[v3,05/11] commit-reach: start reporting errors in `paint_down_to_common()`

Commit Message

Johannes Schindelin Feb. 27, 2024, 1:28 p.m. UTC

From: Johannes Schindelin <[e-mail]>

If a commit cannot be parsed, it is currently ignored when looking for
merge bases. That's undesirable as the operation can pretend success in
a corrupt repository, even though the command should fail with an error
message.

Let's start at the bottom of the stack by teaching the
`paint_down_to_common()` function to return an `int`: if negative, it
indicates fatal error, if 0 success.

This requires a couple of callers to be adjusted accordingly.

Signed-off-by: Johannes Schindelin <[e-mail]>
---
 commit-reach.c | 66 ++++++++++++++++++++++++++++++++++----------------
 1 file changed, 45 insertions(+), 21 deletions(-)

Comments

Dirk Gouders Feb. 27, 2024, 2:56 p.m. UTC | #1

"Johannes Schindelin via GitGitGadget" <[e-mail]> writes:

> Let's start at the bottom of the stack by teaching the
> `paint_down_to_common()` function to return an `int`: if negative, it
> indicates fatal error, if 0 success.

Kind of pedantic but the above doesn't describe the real change, i.e. a
value != 0 indicates a fatal error:

> -		common = paint_down_to_common(r, array[i], filled,
> -					      work, min_generation, 0);
> +		if (paint_down_to_common(r, array[i], filled,
> +					 work, min_generation, 0, &common)) {
> +			clear_commit_marks(array[i], all_flags);
> +			clear_commit_marks_many(filled, work, all_flags);
> +			free_commit_list(common);
> +			free(work);
> +			free(redundant);
> +			free(filled_index);
> +			return -1;
> +		}

Dirk

Johannes Schindelin Feb. 27, 2024, 3:08 p.m. UTC | #2

Hi Dirk,

On Tue, 27 Feb 2024, Dirk Gouders wrote:

> "Johannes Schindelin via GitGitGadget" <[e-mail]> writes:
>
> > Let's start at the bottom of the stack by teaching the
> > `paint_down_to_common()` function to return an `int`: if negative, it
> > indicates fatal error, if 0 success.
>
> Kind of pedantic but the above doesn't describe the real change, i.e. a
> value != 0 indicates a fatal error:
>
> > -		common = paint_down_to_common(r, array[i], filled,
> > -					      work, min_generation, 0);
> > +		if (paint_down_to_common(r, array[i], filled,
> > +					 work, min_generation, 0, &common)) {

The fact that we do not bother to verify that the return value is
negative, but only check for a non-zero one instead, does not change the
fact that in the form this patch leaves the code, `paint_down_to_common()`
returns -1 for fatal errors and 0 for success, as advertised, though.

Is it lazy to omit the `< 0` here? Not actually, the reason why I omitted
it here was to stay under 80 columns per line.

Good eyes, though. If `paint_down_to_common()` _did_ return values other
than -1 and 0, in particular positive ones that would not indicate a fatal
error, the hunk under discussion would have introduced a problematic bug.

Ciao,
Johannes

> > +			clear_commit_marks(array[i], all_flags);
> > +			clear_commit_marks_many(filled, work, all_flags);
> > +			free_commit_list(common);
> > +			free(work);
> > +			free(redundant);
> > +			free(filled_index);
> > +			return -1;
> > +		}
>
> Dirk
>
>

Junio C Hamano Feb. 27, 2024, 6:24 p.m. UTC | #3

Johannes Schindelin <[e-mail]> writes:

> Is it lazy to omit the `< 0` here? Not actually, the reason why I omitted
> it here was to stay under 80 columns per line.
>
> Good eyes, though. If `paint_down_to_common()` _did_ return values other
> than -1 and 0, in particular positive ones that would not indicate a fatal
> error, the hunk under discussion would have introduced a problematic bug.

The same patch does compare the returned value with '< 0' in another
function (that is far from this place), which probably made the hunk
stand out during a review, I suspect.

After having fixed a bug elsewhere about a codepath that mixed the
"non-zero is an error" and "negative is an error" conventions during
the last cycle, I would have to say that being consistent would be
nice.  I think we at the end decided to make the callee be more
strict (returning negative to signal an error) while allowing the
caller to be more lenient (taking non-zero as an error) in that
case.

Thanks.

Patch

diff --git a/commit-reach.c b/commit-reach.c
index 7112b10eeea..9148a7dcbc0 100644
--- a/commit-reach.c
+++ b/commit-reach.c
@@ -50,14 +50,14 @@  static int queue_has_nonstale(struct prio_queue *queue)
 }
 
 /* all input commits in one and twos[] must have been parsed! */
-static struct commit_list *paint_down_to_common(struct repository *r,
-						struct commit *one, int n,
-						struct commit **twos,
-						timestamp_t min_generation,
-						int ignore_missing_commits)
+static int paint_down_to_common(struct repository *r,
+				struct commit *one, int n,
+				struct commit **twos,
+				timestamp_t min_generation,
+				int ignore_missing_commits,
+				struct commit_list **result)
 {
 	struct prio_queue queue = { compare_commits_by_gen_then_commit_date };
-	struct commit_list *result = NULL;
 	int i;
 	timestamp_t last_gen = GENERATION_NUMBER_INFINITY;
 
@@ -66,8 +66,8 @@  static struct commit_list *paint_down_to_common(struct repository *r,
 
 	one->object.flags |= PARENT1;
 	if (!n) {
-		commit_list_append(one, &result);
-		return result;
+		commit_list_append(one, result);
+		return 0;
 	}
 	prio_queue_put(&queue, one);
 
@@ -95,7 +95,7 @@  static struct commit_list *paint_down_to_common(struct repository *r,
 		if (flags == (PARENT1 | PARENT2)) {
 			if (!(commit->object.flags & RESULT)) {
 				commit->object.flags |= RESULT;
-				commit_list_insert_by_date(commit, &result);
+				commit_list_insert_by_date(commit, result);
 			}
 			/* Mark parents of a found merge stale */
 			flags |= STALE;
@@ -108,7 +108,8 @@  static struct commit_list *paint_down_to_common(struct repository *r,
 				continue;
 			if (repo_parse_commit(r, p)) {
 				clear_prio_queue(&queue);
-				free_commit_list(result);
+				free_commit_list(*result);
+				*result = NULL;
 				/*
 				 * At this stage, we know that the commit is
 				 * missing: `repo_parse_commit()` uses
@@ -116,7 +117,10 @@  static struct commit_list *paint_down_to_common(struct repository *r,
 				 * corrupt commits would already have been
 				 * dispatched with a `die()`.
 				 */
-				return NULL;
+				if (ignore_missing_commits)
+					return 0;
+				return error(_("could not parse commit %s"),
+					     oid_to_hex(&p->object.oid));
 			}
 			p->object.flags |= flags;
 			prio_queue_put(&queue, p);
@@ -124,7 +128,7 @@  static struct commit_list *paint_down_to_common(struct repository *r,
 	}
 
 	clear_prio_queue(&queue);
-	return result;
+	return 0;
 }
 
 static struct commit_list *merge_bases_many(struct repository *r,
@@ -151,7 +155,10 @@  static struct commit_list *merge_bases_many(struct repository *r,
 			return NULL;
 	}
 
-	list = paint_down_to_common(r, one, n, twos, 0, 0);
+	if (paint_down_to_common(r, one, n, twos, 0, 0, &list) < 0) {
+		free_commit_list(list);
+		return NULL;
+	}
 
 	while (list) {
 		struct commit *commit = pop_commit(&list);
@@ -205,7 +212,7 @@  static int remove_redundant_no_gen(struct repository *r,
 	for (i = 0; i < cnt; i++)
 		repo_parse_commit(r, array[i]);
 	for (i = 0; i < cnt; i++) {
-		struct commit_list *common;
+		struct commit_list *common = NULL;
 		timestamp_t min_generation = commit_graph_generation(array[i]);
 
 		if (redundant[i])
@@ -221,8 +228,16 @@  static int remove_redundant_no_gen(struct repository *r,
 			if (curr_generation < min_generation)
 				min_generation = curr_generation;
 		}
-		common = paint_down_to_common(r, array[i], filled,
-					      work, min_generation, 0);
+		if (paint_down_to_common(r, array[i], filled,
+					 work, min_generation, 0, &common)) {
+			clear_commit_marks(array[i], all_flags);
+			clear_commit_marks_many(filled, work, all_flags);
+			free_commit_list(common);
+			free(work);
+			free(redundant);
+			free(filled_index);
+			return -1;
+		}
 		if (array[i]->object.flags & PARENT2)
 			redundant[i] = 1;
 		for (j = 0; j < filled; j++)
@@ -422,6 +437,10 @@  static struct commit_list *get_merge_bases_many_0(struct repository *r,
 	clear_commit_marks_many(n, twos, all_flags);
 
 	cnt = remove_redundant(r, rslt, cnt);
+	if (cnt < 0) {
+		free(rslt);
+		return NULL;
+	}
 	result = NULL;
 	for (i = 0; i < cnt; i++)
 		commit_list_insert_by_date(rslt[i], &result);
@@ -491,7 +510,7 @@  int repo_in_merge_bases_many(struct repository *r, struct commit *commit,
 			     int nr_reference, struct commit **reference,
 			     int ignore_missing_commits)
 {
-	struct commit_list *bases;
+	struct commit_list *bases = NULL;
 	int ret = 0, i;
 	timestamp_t generation, max_generation = GENERATION_NUMBER_ZERO;
 
@@ -510,10 +529,11 @@  int repo_in_merge_bases_many(struct repository *r, struct commit *commit,
 	if (generation > max_generation)
 		return ret;
 
-	bases = paint_down_to_common(r, commit,
-				     nr_reference, reference,
-				     generation, ignore_missing_commits);
-	if (commit->object.flags & PARENT2)
+	if (paint_down_to_common(r, commit,
+				 nr_reference, reference,
+				 generation, ignore_missing_commits, &bases))
+		ret = -1;
+	else if (commit->object.flags & PARENT2)
 		ret = 1;
 	clear_commit_marks(commit, all_flags);
 	clear_commit_marks_many(nr_reference, reference, all_flags);
@@ -566,6 +586,10 @@  struct commit_list *reduce_heads(struct commit_list *heads)
 		}
 	}
 	num_head = remove_redundant(the_repository, array, num_head);
+	if (num_head < 0) {
+		free(array);
+		return NULL;
+	}
 	for (i = 0; i < num_head; i++)
 		tail = &commit_list_insert(array[i], tail)->next;
 	free(array);