From: "Maik Riechert via GitGitGadget" <gitgitgadget@gmail.com>
To: git@vger.kernel.org
Cc: Ilia K <ki.stfu@gmail.com>, Derrick Stolee <stolee@gmail.com>,
Junio C Hamano <gitster@pobox.com>,
Maik Riechert <letmaik@outlook.com>,
Maik Riechert <letmaik@outlook.com>
Subject: [PATCH] dir: invalidate parents of stale cached directories
Date: Fri, 20 Nov 2026 05:33:03 +0000 [thread overview]
Message-ID: <pull.2321.git.1795152783221.gitgitgadget@gmail.com> (raw)
From: Maik Riechert <letmaik@outlook.com>
After a failed fsmonitor query or a trivial response, "git status"
can report both an untracked directory and a directory below it,
for example "?? three/" and "?? three/deep/".
1cf98cb4d9 (fsmonitor: check the untracked cache after a trivial
response, 2026-11-09) added a stat check before writing the untracked
cache alongside a new fsmonitor token after these responses. The check
invalidates stale directories, but does not propagate the invalidation
to their parents.
In normal mode, a parent's cached untracked list can contain "three/"
from a check-only scan of the child. Writing the invalid child clears
its check_only flag. The next cached walk then recurses into the child
and reports "three/deep/", while the parent's list still reports
"three/".
Invalidate ancestors too when DIR_SHOW_OTHER_DIRECTORIES is set, as
invalidate_one_component() does. Leave all-files mode unchanged and
reuse the existing traversal without adding stat calls.
Add normal-mode regression tests for failed queries and trivial
responses. Use fixed directory mtimes and force an index write without
scanning untracked files.
Signed-off-by: Maik Riechert <letmaik@outlook.com>
---
dir: invalidate parents of stale cached directories
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-2321%2Fletmaik%2Funtracked-cache-stale-dirs-parents-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-2321/letmaik/untracked-cache-stale-dirs-parents-v1
Pull-Request: https://github.com/gitgitgadget/git/pull/2321
dir.c | 6 +++++
t/t7519-status-fsmonitor.sh | 46 +++++++++++++++++++++++++++++++++++++
2 files changed, 52 insertions(+)
diff --git a/dir.c b/dir.c
index 4f748800b1..633d259d7a 100644
--- a/dir.c
+++ b/dir.c
@@ -4094,6 +4094,12 @@ static int invalidate_stale_dirs(struct untracked_cache *uc,
strbuf_setlen(path, len);
}
+ if (nr_invalidated && ucd->valid &&
+ (uc->dir_flags & DIR_SHOW_OTHER_DIRECTORIES)) {
+ invalidate_one_directory(uc, ucd);
+ nr_invalidated++;
+ }
+
return nr_invalidated;
}
diff --git a/t/t7519-status-fsmonitor.sh b/t/t7519-status-fsmonitor.sh
index 27c3194ae0..3877140b77 100755
--- a/t/t7519-status-fsmonitor.sh
+++ b/t/t7519-status-fsmonitor.sh
@@ -388,6 +388,52 @@ test_expect_success UNTRACKED_CACHE 'untracked cache is checked after a trivial
)
'
+for response in trivial failed
+do
+ test_expect_success UNTRACKED_CACHE,!MINGW "stale check-only directories invalidate parents after a $response response" '
+ test_when_finished "rm -rf stale-$response" &&
+ git init stale-$response &&
+ (
+ cd stale-$response &&
+ mkdir -p parent/three/deep &&
+ >parent/keep &&
+ >parent/three/deep/tracked &&
+ git add parent &&
+ git commit -m initial &&
+ git rm --cached parent/three/deep/tracked &&
+ git config core.untrackedCache true &&
+ git config status.showUntrackedFiles normal &&
+ git config core.fsmonitor "$TEST_DIRECTORY/t7519/fsmonitor-none" &&
+ git config core.fsmonitorHookVersion 1 &&
+ test-tool chmtime =1000000000 . parent parent/three parent/three/deep &&
+ cat >../expect <<-\EOF &&
+ D parent/three/deep/tracked
+ ?? parent/three/
+ EOF
+ git status --porcelain >../actual &&
+ test_cmp ../expect ../actual &&
+
+ # Change only the check-only directory after caching it.
+ # Write the index without scanning untracked files.
+ test-tool chmtime +60 parent/three &&
+ if test "$response" = trivial
+ then
+ hook_version=1
+ else
+ # fsmonitor-all fails for version 2.
+ hook_version=2
+ fi &&
+ git -c core.fsmonitor="$TEST_DIRECTORY/t7519/fsmonitor-all" \
+ -c core.fsmonitorHookVersion=$hook_version \
+ update-index --force-write-index &&
+ git status --porcelain >../actual &&
+ test_cmp ../expect ../actual &&
+ git status --porcelain >../actual &&
+ test_cmp ../expect ../actual
+ )
+ '
+done
+
# test that splitting the index doesn't interfere
test_expect_success 'splitting the index results in the same state' '
write_integration_script &&
base-commit: f7de81456940dd6f3274e461149bba12965de609
--
gitgitgadget
reply other threads:[~2026-11-20 5:33 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=pull.2321.git.1795152783221.gitgitgadget@gmail.com \
--to=gitgitgadget@gmail.com \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=ki.stfu@gmail.com \
--cc=letmaik@outlook.com \
--cc=stolee@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox