This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Refactor expansion of -D and -I specs

Paolo Bonzini <> writes:

> This patch is the first of two patches to support top-level
> bootstrap in the GCC driver.  This one has no behavioral
> change, it only refactors gcc.c so that the same routine
> used to output multilib paths is used to output include
> paths.  The suffix used (like for example "nof" or "le"
> in ppc multilibs) is "include", so the include_prefixes
> will not end with "include" anymore.
> This part also includes the removal and poisoning of
> I also attach a sdiff between the old code for do_spec_1 and
> the new do_spec_path routine, with the whitespace a bit
> tweaked in the hope of aiding the review of the patch.
> You can see that multilib_dir becomes dir_for_machine_suffix,
> and multi_dir become dir_for_no_suffix.
> Bootstrapped/regtested i686-pc-linux-gnu, ok for mainline?
> Paolo
> 2004-04-30 Paolo Bonzini <>
> 	* gcc.c (struct prefix_list): Add forward declaration.
> 	(do_spec_path): New function, extracted from...
> 	(do_spec_1) <'D'>: ... here.  Drop support for
> 	(do_spec_1) <'I'>: Use do_spec_path.
> 	(process_command): Do not store the 'include' suffix
> 	in include_prefixes.
> 	* system.h: Poison SPACE_AFTER_L_OPTION.

This is OK.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]