I'm reading a white paper published by Sun regarding the porting efforts from SPARC to x86. I summarized as below but has questions inlined.
Easy part (1) If the application is in Java language, then it's non-port. (2) For driver code, if it uses standard DDI/DDK kernel interface, it's just recompile, no need to change. Defore Sun open source the code 2+ years ago, is it likely to develop such driver using non-standard kernel interface for a driver? (3) If the application on SPARC using published Solaris OS interfaces on SPARC, it doesn't need any change, just recompile. (4) If the application using Sun Studio to compile, no need to change compiler options, only test options to optimize for x86. How about those compiled using gcc on Solaris/SPARC, are there big efforts to modify the compiler options if keep using the same gcc on x86? Modifications required (1) Needs to check if their apps depend on any third-part apps/libs/tools, if so make sure these apps are supported on both SPARC and x86. (2) If there is SPARC assemble code in apps, they need to be rewritten (better to move from Linux/x86). (3) byte-order code (big/little endian) needs to be changed. Are there any code examples of this? (4) Platform dependent code like "#ifdef SPARC" code. What are the typical example of platform (SPARC vs. x86) dependent code? (5) what are the other efforts may required? (6) If port from Solaris9/SPARC, does it take more efforts compared with porting from Solaris10/SPARC to Solaris10/x86? Any inputs or suggestions that helps to scope the efforts will be very welcome!
_______________________________________________ opensolaris-discuss mailing list [email protected]
