'struct split_index' contains the field 'nr_deletions', whose name
with the 'nr_' prefix suggests that it contains the number of deleted
cache entries. However, barring its initialization to 0, this field
is only ever set to 1, indicating that there is at least one deleted
entry, but not the number of deleted entries. Luckily, this doesn't
cause any issues (other than confusing the reader, that is), because
the only place reading this field uses it in the same sense, i.e.: 'if
(si->nr_deletions)'.
To avoid confusion, we could either rename this field to something
like 'has_deletions' to make its name match its role, or make it a
counter of deleted cache entries to match its name.
Let's make it a counter, to keep it in sync with the related field
'nr_replacements', which does contain the number of replaced cache
entries. This will also give developers debugging the split index
code easy access to the number of deleted cache entries.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
't1700-split-index.sh' checks that the index was split correctly under
various circumstances and that all the different ways to turn the
split index feature on and off work correctly. To do so, most of its
tests use 'test-tool dump-split-index' to see which files have their
cache entries in the split index. All these tests assume that all
cache entries are written to the shared index (called "base"
throughout these tests) when a new shared index is created. This is
an implementation detail: most git commands (basically all except 'git
update-index') don't care or know at all about split index or whether
a cache entry is stored in the split or shared index.
As demonstrated in the previous patch, refreshing a split index is
prone to a variant of the classic racy git issue. The next patch will
fix this issue, but while doing so it will also slightly change this
behaviour: only cache entries with mtime in the past will be written
only to the newly created shared index, but racily clean cache entries
will be written to the new split index (with smudged stat data).
While this upcoming change won't at all affect any git commands, it
will violate the above mentioned assumption of 't1700's tests. Since
these tests create or modify files and create or refresh the split
index in rapid succession, there are plenty of racily clean cache
entries to be dealt with, which will then be written to the new split
indexes, and, ultimately, will cause several tests in 't1700' to fail.
Let's prepare 't1700-split-index.sh' for this upcoming change and
modify its tests to avoid racily clean files by backdating the mtime
of any file modifications (and since a lot of tests create or modify
files, encapsulate it into a helper function).
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Ever since the split index feature was introduced [1], refreshing a
split index is prone to a variant of the classic racy git problem.
There are a couple of unrelated tests in the test suite that
occasionally fail when run with 'GIT_TEST_SPLIT_INDEX=yes', but
't1700-split-index.sh', the only test script focusing solely on split
index, has never noticed this issue, because it only cares about how
the index is split under various circumstances and all the different
ways to turn the split index feature on and off.
Add a dedicated test script 't1701-racy-split-index.sh' to exercise
the split index feature in racy situations as well; kind of a
"t0010-racy-git.sh for split index" but with modern style (the tests
do everything in &&-chained list of commands in 'test_expect_...'
blocks, and use 'test_cmp' for more informative output on failure).
The tests cover the following sequences of index splitting, updating,
and racy file modifications, with the last two cases demonstrating the
racy split index problem:
1. Split the index while adding a racily clean file:
echo "cached content" >file
git update-index --split-index --add file
echo "dirty worktree" >file # size stays the same
This case already works properly. Even though the cache entry's
stat data matches with the modifid file in the worktree,
subsequent git commands will notice that the (split) index and
the file have the same mtime, and then will go on to check the
file's content and notice its dirtiness.
2. Add a racily clean file to an already split index:
git update-index --split-index
echo "cached content" >file
git update-index --add file
echo "dirty worktree" >file
This case already works properly. After the second 'git
update-index' writes the newly added file's cache entry to the
new split index, it basically works in the same way as case #1.
3. Split the index when it (i.e. the not yet splitted index)
contains a racily clean cache entry, i.e. an entry whose cached
stat data matches with the corresponding file in the worktree and
the cached mtime matches that of the index:
echo "cached content" >file
git update-index --add file
echo "dirty worktree" >file
# ... wait ...
git update-index --split-index --add other-file
This case already works properly. The shared index is written by
do_write_index(), i.e. the same function that is responsible for
writing "regular" and split indexes as well. This function
cleverly notices the racily clean cache entry, and writes the
entry to the new shared index with smudged stat data, i.e. file
size set to 0. When subsequent git commands read the index, they
will notice that the smudged stat data doesn't match with the
file in the worktree, and then go on to check the file's content
and notice its dirtiness.
4. Update the split index when it contains a racily clean cache
entry:
git update-index --split-index
echo "cached content" >file
git update-index --add file
echo "dirty worktree" >file
# ... wait ...
git update-index --add other-file
This case already works properly. After the second 'git
update-index' the newly added file's cache entry is only stored
in the split index. If a cache entry is present in the split
index (even if it is a replacement of an outdated entry in the
shared index), then it will always be included in the new split
index on subsequent split index updates (until the file is
removed or a new shared index is written), independently from
whether the entry is racily clean or not. When do_write_index()
writes the new split index, it notices the racily clean cache
entry, and smudges its stat date. Subsequent git commands
reading the index will notice the smudged stat data and then go
on to check the file's content and notice its dirtiness.
5. Update the split index when a racily clean cache entry is stored
only in the shared index:
echo "cached content" >file
git update-index --split-index --add file
echo "dirty worktree" >file
# ... wait ...
git update-index --add other-file
This case fails due to the racy split index problem. In the
second 'git update-index' prepare_to_write_split_index() decides,
among other things, which cache entries stored only in the shared
index should be replaced in the new split index. Alas, this
function never looks out for racily clean cache entries, and
since the file's stat data in the worktree hasn't changed since
the shared index was written, the entry won't be replaced in the
new split index. Consequently, do_write_index() doesn't even get
this racily clean cache entry, and can't smudge its stat data.
Subsequent git commands will then see that the index has more
recent mtime than the file and that the (not smudged) cached stat
data still matches with the file in the worktree, and,
ultimately, will erroneously consider the file clean.
6. Update the split index after unpack_trees() copied a racily clean
cache entry from the shared index:
echo "cached content" >file
git update-index --split-index --add file
echo "dirty worktree" >file
# ... wait ...
git read-tree -m HEAD
This case fails due to the racy split index problem. This
basically fails for the same reason as case #5 above, but there
is one important difference, which warrants the dedicated test.
While that second 'git update-index' in case #5 updates
index_state in place, in this case 'git read-tree -m' calls
unpack_trees(), which throws out the entire index, and constructs
a new one from the (potentially updated) copies of the original's
cache entries. Consequently, when prepare_to_write_split_index()
gets to work on this reconstructed index, it takes a different
code path than in case #5 when deciding which cache entries in
the shared index should be replaced. The result is the same,
though: the racily clean cache entry goes unnoticed, it isn't
added to the split index with smudged stat data, and subsequent
git commands will then erroneously consider the file clean.
Note that in the last two 'test_expect_failure' cases I omitted the
'#' (as in nr. of trial) from the tests' description on purpose for
now, as it breakes the TAP output [2]; it will be added at the end of
the series, when those two tests will be flipped to
'test_expect_success'.
[1] In the branch leading to the merge commit v2.1.0-rc0~45 (Merge
branch 'nd/split-index', 2014-07-16).
[2] In the TAP output a '#' should separate the test's description
from the TODO directive emitted by 'test_expect_failure'. The
additional '#' in "#$trial" interferes with this, the test harness
won't recognize the TODO directive, and will report that those
tests failed unexpectedly.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The earlier attempt barfed when given a CONTENT_LENGTH that is
set to an empty string. RFC 3875 is fairly clear that in this
case we should not read any message body, but we've been reading
through to the EOF in previous versions (which did not even pay
attention to the environment variable), so keep that behaviour for
now in this late update.
* mk/http-backend-content-length:
http-backend: allow empty CONTENT_LENGTH
This reverts commit 7e25437d35, reversing
changes made to 00624d608c.
v2.19.0-rc0~165^2~1 (submodule: ensure core.worktree is set after
update, 2018-06-18) assumes an "absorbed" submodule layout, where the
submodule's Git directory is in the superproject's .git/modules/
directory and .git in the submodule worktree is a .git file pointing
there. In particular, it uses $GIT_DIR/modules/$name to find the
submodule to find out whether it already has core.worktree set, and it
uses connect_work_tree_and_git_dir if not, resulting in
fatal: could not open sub/.git for writing
The context behind that patch: v2.19.0-rc0~165^2~2 (submodule: unset
core.worktree if no working tree is present, 2018-06-12) unsets
core.worktree when running commands like "git checkout
--recurse-submodules" to switch to a branch without the submodule. If
a user then uses "git checkout --no-recurse-submodules" to switch back
to a branch with the submodule and runs "git submodule update", this
patch is needed to ensure that commands using the submodule directly
are aware of the path to the worktree.
It is late in the release cycle, so revert the whole 3-patch series.
We can try again later for 2.20.
Reported-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Helped-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
According to RFC3875, empty environment variable is equivalent to unset,
and for CONTENT_LENGTH it should mean zero body to read.
However, unset CONTENT_LENGTH is also used for chunked encoding to indicate
reading until EOF. At least, the test "large fetch-pack requests can be split
across POSTs" from t5551 starts faliing, if unset or empty CONTENT_LENGTH is
treated as zero length body. So keep the existing behavior as much as possible.
Add a test for the case.
Reported-By: Jelmer Vernooij <jelmer@jelmer.uk>
Signed-off-by: Max Kirillov <max@max630.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"cette" can be only be used before a word (like in "cette bouteille" for
"this bottle"), but here "this" refers to the current step and we have
to use "ceci" in French.
Signed-off-by: Raphaël Hertzog <hertzog@debian.org>
The test linter code has learned that the end of here-doc mark
"EOF" can be quoted in a double-quote pair, not just in a
single-quote pair.
* es/chain-lint-more:
chainlint: match "quoted" here-doc tags
Portability fix.
* ab/portable-more:
tests: fix non-portable iconv invocation
tests: fix non-portable "${var:-"str"}" construct
tests: fix and add lint for non-portable grep --file
tests: fix version-specific portability issue in Perl JSON
tests: use shorter labels in chainlint.sed for AIX sed
tests: fix comment syntax in chainlint.sed for AIX sed
tests: fix and add lint for non-portable seq
tests: fix and add lint for non-portable head -c N
"git merge-base" in 2.19-rc1 has performance regression when the
(experimental) commit-graph feature is in use, which has been
mitigated.
* ds/commit-graph-lockfile-fix:
commit: don't use generation numbers if not needed
Recent addition of "directory rename" heuristics to the
merge-recursive backend makes the command susceptible to false
positives and false negatives. In the context of "git am -3",
which does not know about surrounding unmodified paths and thus
cannot inform the merge machinery about the full trees involved,
this risk is particularly severe. As such, the heuristic is
disabled for "git am -3" to keep the machinery "more stupid but
predictable".
* en/directory-renames-nothanks:
am: avoid directory rename detection when calling recursive merge machinery
merge-recursive: add ability to turn off directory rename detection
t3401: add another directory rename testcase for rebase and am
Recent "git rebase -i" update started to write bogusly formatted
author-script, with a matching broken reading code. These are
fixed.
* pw/rebase-i-author-script-fix:
sequencer: fix quoting in write_author_script
sequencer: handle errors from read_author_ident()
OLD_ICONV has long been needed by FreeBSD so config.mak.uname defines
it unconditionally. However, recent versions do not need it, and its
presence results in compilation warnings. Resolve this issue by defining
OLD_ICONV only for older FreeBSD versions.
Specifically, revision r281550[1], which is part of FreeBSD 11, removed
the need for OLD_ICONV, and r282275[2] back-ported that change to 10.2.
Versions prior to 10.2 do need it.
[1] b0813ee288
[2] b709ec868a
[es: commit message; tweak version check to distinguish 10.x versions]
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In 3afc679b "commit: use generations in paint_down_to_common()",
the queue in paint_down_to_common() was changed to use a priority
order based on generation number before commit date. This served
two purposes:
1. When generation numbers are present, the walk guarantees
correct topological relationships, regardless of clock skew in
commit dates.
2. It enables short-circuiting the walk when the min_generation
parameter is added in d7c1ec3e "commit: add short-circuit to
paint_down_to_common()". This short-circuit helps commands
like 'git branch --contains' from needing to walk to a merge
base when we know the result is false.
The commit message for 3afc679b includes the following sentence:
This change does not affect the number of commits that are
walked during the execution of paint_down_to_common(), only
the order that those commits are inspected.
This statement is incorrect. Because it changes the order in which
the commits are inspected, it changes the order they are added to
the queue, and hence can change the number of loops before the
queue_has_nonstale() method returns true.
This change makes a concrete difference depending on the topology
of the commit graph. For instance, computing the merge-base between
consecutive versions of the Linux kernel has no effect for versions
after v4.9, but 'git merge-base v4.8 v4.9' presents a performance
regression:
v2.18.0: 0.122s
v2.19.0-rc1: 0.547s
HEAD: 0.127s
To determine that this was simply an ordering issue, I inserted
a counter within the while loop of paint_down_to_common() and
found that the loop runs 167,468 times in v2.18.0 and 635,579
times in v2.19.0-rc1.
The topology of this case can be described in a simplified way
here:
v4.9
| \
| \
v4.8 \
| \ \
| \ |
... A B
| / /
| / /
|/__/
C
Here, the "..." means "a very long line of commits". By generation
number, A and B have generation one more than C. However, A and B
have commit date higher than most of the commits reachable from
v4.8. When the walk reaches v4.8, we realize that it has PARENT1
and PARENT2 flags, so everything it can reach is marked as STALE,
including A. B has only the PARENT1 flag, so is not STALE.
When paint_down_to_common() is run using
compare_commits_by_commit_date, A and B are removed from the queue
early and C is inserted into the queue. At this point, C and the
rest of the queue entries are marked as STALE. The loop then
terminates.
When paint_down_to_common() is run using
compare_commits_by_gen_then_commit_date, B is removed from the
queue only after the many commits reachable from v4.8 are explored.
This causes the loop to run longer. The reason for this regression
is simple: the queue order is intended to not explore a commit
until everything that _could_ reach that commit is explored. From
the information gathered by the original ordering, we have no
guarantee that there is not a commit D reachable from v4.8 that
can also reach B. We gained absolute correctness in exchange for
a performance regression.
The performance regression is probably the worse option, since
these incorrect results in paint_down_to_common() are rare. The
topology required for the performance regression are less rare,
but still require multiple merge commits where the parents differ
greatly in generation number. In our example above, the commit A
is as important as the commit B to demonstrate the problem, since
otherwise the commit C will sit in the queue as non-stale just as
long in both orders.
The solution provided uses the min_generation parameter to decide
if we should use generation numbers in our ordering. When
min_generation is equal to zero, it means that the caller has no
known cutoff for the walk, so we should rely on our commit-date
heuristic as before; this is the case with merge_bases_many().
When min_generation is non-zero, then the caller knows a valuable
cutoff for the short-circuit mechanism; this is the case with
remove_redundant() and in_merge_bases_many().
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Let's say you have the following three trees, where Base is from one commit
behind either master or branch:
Base : bar_v1, foo/{file1, file2, file3}
branch: bar_v2, foo/{file1, file2}, goo/file3
master: bar_v3, foo/{file1, file2, file3}
Using git-am (or am-based rebase) to apply the changes from branch onto
master results in the following tree:
Result: bar_merged, goo/{file1, file2, file3}
This is not what users want; they did not rename foo/ -> goo/, they only
renamed one file within that directory. The reason this happens is am
constructs fake trees (via build_fake_ancestor()) of the following form:
Base_bfa : bar_v1, foo/file3
branch_bfa: bar_v2, goo/file3
Combining these two trees with master's tree:
master: bar_v3, foo/{file1, file2, file3},
You can see that merge_recursive_generic() would see branch_bfa as renaming
foo/ -> goo/, and master as just adding both foo/file1 and foo/file2. As
such, it ends up with goo/{file1, file2, file3}
The core problem is that am does not have access to the original trees; it
can only construct trees using the blobs involved in the patch. As such,
it is not safe to perform directory rename detection within am -3.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Similar to commit 16346883ab ("t3401: add directory rename testcases for
rebase and am", 2018-06-27), add another testcase for directory rename
detection. This new testcase differs in that it showcases a situation
where no directory rename was performed, but which some backends
incorrectly detect.
As with the other testcase, run this in conjunction with each of the
types of rebases:
git-rebase--interactive
git-rebase--am
git-rebase--merge
and also use the same testcase for
git am --3way
Reported-by: Nikolay Kasyanov <corrmage@gmail.com>
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A here-doc tag can be quoted ('EOF'/"EOF") or escaped (\EOF) to suppress
interpolation within the body. chainlint recognizes single-quoted and
escaped tags, but does not know about double-quoted tags. For
completeness, teach it to recognize double-quoted tags, as well.
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The iconv that comes with a FreeBSD 11.2-RELEASE-p2 box I have access
to doesn't support the SHIFT-JIS encoding. Guard a test added in
e92d62253 ("convert: add round trip check based on
'core.checkRoundtripEncoding'", 2018-04-15) first released with Git
v2.18.0 with a prerequisite that checks for its availability.
The iconv command is in POSIX, and we have numerous tests
unconditionally relying on its ability to convert ASCII, UTF-8 and
UTF-16, but unconditionally relying on the presence of more obscure
encodings isn't portable.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
On both AIX 7200-00-01-1543 and FreeBSD 11.2-RELEASE-p2 the
"${var:-"str"}" syntax means something different than what it does
under the bash or dash shells.
Both will consider the start of the new unescaped quotes to be a new
argument to test_expect_success, resulting in the following error:
error: bug in the test script: 'git diff-tree initial # magic
is (not' does not look like a prereq
Fix this by removing the redundant quotes. There's no need for them,
and the resulting code works under all the aforementioned shells. This
fixes a regression in c2f1d3989 ("t4013: test new output from diff
--abbrev --raw", 2017-12-03) first released with Git v2.16.0.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Partially revert the support for multiple hash functions to regain
hash comparison performance; we'd think of a way to do this better
in the next cycle.
* jk/hashcmp-optim-for-2.19:
hashcmp: assert constant hash size
