Re: [PATCH 1/8] t/helper: add 'test-tool bitmap write' subcommand
- From
Junio C Hamano <gitster@pobox.com>
- Date
- Apr 14, 2026, 20:08 UTC
- Message-ID
- <xmqqeckh9uew.fsf@gitster.g>
- In-Reply-To
- <d5ef6b959fd7c05c73bd33aa2b394558320aceac.1776124588.git.me@ttaylorr.com>
Taylor Blau <me@ttaylorr.com> writes:
Show 19 quoted lines
> +struct bitmap_writer_data {
> + struct packing_data packed;
> + struct pack_idx_entry **index;
> + uint32_t nr;
> +};
> +
> +static int add_packed_object(const struct object_id *oid,
> + struct packed_git *pack,
> + uint32_t pos,
> + void *_data)
> +{
> + struct bitmap_writer_data *data = _data;
> + struct object_entry *entry;
> + struct object_info oi = OBJECT_INFO_INIT;
> + enum object_type type;
> +
> + oi.typep = &type;
> +
> + entry = packlist_alloc(&data->packed, oid);data->packed is "packing data" that has a pointer "objects" that is a flat array of "struct object_entry". This array is dynamically expanded with realloc() in packlist_alloc(), and it returns a pointer into this data->packed->objects[] array. We receive it in a local variable "entry" here.
Show 6 quoted lines
> + entry->idx.offset = nth_packed_object_offset(pack, pos);
> + if (packed_object_info(pack, entry->idx.offset, &oi) < 0)
> + die("could not get type of object %s",
> + oid_to_hex(oid));
> + oe_set_type(entry, type);
> + oe_set_in_pack(&data->packed, entry, pack);And populate the entry.
> + data->index[data->nr++] = &entry->idx;
And then store the pointer to one of the members (actually the first member) in that "struct object_entry" instance in that data->packed->objects[] array we took from.
What happens when a repeated call to this function to add many objects (those contained within the pack we are iterating over) caused the packlist_alloc() to realloc data->packed->objects[] array eventually? Wouldn't it invalidate the address of &entry->idx we are taking from before the realloc() happens?
I must be missing something?