Git development
 help / color / mirror / Atom feed
From: Junio C Hamano <gitster@pobox.com>
To: "Harald Nordgren via GitGitGadget" <gitgitgadget@gmail.com>
Cc: git@vger.kernel.org,
	 Kristoffer Haugsbakk <kristofferhaugsbakk@fastmail.com>,
	 "D. Ben Knoble" <ben.knoble@gmail.com>,
	 Phillip Wood <phillip.wood123@gmail.com>,
	 Jasper Mayone <me@jaspermayone.com>,
	 Tuomas Ahola <taahol@utu.fi>,
	 Harald Nordgren <haraldnordgren@gmail.com>
Subject: Re: [PATCH v8 4/4] branch: let --delete-integrated find branches merged onto changed files
Date: Wed, 18 Nov 2026 15:05:38 -0800	[thread overview]
Message-ID: <xmqqik1t2119.fsf@gitster.g> (raw)
In-Reply-To: <22658105edeacf296a18373400e7b35141e053ef.1794921876.git.gitgitgadget@gmail.com> (Harald Nordgren via GitGitGadget's message of "Tue, 17 Nov 2026 13:24:36 +0000")

"Harald Nordgren via GitGitGadget" <gitgitgadget@gmail.com> writes:

> @@ -888,6 +897,8 @@ static void collect_branch_changes(struct landing_search *search,
>  			   search->changes_alloc);
>  		change = &search->changes[search->changes_nr++];
>  		change->path = xstrdup(p->two->path);
> +		oidcpy(&change->base_oid, DIFF_FILE_VALID(p->one) ?
> +		       &p->one->oid : null_oid(the_hash_algo));
>  		oidcpy(&change->branch_oid, DIFF_FILE_VALID(p->two) ?
>  		       &p->two->oid : null_oid(the_hash_algo));
>  		change->branch_mode = p->two->mode;

Mental note.  We are recording the mode in the "branch", but never
record the mode of the path in the base here.

> @@ -895,25 +906,65 @@ static void collect_branch_changes(struct landing_search *search,
>  	diff_flush(&diffopt);
>  }
>  
> +static int merge_keeps_upstream(const struct branch_change *change)
> +{
> +	mmfile_t base, upstream, branch;
> +	mmfile_t result = { 0 };
> +	int ret;
> +
> +	read_mmblob(&base, the_repository->objects, &change->base_oid);
> +	read_mmblob(&upstream, the_repository->objects, &change->upstream_oid);
> +	read_mmblob(&branch, the_repository->objects, &change->branch_oid);

Imagine that a path that originally was a gitlink has been replaced
with different regular files on our branch and the upstream.  If
this function ever gets called on such a path from changes_landed(),
the base_oid would record the submodule's commit object name, and
read_mmblob() would probably crash.

> +	ret = ll_merge(&result, change->path, &base, "base",
> +		       &upstream, "upstream", &branch, "branch",
> +		       the_repository->index, NULL) == LL_MERGE_OK &&
> +	      result.size == upstream.size &&
> +	      !memcmp(result.ptr, upstream.ptr, upstream.size);
> +
> +	free(base.ptr);
> +	free(upstream.ptr);
> +	free(branch.ptr);
> +	free(result.ptr);
> +	return ret;
> +}
> +
> +static enum change_state change_state_at(struct branch_change *change,
> +					 struct commit *commit)
>  {
>  	unsigned short mode;
>  
>  	if (get_tree_entry(the_repository, get_commit_tree_oid(commit),
> +			   change->path, &change->upstream_oid, &mode))
> +		return is_null_oid(&change->branch_oid) ?
> +			CHANGE_LANDED : CHANGE_MISSING;
> +	if (oideq(&change->upstream_oid, &change->branch_oid))
> +		return mode == change->branch_mode ?
> +			CHANGE_LANDED : CHANGE_MISSING;
>
> +	if (is_null_oid(&change->base_oid) ||
> +	    is_null_oid(&change->branch_oid) ||
> +	    oideq(&change->upstream_oid, &change->base_oid) ||
> +	    mode != change->branch_mode || !S_ISREG(mode))
> +		return CHANGE_MISSING;
> +	return CHANGE_NEEDS_MERGE;
>  }

Now, if a gitlink was turned into different regular files, as object
names in upstream/branch/base are all different, none of them is
null_oid, and as we saw earlier, branch_mode is from p->two->mode
found in collect_branch_changes(), and mode would also be for the
regular file.  As the mode is the same, we end up returning
CHANGE_NEEDS_NERGE from here.  Which would cause ...

>  static int changes_landed(struct landing_search *search,
>  			  struct commit *commit)
>  {
> -	for (size_t i = 0; i < search->changes_nr; i++)
> -		if (!change_landed(&search->changes[i], commit))
> +	for (size_t i = 0; i < search->changes_nr; i++) {
> +		struct branch_change *change = &search->changes[i];
> +
> +		change->state = change_state_at(change, commit);
> +		if (change->state == CHANGE_MISSING)
>  			return 0;
> +	}
> +	for (size_t i = 0; i < search->changes_nr; i++) {
> +		struct branch_change *change = &search->changes[i];
> +
> +		if (change->state == CHANGE_NEEDS_MERGE &&
> +		    !merge_keeps_upstream(change))
> +			return 0;
> +	}

... this code to call merge_keeps_upstream(), causing read_mmfile()
on the gitlink commit from submodule object database, leading to a
crash, no?

Or am I misreading the code flow?

>  	return 1;
>  }

      reply	other threads:[~2026-11-18 23:05 UTC|newest]

Thread overview: 92+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-29  7:30 [PATCH] branch: let --delete-merged find squash merged branches Harald Nordgren via GitGitGadget
2026-09-29  7:46 ` Kristoffer Haugsbakk
2026-09-29  7:52   ` Harald Nordgren
2026-09-29  8:08     ` Kristoffer Haugsbakk
2026-09-29 11:18       ` D. Ben Knoble
2026-09-29 13:31         ` Kristoffer Haugsbakk
2026-09-29 11:26 ` D. Ben Knoble
2026-09-29 13:44   ` Harald Nordgren
2026-09-29 13:52   ` Harald Nordgren
2026-09-29 15:28     ` D. Ben Knoble
2026-09-29 18:34       ` Harald Nordgren
2026-10-04  9:54   ` Phillip Wood
2026-10-04 22:28     ` Harald Nordgren
2026-10-08 15:25       ` Phillip Wood
2026-10-08 18:42         ` Harald Nordgren
2026-10-09  8:16           ` Kristoffer Haugsbakk
2026-10-09 11:11             ` Harald Nordgren
2026-10-09 11:18           ` Harald Nordgren
2026-10-09 13:40             ` Phillip Wood
2026-10-09 13:37           ` Phillip Wood
2026-10-09 15:27             ` Harald Nordgren
2026-10-08 17:51     ` D. Ben Knoble
2026-10-09 13:34       ` Phillip Wood
2026-10-11  9:41 ` [PATCH v2 0/2] " Harald Nordgren via GitGitGadget
2026-10-11  9:41   ` [PATCH v2 1/2] " Harald Nordgren via GitGitGadget
2026-10-11  9:41   ` [PATCH v2 2/2] branch: add --no-squashed to skip the squash search Harald Nordgren via GitGitGadget
2026-10-12  0:30   ` [PATCH v2 0/2] branch: let --delete-merged find squash merged branches Junio C Hamano
2026-10-12  9:50     ` Phillip Wood
2026-10-12 14:31       ` Harald Nordgren
2026-10-12 16:50       ` Junio C Hamano
2026-10-12 17:44         ` Harald Nordgren
2026-10-13  2:23           ` Junio C Hamano
2026-10-17 22:23             ` Harald Nordgren
2026-10-19  4:40             ` Jasper Mayone
2026-10-18 15:09         ` Harald Nordgren
2026-10-12 12:41     ` Harald Nordgren
2026-10-12 15:49       ` D. Ben Knoble
2026-10-21 12:28 ` [PATCH v3 0/3] " Harald Nordgren via GitGitGadget
2026-10-21 12:28   ` [PATCH v3 1/3] branch: move the deleteMerged opt-out check into a helper Harald Nordgren via GitGitGadget
2026-10-21 12:28   ` [PATCH v3 2/3] branch: let --delete-merged find rebase merged branches Harald Nordgren via GitGitGadget
2026-10-21 12:28   ` [PATCH v3 3/3] branch: add --squashed to find squash " Harald Nordgren via GitGitGadget
2026-10-21 17:35   ` [PATCH v3 0/3] branch: let --delete-merged " Jasper Mayone
2026-10-21 18:51     ` Harald Nordgren
2026-10-22 12:55       ` D. Ben Knoble
2026-10-22 14:41         ` Harald Nordgren
2026-10-22  5:49   ` Junio C Hamano
2026-10-22  7:18     ` Harald Nordgren
2026-10-22 16:33       ` Junio C Hamano
2026-10-22 16:57         ` Junio C Hamano
2026-10-22 17:31           ` Harald Nordgren
2026-10-22 17:44             ` Jasper Mayone
2026-10-22 17:56               ` Junio C Hamano
2026-10-23  9:07               ` Harald Nordgren
2026-10-22 17:44         ` Harald Nordgren
2026-10-22 12:59   ` D. Ben Knoble
2026-10-26  8:33 ` [PATCH v4 0/4] " Harald Nordgren via GitGitGadget
2026-10-26  8:33   ` [PATCH v4 1/4] branch: move the deleteMerged opt-out check into a helper Harald Nordgren via GitGitGadget
2026-10-26  8:33   ` [PATCH v4 2/4] branch: let delete_branches() name where a branch landed Harald Nordgren via GitGitGadget
2026-10-26  8:33   ` [PATCH v4 3/4] branch: add --delete-integrated to delete rebase and squash merged branches Harald Nordgren via GitGitGadget
2026-10-26  8:33   ` [PATCH v4 4/4] branch: let --delete-integrated find branches merged onto changed files Harald Nordgren via GitGitGadget
2026-10-29  4:30   ` [PATCH v4 0/4] branch: let --delete-merged find squash merged branches Jasper Mayone
2026-10-29  8:30     ` Harald Nordgren
2026-10-30 10:52 ` [PATCH v5 " Harald Nordgren via GitGitGadget
2026-10-30 10:52   ` [PATCH v5 1/4] branch: move the deleteMerged opt-out check into a helper Harald Nordgren via GitGitGadget
2026-10-30 10:53   ` [PATCH v5 2/4] branch: let delete_branches() name where a branch landed Harald Nordgren via GitGitGadget
2026-10-30 10:53   ` [PATCH v5 3/4] branch: add --delete-integrated to delete rebase and squash merged branches Harald Nordgren via GitGitGadget
2026-10-30 19:14     ` Junio C Hamano
2026-10-31  8:09       ` Harald Nordgren
2026-10-30 10:53   ` [PATCH v5 4/4] branch: let --delete-integrated find branches merged onto changed files Harald Nordgren via GitGitGadget
2026-10-30 19:25     ` Junio C Hamano
2026-10-31  8:10       ` Harald Nordgren
2026-11-01  8:22 ` [PATCH v6 0/4] branch: let --delete-merged find squash merged branches Harald Nordgren via GitGitGadget
2026-11-01  8:22   ` [PATCH v6 1/4] branch: move the deleteMerged opt-out check into a helper Harald Nordgren via GitGitGadget
2026-11-01  8:22   ` [PATCH v6 2/4] branch: let delete_branches() name where a branch landed Harald Nordgren via GitGitGadget
2026-11-01  8:22   ` [PATCH v6 3/4] branch: add --delete-integrated to delete rebase and squash merged branches Harald Nordgren via GitGitGadget
2026-11-01  8:22   ` [PATCH v6 4/4] branch: let --delete-integrated find branches merged onto changed files Harald Nordgren via GitGitGadget
2026-11-11  8:30     ` Harald Nordgren
2026-11-11 15:27       ` Jasper Mayone
2026-11-14 11:08         ` Harald Nordgren
2026-11-14 13:24 ` [PATCH v7 0/4] branch: let --delete-merged find squash merged branches Harald Nordgren via GitGitGadget
2026-11-14 13:24   ` [PATCH v7 1/4] branch: move the deleteMerged opt-out check into a helper Harald Nordgren via GitGitGadget
2026-11-14 13:24   ` [PATCH v7 2/4] branch: let delete_branches() name where a branch landed Harald Nordgren via GitGitGadget
2026-11-14 13:24   ` [PATCH v7 3/4] branch: add --delete-integrated to delete rebase and squash merged branches Harald Nordgren via GitGitGadget
2026-11-14 13:24   ` [PATCH v7 4/4] branch: let --delete-integrated find branches merged onto changed files Harald Nordgren via GitGitGadget
2026-11-14 17:14     ` Tuomas Ahola
2026-11-16 16:16     ` Jasper Mayone
2026-11-17 13:24 ` [PATCH v8 0/4] branch: let --delete-merged find squash merged branches Harald Nordgren via GitGitGadget
2026-11-17 13:24   ` [PATCH v8 1/4] branch: move the deleteMerged opt-out check into a helper Harald Nordgren via GitGitGadget
2026-11-17 13:24   ` [PATCH v8 2/4] branch: let delete_branches() name where a branch landed Harald Nordgren via GitGitGadget
2026-11-17 13:24   ` [PATCH v8 3/4] branch: add --delete-integrated to delete rebase and squash merged branches Harald Nordgren via GitGitGadget
2026-11-17 13:24   ` [PATCH v8 4/4] branch: let --delete-integrated find branches merged onto changed files Harald Nordgren via GitGitGadget
2026-11-18 23:05     ` Junio C Hamano [this message]

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=xmqqik1t2119.fsf@gitster.g \
    --to=gitster@pobox.com \
    --cc=ben.knoble@gmail.com \
    --cc=git@vger.kernel.org \
    --cc=gitgitgadget@gmail.com \
    --cc=haraldnordgren@gmail.com \
    --cc=kristofferhaugsbakk@fastmail.com \
    --cc=me@jaspermayone.com \
    --cc=phillip.wood123@gmail.com \
    --cc=taahol@utu.fi \
    /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