When we call show_signature or show_mergetag, we read the
commit object fresh via read_sha1_file and reparse its
headers. However, in most cases we already have the object
data available, attached to the "struct commit". This is
partially laziness in dealing with the memory allocation
issues, but partially defensive programming, in that we
would always want to verify a clean version of the buffer
(not one that might have been munged by other users of the
commit).
However, we do not currently ever munge the commit buffer,
and not using the already-available buffer carries a fairly
big performance penalty when we are looking at a large
number of commits. Here are timings on linux.git:
[baseline, no signatures]
$ time git log >/dev/null
real 0m4.902s
user 0m4.784s
sys 0m0.120s
[before]
$ time git log --show-signature >/dev/null
real 0m14.735s
user 0m9.964s
sys 0m0.944s
[after]
$ time git log --show-signature >/dev/null
real 0m9.981s
user 0m5.260s
sys 0m0.936s
Note that our user CPU time drops almost in half, close to
the non-signature case, but we do still spend more
wall-clock and system time, presumably from dealing with
gpg.
An alternative to this is to note that most commits do not
have signatures (less than 1% in this repo), yet we pay the
re-parsing cost for every commit just to find out if it has
a mergetag or signature. If we checked that when parsing the
commit initially, we could avoid re-examining most commits
later on. Even if we did pursue that direction, however,
this would still speed up the cases where we _do_ have
signatures. So it's probably worth doing either way.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Most callsites which use the commit buffer try to use the
cached version attached to the commit, rather than
re-reading from disk. Unfortunately, that interface provides
only a pointer to the NUL-terminated buffer, with no
indication of the original length.
For the most part, this doesn't matter. People do not put
NULs in their commit messages, and the log code is happy to
treat it all as a NUL-terminated string. However, some code
paths do care. For example, when checking signatures, we
want to be very careful that we verify all the bytes to
avoid malicious trickery.
This patch just adds an optional "size" out-pointer to
get_commit_buffer and friends. The existing callers all pass
NULL (there did not seem to be any obvious sites where we
could avoid an immediate strlen() call, though perhaps with
some further refactoring we could).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This will make it easier to manage the buffer cache
independently of the "struct commit" objects. It also
shrinks "struct commit" by one pointer, which may be
helpful.
Unfortunately it does not reduce the max memory size of
something like "rev-list", because rev-list uses
get_cached_commit_buffer() to decide not to show each
commit's output (and due to the design of slab_at, accessing
the slab requires us to extend it, allocating exactly the
same number of buffer pointers we dropped from the commit
structs).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Callers currently must use init_foo_slab() at runtime before
accessing a slab. For global slabs, it's much nicer if we
can initialize them in BSS, so that each user does not have
to add code to check-and-initialize.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Each of these sites assumes that commit->buffer is valid.
Since they would segfault if this was not the case, they are
likely to be correct in practice. However, we can
future-proof them by using get_commit_buffer.
And as a side effect, we abstract away the final bare uses
of commit->buffer.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Like the callsites in the previous commit, logmsg_reencode
already falls back to read_sha1_file when necessary.
However, I split its conversion out into its own commit
because it's a bit more complex.
We return either:
1. The original commit->buffer
2. A newly allocated buffer from read_sha1_file
3. A reencoded buffer (based on either 1 or 2 above).
while trying to do as few extra reads/allocations as
possible. Callers currently free the result with
logmsg_free, but we can simplify this by pointing them
straight to unuse_commit_buffer. This is a slight layering
violation, in that we may be passing a buffer from (3).
However, since the end result is to free() anything except
(1), which is unlikely to change, and because this makes the
interface much simpler, it's a reasonable bending of the
rules.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
For both of these sites, we already do the "fallback to
read_sha1_file" trick. But we can shorten the code by just
using get_commit_buffer.
Note that the error cases are slightly different when
read_sha1_file fails. get_commit_buffer will die() if the
object cannot be loaded, or is a non-commit.
For get_sha1_oneline, this will almost certainly never
happen, as we will have just called parse_object (and if it
does, it's probably worth complaining about).
For record_author_date, the new behavior is probably better;
we notify the user of the error instead of silently ignoring
it. And because it's used only for sorting by author-date,
somebody examining a corrupt repo can fallback to the
regular traversal order.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Some call sites check commit->buffer to see whether we have
a cached buffer, and if so, do some work with it. In the
long run we may want to switch these code paths to make
their decision on a different boolean flag (because checking
the cache may get a little more expensive in the future).
But for now, we can easily support them by converting the
calls to use get_cached_commit_buffer.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Many sites look at commit->buffer to get more detailed
information than what is in the parsed commit struct.
However, we sometimes drop commit->buffer to save memory,
in which case the caller would need to read the object
afresh. Some callers do this (leading to duplicated code),
and others do not (which opens the possibility of a segfault
if somebody else frees the buffer).
Let's provide a pair of helpers, "get" and "unuse", that let
callers easily get the buffer. They will use the cached
buffer when possible, and otherwise load from disk using
read_sha1_file.
Note that we also need to add a "get_cached" variant which
returns NULL when we do not have a cached buffer. At first
glance this seems to defeat the purpose of "get", which is
to always provide a return value. However, some log code
paths actually use the NULL-ness of commit->buffer as a
boolean flag to decide whether to try printing the
commit. At least for now, we want to continue supporting
that use.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Right now this is just a one-liner, but abstracting it will
make it easier to change later.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This converts two lines into one at each caller. But more
importantly, it abstracts the concept of freeing the buffer,
which will make it easier to change later.
Note that we also need to provide a "detach" mechanism for a
tricky case in index-pack. We are passed a buffer for the
object generated by processing the incoming pack. If we are
not using --strict, we just calculate the sha1 on that
buffer and return, leaving the caller to free it. But if we
are using --strict, we actually attach that buffer to an
object, pass the object to the fsck functions, and then
detach the buffer from the object again (so that the caller
can free it as usual). In this case, we don't want to free
the buffer ourselves, but just make sure it is no longer
associated with the commit.
Note that we are making the assumption here that the
attach/detach process does not impact the buffer at all
(e.g., it is never reallocated or modified). That holds true
now, and we have no plans to change that. However, as we
abstract the commit_buffer code, this dependency becomes
less obvious. So when we detach, let's also make sure that
we get back the same buffer that we gave to the
commit_buffer code.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Version tests only make sense when all entries are in the same file,
so we can see if version is downgraded to 2 if 3 is not required.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This could be used to run the whole test suite with split
indexes. Index splitting is carried out at random. "git read-tree"
also resets the index and forces splitting at the next update.
I had a lot of headaches with the test suite, which proves it
exercises split index pretty good.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Normally scripts do not have to be aware about split indexes because
all shared indexes are in $GIT_DIR. A simple "mv $tmp_index
$GIT_DIR/somewhere" is enough. Scripts that generate temporary indexes
and move them across repos must be aware about split index and copy
the shared file as well. This option enables that.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If $GIT_DIR is read only, we can't write $GIT_DIR/sharedindex. This
could happen when $GIT_INDEX_FILE is set to somehwere outside
$GIT_DIR.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If you have a large work tree but only make changes in a subset, then
$GIT_DIR/index's size should be stable after a while. If you change
branches that touch something else, $GIT_DIR/index's size may grow
large that it becomes as slow as the unified index. Do --split-index
again occasionally to force all changes back to the shared index and
keep $GIT_DIR/index small.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We know the positions of replaced entries via the replace bitmap in
"link" extension, so the "name" path does not have to be stored (it's
still in the shared index). With this, we also have a way to
distinguish additions vs replacements at load time and can catch
broken "link" extensions.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We are sure that after merge_base_index() is done. cache-tree can
still be used with the final index. So don't destroy cache tree.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
CE_REMOVE'd entries are removed here because only parts of the code
base (unpack_trees in fact) test this bit when they look for the
presence of an entry. Leaving them may confuse the code ignores this
bit and expects to see a real entry.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
prepare_to_write_split_index() does the major work, classifying
deleted, updated and added entries. write_link_extension() then just
writes it down.
An observation is, deleting an entry, then adding it back is recorded
as "entry X is deleted, entry X is added", not "entry X is replaced".
This is simpler, with small overhead: a replaced entry is stored
without its path, a new entry is store with its path.
A note about unpack_trees() and the deduplication code inside
prepare_to_write_split_index(). Usually tracking updated/removed
entries via read-cache API is enough. unpack_trees() manipulates the
index in a different way: it throws the entire source index out,
builds up a new one, copying/duplicating entries (using dup_entry)
from the source index over if necessary, then returns the new index.
A naive solution would be marking the entire source index "deleted"
and add their duplicates as new. That could bring $GIT_DIR/index back
to the original size. So we try harder and memcmp() between the
original and the duplicate to see if it needs updating.
We could avoid memcmp() too, by avoiding duplicating the original
entry in dup_entry(). The performance gain this way is within noise
level and it complicates unpack-trees.c. So memcmp() is the preferred
way to deal with deduplication.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The large part of this patch just follows CE_ENTRY_CHANGED
marks. replace_index_entry() is updated to update
split_index->base->cache[] as well so base->cache[] does not reference
to a freed entry.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Entries that belong to the base index should not be freed. Mark
CE_REMOVE to track them.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make sure entry addition does not lead to unifying the index. We don't
need to explicitly keep track of new entries. If ce->index is zero,
they're new. Otherwise it's unlikely that they are new, but we'll do a
thorough check later at writing time.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This split-index mode is designed to keep write cost proportional to
the number of changes the user has made, not the size of the work
tree. (Read cost is another matter, to be dealt separately.)
This mode stores index info in a pair of $GIT_DIR/index and
$GIT_DIR/sharedindex.<SHA-1>. sharedindex is large and unchanged over
time while "index" is smaller and updated often. Format details are in
index-format.txt, although not everything is implemented in this
patch.
Shared indexes are not automatically removed, because it's unclear if
the shared index is needed by any (even temporary) indexes by just
looking at it. After a while you'll collect stale shared indexes. The
good news is one shared index is useable for long, until
$GIT_DIR/index becomes too big and sluggish that the new shared index
must be created.
The safest way to clean shared indexes is to turn off split index
mode, so shared files are all garbage, delete them all, then turn on
split index mode again.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Also update SHA-1 after writing. If we do not do that, the second
read_index() will see "initialized" variable already set and not read
.git/index again, which is fine, except istate->sha1 now has a stale
value.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Other fill_stat_cache_info() is on new entries, which should set
CE_ENTRY_ADDED in cache_changed, so we're safe.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
cache entry additions, removals and modifications are separated
out. The rest of changes are still in the catch-all flag
SOMETHING_CHANGED, which would be more specific later.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
remove_marked_cache_entries() deletes entries marked with
CE_REMOVE. But if there is no such entry, do not mark the index as
"changed" because that could trigger an index update unnecessarily.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We're running out of room for in-memory flags. But since b60e188
(Strip namelen out of ce_flags into a ce_namelen field - 2012-07-11),
we copy the namelen (first 12 bits) to ce_namelen field. So those bits
are free to use. Just make sure we do not accidentally write any
in-memory flags back.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This function is now only used by write_locked_index(). Move it to
read-cache.c (because read-cache.c will need to be aware of
alternate_index_output later) and unexport it.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
b4073bb3 (git-p4: Do not include diff in spec file when just
preparing p4, 2014-05-24) broke git p4 submit, here is a proper
fix, including proper handling for windows end of lines.
Signed-off-by: Maxime Coste <frrrwww@gmail.com>
Acked-by: Pete Wyckoff <pw@padd.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since git 2.0.0 starting git gui in a submodule using a gitfile fails with
the following error:
No working directory ../../../<path>
couldn't change working directory
to "../../../<path>": no such file or
directory
This is because "git rev-parse --show-toplevel" is only run when git gui
sees a git version of at least 1.7.0 (which is the version in which the
--show-toplevel option was introduced). But "package vsatisfies" returns
false when the major version changes, which is not what we want here.
Fix that for both places where the git version is checked using vsatisfies
by appending a '-' to the version number. This tells vsatisfies that a
change of the major version is not considered to be a problem, as long as
the new major version is larger. This is done for both the place that
caused the reported bug and another spot where the git version is tested
for another feature.
Reported-by: Chris Packham <judge.packham@gmail.com>
Reported-by: Yann Dirson <ydirson@free.fr>
Signed-off-by: Jens Lehmann <Jens.Lehmann@web.de>
Currently setting submodule.<name>.ignore and/or diff.ignoreSubmodules to
"all" suppresses all output of submodule changes for git-gui. This is
really confusing, as even when the user chooses to record a new commit for
an ignored submodule by adding it manually this change won't show up under
"Staged Changes (Will Commit)".
Fix that by using the '--ignore-submodules=dirty' option for both callers
of "git diff-index --cached" when the underlying git version supports that
option.
Signed-off-by: Jens Lehmann <Jens.Lehmann@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Pat Thoyts <patthoyts@users.sourceforge.net>
We want to make sure that the default behavior of git-repack,
without any options, continues to treat .keep files as it
always has. Adding an explicit --no-pack-kept-objects, as
ee34a2b did, is a much less interesting test, and prevented
us from noticing the bug fixed by 64d3dc9 (repack: do not
accidentally pack kept objects by default, 2014-06-10).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It is not C. The code would break under mksh when 'pull.ff' is set:
$ git pull
/usr/lib/git-core/git-pull[67]: break: can't break
Already up-to-date.
Signed-off-by: Jacek Konieczny <jajcus@jajcus.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This simplifies the code, as logmsg_reencode handles the
reencoding for us in a single call. It also means we learn
logmsg_reencode's trick of pulling the buffer from disk when
commit->buffer is NULL (we currently just silently return!).
It is doubtful this matters in practice, though, as
sequencer operations would not generally turn off
save_commit_buffer.
Note that we may be fixing a bug here. The existing code
does:
if (same_encoding(to, from))
reencode_string(buf, to, from);
That probably should have been "!same_encoding".
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The return value from logmsg_reencode may be either a newly
allocated buffer or a pointer to the existing commit->buffer.
We would not want the caller to accidentally free() or
modify the latter, so let's mark it as const. We can cast
away the constness in logmsg_free, but only once we have
determined that it is a free-able buffer.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In both blame and merge-recursive, we sometimes create a
"fake" commit struct for convenience (e.g., to represent the
HEAD state as if we would commit it). By allocating
ourselves rather than using alloc_commit_node, we do not
properly set the "index" field of the commit. This can
produce subtle bugs if we then use commit-slab on the
resulting commit, as we will share the "0" index with
another commit.
We can fix this by using alloc_commit_node() to allocate.
Note that we cannot free the result, as it is part of our
commit allocator. However, both cases were already leaking
the allocated commit anyway, so there's nothing to fix up.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
