ivampiresp/Spdk
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

test/blob: verify number of extent page writes

Browse Source 
New UT verifies number of extent page writes for
thin provisioned blobs.

Signed-off-by: Jim Harris <james.r.harris@intel.com>
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I6b52195d26c3c1414ca862b03e8a3043bbe29b42
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7017
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Community-CI: Mellanox Build Bot
This commit is contained in:
Jim Harris 2021-03-23 17:11:01 +00:00 committed by Tomasz Zawadzki
parent ae10cf633c
commit 1f5364e821
1 changed files with 134 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

134
test/unit/lib/blob/blob.c/blob_ut.c
View File

@ -3863,6 +3863,139 @@ blob_thin_prov_rw(void)
g_blobid = 0;
}
static void
blob_thin_prov_write_count_io(void)
{
struct spdk_blob_store *bs;
struct spdk_blob *blob;
struct spdk_io_channel *ch;
struct spdk_bs_dev *dev;
struct spdk_bs_opts bs_opts;
struct spdk_blob_opts opts;
uint64_t free_clusters;
uint64_t page_size;
uint8_t payload_write[4096];
uint64_t write_bytes;
uint64_t read_bytes;
const uint32_t CLUSTER_SZ = 16384;
uint32_t pages_per_cluster;
uint32_t pages_per_extent_page;
uint32_t i;
/* Use a very small cluster size for this test. This ensures we need multiple
* extent pages to hold all of the clusters even for relatively small blobs like
* we are restricted to for the unit tests (i.e. we don't want to allocate multi-GB
* buffers).
*/
dev = init_dev();
spdk_bs_opts_init(&bs_opts, sizeof(bs_opts));
bs_opts.cluster_sz = CLUSTER_SZ;
spdk_bs_init(dev, &bs_opts, bs_op_with_handle_complete, NULL);
poll_threads();
CU_ASSERT(g_bserrno == 0);
SPDK_CU_ASSERT_FATAL(g_bs != NULL);
bs = g_bs;
free_clusters = spdk_bs_free_cluster_count(bs);
page_size = spdk_bs_get_page_size(bs);
pages_per_cluster = CLUSTER_SZ / page_size;
pages_per_extent_page = SPDK_EXTENTS_PER_EP * pages_per_cluster;
ch = spdk_bs_alloc_io_channel(bs);
SPDK_CU_ASSERT_FATAL(ch != NULL);
ut_spdk_blob_opts_init(&opts);
opts.thin_provision = true;
blob = ut_blob_create_and_open(bs, &opts);
CU_ASSERT(free_clusters == spdk_bs_free_cluster_count(bs));
/* Resize the blob so that it will require 8 extent pages to hold all of
* the clusters.
*/
g_bserrno = -1;
spdk_blob_resize(blob, SPDK_EXTENTS_PER_EP * 8, blob_op_complete, NULL);
poll_threads();
CU_ASSERT(g_bserrno == 0);
g_bserrno = -1;
spdk_blob_sync_md(blob, blob_op_complete, NULL);
poll_threads();
CU_ASSERT(g_bserrno == 0);
CU_ASSERT(free_clusters == spdk_bs_free_cluster_count(bs));
CU_ASSERT(blob->active.num_clusters == SPDK_EXTENTS_PER_EP * 8);
memset(payload_write, 0, sizeof(payload_write));
for (i = 0; i < 8; i++) {
write_bytes = g_dev_write_bytes;
read_bytes = g_dev_read_bytes;
g_bserrno = -1;
spdk_blob_io_write(blob, ch, payload_write, pages_per_extent_page * i, 1, blob_op_complete, NULL);
poll_threads();
CU_ASSERT(g_bserrno == 0);
CU_ASSERT(free_clusters - (2 * i + 1) == spdk_bs_free_cluster_count(bs));
CU_ASSERT(g_dev_read_bytes == read_bytes);
if (!g_use_extent_table) {
/* For legacy metadata, we should have written two pages - one for the
* write I/O itself, another for the blob's primary metadata.
*/
CU_ASSERT((g_dev_write_bytes - write_bytes) / page_size == 2);
} else {
/* For extent table metadata, we should have written three pages - one
* for the write I/O, one for the extent page, one for the blob's primary
* metadata.
*/
CU_ASSERT((g_dev_write_bytes - write_bytes) / page_size == 3);
}
/* The write should have synced the metadata already. Do another sync here
* just to confirm.
*/
write_bytes = g_dev_write_bytes;
read_bytes = g_dev_read_bytes;
g_bserrno = -1;
spdk_blob_sync_md(blob, blob_op_complete, NULL);
poll_threads();
CU_ASSERT(g_bserrno == 0);
CU_ASSERT(free_clusters - (2 * i + 1) == spdk_bs_free_cluster_count(bs));
CU_ASSERT(g_dev_read_bytes == read_bytes);
CU_ASSERT(g_dev_write_bytes == write_bytes);
/* Now write to another unallocated cluster that is part of the same extent page. */
g_bserrno = -1;
spdk_blob_io_write(blob, ch, payload_write, pages_per_extent_page * i + pages_per_cluster,
1, blob_op_complete, NULL);
poll_threads();
CU_ASSERT(g_bserrno == 0);
CU_ASSERT(free_clusters - (2 * i + 2) == spdk_bs_free_cluster_count(bs));
CU_ASSERT(g_dev_read_bytes == read_bytes);
/*
* For legacy metadata, we should have written the I/O and the primary metadata page.
* For extent table metadata, we should have written the I/O and the extent metadata page.
*/
CU_ASSERT((g_dev_write_bytes - write_bytes) / page_size == 2);
}
ut_blob_close_and_delete(bs, blob);
CU_ASSERT(free_clusters == spdk_bs_free_cluster_count(bs));
spdk_bs_free_io_channel(ch);
poll_threads();
g_blob = NULL;
g_blobid = 0;
spdk_bs_unload(bs, bs_op_complete, NULL);
poll_threads();
CU_ASSERT(g_bserrno == 0);
g_bs = NULL;
}
static void
blob_thin_prov_rle(void)
{
@ -6784,6 +6917,7 @@ int main(int argc, char **argv)
CU_ADD_TEST(suite_bs, blob_thin_prov_alloc);
CU_ADD_TEST(suite_bs, blob_insert_cluster_msg_test);
CU_ADD_TEST(suite_bs, blob_thin_prov_rw);
CU_ADD_TEST(suite, blob_thin_prov_write_count_io);
CU_ADD_TEST(suite_bs, blob_thin_prov_rle);
CU_ADD_TEST(suite_bs, blob_thin_prov_rw_iov);
CU_ADD_TEST(suite, bs_load_iter_test);