Issues with porting sparc Solaris 8 application to intel Solaris 10.

From what I know of the Solaris architecture after kicking around in it for about 10 years, I don't think this will be very difficult. Of Solaris itself, only about 5% of the code is hardware/CPU-architecture-dependent. The rest is hardware/CPU-architecture-independent. Unless the application in question is making some deep system calls, and hitting that 5% of the OS, I'm not aware of any particular roadblocks from the OS.

This is, of course, affected by how your application is coded/architected. If it likes to do a lot of Ring 0 work, if it likes to talk to hardware directly, then porting will be hard/very difficult. But if it lives in Ring 3 and talks to the OS as opposed to the hardware, then my estimation is it should be fairly easy.

Disclaimer: IANAPAM (I Am Not A Programmer Any More)

Thanks for you response. So what are guidelines you suggest porting my application from Sparc Solaris to Intel Solaris and are there known issues to be taken into consideration? I tried searching this on net but failed to find good pointers for porting Sparc Solaris application to Intel Solaris. What I am really worried about is something like endianness (and other such issue which I might not even be aware of), which is different on Sparc Solaris then on Intel Solaris. What effect does endianness has on something like socket programming. Since endianness of bytes received from network on Sparc Solaris doesn’t has to be changed to use, but endianness of the bytes received from network on Intel Solaris has to be changed. Really appreciate a quick response on this.

PsiCop,

How do I find if my code is doing lot of Ring 0/hardware/deep system calls. I am new to Solaris and am not familiar with the calls. Can you provide something like list of calls - & we could simply grep -r through the code.

-Sunil