[PATCH v2] mingw: align symlinks-related rmdir() behavior with Linux
flat view
DORMANTno replies
From: Johannes Schindelin via GitGitGadget <hidden>
Date: 2021-08-02 21:07:36
Subsystem:
the rest · Maintainer:
Linus Torvalds
From: =?UTF-8?q?Thomas=20B=C3=A9tous?= <redacted>
When performing a rebase, rmdir() is called on the folder .git/logs. On
Unix rmdir() exits without deleting anything in case .git/logs is a
symbolic link but the equivalent functions on Windows (_rmdir, _wrmdir
and RemoveDirectoryW) do not behave the same and remove the folder if it
is symlinked even if it is not empty.
This creates issues when folders in .git/ are symlinks which is
especially the case when git-repo[1] is used: It replaces `.git/logs/`
with a symlink.
One such issue is that the _target_ of that symlink is removed e.g.
during a `git rebase`, where `delete_reflog("REBASE_HEAD")` will not
only try to remove `.git/logs/REBASE_HEAD` but then recursively try to
remove the parent directories until an error occurs, a technique that
obviously relies on `rmdir()` refusing to remove a symlink.
This was reported in https://github.com/git-for-windows/git/issues/2967.
This commit updates mingw_rmdir() so that its behavior is the same as
Linux rmdir() in case of symbolic links.
To verify that Git does not regress on the reported issue, this patch
adds a regression test for the `git rebase` symptom, even if the same
`rmdir()` behavior is quite likely to cause potential problems in other
Git commands as well.
[1]: git-repo is a python tool built on top of Git which helps manage
many Git repositories. It stores all the .git/ folders in a central
place by taking advantage of symbolic links.
More information: https://gerrit.googlesource.com/git-repo/
Signed-off-by: Thomas Bétous <redacted>
Signed-off-by: Johannes Schindelin <redacted>
---
mingw: support the git-repo tool better
This addresses an issue, originally reported at
https://github.com/git-for-windows/git/issues/2967, where the git-repo
tool [https://gerrit.googlesource.com/git-repo/] replaces folders in
.git/ with symlinks and mingw_rmdir() erroneously removes the symlink
target directory's contents.
Changes since v1:
* Fixed the authorship
* Augmented the commit message to elaborate on what the issues are,
concretely, with the current behavior of mingw_rmdir()
* Added an explanation to the commit message as to what happens during
a git rebase that would trigger the rmdir() code path to misbehave
* Adjusted the code comment to talk specifically about the
remove_path() caller and why we want to align with Linux' rmdir()
behavior
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1008%2Fdscho%2Ffix-rmdir-with-symlinks-on-windows-v2
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1008/dscho/fix-rmdir-with-symlinks-on-windows-v2
Pull-Request: https://github.com/gitgitgadget/git/pull/1008
Range-diff vs v1:
1: c5a7fe0007f ! 1: 0a9cb1bcc14 mingw: align symlinks-related rmdir() behavior with Linux
@@
## Metadata ##
-Author: Johannes Schindelin [off-list ref]
+Author: Thomas Bétous [off-list ref]
## Commit message ##
mingw: align symlinks-related rmdir() behavior with Linux
@@ Commit message
is symlinked even if it is not empty.
This creates issues when folders in .git/ are symlinks which is
- especially the case when git-repo[1] is used.
+ especially the case when git-repo[1] is used: It replaces `.git/logs/`
+ with a symlink.
+
+ One such issue is that the _target_ of that symlink is removed e.g.
+ during a `git rebase`, where `delete_reflog("REBASE_HEAD")` will not
+ only try to remove `.git/logs/REBASE_HEAD` but then recursively try to
+ remove the parent directories until an error occurs, a technique that
+ obviously relies on `rmdir()` refusing to remove a symlink.
+
+ This was reported in https://github.com/git-for-windows/git/issues/2967.
This commit updates mingw_rmdir() so that its behavior is the same as
Linux rmdir() in case of symbolic links.
- This fixes https://github.com/git-for-windows/git/issues/2967
+ To verify that Git does not regress on the reported issue, this patch
+ adds a regression test for the `git rebase` symptom, even if the same
+ `rmdir()` behavior is quite likely to cause potential problems in other
+ Git commands as well.
[1]: git-repo is a python tool built on top of Git which helps manage
many Git repositories. It stores all the .git/ folders in a central
@@ compat/mingw.c: int mingw_rmdir(const char *pathname)
+ struct stat st;
+
+ /*
-+ * Contrary to Linux rmdir(), Windows' _wrmdir() and _rmdir()
-+ * will remove the directory at the path if it is a symbolic link
-+ * which leads to issues when symlinks are used in the .git folder
-+ * (in the context of git-repo for instance). So before calling _wrmdir()
-+ * we first check if the path is a symbolic link. If it is, we exit
-+ * and return the same error as Linux rmdir() in this case (ENOTDIR).
-+ */
++ * Contrary to Linux' `rmdir()`, Windows' _wrmdir() and _rmdir()
++ * (and `RemoveDirectoryW()`) will attempt to remove the target of a
++ * symbolic link (if it points to a directory).
++ *
++ * This behavior breaks the assumption of e.g. `remove_path()` which
++ * upon successful deletion of a file will attempt to remove its parent
++ * directories recursively until failure (which usually happens when
++ * the directory is not empty).
++ *
++ * Therefore, before calling `_wrmdir()`, we first check if the path is
++ * a symbolic link. If it is, we exit and return the same error as
++ * Linux' `rmdir()` would, i.e. `ENOTDIR`.
++ */
+ if (!mingw_lstat(pathname, &st) && S_ISLNK(st.st_mode)) {
+ errno = ENOTDIR;
+ return -1;
compat/mingw.c | 21 +++++++++++++++++++++
t/t3400-rebase.sh | 10 ++++++++++
t/test-lib.sh | 6 ++++++
3 files changed, 37 insertions(+)
diff --git a/compat/mingw.c b/compat/mingw.c
index aa647b367b0..9e0cd1e097f 100644
--- a/compat/mingw.c
+++ b/compat/mingw.c@@ -341,6 +341,27 @@ int mingw_rmdir(const char *pathname) { int ret, tries = 0; wchar_t wpathname[MAX_PATH]; + struct stat st; + + /* + * Contrary to Linux' `rmdir()`, Windows' _wrmdir() and _rmdir() + * (and `RemoveDirectoryW()`) will attempt to remove the target of a + * symbolic link (if it points to a directory). + * + * This behavior breaks the assumption of e.g. `remove_path()` which + * upon successful deletion of a file will attempt to remove its parent + * directories recursively until failure (which usually happens when + * the directory is not empty). + * + * Therefore, before calling `_wrmdir()`, we first check if the path is + * a symbolic link. If it is, we exit and return the same error as + * Linux' `rmdir()` would, i.e. `ENOTDIR`. + */ + if (!mingw_lstat(pathname, &st) && S_ISLNK(st.st_mode)) { + errno = ENOTDIR; + return -1; + } + if (xutftowcs_path(wpathname, pathname) < 0) return -1;
diff --git a/t/t3400-rebase.sh b/t/t3400-rebase.sh
index 0bb88aa982b..23dbd3c82ed 100755
--- a/t/t3400-rebase.sh
+++ b/t/t3400-rebase.sh@@ -406,4 +406,14 @@ test_expect_success 'refuse to switch to branch checked out elsewhere' ' test_i18ngrep "already checked out" err ' +test_expect_success MINGW,SYMLINKS_WINDOWS 'rebase when .git/logs is a symlink' ' + git checkout main && + mv .git/logs actual_logs && + cmd //c "mklink /D .git\logs ..\actual_logs" && + git rebase -f HEAD^ && + test -L .git/logs && + rm .git/logs && + mv actual_logs .git/logs +' + test_done
diff --git a/t/test-lib.sh b/t/test-lib.sh
index adaf03543e8..73f6d645b66 100644
--- a/t/test-lib.sh
+++ b/t/test-lib.sh@@ -1513,6 +1513,12 @@ test_lazy_prereq SYMLINKS ' ln -s x y && test -h y ' +test_lazy_prereq SYMLINKS_WINDOWS ' + # test whether symbolic links are enabled on Windows + test_have_prereq MINGW && + cmd //c "mklink y x" &> /dev/null && test -h y +' + test_lazy_prereq FILEMODE ' test "$(git config --bool core.filemode)" = true '
base-commit: ebf3c04b262aa27fbb97f8a0156c2347fecafafb -- gitgitgadget