>>> On 9/18/2008 at 12:50 PM, in message <[EMAIL PROTECTED]>, Lindy Mayfield <[EMAIL PROTECTED]> wrote: -snip- > I asked the lead developer how he started something like that and the > first step was to get gcc to produce z object code. I was thinking > today, but how to compile the changes to the compiler?
If you use gcc, then you take a working gcc package, and the gcc source, and recompile it to be a cross compiler. That is, the newly created compiler runs on whatever platform you currently have, but generates code for the target system. You also have to build cross versions of the binutils package, since they're part of the "tool chain," and contain the assembler and linker. After that, you also have to build a cross version of glibc, which is the hardest part of all. Then, you use the cross versions of all the tool chain to start compiling things like the kernel/nucleus/whateveryoucallit, a "native" version of the tool chain and glibc, and a shell to run it all. Then you transfer these to the target system and try to bring it up. Once there, the first thing I did when I was building a 64-bit version of Slack/390 was to re-build the entire tool chain, and then start creating all the packages that made up the OS. Now, this was just going from 31-bit Linux to 64-bit Linux for the mainframe. That meant a lot of packages would run in 31-bit mode on the 64-bit kernel, which let me start building things natively a lot sooner than I would have been able to if I had been going to a totally different architecture. Even so, getting to that point was a _giant_ pain in the neck. Going from SPARC to System z would have been orders of magnitude more so. The primary developer, Neale Ferguson, had also worked on the Bigfoot port of Linux to S/370 prior to IBM releasing it's patches. So, he's been through this more than once. He has far more brain power (and tolerance for pain) than I do, that's for certain. Mark Post ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html