git/worktree.h, branch main

Merge branch 'pw/worktree-reduce-the-repository'

2026-04-03T20:01:09Z

Reduce the reference to the_repository in the worktree subsystem. * pw/worktree-reduce-the-repository: worktree: reject NULL worktree in get_worktree_git_dir() worktree add: stop reading ".git/HEAD" worktree: remove "the_repository" from is_current_worktree()

worktree: reject NULL worktree in get_worktree_git_dir()

2026-03-26T15:20:50Z

This removes the final dependence on "the_repository" in get_worktree_git_dir(). The last commit removed only caller that passed a NULL worktree. get_worktree_git_dir() has the following callers: - branch.c:prepare_checked_out_branches() which loops over all worktrees. - builtin/fsck.c:cmd_fsck() which loops over all worktrees. - builtin/receive-pack.c:update_worktree() which is called from update() only when "worktree" is non-NULL. - builtin/worktree.c:validate_no_submodules() which is called from check_clean_worktree() and move_worktree(), both of which supply a non-NULL worktree. - reachable.c:add_rebase_files() which loops over all worktrees. - revision.c:add_index_objects_to_pending() which loops over all worktrees. - worktree.c:is_current_worktree() which expects a non-NULL worktree. Signed-off-by: Phillip Wood Signed-off-by: Junio C Hamano

worktree: remove "the_repository" from is_current_worktree()

2026-03-26T15:20:50Z

The "is_current" member of struct worktree was added in 750e8a60d69 (worktree.c: mark current worktree, 2016-04-22) and was used in 8d9fdd7087d (worktree.c: check whether branch is rebased in another worktree, 2016-04-22) to optionally skip the current worktree when seeing if a branch is already checked out in die_if_checked_out(). To determine if a worktree is "current" is_current_worktree() compares the gitdir of the worktree to the gitdir of "the_repository" and returns true when they match. To get the gitdir of the worktree it calls get_workree_git_dir() which also depends on "the_repository". This means that even if "wt->path" matches "wt->repo->worktree" is_current_worktree(wt) will return false when "wt->repo" is not "the_repository". Consequently die_if_checked_out() will fail to skip such a worktree when checking if a branch is already checked out and may die errounously. Fix this by using the worktree's repository instance instead of "the_repository" when comparing gitdirs. The use of "the_repository" in is_current_wortree() comes from replacing get_git_dir() with repo_get_git_dir() in 246deeac951 (environment: make `get_git_dir()` accept a repository, 2024-09-12). In get_worktree_git_dir() it comes from replacing git_common_path() with repo_common_path() in 07242c2a5af (path: drop `git_common_path()` in favor of `repo_common_path()`, 2025-02-07). In both cases the replacements appear to have been mechanical. Signed-off-by: Phillip Wood Signed-off-by: Junio C Hamano

Merge branch 'dd/list-objects-filter-options-wo-strbuf-split'

2026-03-19T16:54:55Z

The way combined list-object filter options are parsed has been revamped. * dd/list-objects-filter-options-wo-strbuf-split: list-objects-filter-options: avoid strbuf_split_str() worktree: do not pass strbuf by value

worktree: do not pass strbuf by value

2026-03-11T18:08:53Z

write_worktree_linking_files() takes two struct strbuf parameters by value, even though it only reads path strings from them. Passing a strbuf by value is misleading and dangerous. The structure carries a pointer to its underlying character array; caller and callee end up sharing that storage. If the callee ever causes the strbuf to be reallocated, the caller's copy becomes a dangling pointer, which results in a double-free when the caller does strbuf_release(). The function only needs the string values, not the strbuf machinery. Switch it to take const char * and update all callers to pass .buf. Signed-off-by: Deveshi Dwivedi Signed-off-by: Junio C Hamano

Merge branch 'ps/refs-for-each'

2026-03-09T21:36:55Z

Code refactoring around refs-for-each-* API functions. * ps/refs-for-each: refs: replace `refs_for_each_fullref_in()` refs: replace `refs_for_each_namespaced_ref()` refs: replace `refs_for_each_glob_ref()` refs: replace `refs_for_each_glob_ref_in()` refs: replace `refs_for_each_rawref_in()` refs: replace `refs_for_each_rawref()` refs: replace `refs_for_each_ref_in()` refs: improve verification for-each-ref options refs: generalize `refs_for_each_fullref_in_prefixes()` refs: generalize `refs_for_each_namespaced_ref()` refs: speed up `refs_for_each_glob_ref_in()` refs: introduce `refs_for_each_ref_ext` refs: rename `each_ref_fn` refs: rename `do_for_each_ref_flags` refs: move `do_for_each_ref_flags` further up refs: move `refs_head_ref_namespaced()` refs: remove unused `refs_for_each_include_root_ref()`

refs: rename `each_ref_fn`

2026-02-23T21:21:18Z

Similar to the preceding commit, rename `each_ref_fn` to better match our current best practices around how we name things. Signed-off-by: Patrick Steinhardt Signed-off-by: Junio C Hamano

wt-status: avoid passing NULL worktree

2026-02-19T19:03:23Z

In preparation for removing the repository argument from worktree_git_path() add a function to construct a "struct worktree" from a "struct repository" using its "gitdir" and "worktree" members. This function is then used to avoid passing a NULL worktree to wt_status_check_bisect() and wt_status_check_rebase(). In general the "struct worktree" returned may not correspond to the "current" worktree defined by is_current_worktree() as that function uses "the_repository" rather than "wt->repo" when deciding which worktree is "current". In practice the "struct repository" we pass corresponds to "the_repository" as we only ever operate on a single repository at the moment. wt_status_check_bisect() and wt_status_check_rebase() have the following callers: - branch.c:prepare_checked_out_branches() which loops over all worktrees. - worktree.c:is_worktree_being_rebased() which is called from builtin/branch.c:reject_rebase_or_bisect_branch() that loops over all worktrees and worktree.c:is_shared_symref() which dereferences wt earlier in the function. - wt-status:wt_status_get_state() which is updated to avoid passing a NULL worktree by this patch. This updates the only callers that pass a NULL worktree to worktree_git_path(). A new test is added to check that "git status" detects a rebase in a linked worktree. Signed-off-by: Phillip Wood Signed-off-by: Junio C Hamano

Merge branch 'sj/ref-consistency-checks-more'

2025-03-26T07:26:10Z

"git fsck" becomes more careful when checking the refs. * sj/ref-consistency-checks-more: builtin/fsck: add `git refs verify` child process packed-backend: check whether the "packed-refs" is sorted packed-backend: add "packed-refs" entry consistency check packed-backend: check whether the refname contains NUL characters packed-backend: add "packed-refs" header consistency check packed-backend: check if header starts with "# pack-refs with: " packed-backend: check whether the "packed-refs" is regular file builtin/refs: get worktrees without reading head information t0602: use subshell to ensure working directory unchanged

builtin/refs: get worktrees without reading head information

2025-02-27T22:03:07Z

In "packed-backend.c", there are some functions such as "create_snapshot" and "next_record" which would check the correctness of the content of the "packed-ref" file. When anything is bad, the program will die. It may seem that we have nothing relevant to above feature, because we are going to read and parse the raw "packed-ref" file without creating the snapshot and using the ref iterator to check the consistency. However, when using "get_worktrees" in "builtin/refs", we would parse the "HEAD" information. If the referent of the "HEAD" is inside the "packed-ref", we will call "create_snapshot" function to parse the "packed-ref" to get the information. No matter whether the entry of "HEAD" in "packed-ref" is correct, "create_snapshot" would call "verify_buffer_safe" to check whether there is a newline in the last line of the file. If not, the program will die. Although this behavior has no harm for the program, it will short-circuit the program. When the users execute "git refs verify" or "git fsck", we should avoid reading the head information, which may execute the read operation in packed backend with stricter checks to die the program. Instead, we should continue to check other parts of the "packed-refs" file completely. Fortunately, in 465a22b338 (worktree: skip reading HEAD when repairing worktrees, 2023-12-29), we have introduced a function "get_worktrees_internal" which allows us to get worktrees without reading head information. Create a new exposed function "get_worktrees_without_reading_head", then replace the "get_worktrees" in "builtin/refs" with the new created function. Mentored-by: Patrick Steinhardt Mentored-by: Karthik Nayak Signed-off-by: shejialuo Signed-off-by: Junio C Hamano