FLEX_ARRAY, in order to fill it with expand_base_dir()
- we'll call prepare_alt_odb() earlier in many functions (i.e.,
outside of the loop). This may result in us calling it even when our
function would be satisfied looking only at the main odb.
But this doesn't matter in practice. It's not a very expensive
operation in the first place, and in the majority of cases it will
be a noop. We call it already (and cache its results) in
prepare_packed_git(), and we'll generally check packs before loose
objects. So essentially every program is going to call it
immediately once per program.
Arguably we should just prepare_alt_odb() immediately upon setting
up the repository's object directory, which would save us sprinkling
calls throughout the code base (and forgetting to do so has been a
source of subtle bugs in the past). But I've stopped short of that
here, since there are already a lot of other moving parts in this
patch.
- Most call sites just get shorter. The check_and_freshen() functions
are an exception, because they have entry points to handle local and
nonlocal directories separately.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we generate loose file paths for the main object directory, the
caller provides a buffer to loose_object_path (formerly sha1_file_name).
The callers generally keep their own static buffer to avoid excessive
reallocations.
But for alternate directories, each struct carries its own scratch
buffer. This is needlessly different; let's unify them.
We could go either direction here, but this patch moves the alternates
struct over to the main directory style (rather than vice-versa).
Technically the alternates style is more efficient, as it avoids
rewriting the object directory name on each call. But this is unlikely
to matter in practice, as we avoid reallocations either way (and nobody
has ever noticed or complained that the main object directory is copying
a few extra bytes before making a much more expensive system call).
And this has the advantage that the reusable buffers are tied to
particular calls, which makes the invalidation rules simpler (for
example, the return value from stat_sha1_file() used to be invalidated
by basically any other object call, but now it is affected only by other
calls to stat_sha1_file()).
We do steal the trick from alt_sha1_path() of returning a pointer to the
filled buffer, which makes a few conversions more convenient.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The sha1_file_name() function is used to generate the path to a loose
object in the object directory. It doesn't make much sense for it to
append, since the the path we write may be absolute (i.e., you cannot
reliably build up a path with it). Because many callers use it with a
static buffer, they have to strbuf_reset() manually before each call
(and the other callers always use an empty buffer, so they don't care
either way). Let's handle this automatically.
Since we're changing the semantics, let's take the opportunity to give
it a more hash-neutral name (which will also catch any callers from
topics in flight).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In preparation for unifying the handling of alt odb's and the normal
repo object directory, let's use a more neutral name. This patch is
purely mechanical, swapping the type name, and converting any variables
named "alt" to "odb". There should be no functional change, but it will
reduce the noise in subsequent diffs.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If a repo contains a multi-pack-index, then the packed_git list
does not contain the packfiles that are covered by the multi-pack-index.
This is important for doing object lookups, abbreviations, and
approximating object count. However, there are many operations that
really want to iterate over all packfiles.
Create a new 'all_packs' linked list that contains this list, starting
with the packfiles in the multi-pack-index and then continuing along
the packed_git linked list.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Carlo Marcelo Arenas Belón
Eric Wong
Jeff King
Taylor Blau
Jonathan Tan
brian m. carlson
Junio C Hamano
Matheus Tavares
Jonathan Nieder
Colin Stolley
Derrick Stolee
Denton Liu
René Scharfe
Stefan Beller