Re: Debian on mac68k
On Sat, 20 Feb 2016, Mac User wrote:
> I've installed Debian 3.1 (sarge) on a Mac IIfx with 32 MB memory and on
> two Mac LC III systems each with 36 MB memory.
The LC III uses Egret ADB, and the driver for that chip is known to be
unstable. Both the LC III and the IIfx use the NCR 5380 SCSI driver, which
is a work-in-progress. PDMA on the LC III is broken and DMA on the IIfx
was never implemented, so SCSI throughput is low. These are both known
issues, and are documented at http://www.mac.linux-m68k.org/status/
> The installations (text only, no X-Windows, Linux 2.2.25 kernel) are
> relatively stable (more on that later).
Linux v2.2 has many known bugs that have been fixed in v2.6 or v3. I would
advise that you avoid old kernels.
> I read that there's an effort to restart Debian support for mac68k after
> Debian 3.1, but I haven't found much information about it.
There is no Debian port for Apple Mac 68k hardware as such. There is a
Debian port for the m68k architecture, which covers hardware from Apollo,
Apple, Atari, Commodore, NeXT, Sun etc. Debian/m68k is actively developed.
Please see https://wiki.debian.org/M68k
> I'm currently compiling Linux kernel 3.19.8 on the IIfx, but I may need
> binutils > 2.15, so I'm compiling binutils-2.26 on an LC III. The LC
> III successfully compiled Linux kernel 18.104.22.168 but I couldn't get it to
> boot (may be an issue with binutils).
> If there's an effort to build packages for etch or later, I'd like to
> help, though I currently don't know much about how to build Debian
> packages. And I don't have a cross compiler for mac68k, only the IIfx
> and LC III systems.
Modern compilers are more CPU intensive, package code bases have grown
(requiring more RAM to build), and developers continue to adopt new
compiler features. So native compilation on hardware like yours is not
realistic in many cases.
If you want to build your own kernels, I suggest you obtain a cross
toolchain from kernel.org: https://www.kernel.org/pub/tools/crosstool/
> I was surprised that the IIfx is considerably slower than the LC III for
> almost everything, and the internal clock on the IIfx comes to a near
> stop when its CPU is busy (not so bad on the LC III). All the systems
> have SCSI stability issues if two or more disks are connected (phase
> errors, timeouts). Hopefully these will be fixed in some of the later
The scsi stability issues can be resolved with the right patches. A
timekeeping issue suggests high timer interrupt latency. If this problem
affects recent kernels, we could look into it. If you are willing to test
a recent kernel binary, I'd be happy to build one for you.