* sp/mmap: (27 commits)
Spell default packedgitlimit slightly differently
Increase packedGit{Limit,WindowSize} on 64 bit systems.
Update packedGit config option documentation.
mmap: set FD_CLOEXEC for file descriptors we keep open for mmap()
pack-objects: fix use of use_pack().
Fix random segfaults in pack-objects.
Cleanup read_cache_from error handling.
Replace mmap with xmmap, better handling MAP_FAILED.
Release pack windows before reporting out of memory.
Default core.packdGitWindowSize to 1 MiB if NO_MMAP.
Test suite for sliding window mmap implementation.
Create pack_report() as a debugging aid.
Support unmapping windows on 'temporary' packfiles.
Improve error message when packfile mmap fails.
Ensure core.packedGitWindowSize cannot be less than 2 pages.
Load core configuration in git-verify-pack.
Fully activate the sliding window pack access.
Unmap individual windows rather than entire files.
Document why header parsing won't exceed a window.
Loop over pack_windows when inflating/accessing data.
...
* jr/status:
Improve cached content header of status output
Support --amend on initial commit in status output
Improve "nothing to commit" part of status output
Clarify syntax and role of git-add in status output
git-reset <tree> -- <path> restores absense of <path> in <tree>
When <path> exists in the index (either merged or unmerged), and
<tree> does not have it, git-reset should be usable to restore
the absense of it from the tree. This implements it.
diff-index --cached --raw: show tree entry on the LHS for unmerged entries.
This updates the way diffcore represents an unmerged pair
somewhat. It used to be that entries with mode=0 on both sides
were used to represent an unmerged pair, but now it has an
explicit flag. This is to allow diff-index --cached to report
the entry from the tree when the path is unmerged in the index.
This is used in updating "git reset <tree> -- <path>" to restore
absense of the path in the index from the tree.
gitweb: Fix split patches output (e.g. file to symlink)
Do not replace /dev/null in two-line from-file/to-file diff header for
split patches ("split" patch mean more than one patch per one
diff-tree raw line) by a/file or b/file link.
Split patches differ from pair of deletion/creation patch in git diff
header: both a/file and b/file are hyperlinks, in all patches in a
split.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Fix errors in git_patchset_body for empty patches
We now do not skip over empty patches in git_patchset_body (where
empty means that they consist only of git diff header, and of extended
diff header, for example "pure rename" patch). This means that after
extended diff header there can be next patch (i.e. /^diff /) or end of
patchset, and not necessary patch body (i.e. /^--- /).
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Fix error in git_patchest_body for file creation/deletion patch
$from_id, $to_id variables should be local per PATCH.
Fix error in git_patchset_body for file creation (deletion) patches,
where instead of /dev/null as from-file (to-file) diff header line, it
had link to previous file with current file name. This error occured
only if there was another patch before file creation (deletion) patch.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Documentation/git-svn: clarify dcommit, rebase vs pull/merge
Clarify that dcommit creates a revision in SVN for every commit
in git. Also, add 'merge' to the rebase vs pull section because
git-merge is now a first-class UI.
Signed-off-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
This adds ability to do import "in chunks" (default 1000 revisions),
after each chunk git repo will be repacked. The option -R is used to
change default value of chunk size (or how often repository will
repacked).
Signed-off-by: Sasha Khapyorsky <sashak@voltaire.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Describe git-clone's actual behavior in the summary
If a branch other than "master" is checked out in the origin repository,
git-clone makes a local copy of that branch rather than the origin's
"master"
branch. This patch describes the actual behavior.
Signed-off-by: Steven Grimm <koreth@midwinter.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Set default "tar" umask to 002 and owner.group to root.root
In order to make the generated tar files more friendly to users who
extract them as root using GNU tar and its implied -p option, change
the default umask to 002 and change the owner name and group name to
root. This ensures that a) the extracted files and directories are
not world-writable and b) that they belong to user and group root.
Before they would have been assigned to a user and/or group named
git if it existed. This also answers the question in the removed
comment: uid=0, gid=0, uname=root, gname=root is exactly what we
want.
Normal users who let tar apply their umask while extracting are
only affected if their umask allowed the world to change their
files (e.g. a umask of zero). This case is so unlikely and strange
that we don't need to support it.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx> Signed-off-by: Junio C Hamano <junkio@cox.net>
Increase packedGit{Limit,WindowSize} on 64 bit systems.
If we have a 64 bit address space we can easily afford to commit
a larger amount of virtual address space to pack file access.
So on these platforms we should increase the default settings of
core.packedGit{Limit,WindowSize} to something that will better
handle very large projects.
Thanks to Andy Whitcroft for pointing out that we can safely
increase these defaults on such systems.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
The earlier test timestamp was too old; I forgot that the bare
unixtime integer had to be after Jan 1, 2000. This changes
test_tick to use the git-epoch timestamp.
Somehow we forgot to turn save_commit_buffer off while walking
the reachable objects. Releasing the memory for commit object
data that we do not use matters for large projects (for example,
about 90MB is saved while traversing linux-2.6 history).
It might be handy to have a single command that helps you manage
your configuration that relates to downloading from remote
repositories. This currently does only about 20% of what I want
it to do.
$ git remote
shows the list of 'remotes' you have defined somewhere, and
$ git remote origin
shows the details about the named remote (in this case
"origin"). How the branches are tracked, if you have a
tracking branch that is stale, etc.
$ git add another git://git.kernel.org/pub/...
defines the default remote.another.url and remote.another.fetch
entries just like a clone does; you can say "git fetch another"
afterwards.
For it to be useful, I think it should be enhanced to:
- check overlaps of tracking branches and warn;
- offer to remove stale tracking branches in one go;
- offer ways to remove or rename remote;
- offer ways to update an existing remote, perhaps have an
interactive mode;
Other enhancements might be also possible, but I do not think of
anything that is absolutely necessary other than the above right
now.
Blame "linenr" link jumps to previous state at "orig_lineno"
Blame currently displays the commit id which introduced a
block of one or more lines, the line numbers wrt the current
listing of the file and the file's line contents.
The commit id displayed is hyperlinked to the commit.
Currently the linenr links are hyperlinked to the same
commit id displayed to the left, which is _no_ different
than the block of lines displayed, since it is the _same
commit_ that is hyperlinked. And thus clicking on it leads
to the same state of the file for that chunk of
lines. I.e. data mining is not currently possible with
gitweb given a chunk of lines introduced by a commit.
This patch makes such data mining possible.
The line numbers are now hyperlinked to the parent of the
commit id of the block of lines. Furthermore they are
linked to the line where that block was introduced.
Thus clicking on a linenr link will show you the file's
line(s) state prior to the commit id you were viewing.
So clicking continually on a linenr link shows you how this
line and its line number changed over time, leading to the
initial commit where it was first introduced.
Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Fix "Use of uninitialized value" warning in git_tags_body
Fix "Use of uninitialized value" warning in git_tags_body generated
for lightweight tags of tree and blob object; those don't have age
($tag{'age'}) defined.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-svn: make --repack work consistently between fetch and multi-fetch
Since fetch reforks itself at most every 1000 revisions, we
need to update the counter in the parent process to have a
working count if we set our repack interval to be > ~1000
revisions. multi-fetch has always done this correctly
because of an extra process; now fetch uses the extra process;
as well.
While we're at it, only compile the $sha1 regex that checks for
repacking once.
Signed-off-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
It now requires at least one of the (trunk|branch|tags) arguments
(either from the command-line or in .git/config). Also we make
sure that anything that is passed as a URL ('help') in David's
case is actually a URL.
Thanks to David Kågedal for reporting this issue.
Signed-off-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
The variable named entry is allocated using malloc() and then
forgotten, it being shadowed by an automatic variable of the
same name. Fixing the array size at 3 worked so far because
the only caller of traverse_trees() needed only as much
entries. Simply remove the shadowing varaible and we're able
to traverse more than three trees and save stack space at the
same time!
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx> Signed-off-by: Junio C Hamano <junkio@cox.net>
pack-check.c::verify_packfile(): don't run SHA-1 update on huge data
Running the SHA1_Update() on the whole packfile in a single call
revealed an overflow problem we had in the SHA-1 implementation
on POWER architecture some time ago, which was fixed with commit b47f509b (June 19, 2006). Other SHA-1 implementations may have
a similar problem.
The sliding mmap() series already makes chunked calls to
SHA1_Update(), so this patch itself will become moot when it
graduates to "master", but in the meantime, run the hash
function in smaller chunks to prevent possible future problems.
My change in 190d7fdcf325bb444fa806f09ebbb403a4ae4ee6 had a small bug
found by Michael Krufky which caused the passed in hash value to be
ignored, so shortlog would only show the HEAD revision.
Signed-off-by: Robert Fitzsimons <robfitz@273k.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: There can be empty patches (in git_patchset_body)
We now do not skip over empty patches in git_patchset_body
(where empty means that they consist only of git diff header,
and of extended diff header), so uncomment branch of code dealing
with empty patches (patches which do not have even two-line
from/to header)
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
- Teach how to delete a branch with "git branch -d name".
- Usually a commit has one parent; merge has more.
- Teach "git show" instead of "git cat-file -p".
Signed-off-by: Santi Béjar <sbejar@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
This modifies pretty_print_commit() to make the output of git-rev-list and
friends a bit more predictable.
A commit body starting with blank lines might be unheard-of, but still possible
to create using git-commit-tree (so is bound to appear somewhere, sometime).
Signed-off-by: Lars Hjemli <hjemli@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Instead of "$projectroot/$pr->{'path'}" to get the path to project
GIT_DIR, it was used "$projectroot/$project" which is valid only
for actions where project parameter is set, and 'project_index' is not
one of them.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Fix infinite loop when deleting multiple packed refs.
It was stupid to link the same element twice to lock_file_list
and end up in a loop, so we certainly need a fix.
But it is not like we are taking a lock on multiple files in
this case. It is just that we leave the linked element on the
list even after commit_lock_file() successfully removes the
cruft.
We cannot remove the list element in commit_lock_file(); if we
are interrupted in the middle of list manipulation, the call to
remove_lock_file_on_signal() will happen with a broken list
structure pointed by lock_file_list, which would cause the cruft
to remain, so not removing the list element is the right thing
to do. Instead we should be reusing the element already on the
list.
There is already a code for that in lock_file() function in
lockfile.c. The code checks lk->next and the element is linked
only when it is not already on the list -- which is incorrect
for the last element on the list (which has NULL in its next
field), but if you read the check as "is this element already on
the list?" it actually makes sense. We do not want to link it
on the list again, nor we would want to set up signal/atexit
over and over.
This tries to be more to the point while also including a pointer on how to
unstage changes from the index.
Since this header is printed in two different code paths and the name of the
reference commit is needed for the unstage part, provide a new printing
function.
Signed-off-by: Jürgen Rühle <j-r@online.de> Signed-off-by: Junio C Hamano <junkio@cox.net>
Support --amend on initial commit in status output
We check the existence of the parent commit to determine whether the status is
requested for an initial commit. Since the parent commit depends on the
presence of the --amend switch do initial commit detection after command line
arguments have been handled.
Signed-off-by: Jürgen Rühle <j-r@online.de> Signed-off-by: Junio C Hamano <junkio@cox.net>
Previously git-status in a clean working directory would advice the user to use
git add. This isn't very helpful when there is nothing to add in the working
directory, therefore note a clean working directory while displaying the other
sections and print the appropriate message for each case.
Signed-off-by: Jürgen Rühle <j-r@online.de> Signed-off-by: Junio C Hamano <junkio@cox.net>
send pack check for failure to send revisions list
When passing the revisions list to pack-objects we do not check for
errors nor short writes. Introduce a new write_in_full which will
handle short writes and report errors to the caller. Use this to
short cut the send on failure, allowing us to wait for and report
the child in case the failure is its fault.
Signed-off-by: Andy Whitcroft <apw@shadowen.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
* 'sp/merge' (early part):
Use merge-recursive in git-am -3.
Allow merging bare trees in merge-recursive.
Move better_branch_name above get_ref in merge-recursive.
Documentation: update git-pull.txt for new clone behavior
Update examples, stop using branch named "origin" as an example.
Remove large example of use of remotes; that particular case is
nicely automated by default, so it's not so pressing to explain, and
we can refer to git-repo-config for the details.
Signed-off-by: "J. Bruce Fields" <bfields@citi.umich.edu> Signed-off-by: Junio C Hamano <junkio@cox.net>
Removal of them is needed regardless of errors. The original
code had the removal outside of the process which sets the flag
to tell the later step what to remove, but it runs as a
downstream of a pipeline and its effect was lost.
fail pull/merge early in the middle of conflicted merge
After a pull that results in a conflicted merge, a new user
often tries another "git pull" in desperation. When the index
is unmerged, merge backends correctly bail out without touching
either index nor the working tree, so this does not make the
wound any worse.
The user will however see several lines of messsages during this
process, such as "filename: needs merge", "you need to resolve
your current index first", "Merging...", and "Entry ... would be
overwritten by merge. Cannot merge.". They are unnecessarily
alarming, and cause useful conflict messages from the first pull
scroll off the top of the terminal.
This changes pull and merge to run "git-ls-files -u" upfront and
stop them much earlier than we currently do. Old timers may
know better and would not to try pulling again before cleaning
things up; this change adds extra overhead that is unnecessary
for them. But this would be worth paying for to save new people
from needless confusion.
This removes some unnecessary 'svn up' calls throughout
t9103-git-svn-graft-branches.sh:
* removed an 'svn log' call that was leftover from debugging
* removed multiple git-svn calls with a multi-init / multi-fetch
combination (which weren't tested before, either)
* replaced `rev-list ... | head -n1` with `rev-parse ...`
(not sure what I was thinking when I wrote that)
All this saves about 9 seconds from a test run
(53s -> 44s for 'make t91*') on my 1.3GHz Athlon
Signed-off-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-svn: t/t9100-git-svn-basic: remove old check for NO_SYMLINK
We don't support the svn command-line client anymore; nor
do we support anything before SVN 1.1.0, so we can be certain
symlinks will be supported in the SVN repository.
Signed-off-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
i18n: do not leak 'encoding' header even when we cheat the conversion.
We special case the case where encoding recorded in the commit
and the output encoding are the same and do not call iconv().
But we should drop 'encoding' header for this case as well for
consistency.
Do not merge random set of refs out of wildcarded refs
When your fetch configuration has only the wildcards, we would
pick the lexicographically first ref from the remote side for
merging, which was complete nonsense. Make sure nothing except
the one that is specified with branch.*.merge is merged in this
case.
Verify that the update hooks work as documented/advertised. This is
a simple set of tests to check that the update hooks run with the
parameters expected, have their STDOUT and STDERR redirected to
the client side of the connection, and that their STDIN does not
contain any data (as its actually /dev/null).
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
* master:
Documentation/config.txt (and repo-config manpage): mark-up fix.
Teach Git how to parse standard power of 2 suffixes.
Use /dev/null for update hook stdin.
Redirect update hook stdout to stderr.
Remove unnecessary argc parameter from run_command_v.
Automatically detect a bare git repository.
Replace "GIT_DIR" with GIT_DIR_ENVIRONMENT.
Use PATH_MAX constant for --bare.
Force core.filemode to false on Cygwin.
Fix formatting for urls section of fetch, pull, and push manpages
Fix yet another subtle xdl_merge() bug
i18n: drop "encoding" header in the output after re-coding.
commit-tree: cope with different ways "utf-8" can be spelled.
Move commit reencoding parameter parsing to revision.c
Documentation: minor rewording for git-log and git-show pages.
Documentation: i18n commit log message notes.
t3900: test log --encoding=none
commit re-encoding: fix confusion between no and default conversion.
Teach Git how to parse standard power of 2 suffixes.
Sometimes its necessary to supply a value as a power of two in a
configuration parameter. In this case the user may want to use the
standard suffixes such as K, M, or G to indicate that the numerical
value should be multiplied by a constant base before being used.
Shell scripts/etc. can also benefit from this automatic option
parsing with `git repo-config --int`.
[jc: with a couple of test and a slight input tightening]
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Currently the update hook invoked by receive-pack has its stdin
connected to the pushing client. The hook shouldn't attempt to
read from this stream, and doing so may consume data that was
meant for receive-pack. Instead we should give the update hook
/dev/null as its stdin, ensuring that it always receives EOF and
doesn't disrupt the protocol if it attempts to read any data.
The post-update hook is similar, as it gets invoked with /dev/null
on stdin to prevent the hook from reading data from the client.
Previously we had invoked it with stdout also connected to /dev/null,
throwing away anything on stdout, to prevent client protocol errors.
Instead we should redirect stdout to stderr, like we do with the
update hook.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
If an update hook outputs to stdout then that output will be sent
back over the wire to the push client as though it were part of
the git protocol. This tends to cause protocol errors on the
client end of the connection, as the hook output is not expected
in that context. Most hook developers work around this by making
sure their hook outputs everything to stderr.
But hooks shouldn't need to perform such special behavior. Instead
we can just dup stderr to stdout prior to invoking the update hook.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Remove unnecessary argc parameter from run_command_v.
The argc parameter is never used by the run_command_v family of
functions. Instead they require that the passed argv[] be NULL
terminated so they can rely on the operating system's execvp
function to correctly pass the arguments to the new process.
Making the caller pass the argc is just confusing, as the caller
could be mislead into believing that the argc might take precendece
over the argv, or that the argv does not need to be NULL terminated.
So goodbye argc. Don't come back.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Many users find it unfriendly that they can create a bare git
repository easily with `git clone --bare` but are then unable to
run simple commands like `git log` once they cd into that newly
created bare repository. This occurs because we do not check to
see if the current working directory is a git repository.
Instead of failing out with "fatal: Not a git repository" we should
try to automatically detect if the current working directory is
a bare repository and use that for GIT_DIR, and fail out only if
that doesn't appear to be true.
We test the current working directory only after we have tried
searching up the directory tree. This is to retain backwards
compatibility with our previous behavior on the off chance that
a user has a 'refs' and 'objects' subdirectories and a 'HEAD'
file that looks like a symref, all stored within a repository's
associated working directory.
This change also consolidates the validation logic between the case
of GIT_DIR being supplied and GIT_DIR not being supplied, cleaning
up the code.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Many users have noticed that core.filemode doesn't appear to be
automatically set right on Cygwin when using a repository stored
on NTFS. The issue is that Cygwin and NTFS correctly supports
the executable mode bit, and Git properly detected that, but most
native Windows applications tend to create files such that Cygwin
sees the executable bit set when it probably shouldn't be.
This is especially bad if the user's favorite editor deletes the
file then recreates it whenever they save (vs. just overwriting)
as now a file that was created with mode 0644 by checkout-index
appears to have mode 0755.
So we introduce NO_TRUSTABLE_FILEMODE, settable at compile time.
Setting this option forces core.filemode to false, even if the
detection code would have returned true. This option should be
enabled by default on Cygwin.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Fix formatting for urls section of fetch, pull, and push manpages
The line:
[remote "<remote>"]
was getting swallowed up by asciidoc, causing a critical line in the
explanation for how to store the .git/remotes information in .git/config
to go missing from the git-fetch, git-pull, and git-push manpages.
Put all of the examples into delimited blocks to fix this problem and to
make them look nicer.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Signed-off-by: Junio C Hamano <junkio@cox.net>
In very obscure cases, a merge can hit an unexpected code path (where the
original code went as far as saying that this was a bug). This failing
merge was noticed by Alexandre Juillard.
The problem is that the original file contains something like this:
-- snip --
two non-empty lines
before two empty lines
after two empty lines
-- snap --
and this snippet is reduced to _one_ empty line in _both_ new files.
However, it is ambiguous as to which hunk takes the empty line: the first
or the second one?
Indeed in Alexandre's example files, the xdiff algorithm attributes the
empty line to the first hunk in one case, and to the second hunk in the
other case.
(Trimming down the example files _changes_ that behaviour!)
Thus, the call to xdl_merge_cmp_lines() has no chance to realize that the
change is actually identical in both new files. Therefore,
xdl_refine_conflicts() finds an empty diff script, which was not expected
there, because (the original author of xdl_merge() thought)
xdl_merge_cmp_lines() would catch that case earlier.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <junkio@cox.net>
i18n: drop "encoding" header in the output after re-coding.
After re-coding the commit message into the encoding the user
specified (either with core.logoutputencidng or --encoding
option), this drops the "encoding" header altogether. The
output is after re-coding as the user asked (either with the
config or --encoding=<encoding> option), and the extra header
becomes redundant information.