Optimize the code to find shortest unique prefix of object names.
* ds/find-unique-abbrev-optim:
sha1_name: minimize OID comparisons during disambiguation
sha1_name: parse less while finding common prefix
sha1_name: unroll len loop in find_unique_abbrev_r()
p4211-line-log.sh: add log --online --raw --parents perf test
* pc/submodule-helper:
submodule: port submodule subcommand 'status' from shell to C
submodule--helper: introduce for_each_listed_submodule()
submodule--helper: introduce get_submodule_displaypath()
An early part of piece-by-piece rewrite of "git bisect".
* pb/bisect-helper:
bisect--helper: `is_expected_rev` & `check_expected_revs` shell function in C
t6030: explicitly test for bisection cleanup
bisect--helper: `bisect_clean_state` shell function in C
bisect--helper: `write_terms` shell function in C
bisect--helper: rewrite `check_term_format` shell function in C
bisect--helper: use OPT_CMDMODE instead of OPT_BOOL
A hook script that is set unexecutable is simply ignored. Git
notifies when such a file is ignored, unless the message is
squelched via advice.ignoredHook configuration.
* dm/run-command-ignored-hook-advise:
run-command: add hint when a hook is ignored
The final step to make an empty string as a pathspec element
illegal. We started this by first deprecating and warning a
pathspec that has such an element in 2.11 (Nov 2016).
Hopefully we can merge this down to the 'master' by the end of the
year? A deprecation warning period that is about 1 year does not
sound too bad.
* ex/deprecate-empty-pathspec-as-match-all:
pathspec: die on empty strings as pathspec
t0027: do not use an empty string as a pathspec element
"git status --ignored -u" did not stop at a working tree of a
separate project that is embedded in an ignored directory and
listed files in that other project, instead of just showing the
directory itself as ignored.
* js/submodule-in-excluded:
status: do not get confused by submodules in excluded directories
Instead of using custom line comparison and hashing functions to
implement "moved lines" coloring in the diff output, use the pair
of these functions from lower-layer xdiff/ code.
* sb/diff-color-moved-use-xdl-recmatch:
diff.c: get rid of duplicate implementation
xdiff-interface: export comparing and hashing strings
The experimental "color moved lines differently in diff output"
feature was buggy around "ignore whitespace changes" edges, whihch
has been corrected.
* jk/diff-color-moved-fix:
diff: handle NULs in get_string_hash()
diff: fix whitespace-skipping with --color-moved
t4015: test the output of "diff --color-moved -b"
t4015: check "negative" case for "-w --color-moved"
t4015: refactor --color-moved whitespace test
"auto" as a value for the columnar output configuration ought to
judge "is the output consumed by humans?" with the same criteria as
"auto" for coloured output configuration, i.e. either the standard
output stream is going to tty, or a pager is in use. We forgot the
latter, which has been fixed.
* kd/auto-col-with-pager-fix:
column: do not include pager.c
column: show auto columns when pager is active
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`
With --recurse-submodules, we add each submodule that we encounter to
the list of alternate object databases. With threading, our changes to
the list are not protected against races. Indeed, ThreadSanitizer
reports a race when we call `add_to_alternates_memory()` around the same
time that another thread is reading in the list through
`read_sha1_file()`.
Take the grep read-lock while adding the submodule. The lock is used to
serialize uses of non-thread-safe parts of Git's API, including
`read_sha1_file()`.
Helped-by: Brandon Williams <bmwill@google.com> Signed-off-by: Martin Ågren <martin.agren@gmail.com> Acked-by: Brandon Williams <bmwill@google.com> Reviewed-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we replaced the old shell script based interactive rebase in
commmit 18633e1a22a6 ("rebase -i: use the rebase--helper builtin",
2017-02-09) we introduced a regression of functionality in that the
GIT_DIR would be sent to the environment of the exec command as-is.
This generally meant that it would be passed as "GIT_DIR=.git", which
causes problems for any exec command that wants to run git commands in
a subdirectory.
This isn't a very large regression, since it is not that likely that the
exec command will run a git command, and even less likely that it will
need to do so in a subdir. This regression was discovered by a build
system which uses git-describe to find the current version of the build
system, and happened to do so from the src/ sub directory of the
project.
Fix this by passing in the absolute path of the git directory into the
child environment.
Signed-off-by: Jacob Keller <jacob.keller@gmail.com> Acked-by: Johannes Schindelin <Johannes.Schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
t0000: check whether the shell supports the "local" keyword
Add a test balloon to see if we get complaints from anybody who is
using a shell that doesn't support the "local" keyword. If so, this
test can be reverted. If not, we might want to consider using "local"
in shell code throughout the git code base.
Signed-off-by: Michael Haggerty <mhagger@alum.mit.edu> Signed-off-by: Junio C Hamano <gitster@pobox.com>
rev-list-options.txt: use correct directional reference
The descriptions of the options '--parents', '--children' and
'--graph' say "see 'History Simplification' below", although the
referred section is in fact above the description of these options.
Send readers in the right direction by saying "above" instead of
"below".
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
status: do not get confused by submodules in excluded directories
We meticulously pass the `exclude` flag to the `treat_directory()`
function so that we can indicate that files in it are excluded rather
than untracked when recursing.
But we did not yet treat submodules the same way.
Because of that, `git status --ignored --untracked` with a submodule
`submodule` in a gitignored `tracked/` would show the submodule in the
"Untracked files" section, e.g.
On branch master
Untracked files:
(use "git add <file>..." to include in what will be committed)
tracked/submodule/
Ignored files:
(use "git add -f <file>..." to include in what will be committed)
tracked/submodule/initial.t
Instead, we would want it to show the submodule in the "Ignored files"
section:
On branch master
Ignored files:
(use "git add -f <file>..." to include in what will be committed)
tracked/submodule/
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The implementations in diff.c to detect moved lines needs to compare
strings and hash strings, which is implemented in that file, as well
as in the xdiff library.
Remove the rather recent implementation in diff.c and rely on the well
exercised code in the xdiff lib.
With this change the hash used for bucketing the strings for the moved
line detection changes from FNV32 (that is provided via the hashmaps
memhash) to DJB2 (which is used internally in xdiff). Benchmarks found
on the web[1] do not indicate that these hashes are different in
performance for readable strings.
xdiff-interface: export comparing and hashing strings
This will turn out to be useful in a later patch.
xdl_recmatch is exported in xdiff/xutils.h, to be used by various
xdiff/*.c files, but not outside of xdiff/. This one makes it available
to the outside, too.
While at it, add documentation.
Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
files_transaction_prepare(): fix handling of ref lock failure
Since dc39e09942 (files_ref_store: use a transaction to update packed
refs, 2017-09-08), failure to lock a reference has been handled
incorrectly by `files_transaction_prepare()`. If
`lock_ref_for_update()` fails in the lock-acquisition loop of that
function, it sets `ret` then breaks out of that loop. Prior to dc39e09942, that was OK, because the only thing following the loop was
the cleanup code. But dc39e09942 added another blurb of code between
the loop and the cleanup. That blurb sometimes resets `ret` to zero,
making the cleanup code think that the locking was successful.
Specifically, whenever
* One or more reference deletions have been processed successfully in
the lock-acquisition loop. (Processing the first such reference
causes a packed-ref transaction to be initialized.)
* Then `lock_ref_for_update()` fails for a subsequent reference. Such
a failure can happen for a number of reasons, such as the old SHA-1
not being correct, lock contention, etc. This causes a `break` out
of the lock-acquisition loop.
* The `packed-refs` lock is acquired successfully and
`ref_transaction_prepare()` succeeds for the packed-ref transaction.
This has the effect of resetting `ret` back to 0, and making the
cleanup code think that lock acquisition was successful.
In that case, any reference updates that were processed prior to
breaking out of the loop would be carried out (loose and packed), but
the reference that couldn't be locked and any subsequent references
would silently be ignored.
This can easily cause data loss if, for example, the user was trying
to push a new name for an existing branch while deleting the old name.
After the push, the branch could be left unreachable, and could even
subsequently be garbage-collected.
This problem was noticed in the context of deleting one reference and
creating another in a single transaction, when the two references D/F
conflict with each other, like
git update-ref --stdin <<EOF
delete refs/foo
create refs/foo/bar HEAD
EOF
This triggers the above bug because the deletion is processed
successfully for `refs/foo`, then the D/F conflict causes
`lock_ref_for_update()` to fail when `refs/foo/bar` is processed. In
this case the transaction *should* fail, but instead it causes
`refs/foo` to be deleted without creating `refs/foo`. This could
easily result in data loss.
The fix is simple: instead of just breaking out of the loop, jump
directly to the cleanup code. This fixes some tests in t1404 that were
added in the previous commit.
Signed-off-by: Michael Haggerty <mhagger@alum.mit.edu> Signed-off-by: Junio C Hamano <gitster@pobox.com>
It is currently not allowed, in a single transaction, to add one
reference and delete another reference if the two reference names D/F
conflict with each other (e.g., like `refs/foo/bar` and `refs/foo`).
The reason is that the code would need to take locks
$GIT_DIR/refs/foo.lock
$GIT_DIR/refs/foo/bar.lock
But the latter lock couldn't coexist with the loose reference file
$GIT_DIR/refs/foo
, because `$GIT_DIR/refs/foo` cannot be both a directory and a file at
the same time (hence the name "D/F conflict).
Add a bunch of tests that we cleanly reject such transactions.
In fact, many of the new tests currently fail. They will be fixed in
the next commit along with an explanation.
Reported-by: Jeff King <peff@peff.net> Signed-off-by: Michael Haggerty <mhagger@alum.mit.edu> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Merge branch 'jc/fetch-refspec-doc-update' into maint
"git fetch <there> <src>:<dst>" allows an object name on the <src>
side when the other side accepts such a request since Git v2.5, but
the documentation was left stale.
* jc/fetch-refspec-doc-update:
fetch doc: src side of refspec could be full SHA-1
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
The scripts to drive TravisCI has been reorganized and then an
optimization to avoid spending cycles on a branch whose tip is
tagged has been implemented.
* ls/travis-scriptify:
travis-ci: fix "skip_branch_tip_with_tag()" string comparison
travis: dedent a few scripts that are indented overly deeply
travis-ci: skip a branch build if equal tag is present
travis-ci: move Travis CI code into dedicated scripts
Merge branch 'er/fast-import-dump-refs-on-checkpoint' into maint
The checkpoint command "git fast-import" did not flush updates to
refs and marks unless at least one object was created since the
last checkpoint, which has been corrected, as these things can
happen without any new object getting created.
* er/fast-import-dump-refs-on-checkpoint:
fast-import: checkpoint: dump branches/tags/marks even if object_count==0
Merge branch 'nd/worktree-kill-parse-ref' into maint
"git branch -M a b" while on a branch that is completely unrelated
to either branch a or branch b misbehaved when multiple worktree
was in use. This has been fixed.
* nd/worktree-kill-parse-ref:
branch: fix branch renaming not updating HEADs correctly
worktree: handle broken symrefs in find_shared_symref()
The refs_resolve_ref_unsafe() function may return NULL even
with a REF_ISSYMREF flag if a symref points to a broken ref.
As a result, it's possible for find_shared_symref() to
segfault when it passes NULL to strcmp().
This is hard to trigger for most code paths. We typically
pass HEAD to the function as the symref to resolve, and
programs like "git branch" will bail much earlier if HEAD
isn't valid.
I did manage to trigger it through one very obscure
sequence:
# You have multiple notes refs which conflict.
git notes add -m base
git notes --ref refs/notes/foo add -m foo
# There's left-over cruft in NOTES_MERGE_REF that
# makes it a broken symref (in this case we point
# to a syntactically invalid ref).
echo "ref: refs/heads/master.lock" >.git/NOTES_MERGE_REF
# You try to merge the notes. We read the broken value in
# order to complain that another notes-merge is
# in-progress, but we segfault in find_shared_symref().
git notes merge refs/notes/foo
This is obviously silly and almost certainly impossible to
trigger accidentally, but it does show that the bug is
triggerable from at least one code path. In addition, it
would trigger if we saw a transient filesystem error when
resolving the pointed-to ref.
We can fix this by treating NULL the same as a non-matching
symref. Arguably we'd prefer to know if a symref points to
"refs/heads/foo", but "refs/heads/foo" is broken. But
refs_resolve_ref_unsafe() isn't capable of giving us that
information, so this is the best we can do.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The resolve_ref_unsafe() function may return NULL even with
a REF_ISSYMREF flag if a symref points to a broken ref. As a
result, it's possible for the decoration code's "is this
branch the current HEAD" check to segfault when it passes
the NULL to starts_with().
This is unlikely in practice, since we can only reach this
code if we already resolved HEAD to a matching sha1 earlier.
But it's possible if HEAD racily becomes broken, or if
there's a transient filesystem error.
We can fix this by returning early in the broken case, since
NULL could not possibly match any of our branch names.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
It's possible for resolve_ref_unsafe() to return NULL with a
REF_ISSYMREF flag if a symref points to a broken ref. In
this case, the read_remote_branches() function will segfault
passing the name to xstrdup().
This is hard to trigger in practice, since this function is
used as a callback to for_each_ref(), which will skip broken
refs in the first place (so it would have to be broken
racily, or for us to see a transient filesystem error).
If we see such a racy broken outcome let's treat it as "not
a symref". This is exactly the same thing that would happen
in the non-racy case (our function would not be called at
all, as for_each_ref would skip the broken symref).
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
It's possible for resolve_ref_unsafe() to return NULL (e.g.,
if we are reading and the ref does not exist), in which case
we'll pass NULL to printf. On glibc systems this produces
"(null)", but on others it may segfault.
The tests don't expect any such case, but if we ever did
trigger this, we would prefer to cleanly fail the test with
unexpected input rather than segfault. Let's manually
replace NULL with "(null)". The exact value doesn't matter,
as it won't match any possible ref the caller could expect
(and anyway, the exit code of the program will tell whether
"ref" is valid or not).
Signed-off-by: Jeff King <peff@peff.net> 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>
For computing moved lines, we feed the characters of each
line into a hash. When we've been asked to ignore
whitespace, then we pick each character using next_byte(),
which returns -1 on end-of-string, which it determines using
the start/end pointers we feed it.
However our check of its return value treats "0" the same as
"-1", meaning we'd quit if the string has an embedded NUL.
This is unlikely to ever come up in practice since our line
boundaries generally come from calling strlen() in the first
place.
But it was a bit surprising to me as a reader of the
next_byte() code. And it's possible that we may one day feed
this function with more exotic input, which otherwise works
with arbitrary ptr/len pairs.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The code for handling whitespace with --color-moved
represents partial strings as a pair of pointers. There are
two possible conventions for the end pointer:
1. It points to the byte right after the end of the
string.
2. It points to the final byte of the string.
But we seem to use both conventions in the code:
a. we assign the initial pointers from the NUL-terminated
string using (1)
b. we eat trailing whitespace by checking the second
pointer for isspace(), which needs (2)
c. the next_byte() function checks for end-of-string with
"if (cp > endp)", which is (2)
d. in next_byte() we skip past internal whitespace with
"while (cp < end)", which is (1)
This creates fewer bugs than you might think, because there
are some subtle interactions. Because of (a) and (c), we
always return the NUL-terminator from next_byte(). But all
of the callers of next_byte() happen to handle that
gracefully.
Because of the mismatch between (d) and (c), next_byte()
could accidentally return a whitespace character right at
endp. But because of the interaction of (a) and (b), we fail
to actually chomp trailing whitespace, meaning our endp
_always_ points to a NUL, canceling out the problem.
But that does leave (b) as a real bug: when ignoring
whitespace only at the end-of-line, we don't correctly trim
it, and fail to match up lines.
We can fix the whole thing by moving consistently to one
convention. Since convention (1) is idiomatic in our code
base, we'll pick that one.
The existing "-w" and "-b" tests continue to pass, and a new
"--ignore-space-at-eol" shows off the breakage we're fixing.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit fa5ba2c1dd (diff: fix infinite loop with
--color-moved --ignore-space-change, 2017-10-12) added a
test to make sure that "--color-moved -b" doesn't run
forever, but the test in question doesn't actually have any
moved lines in it.
Let's scrap that test and add a variant of the existing
"--color-moved -w" test, but this time we'll check that we
find the move with whitespace changes, but not arbitrary
whitespace additions.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
t4015: check "negative" case for "-w --color-moved"
We test that lines with whitespace changes are not found by
"--color-moved" by default, but are found if "-w" is added.
Let's add one more twist: a line that has non-whitespace
changes should not be marked as a pure move.
This is perhaps an obvious case for us to get right (and we
do), but as we add more whitespace tests, they will form a
pattern of "make sure this case is a move and this other
case is not".
Note that we have to add a line to our moved block, since
having a too-small block doesn't trigger the "moved"
heuristics. And we also add a line of context to ensure
that there's more context lines than moved lines (so the
diff shows us moving the lines up, rather than moving the
context down).
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
In preparation for testing several different whitespace
options, let's split out the setup and cleanup steps of the
whitespace test.
While we're here, let's also switch to using "<<-" to indent
our here-documents properly, and use q_to_tab to more
explicitly mark where we expect whitespace to appear.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Update the documentation for "git filter-branch" so that the filter
options are listed in the same order as they are applied, as
described in an earlier part of the doc.
* dg/filter-branch-filter-order-doc:
doc: list filter-branch subdirectory-filter first
"git fetch <there> <src>:<dst>" allows an object name on the <src>
side when the other side accepts such a request since Git v2.5, but
the documentation was left stale.
* jc/fetch-refspec-doc-update:
fetch doc: src side of refspec could be full SHA-1
Change the word "bla" to "section.variable"; "bla" is a placeholder
for a variable name but it wasn't clear for everyone.
While we're here, also reformat this sample command line to use
monospace instead of italics, to better match the rest of the file.
Use a space instead of a dash in "git config", as is common in the
rest of Git's documentation.
Reported-by: Robert P. J. Day <rpjday@crashcourse.ca> Signed-off-by: MOY Matthieu <matthieu.moy@univ-lyon1.fr> Signed-off-by: Daniel Bensoussan <daniel.bensoussan--bohm@etu.univ-lyon1.fr> Signed-off-by: Timothee Albertin <timothee.albertin@etu.univ-lyon1.fr> Signed-off-by: Nathan Payre <nathan.payre@etu.univ-lyon1.fr> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Merge branch 'jk/ref-filter-colors-fix' into maint
This is the "theoretically more correct" approach of simply
stepping back to the state before plumbing commands started paying
attention to "color.ui" configuration variable.
* jk/ref-filter-colors-fix:
tag: respect color.ui config
Revert "color: check color.ui in git_default_config()"
Revert "t6006: drop "always" color config tests"
Revert "color: make "always" the same as "auto" in config"
color: make "always" the same as "auto" in config
provide --color option for all ref-filter users
t3205: use --color instead of color.branch=always
t3203: drop "always" color test
t6006: drop "always" color config tests
t7502: use diff.noprefix for --verbose test
t7508: use test_terminal for color output
t3701: use test-terminal to collect color output
t4015: prefer --color to -c color.diff=always
test-terminal: set TERM=vt100