On Fri, 28 Mar 2025, Jason Merrill wrote:
> Tested x86_64-pc-linux-gnu, initially with extra checking to make sure that
> indeed nothing got saved from a namespace level.
>
> This isn't a regression, but a 20% speedup for a simple change is pretty
> attractive; what do people think about this change for GCC 15?
>
> -- 8< --
>
> Profiling showed that the loop to save away IDENTIFIER_BINDINGs from open
> binding levels was taking 5% of total compilation time in the PR116285
> testcase. This turned out to be because we were unnecessarily trying to do
> this for namespaces, whose bindings are found through
> DECL_NAMESPACE_BINDINGS, not IDENTIFIER_BINDING.
>
> As a result we would frequently loop through everything in std::, checking
> whether it needs to be stored, and never storing anything.
>
> This change actually appears to speed up compilation for the PR116285
> testcase by ~20%.
Nice, this also speeds up compilation of stdc++.h by ~20%! And std
module compilation by ~10%. +1
>
> The replaced comments referred either to long-replaced handling of classes
> and templates, or to wanting b to point to :: when the loop exits.
>
> PR c++/64500
> PR c++/116285
>
> gcc/cp/ChangeLog:
>
> * name-lookup.cc (push_to_top_level): Don't try to store_bindings
> for namespace levels.
> ---
> gcc/cp/name-lookup.cc | 20 ++++++++++++--------
> 1 file changed, 12 insertions(+), 8 deletions(-)
>
> diff --git a/gcc/cp/name-lookup.cc b/gcc/cp/name-lookup.cc
> index 7f1ee869d52..61b7bfcaf94 100644
> --- a/gcc/cp/name-lookup.cc
> +++ b/gcc/cp/name-lookup.cc
> @@ -8675,6 +8675,9 @@ store_class_bindings (vec<cp_class_binding, va_gc>
> *names,
>
> static GTY((deletable)) struct saved_scope *free_saved_scope;
>
> +/* Temporarily make the current scope the global namespace, saving away
> + the current scope for pop_from_top_level. */
> +
> void
> push_to_top_level (void)
> {
> @@ -8716,18 +8719,19 @@ push_to_top_level (void)
> store_class_bindings (previous_class_level->class_shadowed,
> &s->old_bindings);
>
> - /* Have to include the global scope, because class-scope decls
> - aren't listed anywhere useful. */
> + /* Save and clear any IDENTIFIER_BINDING from local scopes. */
> for (; b; b = b->level_chain)
> {
> tree t;
>
> - /* Template IDs are inserted into the global level. If they were
> - inserted into namespace level, finish_file wouldn't find them
> - when doing pending instantiations. Therefore, don't stop at
> - namespace level, but continue until :: . */
> - if (global_scope_p (b))
> - break;
> + /* We don't need to consider namespace scopes, they don't affect
> + IDENTIFIER_BINDING. */
> + if (b->kind == sk_namespace)
> + {
> + /* Jump straight to '::'. */
> + b = NAMESPACE_LEVEL (global_namespace);
> + break;
> + }
>
> store_bindings (b->names, &s->old_bindings);
> /* We also need to check class_shadowed to save class-level type
>
> base-commit: d9b56c65a2697e0d7a6c0f15f1977803dc94579b
> --
> 2.49.0
>
>
