[Simh] gcc c99 errors in vaxmod_defs.h and vax780_defs.h

[Gregory Wright]

Hi,

On Jul 21, 2008, at 11:20 AM, HoffmanLabs wrote:

>
> Mac OS X 10.5.4 Leopard with Xtools 3.1 (current) with gcc 4.0.1.
>
> This code:
>
> /* Function prototypes for physical memory interface (inlined) */
>
> SIM_INLINE int32 ReadB (uint32 pa);
> SIM_INLINE int32 ReadW (uint32 pa);
> SIM_INLINE int32 ReadL (uint32 pa);
> SIM_INLINE int32 ReadLP (uint32 pa);
> SIM_INLINE void WriteB (uint32 pa, int32 val);
> SIM_INLINE void WriteW (uint32 pa, int32 val);
> SIM_INLINE void WriteL (uint32 pa, int32 val);
> ...
>
> runs afoul of c99 rules for inline, and gcc 4.0.1 is now (per its
> release notes) now enforcing these rules.
>
> Compilation errors are akin to the following:
>
> VAX/vaxmod_defs.h:455: warning: inline function ‘WriteB’ declared but
> never defined
> VAX/vaxmod_defs.h:454: warning: inline function ‘ReadLP’ declared but
> never defined
> VAX/vaxmod_defs.h:453: warning: inline function ‘ReadL’ declared but
> never defined
> ....
>
> I've implemented the following brute-force hack (in vaxmod_defs.h and
> vax780_defs.h) to get the code to build and run, but it's certainly  
> not
> the right fix.
>
> /* Function prototypes for physical memory interface (inlined) */
>
> #ifdef SIM_INLINE
> #undef SIM_INLINE
> #define SIM_INLINE extern
> #endif /* Hoff */
> SIM_INLINE int32 ReadB (uint32 pa);
> SIM_INLINE int32 ReadW (uint32 pa);
> SIM_INLINE int32 ReadL (uint32 pa);
> SIM_INLINE int32 ReadLP (uint32 pa);
> SIM_INLINE void WriteB (uint32 pa, int32 val);
> SIM_INLINE void WriteW (uint32 pa, int32 val);
> SIM_INLINE void WriteL (uint32 pa, int32 val);
> ...
>
>
> $ gcc --version
> i686-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1 (Apple Inc. build 5484)
>
> I've a set of Mac OS X build and install directions posted which  
> include
> this detail, and am working on some related pieces.
>
> I've also tweaked the local version of the build slightly as libpcap  
> is
> around.
>
> I don't immediately see any list archives referenced, so this c99-ism
> may be known and discussed elsewhere.
>
>

The issue came up last month, when I was porting the simh 3.8 to  
MacPorts.
Here's the reference:

http://mailman.trailing-edge.com/pipermail/simh/2008-June/001936.html

Something along the lines of the patch shown is probably the right  
thing to
use, but it needs to be checked for compatibility earlier gcc releases.

Greg