This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: RX cleanup rx_option_optimization
- From: Nick Clifton <nickc at redhat dot com>
- To: "Joseph S. Myers" <joseph at codesourcery dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 15 Oct 2010 09:13:15 +0100
- Subject: Re: RX cleanup rx_option_optimization
- References: <Pine.LNX.4.64.1010150037540.2442@digraph.polyomino.org.uk>
Hi Joseph,
In general this setting -ffast-math seems suspicious. What exactly
are the non-IEEE properties of the RX FPU? Is it really the case that
when results are normal, finite values, the FPU will still return
inaccurate results?
No.
> If not, then only the relevant flags should be
set, not the whole of -ffast-math. For example, if it doesn't support
NaNs and infinities, set flag_finite_math_only;
This is what should be happening. If you would care to make that change
to your patch then that is pre-approved.
2010-10-14 Joseph Myers<joseph@codesourcery.com>
* config/rx/rx.c (rx_option_optimization): Change to
rx_override_options_after_change. Don't change
flag_lto_compression_level. Don't check for changes to whether
FPU instructions can be used.
(rx_option_override): Call rx_override_options_after_change.
(TARGET_OPTION_OPTIMIZATION): Remove.
(TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
Approved.
Cheers
Nick