#9053: BUG gmp: crash in config.guess due to mfpvr instruction ----------------------------------------+----------------------------------- Reporter: vincent-opdarw@… | Owner: gwright@… Type: defect | Status: assigned Priority: Low | Milestone: Port Bugs Component: ports | Version: 1.0 Resolution: | Keywords: Port: gmp | ----------------------------------------+----------------------------------- Comment(by mcalhoun@…): Replying to [comment:21 vinc17@…]:

Replying to [comment:20 mcalhoun@…]:

...

-force_cpusubtype_ALL is required on PPC architectures or else gmp won't work properly.

What is the cause? (I think it may be related to broken CPU subtype detection.) In the configure script is the following: {{{ *-*-darwin*) # On Darwin we can use 64-bit instructions with a longlong limb, # but the chip still in 32-bit mode. # In theory this can be used on any OS which knows how to save # 64-bit registers in a context switch. # # Note that we must use -mpowerpc64 with gcc, since the # longlong.h macros expect limb operands in a single 64-bit # register, not two 32-bit registers as would be given for a # long long without -mpowerpc64. In theory we could detect and # accomodate both styles, but the proper 64-bit registers will # be fastest and are what we really want to use. # # One would think -mpowerpc64 would set the assembler in the right # mode to handle 64-bit instructions. But for that, also # -force_cpusubtype_ALL is needed. }}} So it is possible that -force_cpusubtype_ALL is required to get the PowePC to be in the correct state.[[BR]] I admit I do not understand the intricacies of processor architectures.

I think the bottom line is that as far a I know, gmp builds and runs just as the developers intended.[[BR]] If there is a particular patch, however, we can reopen the discussion. -- Ticket URL: <http://trac.macports.org/ticket/9053#comment:22> MacPorts <http://www.macports.org/> Ports system for Mac OS