| Re: [hatari-devel] Hatari and OUTSIDE (virtual memory manager) |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/hatari-devel Archives
]
- To: hatari-devel@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [hatari-devel] Hatari and OUTSIDE (virtual memory manager)
- From: Uwe Seimet <Uwe.Seimet@xxxxxxxxx>
- Date: Mon, 8 Oct 2018 22:51:50 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; t=1539031911; s=strato-dkim-0002; d=seimet.de; h=In-Reply-To:References:Message-ID:Subject:To:From:Date: X-RZG-CLASS-ID:X-RZG-AUTH:From:Subject:Sender; bh=sHI311XT3BFGjJDjcXIjz3fxvnvlUVR64BzckVTDJZ4=; b=L06aO8+Ophj5+pF33v3wcyLZOP/mKJQ86w5zmzM0SQZ2f9bpTHiV+0FGmahXA16tLU oajWlKqBrXE8SydBd8BQXpwr9GRAegz0fCvSgsQhGfV5pta1RULo8FuVcKbWEG3ZIibj /2ZAYQd+k14ukDqUDJ/xM0AkVf7+F/8AQwpSrg+Z0qDyy03uZMVDpujTX6En1PKNscHt lxrj66k8q64xQQYF+8N5w3WMI+rFnXa8nTDbXRPL5F1XdgK/6/jlLDV8eF+9dY9qUy+t nr4WMA18yg4yBJhmDmtU5eRsYDiI7hFG8AjBuN2iQZgebJmKZvp+D4BmNTdaBqqiW9Oq ggVg==
Hi,
Don't understand me wrong, but what just surprises me is that simply trying
to launch two programs using the PMMU has already revealed several issues
with the PMMU code.
Regarding virtual memory, I would expect NeXTstep not to differ a lot
from OUTSIDE (or VRAM, which is the other virtual memory manager for the
Atari), because all in all there are not that many choices on how to
implement virtual memory. But for NeXTstep it was probably more
straightforward because in contrast to TOS it was designed (at least I
assume so) to use virtual memory.
You only had access to the data sheets, not to the full manuals? I
haven't looked at the data sheets for ages, but if I remember correctly
they contain much less information than the manuals, which nowadays can
be downloaded.
It's not possible to trace PMMU usage with Hatari, is it? This might
help to identify problems, in particular with software like ASV where no
sources are available.
Best regards
Uwe
>
> Am 08.10.2018 um 19:11 schrieb Uwe Seimet <Uwe.Seimet@xxxxxxxxx>:
>
> > Hi,
> >
> >> did you try outside too ?
> >
> > OUTSIDE is very instable, crashes the system very often. Considering
> > that OUTSIDE make much more use of the PMMU than ROMSPEED does, it's
> > unlikely that OUTSIDE will work when ROMSPEED does not.
> >
> > I think I already asked this, but no one responded: Is the PMMU
> > emulation actually complete, including function codes, master stack pointer,
> > stack frames etc.? How has the emulation been tested?
> >
> 68030 MMU emulation was made for Previous (NeXT Computer emulator), but I did not intend to write some minimalistic emulation just to make NeXTstep happy. When I wrote 68030 MMU emulation I tried to make it complete using Motorola's data sheets. Some code was added later by Toni: checking for illegal instructions (addressing, illegal combinations of parameters) and handling of cache (Previous does not use cache emulation).
> Of course I can't guarantee that some bug hides in the code, but there are no parts missing intentionally. NeXTstep uses quite a lot of the MMU features and works fine.
>