This is the mail archive of the
mailing list for the GCC project.
Re: RFA: Use .opt files for the ARM port
- From: Richard Earnshaw <rearnsha at gcc dot gnu dot org>
- To: Richard Sandiford <rsandifo at redhat dot com>
- Cc: gcc-patches <gcc-patches at gcc dot gnu dot org>
- Date: Wed, 04 May 2005 11:40:49 +0100
- Subject: Re: RFA: Use .opt files for the ARM port
- References: <email@example.com>
On Sat, 2005-04-30 at 20:44, Richard Sandiford wrote:
> This patch makes the ARM port define its options using .opt files.
> Main points:
> - Most mask macros were called things like ARM_FLAG_* or THUMB_FLAG_*
> and often didn't use the same suffix as the associated TARGET_* macro.
> The patch uses the standard MASK_* names instead. Various definitions
> of TARGET_DEFAULT have to change as a result.
> - -mtpcs-frame and -mtpcs-leaf-frame were associated with
> THUMB_FLAG_BACKTRACE and THUMB_FLAG_LEAF_BACKTRACE but
> had no TARGET_* macro. The patch changes them to use
> TARGET/MASK_TPCS_FRAME and TARGET/MASK_TPCS_LEAF_FRAME
> respectively. (This avoids a potential collision with
> TARGET_BACKTRACE, which is defined in terms of both options.)
> - Most members of TARGET_OPTIONS have some interrelationship with
> other options. I've therefore left the parsing and checking code
> in its current place (override_options) and simply made handle_option
> set the same strings as TARGET_OPTIONS did. The two exceptions are
> -mhard-float and -msoft-float, which are now true aliases for
> -mfloat-abi=hard and -mfloat-abi=soft respectively.
> Tested by building cc1 for arm-elf, arm-pe, arm-wince-pe, strongarm-pe
> and xscale-coff. I checked each option by hand to make sure it had
> the intended effect. I also checked for stray warnings. OK to install?
One quick question.
Is it not possible to merge:
+Target RejectNegative Report Mask(NO_SCHED_PRO)
+Do not move instructions into a function's prologue
+Target RejectNegative InverseMask(NO_SCHED_PRO) Undocumented
into a single flag:
TARGET Report InverseMask(NO_SCHED_PRO)
Permit scheduling of a function's prologue sequence