* maint-2.5:
Git 2.5.5
Git 2.4.11
list-objects: pass full pathname to callbacks
list-objects: drop name_path entirely
list-objects: convert name_path to a strbuf
show_object_with_name: simplify by using path_name()
http-push: stop using name_path
tree-diff: catch integer overflow in combine_diff_path allocation
add helpers for detecting size_t overflow
* maint-2.4:
Git 2.4.11
list-objects: pass full pathname to callbacks
list-objects: drop name_path entirely
list-objects: convert name_path to a strbuf
show_object_with_name: simplify by using path_name()
http-push: stop using name_path
tree-diff: catch integer overflow in combine_diff_path allocation
add helpers for detecting size_t overflow
Merge branch 'jk/path-name-safety-2.4' into maint-2.4
Bugfix patches were backported from the 'master' front to plug heap
corruption holes, to catch integer overflow in the computation of
pathname lengths, and to get rid of the name_path API. Both of
these would have resulted in writing over an under-allocated buffer
when formulating pathnames while tree traversal.
* jk/path-name-safety-2.4:
list-objects: pass full pathname to callbacks
list-objects: drop name_path entirely
list-objects: convert name_path to a strbuf
show_object_with_name: simplify by using path_name()
http-push: stop using name_path
tree-diff: catch integer overflow in combine_diff_path allocation
add helpers for detecting size_t overflow
Translate 22 new messages came from git.pot update in f1522b2
(l10n: git.pot: v2.8.0 round 2 (21 new, 1 removed)) and a5a4168
(l10n: git.pot: Add one new message for Git 2.8.0).
* maint:
list-objects: pass full pathname to callbacks
list-objects: drop name_path entirely
list-objects: convert name_path to a strbuf
show_object_with_name: simplify by using path_name()
http-push: stop using name_path
tree-diff: catch integer overflow in combine_diff_path allocation
add helpers for detecting size_t overflow
Recent versions of GNU grep is pickier than before to decide if a
file is "binary" and refuse to give line-oriented hits when we
expect it to, unless explicitly told with "-a" option. As our
scripted Porcelains use sane_grep wrapper for line-oriented data,
even when the line may contain non-ASCII payload we took from
end-user data, use "grep -a" to implement sane_grep wrapper when
using an implementation of "grep" that takes the "-a" option.
* jc/sane-grep:
rebase-i: clarify "is this commit relevant?" test
sane_grep: pass "-a" if grep accepts it
The two alternative ways to spell "ssh://" transport have been
deprecated for a long time. The last mention of them has finally
removed from the documentation.
* cn/deprecate-ssh-git-url:
Disown ssh+git and git+ssh
git-svn: fix URL canonicalization during init w/ SVN 1.7+
URL canonicalization when full URLs are passed became broken
when using SVN::_Core::svn_dirent_canonicalize under SVN 1.7.
Ensure we canonicalize paths and URLs with appropriate functions
for each type from now on as the path/URL-agnostic
SVN::_Core::svn_path_canonicalize function is deprecated in SVN.
Reported-by: Adam Dinwoodie <adam@dinwoodie.org>
http://mid.gmane.org/20160315162344.GM29016@dinwoodie.org Signed-off-by: Eric Wong <normalperson@yhbt.net>
* jk/path-name-safety-2.7:
list-objects: pass full pathname to callbacks
list-objects: drop name_path entirely
list-objects: convert name_path to a strbuf
show_object_with_name: simplify by using path_name()
http-push: stop using name_path
tree-diff: catch integer overflow in combine_diff_path allocation
add helpers for detecting size_t overflow
t9117: test specifying full url to git svn init -T
According to the documentation, full URLs can be specified in the `-T`
argument to `git svn init`. However, the canonicalization of such
arguments squashes together consecutive "/"s, which unsurprisingly
breaks http://, svn://, etc URLs. Add a failing test case to provide
evidence of that.
On systems where Subversion provides svn_path_canonicalize but not
svn_dirent_canonicalize (Subversion 1.6 and earlier?), this test passes,
as svn_path_canonicalize doesn't mangle the consecutive "/"s.
[ew: fixed whitespace]
Signed-off-by: Adam Dinwoodie <adam@dinwoodie.org> Signed-off-by: Eric Wong <normalperson@yhbt.net>
Merge branch 'jk/path-name-safety-2.6' into jk/path-name-safety-2.7
* jk/path-name-safety-2.6:
list-objects: pass full pathname to callbacks
list-objects: drop name_path entirely
list-objects: convert name_path to a strbuf
show_object_with_name: simplify by using path_name()
http-push: stop using name_path
tree-diff: catch integer overflow in combine_diff_path allocation
add helpers for detecting size_t overflow
Merge branch 'jk/path-name-safety-2.5' into jk/path-name-safety-2.6
* jk/path-name-safety-2.5:
list-objects: pass full pathname to callbacks
list-objects: drop name_path entirely
list-objects: convert name_path to a strbuf
show_object_with_name: simplify by using path_name()
http-push: stop using name_path
tree-diff: catch integer overflow in combine_diff_path allocation
add helpers for detecting size_t overflow
Merge branch 'jk/path-name-safety-2.4' into jk/path-name-safety-2.5
* jk/path-name-safety-2.4:
list-objects: pass full pathname to callbacks
list-objects: drop name_path entirely
list-objects: convert name_path to a strbuf
show_object_with_name: simplify by using path_name()
http-push: stop using name_path
tree-diff: catch integer overflow in combine_diff_path allocation
add helpers for detecting size_t overflow
When we find a blob at "a/b/c", we currently pass this to
our show_object_fn callbacks as two components: "a/b/" and
"c". Callbacks which want the full value then call
path_name(), which concatenates the two. But this is an
inefficient interface; the path is a strbuf, and we could
simply append "c" to it temporarily, then roll back the
length, without creating a new copy.
So we could improve this by teaching the callsites of
path_name() this trick (and there are only 3). But we can
also notice that no callback actually cares about the
broken-down representation, and simply pass each callback
the full path "a/b/c" as a string. The callback code becomes
even simpler, then, as we do not have to worry about freeing
an allocated buffer, nor rolling back our modification to
the strbuf.
This is theoretically less efficient, as some callbacks
would not bother to format the final path component. But in
practice this is not measurable. Since we use the same
strbuf over and over, our work to grow it is amortized, and
we really only pay to memcpy a few bytes.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
In the previous commit, we left name_path as a thin wrapper
around a strbuf. This patch drops it entirely. As a result,
every show_object_fn callback needs to be adjusted. However,
none of their code needs to be changed at all, because the
only use was to pass it to path_name(), which now handles
the bare strbuf.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The "struct name_path" data is examined in only two places:
we generate it in process_tree(), and we convert it to a
single string in path_name(). Everyone else just passes it
through to those functions.
We can further note that process_tree() already keeps a
single strbuf with the leading tree path, for use with
tree_entry_interesting().
Instead of building a separate name_path linked list, let's
just use the one we already build in "base". This reduces
the amount of code (especially tricky code in path_name()
which did not check for integer overflows caused by deep
or large pathnames).
It is also more efficient in some instances. Any time we
were using tree_entry_interesting, we were building up the
strbuf anyway, so this is an immediate and obvious win
there. In cases where we were not, we trade off storing
"pathname/" in a strbuf on the heap for each level of the
path, instead of two pointers and an int on the stack (with
one pointer into the tree object). On a 64-bit system, the
latter is 20 bytes; so if path components are less than that
on average, this has lower peak memory usage. In practice
it probably doesn't matter either way; we are already
holding in memory all of the tree objects leading up to each
pathname, and for normal-depth pathnames, we are only
talking about hundreds of bytes.
This patch leaves "struct name_path" as a thin wrapper
around the strbuf, to avoid disrupting callbacks. We should
fix them, but leaving it out makes this diff easier to view.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
show_object_with_name: simplify by using path_name()
When "git rev-list" shows an object with its associated path
name, it does so by walking the name_path linked list and
printing each component (stopping at any embedded NULs or
newlines).
We'd like to eventually get rid of name_path entirely in
favor of a single buffer, and dropping this custom printing
code is part of that. As a first step, let's use path_name()
to format the list into a single buffer, and print that.
This is strictly less efficient than the original, but it's
a temporary step in the refactoring; our end game will be to
get the fully formatted name in the first place.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Performing computations on size_t variables that we feed to
xmalloc and friends can be dangerous, as an integer overflow
can cause us to allocate a much smaller chunk than we
realized.
We already have unsigned_add_overflows(), but let's add
unsigned_mult_overflows() to that. Furthermore, rather than
have each site manually check and die on overflow, we can
provide some helpers that will:
- promote the arguments to size_t, so that we know we are
doing our computation in the same size of integer that
will ultimately be fed to xmalloc
- check and die on overflow
- return the result so that computations can be done in
the parameter list of xmalloc.
These functions are a lot uglier to use than normal
arithmetic operators (you have to do "st_add(foo, bar)"
instead of "foo + bar"). To at least limit the damage, we
also provide multi-valued versions. So rather than:
st_add(st_add(a, b), st_add(c, d));
you can write:
st_add4(a, b, c, d);
This isn't nearly as elegant as a varargs function, but it's
a lot harder to get it wrong. You don't have to remember to
add a sentinel value at the end, and the compiler will
complain if you get the number of arguments wrong. This
patch adds only the numbered variants required to convert
the current code base; we can easily add more later if
needed.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The graph traversal code here passes along a name_path to
build up the pathname at which we find each blob. But we
never actually do anything with the resulting names, making
it a waste of code and memory.
This usage came in aa1dbc9 (Update http-push functionality,
2006-03-07), and originally the result was passed to
"add_object" (which stored it, but didn't really use it,
either). But we stopped using that function in 1f1e895 (Add
"named object array" concept, 2006-06-19) in favor of
storing just the objects themselves.
Moreover, the generation of the name in process_tree() is
buggy. It sticks "name" onto the end of the name_path linked
list, and then passes it down again as it recurses (instead
of "entry.path"). So it's a good thing this was unused, as
the resulting path for "a/b/c/d" would end up as "a/a/a/a".
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
tree-diff: catch integer overflow in combine_diff_path allocation
A combine_diff_path struct has two "flex" members allocated
alongside the struct: a string to hold the pathname, and an
array of parent pointers. We use an "int" to compute this,
meaning we may easily overflow it if the pathname is
extremely long.
We can fix this by using size_t, and checking for overflow
with the st_add helper.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When trying to find a good spot for testing clone with submodules, I
got confused where to add a new test file. There are both tests in t560*
as well as t57* both testing the clone command. t/README claims the
second digit is to indicate the command, which is inconsistent to the
current naming structure.
Rename all t57* tests to be in t56* to follow the pattern of the digits
as laid out in t/README.
It would have been less work to rename t56* => t57* because there are less
files, but the tests in t56* look more basic and I assumed the higher the
last digits the more complicated niche details are tested, so with the patch
now it looks more in order to me.
Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Error messages should attempt to fit within the confines of
an 80-column terminal to avoid compatibility and accessibility
problems. Furthermore the word "directories" can be misleading
when used in the context of git refnames.
Expand the area of globs applicability for branches and tags
in git-svn. It is now possible to use globs like 'a*e', or 'release_*'.
This allows users to avoid long lines in config like:
While I was checking all the call sites of sane_grep and sane_egrep,
I noticed this one is somewhat strangely written. The lines in the
file sane_grep works on all begin with 40-hex object name, so there
is no real risk of confusing "test $(...) = ''" by finding something
that begins with a dash, but using the status from sane_grep makes
it a lot clearer what is going on.
Newer versions of GNU grep is reported to be pickier when we feed a
non-ASCII input and break some Porcelain scripts. As we know we do
not feed random binary file to our own sane_grep wrapper, allow us
to always pass "-a" by setting SANE_TEXT_GREP=-a Makefile variable
to work it around.
If two branches each move a file into different directories then
mergetool will fail because it assumes that the file being merged, and
its parent directory, are present in the worktree.
Create the merge file's parent directory to allow using the
deleted base version of the file for merge resolution when
encountering a delete/delete conflict.
The end result is that a delete/delete conflict is presented for the
user to resolve.
Reported-by: Joe Einertson <joe@kidblog.org> Signed-off-by: David Aguilar <davvid@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Merge branch 'jk/pack-idx-corruption-safety' into maint
The code to read the pack data using the offsets stored in the pack
idx file has been made more carefully check the validity of the
data in the idx.
* jk/pack-idx-corruption-safety:
sha1_file.c: mark strings for translation
use_pack: handle signed off_t overflow
nth_packed_object_offset: bounds-check extended offset
t5313: test bounds-checks of corrupted/malicious pack/idx files
Merge branch 'js/config-set-in-non-repository' into maint
"git config section.var value" to set a value in per-repository
configuration file failed when it was run outside any repository,
but didn't say the reason correctly.
* js/config-set-in-non-repository:
git config: report when trying to modify a non-existing repo config
Merge branch 'sb/submodule-module-list-fix' into maint
A helper function "git submodule" uses since v2.7.0 to list the
modules that match the pathspec argument given to its subcommands
(e.g. "submodule add <repo> <path>") has been fixed.
Merge branch 'jk/grep-binary-workaround-in-test' into maint
Recent versions of GNU grep are pickier when their input contains
arbitrary binary data, which some of our tests uses. Rewrite the
tests to sidestep the problem.
* jk/grep-binary-workaround-in-test:
t9200: avoid grep on non-ASCII data
t8005: avoid grep on non-ASCII data
* jk/tighten-alloc: (23 commits)
compat/mingw: brown paper bag fix for 50a6c8e
ewah: convert to REALLOC_ARRAY, etc
convert ewah/bitmap code to use xmalloc
diff_populate_gitlink: use a strbuf
transport_anonymize_url: use xstrfmt
git-compat-util: drop mempcpy compat code
sequencer: simplify memory allocation of get_message
test-path-utils: fix normalize_path_copy output buffer size
fetch-pack: simplify add_sought_entry
fast-import: simplify allocation in start_packfile
write_untracked_extension: use FLEX_ALLOC helper
prepare_{git,shell}_cmd: use argv_array
use st_add and st_mult for allocation size computation
convert trivial cases to FLEX_ARRAY macros
use xmallocz to avoid size arithmetic
convert trivial cases to ALLOC_ARRAY
convert manual allocations to argv_array
argv-array: add detach function
add helpers for allocating flex-array structs
harden REALLOC_ARRAY and xcalloc against size_t overflow
...
"git merge-tree" used to mishandle "both sides added" conflict with
its own "create a fake ancestor file that has the common parts of
what both sides have added and do a 3-way merge" logic; this has
been updated to use the usual "3-way merge with an empty blob as
the fake common ancestor file" approach used in the rest of the
system.
* jk/no-diff-emit-common:
xdiff: drop XDL_EMIT_COMMON
merge-tree: drop generate_common strategy
merge-one-file: use empty blob for add/add base
Merge branch 'nd/dwim-wildcards-as-pathspecs' into maint
"git show 'HEAD:Foo[BAR]Baz'" did not interpret the argument as a
rev, i.e. the object named by the the pathname with wildcard
characters in a tree object.
Handling of errors while writing into our internal asynchronous
process has been made more robust, which reduces flakiness in our
tests.
* jk/epipe-in-async:
t5504: handle expected output from SIGPIPE death
test_must_fail: report number of unexpected signal
fetch-pack: ignore SIGPIPE in sideband demuxer
write_or_die: handle EPIPE in async threads
Many codepaths forget to check return value from git_config_set();
the function is made to die() to make sure we do not proceed when
setting a configuration variable failed.
* ps/config-error:
config: rename git_config_set_or_die to git_config_set
config: rename git_config_set to git_config_set_gently
compat: die when unable to set core.precomposeunicode
sequencer: die on config error when saving replay opts
init-db: die on config errors when initializing empty repo
clone: die on config error in cmd_clone
remote: die on config error when manipulating remotes
remote: die on config error when setting/adding branches
remote: die on config error when setting URL
submodule--helper: die on config error when cloning module
submodule: die on config error when linking modules
branch: die on config error when editing branch description
branch: die on config error when unsetting upstream
branch: report errors in tracking branch setup
config: introduce set_or_die wrappers
Traditionally, the tests that try commands that work on the
contents in the working tree were named with "worktree" in their
filenames, but with the recent addition of "git worktree"
subcommand, whose tests are also named similarly, it has become
harder to tell them apart. The traditional tests have been renamed
to use "work-tree" instead in an attempt to differentiate them.
* mg/work-tree-tests:
tests: rename work-tree tests to *work-tree*
Some people argue that these were silly from the beginning (see
http://thread.gmane.org/gmane.comp.version-control.git/285590/focus=285601
for example), but we have to support them for compatibility.
That doesn't mean we have to show them in the documentation. These
were already left out of the main list, but a reference in the main
manpage was left, so remove that.
Also add a note to discourage their use if anybody goes looking for them
in the source code.
Signed-off-by: Carlos Martín Nieto <cmn@dwim.me> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Some of the test-* programs rely on examining refs, but did
not bother to make sure we are actually in a git repository.
Let's have them call setup_git_directory() to do so.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
If we are running "git grep --no-index" outside of a git
repository, we behave roughly like "grep -r", examining all
files in the current directory and its subdirectories.
However, because we use fill_directory() to do the
recursion, it will skip over any directories which look like
sub-repositories.
For a normal git operation (like "git grep" in a repository)
this makes sense; we do not want to cross the boundary out
of our current repository into a submodule. But for
"--no-index" without a repository, we should look at all
files, including embedded repositories.
There is one exception, though: we probably should _not_
descend into ".git" directories. Doing so is inefficient and
unlikely to turn up useful hits.
This patch drops our use of dir.c's gitlink-detection, but
we do still avoid ".git". That makes us more like tools such
as "ack" or "ag", which also know to avoid cruft in .git.
As a bonus, this also drops our usage of the ref code
when we are outside of a repository, making the transition
to pluggable ref backends cleaner.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The mailmap code may be triggered outside of a repository by
git-shortlog. There is no point in looking up a name like
"HEAD:.mailmap" there; without a repository, we have no
refs.
This is unlikely to matter much in practice for the current
code, as we would simply fail to find the ref. But as the
refs code learns about new backends, this is more important;
without a repository, we do not even know which backend to
look at.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The remote-config code wants to look at HEAD to mark the
current branch specially. But if we are not in a repository
(e.g., running "git archive --remote"), this makes no sense;
there is no HEAD to look at, and we have no current branch.
This doesn't really cause any bugs in practice (if you are
not in a repo, you probably don't have a .git/HEAD file),
but we should be more careful about triggering the refs code
at all in a non-repo. As we grow new ref backends, we would
not even know which backend to use.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
setup: set startup_info->have_repository more reliably
When setup_git_directory() is called, we set a flag in
startup_info to indicate we have a repository. But there are
a few other mechanisms by which we might set up a repo:
1. When creating a new repository via init_db(), we
transition from no-repo to being in a repo. We should
tweak this flag at that moment.
2. In enter_repo(), a stricter form of
setup_git_directory() used by server-side programs, we
check the repository format config. After doing so, we
know we're in a repository, and can set the flag.
With these changes, library code can now reliably tell
whether we are in a repository and act accordingly. We'll
leave the "prefix" field as NULL, which is what happens when
setup_git_directory() finds there is no prefix.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit a60645f (setup: remember whether repository was
found, 2010-08-05) introduced the startup_info structure,
which records some parts of the setup_git_directory()
process (notably, whether we actually found a repository or
not).
One of the uses of this data is for functions to behave
appropriately based on whether we are in a repo. But the
startup_info struct is just a pointer to storage provided by
the main program, and the only program that sets it up is
the git.c wrapper. Thus builtins have access to
startup_info, but externally linked programs do not.
Worse, library code which is accessible from both has to be
careful about accessing startup_info. This can be used to
trigger a die("BUG") via get_sha1():
$ git fast-import <<-\EOF
tag foo
from HEAD:./whatever
EOF
fatal: BUG: startup_info struct is not initialized.
Obviously that's fairly nonsensical input to feed to
fast-import, but we should never hit a die("BUG"). And there
may be other ways to trigger it if other non-builtins
resolve sha1s.
So let's point the storage for startup_info to a static
variable in setup.c, making it available to all users of the
library code. We _could_ turn startup_info into a regular
extern struct, but doing so would mean tweaking all of the
existing use sites. So let's leave the pointer indirection
in place. We can, however, drop any checks for NULL, as
they will always be false (and likewise, we can drop the
test covering this case, which was a rather artificial
situation using one of the test-* programs).
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
strbuf_getwholeline: NUL-terminate getdelim buffer on error
Commit 0cc30e0 (strbuf_getwholeline: use getdelim if it is
available, 2015-04-16) tries to clean up after getdelim()
returns EOF, but gets one case wrong, which can lead in some
obscure cases to us reading uninitialized memory.
After getdelim() returns -1, we re-initialize the strbuf
only if sb->buf is NULL. The thinking was that either:
1. We fed an existing allocated buffer to getdelim(), and
at most it would have realloc'd, leaving our NUL in
place.
2. We didn't have a buffer to feed, so we gave getdelim()
NULL; sb->buf will remain NULL, and we just want to
restore the empty slopbuf.
But that second case isn't quite right. getdelim() may
allocate a buffer, write nothing into it, and then return
EOF. The resulting strbuf rightfully has sb->len set to "0",
but is missing the NUL terminator in the first byte.
Most call-sites are fine with this. They see the EOF and
don't bother looking at the strbuf. Or they notice that
sb->len is empty, and don't look at the contents. But
there's at least one case that does neither, and relies on
parsing the resulting (possibly zero-length) string:
fast-import. You can see this in action with the new test
(though we probably only notice failure there when run with
--valgrind or ASAN).
We can fix this by unconditionally resetting the strbuf when
we have a buffer after getdelim(). That fixes case 2 above.
Case 1 is probably already fine in practice, but it does not
hurt for us to re-assert our invariants (especially because
we are relying on whatever getdelim() happens to do, which
may vary from platform to platform). Our fix covers that
case, too.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>