This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: Changes for a `m68k-motorola-sysv'
- To: law at cygnus dot com, manfred at s-direktnet dot de
- Subject: Re: Changes for a `m68k-motorola-sysv'
- From: meissner at cygnus dot com
- Date: Sun, 7 Sep 1997 21:08:39 -0400 (EDT)
- Cc: egcs-bugs at cygnus dot com
| >
| > #ifndef TARGET_DEFAULT
| > #define TARGET_DEFAULT (MASK_68040|MASK_BITFIELD|MASK_68881|MASK_68020)
| > #endif ^^^^^^^^^^
| >
| > how do you assume that these bits will be removed from the default,
| > if gcc will be called with an argument specifying a cpu that doesn't
| > actually have such an fpu?
| I would think the right thing to do is to have t-mot3300 arrange for
| -msoft-float to be included when building multilibs for the pure
| 68000.
|
| I'm not that familiar with the multilib stuff, but I would think this
| something it should be able to handle.
Yes, MULTILIB_MATCHES is what you want. For example on the rs6000/powerpc
configs, MULTILIB_MATCHES is eventually set to:
msoft-float=mcpu?403 \
msoft-float=mcpu?821 \
msoft-float=mcpu?860 \
mlittle=mlittle-endian \
mbig=mbig-endian \
mcall-sysv=mcall-sysv-eabi \
mcall-sysv=mcall-sysv-noeabi
This says that if the compiler sees -mcpu=403, it should assume that it also
sees -msoft-float (the '?' is replaced by an '=').