On Tue, Sep 15, 2026 at 10:32:36AM +0000, [email protected]
wrote:
> gcc/ChangeLog:
>
> * ipa-inline.cc (report_inline_failed_reason): Check for NULL
> before calling cl_target_option_print_diff.
>
> diff --git gcc/ipa-inline.cc gcc/ipa-inline.cc
> index 5a7615666cc..111593f45fd 100644
> --- gcc/ipa-inline.cc
> +++ gcc/ipa-inline.cc
> @@ -294,10 +294,14 @@ report_inline_failed_reason (struct cgraph_edge *e)
> e->callee->ultimate_alias_target
> ()->lto_file_data->file_name);
> }
> if (e->inline_failed == CIF_TARGET_OPTION_MISMATCH)
> - if (dump_file)
> - cl_target_option_print_diff
> - (dump_file, 2, target_opts_for_fn (e->caller->decl),
> - target_opts_for_fn (e->callee->ultimate_alias_target ()->decl));
> + {
> + struct cl_target_option *opt_caller
> + = target_opts_for_fn (e->caller->decl);
> + struct cl_target_option *opt_callee
> + = target_opts_for_fn (e->callee->ultimate_alias_target ()->decl);
> + if (dump_file && opt_caller != NULL && opt_callee != NULL)
> + cl_target_option_print_diff (dump_file, 2, opt_caller, opt_callee);
> + }
Isn't it a waste of compile time to call target_opts_for_fn when it actually
isn't needed in the dump_file == NULL case?
Wouldn't it be better to
add && dump_file to the if (e->inline_failed == CIF_TARGET_OPTION_MISMATCH)
condition and leave it out from the inner one?
> if (e->inline_failed == CIF_OPTIMIZATION_MISMATCH)
> if (dump_file)
> cl_optimization_print_diff
Jakub
