Re: virt68k/68030 problem

[Jason Thorpe <thorpej%me.com@localhost>]

> On Jan 17, 2024, at 5:30 AM, Tetsuya Isaki <isaki%pastel-flower.jp@localhost> wrote:
> 
> Hello,
> 
> virt68k started to boot on my 68030 emulator (not qemu).
> But it does not set TT register for I/O space in 68030 case,
> So that it lost console output just after mmu enabled.
> The attached patch solves this issue.

Oh, that’s great!

> The WIP screen shot is https://misskey.io/notes/9olbbm9b1hg00cuy
> 
> May I commit it? (or would you take?)

Please go ahead.  Although, are the “.long” statments necessary?  Or is the assembler not configured correctly to take the pmove statements into tt0 directly?  I assume you also have associated changes for pmap.h and genassym.cf.

Thanks!

> ---
> Tetsuya Isaki <isaki%pastel-flower.jp@localhost / isaki%NetBSD.org@localhost>
> 
> Index: sys/arch/virt68k/virt68k/locore.s
> ===================================================================
> RCS file: /cvsroot/src/sys/arch/virt68k/virt68k/locore.s,v
> retrieving revision 1.9
> diff -u -r1.9 locore.s
> --- sys/arch/virt68k/virt68k/locore.s 13 Jan 2024 17:10:58 -0000 1.9
> +++ sys/arch/virt68k/virt68k/locore.s 17 Jan 2024 13:04:00 -0000
> @@ -199,8 +199,13 @@
> jmp Lenab1
> 
> Lmotommu2:
> + movl #VIRT68K_TT30_IO,%sp@- | TT0 maps the I/O space
> + .long 0xf0170800 | pmove %sp@,mmutt0
> + clrl %sp@ | ensure TT1 is disabled
> + .long 0xf0170c00 | pmove %sp@,mmutt1
> +
> pflusha
> - movl #MMU51_TCR_BITS,%sp@- | value to load TC with
> + movl #MMU51_TCR_BITS,%sp@ | value to load TC with
> pmove %sp@,%tc | load it
> 
> /*
> 

-- thorpej