From: René Scharfe <l.s.r@web.de>
Date: Sat, 03 Jun 2023 06:02:51 GMT
Subject: Re: [PATCH] Fix memory leak in get_reachable_subset
Message-ID: <21d3f105-ed32-e397-a1fe-53344894bd44@web.de>
In-Reply-To: <xmqqedmt7167.fsf@gitster.g>

Am 03.06.23 um 04:02 schrieb Junio C Hamano:
> Mike Hommey <mh@glandium.org> writes:
>
>> Signed-off-by: Mike Hommey <mh@glandium.org>
>> ---
>
> As most lines in the affected function seems to come from your
> fcb2c076 (commit-reach: implement get_reachable_subset, 2018-11-02),
> I'll redirect the review of this patch to you.

Stolee reviewed it already when the patch was sent the first time, here:
https://lore.kernel.org/git/20230421234409.1925489-1-mh@glandium.org/T/#u

>
> Thanks.
>
>>  commit-reach.c | 2 ++
>>  1 file changed, 2 insertions(+)
>>
>> diff --git a/commit-reach.c b/commit-reach.c
>> index 70bde8af05..f15d84566b 100644
>> --- a/commit-reach.c
>> +++ b/commit-reach.c
>> @@ -944,6 +944,8 @@ struct commit_list *get_reachable_subset(struct commit **from, int nr_from,
>>  		}
>>  	}
>>
>> +	clear_prio_queue(&queue);

Makes sense: The loop that dequeues would end before consuming all items
when all "to" commits are found to be reachable.

>> +
>>  	clear_commit_marks_many(nr_to, to, PARENT1);
>>  	clear_commit_marks_many(nr_from, from, PARENT2);