Change the "nr" and "alloc" members of "struct string_list" to use
"size_t" instead of "nr". On some platforms the size of an "unsigned
int" will be smaller than a "size_t", e.g. a 32 bit unsigned v.s. 64
bit unsigned. As "struct string_list" is a generic API we use in a lot
of places this might cause overflows.
As one example: code in "refs.c" keeps track of the number of refs
with a "size_t", and auxiliary code in builtin/remote.c in
get_ref_states() appends those to a "struct string_list".
While we're at it split the "nr" and "alloc" in string-list.h across
two lines, which is the case for most such struct member
declarations (e.g. in "strbuf.h" and "strvec.h").
Changing e.g. "int i" to "size_t i" in run_and_feed_hook() isn't
strictly necessary, and there are a lot more cases where we'll use a
local "int", "unsigned int" etc. variable derived from the "nr" in the
"struct string_list". But in that case as well as
add_wrapped_shortlog_msg() in builtin/shortlog.c we need to adjust the
printf format referring to "nr" anyway, so let's also change the other
variables referring to it.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change a few stray users of the inline gettext.h Q_() function to stop
casting its "n" argument, the vast majority of the users of that
wrapper API use the implicit cast to "unsigned long".
The ngettext() function (which Q_() resolves to) takes an "unsigned
long int", and so does our Q_() wrapper for it, see 0c9ea33b90 (i18n:
add stub Q_() wrapper for ngettext, 2011-03-09). The function isn't
ours, but provided by e.g. GNU libintl.
This amends code added in added in 7171a0b0cf (index-pack: correct
"len" type in unpack_data(), 2016-07-13). The cast it added for the
printf format to die() was needed, but not the cast to Q_().
Likewise the casts in strbuf.c added in 8f354a1fae (l10n: localizable
upload progress messages, 2019-07-02) and for
builtin/merge-recursive.c in ccf7813139 (i18n: merge-recursive: mark
error messages for translation, 2016-09-15) weren't needed.
In the latter case the cast was copy/pasted from the argument to
warning() itself, added in b74d779bd9 (MinGW: Fix compiler warning in
merge-recursive, 2009-05-23). The cast for warning() is needed, but
not the one for ngettext()'s "n" argument.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The Generation Data (GDAT) and Generation Data Overflow (GDOV) chunks
store corrected commit date offsets, used for generation number v2.
Recent changes have demonstrated that previous versions of Git were
incorrectly parsing data from these chunks, but might have also been
writing them incorrectly.
I asserted [1] that the previous fixes were sufficient because the known
reasons for incorrectly writing generation number v2 data relied on
parsing the information incorrectly out of a commit-graph file, but the
previous versions of Git were not reading the generation number v2 data.
However, Patrick demonstrated [2] a case where in split commit-graphs
across an alternate boundary (and possibly some other special
conditions) it was possible to have a commit-graph that was generated by
a previous version of Git have incorrect generation number v2 data which
results in errors like the following:
commit-graph generation for commit <oid> is 1623273624 < 1623273710
[1] https://lore.kernel.org/git/f50e74f0-9ffa-f4f2-4663-269801495ed3@github.com/
[2] https://lore.kernel.org/git/Yh93vOkt2DkrGPh2@ncase/
Clearly, there is something else going on. The situation is not
completely understood, but the errors do not reproduce if the
commit-graphs are all generated by a Git version including these recent
fixes.
If we cannot trust the existing data in the GDAT and GDOV chunks, then
we can alter the format to change the chunk IDs for these chunks. This
causes the new version of Git to silently ignore the older chunks (and
disabling generation number v2 in the process) while writing new
commit-graph files with correct data in the GDA2 and GDO2 chunks.
Update commit-graph-format.txt including a historical note about these
deprecated chunks.
Reported-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Derrick Stolee <derrickstolee@github.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Many output modes of "ls-files" do not work with its
"--recurse-submodules" option, but the "-s" mode has been taught to
work with it.
* jt/ls-files-stage-recurse:
ls-files: support --recurse-submodules --stage
"git checkout -b branch/with/multi/level/name && git stash" only
recorded the last level component of the branch name, which has
been corrected.
* gc/stash-on-branch-with-multi-level-name:
stash: strip "refs/heads/" with skip_prefix
The error message given by "git switch HEAD~4" has been clarified
to suggest the "--detach" option that is required.
* ah/advice-switch-requires-detach-to-detach:
switch: mention the --detach option when dying due to lack of a branch
Use designated initializers we started using in mid 2017 in more
parts of the codebase that are relatively quiescent.
* ab/c99-designated-initializers:
fast-import.c: use designated initializers for "partial" struct assignments
refspec.c: use designated initializers for "struct refspec_item"
convert.c: use designated initializers for "struct stream_filter*"
userdiff.c: use designated initializers for "struct userdiff_driver"
archive-*.c: use designated initializers for "struct archiver"
object-file: use designated initializers for "struct git_hash_algo"
trace2: use designated initializers for "struct tr2_dst"
trace2: use designated initializers for "struct tr2_tgt"
imap-send.c: use designated initializers for "struct imap_server_conf"
When "index-pack" dies due to incoming data exceeding the maximum
allowed input size, include the value of the limit in the error
message.
* mc/index-pack-report-max-size:
index-pack: clarify the breached limit
Tighten the language around "working tree" and "worktree" in the
docs.
* ds/worktree-docs:
worktree: use 'worktree' over 'working tree'
worktree: use 'worktree' over 'working tree'
worktree: use 'worktree' over 'working tree'
worktree: use 'worktree' over 'working tree'
worktree: use 'worktree' over 'working tree'
worktree: use 'worktree' over 'working tree'
worktree: use 'worktree' over 'working tree'
worktree: extract checkout_worktree()
worktree: extract copy_sparse_checkout()
worktree: extract copy_filtered_worktree_config()
worktree: combine two translatable messages
A not-so-common mistake is to write a script to feed "git bisect
run" without making it executable, in which case all tests will
exit with 126 or 127 error codes, even on revisions that are marked
as good. Try to recognize this situation and stop iteration early.
* rs/bisect-executable-not-found:
bisect--helper: double-check run command on exit code 126 and 127
bisect: document run behavior with exit codes 126 and 127
bisect--helper: release strbuf and strvec on run error
bisect--helper: report actual bisect_state() argument on error
Further polishing of "git sparse-checkout".
* en/sparse-checkout-fixes:
sparse-checkout: reject arguments in cone-mode that look like patterns
sparse-checkout: error or warn when given individual files
sparse-checkout: pay attention to prefix for {set, add}
sparse-checkout: correctly set non-cone mode when expected
sparse-checkout: correct reapply's handling of options
Test modernization.
* cg/t3903-modernize:
tests: make the code more readable
tests: allow testing if a path is truly a file or a directory
t/t3903-stash.sh: replace test [-d|-f] with test_path_is_*
The "struct path_cache" added in 102de880d2 (path.c: migrate global
git_path_* to take a repository argument, 2018-05-17) is only used
directly by code in repository.[ch] (but populated in path.[ch]).
Let's move this code to repository.[ch], and stop leaking this memory
when we run repo_clear(). To avoid the cast change it from a "const
char *" to a "char *".
This also removes the "PATH_CACHE_INIT" macro, which has never been
used for anything. For the "struct repository" we already make a hard
assumption that it (and "the_repository") can be identically
initialized by making it a "static" variable, so making use of a
"PATH_CACHE_INIT" somewhere would have been confusing.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Amend code added in d9c66f0b5b (range-diff: first rudimentary
implementation, 2018-08-13) to use a "goto cleanup" pattern. This
makes for less code, and frees memory that we'd previously leak.
The reason for changing free(util) to FREE_AND_NULL(util) is because
at the end of the function we append the contents of "util" to a
"struct string_list" if it's non-NULL.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Create a public release_patch() version of the private free_patch()
function added in 13b5af22f3 (apply: move libified code from
builtin/apply.c to apply.{c,h}, 2016-04-22). Unlike the existing
function this one doesn't free() the "struct patch" itself, so we can
use it for variables on the stack.
Use it in range-diff.c to fix a memory leak in common range-diff
invocations, e.g.:
git -P range-diff origin/master origin/next origin/seen
Would emit several errors when compiled with SANITIZE=leak, but now
runs cleanly.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix a memory leak in code added in 6c622f9f0b (commit-graph: write
commit-graph chains, 2019-06-18). We needed to free the "lock_name" if
we encounter errors, and the "graph_name" after we'd run unlink() on
it.
For the case of write_commit_graph_file() refactoring the code to free
the "lock_name" after we were done using the "struct lock_file lk"
would have made the control flow more complex. Luckily we can free the
"lock_file" right after the hold_lock_file_for_update() call, if it
makes use of "path" at all it'll have copied its contents to a "struct
strbuf" of its own.
While I'm at it let's fix code added in fb10ca5b54 (sparse-checkout:
write using lockfile, 2019-11-21) in write_patterns_and_update() to
avoid the same complexity that I thought I needed when I wrote the
initial fix for write_commit_graph_file(). We can free the
"sparse_filename" right after calling hold_lock_file_for_update(), we
don't need to wait until we're exiting the function to do so.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix a bug in fill_oids_from_packs(), we should always stop_progress(),
but did not do so if we returned an error here. This also plugs a
memory leak in those cases by releasing the two "struct strbuf"
variables the function uses.
While I'm at it stop hardcoding "-1" here and just use the return
value of error() instead, which happens to be "-1".
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When this code was migrated to the string_list API in
d88b14b3fd (commit-graph: use string-list API for input, 2018-06-27)
it was made to use used both STRING_LIST_INIT_NODUP and a
strbuf_detach() pattern.
Those should not be used together if string_list_clear() is expected
to free the memory, instead we need to either use STRING_LIST_INIT_DUP
with a string_list_append_nodup(), or a STRING_LIST_INIT_NODUP and
manually fiddle with the "strdup_strings" member before calling
string_list_clear(). Let's do the former.
Since "strdup_strings = 1" is set now other code might be broken by
relying on "pack_indexes" not to duplicate it strings, but that
doesn't happen. When we pass this down to write_commit_graph() that
code uses the "struct string_list" without modifying it. Let's add a
"const" to the variable to have the compiler enforce that assumption.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix a memory leak in code added in a6226fd772 (submodule--helper:
convert the bulk of cmd_add() to C, 2021-08-10). If "realrepo" isn't a
copy of the "repo" member we should free() it.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Amend the logic added in fddf2ebe38 (transport: teach all vtables to
allow fetch first, 2019-08-21) and save ourselves pointless work in
fetch_refs_from_bundle().
The fetch_refs_from_bundle() caller doesn't care about the "struct
ref *result" return value of get_refs_from_bundle(), and doesn't need
any of the work we were doing in looping over the
"data->header.references" in get_refs_from_bundle().
So this change saves us work, and also fixes a memory leak that we had
when called from fetch_refs_from_bundle(). The other caller of
get_refs_from_bundle() is the "get_refs_list" member we set up for the
"struct transport_vtable bundle_vtable". That caller does care about
the "struct ref *result" return value.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fixing this small memory leak in cmd_bundle_create() gets
"t5607-clone-bundle.sh" closer to passing under SANITIZE=leak.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Plug a trivial memory leak in code added in a2d725b7bd (Use an
external program to implement fetching with curl, 2009-08-05).
To do this have the cmd_main() use a "goto cleanup" pattern, and to
return an error of 1 unless we can fall through to the http_cleanup()
at the end.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Plug a memory leak in credential_apply_config() by adding and using a
new urlmatch_config_release() function. This just does a
string_list_clear() on the "vars" member.
This finished up work on normalizing the init/free pattern in this
API, started in 73ee449bbf (urlmatch.[ch]: add and use
URLMATCH_CONFIG_INIT, 2021-10-01).
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Amend the freeing logic added in e6e045f803 (diff.c: buffer all
output if asked to, 2017-06-29) to free the containing "buf" in
addition to its members.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix a memory leak in 53eda89b2f (merge-base: teach "git merge-base"
to drive underlying merge_bases_many(), 2008-07-30) by calling free()
on the "struct commit **" list used by "git merge-base".
This gets e.g. "t6010-merge-base.sh" closer to passing under
SANITIZE=leak, it failed 8 tests before when compiled with that
option, and now fails only 5 tests.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix various memory leaks in "git index-pack", due to how tightly
coupled this command is with the revision walking this doesn't make
any new tests pass.
But e.g. this now passes, and had several failures before, i.e. we
still have failures in tests 3, 5 etc., which are being skipped here.
./t5300-pack-object.sh --run=1-2,4,6-27,30-42
It is a bit odd that we'll free "opts.anomaly", since the "opts" is a
"struct pack_idx_option" declared in pack.h. In pack-write.c there's a
reset_pack_idx_option(), but it only wipes the contents, but doesn't
free() anything.
Doing this here in cmd_index_pack() is correct because while the
struct is declared in pack.h, this code in builtin/index-pack.c (in
read_v2_anomalous_offsets()) is what allocates the "opts.anomaly", so
we should also free it here.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The gpg-agent is one of several processes that newer releases of GnuPG
start automatically. Issue a kill to each of them to ensure they do not
affect separate tests. (Yes, the separate GNUPGHOME should do that
already. If we find that is case, we could drop the --kill entirely.)
In terms of compatibility, the 'all' keyword was added to the --kill &
--reload options in GnuPG 2.1.18. Debian and RHEL are often used as
indicators of how a change might affect older systems we often try to
support.
- Debian Strech (old old stable), which has limited security support
until June 2022, has GnuPG 2.1.18 (or 2.2.x in backports).
- CentOS/RHEL 7, which is supported until June 2024, has GnuPG
2.0.22, which lacks the --kill option, so the change won't have
any impact.
Signed-off-by: Todd Zullinger <tmz@pobox.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
With gpgsm from gnupg-2.3, the changes to the trustlist.txt do not
appear to be picked up without refreshing the gpg-agent. Use the 'all'
keyword to reload all of the gpg components. The scdaemon is started as
a child of gpg-agent, for example.
We used to have a --kill at this spot, but I removed it in 2e285e7803
(t/lib-gpg: drop redundant killing of gpg-agent, 2019-02-07). It seems
like it might be necessary (again) for 2.3.
Signed-off-by: Todd Zullinger <tmz@pobox.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Checking if signing was successful will now accept '[GNUPG]:
SIG_CREATED' on the beginning of the first or any subsequent line. Not
just explictly the second one anymore.
Gpgsm v2.3 changed its output when listing keys from `fingerprint` to
`sha1/2 fpr`. This leads to the gpgsm tests silently not being executed
because of a failed prerequisite.
Switch to gpg's `--with-colons` output format when evaluating test
prerequisites to make parsing more robust. This also allows us to
combine the existing grep/cut/tr/echo pipe for writing the trustlist.txt
into a single awk expression.
Adjust error message checking in test for v2.3 specific output changes.
Helped-By: Junio C Hamano <gitster@pobox.com>
Helped-By: Todd Zullinger <tmz@pobox.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix a regression in ff37a60c36 (log tests: check if grep_config() is
called by "log"-like cmds, 2022-02-16), a "test_done" command used
during development made it into a submitted patch causing tests 41-136
in t/t4202-log.sh to be skipped.
Reported-by: Fabian Stelzer <fs@gigacodes.de>
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The usage help for --type option of `git config` is missing `type`
in the argument placeholder (`<>`). Add it.
Signed-off-by: Matheus Felipe <matheusfelipeog@protonmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When renaming a remote, Git needs to rename all remote tracking
references to the remote's new name (e.g., renaming
"refs/remotes/old/foo" to "refs/remotes/new/foo" when renaming a remote
from "old" to "new").
This can be somewhat slow when there are many references to rename,
since each rename is done in a separate call to rename_ref() as opposed
to grouping all renames together into the same transaction. It would be
nice to execute all renames as a single transaction, but there is a
snag: the reference transaction backend doesn't support renames during a
transaction (only individually, via rename_ref()).
The reasons there are described in more detail in [1], but the main
problem is that in order to preserve the existing reflog, it must be
moved while holding both locks (i.e., on "oldname" and "newname"), and
the ref transaction code doesn't support inserting arbitrary actions
into the middle of a transaction like that.
As an aside, adding support for this to the ref transaction code is
less straightforward than inserting both a ref_update() and ref_delete()
call into the same transaction. rename_ref()'s special handling to
detect D/F conflicts would need to be rewritten for the transaction code
if we wanted to proactively catch D/F conflicts when renaming a
reference during a transaction. The reftable backend could support this
much more readily because of its lack of D/F conflicts.
Instead of a more complex modification to the ref transaction code,
display a progress meter when running verbosely in order to convince the
user that Git is doing work while renaming a remote.
This is mostly done as-expected, with the minor caveat that we
intentionally count symrefs renames twice, since renaming a symref takes
place over two separate calls (one to delete the old one, and another to
create the new one).
[1]: https://lore.kernel.org/git/572367B4.4050207@alum.mit.edu/
Suggested-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The 'git remote rename' command doesn't currently take any command-line
arguments besides the existing and new name of a remote, and so has no
need to call parse_options().
But the subsequent patch will add a `--[no-]progress` option, in which
case we will need to call parse_options().
Do so now so as to avoid cluttering the following patch with noise, like
adjusting setting `rename.{old,new}_name` to argv[0] and argv[1], since
parse_options handles advancing argv past the name of the sub-command.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In the testcase to exercise backfilling of tags for fetches we evoke a
failure of the backfilling mechanism by creating a reference that later
on causes a D/F conflict. Because the assumption was that git-fetch(1)
would notice the D/F conflict early on this conflicting reference was
created via the reference-transaction hook just when we were about to
write the backfilled tag. As it turns out though this is not the case,
and the fetch fails in the same way when we create the conflicting ref
up front.
Simplify the test setup creating the reference up front, which allows us
to get rid of the hook script.
Signed-off-by: Patrick Steinhardt <ps@pks.im>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This makes it consistent with the rest of the documentation.
Signed-off-by: Nihal Jere <nihal@nihaljere.xyz>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a template to do the "mkdir -p" of $(@D) (the parent dir of $@)
for us, and use it for the "make lint-docs" targets I added in
8650c6298c (doc lint: make "lint-docs" non-.PHONY, 2021-10-15).
As seen in 4c64fb5aad (Documentation/Makefile: fix lint-docs mkdir
dependency, 2021-10-26) maintaining these manual lists of parent
directory dependencies is fragile, in addition to being obviously
verbose.
I used this pattern at the time because I couldn't find another method
than "order-only" prerequisites to avoid doing a "mkdir -p $(@D)" for
every file being created, which as noted in [1] would be significantly
slower.
But as it turns out we can use this neat trick of only doing a "mkdir
-p" if the $(wildcard) macro tells us the path doesn't exist. A re-run
of a performance test similar to that noted downthread of [1] in [2]
shows that this is faster, in addition to being less verbose and more
reliable (this uses my "git-hyperfine" thin wrapper for "hyperfine"[3]):
$ git -c hyperfine.hook.setup= hyperfine -L rev HEAD~1,HEAD~0 -s 'make -C Documentation lint-docs' -p 'rm -rf Documentation/.build' 'make -C Documentation -j1 lint-docs'
Benchmark 1: make -C Documentation -j1 lint-docs' in 'HEAD~1
Time (mean ± σ): 2.914 s ± 0.062 s [User: 2.449 s, System: 0.489 s]
Range (min … max): 2.834 s … 3.020 s 10 runs
Benchmark 2: make -C Documentation -j1 lint-docs' in 'HEAD~0
Time (mean ± σ): 2.315 s ± 0.062 s [User: 1.950 s, System: 0.386 s]
Range (min … max): 2.229 s … 2.397 s 10 runs
Summary
'make -C Documentation -j1 lint-docs' in 'HEAD~0' ran
1.26 ± 0.04 times faster than 'make -C Documentation -j1 lint-docs' in 'HEAD~1'
So let's use that pattern both for the "lint-docs" target, and a few
miscellaneous other targets.
This method of creating parent directories is explicitly racy in that
we don't know if we're going to say always create a "foo" followed by
a "foo/bar" under parallelism, or skip the "foo" because we created
"foo/bar" first. In this case it doesn't matter for anything except
that we aren't guaranteed to get the same number of rules firing when
running make in parallel.
1. https://lore.kernel.org/git/211028.861r45y3pt.gmgdl@evledraar.gmail.com/
2. https://lore.kernel.org/git/211028.86o879vvtp.gmgdl@evledraar.gmail.com/
3. https://gitlab.com/avar/git-hyperfine/
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The $(QUIET) variables we define are largely duplicated between our
various Makefiles, let's define them in the new "shared.mak" instead.
Since we're not using the environment to pass these around we don't
need to export the "QUIET_GEN" and "QUIET_BUILT_IN" variables
anymore. The "QUIET_GEN" variable is used in "git-gui/Makefile" and
"gitweb/Makefile", but they've got their own definition for those. The
"QUIET_BUILT_IN" variable is only used in the top-level "Makefile". We
still need to export the "V" variable.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move these variables over to the shared.mak, we'll make use of them in
a subsequent commit.
Note that there's reason for these to be "simply expanded variables",
i.e. to use ":=" assignments instead of lazily expanded "="
assignments. We could use "=", but let's leave this as-is for now for
ease of review.
See 425ca6710b (Makefile: allow combining UBSan with other
sanitizers, 2017-07-15) for the commit that introduced these.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This was added in 30248886ce (Makefile: disable default implicit
rules, 2010-01-26), let's move it to the top of "shared.mak" so it'll
apply to all our Makefiles.
This doesn't benefit the main Makefile at all, since it already had
the rule, but since we're including shared.mak in other Makefiles
starts to benefit them. E.g. running the 'man" target is now faster:
$ git -c hyperfine.hook.setup= hyperfine -L rev HEAD~1,HEAD~0 -s 'make -C Documentation man' 'make -C Documentation -j1 man'
Benchmark 1: make -C Documentation -j1 man' in 'HEAD~1
Time (mean ± σ): 121.7 ms ± 8.8 ms [User: 105.8 ms, System: 18.6 ms]
Range (min … max): 112.8 ms … 148.4 ms 26 runs
Benchmark 2: make -C Documentation -j1 man' in 'HEAD~0
Time (mean ± σ): 97.5 ms ± 8.0 ms [User: 80.1 ms, System: 20.1 ms]
Range (min … max): 89.8 ms … 111.8 ms 32 runs
Summary
'make -C Documentation -j1 man' in 'HEAD~0' ran
1.25 ± 0.14 times faster than 'make -C Documentation -j1 man' in 'HEAD~1'
The reason for that can be seen when comparing that run with
"--debug=a". Without this change making a target like "git-status.1"
will cause "make" to consider not only "git-status.txt", but
"git-status.txt.o", as well as numerous other implicit suffixes such
as ".c", ".cc", ".cpp" etc. See [1] for a more detailed before/after
example.
So this is causing us to omit a bunch of work we didn't need to
do. For making "git-status.1" the "--debug=a" output is reduced from
~140k lines to ~6k.
1. https://lore.kernel.org/git/220222.86bkyz875k.gmgdl@evledraar.gmail.com/
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Combine the definitions of $(FIND_SOURCE_FILES) and $(LIB_H) to speed
up the Makefile, as these are the two main expensive $(shell) commands
that we execute unconditionally.
When see what was in $(FOUND_SOURCE_FILES) that wasn't in $(LIB_H) via
the ad-hoc test of:
$(error $(filter-out $(LIB_H),$(filter %.h,$(ALL_SOURCE_FILES))))
$(error $(filter-out $(ALL_SOURCE_FILES),$(filter %.h,$(LIB_H))))
We'll get, respectively:
Makefile:850: *** t/helper/test-tool.h. Stop.
Makefile:850: *** . Stop.
I.e. we only had a discrepancy when it came to
t/helper/test-tool.h. In terms of correctness this was broken before,
but now works:
$ make t/helper/test-tool.hco
HDR t/helper/test-tool.h
This speeds things up a lot:
$ git -c hyperfine.hook.setup= hyperfine -L rev HEAD~1,HEAD~0 -s 'make NO_TCLTK=Y' 'make -j1 NO_TCLTK=Y' --warmup 10 -M 10
Benchmark 1: make -j1 NO_TCLTK=Y' in 'HEAD~1
Time (mean ± σ): 159.9 ms ± 6.8 ms [User: 137.2 ms, System: 28.0 ms]
Range (min … max): 154.6 ms … 175.9 ms 10 runs
Benchmark 2: make -j1 NO_TCLTK=Y' in 'HEAD~0
Time (mean ± σ): 100.0 ms ± 1.3 ms [User: 84.2 ms, System: 20.2 ms]
Range (min … max): 98.8 ms … 102.8 ms 10 runs
Summary
'make -j1 NO_TCLTK=Y' in 'HEAD~0' ran
1.60 ± 0.07 times faster than 'make -j1 NO_TCLTK=Y' in 'HEAD~1'
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Override built-in rules of GNU make that use a wildcard target. This
can speeds things up significantly as we don't need to stat() so many
files. GNU make does that by default to see if it can retrieve their
contents from RCS or SCCS. See [1] for an old mailing list discussion
about how to disable these.
The speed-up may vary. I've seen 1-10% depending on the speed of the
local disk, caches, -jN etc. Running:
strace -f -c -S calls make -j1 NO_TCLTK=Y
Shows that we reduce the number of syscalls we make, mostly in "stat"
calls.
We could also invoke make with "-r" by setting "MAKEFLAGS = -r"
early. Doing so might make us a bit faster still. But doing so is a
much bigger hammer, since it will disable all built-in rules,
some (all?) of which can be seen with:
make -f/dev/null -p | grep -v -e ^# -e ^$
We may have something that relies on them, so let's go for the more
isolated optimization here that gives us most or all of the wins.
1. https://lists.gnu.org/archive/html/help-make/2002-11/msg00063.html
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We have various behavior that's shared across our Makefiles, or that
really should be (e.g. via defined templates). Let's create a
top-level "shared.mak" to house those sorts of things, and start by
adding the ".DELETE_ON_ERROR" flag to it.
See my own 7b76d6bf22 (Makefile: add and use the ".DELETE_ON_ERROR"
flag, 2021-06-29) and db10fc6c09 (doc: simplify Makefile using
.DELETE_ON_ERROR, 2021-05-21) for the addition and use of the
".DELETE_ON_ERROR" flag.
I.e. this changes the behavior of existing rules in the altered
Makefiles (except "Makefile" & "Documentation/Makefile"). I'm
confident that this is safe having read the relevant rules in those
Makfiles, and as the GNU make manual notes that it isn't the default
behavior is out of an abundance of backwards compatibility
caution. From edition 0.75 of its manual, covering GNU make 4.3:
[Enabling '.DELETE_ON_ERROR' is] almost always what you want
'make' to do, but it is not historical practice; so for
compatibility, you must explicitly request it.
This doesn't introduce a bug by e.g. having this
".DELETE_ON_ERROR" flag only apply to this new shared.mak, Makefiles
have no such scoping semantics.
It does increase the danger that any Makefile without an explicit "The
default target of this Makefile is..." snippet to define the default
target as "all" could have its default rule changed if our new
shared.mak ever defines a "real" rule. In subsequent commits we'll be
careful not to do that, and such breakage would be obvious e.g. in the
case of "make -C t".
We might want to make that less fragile still (e.g. by using
".DEFAULT_GOAL" as noted in the preceding commit), but for now let's
simply include "shared.mak" without adding that boilerplate to all the
Makefiles that don't have it already. Most of those are already
exposed to that potential caveat e.g. due to including "config.mak*".
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make the "contrib/scalar/Makefile" be stylistically consistent with
the top-level "Makefile" in first declaring "all" to be the default
rule, followed by including other Makefile snippets.
This adjusts code added in 0a43fb2202 (scalar: create a rudimentary
executable, 2021-12-03), it further ensures that when we add another
"include" file in a subsequent commit that the included file won't be
the one to define our default target.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Now that cmd_reflog_delete has been libified an exported it into a new
reflog.c library so we can call it directly from builtin/stash.c. This
not only gives us a performance gain since we don't need to create a
subprocess, but it also allows us to use the ref transactions api in the
future.
Helped-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: John Cai <johncai86@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
