Re: Booter for 9.3 kernel not working
Matthew Stock <[email protected]> Thu, 25 Jan 2024 23:09:14 -0500
| Newsgroups | gmane.os.netbsd.ports.mac68k |
|---|---|
| Message-ID | <CAJEih48u37wJ+wB0jyBkBVRH17MgVFVN7cn0ZyWkRz=ncFH9RA@mail.gmail.com> |
On Thu, Jan 25, 2024 at 3:21 PM Hauke Fath <[email protected]> wrote: > > ISTR Think C doesn't work any more, and you need Codewarrior (in the > right version to read the enclosed project file). That appears to be the case. I was able to get a build with CW and the 2.0.0.a10 source to build. I believe I have figured out the problem, and it essentially boils down to big numbers two ways. First, I was getting an lseek issue with the native filesystem image boot. I realized that it would boot, but only if the kernel was compressed. I believe this was in some way due to the file size, but I haven't confirmed that yet. All I can say is that I was able to get a compiled kernel to boot. So why wouldn't it work on the BSD filesystem? The standard kernel isn't compressed, so that isn't it. The Booter logs indicate that it was able to parse the filesystem, find the file, but that the magic number was all 0's vs the standard ELF. In this case, I think the problem is that for the default kernel, the inode number and/or the blocks used to store the kernel are small. When I built a new kernel, the blocks being used on the filesystem (which is fairly large) may have been beyond some power of 2, and the filesystem handling code couldn't find the executable and read it properly. Time permitting, I'll see if I can confirm these theories, and if I'm really lucky I'll be able to fix the filesystem code. It may be that I just need a smaller root filesystem. Thanks to everyone for assisting in getting this far, and with all of the code and the ability to rebuild Booter I feel like I should be able to resolve the issues I'm seeing. Matt