am --skip/--abort: merge HEAD/ORIG_HEAD tree into index
f8da6801 (am --skip: support skipping while on unborn branch,
2015-06-06) introduced a performance regression to "git am --skip",
where it used "read-tree" to reconstruct the index from scratch
without reusing the cached stat information.
This is a backport of the corresponding patch to the builtin am in 2.6: 3ecc704 (am --skip/--abort: merge HEAD/ORIG_HEAD tree into index,
2015-08-19).
Reportedly, it can make a huge difference on Windows, in one case a `git
rebase --skip` took 1m40s without, and 5s with, this patch.
Reported-and-suggested-by: Kim Gybels <kgybels@infogroep.be> Acked-by: Paul Tan <pyokagan@gmail.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Recent "git am" had regression when adding a Signed-off-by line
with its "-s" option by an unintended tightening of how an existing
trailer block is detected.
* jc/builtin-am-signoff-regression-fix:
am: match --signoff to the original scripted version
When executing user-specified programs, we generally always
want to use a shell, for flexibility and consistency. One
big exception is executing $GIT_SSH, which for historical
reasons must not use a shell.
Once upon a time the logic in git_connect looked like:
But over time the PROTO_SSH block has grown, and the "local"
block has shrunk so that it contains only conn->use_shell;
it's easy to miss at the end of the large block. Moreover,
PROTO_SSH now also sometimes sets use_shell, when the new
GIT_SSH_COMMAND is used.
Let's just set conn->use_shell when we're setting up the "conn"
struct, and unset it (with a comment) in the historical GIT_SSH
case. This will make the flow easier to follow.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
am: match --signoff to the original scripted version
Linus noticed that the recently reimplemented "git am -s" defines
the trailer block too rigidly, resulting in an unnecessary blank
line between the existing sign-offs and his new sign-off. An e-mail
submission sent to Linus in real life ends with mixture of sign-offs
and commentaries, e.g.
title here
message here
Signed-off-by: Original Author <original@auth.or>
[rv: tweaked frotz and nitfol] Signed-off-by: Re Viewer <rv@ew.er> Signed-off-by: Other Reviewer <other@rev.ewer>
---
patch here
Because the reimplementation reused append_signoff() helper that is
used by other codepaths, which is unaware that people intermix such
comments with their sign-offs in the trailer block, such a message
was judged to end with a non-trailer, resulting in an extra blank
line before adding a new sign-off.
The original scripted version of "git am" used a lot looser
definition, i.e. "if and only if there is no line that begins with
Signed-off-by:, add a blank line before adding a new sign-off". For
the upcoming release, stop using the append_signoff() in "git am"
and reimplement the looser definition used by the scripted version
to use only in "git am" to fix this regression in "am" while
avoiding new regressions to other users of append_signoff().
In the longer term, we should look into loosening append_signoff()
so that other codepaths that add a new sign-off behave the same way
as "git am -s", but that is a task for post-release.
Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we "switch" to another local repository to run the server
side of a fetch or push, we must clear the variables in
local_repo_env so that our local $GIT_DIR, etc, do not
pollute the upload-pack or receive-pack that is executing in
the "remote" repository.
We have never done so for ssh connections. For the most
part, nobody has noticed because ssh will not pass unknown
environment variables by default. However, it is not out of
the question for a user to configure ssh to pass along GIT_*
variables using SendEnv/AcceptEnv.
We can demonstrate the problem by using "git -c" on a local
command and seeing its impact on a remote repository. This
config ends up in $GIT_CONFIG_PARAMETERS. In the local case,
the config has no impact, but in the ssh transport, it does
(our test script has a fake ssh that passes through all
environment variables; this isn't normal, but does simulate
one possible setup).
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
pager: don't use unsafe functions in signal handlers
Since the commit a3da8821208d (pager: do wait_for_pager on signal
death), we call wait_for_pager() in the pager's signal handler. The
recent bug report revealed that this causes a deadlock in glibc at
aborting "git log" [*1*]. When this happens, git process is left
unterminated, and it can't be killed by SIGTERM but only by SIGKILL.
The problem is that wait_for_pager() function does more than waiting
for pager process's termination, but it does cleanups and printing
errors. Unfortunately, the functions that may be used in a signal
handler are very limited [*2*]. Particularly, malloc(), free() and the
variants can't be used in a signal handler because they take a mutex
internally in glibc. This was the cause of the deadlock above. Other
than the direct calls of malloc/free, many functions calling
malloc/free can't be used. strerror() is such one, either.
Also the usage of fflush() and printf() in a signal handler is bad,
although it seems working so far. In a safer side, we should avoid
them, too.
This patch tries to reduce the calls of such functions in signal
handlers. wait_for_signal() takes a flag and avoids the unsafe
calls. Also, finish_command_in_signal() is introduced for the
same reason. There the free() calls are removed, and only waits for
the children without whining at errors.
When we show "branch@{0}", we format into a fixed-size
buffer using sprintf. This can overflow if you have long
branch names. We can fix it by using a temporary strbuf.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
read_info_alternates: handle paths larger than PATH_MAX
This function assumes that the relative_base path passed
into it is no larger than PATH_MAX, and writes into a
fixed-size buffer. However, this path may not have actually
come from the filesystem; for example, add_submodule_odb
generates a path using a strbuf and passes it in. This is
hard to trigger in practice, though, because the long
submodule directory would have to exist on disk before we
would try to open its info/alternates file.
We can easily avoid the bug, though, by simply creating the
filename on the heap.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we are loading a notes tree into our internal hash
table, we also collect any files that are clearly non-notes.
We format the name of the file into a PATH_MAX buffer, but
unlike true notes (which cannot be larger than a fanned-out
sha1 hash), these tree entries can be arbitrarily long,
overflowing our buffer.
We can fix this by switching to a strbuf. It doesn't even
cost us an extra allocation, as we can simply hand ownership
of the buffer over to the non-note struct.
This is of moderate security interest, as you might fetch
notes trees from an untrusted remote. However, we do not do
so by default, so you would have to manually fetch into the
notes namespace.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
verify_absent: allow filenames longer than PATH_MAX
When unpack-trees wants to know whether a path will
overwrite anything in the working tree, we use lstat() to
see if there is anything there. But if we are going to write
"foo/bar", we can't just lstat("foo/bar"); we need to look
for leading prefixes (e.g., "foo"). So we use the lstat cache
to find the length of the leading prefix, and copy the
filename up to that length into a temporary buffer (since
the original name is const, we cannot just stick a NUL in
it).
The copy we make goes into a PATH_MAX-sized buffer, which
will overflow if the prefix is longer than PATH_MAX. How
this happens is a little tricky, since in theory PATH_MAX is
the biggest path we will have read from the filesystem. But
this can happen if:
- the compiled-in PATH_MAX does not accurately reflect
what the filesystem is capable of
- the leading prefix is not _quite_ what is on disk; it
contains the next element from the name we are checking.
So if we want to write "aaa/bbb/ccc/ddd" and "aaa/bbb"
exists, the prefix of interest is "aaa/bbb/ccc". If
"aaa/bbb" approaches PATH_MAX, then "ccc" can overflow
it.
So this can be triggered, but it's hard to do. In
particular, you cannot just "git clone" a bogus repo. The
verify_absent checks happen before unpack-trees writes
anything to the filesystem, so there are never any leading
prefixes during the initial checkout, and the bug doesn't
trigger. And by definition, these files are larger than
PATH_MAX, so writing them will fail, and clone will
complain (though it may write a partial path, which will
cause a subsequent "git checkout" to hit the bug).
We can fix it by creating the temporary path on the heap.
The extra malloc overhead is not important, as we are
already making at least one stat() call (and probably more
for the prefix discovery).
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Merge branch 'cc/trailers-corner-case-fix' into maint
The "interpret-trailers" helper mistook a multi-paragraph title of
a commit log message with a colon in it as the end of the trailer
block.
* cc/trailers-corner-case-fix:
trailer: support multiline title
trailer: retitle a test and correct an in-comment message
trailer: ignore first line of message
Merge branch 'dt/commit-preserve-base-index-upon-opportunistic-cache-tree-update' into maint
When re-priming the cache-tree opportunistically while committing
the in-core index as-is, we mistakenly invalidated the in-core
index too aggressively, causing the experimental split-index code
to unnecessarily rewrite the on-disk index file(s).
* dt/commit-preserve-base-index-upon-opportunistic-cache-tree-update:
commit: don't rewrite shared index unnecessarily
"git archive" did not use zip64 extension when creating an archive
with more than 64k entries, which nobody should need, right ;-)?
* rs/archive-zip-many:
archive-zip: support more than 65535 entries
archive-zip: use a local variable to store the creator version
t5004: test ZIP archives with many entries
Merge branch 'jk/fix-alias-pager-config-key-warnings' into maint
Because the configuration system does not allow "alias.0foo" and
"pager.0foo" as the configuration key, the user cannot use '0foo'
as a custom command name anyway, but "git 0foo" tried to look these
keys up and emitted useless warnings before saying '0foo is not a
git command'. These warning messages have been squelched.
* jk/fix-alias-pager-config-key-warnings:
config: silence warnings for command names with invalid keys
Merge branch 'cb/open-noatime-clear-errno' into maint
When trying to see that an object does not exist, a state errno
leaked from our "first try to open a packfile with O_NOATIME and
then if it fails retry without it" logic on a system that refuses
O_NOATIME. This confused us and caused us to die, saying that the
packfile is unreadable, when we should have just reported that the
object does not exist in that packfile to the caller.
* cb/open-noatime-clear-errno:
git_open_noatime: return with errno=0 on success
An off-by-one error made "git remote" to mishandle a remote with a
single letter nickname.
* mh/get-remote-group-fix:
get_remote_group(): use skip_prefix()
get_remote_group(): eliminate superfluous call to strcspn()
get_remote_group(): rename local variable "space" to "wordlen"
get_remote_group(): handle remotes with single-character names
The spec is very inconsistent about which PKT-LINE() parts
of the grammar include a LF. On top of that, the code is not
consistent, either (e.g., send-pack does not put newlines
into the ref-update commands it sends).
Let's make explicit the long-standing expectation that we
generally expect pkt-lines to end in a newline, but that
receivers should be lenient. This makes the spec consistent,
and matches what git already does (though it does not always
fulfill the SHOULD).
We do make an exception for the push-cert, where the
receiving code is currently a bit pickier. This is a
reasonable way to be, as the data needs to be byte-for-byte
compatible with what was signed. We _could_ make up some
rules about signing a canonicalized version including
newlines, but that would require a code change, and is out
of scope for this patch.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
A test was designed for "git diff-index --cached -M" but the command is
run without the "-M" option (which makes the test essentially identical
to its preceding counterpart).
Signed-off-by: Matthieu Prat <matthieuprat@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When re-priming the cache-tree opportunistically while committing
the in-core index as-is, we mistakenly invalidated the in-core
index too aggressively, causing the experimental split-index code
to unnecessarily rewrite the on-disk index file(s).
* dt/commit-preserve-base-index-upon-opportunistic-cache-tree-update:
commit: don't rewrite shared index unnecessarily
"git archive" did not use zip64 extension when creating an archive
with more than 64k entries, which nobody should need, right ;-)?
* rs/archive-zip-many:
archive-zip: support more than 65535 entries
archive-zip: use a local variable to store the creator version
t5004: test ZIP archives with many entries
There's a bug in builtin/am.c in which we take a lock on
MERGE_RR recursively. But rather than fix am.c, this patch
fixes the confusing interface from rerere.c that caused the
bug. Read on for the gory details.
The setup_rerere() function both reads the existing MERGE_RR
file, and takes MERGE_RR.lock. In the rerere() and
rerere_forget() functions, we end up in write_rr(), which
will then commit the lock file.
But for functions like rerere_clear() that do not write to
MERGE_RR, we expect the caller to have handled
setup_rerere(). That caller would then need to release the
lockfile, but it can't; the lock struct is local to
rerere.c.
For builtin/rerere.c, this is OK. We run a single rerere
operation and then exit immediately, which has the side
effect of rolling back the lockfile.
But in builtin/am.c, this is actively wrong. If we run "git
am -3 --skip", we call setup-rerere twice without releasing
the lock:
1. The "--skip" causes us to call am_rerere_clear(), which
calls setup_rerere(), but never drops the lock.
2. We then proceed to the next patch.
3. The "--3way" may cause us to call rerere() to handle
conflicts in that patch, but we are already holding the
lock. The lockfile code dies with:
BUG: prepare_tempfile_object called for active object
We could fix this by having rerere_clear() call
rollback_lock_file(). But it feels a bit odd for it to roll
back a lockfile that it did not itself take. So let's
simplify the interface further, and handle setup_rerere in
the function itself, taking away the question from the
caller over whether they need to do so.
We can give rerere_gc() the same treatment, as well (even
though it doesn't have any callers besides builtin/rerere.c
at this point). Note that these functions don't take flags
from their callers to pass along to setup_rerere; that's OK,
because the flags would not be meaningful for what they are
doing.
Both of those functions need to hold the lock because even
though they do not write to MERGE_RR, they are still writing
and should be protected from a simultaneous "rerere" run.
But rerere_remaining(), "rerere diff", and "rerere status"
are all read-only operations. They want to setup_rerere(),
but do not care about taking the lock in the first place.
Since our update of MERGE_RR is the usual atomic rename done
by commit_lock_file, they can just do a lockless read. For
that, we teach setup_rerere a READONLY flag to avoid the
lock.
As a bonus, this pushes builtin/rerere.c's setup_rerere call
closer to the functions that use it. Which means that "git
rerere totally-bogus-command" will no longer silently
exit(0) in a repository without rerere enabled.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git describe" without argument defaulted to describe the HEAD
commit, but "git describe --contains" didn't. Arguably, in a
repository used for active development, such defaulting would not
be very useful as the tip of branch is typically not tagged, but it
is better to be consistent.
* sg/describe-contains:
describe --contains: default to HEAD when no commit-ish is given
The client side codepaths in "git push" have been cleaned up
and the user can request to perform an optional "signed push",
i.e. sign only when the other end accepts signed push.
* db/push-sign-if-asked:
push: add a config option push.gpgSign for default signed pushes
push: support signing pushes iff the server supports it
builtin/send-pack.c: use parse_options API
config.c: rename git_config_maybe_bool_text and export it as git_parse_maybe_bool
transport: remove git_transport_options.push_cert
gitremote-helpers.txt: document pushcert option
Documentation/git-send-pack.txt: document --signed
Documentation/git-send-pack.txt: wrap long synopsis line
Documentation/git-push.txt: document when --signed may fail
"git notes merge" can be told with "--strategy=<how>" option how to
automatically handle conflicts; this can now be configured by
setting notes.mergeStrategy configuration variable.
* jk/notes-merge-config:
notes: teach git-notes about notes.<name>.mergeStrategy option
notes: add notes.mergeStrategy option to select default strategy
notes: add tests for --commit/--abort/--strategy exclusivity
notes: extract parse_notes_merge_strategy to notes-utils
notes: extract enum notes_merge_strategy to notes-utils.h
notes: document cat_sort_uniq rewriteMode
Recent reimplementation of "git am" changed the format of state
files kept in $GIT_DIR/rebase-apply/ without meaning to do so,
primarily because write_file() API was cumbersome to use and it was
easy to mistakenly make text files with incomplete lines. Update
write_file() interface to make it harder to misuse.
* jc/am-state-fix:
write_file(): drop caller-supplied LF from calls to create a one-liner file
write_file_v(): do not leave incomplete line at the end
write_file(): drop "fatal" parameter
builtin/am: make sure state files are text
builtin/am: introduce write_state_*() helper functions
"git log --cc" did not show any patch, even though most of the time
the user meant "git log --cc -p -m" to see patch output for commits
with a single parent, and combined diff for merge commits. The
command is taught to DWIM "--cc" (without "--raw" and other forms
of output specification) to "--cc -p -m".
* jc/log-p-cc:
builtin/log.c: minor reformat
log: show merge commit when --cc is given
log: when --cc is given, default to -p unless told otherwise
log: rename "tweak" helpers
Because the configuration system does not allow "alias.0foo" and
"pager.0foo" as the configuration key, the user cannot use '0foo'
as a custom command name anyway, but "git 0foo" tried to look these
keys up and emitted useless warnings before saying '0foo is not a
git command'. These warning messages have been squelched.
* jk/fix-alias-pager-config-key-warnings:
config: silence warnings for command names with invalid keys
The gitmodules API accessed from the C code learned to cache stuff
lazily.
* hv/submodule-config:
submodule: allow erroneous values for the fetchRecurseSubmodules option
submodule: use new config API for worktree configurations
submodule: extract functions for config set and lookup
submodule: implement a config API for lookup of .gitmodules values
If you init or clone an empty repository, the initial
message from running "git log" is not very friendly:
$ git init
Initialized empty Git repository in /home/peff/foo/.git/
$ git log
fatal: bad default revision 'HEAD'
Let's detect this situation and write a more friendly
message:
$ git log
fatal: your current branch 'master' does not have any commits yet
We also detect the case that 'HEAD' points to a broken ref;
this should be even less common, but is easy to see. Note
that we do not diagnose all possible cases. We rely on
resolve_ref, which means we do not get information about
complex cases. E.g., "--default master" would use dwim_ref
to find "refs/heads/master", but we notice only that
"master" does not exist. Similarly, a complex sha1
expression like "--default HEAD^2" will not resolve as a
ref.
But that's OK. We fall back to a generic error message in
those cases, and they are unlikely to be used anyway.
Catching an empty or broken "HEAD" improves the common case,
and the other cases are not regressed.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Remove a cache invalidation which would cause the shared index to be
rewritten on as-is commits.
When the cache-tree has changed, we need to update it. But we don't
necessarily need to update the shared index. So setting
active_cache_changed to SOMETHING_CHANGED is unnecessary. Instead, we
let update_main_cache_tree just update the CACHE_TREE_CHANGED bit.
In order to test this, make test-dump-split-index not segfault on
missing replace_bitmap/delete_bitmap. This new codepath is not called
now that the test passes, but is necessary to avoid a segfault when the
new test is run with the old builtin/commit.c code.
Signed-off-by: David Turner <dturner@twopensource.com> Acked-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
read-tree: replace bracket set with parentheses to clarify usage
-u and -i can only be given if -m, --reset, or --prefix is given.
Without parentheses, it looks like -u and -i can be used no matter
what, and the second pair of brackets is confusing.
Signed-off-by: Alex Henrie <alexhenrie24@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-submodule: remove extraneous space from error message
Signed-off-by: Alex Henrie <alexhenrie24@gmail.com> Acked-by: Chris Packham <judge.packham@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
lockfile: remove function "hold_lock_file_for_append"
With 77b9b1d (add_to_alternates_file: don't add duplicate entries,
2015-08-10) the last caller of function "hold_lock_file_for_append"
has been removed, so we can remove the function as well.
Signed-off-by: Ralf Thielow <ralf.thielow@gmail.com> Acked-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The code to open and test the second end of the pipe clearly imitates
the code for the first end. A little too closely, though... Let's fix
the obvious copy-edit bug.
Signed-off-by: Jose F. Morales <jfmcjf@gmail.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Reviewed-by: Jonathan Nieder <jrnieder@gmail.com> Acked-by: Johannes Sixt <j6t@kdbg.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-p4: honor core.ignorecase when using P4 client specs
Perforce depot may record paths in mixed cases, e.g. "p4 files" may
show that there are these two paths:
//depot/Path/to/file1
//depot/pATH/to/file2
and with "p4" or "p4v", these end up in the same directory, e.g.
//depot/Path/to/file1
//depot/Path/to/file2
which is the desired outcome on case insensitive systems.
If git-p4 is used with client spec "//depot/Path/...", however, then
all files not matching the case in the client spec are ignored (in
the example above "//depot/pATH/to/file2").
Fix this by using the path case that appears first in lexicographical
order when core.ignorecase is set to true. This behavior is consistent
with "p4" and "p4v".
Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Acked-by: Luke Diamand <luke@diamand.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Support more than 65535 entries cleanly by writing a "zip64 end of
central directory record" (with a 64-bit field for the number of
entries) before the usual "end of central directory record" (which
contains only a 16-bit field). InfoZIP's zip does the same.
Archives with 65535 or less entries are not affected.
Programs that extract all files like InfoZIP's zip and 7-Zip
ignored the field and could extract all files already. Software
that relies on the ZIP file directory to show a list of contained
files quickly to simulate to normal directory like Windows'
built-in ZIP functionality only saw a subset of the included files.
Windows supports ZIP64 since Vista according to
https://en.wikipedia.org/wiki/Zip_%28file_format%29#ZIP64.
Suggested-by: Johannes Schauer <josch@debian.org> Signed-off-by: Rene Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
archive-zip: use a local variable to store the creator version
Use a simpler conditional right next to the code which makes a higher
creator version necessary -- namely symlink handling and support for
executable files -- instead of a long line with a ternary operator.
The resulting code has more lines but is simpler and allows reuse of
the value easily.
Signed-off-by: Rene Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
A ZIP file directory has a 16-bit field for the number of entries it
contains. There are 64-bit extensions to deal with that. Demonstrate
that git archive --format=zip currently doesn't use them and instead
overflows the field.
InfoZIP's unzip doesn't care about this field and extracts all files
anyway. Software that uses the directory for presenting a filesystem
like view quickly -- notably Windows -- depends on it, but doesn't
lend itself to an automatic test case easily. Use InfoZIP's zipinfo,
which probably isn't available everywhere but at least can provides
*some* way to check this field.
To speed things up a bit create and commit only a subset of the files
and build a fake tree out of duplicates and pass that to git archive.
Signed-off-by: Rene Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>