"cd sub/dir && git commit ../path" ought to record the changes to
the file "sub/path", but this regressed long time ago.
* bw/commit-partial-from-subdirectory-fix:
commit: allow partial commits with relative paths
Commit 8894d53580 (commit: allow partial commits with relative paths,
2011-07-30) ensured that partial commits were allowed when a user
supplies a relative pathspec but then this was regressed in 5879f5684c
(remove prefix argument from pathspec_prefix, 2011-09-04) when the
prefix argument to 'pathspec_prefix' removed and the 'list_paths'
function wasn't properly adjusted to cope with the change, resulting in
over-eager pruning of the tree that is overlayed on the index.
This fixes the regression and adds a regression test so this can be
prevented in the future.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
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
"git commit" used to run "gc --auto" near the end, which was lost
when the command was reimplemented in C by mistake.
* ab/gc-auto-in-commit:
commit: run git gc --auto just before the post-commit hook
"git status" can spend a lot of cycles to compute the relation
between the current branch and its upstream, which can now be
disabled with "--no-ahead-behind" option.
* jh/status-no-ahead-behind:
status: support --no-ahead-behind in long format
status: update short status to respect --no-ahead-behind
status: add --[no-]ahead-behind to status and commit for V2 format.
stat_tracking_info: return +1 when branches not equal
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>
Change the behavior of git-commit back to what it was back in
d4bb43ee27 ("Invoke "git gc --auto" from commit, merge, am and
rebase.", 2007-09-05) when it was git-commit.sh.
Shortly afterwards in f5bbc3225c ("Port git commit to C.", 2007-11-08)
when it was ported to C, the "git gc --auto" invocation went away.
Since that unintended regression, git gc --auto only ran for git-am,
git-merge, git-fetch, and git-receive-pack. It was possible to
write a script that would "git commit" a lot of data locally, and gc
would never run.
One such repository that was locally committing generated zone file
changes had grown to a size of ~60GB before a daily cronjob was added
to "git gc", bringing it down to less than 1GB. This will make such
cases work without intervention.
I think fixing such pathological cases where the repository will grow
forever is a worthwhile trade-off for spending a couple of
milliseconds calling "git gc --auto" (in the common cases where it
doesn't do anything).
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git commit --fixup" did not allow "-m<message>" option to be used
at the same time; allow it to annotate resulting commit with more
text.
* ab/commit-m-with-fixup:
commit: add support for --fixup <commit> -m"<extra message>"
commit doc: document that -c, -C, -F and --fixup with -m error
"git status" after moving a path in the working tree (hence making
it appear "removed") and then adding with the -N option (hence
making that appear "added") detected it as a rename, but did not
report the old and new pathnames correctly.
* nd/ita-wt-renames-in-status:
wt-status.c: handle worktree renames
wt-status.c: rename rename-related fields in wt_status_change_data
wt-status.c: catch unhandled diff status codes
wt-status.c: coding style fix
Use DIFF_DETECT_RENAME for detect_rename assignments
t2203: test status output with porcelain v2 format
The sequencer infrastructure is shared across "git cherry-pick",
"git rebase -i", etc., and has always spawned "git commit" when it
needs to create a commit. It has been taught to do so internally,
when able, by reusing the codepath "git commit" itself uses, which
gives performance boost for a few tens of percents in some sample
scenarios.
* pw/sequencer-in-process-commit:
sequencer: run 'prepare-commit-msg' hook
t7505: add tests for cherry-pick and rebase -i/-p
t7505: style fixes
sequencer: assign only free()able strings to gpg_sign
sequencer: improve config handling
t3512/t3513: remove KNOWN_FAILURE_CHERRY_PICK_SEES_EMPTY_COMMIT=1
sequencer: try to commit without forking 'git commit'
sequencer: load commit related config
sequencer: simplify adding Signed-off-by: trailer
commit: move print_commit_summary() to libgit
commit: move post-rewrite code to libgit
Add a function to update HEAD after creating a commit
commit: move empty message checks to libgit
t3404: check intermediate squash messages
Convert the definitions and declarations of commit_tree and
commit_tree_extended to use struct object_id and adjust all usages of
these functions.
Signed-off-by: Patryk Obara <patryk.obara@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Teach "git status" and "git commit" to accept "--no-ahead-behind"
and "--ahead-behind" arguments to request quick or full ahead/behind
reporting.
When "--no-ahead-behind" is given, the existing porcelain V2 line
"branch.ab +x -y" is replaced with a new "branch.ab +? -?" line.
This indicates that the branch and its upstream are or are not equal
without the expense of computing the full ahead/behind values.
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit 356ee4659b ("sequencer: try to commit without forking 'git
commit'", 2017-11-24) forgot to run the 'prepare-commit-msg' hook when
creating the commit. Fix this by writing the commit message to a
different file and running the hook. Using a different file means that
if the commit is cancelled the original message file is
unchanged. Also move the checks for an empty commit so the order
matches 'git commit'.
Reported-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Helped-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git commit --fixup" did not allow "-m<message>" option to be used
at the same time; allow it to annotate resulting commit with more
text.
* ab/commit-m-with-fixup:
commit: add support for --fixup <commit> -m"<extra message>"
commit doc: document that -c, -C, -F and --fixup with -m error
"git status" after moving a path in the working tree (hence making
it appear "removed") and then adding with the -N option (hence
making that appear "added") detected it as a rename, but did not
report the old and new pathnames correctly.
* nd/ita-wt-renames-in-status:
wt-status.c: handle worktree renames
wt-status.c: rename rename-related fields in wt_status_change_data
wt-status.c: catch unhandled diff status codes
wt-status.c: coding style fix
Use DIFF_DETECT_RENAME for detect_rename assignments
t2203: test status output with porcelain v2 format
This field can have two values (2 for copy). Use this name instead for
clarity. Many places have already used this constant.
Note, the detect_rename assignments in merge-recursive.c remain
unchanged because it's actually a boolean there.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add support for supplying the -m option with --fixup. Doing so has
errored out ever since --fixup was introduced. Before this, the only
way to amend the fixup message while committing was to use --edit and
amend it in the editor.
The use-case for this feature is one of:
* Leaving a quick note to self when creating a --fixup commit when
it's not self-evident why the commit should be squashed without a
note into another one.
* (Ab)using the --fixup feature to "fix up" commits that have already
been pushed to a branch that doesn't allow non-fast-forwards,
i.e. just noting "this should have been part of that other commit",
and if the history ever got rewritten in the future the two should
be combined.
In such a case you might want to leave a small message,
e.g. "forgot this part, which broke XYZ".
With this, --fixup <commit> -m"More" -m"Details" will result in a
commit message like:
!fixup <subject of <commit>>
More
Details
The reason the test being added here seems to squash "More" at the end
of the subject line of the commit being fixed up is because the test
code is using "%s%b" so the body immediately follows the subject, it's
not a bug in this code, and other tests t7500-commit.sh do the same
thing.
When the --fixup option was initially added the "Option -m cannot be
combined" error was expanded from -c, -C and -F to also include
--fixup[1]
Those options could also support combining with -m, but given what
they do I can't think of a good use-case for doing that, so I have not
made the more invasive change of splitting up the logic in commit.c to
first act on those, and then on -m options.
1. d71b8ba7c9 ("commit: --fixup option for use with rebase
--autosquash", 2010-11-02)
Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move print_commit_summary() from builtin/commit.c to sequencer.c so it
can be shared with other commands. The function is modified by
changing the last argument to a flag so callers can specify whether
they want to show the author date in addition to specifying if this is
an initial commit.
If the sequencer dies in print_commit_summary() (which can only happen
when cherry-picking or reverting) then neither the todo list nor the
abort safety file are updated to reflect the commit that was just
made. print_commit_summary() can die if:
- The commit that was just created cannot be found or parsed.
- HEAD cannot be resolved either because some other process is
updating it (which is bad news in the middle of a cherry-pick) or
because it is corrupt.
- log_tree_commit() cannot read some objects.
In all those cases dying will leave the sequencer in a sane state for
aborting; 'git cherry-pick --abort' will rewind HEAD to the last
successful commit before there was a problem with HEAD or the object
database. If the user somehow fixes the problem and runs 'git
cherry-pick --continue' then the sequencer will try and pick the same
commit again which may or may not be what the user wants depending on
what caused print_commit_summary() to die. If print_commit_summary()
returned an error instead then update_abort_safety_file() would try to
resolve HEAD which may or may not be successful. If it is successful
then running 'git rebase --abort' would not rewind HEAD to the last
successful commit which is not what we want.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move run_rewrite_hook() from bulitin/commit.c to sequencer.c so it can
be shared with other commands and add a new function
commit_post_rewrite() based on the code in builtin/commit.c that
encapsulates rewriting notes and running the post-rewrite hook. Once
the sequencer learns how to create commits without forking 'git
commit' these functions will be used when squashing commits.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add update_head_with_reflog() based on the code that updates HEAD
after committing in builtin/commit.c that can be called by 'git
commit' and other commands.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The set of paths output from "git status --ignored" was tied
closely with its "--untracked=<mode>" option, but now it can be
controlled more flexibly. Most notably, a directory that is
ignored because it is listed to be ignored in the ignore/exclude
mechanism can be handled differently from a directory that ends up
to be ignored only because all files in it are ignored.
* jm/status-ignored-files-list:
status: test ignored modes
status: document options to show matching ignored files
status: report matching ignored and normal untracked
status: add option to show ignored files differently
Move the functions that check for empty messages from bulitin/commit.c
to sequencer.c so they can be shared with other commands. The
functions are refactored to take an explicit cleanup mode and template
filename passed by the caller.
Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A single-word "unsigned flags" in the diff options is being split
into a structure with many bitfields.
* bw/diff-opt-impl-to-bitfields:
diff: make struct diff_flags members lowercase
diff: remove DIFF_OPT_CLR macro
diff: remove DIFF_OPT_SET macro
diff: remove DIFF_OPT_TST macro
diff: remove touched flags
diff: add flag to indicate textconv was set via cmdline
diff: convert flags to be stored in bitfields
add, reset: use DIFF_OPT_SET macro to set a diff flag
We currently have seven callers of `reduce_heads(foo)`. Six of them do
not use the original list `foo` again, and actually, all six of those
end up leaking it.
Introduce and use `reduce_heads_replace(&foo)` as a leak-free version of
`foo = reduce_heads(foo)` to fix several of these. Fix the remaining
leaks using `free_commit_list()`.
While we're here, document `reduce_heads()` and mark it as `extern`.
Signed-off-by: Martin Ågren <martin.agren@gmail.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
...
"git commit", after making a commit, did not check for errors when
asking on what branch it made the commit, which has been correted.
* ao/check-resolve-ref-unsafe-result:
commit: check result of resolve_ref_unsafe
An earlier update made it possible to use an on-stack in-core
lockfile structure (as opposed to having to deliberately leak an
on-heap one). Many codepaths have been updated to take advantage
of this new facility.
* ma/lockfile-fixes:
read_cache: roll back lock in `update_index_if_able()`
read-cache: leave lock in right state in `write_locked_index()`
read-cache: drop explicit `CLOSE_LOCK`-flag
cache.h: document `write_locked_index()`
apply: remove `newfd` from `struct apply_state`
apply: move lockfile into `apply_state`
cache-tree: simplify locking logic
checkout-index: simplify locking logic
tempfile: fix documentation on `delete_tempfile()`
lockfile: fix documentation on `close_lock_file_gently()`
treewide: prefer lockfiles on the stack
sha1_file: do not leak `lock_file`
We cannot add many more flags to the diff machinery due to the
limitations of the number of flags that can be stored in a single
unsigned int. In order to allow for more flags to be added to the diff
machinery in the future this patch converts the flags to be stored in
bitfields in 'struct diff_flags'.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Teach the status command more flexibility in how ignored files are
reported. Currently, the reporting of ignored files and untracked
files are linked. You cannot control how ignored files are reported
independently of how untracked files are reported (i.e. `all` vs
`normal`). This makes it impossible to show untracked files with the
`all` option, but show ignored files with the `normal` option.
This work 1) adds the ability to control the reporting of ignored
files independently of untracked files and 2) introduces the concept
of status reporting ignored paths that explicitly match an ignored
pattern. There are 2 benefits to these changes: 1) if a consumer needs
all untracked files but not all ignored files, there is a performance
benefit to not scanning all contents of an ignored directory and 2)
returning ignored files that explicitly match a path allow a consumer
to make more informed decisions about when a status result might be
stale.
This commit implements --ignored=matching with --untracked-files=all.
The following commit will implement --ignored=matching with
--untracked=files=normal.
As an example of where this flexibility could be useful is that our
application (Visual Studio) runs the status command and presents the
output. It shows all untracked files individually (e.g. using the
'--untracked-files==all' option), and would like to know about which
paths are ignored. It uses information about ignored paths to make
decisions about when the status result might have changed.
Additionally, many projects place build output into directories inside
a repository's working directory (e.g. in "bin/" and "obj/"
directories). Normal usage is to explicitly ignore these 2 directory
names in the .gitignore file (rather than or in addition to the *.obj
pattern).If an application could know that these directories are
explicitly ignored, it could infer that all contents are ignored as
well and make better informed decisions about files in these
directories. It could infer that any changes under these paths would
not affect the output of status. Additionally, there can be a
significant performance benefit by avoiding scanning through ignored
directories.
When status is set to report matching ignored files, it has the
following behavior. Ignored files and directories that explicitly
match an exclude pattern are reported. If an ignored directory matches
an exclude pattern, then the path of the directory is returned. If a
directory does not match an exclude pattern, but all of its contents
are ignored, then the contained files are reported instead of the
directory.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add check of the resolved HEAD reference while printing of a commit summary.
resolve_ref_unsafe() may return NULL pointer if underlying calls of lstat() or
open() fail in files_read_raw_ref().
Such situation can be caused by race: file becomes inaccessible to this moment.
Signed-off-by: Andrey Okoshkin <a.okoshkin@samsung.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Update the ref transaction code to use struct object_id. Remove one
NULL pointer check which was previously inserted around a dereference;
since we now pass a pointer to struct object_id directly through, the
code we're calling handles this for us.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
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>
Some commands (most notably "git status") makes an opportunistic
update when performing a read-only operation to help optimize later
operations in the same repository. The new "--no-optional-locks"
option can be passed to Git to disable them.
* jk/no-optional-locks:
git: add --no-optional-locks option
Memory leaks in various codepaths have been plugged.
* ma/leakplugs:
pack-bitmap[-write]: use `object_array_clear()`, don't leak
object_array: add and use `object_array_pop()`
object_array: use `object_array_clear()`, not `free()`
leak_pending: use `object_array_clear()`, not `free()`
commit: fix memory leak in `reduce_heads()`
builtin/commit: fix memory leak in `prepare_index()`
Some tools like IDEs or fancy editors may periodically run
commands like "git status" in the background to keep track
of the state of the repository. Some of these commands may
refresh the index and write out the result in an
opportunistic way: if they can get the index lock, then they
update the on-disk index with any updates they find. And if
not, then their in-core refresh is lost and just has to be
recomputed by the next caller.
But taking the index lock may conflict with other operations
in the repository. Especially ones that the user is doing
themselves, which _aren't_ opportunistic. In other words,
"git status" knows how to back off when somebody else is
holding the lock, but other commands don't know that status
would be happy to drop the lock if somebody else wanted it.
There are a couple possible solutions:
1. Have some kind of "pseudo-lock" that allows other
commands to tell status that they want the lock.
This is likely to be complicated and error-prone to
implement (and maybe even impossible with just
dotlocks to work from, as it requires some
inter-process communication).
2. Avoid background runs of commands like "git status"
that want to do opportunistic updates, preferring
instead plumbing like diff-files, etc.
This is awkward for a couple of reasons. One is that
"status --porcelain" reports a lot more about the
repository state than is available from individual
plumbing commands. And two is that we actually _do_
want to see the refreshed index. We just don't want to
take a lock or write out the result. Whereas commands
like diff-files expect us to refresh the index
separately and write it to disk so that they can depend
on the result. But that write is exactly what we're
trying to avoid.
3. Ask "status" not to lock or write the index.
This is easy to implement. The big downside is that any
work done in refreshing the index for such a call is
lost when the process exits. So a background process
may end up re-hashing a changed file multiple times
until the user runs a command that does an index
refresh themselves.
This patch implements the option 3. The idea (and the test)
is largely stolen from a Git for Windows patch by Johannes
Schindelin, 67e5ce7f63 (status: offer *not* to lock the
index and update it, 2016-08-12). The twist here is that
instead of making this an option to "git status", it becomes
a "git" option and matching environment variable.
The reason there is two-fold:
1. An environment variable is carried through to
sub-processes. And whether an invocation is a
background process or not should apply to the whole
process tree. So you could do "git --no-optional-locks
foo", and if "foo" is a script or alias that calls
"status", you'll still get the effect.
2. There may be other programs that want the same
treatment.
I've punted here on finding more callers to convert,
since "status" is the obvious one to call as a repeated
background job. But "git diff"'s opportunistic refresh
of the index may be a good candidate.
The test is taken from 67e5ce7f63, and it's worth repeating
Johannes's explanation:
Note that the regression test added in this commit does
not *really* verify that no index.lock file was written;
that test is not possible in a portable way. Instead, we
verify that .git/index is rewritten *only* when `git
status` is run without `--no-optional-locks`.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This allows us to get rid of some write-only variables, among them seven
SHA1 buffers.
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Release `pathspec` and the string list `partial`.
When we clear the string list, make sure we do not free the `util`
pointers. That would result in double-freeing, since we set them up as
`item->util = item` in `list_paths()`.
Initialize the string list early, so that we can always release it. That
introduces some unnecessary overhead in various code paths, but means
there is one and only one way out of the function. If we ever accumulate
more things we need to free, it should be straightforward to do so.
Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It's a common pattern in git commands to allocate some
memory that should last for the lifetime of the program and
then not bother to free it, relying on the OS to throw it
away.
This keeps the code simple, and it's fast (we don't waste
time traversing structures or calling free at the end of the
program). But it also triggers warnings from memory-leak
checkers like valgrind or LSAN. They know that the memory
was still allocated at program exit, but they don't know
_when_ the leaked memory stopped being useful. If it was
early in the program, then it's probably a real and
important leak. But if it was used right up until program
exit, it's not an interesting leak and we'd like to suppress
it so that we can see the real leaks.
This patch introduces an UNLEAK() macro that lets us do so.
To understand its design, let's first look at some of the
alternatives.
Unfortunately the suppression systems offered by
leak-checking tools don't quite do what we want. A
leak-checker basically knows two things:
1. Which blocks were allocated via malloc, and the
callstack during the allocation.
2. Which blocks were left un-freed at the end of the
program (and which are unreachable, but more on that
later).
Their suppressions work by mentioning the function or
callstack of a particular allocation, and marking it as OK
to leak. So imagine you have code like this:
int cmd_foo(...)
{
/* this allocates some memory */
char *p = some_function();
printf("%s", p);
return 0;
}
You can say "ignore allocations from some_function(),
they're not leaks". But that's not right. That function may
be called elsewhere, too, and we would potentially want to
know about those leaks.
So you can say "ignore the callstack when main calls
some_function". That works, but your annotations are
brittle. In this case it's only two functions, but you can
imagine that the actual allocation is much deeper. If any of
the intermediate code changes, you have to update the
suppression.
What we _really_ want to say is that "the value assigned to
p at the end of the function is not a real leak". But
leak-checkers can't understand that; they don't know about
"p" in the first place.
However, we can do something a little bit tricky if we make
some assumptions about how leak-checkers work. They
generally don't just report all un-freed blocks. That would
report even globals which are still accessible when the
leak-check is run. Instead they take some set of memory
(like BSS) as a root and mark it as "reachable". Then they
scan the reachable blocks for anything that looks like a
pointer to a malloc'd block, and consider that block
reachable. And then they scan those blocks, and so on,
transitively marking anything reachable from a global as
"not leaked" (or at least leaked in a different category).
So we can mark the value of "p" as reachable by putting it
into a variable with program lifetime. One way to do that is
to just mark "p" as static. But that actually affects the
run-time behavior if the function is called twice (you
aren't likely to call main() twice, but some of our cmd_*()
functions are called from other commands).
Instead, we can trick the leak-checker by putting the value
into _any_ reachable bytes. This patch keeps a global
linked-list of bytes copied from "unleaked" variables. That
list is reachable even at program exit, which confers
recursive reachability on whatever values we unleak.
In other words, you can do:
int cmd_foo(...)
{
char *p = some_function();
printf("%s", p);
UNLEAK(p);
return 0;
}
to annotate "p" and suppress the leak report.
But wait, couldn't we just say "free(p)"? In this toy
example, yes. But UNLEAK()'s byte-copying strategy has
several advantages over actually freeing the memory:
1. It's recursive across structures. In many cases our "p"
is not just a pointer, but a complex struct whose
fields may have been allocated by a sub-function. And
in some cases (e.g., dir_struct) we don't even have a
function which knows how to free all of the struct
members.
By marking the struct itself as reachable, that confers
reachability on any pointers it contains (including those
found in embedded structs, or reachable by walking
heap blocks recursively.
2. It works on cases where we're not sure if the value is
allocated or not. For example:
char *p = argc > 1 ? argv[1] : some_function();
It's safe to use UNLEAK(p) here, because it's not
freeing any memory. In the case that we're pointing to
argv here, the reachability checker will just ignore
our bytes.
3. Likewise, it works even if the variable has _already_
been freed. We're just copying the pointer bytes. If
the block has been freed, the leak-checker will skip
over those bytes as uninteresting.
4. Because it's not actually freeing memory, you can
UNLEAK() before we are finished accessing the variable.
This is helpful in cases like this:
char *p = some_function();
return another_function(p);
Writing this with free() requires:
int ret;
char *p = some_function();
ret = another_function(p);
free(p);
return ret;
But with unleak we can just write:
char *p = some_function();
UNLEAK(p);
return another_function(p);
This patch adds the UNLEAK() macro and enables it
automatically when Git is compiled with SANITIZE=leak. In
normal builds it's a noop, so we pay no runtime cost.
It also adds some UNLEAK() annotations to show off how the
feature works. On top of other recent leak fixes, these are
enough to get t0000 and t0001 to pass when compiled with
LSAN.
Note the case in commit.c which actually converts a
strbuf_release() into an UNLEAK. This code was already
non-leaky, but the free didn't do anything useful, since
we're exiting. Converting it to an annotation means that
non-leak-checking builds pay no runtime cost. The cost is
minimal enough that it's probably not worth going on a
crusade to convert these kinds of frees to UNLEAKS. I did it
here for consistency with the "sb" leak (though it would
have been equally correct to go the other way, and turn them
both into strbuf_release() calls).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Code clean-up to avoid mixing values read from the .gitmodules file
and values read from the .git/config file.
* bw/submodule-config-cleanup:
submodule: remove gitmodules_config
unpack-trees: improve loading of .gitmodules
submodule-config: lazy-load a repository's .gitmodules file
submodule-config: move submodule-config functions to submodule-config.c
submodule-config: remove support for overlaying repository config
diff: stop allowing diff to have submodules configured in .git/config
submodule: remove submodule_config callback routine
unpack-trees: don't respect submodule.update
submodule: don't rely on overlayed config when setting diffopts
fetch: don't overlay config with submodule-config
submodule--helper: don't overlay config in update-clone
submodule--helper: don't overlay config in remote_submodule_branch
add, reset: ensure submodules can be added or reset
submodule: don't use submodule_from_name
t7411: check configuration parsing errors
"git commit" when seeing an totally empty message said "you did not
edit the message", which is clearly wrong. The message has been
corrected.
* ks/commit-abort-on-empty-message-fix:
commit: check for empty message before the check for untouched template
"git commit" used to discard the index and re-read from the filesystem
just in case the pre-commit hook has updated it in the middle; this
has been optimized out when we know we do not run the pre-commit hook.
* kw/commit-keep-index-when-pre-commit-is-not-run:
commit: skip discarding the index if there is no pre-commit hook
If there is not a pre-commit hook, there is no reason to discard
the index and reread it.
This change checks to presence of a pre-commit hook and then only
discards the index if there was one.
Signed-off-by: Kevin Willford <kewillf@microsoft.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git commit" when seeing an totally empty message said "you did not
edit the message", which is clearly wrong. The message has been
corrected.
* ks/commit-abort-on-empty-message-fix:
commit: check for empty message before the check for untouched template
Now that the submodule-config subsystem can lazily read the gitmodules
file we no longer need to explicitly pre-read the gitmodules by calling
'gitmodules_config()' so let's remove it.
Signed-off-by: Brandon Williams <bmwill@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The check for whether the template given to 'git commit' is untouched
is done before the empty message check. This results in a wrong error
message being displayed in the following case. When the user removes
everything in template completely to abort the commit he is shown the
"template untouched" error which is wrong. He should be shown the
"empty message" error.
Do the empty message check before checking for an untouched template
thus fixing this issue.
Signed-off-by: Kaartic Sivaraam <kaarticsivaraam91196@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
