From: "Maik Riechert via GitGitGadget" <gitgitgadget@gmail.com>
To: git@vger.kernel.org
Cc: Josip Sokcevic <sokcevic@google.com>,
Maik Riechert <letmaik@outlook.com>,
Maik Riechert <letmaik@outlook.com>
Subject: [PATCH v2] diff-lib: skip leading path check for fsmonitor-valid entries
Date: Thu, 19 Nov 2026 05:43:56 +0000 [thread overview]
Message-ID: <pull.2318.v2.git.1795067037014.gitgitgadget@gmail.com> (raw)
In-Reply-To: <pull.2318.git.1795026168007.gitgitgadget@gmail.com>
From: Maik Riechert <letmaik@outlook.com>
Since 4f3d6d0261 (fsmonitor: skip lstat deletion check during git
diff-index, 2021-03-17), check_removed() does not lstat() entries
that fsmonitor says are unchanged. But it still calls
has_symlink_leading_path() for them, which lstat()s the directories
leading up to each path.
That check is there to notice a leading directory that was replaced
by a symlink. fsmonitor would have reported such a change, and
refresh_fsmonitor() would have cleared CE_FSMONITOR_VALID for every
entry below it, so we would not get here. preload_index() already
trusts the flag this way and skips these entries altogether.
The S_ISDIR() check that follows cannot trigger either: the mode
comes from the index, and gitlinks are never marked valid.
So return as soon as fake_lstat() has filled in the stat data.
The first new test checks the shortcut itself by replacing a parent
directory with a symlink without telling fsmonitor. The other two make
sure we still notice when fsmonitor does report that directory, with
and without a trailing slash, since nothing else checks for it now.
On Linux, with 100,000 files (ten per directory), a hook that
reports nothing and a warm cache, "git diff HEAD" goes from 57ms to
26ms.
Signed-off-by: Maik Riechert <letmaik@outlook.com>
---
diff-lib: skip leading path check for fsmonitor-valid entries
Changes since v1:
* Dropped the stat_err variable and called lstat() directly in the if()
condition (suggested by Junio).
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-2318%2Fletmaik%2Fdiff-lib-fsmonitor-leading-path-v2
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-2318/letmaik/diff-lib-fsmonitor-leading-path-v2
Pull-Request: https://github.com/gitgitgadget/git/pull/2318
Range-diff vs v1:
1: 22f99f7189 ! 1: 4f8cbd869d diff-lib: skip leading path check for fsmonitor-valid entries
@@ Commit message
Signed-off-by: Maik Riechert <letmaik@outlook.com>
## diff-lib.c ##
-@@ diff-lib.c: static int check_removed(const struct cache_entry *ce, struct stat *st)
+@@
+ */
+ static int check_removed(const struct cache_entry *ce, struct stat *st)
{
- int stat_err;
+- int stat_err;
++ if (ce->ce_flags & CE_FSMONITOR_VALID)
++ return fake_lstat(ce, st);
- if (!(ce->ce_flags & CE_FSMONITOR_VALID))
- stat_err = lstat(ce->name, st);
- else
- stat_err = fake_lstat(ce, st);
-+ if (ce->ce_flags & CE_FSMONITOR_VALID)
-+ return fake_lstat(ce, st);
-+
-+ stat_err = lstat(ce->name, st);
- if (stat_err < 0) {
+- if (stat_err < 0) {
++ if (lstat(ce->name, st) < 0) {
if (!is_missing_file_error(errno))
return -1;
+ return 1;
## t/t7519-status-fsmonitor.sh ##
@@ t/t7519-status-fsmonitor.sh: do
diff-lib.c | 9 +++-----
t/t7519-status-fsmonitor.sh | 44 +++++++++++++++++++++++++++++++++++++
2 files changed, 47 insertions(+), 6 deletions(-)
diff --git a/diff-lib.c b/diff-lib.c
index 086476bd77..b644154397 100644
--- a/diff-lib.c
+++ b/diff-lib.c
@@ -41,13 +41,10 @@
*/
static int check_removed(const struct cache_entry *ce, struct stat *st)
{
- int stat_err;
+ if (ce->ce_flags & CE_FSMONITOR_VALID)
+ return fake_lstat(ce, st);
- if (!(ce->ce_flags & CE_FSMONITOR_VALID))
- stat_err = lstat(ce->name, st);
- else
- stat_err = fake_lstat(ce, st);
- if (stat_err < 0) {
+ if (lstat(ce->name, st) < 0) {
if (!is_missing_file_error(errno))
return -1;
return 1;
diff --git a/t/t7519-status-fsmonitor.sh b/t/t7519-status-fsmonitor.sh
index 93973ed25a..6c9df9f333 100755
--- a/t/t7519-status-fsmonitor.sh
+++ b/t/t7519-status-fsmonitor.sh
@@ -336,6 +336,50 @@ do
done
done
+# Create a repository whose only entry, dir1/tracked, is marked valid
+# by an fsmonitor hook that reports no changes, then replace dir1 with
+# a symlink to a directory that has the same contents.
+setup_symlinked_leading_dir () {
+ test_when_finished "rm -rf repo" &&
+ git init repo &&
+ mkdir repo/dir1 &&
+ : >repo/dir1/tracked &&
+ git -C repo add dir1/tracked &&
+ git -C repo commit -m initial &&
+ test_hook -C repo --setup fsmonitor-test <<-\EOF &&
+ printf "last_update_token\0"
+ EOF
+ git -C repo config core.fsmonitor .git/hooks/fsmonitor-test &&
+ git -C repo update-index --refresh --force-write-index &&
+ echo "h dir1/tracked" >expect &&
+ git -C repo ls-files -f >actual &&
+ test_cmp expect actual &&
+ mv repo/dir1 repo/dir2 &&
+ ln -s dir2 repo/dir1
+}
+
+# Like "status" above, diff-index should trust fsmonitor and not look
+# for symlinks in leading directories that it was not told about.
+test_expect_success SYMLINKS "diff-index doesn't detect unreported symlinked directories" '
+ setup_symlinked_leading_dir &&
+ git -C repo diff-index --name-status HEAD >actual &&
+ test_must_be_empty actual
+'
+
+for report in dir1 dir1/
+do
+ test_expect_success SYMLINKS "diff-index detects symlinked directory reported as $report" '
+ setup_symlinked_leading_dir &&
+ test_hook -C repo --clobber fsmonitor-test <<-EOF &&
+ printf "last_update_token\0"
+ printf "$report\0"
+ EOF
+ printf "D\tdir1/tracked\n" >expect &&
+ git -C repo diff-index --name-status HEAD >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: e1ec09b23c8e0deb21a0b95ac5996422277395d4
--
gitgitgadget
next prev parent reply other threads:[~2026-11-19 5:43 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-18 18:22 [PATCH] diff-lib: skip leading path check for fsmonitor-valid entries Maik Riechert via GitGitGadget
2026-11-18 19:01 ` Junio C Hamano
2026-11-19 5:43 ` Maik Riechert via GitGitGadget [this message]
2026-11-19 22:06 ` [PATCH v2] " Junio C Hamano
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.2318.v2.git.1795067037014.gitgitgadget@gmail.com \
--to=gitgitgadget@gmail.com \
--cc=git@vger.kernel.org \
--cc=letmaik@outlook.com \
--cc=sokcevic@google.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.