Re: Anybody working on xfree 3.3.2??
Hi,
>> Where does that kernel get stuck?
>
>Right after amiboot has loadad the kernel - there is not a singe line
>of text output.
That leaves quite some margin. Anything between head.S and console_init :-)
>> The one thing that seems likely is inline asm wich isn't often used
>> in normal binaries.
>
>I think the problem is really related to asm: The following object files
>are different when building the kernel using binutils 2.9-0.1:
>
>arch/m68k/ifpsp060/fskeleton.o
>arch/m68k/ifpsp060/iskeleton.o
>arch/m68k/ifpsp060/os.o
Saw a posting on c.o.l.m68k today - Jes suggested to use newer binutils than
the ones on watchtower. But I'd hope that the 2.9 binutils are fine....
>arch/m68k/kernel/entry.o
>arch/m68k/kernel/kernel.o
>arch/m68k/kernel/head.o
>arch/m68k/lib/semaphore.o
entry.S and head.S are pure assembler files, yet different? Can you objdump -d
them and compare the disassembly?
It seems the 2.9 as is broken then. Nothing to do with inline, just plain
problem with as.
>drivers/scsi/wd33c93.o
Strange - haven't seen inline asm in there except for save_flags and friends.
And if these are different, pretty much all drivers are affected.
>init/version.o
That's normal - you probably didn't reset the compile.h version number.
I'd focus on differences in head.o and entry.o first. You could hardcode your
screen base address (if it's constant) in head.S and plot patterns on the
screen to trace where the kernel stops there (see the Mac head.S for ideas).
Or use the serial port debug option in head.S.
Michael
--
To UNSUBSCRIBE, email to debian-68k-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: