adding a merge map function #6

Merged
greg merged 12 commits from gburd/merge-map into main 2024-04-30 15:20:48 +00:00
3 changed files with 95 additions and 17 deletions
Showing only changes of commit e5e4a0f960 - Show all commits

View file

@ -264,7 +264,7 @@ void sparsemap_scan(sparsemap_t *map, void (*scanner)(sm_idx_t vec[], size_t n),
* \b other bitmap while removing them from \b map. * \b other bitmap while removing them from \b map.
* *
* @param[in] map The sparsemap reference. * @param[in] map The sparsemap reference.
* @param[in] other The bitmap to merge into \b map. * @param[in] other The bitmap to merge into \b map.
*/ */
void sparsemap_merge(sparsemap_t *map, sparsemap_t *other); void sparsemap_merge(sparsemap_t *map, sparsemap_t *other);

View file

@ -109,6 +109,7 @@ enum __SM_CHUNK_INFO {
}; };
#define SM_CHUNK_GET_FLAGS(from, at) (((from)) & ((sm_bitvec_t)SM_FLAG_MASK << ((at)*2))) >> ((at)*2) #define SM_CHUNK_GET_FLAGS(from, at) (((from)) & ((sm_bitvec_t)SM_FLAG_MASK << ((at)*2))) >> ((at)*2)
#define SM_NEXT_CHUNK_ADDR(chunk) (uint8_t *)((chunk) + sizeof(sm_idx_t) + __sm_chunk_map_get_size((__sm_chunk_t *)&(chunk)));
typedef struct { typedef struct {
sm_bitvec_t *m_data; sm_bitvec_t *m_data;
@ -763,7 +764,6 @@ __sm_get_chunk_map_data(sparsemap_t *map, size_t offset)
* *
* @param[in] map The sparsemap_t in question. * @param[in] map The sparsemap_t in question.
* @returns a pointer after the end of the used data * @returns a pointer after the end of the used data
* @todo could this simply use m_data_used?
*/ */
static uint8_t * static uint8_t *
__sm_get_chunk_map_end(sparsemap_t *map) __sm_get_chunk_map_end(sparsemap_t *map)
@ -1247,30 +1247,57 @@ sparsemap_scan(sparsemap_t *map, void (*scanner)(sm_idx_t[], size_t), size_t ski
} }
} }
void
__sm_chunk_map_merge(sparsemap_t *map, sparsemap_idx_t offset, __sm_chunk_t src) {
size_t capacity = __sm_chunk_map_get_capacity(&src);
for (sparsemap_idx_t j = 0; j < capacity; j++, offset++) {
if (__sm_chunk_map_is_set(&src, j)) {
sparsemap_set(map, offset, true);
}
}
}
void void
sparsemap_merge(sparsemap_t *map, sparsemap_t *other) sparsemap_merge(sparsemap_t *map, sparsemap_t *other)
{ {
uint8_t *next_dst, *dst = __sm_get_chunk_map_data(map, 0); uint8_t *dst = __sm_get_chunk_map_data(map, 0),
*dst_tail = __sm_get_chunk_map_end(map),
/* |src| points to the |other| (source) chunk map */ *src = __sm_get_chunk_map_data(other, 0);
uint8_t *next_src, *src = __sm_get_chunk_map_data(other, 0);
size_t i, src_count = __sm_get_chunk_map_count(other), dst_count = __sm_get_chunk_map_count(map); size_t i, src_count = __sm_get_chunk_map_count(other), dst_count = __sm_get_chunk_map_count(map);
for (i = 0; i < dst_count; i++) { for (i = 0; i < src_count + dst_count;) {
__sm_chunk_t src_chunk, dst_chunk;
sm_idx_t src_start = *(sm_idx_t *)src; sm_idx_t src_start = *(sm_idx_t *)src;
sm_idx_t dst_start = *(sm_idx_t *)dst; sm_idx_t dst_start = *(sm_idx_t *)dst;
__sm_chunk_t src_chunk, dst_chunk;
__sm_chunk_map_init(&src_chunk, src + sizeof(sm_idx_t)); __sm_chunk_map_init(&src_chunk, src + sizeof(sm_idx_t));
__sm_chunk_map_init(&dst_chunk, dst + sizeof(sm_idx_t)); __sm_chunk_map_init(&dst_chunk, dst + sizeof(sm_idx_t));
if (src_start == dst_start) {
__sm_chunk_map_merge(dst, src);
} else {
next_dst = dst + sizeof(sm_idx_t) + __sm_chunk_map_get_size(&dst_chunk);
next_src = dst + sizeof(sm_idx_t) + __sm_chunk_map_get_size(&src_chunk);
if (src_start != next_dst) {
// Copy the chunk.
} else {
if (src_start == dst_start) {
/* Chunks overlap, merge them. */
__sm_chunk_map_merge(map, src_start, src_chunk);
src = SM_NEXT_CHUNK_ADDR(src);
dst = SM_NEXT_CHUNK_ADDR(dst);
i += 2;
} else if (src_start < dst_start || i > dst_count) {
/* Copy the src chunk into the dest map. */
size_t s = __sm_chunk_map_get_size(&src_chunk);
/* Copy the index (sm_idx_t) */
*(sm_idx_t *)dst_tail = src_start;
dst_tail += sizeof(sm_idx_t);
/* Copy the chunk data to the buffer. */
memcpy(dst_tail, src, s);
/* Update the chunk count and data_used. */
map->m_data_used = 0;
__sm_set_chunk_map_count(map, __sm_get_chunk_map_count(map) + 1);
src = SM_NEXT_CHUNK_ADDR(src);
i += 1;
if (i > dst_count) {
dst = SM_NEXT_CHUNK_ADDR(dst);
i += 1;
} }
} }
} }
@ -1363,7 +1390,7 @@ sparsemap_split(sparsemap_t *map, sparsemap_idx_t offset, sparsemap_t *other)
__sm_chunk_map_set_capacity(&s_chunk, offset % capacity); __sm_chunk_map_set_capacity(&s_chunk, offset % capacity);
} }
/* Now continue with all remaining minimaps. */ /* Now continue with all remaining chunk maps. */
for (; i < count; i++) { for (; i < count; i++) {
sm_idx_t start = *(sm_idx_t *)src; sm_idx_t start = *(sm_idx_t *)src;
src += sizeof(sm_idx_t); src += sizeof(sm_idx_t);

View file

@ -605,6 +605,56 @@ test_api_split(const MunitParameter params[], void *data)
return MUNIT_OK; return MUNIT_OK;
} }
static void *
test_api_merge_setup(const MunitParameter params[], void *user_data)
{
(void)params;
(void)user_data;
sparsemap_t *map = sparsemap(10 * 1024);
assert_ptr_not_null(map);
return (void *)map;
}
static void
test_api_merge_tear_down(void *fixture)
{
sparsemap_t *map = (sparsemap_t *)fixture;
assert_ptr_not_null(map);
munit_free(map);
}
static MunitResult
test_api_merge(const MunitParameter params[], void *data)
{
sparsemap_t *map = (sparsemap_t *)data;
sparsemap_t *other = sparsemap(1024);
(void)params;
assert_ptr_not_null(map);
assert_ptr_not_null(other);
sparsemap_set(map, 0, true);
sparsemap_set(other, 1, true);
sparsemap_set(other, 2049, true);
sparsemap_set(other, 8193, true);
sparsemap_set(other, 8194, true);
sparsemap_merge(map,other);
assert_true(sparsemap_is_set(map, 0));
assert_true(sparsemap_is_set(map, 1));
assert_true(sparsemap_is_set(map, 2049));
assert_true(sparsemap_is_set(map, 8193));
assert_true(sparsemap_is_set(map, 8194));
for (int i = 0; i < 10000; i++) {
if (i == 0 || i == 1 || i == 2049 || i == 8193 || i == 8194)
continue;
else
assert_false(sparsemap_is_set(map, i));
}
free(other);
return MUNIT_OK;
}
static void * static void *
test_api_select_setup(const MunitParameter params[], void *user_data) test_api_select_setup(const MunitParameter params[], void *user_data)
{ {
@ -919,6 +969,7 @@ static MunitTest api_test_suite[] = {
{ (char *)"/get_size", test_api_get_size, test_api_get_size_setup, test_api_get_size_tear_down, MUNIT_TEST_OPTION_NONE, NULL }, { (char *)"/get_size", test_api_get_size, test_api_get_size_setup, test_api_get_size_tear_down, MUNIT_TEST_OPTION_NONE, NULL },
{ (char *)"/scan", test_api_scan, test_api_scan_setup, test_api_scan_tear_down, MUNIT_TEST_OPTION_NONE, NULL }, { (char *)"/scan", test_api_scan, test_api_scan_setup, test_api_scan_tear_down, MUNIT_TEST_OPTION_NONE, NULL },
{ (char *)"/split", test_api_split, test_api_split_setup, test_api_split_tear_down, MUNIT_TEST_OPTION_NONE, NULL }, { (char *)"/split", test_api_split, test_api_split_setup, test_api_split_tear_down, MUNIT_TEST_OPTION_NONE, NULL },
{ (char *)"/merge", test_api_merge, test_api_merge_setup, test_api_merge_tear_down, MUNIT_TEST_OPTION_NONE, NULL },
{ (char *)"/select/true", test_api_select, test_api_select_setup, test_api_select_tear_down, MUNIT_TEST_OPTION_NONE, NULL }, { (char *)"/select/true", test_api_select, test_api_select_setup, test_api_select_tear_down, MUNIT_TEST_OPTION_NONE, NULL },
#ifdef SELECT_FALSE #ifdef SELECT_FALSE
{ (char *)"/select/false", test_api_select_false, test_api_select_false_setup, test_api_select_false_tear_down, MUNIT_TEST_OPTION_NONE, NULL }, { (char *)"/select/false", test_api_select_false, test_api_select_false_setup, test_api_select_false_tear_down, MUNIT_TEST_OPTION_NONE, NULL },