The "bitmapPseudoMerge.*.sampleRate" configuration controls what
fraction of unstable commits are included in each pseudo-merge group.
The config validation accepts values in the range `[0, 1]`, but a value
of exactly 0 causes a division by zero in `select_pseudo_merges_1()`:
if (j % (uint32_t)(1.0 / group->sample_rate))
When `sample_rate` is 0, `1.0 / 0.0` produces `+inf`, and casting
infinity to `uint32_t` is undefined behavior in C. On most platforms
this yields 0, making the subsequent modulo operation (`j % 0`) a
fatal arithmetic trap.
This path was not previously reachable because an earlier bug caused
all pseudo-merge candidates to be classified as "stable" (where the
sampling rate is not used), regardless of their actual commit date. Now
that the date classification is fixed, the unstable path is exercised
and the division by zero can fire.
Fix this by changing the validation to require a strict lower bound and
thus reject 0.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
`find_pseudo_merge_group_for_ref()` uses the commit's date to classify
it as either "stable" (older than the stable threshold) or "unstable"
(otherwise).
However, to find the relevant commit from a given OID, the function
`find_pseudo_merge_group_for_ref()` uses `lookup_commit()` which does
not parse commits.
Because an unparsed commit has its "date" set to zero, every candidate
is placed in the "stable" bucket regardless of its actual committer
timestamp. This means the `bitmapPseudoMerge.*.threshold` and
`stableThreshold` configuration options have no effect: the
stable/unstable split is always determined by comparing against zero
rather than the real commit date.
The net result is that pseudo-merge groups are partitioned by
`stableSize` instead of the intended decay-based sizing, and the
`sampleRate` knob (which only applies to the unstable path) is never
exercised.
Fix this by calling `repo_parse_commit()` after `lookup_commit()`,
bailing out of the callback if parsing fails.
The corresponding test configures two pseudo-merge groups that both
match all tags. The "stable" group uses `threshold=1.month.ago`, and the
"all" group uses `threshold=now`. The test use our custom
"GIT_TEST_DATE_NOW" environment variable by setting it to the value of
"$test_tick" to align Git's notion of "now" (and therefore
"1.month.ago") with the `test_tick` timestamps, so the commits appear to
be younger than one month: only the "all" group matches them, producing
exactly one pseudo-merge.
Without the fix every commit has `date == 0`, which satisfies `date <=
threshold` for both groups (since 0 is older than one month ago), and
the "stable" group erroneously matches as well.
Now that commits are correctly classified as "unstable", the bug
described in the test exercising the "sampleRate=0" test is reachable,
and the test is marked as failing. It will be fixed in a following
commit.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When a commit appears in more than one pseudo-merge group, its entry in
the commit lookup table has the high bit set in its offset field,
indicating that the offset points to an "extended" table containing the
set of pseudo-merges for that commit.
There are three bugs in this path:
* The `next_ext` offset in `write_pseudo_merges()` undercounts the
per-entry size of the lookup table (8 vs. 12 bytes).
* `nth_pseudo_merge_ext()` calls `read_pseudo_merge_commit_at()` on a
pseudo-merge bitmap offset, misinterpreting it as a 12-byte commit
table entry.
* The error check after `pseudo_merge_ext_at()` in
`apply_pseudo_merges_for_commit()` tests `< -1` instead of `< 0`,
silently swallowing errors from `error()`.
The first bug is on the write side: each commit lookup entry contains a
4- and 8-byte unsigned value for a total of 12 bytes, but the
calculation assumes that the entry only contains 8 bytes of data. This
makes `next_ext` too small, so the extended-table offsets that get
written point into the middle of the non-extended lookup table rather
than past it. The reader then interprets non-extended lookup data as
extended entries, producing garbage.
The second bug is on the read side and is independently fatal: even with
a correctly positioned extended table, `nth_pseudo_merge_ext()` feeds
the offset it reads (which points at pseudo-merge bitmap data) to
`read_pseudo_merge_commit_at()`. That function tries to parse 12 bytes
as a `pseudo_merge_commit` struct, clobbering `merge->pseudo_merge_ofs`
with whatever happens to be at that location. The caller only needs
`pseudo_merge_ofs`, so the fix is to store the offset directly rather
than re-parsing a commit table entry. The `commit_pos` field is left
untouched, retaining the value that `find_pseudo_merge()` set earlier.
The third bug is latent. With the first two fixes applied, the extended
table is correctly written and read, so `pseudo_merge_ext_at()` does not
fail during normal operation. The `< -1` vs `< 0` distinction only
matters when the bitmap file is corrupt or truncated, in which case the
error would be silently ignored and the code would proceed with
uninitialized data.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The binary search in `pseudo_merge_at()` has its "lo" and "hi" updates
swapped: when the midpoint's offset is less than the target, it sets `hi
= mi` (searching left) instead of `lo = mi + 1` (searching right), and
vice versa.
This means that lookups for pseudo-merges whose offset is not near the
midpoint of the pseudo-merge table are likely to fail.
In practice, with a single pseudo-merge group this is masked because the
lone entry is always at the midpoint. With multiple groups, the inverted
comparisons cause lookups to search in the wrong direction, potentially
missing entries.
Swap the "lo" and "hi" assignments to search in the correct direction,
making it possible to apply pseudo-merges during fill-in when more than
one pseudo-merge exists in a group.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The pseudo-merge commit lookup table stores each commit's position in
the pack- or pseudo-pack order, and is used to perform a binary search
in order to determine which pseudo-merge(s) a given commit belongs to.
However, the table was previously sorted in lexical order (via
`oid_array_sort()`), causing the binary search to fail.
While this causes pseudo-merge bitmaps to be de-facto broken for fill-in
traversal, there are a couple of important points to keep in mind:
* Pseudo-merge application during the initial phases of a bitmap-based
traversal are applied via `cascade_pseudo_merges_1()`. This function
enumerates the known pseudo-merges and determines if its parents are
a subset of the traversal roots.
This is a different path than the fill-in traversal, where we are
looking for any pseudo-merges which may be satisfied after visiting
some commit along an object walk, which involves the aforementioned
(broken) binary search.
As a consequence, any pseudo-merges we apply at this stage are done
so correctly.
* While this bug makes applying pseudo-merges during fill-in traversal
effectively broken, it does not produce wrong results. Instead of
applying the *wrong* pseudo-merge, we will simply fail to find
satisfied pseudo-merges, leaving the traversal to use the existing
fill-in routines.
Fix this by sorting the table by bit position before writing, matching
the order that the reader's binary search expects.
This does produce a change the on-disk format insofar as the actual code
now complies with the documented format (for more details, refer to:
Documentation/technical/bitmap-format.adoc). Given that this never
worked in the first place, such a change should be OK to perform.
If an out-of-tree implementation of pseudo-merges happened to generate
bitmaps that comply with the documented format, they will continue to be
read and interpreted as normal.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Using the test helper introduced via the previous commit, add various
failing tests demonstrating bugs in the pseudo-merge implementation.
These are all marked as failing with one exception. The "sampleRate=0"
test describes a latent bug, which is only reachable through a code path
that is itself masked by a separate bug. A future commit will fix that
bug, and, in turn, cause the aforementioned test to fail. Accordingly,
that commit will mark the test as failing, and it will be re-marked as
passing in a separate commit which fixes the once-latent bug.
For the rest: the following commits will explain and fix the underlying
bugs in detail.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In f16eb1c091 (pseudo-merge: fix disk reads from find_pseudo_merge(),
2026-03-31), we noted that `apply_pseudo_merges_for_commit()` is never
triggered by the existing test suite, and that this bears further
investigation.
This patch is the first one to begin that investigation. The following
patches will expose and fix a variety of bugs in the implementation of
pseudo-merge bitmaps.
In order to do so, however, many of these tests require very precise
selection of which commits receive bitmaps and which do not. To date,
there isn't a standard approach to easily facilitate this. Address this
by introducing a `test-tool bitmap write` subcommand that writes a
bitmap for a given packfile, reading the set of commits which should
receive individual bitmaps from stdin like so:
test-tool bitmap write <pack-basename> </path/to/commits.list
, where "<pack-basename>" is the filename for a specific packfile (e.g.,
"pack-abc123.pack"), and "/path/to/commits.list" is a list of commit
OIDs which will receive bitmaps.
The helper respects `bitmapPseudoMerge.*` configuration for creating
pseudo-merge bitmaps alongside the regular commit bitmaps.
Signed-off-by: Taylor Blau <me@ttaylorr.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We used writev() in limited code paths and supplied emulation for
platforms without working writev(), but the emulation was too
faithful to the spec to make the result useless to send even 64kB;
revert the topic and plan to restart the effort later.
* jc/no-writev-does-not-work:
Revert "compat/posix: introduce writev(3p) wrapper"
Revert "wrapper: introduce writev(3p) wrappers"
Revert "sideband: use writev(3p) to send pktlines"
Revert "cmake: use writev(3p) wrapper as needed"
Earlier we timelined that we'd tune our build procedures to build
with Rust by default in Git 2.53, but we are already in prerelease
freeze for 2.54 now. Update the BreakingChanges document to delay
it until Git 2.55 (slated for the end of June 2026).
Noticed-by: brian m. carlson <sandals@crustytoothpaste.net>
Helped-by: Derrick Stolee <stolee@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The emulation layer we added for writev(3p) tries to be too faithful
to the spec that on systems with SSIZE_MAX set to lower than 64kB to
fit a single sideband packet would fail just like the real system
writev(), which makes our use of writev() for sideband messages
unworkable.
Let's revert them and reboot the effort after the release. The
reverted commits are:
$ git log -Swritev --oneline 8023abc632^..v2.52.0-rc1
89152af176 cmake: use writev(3p) wrapper as needed
26986f4cba sideband: use writev(3p) to send pktlines
1970fcef93 wrapper: introduce writev(3p) wrappers
3b9b2c2a29 compat/posix: introduce writev(3p) wrapper
8023abc632 is the merge of ps/upload-pack-buffer-more-writes topic to
the mainline.
Helped-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git rev-list --maximal-only" has been optimized by borrowing the
logic used by "git show-branch --independent", which computes the
same kind of information much more efficiently.
* ds/rev-list-maximal-only-optim:
rev-list: use reduce_heads() for --maximal-only
p6011: add perf test for rev-list --maximal-only
t6600: test --maximal-only and --independent
"git config list" is the official way to spell "git config -l" and
"git config --list". Use it to update the documentation.
* kh/doc-config-list:
doc: gitcvs-migration: rephrase “man page”
doc: replace git config --list/-l with `list`
Further work to adjust the codebase for C23 that changes functions
like strchr() that discarded constness when they return a pointer into
a const string to preserve constness.
* jk/c23-const-preserving-fixes-more:
git-compat-util: fix CONST_OUTPARAM typo and indentation
refs/files-backend: drop const to fix strchr() warning
http: drop const to fix strstr() warning
range-diff: drop const to fix strstr() warnings
pkt-line: make packet_reader.line non-const
skip_prefix(): check const match between in and out params
pseudo-merge: fix disk reads from find_pseudo_merge()
find_last_dir_sep(): convert inline function to macro
run-command: explicitly cast away constness when assigning to void
pager: explicitly cast away strchr() constness
transport-helper: drop const to fix strchr() warnings
http: add const to fix strchr() warnings
convert: add const to fix strchr() warnings
Clarify that --prefix is used as given and is not normalized,
and may include leading slashes or parent directory components.
Signed-off-by: Pushkar Singh <pushkarkumarsingh1970@gmail.com>
Acked-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In commit ec0becacc9 (run-command: add stdin callback for
parallelization, 2026-01-28), we taught run_processes_parallel() to
ignore SIGPIPE, since we wouldn't want a write() to a broken pipe of one
of the children to take down the whole process.
But there's a subtle ordering issue. After we ignore SIGPIPE, we call
pp_init(), which installs its own cleanup handler for multiple signals
using sigchain_push_common(), which includes SIGPIPE. So if we receive
SIGPIPE while writing to a child, we'll trigger that handler first, pop
it off the stack, and then re-raise (which is then ignored because of
the SIG_IGN we pushed first).
But what does that handler do? It tries to clean up all of the child
processes, under the assumption that when we re-raise the signal we'll
be exiting the process!
So a hook that exits without reading all of its input will cause us to
get SIGPIPE, which will put us in a signal handler that then tries to
kill() that same child.
This seems to be mostly harmless on Linux. The process has already
exited by this point, and though kill() does not complain (since the
process has not been reaped with a wait() call), it does not affect the
exit status of the process.
However, this seems not to be true on all platforms. This case is
triggered by t5401.13, "pre-receive hook that forgets to read its
input". This test fails on NonStop since that hook was converted to the
run_processes_parallel() API.
We can fix it by reordering the code a bit. We should run pp_init()
first, and then push our SIG_IGN onto the stack afterwards, so that it
is truly ignored while feeding the sub-processes.
Note that we also reorder the popping at the end of the function, too.
This is not technically necessary, as we are doing two pops either way,
but now the pops will correctly match their pushes.
This also fixes a related case that we can't test yet. If we did have
more than one process to run, then one child causing SIGPIPE would cause
us to kill() all of the children (which might still actually be
running). But the hook API is the only user of the new feed_pipe
feature, and it does not yet support parallel hook execution. So for now
we'll always execute the processes sequentially. Once parallel hook
execution exists, we'll be able to add a test which covers this.
Reported-by: Randall S. Becker <rsbecker@nexbridge.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
During Git 2.52 timeframe, we broke streaming computation of object
hash outside a repository, which has been corrected.
* jt/index-fd-wo-repo-regression-fix:
During Git 2.52 timeframe, we broke streaming computation of object
hash outside a repository, which has been corrected.
* jt/index-fd-wo-repo-regression-fix-maint:
object-file: avoid ODB transaction when not writing objects
The experimental `git replay` command learned the `--ref=<ref>` option
to allow specifying which ref to update, overriding the default behavior.
* tc/replay-ref:
replay: allow to specify a ref with option --ref
replay: use stuck form in documentation and help message
builtin/replay: mark options as not negatable
add_files_to_cache() used diff_files() to detect only the paths that
are different between the index and the working tree and add them,
which does not need rename detection, which interfered with unnecessary
conflicts.
* ng/add-files-to-cache-wo-rename:
read-cache: disable renames in add_files_to_cache
Update reftable library part with what is used in libgit2 to improve
portability to different target codebases and platforms.
* ps/reftable-portability:
reftable/system: add abstraction to mmap files
reftable/system: add abstraction to retrieve time in milliseconds
reftable/fsck: use REFTABLE_UNUSED instead of UNUSED
reftable/stack: provide fsync(3p) via system header
reftable: introduce "reftable-system.h" header
Various code clean-up around odb subsystem.
* ps/odb-cleanup:
odb: drop unneeded headers and forward decls
odb: rename `odb_has_object()` flags
odb: use enum for `odb_write_object` flags
odb: rename `odb_write_object()` flags
treewide: use enum for `odb_for_each_object()` flags
CodingGuidelines: document our style for flags
Add the missing &&'s so we properly propagate failures
between commands in the hook helper functions.
Also add a missing mkdir -p arg (found by adding the &&).
Reported-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Our description of the reftable format is that it is experimental and
subject to change, but that is no longer true. Remove this statement so
as not to mislead users.
In addition, the documentation says that the files format is the
default, but that is not true if breaking changes mode is on. Correct
this information with a conditional.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In ce1661f9da (odb: add transaction interface, 2025-09-16), existing
ODB transaction logic is adapted to create a transaction interface
at the ODB layer. The intent here is for the ODB transaction
interface to eventually provide an object source agnostic means to
manage transactions.
An unintended consequence of this change though is that
`object-file.c:index_fd()` may enter the ODB transaction path even
when no object write is requested. In non-repository contexts, this
can result in a NULL dereference and segfault. One such case occurs
when running git-diff(1) outside of a repository with
"core.bigFileThreshold" forcing the streaming path in `index_fd()`:
$ echo foo >foo
$ echo bar >bar
$ git -c core.bigFileThreshold=1 diff -- foo bar
In this scenario, the caller only needs to compute the object ID. Object
hashing does not require an ODB, so starting a transaction is both
unnecessary and invalid.
Fix the bug by avoiding the use of ODB transactions in `index_fd()` when
callers are only interested in computing the object hash.
Reported-by: Luca Stefani <luca.stefani.ge1@gmail.com>
Signed-off-by: Justin Tobler <jltobler@gmail.com>
[jc: adjusted to fd13909e (Merge branch 'jt/odb-transaction', 2025-10-02)]
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git backfill" is capable of auto-detecting a sparsely checked out
working tree, which was broken.
* th/backfill-auto-detect-sparseness-fix:
backfill: auto-detect sparse-checkout from config
The check in "receive-pack" to prevent a checked out branch from
getting updated via updateInstead mechanism has been corrected.
* ps/receive-pack-updateinstead-in-worktree:
receive-pack: use worktree HEAD for updateInstead
t5516: clean up cloned and new-wt in denyCurrentBranch and worktrees test
t5516: test updateInstead with worktree and unborn bare HEAD
Handling of signed commits and tags in fast-import has been made more
configurable.
* jt/fast-import-signed-modes:
fast-import: add 'abort-if-invalid' mode to '--signed-tags=<mode>'
fast-import: add 'sign-if-invalid' mode to '--signed-tags=<mode>'
fast-import: add 'strip-if-invalid' mode to '--signed-tags=<mode>'
fast-import: add 'abort-if-invalid' mode to '--signed-commits=<mode>'
fast-export: check for unsupported signing modes earlier
The way the "git log -L<range>:<file>" feature is bolted onto the
log/diff machinery is being reworked a bit to make the feature
compatible with more diff options, like -S/G.
* mm/line-log-use-standard-diff-output:
doc: note that -L supports patch formatting and pickaxe options
t4211: add tests for -L with standard diff options
line-log: route -L output through the standard diff pipeline
line-log: fix crash when combined with pickaxe options
Reduce dependency on `the_repository` in add-patch.c file.
* sp/add-patch-with-fewer-the-repository:
add-patch: use repository instance from add_i_state instead of the_repository
Internals of "git fsck" have been refactored to not depend on the
global `the_repository` variable.
* ps/fsck-wo-the-repository:
builtin/fsck: stop using `the_repository` in error reporting
builtin/fsck: stop using `the_repository` when marking objects
builtin/fsck: stop using `the_repository` when checking packed objects
builtin/fsck: stop using `the_repository` with loose objects
builtin/fsck: stop using `the_repository` when checking reflogs
builtin/fsck: stop using `the_repository` when checking refs
builtin/fsck: stop using `the_repository` when snapshotting refs
builtin/fsck: fix trivial dependence on `the_repository`
fsck: drop USE_THE_REPOSITORY
fsck: store repository in fsck options
fsck: initialize fsck options via a function
fetch-pack: move fsck options into function scope
The value of a wrong pointer variable was referenced in an error
message that reported that it shouldn't be NULL.
* yc/path-walk-fix-error-reporting:
path-walk: fix NULL pointer dereference in error message
Fix a regression in writing the commit-graph where commits with dates
exceeding 34 bits (beyond year 2514) could cause an underflow and
crash Git during the generation data overflow chunk writing.
* ps/commit-graph-overflow-fix:
commit-graph: fix writing generations with dates exceeding 34 bits
A handful of inappropriate uses of the_repository have been
rewritten to use the right repository structure instance in the
read-cache.c codepath.
* jd/read-cache-trace-wo-the-repository:
read-cache: use istate->repo for trace2 logging
Adjust the codebase for C23 that changes functions like strchr()
that discarded constness when they return a pointer into a const
string to preserve constness.
* jk/c23-const-preserving-fixes:
config: store allocated string in non-const pointer
rev-parse: avoid writing to const string for parent marks
revision: avoid writing to const string for parent marks
rev-parse: simplify dotdot parsing
revision: make handle_dotdot() interface less confusing
Taylor Blau