git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[PATCH] am --abort: merge ORIG_HEAD tree into index

From
Paul Tan <pyokagan@gmail.com>
Date
Aug 17, 2015, 09:48 UTC
Message-ID
<20150817094819.GA10375@yoshi.chippynet.com>
In-Reply-To
<03631611149f05dbcd862b4c1e8e9d6b@www.dscho.org>
On Mon, Aug 17, 2015 at 10:01:29AM +0200, Johannes Schindelin wrote:
Show 15 quoted lines
> Hi Linus,
> 
> On 2015-08-17 01:33, Linus Torvalds wrote:
> > On Sun, Aug 16, 2015 at 12:46 PM, Linus Torvalds
> > <torvalds@linux-foundation.org> wrote:
> >>
> >> Maybe it has always done this, and I just haven't noticed (I usually
> >> _just_ do the "git reset --hard" thing, don't ask me why I wanted to
> >> be doubly sure this time). But maybe it's an effect of the new
> >> built-in "am".
> > 
> > I bisected this. It's definitely used to work, and the regression is
> > from the new built-in am.
> 
> This patch is a reproducer:
Thanks Johannes for the test, and sorry all for the oversight. ><

It's true that we need to merge the ORIG_HEAD tree into the index instead of overwriting it. Patch below.

Regards, Paul

-- >8 --
Subject: [PATCH] am --abort: merge ORIG_HEAD tree into index

After running "git am --abort", and then running "git reset --hard", files that were not modified would still be re-checked out.

This is because clean_index() in builtin/am.c mistakenly called the read_tree() function, which overwrites all entries in the index, including the stat info.

Fix this by using unpack_trees() instead to merge the tree into the index, so that the stat info from the index is kept.

Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Paul Tan <pyokagan@gmail.com>
---
 builtin/am.c        | 49 ++++++++++++++++++++++++++++++++++++-------------
 t/t4151-am-abort.sh | 12 ++++++++++++
 2 files changed, 48 insertions(+), 13 deletions(-)
diff --git a/builtin/am.c b/builtin/am.c
index 1399c8d..6aaa85d 100644
--- a/builtin/am.c
+++ b/builtin/am.c
@@ -1940,15 +1940,48 @@ static int fast_forward_to(struct tree *head, struct tree *remote, int reset)
 }
 
 /**
+ * Merges a tree into the index. The index's stat info will take precedence
+ * over the merged tree's. Returns 0 on success, -1 on failure.
+ */
+static int merge_tree(struct tree *tree)
+{
+	struct lock_file *lock_file;
+	struct unpack_trees_options opts;
+	struct tree_desc t[2];
+
+	if (parse_tree(tree))
+		return -1;
+
+	lock_file = xcalloc(1, sizeof(struct lock_file));
+	hold_locked_index(lock_file, 1);
+
+	memset(&opts, 0, sizeof(opts));
+	opts.head_idx = 1;
+	opts.src_index = &the_index;
+	opts.dst_index = &the_index;
+	opts.merge = 1;
+	opts.fn = oneway_merge;
+	init_tree_desc(&t[0], tree->buffer, tree->size);
+
+	if (unpack_trees(1, t, &opts)) {
+		rollback_lock_file(lock_file);
+		return -1;
+	}
+
+	if (write_locked_index(&the_index, lock_file, COMMIT_LOCK))
+		die(_("unable to write new index file"));
+
+	return 0;
+}
+
+/**
  * Clean the index without touching entries that are not modified between
  * `head` and `remote`.
  */
 static int clean_index(const unsigned char *head, const unsigned char *remote)
 {
-	struct lock_file *lock_file;
 	struct tree *head_tree, *remote_tree, *index_tree;
 	unsigned char index[GIT_SHA1_RAWSZ];
-	struct pathspec pathspec;
 
 	head_tree = parse_tree_indirect(head);
 	if (!head_tree)
@@ -1973,18 +2006,8 @@ static int clean_index(const unsigned char *head, const unsigned char *remote)
 	if (fast_forward_to(index_tree, remote_tree, 0))
 		return -1;
 
-	memset(&pathspec, 0, sizeof(pathspec));
-
-	lock_file = xcalloc(1, sizeof(struct lock_file));
-	hold_locked_index(lock_file, 1);
-
-	if (read_tree(remote_tree, 0, &pathspec)) {
-		rollback_lock_file(lock_file);
+	if (merge_tree(remote_tree))
 		return -1;
-	}
-
-	if (write_locked_index(&the_index, lock_file, COMMIT_LOCK))
-		die(_("unable to write new index file"));
 
 	remove_branch_state();
 
diff --git a/t/t4151-am-abort.sh b/t/t4151-am-abort.sh
index 05bdc3e..9c3bbd1 100755
--- a/t/t4151-am-abort.sh
+++ b/t/t4151-am-abort.sh
@@ -168,4 +168,16 @@ test_expect_success 'am --abort on unborn branch will keep local commits intact'
 	test_cmp expect actual
 '
 
+test_expect_success 'am --abort leaves index stat info alone' '
+	git checkout -f --orphan stat-info &&
+	git reset &&
+	test_commit should-be-untouched &&
+	test-chmtime =0 should-be-untouched.t &&
+	git update-index --refresh &&
+	git diff-files --exit-code --quiet &&
+	test_must_fail git am 0001-*.patch &&
+	git am --abort &&
+	git diff-files --exit-code --quiet
+'
+
 test_done
-- 
2.5.0.331.g11c07ce
Previous: Johannes SchindelinNext: Johannes Schindelin
Message 4 of 9 in “"git am --abort" screwing up index?”
  1. Linus TorvaldsAug 16, 2015
  2. Linus TorvaldsAug 16, 2015
  3. Johannes SchindelinAug 17, 2015
  4. am --abort: merge ORIG_HEAD tree into indexPaul Tan, Aug 17, 2015
  5. Johannes SchindelinAug 17, 2015
  6. Linus TorvaldsAug 17, 2015
  7. Junio C HamanoAug 17, 2015
  8. am --skip/--abort: merge HEAD/ORIG_HEAD tree into indexPaul Tan, Aug 19, 2015
  9. Junio C HamanoAug 19, 2015

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.