Re: [PATCH v1 0/7] perf symbol: Reference counting, flat array storage, and LRU shrinking

[Ian Rogers]

On Mon, Sep 28, 2026 at 12:52 AM Ian Rogers <irogers@xxxxxxxxxx> wrote:
>
> This series modernizes struct symbol memory management and backing storage in
> tools/perf to support reference counting (DECLARE_RC_STRUCT(symbol)) and
> periodic LRU memory shrinking during long or large profiling sessions:
>
> 1. Field accessors (Patch 1): Introduces inline getters/setters for
>    struct symbol fields across tools/perf in preparation for wrapping
>    struct symbol with DECLARE_RC_STRUCT(symbol).
> 2. Remove negative-offset private allocations (Patch 2): Replaces
>    symbol_conf.priv_size and symbol__priv() negative-offset allocations in
>    util/annotate.c and util/db-export.c with hashmaps keyed by symbol
>    pointer (struct annotated_symbol protected by an rw_semaphore, and
>    sym_db_ids on struct db_export).
> 3. Flat sorted array backing storage (Patch 3): Replaces the dual
>    address-sorted and name-sorted red-black trees (rb_node) in struct dso
>    with a struct symbols container backed by dynamically sorted flat arrays
>    protected by struct rw_semaphore (util/rwsem.h) with Clang thread-safety
>    annotations. Saves 48 bytes per struct symbol (two rb_nodes) and provides
>    cache-friendly binary search (bsearch) and callback-based iteration.
> 4. Reference counting and DECLARE_RC_STRUCT(symbol) (Patch 4): Wraps
>    struct symbol with DECLARE_RC_STRUCT(symbol) and adds refcount_t refcnt
>    along with symbol__get(), symbol__put(), and symbol__zput(). Lookup
>    functions return a counted reference acquired under symbols->lock, and
>    lifetime holders (addr_location, map_symbol, hist_entry, call_path,
>    block_info, disasm_line, thread_stack, db_export, etc.) properly
>    manage symbol references.
> 5. LRU memory shrinking (Patch 5): Adds an accessed bit to struct dso
>    and implements symbols__shrink(), dso__shrink_symbols(), dsos__shrink(),
>    machine__shrink(), and machines__shrink(). Unreferenced symbols (refcnt == 1)
>    in cold DSOs are reclaimed, and shrunken DSOs transparently reload on demand
>    via dso__load() while preserving live symbols and their RC_STRUCT(symbol)
>    identity.
> 6. Periodic session shrinking (Patch 6): Hooks machines__shrink() into
>    perf_session event processing so long-running sessions periodically reclaim
>    cold symbols and DSOs.
> 7. Unit tests (Patch 7): Adds unit tests in tools/perf/tests/symbols.c
>    covering symbol reference counting, symbols__shrink() retention of active
>    symbols, dsos__shrink() two-pass LRU clock behavior, and on-demand reload
>    preserving live symbol identity.

This needs a rebase to after the changes to the python integration,
hence no Sashiko reviews. I'll do that asap.

Thanks,
Ian

> Ian Rogers (7):
>   perf symbol: Add accessor functions for struct symbol fields
>   perf symbol: Remove symbol_conf.priv_size and negative-offset
>     allocations
>   perf symbol: Switch backing storage from rbtree to struct symbols
>     array
>   perf symbol: Add reference counting and DECLARE_RC_STRUCT(symbol)
>   perf symbol: Add LRU memory shrinking for symbols, DSOs, and machines
>   perf session: Periodically shrink symbols and DSOs during event
>     processing
>   perf test symbols: Add tests for symbol and DSO LRU shrinking
>
>  .../arch/powerpc/util/skip-callchain-idx.c    |   2 +-
>  tools/perf/arch/powerpc/util/sym-handling.c   |  40 +-
>  tools/perf/builtin-annotate.c                 |  15 +-
>  tools/perf/builtin-c2c.c                      |  14 -
>  tools/perf/builtin-diff.c                     |   5 +-
>  tools/perf/builtin-kallsyms.c                 |  10 +-
>  tools/perf/builtin-kmem.c                     |  66 +-
>  tools/perf/builtin-kwork.c                    |   4 +-
>  tools/perf/builtin-lock.c                     |  27 +-
>  tools/perf/builtin-mem.c                      |   2 +-
>  tools/perf/builtin-report.c                   |  14 -
>  tools/perf/builtin-sched.c                    |   6 +-
>  tools/perf/builtin-script.c                   |  18 +-
>  tools/perf/builtin-timechart.c                |   2 +-
>  tools/perf/builtin-top.c                      |  29 +-
>  tools/perf/builtin-trace.c                    |  12 +-
>  tools/perf/tests/builtin-test.c               |   1 -
>  tools/perf/tests/dlfilter-test.c              |  12 +-
>  tools/perf/tests/dwarf-unwind.c               |   2 +-
>  tools/perf/tests/hists_common.c               |   4 +-
>  tools/perf/tests/hists_cumulate.c             |   8 +-
>  tools/perf/tests/hists_filter.c               |  10 +-
>  tools/perf/tests/hists_link.c                 |  16 +-
>  tools/perf/tests/hists_output.c               |   9 +-
>  tools/perf/tests/kallsyms-split.c             |   5 +-
>  tools/perf/tests/pe-file-parsing.c            |   1 +
>  tools/perf/tests/symbols.c                    | 289 +++++-
>  tools/perf/tests/vmlinux-kallsyms.c           | 191 ++--
>  tools/perf/ui/browsers/annotate.c             |  11 +-
>  tools/perf/ui/browsers/hists.c                |  17 +-
>  tools/perf/ui/browsers/map.c                  |  87 +-
>  tools/perf/ui/gtk/annotate.c                  |   7 +-
>  tools/perf/ui/stdio/hist.c                    |   6 +-
>  tools/perf/util/addr_location.c               |   4 +
>  .../perf/util/annotate-arch/annotate-alpha.c  |   2 +-
>  .../util/annotate-arch/annotate-loongarch.c   |   8 +-
>  tools/perf/util/annotate-arch/annotate-s390.c |   2 +-
>  tools/perf/util/annotate-arch/annotate-x86.c  |  14 +-
>  tools/perf/util/annotate-data.c               |  14 +-
>  tools/perf/util/annotate.c                    | 211 ++++-
>  tools/perf/util/annotate.h                    |   6 +-
>  tools/perf/util/auxtrace.c                    | 111 ++-
>  tools/perf/util/block-info.c                  |  13 +-
>  tools/perf/util/bpf_lock_contention.c         |  39 +-
>  tools/perf/util/c2c-function.c                |  25 +-
>  tools/perf/util/call-path.c                   |  15 +-
>  tools/perf/util/callchain.c                   |  19 +-
>  tools/perf/util/capstone.c                    |  21 +-
>  tools/perf/util/data-convert-json.c           |   4 +-
>  tools/perf/util/db-export.c                   |  76 +-
>  tools/perf/util/db-export.h                   |   6 +
>  tools/perf/util/debug.c                       |   3 +-
>  tools/perf/util/disasm.c                      |  57 +-
>  tools/perf/util/dlfilter.c                    |  12 +-
>  tools/perf/util/dso.c                         | 236 +++--
>  tools/perf/util/dso.h                         |  85 +-
>  tools/perf/util/dsos.c                        | 100 +++
>  tools/perf/util/dsos.h                        |  12 +-
>  tools/perf/util/event.c                       |  12 +-
>  tools/perf/util/evsel_fprintf.c               |   2 +-
>  tools/perf/util/hist.c                        |  30 +-
>  tools/perf/util/intel-pt.c                    |  74 +-
>  tools/perf/util/libbfd.c                      |   5 +-
>  tools/perf/util/libdw.c                       |   5 +-
>  tools/perf/util/llvm.c                        |   6 +-
>  tools/perf/util/lock-contention.c             |   8 +-
>  tools/perf/util/machine.c                     | 143 ++-
>  tools/perf/util/machine.h                     |   2 +
>  tools/perf/util/map.c                         |  53 +-
>  tools/perf/util/map.h                         |  18 +-
>  tools/perf/util/map_symbol.c                  |   4 +-
>  tools/perf/util/maps.c                        |   3 +-
>  tools/perf/util/probe-event.c                 | 135 +--
>  tools/perf/util/python.c                      |  20 +-
>  .../util/scripting-engines/trace-event-perl.c |   8 +-
>  .../scripting-engines/trace-event-python.c    |  38 +-
>  tools/perf/util/session.c                     |  31 +
>  tools/perf/util/session.h                     |  13 +
>  tools/perf/util/sort.c                        |  44 +-
>  tools/perf/util/srcline.c                     |  19 +-
>  tools/perf/util/symbol-elf.c                  |  22 +-
>  tools/perf/util/symbol.c                      | 845 ++++++++++++------
>  tools/perf/util/symbol.h                      | 203 ++++-
>  tools/perf/util/symbol_conf.h                 |   1 -
>  tools/perf/util/symbol_fprintf.c              |  14 +-
>  tools/perf/util/thread-stack.c                |  18 +-
>  tools/perf/util/threads.c                     |  17 +-
>  tools/perf/util/unwind-libdw.c                |   4 +-
>  tools/perf/util/unwind-libunwind.c            |   2 +-
>  89 files changed, 2644 insertions(+), 1192 deletions(-)
>
> --
> 2.56.0.rc1.315.gc6ed9934b7-goog
>