Re: 8.1 boot failure on Challenge S

[Michael <macallan%netbsd.org@localhost>]

Hello,

On Sat, 16 Nov 2019 20:20:40 +0000
George Harvey <fr30%dial.pipex.com@localhost> wrote:

> Since the failure occurred after finding hpc1, which isn't present on
> an Indy, I built a custom kernel with hpc1 disabled. That booted and
> ran normally on the Challenge S so it looks like the problem is related
> to the second hpc. What can I do to narrow it down further?

Hmm, might be fallout from the bus space rework I did a while ago -
this is a hardware combination I was unable to test.

> mode status=0x20040003, cause=0xc008, epc=0x880b59e4, vaddr=0
> tf=0x88061848 ksp=0x880618e8 ra=0x880b59d8 ppl=0x8
> kernel: TLB miss (load or instr. fetch) trap
> Stopped in pid 0.1 (system) at  880b59e4:       lw      a0,0(v1)

That's a NULL pointer access. You could load the kernel into gdb and
check which function 0x880b59e4 belongs to, then sprinkle 
printf("%s %d\n", __func__, __LINE__);
all over it to see where exactly we crash.

> Compiling the kernel threw a very large number of warnings at link
> stage but the resulting kernel works. I'll put the details in a
> separate thread.

softfloat vs. hardfloat?
The kernel is built as a softfloat binary, but in order to mess with
the FPU some files need to be built with -mhard-float.

have fun
Michael