Merge pull request #453 from arutk/no_cl_gl_lock
Skip cacheline concurrency global lock in fast path
This commit is contained in:
@@ -80,14 +80,18 @@ void cache_mngt_core_deinit_attached_meta(ocf_core_t core)
|
||||
continue;
|
||||
}
|
||||
|
||||
if (!ocf_cache_line_try_lock_wr(cache, curr_cline))
|
||||
if (!ocf_cache_line_try_lock_wr(
|
||||
cache->device->concurrency.cache_line,
|
||||
curr_cline)) {
|
||||
break;
|
||||
}
|
||||
|
||||
if (metadata_test_dirty(cache, curr_cline))
|
||||
ocf_purge_cleaning_policy(cache, curr_cline);
|
||||
ocf_metadata_sparse_cache_line(cache, curr_cline);
|
||||
|
||||
ocf_cache_line_unlock_wr(cache, curr_cline);
|
||||
ocf_cache_line_unlock_wr(cache->device->concurrency.cache_line,
|
||||
curr_cline);
|
||||
|
||||
if (prev_cline != cache->device->collision_table_entries)
|
||||
curr_cline = ocf_metadata_get_collision_next(cache, prev_cline);
|
||||
|
Reference in New Issue
Block a user