The fsmonitor interface got out of sync after the in-core index
file gets discarded, which has been corrected.
* js/fsmonitor-refresh-after-discarding-index:
fsmonitor: force a refresh after the index was discarded
fsmonitor: demonstrate that it is not refreshed after discard_index()
The load_cache_entries_threaded() function takes a src_offset parameter
that it doesn't use. This has been there since its inception in
77ff1127a4 (read-cache: load cache entries on worker threads,
2018-10-10).
Digging on the mailing list, that parameter was part of an earlier
iteration of the series[1], but became unnecessary when the code
switched to using the IEOT extension.
[1] https://public-inbox.org/git/20180906210227.54368-5-benpeart@microsoft.com/
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix a trivial bug that's been here since the shared/do_write_index
tracing was added in 42fee7a388 ("trace2:data: add trace2
instrumentation to index read/write", 2019-02-22). We should have
enter/leave points, not two enter/enter points. This resulted in an
"enter" event without a corresponding "leave" event.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Acked-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Running "git add" on a repository created inside the current
repository is an explicit indication that the user wants to add it
as a submodule, but when the HEAD of the inner repository is on an
unborn branch, it cannot be added as a submodule. Worse, the files
in its working tree can be added as if they are a part of the outer
repository, which is not what the user wants. These problems are
being addressed.
* km/empty-repo-is-still-a-repo:
add: error appropriately on repository with no commits
dir: do not traverse repositories with no commits
submodule: refuse to add repository with no commits
With this change, the `index_state` struct becomes the new home for the
flag that says whether the fsmonitor hook has been run, i.e. it is now
per-index.
It also gets re-set when the index is discarded, fixing the bug
demonstrated by the "test_expect_failure" test added in the preceding
commit. In that case fsmonitor-enabled Git would miss updates under
certain circumstances, see that preceding commit for details.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Conversion from unsigned char[20] to struct object_id continues.
* bc/hash-transition-16: (35 commits)
gitweb: make hash size independent
Git.pm: make hash size independent
read-cache: read data in a hash-independent way
dir: make untracked cache extension hash size independent
builtin/difftool: use parse_oid_hex
refspec: make hash size independent
archive: convert struct archiver_args to object_id
builtin/get-tar-commit-id: make hash size independent
get-tar-commit-id: parse comment record
hash: add a function to lookup hash algorithm by length
remote-curl: make hash size independent
http: replace sha1_to_hex
http: compute hash of downloaded objects using the_hash_algo
http: replace hard-coded constant with the_hash_algo
http-walker: replace sha1_to_hex
http-push: remove remaining uses of sha1_to_hex
http-backend: allow 64-character hex names
http-push: convert to use the_hash_algo
builtin/pull: make hash-size independent
builtin/am: make hash size independent
...
A new hook "post-index-change" is called when the on-disk index
file changes, which can help e.g. a virtualized working tree
implementation.
* bp/post-index-change-hook:
read-cache: add post-index-change hook
The previous commit made 'git add' abort when given a repository that
doesn't have a commit checked out. However, the output upon failure
isn't appropriate:
% git add repo
warning: adding embedded git repository: repo
hint: You've added another git repository inside your current repository.
hint: [...]
error: unable to index file 'repo/'
fatal: adding files failed
The hint doesn't apply in this case, and the error message doesn't
tell the user why 'repo' couldn't be added to the index.
Provide better output by teaching add_to_index() to error when given a
git directory where HEAD can't be resolved. To avoid the embedded
repository warning and hint, call check_embedded_repo() only after
add_file_to_index() succeeds because, in general, its output doesn't
make sense if adding to the index fails.
Signed-off-by: Kyle Meyer <kyle@kyleam.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Index entries are structured with a variety of fields up front, followed
by a hash and one or two flags fields. Because the hash field is stored
in the middle of the structure, it's difficult to use one fixed-size
structure that easily allows access to the hash and flags fields.
Adjust the structure to hold the maximum amount of data that may be
needed using a member called "data" and read and write this field
independently in the various places that need to read and write the
structure.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A more structured way to obtain execution trace has been added.
* jh/trace2:
trace2: add for_each macros to clang-format
trace2: t/helper/test-trace2, t0210.sh, t0211.sh, t0212.sh
trace2:data: add subverb for rebase
trace2:data: add subverb to reset command
trace2:data: add subverb to checkout command
trace2:data: pack-objects: add trace2 regions
trace2:data: add trace2 instrumentation to index read/write
trace2:data: add trace2 hook classification
trace2:data: add trace2 transport child classification
trace2:data: add trace2 sub-process classification
trace2:data: add editor/pager child classification
trace2:data: add trace2 regions to wt-status
trace2: collect Windows-specific process information
trace2: create new combined trace facility
trace2: Documentation/technical/api-trace2.txt
"git checkout --no-overlay" can be used to trigger a new mode of
checking out paths out of the tree-ish, that allows paths that
match the pathspec that are in the current index and working tree
and are not in the tree-ish.
* tg/checkout-no-overlay:
revert "checkout: introduce checkout.overlayMode config"
checkout: introduce checkout.overlayMode config
checkout: introduce --{,no-}overlay option
checkout: factor out mark_cache_entry_for_checkout function
checkout: clarify comment
read-cache: add invalidate parameter to remove_marked_cache_entries
entry: support CE_WT_REMOVE flag in checkout_entry
entry: factor out unlink_entry function
move worktree tests to t24*
Add trace2 events to measure reading and writing the index.
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a post-index-change hook that is invoked after the index is written in
do_write_locked_index().
This hook is meant primarily for notification, and cannot affect
the outcome of git commands that trigger the index write.
The hook is passed a flag to indicate whether the working directory was
updated or not and a flag indicating if a skip-worktree bit could have
changed. These flags enable the hook to optimize its response to the
index change notification.
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since commit 7db118303a (unpack_trees: fix breakage when o->src_index !=
o->dst_index - 2018-04-23) and changes in merge code to use separate
index_state for source and destination, when doing a merge with split
index activated, we may run into this line in unpack_trees():
o->result.split_index = init_split_index(&o->result);
This is by itself not wrong. But this split index information is not
fully populated (and it's only so when move_cache_to_base_index() is
called, aka force splitting the index, or loading index_state from a
file). Both "base_oid" and "base" in this case remain null.
So when writing the main index down, we link to this index with null
oid (default value after init_split_index()), which also means "no split
index" internally. This triggers an incorrect base index refresh:
warning: could not freshen shared index '.../sharedindex.0{40}'
This patch makes sure we will not refresh null base_oid (because the
file is never there). It also makes sure not to write "link" extension
with null base_oid in the first place (no point having it at
all). Read code already has protection against null base_oid.
There is also another side fix in remove_split_index() that causes a
crash when doing "git update-index --no-split-index" when base_oid in
the index file is null. In this case we will not load
istate->split_index->base but we dereference it anyway and are rewarded
with a segfault. This should not happen anymore, but it's still wrong to
dereference a potential NULL pointer, especially when we do check for
NULL pointer in the next code.
Reported-by: Luke Diamand <luke@diamand.org>
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The assumption to work on the single "in-core index" instance has
been reduced from the library-ish part of the codebase.
* nd/the-index-final:
cache.h: flip NO_THE_INDEX_COMPATIBILITY_MACROS switch
read-cache.c: remove the_* from index_has_changes()
merge-recursive.c: remove implicit dependency on the_repository
merge-recursive.c: remove implicit dependency on the_index
sha1-name.c: remove implicit dependency on the_index
read-cache.c: replace update_index_if_able with repo_&
read-cache.c: kill read_index()
checkout: avoid the_index when possible
repository.c: replace hold_locked_index() with repo_hold_locked_index()
notes-utils.c: remove the_repository references
grep: use grep_opt->repo instead of explict repo argument
Commit 9e5da3d055 (add: use separate ADD_CACHE_RENORMALIZE flag,
2019-01-17) switched out using HASH_RENORMALIZE in our flags field for a
new ADD_CACHE_RENORMALIZE flag. However, it forgot to convert one of the
checks for HASH_RENORMALIZE into the new flag, which totally broke "git
add --renormalize".
To make matters even more confusing, the resulting code would racily
pass the tests! The forgotten check was responsible for defeating the
up-to-date check of the index entry. That meant that "git add
--renormalize" would refuse to renormalize things that appeared
stat-clean. But most of the time the test commands run fast enough that
the file mtime is the same as the index mtime. And thus we err on the
conservative side and re-hash the file, which is what "--renormalize"
would have wanted.
But if you're unlucky and cross that one-second boundary between writing
the file and writing the index (which is more likely to happen on a slow
or heavily-loaded system), then the file appears stat-clean. And
"--renormalize" would effectively do nothing.
The fix is straightforward: convert this check to use the right flag.
Noticed-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git add --ignore-errors" did not work as advertised and instead
worked as an unintended synonym for "git add --renormalize", which
has been fixed.
* jk/add-ignore-errors-bit-assignment-fix:
add: use separate ADD_CACHE_RENORMALIZE flag
By default, index compat macros are off from now on, because they
could hide the_index dependency.
Only those in builtin can use it.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit 9472935d81 (add: introduce "--renormalize", 2017-11-16) taught
git-add to pass HASH_RENORMALIZE to add_to_index(), which then passes
the flag along to index_path(). However, the flags taken by
add_to_index() and the ones taken by index_path() are distinct
namespaces. We cannot take HASH_* flags in add_to_index(), because they
overlap with the ADD_CACHE_* flags we already take (in this case,
HASH_RENORMALIZE conflicts with ADD_CACHE_IGNORE_ERRORS).
We can solve this by adding a new ADD_CACHE_RENORMALIZE flag, and using
it to set HASH_RENORMALIZE within add_to_index(). In order to make it
clear that these two flags come from distinct sets, let's also change
the name "newflags" in the function to "hash_flags".
Reported-by: Dmitriy Smirnov <dmitriy.smirnov@jetbrains.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
read_index() shares the same problem as hold_locked_index(): it
assumes $GIT_DIR/index. Move all call sites to repo_read_index()
instead. read_index_preload() and read_index_unmerged() are also
killed as a consequence.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
hold_locked_index() assumes the index path at $GIT_DIR/index. This is
not good for places that take an arbitrary index_state instead of
the_index, which is basically everywhere except builtin/.
Replace it with repo_hold_locked_index(). hold_locked_index() remains
as a wrapper around repo_hold_locked_index() to reduce changes in builtin/
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
More _("i18n") markings.
* nd/i18n:
fsck: mark strings for translation
fsck: reduce word legos to help i18n
parse-options.c: mark more strings for translation
parse-options.c: turn some die() to BUG()
parse-options: replace opterror() with optname()
repack: mark more strings for translation
remote.c: mark messages for translation
remote.c: turn some error() or die() to BUG()
reflog: mark strings for translation
read-cache.c: add missing colon separators
read-cache.c: mark more strings for translation
read-cache.c: turn die("internal error") to BUG()
attr.c: mark more string for translation
archive.c: mark more strings for translation
alias.c: mark split_cmdline_strerror() strings for translation
git.c: mark more strings for translation
When marking cache entries for removal, and later removing them all at
once using 'remove_marked_cache_entries()', cache entries currently
have to be invalidated manually in the cache tree and in the untracked
cache.
Add an invalidate flag to the function. With the flag set, the
function will take care of invalidating the path in the cache tree and
in the untracked cache.
Note that the current callsites already do the invalidation properly
in other places, so we're just passing 0 from there to keep the status
quo.
This will be useful in a subsequent commit.
Signed-off-by: Thomas Gummerer <t.gummerer@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We indent with TABs and sometimes for fine alignment, TABs followed by
spaces, but never all spaces (unless the indentation is less than 8
columns). Indenting with spaces slips through in some places. Fix
them.
Imported code and compat/ are left alone on purpose. The former should
remain as close as upstream as possible. The latter pretty much has
separate maintainers, it's up to them to decide.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The way .git/index and .git/sharedindex* files were initially
created gave these files different perm bits until they were
adjusted for shared repository settings. This was made consistent.
* cc/shared-index-permbits:
read-cache: make the split index obey umask settings
If a user explicitly sets
[index]
threads = true
to read the index using multiple threads, ensure that index writes
include the offset table by default to make that possible. This
ensures that the user's intent of turning on threading is respected.
In other words, permit the following configurations:
- index.threads and index.recordOffsetTable unspecified: do not write
the offset table yet (to avoid alarming the user with "ignoring IEOT
extension" messages when an older version of Git accesses the
repository) but do make use of multiple threads to read the index if
the supporting offset table is present.
This can also be requested explicitly by setting index.threads=true,
0, or >1 and index.recordOffsetTable=false.
- index.threads=false or 1: do not write the offset table, and do not
make use of the offset table.
One can set index.recordOffsetTable=false as well, to be more
explicit.
- index.threads=true, 0, or >1 and index.recordOffsetTable unspecified:
write the offset table and make use of threads at read time.
This can also be requested by setting index.threads=true, 0, >1, or
unspecified and index.recordOffsetTable=true.
Fortunately the complication is temporary: once most Git installations
have upgraded to a version with support for the IEOT and EOIE
extensions, we can flip the defaults for index.recordEndOfIndexEntries
and index.recordOffsetTable to true and eliminate the settings.
Helped-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As with EOIE, popular versions of Git do not support the new IEOT
extension yet. When accessing a Git repository written by a more
modern version of Git, they correctly ignore the unrecognized section,
but in the process they loudly warn
ignoring IEOT extension
resulting in confusion for users. Introduce the index extension more
gently by not writing it yet in this first version with support for
it. Soon, once sufficiently many users are running a modern version
of Git, we can flip the default so users benefit from this index
extension by default.
Introduce a '[index] recordOffsetTable' configuration variable to
control whether the new index extension is written.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since 3b1d9e04 (eoie: add End of Index Entry (EOIE) extension,
2018-10-10) Git defaults to writing the new EOIE section when writing
out an index file. Usually that is a good thing because it improves
threaded performance, but when a Git repository is shared with older
versions of Git, it produces a confusing warning:
$ git status
ignoring EOIE extension
HEAD detached at 371ed0defa
nothing to commit, working tree clean
Let's introduce the new index extension more gently. First we'll roll
out the new version of Git that understands it, and then once
sufficiently many users are using such a version, we can flip the
default to writing it by default.
Introduce a '[index] recordEndOfIndexEntries' configuration variable
to allow interested users to benefit from this index extension early.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make the split index write out its .git/sharedindex_* files with the
same permissions as .git/index. This only changes the behavior when
core.sharedRepository isn't set, i.e. the user's umask settings will
be respected.
This hasn't been the case ever since the split index was originally
implemented in c18b80a0e8 ("update-index: new options to
enable/disable split index mode", 2014-06-13). A mkstemp()-like
function has always been used to create it. First mkstemp() itself,
and then later our own mkstemp()-like in
f6ecc62dbf ("write_shared_index(): use tempfile module", 2015-08-10)
A related bug was fixed in df801f3f9f ("read-cache: use shared perms
when writing shared index", 2017-06-25). Since then the split index
has respected core.sharedRepository.
However, using that setting should not be required simply to make git
obey the user's umask setting. It's intended for the use-case of
overriding whatever that umask is set to. This fixes cases where the
user has e.g. set his umask to 022 on a shared server in anticipation
of other user's needing to run "status", "log" etc. in his repository.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The codebase has been cleaned up to reduce "#ifndef NO_PTHREADS".
* nd/pthreads:
Clean up pthread_create() error handling
read-cache.c: initialize copy_len to shut up gcc 8
read-cache.c: reduce branching based on HAVE_THREADS
read-cache.c: remove #ifdef NO_PTHREADS
pack-objects: remove #ifdef NO_PTHREADS
preload-index.c: remove #ifdef NO_PTHREADS
grep: clean up num_threads handling
grep: remove #ifdef NO_PTHREADS
attr.c: remove #ifdef NO_PTHREADS
name-hash.c: remove #ifdef NO_PTHREADS
index-pack: remove #ifdef NO_PTHREADS
send-pack.c: move async's #ifdef NO_PTHREADS back to run-command.c
run-command.h: include thread-utils.h instead of pthread.h
thread-utils: macros to unconditionally compile pthreads API
The helper function to refresh the cached stat information in the
in-core index has learned to perform the lstat() part of the
operation in parallel on multi-core platforms.
* bp/refresh-index-using-preload:
refresh_index: remove unnecessary calls to preload_index()
speed up refresh_index() by utilizing preload_index()
This case is more interesting than other boring "remove the_repo"
commits because while we need access to the object database, we cannot
simply use r->index because unpack-trees.c can operate on a temporary
index, not $GIT_DIR/index. Ideally we should be able to pass an object
database to lookup_tree() but that ship has sailed.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
typechange_fmt and added_fmt should have a colon before "needs
update". Align the statements to make it easier to read and see. Also
drop the unnecessary ().
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There are a couple other improvements on these strings as well:
- add missing colon (as separator)
- quote paths
- provide more information on error messages
- keep first word in lowercase
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It was reported that when building with NO_PTHREADS=1,
-Wmaybe-uninitialized is triggered. Just initialize the variable from
the beginning to shut the compiler up (because this warning is enabled
in config.dev)
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is a faithful conversion with no attempt to clean up whatsoever.
Code indentation is left broken. There will be another commit to clean
it up and un-indent if we just indent now. It's just more code noise.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Speed up refresh_index() by utilizing preload_index() to do most of the work
spread across multiple threads. This works because most cache entries will
get marked CE_UPTODATE so that refresh_cache_ent() can bail out early when
called from within refresh_index().
On a Windows repo with ~200K files, this drops refresh times from 6.64
seconds to 2.87 seconds for a savings of 57%.
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The codepath to support the experimental split-index mode had
remaining "racily clean" issues fixed.
* sg/split-index-racefix:
split-index: BUG() when cache entry refers to non-existing shared entry
split-index: smudge and add racily clean cache entries to split index
split-index: don't compare cached data of entries already marked for split index
split-index: count the number of deleted entries
t1700-split-index: date back files to avoid racy situations
split-index: add tests to demonstrate the racy split index problem
t1700-split-index: document why FSMONITOR is disabled in this test script
introduced with c46c406ae1 (trace.h: support nested performance tracing)
on Aug 18, 2018 but not affecting maint
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Acked-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git status" learns to show progress bar when refreshing the index
takes a long time.
* nd/status-refresh-progress:
status: show progress bar if refreshing the index takes too long
A new extension to the index file has been introduced, which allows
the file to be read in parallel.
* bp/read-cache-parallel:
read-cache: load cache entries on worker threads
ieot: add Index Entry Offset Table (IEOT) extension
read-cache: load cache extensions on a worker thread
config: add new index.threads config setting
eoie: add End of Index Entry (EOIE) extension
read-cache: clean up casting and byte decoding
read-cache.c: optimize reading index format v4
Various codepaths in the core-ish part learn to work on an
arbitrary in-core index structure, not necessarily the default
instance "the_index".
* nd/the-index: (23 commits)
revision.c: reduce implicit dependency the_repository
revision.c: remove implicit dependency on the_index
ws.c: remove implicit dependency on the_index
tree-diff.c: remove implicit dependency on the_index
submodule.c: remove implicit dependency on the_index
line-range.c: remove implicit dependency on the_index
userdiff.c: remove implicit dependency on the_index
rerere.c: remove implicit dependency on the_index
sha1-file.c: remove implicit dependency on the_index
patch-ids.c: remove implicit dependency on the_index
merge.c: remove implicit dependency on the_index
merge-blobs.c: remove implicit dependency on the_index
ll-merge.c: remove implicit dependency on the_index
diff-lib.c: remove implicit dependency on the_index
read-cache.c: remove implicit dependency on the_index
diff.c: remove implicit dependency on the_index
grep.c: remove implicit dependency on the_index
diff.c: remove the_index dependency in textconv() functions
blame.c: rename "repo" argument to "r"
combine-diff.c: remove implicit dependency on the_index
...
Ever since the split index feature was introduced [1], refreshing a
split index is prone to a variant of the classic racy git problem.
Consider the following sequence of commands updating the split index
when the shared index contains a racily clean cache entry, i.e. an
entry whose cached stat data matches with the corresponding file in
the worktree and the cached mtime matches that of the index:
echo "cached content" >file
git update-index --split-index --add file
echo "dirty worktree" >file # size stays the same!
# ... wait ...
git update-index --add other-file
Normally, when a non-split index is updated, then do_write_index()
(the function responsible for writing all kinds of indexes, "regular",
split, and shared) recognizes racily clean cache entries, and writes
them with smudged stat data, i.e. with file size set to 0. When
subsequent git commands read the index, they will notice that the
smudged stat data doesn't match with the file in the worktree, and
then go on to check the file's content and notice its dirtiness.
In the above example, however, in the second 'git update-index'
prepare_to_write_split_index() decides which cache entries stored only
in the shared index should be replaced in the new split index. Alas,
this function never looks out for racily clean cache entries, and
since the file's stat data in the worktree hasn't changed since the
shared index was written, it won't be replaced in the new split index.
Consequently, do_write_index() doesn't even get this racily clean
cache entry, and can't smudge its stat data. Subsequent git commands
will then see that the index has more recent mtime than the file and
that the (not smudged) cached stat data still matches with the file in
the worktree, and, ultimately, will erroneously consider the file
clean.
Modify prepare_to_write_split_index() to recognize racily clean cache
entries, and mark them to be added to the split index. Note that
there are two places where it should check raciness: first those cache
entries that are only stored in the shared index, and then those that
have been copied by unpack_trees() from the shared index while it
constructed a new index. This way do_write_index() will get these
racily clean cache entries as well, and will then write them with
smudged stat data to the new split index.
This change makes all tests in 't1701-racy-split-index.sh' pass, so
flip the two 'test_expect_failure' tests to success. Also add the '#'
(as in nr. of trial) to those tests' description that were omitted
when the tests expected failure.
Note that after this change if the index is split when it contains a
racily clean cache entry, then a smudged cache entry will be written
both to the new shared and to the new split indexes. This doesn't
affect regular git commands: as far as they are concerned this is just
an entry in the split index replacing an outdated entry in the shared
index. It did affect a few tests in 't1700-split-index.sh', though,
because they actually check which entries are stored in the split
index; a previous patch in this series has already made the necessary
adjustments in 't1700'. And racily clean cache entries and index
splitting are rare enough to not worry about the resulting duplicated
smudged cache entries, and the additional complexity required to
prevent them is not worth it.
Several tests failed occasionally when the test suite was run with
'GIT_TEST_SPLIT_INDEX=yes'. Here are those that I managed to trace
back to this racy split index problem, starting with those failing
more frequently, with a link to a failing Travis CI build job for
each. The highlighted line [2] shows when the racy file was written,
which is not always in the failing test but in a preceeding setup
test.
t3903-stash.sh:
https://travis-ci.org/git/git/jobs/385542084#L5858
t4024-diff-optimize-common.sh:
https://travis-ci.org/git/git/jobs/386531969#L3174
t4015-diff-whitespace.sh:
https://travis-ci.org/git/git/jobs/360797600#L8215
t2200-add-update.sh:
https://travis-ci.org/git/git/jobs/382543426#L3051
t0090-cache-tree.sh:
https://travis-ci.org/git/git/jobs/416583010#L3679
There might be others, e.g. perhaps 't1000-read-tree-m-3way.sh' and
others using 'lib-read-tree-m-3way.sh', but I couldn't confirm yet.
[1] In the branch leading to the merge commit v2.1.0-rc0~45 (Merge
branch 'nd/split-index', 2014-07-16).
[2] Note that those highlighted lines are in the 'after failure' fold,
and your browser might unhelpfully fold it up before you could
take a good look.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch helps address the CPU cost of loading the index by utilizing
the Index Entry Offset Table (IEOT) to divide loading and conversion of
the cache entries across multiple threads in parallel.
I used p0002-read-cache.sh to generate some performance data:
Test w/100,000 files reduced the time by 32.24%
Test w/1,000,000 files reduced the time by -4.77%
Note that on the 1,000,000 files case, multi-threading the cache entry parsing
does not yield a performance win. This is because the cost to parse the
index extensions in this repo, far outweigh the cost of loading the cache
entries.
The high cost of parsing the index extensions is driven by the cache tree
and the untracked cache extensions. As this is currently the longest pole,
any reduction in this time will reduce the overall index load times so is
worth further investigation in another patch series.
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch enables addressing the CPU cost of loading the index by adding
additional data to the index that will allow us to efficiently multi-
thread the loading and conversion of cache entries.
It accomplishes this by adding an (optional) index extension that is a
table of offsets to blocks of cache entries in the index file. To make
this work for V4 indexes, when writing the cache entries, it periodically
"resets" the prefix-compression by encoding the current entry as if the
path name for the previous entry is completely different and saves the
offset of that entry in the IEOT. Basically, with V4 indexes, it
generates offsets into blocks of prefix-compressed entries.
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch helps address the CPU cost of loading the index by loading
the cache extensions on a worker thread in parallel with loading the cache
entries.
In some cases, loading the extensions takes longer than loading the
cache entries so this patch utilizes the new EOIE to start the thread to
load the extensions before loading all the cache entries in parallel.
This is possible because the current extensions don't access the cache
entries in the index_state structure so are OK that they don't all exist
yet.
The CACHE_EXT_TREE, CACHE_EXT_RESOLVE_UNDO, and CACHE_EXT_UNTRACKED
extensions don't even get a pointer to the index so don't have access to the
cache entries.
CACHE_EXT_LINK only uses the index_state to initialize the split index.
CACHE_EXT_FSMONITOR only uses the index_state to save the fsmonitor last
update and dirty flags.
I used p0002-read-cache.sh to generate some performance data:
Test w/100,000 files reduced the time by 0.53%
Test w/1,000,000 files reduced the time by 27.78%
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The End of Index Entry (EOIE) is used to locate the end of the variable
length index entries and the beginning of the extensions. Code can take
advantage of this to quickly locate the index extensions without having
to parse through all of the index entries.
The EOIE extension is always written out to the index file including to
the shared index when using the split index feature. Because it is always
written out, the SHA checksums in t/t1700-split-index.sh were updated
to reflect its inclusion.
It is written as an optional extension to ensure compatibility with other
git implementations that do not yet support it. It is always written out
to ensure it is available as often as possible to speed up index operations.
Because it must be able to be loaded before the variable length cache
entries and other index extensions, this extension must be written last.
The signature for this extension is { 'E', 'O', 'I', 'E' }.
The extension consists of:
- 32-bit offset to the end of the index entries
- 160-bit SHA-1 over the extension types and their sizes (but not
their contents). E.g. if we have "TREE" extension that is N-bytes
long, "REUC" extension that is M-bytes long, followed by "EOIE",
then the hash would be:
SHA-1("TREE" + <binary representation of N> +
"REUC" + <binary representation of M>)
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch does a clean up pass to minimize the casting required to work
with the memory mapped index (mmap).
It also makes the decoding of network byte order more consistent by using
get_be32() where possible.
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Index format v4 requires some more computation to assemble a path
based on a previous one. The current code is not very efficient
because
- it doubles memory copy, we assemble the final path in a temporary
first before putting it back to a cache_entry
- strbuf_remove() in expand_name_field() is not exactly a good fit
for stripping a part at the end, _setlen() would do the same job
and is much cheaper.
- the open-coded loop to find the end of the string in
expand_name_field() can't beat an optimized strlen()
This patch avoids the temporary buffer and writes directly to the new
cache_entry, which addresses the first two points. The last point
could also be avoided if the total string length fits in the first 12
bits of ce_flags, if not we fall back to strlen().
Running "test-tool read-cache 100" on webkit.git (275k files), reading
v2 only takes 4.226 seconds, while v4 takes 5.711 seconds, 35% more
time. The patch reduces read time on v4 to 4.319 seconds.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A new variant repo_diff_setup() is added that takes 'struct repository *'
and diff_setup() becomes a thin macro around it that is protected by
NO_THE_REPOSITORY_COMPATIBILITY_MACROS, similar to NO_THE_INDEX_....
The plan is these macros will always be defined for all library files
and the macros are only accessible in builtin/
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This function calls do_diff_cache() which eventually needs to set this
"istate" to unpack_options->src_index [1]. This is an unfortunate fact
that unpack_trees() _will_ update [2] src_index so we can't really pass a
const index_state there. Just remove 'const'.
[1] Right now diff_cache() in diff-lib.c assigns the_index to
src_index. But the plan is to get rid of the_index, so it should
be 'istate' from here that gets assigned to src_index.
[2] Some transient bits in the source index are touched. Optional
extensions can also be removed. But other than that the source
tree should still be valid.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
spatch transformation to replace boolean uses of !hashcmp() to
newly introduced oideq() is added, and applied, to regain
performance lost due to support of multiple hash algorithms.
* jk/cocci:
show_dirstat: simplify same-content check
read-cache: use oideq() in ce_compare functions
convert hashmap comparison functions to oideq()
convert "hashcmp() != 0" to "!hasheq()"
convert "oidcmp() != 0" to "!oideq()"
convert "hashcmp() == 0" to hasheq()
convert "oidcmp() == 0" to oideq()
introduce hasheq() and oideq()
coccinelle: use <...> for function exclusion
The unpack_trees() API used in checking out a branch and merging
walks one or more trees along with the index. When the cache-tree
in the index tells us that we are walking a tree whose flattened
contents is known (i.e. matches a span in the index), as linearly
scanning a span in the index is much more efficient than having to
open tree objects recursively and listing their entries, the walk
can be optimized, which is done in this topic.
* nd/unpack-trees-with-cache-tree:
Document update for nd/unpack-trees-with-cache-tree
cache-tree: verify valid cache-tree in the test suite
unpack-trees: add missing cache invalidation
unpack-trees: reuse (still valid) cache-tree from src_index
unpack-trees: reduce malloc in cache-tree walk
unpack-trees: optimize walking same trees with cache-tree
unpack-trees: add performance tracing
trace.h: support nested performance tracing
Refreshing the index is usually very fast, but it can still take a
long time sometimes. Cold cache is one. Or copying a repo to a new
place (*). It's good to show something to let the user know "git
status" is not hanging, it's just busy doing something.
(*) In this case, all stat info in the index becomes invalid and git
falls back to rehashing all file content to see if there's any
difference between updating stat info in the index. This is quite
expensive. Even with a repo as small as git.git, it takes 3
seconds.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
These functions return the full oidcmp() value, but the
callers really only care whether it is non-zero. We can use
the more strict !oideq(), which a compiler may be able to
optimize further.
This does change the meaning of the return value subtly, but
it's unlikely that anybody would try to use them for
ordering. They're static-local in this file, and they
already return other error values that would confuse an
ordering (e.g., open() failure gives -1).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This rounds out the previous three patches, covering the
inequality logic for the "hash" variant of the functions.
As with the previous three, the accompanying code changes
are the mechanical result of applying the coccinelle patch;
see those patches for more discussion.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is the flip side of the previous two patches: checking
for a non-zero oidcmp() can be more strictly expressed as
inequality. Like those patches, we write "!= 0" in the
coccinelle transformation, which covers by isomorphism the
more common:
if (oidcmp(E1, E2))
As with the previous two patches, this patch can be achieved
almost entirely by running "make coccicheck"; the only
differences are manual line-wrap fixes to match the original
code.
There is one thing to note for anybody replicating this,
though: coccinelle 1.0.4 seems to miss the case in
builtin/tag.c, even though it's basically the same as all
the others. Running with 1.0.7 does catch this, so
presumably it's just a coccinelle bug that was fixed in the
interim.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Using the more restrictive oideq() should, in the long run,
give the compiler more opportunities to optimize these
callsites. For now, this conversion should be a complete
noop with respect to the generated code.
The result is also perhaps a little more readable, as it
avoids the "zero is equal" idiom. Since it's so prevalent in
C, I think seasoned programmers tend not to even notice it
anymore, but it can sometimes make for awkward double
negations (e.g., we can drop a few !!oidcmp() instances
here).
This patch was generated almost entirely by the included
coccinelle patch. This mechanical conversion should be
completely safe, because we check explicitly for cases where
oidcmp() is compared to 0, which is what oideq() is doing
under the hood. Note that we don't have to catch "!oidcmp()"
separately; coccinelle's standard isomorphisms make sure the
two are treated equivalently.
I say "almost" because I did hand-edit the coccinelle output
to fix up a few style violations (it mostly keeps the
original formatting, but sometimes unwraps long lines).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The more library-ish parts of the codebase learned to work on the
in-core index-state instance that is passed in by their callers,
instead of always working on the singleton "the_index" instance.
* nd/no-the-index: (24 commits)
blame.c: remove implicit dependency on the_index
apply.c: remove implicit dependency on the_index
apply.c: make init_apply_state() take a struct repository
apply.c: pass struct apply_state to more functions
resolve-undo.c: use the right index instead of the_index
archive-*.c: use the right repository
archive.c: avoid access to the_index
grep: use the right index instead of the_index
attr: remove index from git_attr_set_direction()
entry.c: use the right index instead of the_index
submodule.c: use the right index instead of the_index
pathspec.c: use the right index instead of the_index
unpack-trees: avoid the_index in verify_absent()
unpack-trees: convert clear_ce_flags* to avoid the_index
unpack-trees: don't shadow global var the_index
unpack-trees: add a note about path invalidation
unpack-trees: remove 'extern' on function declaration
ls-files: correct index argument to get_convert_attr_ascii()
preload-index.c: use the right index instead of the_index
dir.c: remove an implicit dependency on the_index in pathspec code
...
This makes sure that cache-tree is consistent with the index. The main
purpose is to catch potential problems by saving the index in
unpack_trees() but the line in write_index() would also help spot
missing invalidation in other code.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We do n-way merge by walking the source index and n trees at the same
time and add merge results to a new temporary index called o->result.
The merge result for any given path could be either
- keep_entry(): same old index entry in o->src_index is reused
- merged_entry(): either a new entry is added, or an existing one updated
- deleted_entry(): one entry from o->src_index is removed
For some reason [1] we keep making sure that the source index's
cache-tree is still valid if used by o->result: for all those
merged/deleted entries, we invalidate the same path in o->src_index,
so only cache-trees covering the "keep_entry" parts remain good.
Because of this, the cache-tree from o->src_index can be perfectly
reused in o->result. And in fact we already rely on this logic to
reuse untracked cache in edf3b90553 (unpack-trees: preserve index
extensions - 2017-05-08). Move the cache-tree to o->result before
doing cache_tree_update() to reduce hashing cost.
Since cache_tree_update() has risen up as one of the most expensive
parts in unpack_trees() after the last few patches. This does help
reduce unpack_trees() time significantly (on webkit.git):
before after
--------------------------------------------------------------------
0.080394752 0.051258167 s: read cache .git/index
0.216010838 0.212106298 s: preload index
0.008534301 0.280521764 s: refresh index
0.251992198 0.218160442 s: traverse_trees
0.377031383 0.374948191 s: check_updates
0.372768105 0.037040114 s: cache_tree_update
1.045887251 0.672031609 s: unpack_trees
0.314983512 0.317456290 s: write index, changed mask = 2e
0.062572653 0.038382654 s: traverse_trees
0.000022544 0.000042731 s: check_updates
0.073795585 0.050930053 s: unpack_trees
0.073807557 0.051099735 s: diff-index
1.938191592 1.614241153 s: git command: git checkout -
[1] I'm pretty sure the reason is an oversight in 34110cd4e3 (Make
'unpack_trees()' have a separate source and destination index -
2008-03-06). That patch aims to _not_ update the source index at
all. The invalidation should have been done on o->result in that
patch. But then there was no cache-tree on o->result even then so
it's pointless to do so.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Performance measurements are listed right now as a flat list, which is
fine when we measure big blocks. But when we start adding more and
more measurements, some of them could be just part of a bigger
measurement and a flat list gives a wrong impression that they are
executed at the same level instead of nested.
Add trace_performance_enter() and trace_performance_leave() to allow
indent these nested measurements. For now it does not help much
because the only nested thing is (lazy) name hash initialization
(e.g. called in diff-index from "git status"). This will help more
because I'm going to add some more tracing that's actually nested.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git merge --abort" etc. did not clean things up properly when
there were conflicted entries in the index in certain order that
are involved in D/F conflicts. This has been corrected.
* en/abort-df-conflict-fixes:
read-cache: fix directory/file conflict handling in read_index_unmerged()
t1015: demonstrate directory/file conflict recovery failures
Make the match_patchspec API and friends take an index_state instead
of assuming the_index in dir.c. All external call sites are converted
blindly to keep the patch simple and retain current behavior.
Individual call sites may receive further updates to use the right
index instead of the_index.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The recursive merge strategy did not properly ensure there was no
change between HEAD and the index before performing its operation,
which has been corrected.
* en/dirty-merge-fixes:
merge: fix misleading pre-merge check documentation
merge-recursive: enforce rule that index matches head before merging
t6044: add more testcases with staged changes before a merge is invoked
merge-recursive: fix assumption that head tree being merged is HEAD
merge-recursive: make sure when we say we abort that we actually abort
t6044: add a testcase for index matching head, when head doesn't match HEAD
t6044: verify that merges expected to abort actually abort
index_has_changes(): avoid assuming operating on the_index
read-cache.c: move index_has_changes() from merge.c
For a large tree, the index needs to hold many cache entries
allocated on heap. These cache entries are now allocated out of a
dedicated memory pool to amortize malloc(3) overhead.
* jm/cache-entry-from-mem-pool:
block alloc: add validations around cache_entry lifecyle
block alloc: allocate cache entries from mem_pool
mem-pool: fill out functionality
mem-pool: add life cycle management functions
mem-pool: only search head block for available space
block alloc: add lifecycle APIs for cache_entry structs
read-cache: teach make_cache_entry to take object_id
read-cache: teach refresh_cache_entry to take istate
read_index_unmerged() has two intended purposes:
* return 1 if there are any unmerged entries, 0 otherwise
* drops any higher-stage entries down to stage #0
There are several callers of read_index_unmerged() that check the return
value to see if it is non-zero, all of which then die() if that condition
is met. For these callers, dropping higher-stage entries down to stage #0
is a waste of resources, and returning immediately on first unmerged entry
would be better. But it's probably only a very minor difference and isn't
the focus of this series.
The remaining callers ignore the return value and call this function for
the side effect of dropping higher-stage entries down to stage #0. As
mentioned in commit e11d7b5969 ("'reset --merge': fix unmerged case",
2009-12-31),
The _only_ reason we want to keep a previously unmerged entry in the
index at stage #0 is so that we don't forget the fact that we have
corresponding file in the work tree in order to be able to remove it
when the tree we are resetting to does not have the path.
In fact, prior to commit d1a43f2aa4 ("reset --hard/read-tree --reset -u:
remove unmerged new paths", 2008-10-15), read_index_unmerged() did just
remove unmerged entries from the cache immediately but that had the
unwanted effect of leaving around new untracked files in the tree from
aborted merges.
So, that's the intended purpose of this function. The problem is that
when directory/files conflicts are present, trying to add the file to the
index at stage 0 fails (because there is still a directory in the way),
and the function returns early with a -1 return code to signify the error.
As noted above, none of the callers who want the drop-to-stage-0 behavior
check the return status, though, so this means all remaining unmerged
entries remain in the index and the callers proceed assuming otherwise.
Users then see errors of the form:
error: 'DIR-OR-FILE' appears as both a file and as a directory
error: DIR-OR-FILE: cannot drop to stage #0
and potentially also messages about other unmerged entries which came
lexicographically later than whatever pathname was both a file and a
directory. Google finds a few hits searching for those messages,
suggesting there were probably a couple people who hit this besides me.
Luckily, calling `git reset --hard` multiple times would workaround
this bug.
Since the whole purpose here is to just put the entry *temporarily* into
the index so that any associated file in the working copy can be removed,
we can just skip the DFCHECK and allow both the file and directory to
appear in the index. The temporary simultaneous appearance of the
directory and file entries in the index will be removed by the callers
by calling unpack_trees(), which excludes these unmerged entries marked
with CE_CONFLICTED flag from the resulting index, before they attempt to
write the index anywhere.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The conversion to pass "the_repository" and then "a_repository"
throughout the object access API continues.
* sb/object-store-grafts:
commit: allow lookup_commit_graft to handle arbitrary repositories
commit: allow prepare_commit_graft to handle arbitrary repositories
shallow: migrate shallow information into the object parser
path.c: migrate global git_path_* to take a repository argument
cache: convert get_graft_file to handle arbitrary repositories
commit: convert read_graft_file to handle arbitrary repositories
commit: convert register_commit_graft to handle arbitrary repositories
commit: convert commit_graft_pos() to handle arbitrary repositories
shallow: add repository argument to is_repository_shallow
shallow: add repository argument to check_shallow_file_for_update
shallow: add repository argument to register_shallow
shallow: add repository argument to set_alternate_shallow_file
commit: add repository argument to lookup_commit_graft
commit: add repository argument to prepare_commit_graft
commit: add repository argument to read_graft_file
commit: add repository argument to register_commit_graft
commit: add repository argument to commit_graft_pos
object: move grafts to object parser
object-store: move object access functions to object-store.h
`git merge-recursive` does a three-way merge between user-specified trees
base, head, and remote. Since the user is allowed to specify head, we can
not necesarily assume that head == HEAD.
Modify index_has_changes() to take an extra argument specifying the tree
to compare against. If NULL, it will compare to HEAD. We then use this
from merge-recursive to make sure we compare to the user-specified head.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Modify index_has_changes() to take a struct istate* instead of just
operating on the_index. This is only a partial conversion, though,
because we call do_diff_cache() which implicitly assumes work is to be
done on the_index. Ongoing work is being done elsewhere to do the
remainder of the conversion, and thus is not duplicated here. Instead,
a simple check is put in place until that work is complete.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since index_has_change() is an index-related function, move it to
read-cache.c, only modifying it to avoid uses of the active_cache and
active_nr macros.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add an option (controlled by an environment variable) perform extra
validations on mem_pool allocated cache entries. When set:
1) Invalidate cache_entry memory when discarding cache_entry.
2) When discarding index_state struct, verify that all cache_entries
were allocated from expected mem_pool.
3) When discarding mem_pools, invalidate mem_pool memory.
This should provide extra checks that mem_pools and their allocated
cache_entries are being used as expected.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When reading large indexes from disk, a portion of the time is
dominated in malloc() calls. This can be mitigated by allocating a
large block of memory and manage it ourselves via memory pools.
This change moves the cache entry allocation to be on top of memory
pools.
Design:
The index_state struct will gain a notion of an associated memory_pool
from which cache_entries will be allocated from. When reading in the
index from disk, we have information on the number of entries and
their size, which can guide us in deciding how large our initial
memory allocation should be. When an index is discarded, the
associated memory_pool will be discarded as well - so the lifetime of
a cache_entry is tied to the lifetime of the index_state that it was
allocated for.
In the case of a Split Index, the following rules are followed. 1st,
some terminology is defined:
Terminology:
- 'the_index': represents the logical view of the index
- 'split_index': represents the "base" cache entries. Read from the
split index file.
'the_index' can reference a single split_index, as well as
cache_entries from the split_index. `the_index` will be discarded
before the `split_index` is. This means that when we are allocating
cache_entries in the presence of a split index, we need to allocate
the entries from the `split_index`'s memory pool. This allows us to
follow the pattern that `the_index` can reference cache_entries from
the `split_index`, and that the cache_entries will not be freed while
they are still being referenced.
Managing transient cache_entry structs:
Cache entries are usually allocated for an index, but this is not always
the case. Cache entries are sometimes allocated because this is the
type that the existing checkout_entry function works with. Because of
this, the existing code needs to handle cache entries associated with an
index / memory pool, and those that only exist transiently. Several
strategies were contemplated around how to handle this:
Chosen approach:
An extra field was added to the cache_entry type to track whether the
cache_entry was allocated from a memory pool or not. This is currently
an int field, as there are no more available bits in the existing
ce_flags bit field. If / when more bits are needed, this new field can
be turned into a proper bit field.
Alternatives:
1) Do not include any information about how the cache_entry was
allocated. Calling code would be responsible for tracking whether the
cache_entry needed to be freed or not.
Pro: No extra memory overhead to track this state
Con: Extra complexity in callers to handle this correctly.
The extra complexity and burden to not regress this behavior in the
future was more than we wanted.
2) cache_entry would gain knowledge about which mem_pool allocated it
Pro: Could (potentially) do extra logic to know when a mem_pool no
longer had references to any cache_entry
Con: cache_entry would grow heavier by a pointer, instead of int
We didn't see a tangible benefit to this approach
3) Do not add any extra information to a cache_entry, but when freeing a
cache entry, check if the memory exists in a region managed by existing
mem_pools.
Pro: No extra memory overhead to track state
Con: Extra computation is performed when freeing cache entries
We decided tracking and iterating over known memory pool regions was
less desirable than adding an extra field to track this stae.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It has been observed that the time spent loading an index with a large
number of entries is partly dominated by malloc() calls. This change
is in preparation for using memory pools to reduce the number of
malloc() calls made to allocate cahce entries when loading an index.
Add an API to allocate and discard cache entries, abstracting the
details of managing the memory backing the cache entries. This commit
does actually change how memory is managed - this will be done in a
later commit in the series.
This change makes the distinction between cache entries that are
associated with an index and cache entries that are not associated with
an index. A main use of cache entries is with an index, and we can
optimize the memory management around this. We still have other cases
where a cache entry is not persisted with an index, and so we need to
handle the "transient" use case as well.
To keep the congnitive overhead of managing the cache entries, there
will only be a single discard function. This means there must be enough
information kept with the cache entry so that we know how to discard
them.
A summary of the main functions in the API is:
make_cache_entry: create cache entry for use in an index. Uses specified
parameters to populate cache_entry fields.
make_empty_cache_entry: Create an empty cache entry for use in an index.
Returns cache entry with empty fields.
make_transient_cache_entry: create cache entry that is not used in an
index. Uses specified parameters to populate
cache_entry fields.
make_empty_transient_cache_entry: create cache entry that is not used in
an index. Returns cache entry with
empty fields.
discard_cache_entry: A single function that knows how to discard a cache
entry regardless of how it was allocated.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Refactor refresh_cache_entry() to work on a specific index, instead of
implicitly using the_index. This is in preparation for making the
make_cache_entry function apply to a specific index.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Conversion from uchar[20] to struct object_id continues.
* bc/object-id: (42 commits)
merge-one-file: compute empty blob object ID
add--interactive: compute the empty tree value
Update shell scripts to compute empty tree object ID
sha1_file: only expose empty object constants through git_hash_algo
dir: use the_hash_algo for empty blob object ID
sequencer: use the_hash_algo for empty tree object ID
cache-tree: use is_empty_tree_oid
sha1_file: convert cached object code to struct object_id
builtin/reset: convert use of EMPTY_TREE_SHA1_BIN
builtin/receive-pack: convert one use of EMPTY_TREE_SHA1_HEX
wt-status: convert two uses of EMPTY_TREE_SHA1_HEX
submodule: convert several uses of EMPTY_TREE_SHA1_HEX
sequencer: convert one use of EMPTY_TREE_SHA1_HEX
merge: convert empty tree constant to the_hash_algo
builtin/merge: switch tree functions to use object_id
builtin/am: convert uses of EMPTY_TREE_SHA1_BIN to the_hash_algo
sha1-file: add functions for hex empty tree and blob OIDs
builtin/receive-pack: avoid hard-coded constants for push certs
diff: specify abbreviation size in terms of the_hash_algo
upload-pack: replace use of several hard-coded constants
...
* maint-2.16:
Git 2.16.4
Git 2.15.2
Git 2.14.4
Git 2.13.7
verify_path: disallow symlinks in .gitmodules
update-index: stat updated files earlier
verify_dotfile: mention case-insensitivity in comment
verify_path: drop clever fallthrough
skip_prefix: add case-insensitive variant
is_{hfs,ntfs}_dotgitmodules: add tests
is_ntfs_dotgit: match other .git files
is_hfs_dotgit: match other .git files
is_ntfs_dotgit: use a size_t for traversing string
submodule-config: verify submodule names as paths
* maint-2.15:
Git 2.15.2
Git 2.14.4
Git 2.13.7
verify_path: disallow symlinks in .gitmodules
update-index: stat updated files earlier
verify_dotfile: mention case-insensitivity in comment
verify_path: drop clever fallthrough
skip_prefix: add case-insensitive variant
is_{hfs,ntfs}_dotgitmodules: add tests
is_ntfs_dotgit: match other .git files
is_hfs_dotgit: match other .git files
is_ntfs_dotgit: use a size_t for traversing string
submodule-config: verify submodule names as paths
* maint-2.14:
Git 2.14.4
Git 2.13.7
verify_path: disallow symlinks in .gitmodules
update-index: stat updated files earlier
verify_dotfile: mention case-insensitivity in comment
verify_path: drop clever fallthrough
skip_prefix: add case-insensitive variant
is_{hfs,ntfs}_dotgitmodules: add tests
is_ntfs_dotgit: match other .git files
is_hfs_dotgit: match other .git files
is_ntfs_dotgit: use a size_t for traversing string
submodule-config: verify submodule names as paths
* maint-2.13:
Git 2.13.7
verify_path: disallow symlinks in .gitmodules
update-index: stat updated files earlier
verify_dotfile: mention case-insensitivity in comment
verify_path: drop clever fallthrough
skip_prefix: add case-insensitive variant
is_{hfs,ntfs}_dotgitmodules: add tests
is_ntfs_dotgit: match other .git files
is_hfs_dotgit: match other .git files
is_ntfs_dotgit: use a size_t for traversing string
submodule-config: verify submodule names as paths
There are a few reasons it's not a good idea to make
.gitmodules a symlink, including:
1. It won't be portable to systems without symlinks.
2. It may behave inconsistently, since Git may look at
this file in the index or a tree without bothering to
resolve any symbolic links. We don't do this _yet_, but
the config infrastructure is there and it's planned for
the future.
With some clever code, we could make (2) work. And some
people may not care about (1) if they only work on one
platform. But there are a few security reasons to simply
disallow it:
a. A symlinked .gitmodules file may circumvent any fsck
checks of the content.
b. Git may read and write from the on-disk file without
sanity checking the symlink target. So for example, if
you link ".gitmodules" to "../oops" and run "git
submodule add", we'll write to the file "oops" outside
the repository.
Again, both of those are problems that _could_ be solved
with sufficient code, but given the complications in (1) and
(2), we're better off just outlawing it explicitly.
Note the slightly tricky call to verify_path() in
update-index's update_one(). There we may not have a mode if
we're not updating from the filesystem (e.g., we might just
be removing the file). Passing "0" as the mode there works
fine; since it's not a symlink, we'll just skip the extra
checks.
Signed-off-by: Jeff King <peff@peff.net>
We're more restrictive than we need to be in matching ".GIT"
on case-sensitive filesystems; let's make a note that this
is intentional.
Signed-off-by: Jeff King <peff@peff.net>
We check ".git" and ".." in the same switch statement, and
fall through the cases to share the end-of-component check.
While this saves us a line or two, it makes modifying the
function much harder. Let's just write it out.
Signed-off-by: Jeff King <peff@peff.net>
This should make these functions easier to find and cache.h less
overwhelming to read.
In particular, this moves:
- read_object_file
- oid_object_info
- write_object_file
As a result, most of the codebase needs to #include object-store.h.
In this patch the #include is only added to files that would fail to
compile otherwise. It would be better to #include wherever
identifiers from the header are used. That can happen later
when we have better tooling for it.
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Adjust struct index_state to use struct object_id instead of unsigned
char [20].
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Convert the base_sha1 member of struct split_index to use struct
object_id and rename it base_oid. Include cache.h to make the structure
visible.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
While at there, document about this special mode when running the test
suite.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Conversion from uchar[20] to struct object_id continues.
* bc/object-id: (36 commits)
convert: convert to struct object_id
sha1_file: introduce a constant for max header length
Convert lookup_replace_object to struct object_id
sha1_file: convert read_sha1_file to struct object_id
sha1_file: convert read_object_with_reference to object_id
tree-walk: convert tree entry functions to object_id
streaming: convert istream internals to struct object_id
tree-walk: convert get_tree_entry_follow_symlinks internals to object_id
builtin/notes: convert static functions to object_id
builtin/fmt-merge-msg: convert remaining code to object_id
sha1_file: convert sha1_object_info* to object_id
Convert remaining callers of sha1_object_info_extended to object_id
packfile: convert unpack_entry to struct object_id
sha1_file: convert retry_bad_packed_offset to struct object_id
sha1_file: convert assert_sha1_type to object_id
builtin/mktree: convert to struct object_id
streaming: convert open_istream to use struct object_id
sha1_file: convert check_sha1_signature to struct object_id
sha1_file: convert read_loose_object to use struct object_id
builtin/index-pack: convert struct ref_delta_entry to object_id
...
Code clean-up.
* nd/shared-index-fix:
read-cache: don't write index twice if we can't write shared index
read-cache.c: move tempfile creation/cleanup out of write_shared_index
read-cache.c: change type of "temp" in write_shared_index()
The split-index mode had a few corner case bugs fixed.
* tg/split-index-fixes:
travis: run tests with GIT_TEST_SPLIT_INDEX
split-index: don't write cache tree with null oid entries
read-cache: fix reading the shared index for other repos
The codepath to replace an existing entry in the index had a bug in
updating the name hash structure, which has been fixed.
* bp/refresh-cache-ent-rehash-fix:
Fix bugs preventing adding updated cache entries to the name hash
Internal API clean-up to allow write_locked_index() optionally skip
writing the in-core index when it is not modified.
* ma/skip-writing-unchanged-index:
write_locked_index(): add flag to avoid writing unchanged index
The function ce_write_entry() uses a 'self-initialised' variable
construct, for the symbol 'saved_namelen', to suppress a gcc
'-Wmaybe-uninitialized' warning, given that the warning is a false
positive.
For the purposes of this discussion, the ce_write_entry() function has
three code blocks of interest, that look like so:
/* block #1 */
if (ce->ce_flags & CE_STRIP_NAME) {
saved_namelen = ce_namelen(ce);
ce->ce_namelen = 0;
}
/* block #2 */
/*
* several code blocks that contain, among others, calls
* to copy_cache_entry_to_ondisk(ondisk, ce);
*/
/* block #3 */
if (ce->ce_flags & CE_STRIP_NAME) {
ce->ce_namelen = saved_namelen;
ce->ce_flags &= ~CE_STRIP_NAME;
}
The warning implies that gcc thinks it is possible that the first
block is not entered, the calls to copy_cache_entry_to_ondisk()
could toggle the CE_STRIP_NAME flag on, thereby entering block #3
with saved_namelen unset. However, the copy_cache_entry_to_ondisk()
function does not write to ce->ce_flags (it only reads). gcc could
easily determine this, since that function is local to this file,
but it obviously doesn't.
In order to suppress this warning, we make it clear to the reader
(human and compiler), that block #3 will only be entered when the
first block has been entered, by introducing a new 'stripped_name'
boolean variable. We also take the opportunity to change the type
of 'saved_namelen' to 'unsigned int' to match ce->ce_namelen.
Signed-off-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Update replace_index_entry() to clear the CE_HASHED flag from the new cache
entry so that it can add it to the name hash in set_index_entry()
Fix refresh_cache_ent() to use the copy_cache_entry() macro instead of memcpy()
so that it doesn't incorrectly copy the hash state from the old entry.
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Convert read_sha1_file to take a pointer to struct object_id and rename
it read_object_file. Do the same for read_sha1_file_extended.
Convert one use in grep.c to use the new function without any other code
change, since the pointer being passed is a void pointer that is already
initialized with a pointer to struct object_id. Update the declaration
and definitions of the modified functions, and apply the following
semantic patch to convert the remaining callers:
@@
expression E1, E2, E3;
@@
- read_sha1_file(E1.hash, E2, E3)
+ read_object_file(&E1, E2, E3)
@@
expression E1, E2, E3;
@@
- read_sha1_file(E1->hash, E2, E3)
+ read_object_file(E1, E2, E3)
@@
expression E1, E2, E3, E4;
@@
- read_sha1_file_extended(E1.hash, E2, E3, E4)
+ read_object_file_extended(&E1, E2, E3, E4)
@@
expression E1, E2, E3, E4;
@@
- read_sha1_file_extended(E1->hash, E2, E3, E4)
+ read_object_file_extended(E1, E2, E3, E4)
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We have several callers like
if (active_cache_changed && write_locked_index(...))
handle_error();
rollback_lock_file(...);
where the final rollback is needed because "!active_cache_changed"
shortcuts the if-expression. There are also a few variants of this,
including some if-else constructs that make it more clear when the
explicit rollback is really needed.
Teach `write_locked_index()` to take a new flag SKIP_IF_UNCHANGED and
simplify the callers. Leave the most complicated of the callers (in
builtin/update-index.c) unchanged. Rewriting it to use this new flag
would end up duplicating logic.
We could have made the new flag behave the other way round
("FORCE_WRITE"), but that could break existing users behind their backs.
Let's take the more conservative approach. We can still migrate existing
callers to use our new flag. Later we might even be able to flip the
default, possibly without entirely ignoring the risk to in-flight or
out-of-tree topics.
Suggested-by: Jeff King <peff@peff.net>
Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Rename C++ keyword in order to bring the codebase closer to being able
to be compiled with a C++ compiler.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
More abstraction of hash function from the codepath.
* bc/hash-algo:
hash: update obsolete reference to SHA1_HEADER
bulk-checkin: abstract SHA-1 usage
csum-file: abstract uses of SHA-1
csum-file: rename sha1file to hashfile
read-cache: abstract away uses of SHA-1
pack-write: switch various SHA-1 values to abstract forms
pack-check: convert various uses of SHA-1 to abstract forms
fast-import: switch various uses of SHA-1 to the_hash_algo
sha1_file: switch uses of SHA-1 to the_hash_algo
builtin/unpack-objects: switch uses of SHA-1 to the_hash_algo
builtin/index-pack: improve hash function abstraction
hash: create union for hash context allocation
hash: move SHA-1 macros to hash.h
Code clean-up.
* nd/shared-index-fix:
read-cache: don't write index twice if we can't write shared index
read-cache.c: move tempfile creation/cleanup out of write_shared_index
read-cache.c: change type of "temp" in write_shared_index()
The split-index mode had a few corner case bugs fixed.
* tg/split-index-fixes:
travis: run tests with GIT_TEST_SPLIT_INDEX
split-index: don't write cache tree with null oid entries
read-cache: fix reading the shared index for other repos
Convert various uses of direct calls to SHA-1 and 20- and 40-based
constants to use the_hash_algo instead. Don't yet convert the on-disk
data structures, which will be handled in a future commit.
Adjust some comments so as not to refer explicitly to SHA-1.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
All the known heavy code blocks are measured (except object database
access). This should help identify if an optimization is effective or
not. An unoptimized git-status would give something like below:
0.001791141 s: read cache ...
0.004011363 s: preload index
0.000516161 s: refresh index
0.003139257 s: git command: ... 'status' '--porcelain=2'
0.006788129 s: diff-files
0.002090267 s: diff-index
0.001885735 s: initialize name hash
0.032013138 s: read directory
0.051781209 s: git command: './git' 'status'
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Convert the definition and declaration of write_sha1_file to
struct object_id and adjust usage of this function.
This commit also converts static function write_sha1_file_prepare, as it
is closely related.
Rename these functions to write_object_file and
write_object_file_prepare respectively.
Replace sha1_to_hex, hashcpy and hashclr with their oid equivalents
wherever possible.
Signed-off-by: Patryk Obara <patryk.obara@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a0a967568e ("update-index --split-index: do not split if $GIT_DIR is
read only", 2014-06-13), we tried to make sure we can still write an
index, even if the shared index can not be written.
We did so by just calling 'do_write_locked_index()' just before
'write_shared_index()'. 'do_write_locked_index()' always at least
closes the tempfile nowadays, and used to close or commit the lockfile
if COMMIT_LOCK or CLOSE_LOCK were given at the time this feature was
introduced. COMMIT_LOCK or CLOSE_LOCK is passed in by most callers of
'write_locked_index()'.
After calling 'write_shared_index()', we call 'write_split_index()',
which calls 'do_write_locked_index()' again, which then tries to use the
closed lockfile again, but in fact fails to do so as it's already
closed. This eventually leads to a segfault.
Make sure to write the main index only once.
[nd: most of the commit message and investigation done by Thomas, I only
tweaked the solution a bit]
Helped-by: Thomas Gummerer <t.gummerer@gmail.com>
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Use the helper macro MOVE_ARRAY to move arrays. This is shorter and
safer, as it automatically infers the size of elements.
Patch generated by Coccinelle and contrib/coccinelle/array.cocci in
Travis CI's static analysis build job.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In a96d3cc3f6 ("cache-tree: reject entries with null sha1", 2017-04-21)
we made sure that broken cache entries do not get propagated to new
trees. Part of that was making sure not to re-use an existing cache
tree that includes a null oid.
It did so by dropping the cache tree in 'do_write_index()' if one of
the entries contains a null oid. In split index mode however, there
are two invocations to 'do_write_index()', one for the shared index
and one for the split index. The cache tree is only written once, to
the split index.
As we only loop through the elements that are effectively being
written by the current invocation, that may not include the entry with
a null oid in the split index (when it is already written to the
shared index), where we write the cache tree. Therefore in split
index mode we may still end up writing the cache tree, even though
there is an entry with a null oid in the index.
Fix this by checking for null oids in prepare_to_write_split_index,
where we loop the entries of the shared index as well as the entries for
the split index.
This fixes t7009 with GIT_TEST_SPLIT_INDEX. Also add a new test that's
more specifically showing the problem.
Signed-off-by: Thomas Gummerer <t.gummerer@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
read_index_from() takes a path argument for the location of the index
file. For reading the shared index in split index mode however it just
ignores that path argument, and reads it from the gitdir of the current
repository.
This works as long as an index in the_repository is read. Once that
changes, such as when we read the index of a submodule, or of a
different working tree than the current one, the gitdir of
the_repository will no longer contain the appropriate shared index,
and git will fail to read it.
For example t3007-ls-files-recurse-submodules.sh was broken with
GIT_TEST_SPLIT_INDEX set in 188dce131f ("ls-files: use repository
object", 2017-06-22), and t7814-grep-recurse-submodules.sh was also
broken in a similar manner, probably by introducing struct repository
there, although I didn't track down the exact commit for that.
be489d02d2 ("revision.c: --indexed-objects add objects from all
worktrees", 2017-08-23) breaks with split index mode in a similar
manner, not erroring out when it can't read the index, but instead
carrying on with pruning, without taking the index of the worktree into
account.
Fix this by passing an additional gitdir parameter to read_index_from,
to indicate where it should look for and read the shared index from.
read_cache_from() defaults to using the gitdir of the_repository. As it
is mostly a convenience macro, having to pass get_git_dir() for every
call seems overkill, and if necessary users can have more control by
using read_index_from().
Helped-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Thomas Gummerer <t.gummerer@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
For one thing, we have more consistent cleanup procedure now and always
keep errno intact.
The real purpose is the ability to break out of write_locked_index()
early when mks_tempfile() fails in the next patch. It's more awkward to
do it if this mks_tempfile() is still inside write_shared_index().
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This local variable 'temp' will be passed in from the caller in the next
patch. To reduce patch noise, let's change its type now while it's still
a local variable and get all the trival conversion out of the next patch.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git add --renormalize ." is a new and safer way to record the fact
that you are correcting the end-of-line convention and other
"convert_to_git()" glitches in the in-repository data.
* tb/add-renormalize:
add: introduce "--renormalize"
Various fixes to bp/fsmonitor topic.
* av/fsmonitor:
fsmonitor: simplify determining the git worktree under Windows
fsmonitor: store fsmonitor bitmap before splitting index
fsmonitor: read from getcwd(), not the PWD environment variable
fsmonitor: delay updating state until after split index is merged
fsmonitor: document GIT_TRACE_FSMONITOR
fsmonitor: don't bother pretty-printing JSON from watchman
fsmonitor: set the PWD to the top of the working tree
We learned to talk to watchman to speed up "git status" and other
operations that need to see which paths have been modified.
* bp/fsmonitor:
fsmonitor: preserve utf8 filenames in fsmonitor-watchman log
fsmonitor: read entirety of watchman output
fsmonitor: MINGW support for watchman integration
fsmonitor: add a performance test
fsmonitor: add a sample integration script for Watchman
fsmonitor: add test cases for fsmonitor extension
split-index: disable the fsmonitor extension when running the split index test
fsmonitor: add a test tool to dump the index extension
update-index: add fsmonitor support to update-index
ls-files: Add support in ls-files to display the fsmonitor valid bit
fsmonitor: add documentation for the fsmonitor extension.
fsmonitor: teach git to optionally utilize a file system monitor to speed up detecting new or changed files.
update-index: add a new --force-write-index option
preload-index: add override to enable testing preload-index
bswap: add 64 bit endianness helper get_be64
Make it safer to normalize the line endings in a repository.
Files that had been commited with CRLF will be commited with LF.
The old way to normalize a repo was like this:
# Make sure that there are not untracked files
$ echo "* text=auto" >.gitattributes
$ git read-tree --empty
$ git add .
$ git commit -m "Introduce end-of-line normalization"
The user must make sure that there are no untracked files,
otherwise they would have been added and tracked from now on.
The new "add --renormalize" does not add untracked files:
$ echo "* text=auto" >.gitattributes
$ git add --renormalize .
$ git commit -m "Introduce end-of-line normalization"
Note that "git add --renormalize <pathspec>" is the short form for
"git add -u --renormalize <pathspec>".
While at it, document that the same renormalization may be needed,
whenever a clean filter is added or changed.
Helped-By: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Drop (perhaps overly cautious) sanity check before using the index
read from the filesystem at runtime.
* bp/read-index-from-skip-verification:
read_index_from(): speed index loading by skipping verification of the entry order
ba1b9cac ("fsmonitor: delay updating state until after split index
is merged", 2017-10-27) resolved the problem of the fsmonitor data
being applied to the non-base index when reading; however, a similar
problem exists when writing the index. Specifically, writing of the
fsmonitor extension happens only after the work to split the index
has been applied -- as such, the information in the index is only
for the non-"base" index, and thus the extension information
contains only partial data.
When saving, compute the ewah bitmap before the index is split, and
store it in the fsmonitor_dirty field, mirroring the behavior that
occurred during reading. fsmonitor_dirty is kept from being leaked by
being freed when the extension data is written -- which always happens
precisely once, no matter the split index configuration.
Signed-off-by: Alex Vandiver <alexmv@dropbox.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There is code in post_read_index_from() to catch out of order
entries when reading an index file. This order verification is ~13%
of the cost of every call to read_index_from().
Update check_ce_order() so that it skips this verification unless
the "verify_ce_order" global variable is set.
Teach fsck to force this verification.
The effect can be seen using t/perf/p0002-read-cache.sh:
Test HEAD HEAD~1
--------------------------------------------------------------------------------------
0002.1: read_cache/discard_cache 1000 times 0.41(0.04+0.04) 0.50(0.00+0.10) +22.0%
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Conversion from uchar[20] to struct object_id continues.
* bc/object-id: (25 commits)
refs/files-backend: convert static functions to object_id
refs: convert read_raw_ref backends to struct object_id
refs: convert peel_object to struct object_id
refs: convert resolve_ref_unsafe to struct object_id
worktree: convert struct worktree to object_id
refs: convert resolve_gitlink_ref to struct object_id
Convert remaining callers of resolve_gitlink_ref to object_id
sha1_file: convert index_path and index_fd to struct object_id
refs: convert reflog_expire parameter to struct object_id
refs: convert read_ref_at to struct object_id
refs: convert peel_ref to struct object_id
builtin/pack-objects: convert to struct object_id
pack-bitmap: convert traverse_bitmap_commit_list to object_id
refs: convert dwim_log to struct object_id
builtin/reflog: convert remaining unsigned char uses to object_id
refs: convert dwim_ref and expand_ref to struct object_id
refs: convert read_ref and read_ref_full to object_id
refs: convert resolve_refdup and refs_resolve_refdup to struct object_id
Convert check_connected to use struct object_id
refs: update ref transactions to use struct object_id
...
Convert the declaration and definition of resolve_gitlink_ref to use
struct object_id and apply the following semantic patch:
@@
expression E1, E2, E3;
@@
- resolve_gitlink_ref(E1, E2, E3.hash)
+ resolve_gitlink_ref(E1, E2, &E3)
@@
expression E1, E2, E3;
@@
- resolve_gitlink_ref(E1, E2, E3->hash)
+ resolve_gitlink_ref(E1, E2, E3)
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
`update_index_if_able()` used to always commit the lock or roll it back.
Commit 03b866477 (read-cache: new API write_locked_index instead of
write_index/write_cache, 2014-06-13) stopped rolling it back in case a
write was not even attempted. This change in behavior is not motivated
in the commit message and appears to be accidental: the `else`-path was
removed, although that changed the behavior in case the `if` shortcuts.
Reintroduce the rollback and document this behavior. While at it, move
the documentation on this function from the function definition to the
function declaration in cache.h.
If `write_locked_index(..., COMMIT_LOCK)` fails, it will roll back the
lock for us (see the previous commit).
Noticed-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If the original version of `write_locked_index()` returned with an
error, it didn't roll back the lockfile unless the error occured at the
very end, during closing/committing. See commit 03b866477 (read-cache:
new API write_locked_index instead of write_index/write_cache,
2014-06-13).
In commit 9f41c7a6b (read-cache: close index.lock in do_write_index,
2017-04-26), we learned to close the lock slightly earlier in the
callstack. That was mostly a side-effect of lockfiles being implemented
using temporary files, but didn't cause any real harm.
Recently, commit 076aa2cbd (tempfile: auto-allocate tempfiles on heap,
2017-09-05) introduced a subtle bug. If the temporary file is deleted
(i.e., the lockfile is rolled back), the tempfile-pointer in the `struct
lock_file` will be left dangling. Thus, an attempt to reuse the
lockfile, or even just to roll it back, will induce undefined behavior
-- most likely a crash.
Besides not crashing, we clearly want to make things consistent. The
guarantees which the lockfile-machinery itself provides is A) if we ask
to commit and it fails, roll back, and B) if we ask to close and it
fails, do _not_ roll back. Let's do the same for consistency.
Do not delete the temporary file in `do_write_index()`. One of its
callers, `write_locked_index()` will thereby avoid rolling back the
lock. The other caller, `write_shared_index()`, will delete its
temporary file anyway. Both of these callers will avoid undefined
behavior (crashing).
Teach `write_locked_index(..., COMMIT_LOCK)` to roll back the lock
before returning. If we have already succeeded and committed, it will be
a noop. Simplify the existing callers where we now have a superfluous
call to `rollback_lockfile()`. That should keep future readers from
wondering why the callers are inconsistent.
Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
`write_locked_index()` takes two flags: `COMMIT_LOCK` and `CLOSE_LOCK`.
At most one is allowed. But it is also possible to use no flag, i.e.,
`0`. But when `write_locked_index()` calls `do_write_index()`, the
temporary file, a.k.a. the lockfile, will be closed. So passing `0` is
effectively the same as `CLOSE_LOCK`, which seems like a bug.
We might feel tempted to restructure the code in order to close the file
later, or conditionally. It also feels a bit unfortunate that we simply
"happen" to close the lock by way of an implementation detail of
lockfiles. But note that we need to close the temporary file before
`stat`-ing it, at least on Windows. See 9f41c7a6b (read-cache: close
index.lock in do_write_index, 2017-04-26).
Drop `CLOSE_LOCK` and make it explicit that `write_locked_index()`
always closes the lock. Whether it is also committed is governed by the
remaining flag, `COMMIT_LOCK`.
This means we neither have nor suggest that we have a mode to write the
index and leave the file open. Whatever extra contents we might
eventually want to write, we should probably write it from within
`write_locked_index()` itself anyway.
Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When the index is read from disk, the fsmonitor index extension is used
to flag the last known potentially dirty index entries. The registered
core.fsmonitor command is called with the time the index was last
updated and returns the list of files changed since that time. This list
is used to flag any additional dirty cache entries and untracked cache
directories.
We can then use this valid state to speed up preload_index(),
ie_match_stat(), and refresh_cache_ent() as they do not need to lstat()
files to detect potential changes for those entries marked
CE_FSMONITOR_VALID.
In addition, if the untracked cache is turned on valid_cached_dir() can
skip checking directories for new or changed files as fsmonitor will
invalidate the cache only for those directories that have been
identified as having potential changes.
To keep the CE_FSMONITOR_VALID state accurate during git operations;
when git updates a cache entry to match the current state on disk,
it will now set the CE_FSMONITOR_VALID bit.
Inversely, anytime git changes a cache entry, the CE_FSMONITOR_VALID bit
is cleared and the corresponding untracked cache directory is marked
invalid.
Signed-off-by: Ben Peart <benpeart@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Many codepaths have been updated to squelch -Wimplicit-fallthrough
warnings from Gcc 7 (which is a good code hygiene).
* jk/fallthrough:
consistently use "fallthrough" comments in switches
curl_trace(): eliminate switch fallthrough
test-line-buffer: simplify command parsing
Many codepaths did not diagnose write failures correctly when disks
go full, due to their misuse of write_in_full() helper function,
which have been corrected.
* jk/write-in-full-fix:
read_pack_header: handle signed/unsigned comparison in read result
config: flip return value of store_write_*()
notes-merge: use ssize_t for write_in_full() return value
pkt-line: check write_in_full() errors against "< 0"
convert less-trivial versions of "write_in_full() != len"
avoid "write_in_full(fd, buf, len) != len" pattern
get-tar-commit-id: check write_in_full() return against 0
config: avoid "write_in_full(fd, buf, len) < len" pattern
A hotfix to a topic already in 'master'.
* kw/write-index-reduce-alloc:
read-cache: fix index corruption with index v4
Add t/helper/test-write-cache to .gitignore
Gcc 7 adds -Wimplicit-fallthrough, which can warn when a
switch case falls through to the next case. The general idea
is that the compiler can't tell if this was intentional or
not, so you should annotate any intentional fall-throughs as
such, leaving it to complain about any unannotated ones.
There's a GNU __attribute__ which can be used for
annotation, but of course we'd have to #ifdef it away on
non-gcc compilers. Gcc will also recognize
specially-formatted comments, which matches our current
practice. Let's extend that practice to all of the
unannotated sites (which I did look over and verify that
they were behaving as intended).
Ideally in each case we'd actually give some reasons in the
comment about why we're falling through, or what we're
falling through to. And gcc does support that with
-Wimplicit-fallthrough=2, which relaxes the comment pattern
matching to anything that contains "fallthrough" (or a
variety of spelling variants). However, this isn't the
default for -Wimplicit-fallthrough, nor for -Wextra. In the
name of simplicity, it's probably better for us to support
the default level, which requires "fallthrough" to be the
only thing in the comment (modulo some window dressing like
"else" and some punctuation; see the gcc manual for the
complete set of patterns).
This patch suppresses all warnings due to
-Wimplicit-fallthrough. We might eventually want to add that
to the DEVELOPER Makefile knob, but we should probably wait
until gcc 7 is more widely adopted (since earlier versions
will complain about the unknown warning type).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The return value of write_in_full() is either "-1", or the
requested number of bytes[1]. If we make a partial write
before seeing an error, we still return -1, not a partial
value. This goes back to f6aa66cb95 (write_in_full: really
write in full or return error on disk full., 2007-01-11).
So checking anything except "was the return value negative"
is pointless. And there are a couple of reasons not to do
so:
1. It can do a funny signed/unsigned comparison. If your
"len" is signed (e.g., a size_t) then the compiler will
promote the "-1" to its unsigned variant.
This works out for "!= len" (unless you really were
trying to write the maximum size_t bytes), but is a
bug if you check "< len" (an example of which was fixed
recently in config.c).
We should avoid promoting the mental model that you
need to check the length at all, so that new sites are
not tempted to copy us.
2. Checking for a negative value is shorter to type,
especially when the length is an expression.
3. Linus says so. In d34cf19b89 (Clean up write_in_full()
users, 2007-01-11), right after the write_in_full()
semantics were changed, he wrote:
I really wish every "write_in_full()" user would just
check against "<0" now, but this fixes the nasty and
stupid ones.
Appeals to authority aside, this makes it clear that
writing it this way does not have an intentional
benefit. It's a historical curiosity that we never
bothered to clean up (and which was undoubtedly
cargo-culted into new sites).
So let's convert these obviously-correct cases (this
includes write_str_in_full(), which is just a wrapper for
write_in_full()).
[1] A careful reader may notice there is one way that
write_in_full() can return a different value. If we ask
write() to write N bytes and get a return value that is
_larger_ than N, we could return a larger total. But
besides the fact that this would imply a totally broken
version of write(), it would already invoke undefined
behavior. Our internal remaining counter is an unsigned
size_t, which means that subtracting too many byte will
wrap it around to a very large number. So we'll instantly
begin reading off the end of the buffer, trying to write
gigabytes (or petabytes) of data.
Signed-off-by: Jeff King <peff@peff.net>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
ce012deb98 ("read-cache: avoid allocating every ondisk entry when
writing", 2017-08-21) changed the way cache entries are written to the
index file. While previously it wrote the name to an struct that was
allocated using xcalloc(), it now uses ce_write() directly. Previously
ce_namelen - common bytes were written to the cache entry, which would
automatically make it nul terminated, as it was allocated using calloc.
Now we are writing ce_namelen - common + 1 bytes directly from the
ce->name to the index. If CE_STRIP_NAME however gets set in the split
index case ce->ce_namelen is set to 0 without changing the actual
ce->name buffer. When index-v4, this results in the first character of
ce->name being written out instead of just a terminating nul charcter.
As index-v4 requires the terminating nul character as terminator of
the name when reading it back, this results in a corrupted index.
Fix that by only writing ce_namelen - common bytes directly from
ce->name to the index, and adding the nul terminator in an extra call to
ce_write.
This bug was turned up by setting TEST_GIT_INDEX_VERSION = 4 in
config.mak and running the test suite (t1700 specifically broke).
Signed-off-by: Thomas Gummerer <t.gummerer@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The previous commit taught the tempfile code to give up
ownership over tempfiles that have been renamed or deleted.
That makes it possible to use a stack variable like this:
struct tempfile t;
create_tempfile(&t, ...);
...
if (!err)
rename_tempfile(&t, ...);
else
delete_tempfile(&t);
But doing it this way has a high potential for creating
memory errors. The tempfile we pass to create_tempfile()
ends up on a global linked list, and it's not safe for it to
go out of scope until we've called one of those two
deactivation functions.
Imagine that we add an early return from the function that
forgets to call delete_tempfile(). With a static or heap
tempfile variable, the worst case is that the tempfile hangs
around until the program exits (and some functions like
setup_shallow_temporary rely on this intentionally, creating
a tempfile and then leaving it for later cleanup).
But with a stack variable as above, this is a serious memory
error: the variable goes out of scope and may be filled with
garbage by the time the tempfile code looks at it. Let's
see if we can make it harder to get this wrong.
Since many callers need to allocate arbitrary numbers of
tempfiles, we can't rely on static storage as a general
solution. So we need to turn to the heap. We could just ask
all callers to pass us a heap variable, but that puts the
burden on them to call free() at the right time.
Instead, let's have the tempfile code handle the heap
allocation _and_ the deallocation (when the tempfile is
deactivated and removed from the list).
This changes the return value of all of the creation
functions. For the cleanup functions (delete and rename),
we'll add one extra bit of safety: instead of taking a
tempfile pointer, we'll take a pointer-to-pointer and set it
to NULL after freeing the object. This makes it safe to
double-call functions like delete_tempfile(), as the second
call treats the NULL input as a noop. Several callsites
follow this pattern.
The resulting patch does have a fair bit of noise, as each
caller needs to be converted to handle:
1. Storing a pointer instead of the struct itself.
2. Passing the pointer instead of taking the struct
address.
3. Handling a "struct tempfile *" return instead of a file
descriptor.
We could play games to make this less noisy. For example, by
defining the tempfile like this:
struct tempfile {
struct heap_allocated_part_of_tempfile {
int fd;
...etc
} *actual_data;
}
Callers would continue to have a "struct tempfile", and it
would be "active" only when the inner pointer was non-NULL.
But that just makes things more awkward in the long run.
There aren't that many callers, so we can simply bite
the bullet and adjust all of them. And the compiler makes it
easy for us to find them all.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since the lockfile code is based on the tempfile code, it
has some of the same problems, including that close_lock_file()
erases the tempfile's filename buf, making it hard for the
caller to write a good error message.
In practice this comes up less for lockfiles than for
straight tempfiles, since we usually just report the
refname. But there is at least one buggy case in
write_ref_to_lockfile(). Besides, given the coupling between
the lockfile and tempfile modules, it's less confusing if
their close() functions have the same semantics.
Just as the previous commit did for close_tempfile(), let's
teach close_lock_file() and its wrapper close_ref() not to
rollback on error. And just as before, we'll give them new
"gently" names to catch any new callers that are added.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
