count_valid_references(9)

πŸ“¦ version 7.3.1, in nixpkgs unstable

πŸ’… count_valid_references β€” Kernel API Manual (section 9, bestie) πŸŽ€

Okay so this is count_valid_references(9) from the Kernel API Manual, and honestly? It's a little cleanup angel sent to fix up refcount arrays that got left messy. Let's get into it ✨


🌸 NAME 🌸

count_valid_references β€” process a newly loaded refcount array 🫢


✨ SYNOPSIS ✨

unsigned int count_valid_references (vdo_refcount_t *counters );

πŸŽ€ ARGUMENTS πŸŽ€

  • counters β€” the array of counters that came out of a metadata block. That's the whole input, she's lowkey simple like that πŸ’–

πŸ¦‹ DESCRIPTION πŸ¦‹

So here's the vibe: this function scans an 8-byte-aligned array of counters, and while it's going through it, it fixes up any provisional values that never got cleaned up at shutdown β€” internally it just switches them to zero. Not me doing the janitorial work at 3am πŸ˜­πŸ’…

No cap, that's it. Scan, fix the stragglers, move on.


πŸ”₯ RETURN πŸ”₯

It returns the number of blocks that have a non-zero reference count. Iconic little tally, periodt 🌟


πŸ’œ SEE ALSO πŸ’œ

Kernel file ./drivers/md/dm-vdo/slab-depot.c, plus the entire friend group β€” like genuinely the whole group chat showed up:

get_lock(9), must_make_entries_to_flush(9), is_reaping(9), initialize_tail_block(9), initialize_journal_state(9), block_is_full(9), is_slab_journal_blank(9), mark_slab_journal_dirty(9), compute_fullness_hint(9), check_summary_drain_complete(9), notify_summary_waiters(9), finish_updating_slab_summary_block(9), finish_update(9), handle_write_error(9), launch_write(9), update_slab_summary_entry(9), finish_reaping(9), complete_reaping(9), handle_flush_error(9), flush_for_reaping(9), reap_slab_journal(9), adjust_slab_journal_block_reference(9), release_journal_locks(9), update_tail_block_location(9), reopen_slab_journal(9), complete_write(9), write_slab_journal_block(9), commit_tail(9), encode_slab_journal_entry(9), expand_journal_point(9), add_entry(9), vdo_attempt_replay_into_slab(9), requires_reaping(9), launch_reference_block_write(9), finish_reference_block_write(9), get_reference_counters_for_block(9), pack_reference_block(9), handle_io_error(9), write_reference_block(9), reference_count_to_status(9), dirty_block(9), get_reference_block(9), slab_block_number_from_pbn(9), get_reference_counter(9), adjust_free_block_count(9), increment_for_data(9), decrement_for_data(9), increment_for_block_map(9), update_reference_count(9), add_entry_from_waiter(9), is_next_entry_a_block_map_increment(9), add_entries(9), reset_search_cursor(9), advance_search_cursor(9), vdo_adjust_reference_count_for_rebuild(9), replay_reference_count_change(9), find_zero_byte_in_word(9), find_free_block(9), search_current_reference_block(9), search_reference_blocks(9), make_provisional_reference(9), dirty_all_reference_blocks(9), match_bytes(9), unpack_reference_block(9), finish_reference_block_load(9), load_reference_block_group(9), load_reference_blocks(9).

The list keeps going, don't worry, nobody's left out 🫢:

drain_slab(9), read_slab_journal_tail(9), load_slab_journal(9), get_next_slab(9), has_slabs_to_scrub(9), uninitialize_scrubber_vio(9), finish_scrubbing(9), slab_scrubbed(9), abort_scrubbing(9), handle_scrubber_error(9), apply_block_entries(9), apply_journal_entries(9), start_scrubbing(9), scrub_next_slab(9), scrub_slabs(9), get_depot_slab_iterator(9), next_slab(9), abort_waiter(9), vdo_acquire_provisional_reference(9), open_slab(9), vdo_enqueue_clean_slab_waiter(9), vdo_modify_reference_count(9), vdo_notify_slab_journals_are_recovered(9), make_slab(9), allocate_slabs(9), vdo_abandon_new_slabs(9), release_recovery_journal_lock(9), prepare_for_tail_block_commit(9), schedule_tail_block_commit(9), initialize_slab_scrubber(9), initialize_slab_summary_block(9), vdo_decode_slab_depot(9), vdo_free_slab_depot(9), vdo_record_slab_depot(9), vdo_allocate_reference_counters(9), get_slab_number(9), vdo_get_slab(9), vdo_get_increment_limit(9), vdo_is_physical_data_block(9), vdo_get_slab_depot_allocated_blocks(9), vdo_get_slab_depot_data_blocks(9), finish_combining_zones(9), combine_summaries(9), finish_loading_summary(9), load_slab_summary(9), vdo_load_slab_depot(9), vdo_prepare_slab_depot_to_allocate(9), vdo_update_slab_depot_size(9), vdo_prepare_to_grow_slab_depot(9).

And the last lil batch, main character energy all around πŸ‘‘:

finish_registration(9), vdo_use_new_slabs(9), stop_scrubbing(9), vdo_drain_slab_depot(9), resume_scrubbing(9), vdo_resume_slab_depot(9), vdo_commit_oldest_slab_journal_tail_blocks(9), vdo_scrub_all_unrecovered_slabs(9), get_block_allocator_statistics(9), get_ref_counts_statistics(9), get_slab_journal_statistics(9), vdo_get_slab_depot_statistics(9), vdo_dump_slab_depot(9).


Lowkey the whole slab-depot friend group is in this one. If you're touching refcounts, one of the above literally has your answer. iykyk πŸ’€βœ¨

🌈 That's the whole page, bestie! Source path ./drivers/md/dm-vdo, dated October 2026 β€” go off and count those references πŸ’–πŸŽ€πŸ¦‹

πŸ“¦ where this page comes from