Re: [PATCH v3 5/9] midx: refactor permutation logic and pack sorting
- From
- Jonathan Tan <jonathantanmy@google.com>
- Date
- Jan 23, 2019, 21:00 UTC
- Message-ID
- <20190123210054.118647-1-jonathantanmy@google.com>
- In-Reply-To
- <a0d4cc6cb3fbca0de732dfd34cbec4d765b55228.1547047269.git.gitgitgadget@gmail.com>
Following Stolee's wishes [1], I'll stick to the technical aspects here. Patches 1-4 look correct technically to me, so let me start here.
[1] https://public-inbox.org/git/3aa0a7ea-6c30-2c61-0815-2b9ab8304564@gmail.com/
Show 23 quoted lines
> +struct pack_info {
> + uint32_t orig_pack_int_id;
> + char *pack_name;
> + struct packed_git *p;
> +};
> +
> +static int pack_info_compare(const void *_a, const void *_b)
> +{
> + struct pack_info *a = (struct pack_info *)_a;
> + struct pack_info *b = (struct pack_info *)_b;
> + return strcmp(a->pack_name, b->pack_name);
> +}
> +
> struct pack_list {
> - struct packed_git **list;
> - char **names;
> + struct pack_info *info;
> uint32_t nr;
> - uint32_t alloc_list;
> - uint32_t alloc_names;
> + uint32_t alloc;
> struct multi_pack_index *m;
> };"list" and "names" look like a parallel array, but as far as I can tell, they are actually not, because "names" is eventually sorted but not "list". So combining these two arrays is great, and removes a lot of confusion.
Now the packed_git list will be sorted too when the names are sorted; I was surprised that code did not need to be changed to take this into account, but I see now that it's because the struct packed_gits don't need to be referenced anymore after get_sorted_entries().
Show 33 quoted lines
> -struct pack_pair {
> - uint32_t pack_int_id;
> - char *pack_name;
> -};
> -
> -static int pack_pair_compare(const void *_a, const void *_b)
> -{
> - struct pack_pair *a = (struct pack_pair *)_a;
> - struct pack_pair *b = (struct pack_pair *)_b;
> - return strcmp(a->pack_name, b->pack_name);
> -}
> -
> -static void sort_packs_by_name(char **pack_names, uint32_t nr_packs, uint32_t *perm)
> -{
> - uint32_t i;
> - struct pack_pair *pairs;
> -
> - ALLOC_ARRAY(pairs, nr_packs);
> -
> - for (i = 0; i < nr_packs; i++) {
> - pairs[i].pack_int_id = i;
> - pairs[i].pack_name = pack_names[i];
> - }
> -
> - QSORT(pairs, nr_packs, pack_pair_compare);
> -
> - for (i = 0; i < nr_packs; i++) {
> - pack_names[i] = pairs[i].pack_name;
> - perm[pairs[i].pack_int_id] = i;
> - }
> -
> - free(pairs);
> -}It's nice that this gets removed.
Show 12 quoted lines
> static int nth_midxed_pack_midx_entry(struct multi_pack_index *m,
> - uint32_t *pack_perm,
> struct pack_midx_entry *e,
> uint32_t pos)
> {
> @@ -488,7 +464,7 @@ static int nth_midxed_pack_midx_entry(struct multi_pack_index *m,
> return 1;
>
> nth_midxed_object_oid(&e->oid, m, pos);
> - e->pack_int_id = pack_perm[nth_midxed_pack_int_id(m, pos)];
> + e->pack_int_id = nth_midxed_pack_int_id(m, pos);
> e->offset = nth_midxed_offset(m, pos);nth_midxed_pack_midx_entry() is only called from get_sorted_entries(), which is now called *before* any sorting of pack_info, so there is no longer any need for considering pack_perm. As to why get_sorted_entries() needs to be called before sorting of pack_info and not after, that question will be answered later.
Show 9 quoted lines
> @@ -522,8 +498,7 @@ static void fill_pack_entry(uint32_t pack_int_id, > * of a packfile containing the object). > */ > static struct pack_midx_entry *get_sorted_entries(struct multi_pack_index *m, > - struct packed_git **p, > - uint32_t *perm, > + struct pack_info *info, > uint32_t nr_packs, > uint32_t *nr_objects)
[snip rest of get_sorted_entries]
As stated above, get_sorted_entries() is now called before any sorting of pack_info, so there is no need for perm.
Show 14 quoted lines
> static size_t write_midx_object_offsets(struct hashfile *f, int large_offset_needed,
> + uint32_t *perm,
> struct pack_midx_entry *objects, uint32_t nr_objects)
> {
> struct pack_midx_entry *list = objects;
> @@ -699,7 +675,7 @@ static size_t write_midx_object_offsets(struct hashfile *f, int large_offset_nee
> for (i = 0; i < nr_objects; i++) {
> struct pack_midx_entry *obj = list++;
>
> - hashwrite_be32(f, obj->pack_int_id);
> + hashwrite_be32(f, perm[obj->pack_int_id]);
>
> if (large_offset_needed && obj->offset >> 31)
> hashwrite_be32(f, MIDX_LARGE_OFFSET_NEEDED | nr_large_offset++);This was unexpected to me - why is perm introduced here when it wasn't needed before? This is because get_sorted_entries() is now called before any sorting of pack_info. get_sorted_entries() generates entries using the old pack_int_id, but the resulting packfile is written using new pack numbers, so when writing the entries to the file, they must be written using the new numbers.
The mapping of perm is: perm[old number] = new number.
Show 29 quoted lines
> @@ -795,10 +768,7 @@ int write_midx_file(const char *object_dir)
> if (packs.m && packs.nr == packs.m->num_packs)
> goto cleanup;
>
> - ALLOC_ARRAY(pack_perm, packs.nr);
> - sort_packs_by_name(packs.names, packs.nr, pack_perm);
> -
> - entries = get_sorted_entries(packs.m, packs.list, pack_perm, packs.nr, &nr_entries);
> + entries = get_sorted_entries(packs.m, packs.info, packs.nr, &nr_entries);
>
> for (i = 0; i < nr_entries; i++) {
> if (entries[i].offset > 0x7fffffff)
> @@ -807,8 +777,15 @@ int write_midx_file(const char *object_dir)
> large_offsets_needed = 1;
> }
>
> + QSORT(packs.info, packs.nr, pack_info_compare);
> +
> + ALLOC_ARRAY(pack_perm, packs.nr);
> + for (i = 0; i < packs.nr; i++) {
> + pack_perm[packs.info[i].orig_pack_int_id] = i;
> + }
> +
> for (i = 0; i < packs.nr; i++)
> - pack_name_concat_len += strlen(packs.names[i]) + 1;
> + pack_name_concat_len += strlen(packs.info[i].pack_name) + 1;
>
> if (pack_name_concat_len % MIDX_CHUNK_ALIGNMENT)
> pack_name_concat_len += MIDX_CHUNK_ALIGNMENT -Indeed, the sorting of pack_info is moved to after get_sorted_entries(). Also, pack_perm[old number] = new number, as expected.
I think a comment explaining why the perm is needed would be helpful - something explaining that the entries were generated using the old pack numbers, so we need this mapping to be able to write them using the new numbers.
As part of this review, I did attempt to make everything use the sorted pack_info order, but failed. I got as far as making get_sorted_entries() always use start_pack=0 and skip the pack_infos that didn't have the p pointer (to match the existing behavior of get_sorted_entries() that only operates on new pack_infos being added by add_pack_to_midx), but it still didn't work, and I didn't investigate further.