basilisk-devel — Basilisk II development-related discussion

Re: [B2-devel] Re: Any use of MMU emulation?

[Christian B. <cb...@st...>]

Hi!

On Fri, Jan 25, 2002 at 01:48:06PM -0600, Brian Johnson wrote:
> [MMU support]
> It may be useful as a development vehicle for PPC support, if PPC
> MacOS is more dependent on the MMU.

It isn't.

> BTW, any thoughts on running MacOS X on a PPC BasiliskII?

Forget it. :-)

Bye,
Christian

-- 
  / Coding on PowerPC and proud of it
\/ http://www.uni-mainz.de/~bauec002/

[B2-devel] Re: Any use of MMU emulation?

[Brian J. <bjj...@us...>]

Gwenole Beauchesne <gb...@di...> wrote:
> I was wondering if B2 would benefit from MMU emulation. Some users
> want it because:
>
> 1) It enables the use of virtual memory.
>
> My opinion is that if a user wants X MB of virtual Mac memory, where
> X > physical memory available to the host system, he can already do
> so.  The host system will handle much better extra memory than
> letting MacOS swapping to the hardfile.

Yes.

> 2) It enables other systems to run under B2.
>
> This is a false problem. B2 is "only" a Macintosh emulator and
> doesn't emulate much of a Mac hardware. So, if B2 were to have MMU
> emulation, Linux or other systems will probably not work anyway.

Certainly.

> Other ideas?

It may be useful as a development vehicle for PPC support, if PPC
MacOS is more dependent on the MMU.  But that's a long shot.

BTW, any thoughts on running MacOS X on a PPC BasiliskII?

> There are currently two MMU implementations available on top of
> UAE:
> ...

Would it be possible to do a hardware-assisted implementation?
I.e. have the Mac load directly from Mac virtual addresses (with an
appropriate direct-addressing offset), without translating them to
physical addresses, and use a segfault handler to catch TLB misses and
mmap() the corresponding "physical" page from a file?  We'd need some
way to handle different host and Mac page sizes....

						Brian

--------------------------------------------------------------------

   "Practice random kindness, and senseless acts of beauty."
					   -- seen in Lorene's office

[B2-devel] ticks problem fixed (thanks!)

[Brian J. <bjj...@us...>]

Thanks for fixing the problem with Ticks getting incremented on every
interrupt, instead of just the 60Hz interrupt.  Games are much more
playable now.  8-)

There's only one more Irix-specific issue I know of in BasiliskII:
BII does not set the pthread scope explicitly, instead it defaults to
system scope on Linux (indeed, that's the only scope available in
LinuxThreads), but to process scope on Irix.  That means the Irix
pthreads library does all the thread scheduling itself, instead of
letting threads always block in the kernel.  This is fine for many
apps, but in BasiliskII, the main instruction emulation thread doesn't
get preempted often enough to keep the video and audio running
smoothly...  The fix is to use system scope or, since that requires
root privileges on Irix (high-priority realtime pthreads can starve
out kernel threads, such as the network packet handlers), the
Irix-specific "bound scope".

Anyway, I hope to get a patch together in the next week or so.  If it
can get into BasiliskII 1.0, BII will simply rock on Irix.

Thanks,

						Brian J. Johnson

--------------------------------------------------------------------

  "Great works are performed not by power, but by perseverance."
                                           -- Samuel Johnson

Re: [B2-devel] offtopic - OS 8.1?

[Gwenole B. <gb...@di...>]

Hi,

> I have been searching again for Mac OS 8.1 on eBay, AuctionWatch.com, Yahoo,
> MSN and others and couldn't find Mac OS 8.1 to purchase...

Actually, there were full MacOS 8.1 CDs sold in "recent" hardware
(translate that in the epoch of 68k hardware). However, I had bought MacOS
8.0 and updated to 8.1 with updates that could be found in Mac magazines,
for example.

Bye,
Gwenole.

[B2-devel] offtopic - OS 8.1?

[Hetz B. H. <he...@wi...>]

Hi All,

Well, after talking to my Apple dealer again, it seems they compiled the OS 
against G3 (it's OS 8.1 with Hebrew language support) and they don't have any 
US version available. Not only that - if I don't have a genuine G3 - they 
won't sell it to me. Fucking ego maniacs...

I have been searching again for Mac OS 8.1 on eBay, AuctionWatch.com, Yahoo, 
MSN and others and couldn't find Mac OS 8.1 to purchase...

Anyone knows about a place to buy? a link to download? anything?

Thanks,
Hetz

Re: [B2-devel] Use of ExpandMem lowmem?

[Christian B. <cb...@st...>]

Hi!

On Sat, Jan 19, 2002 at 12:32:02AM +0100, Gwenole Beauchesne wrote:
> Christian, in the series of system patches, does MacOS actually know
> that in 68040 (UAE) mode it has to generate state frames of that type
> when the FPU is enabled? Hadn't you a similar problem for 68040/68060
> native mode on AmigaOS, NetBSD?

MacOS will use the correct state frames according to the values set in
CPUFlag. The problem with the 68060 is that MacOS doesn't know about it
and can't create correct 68060 FPU frames. Hence, the native 68k versions
use 68040 FPU frames on FSAVE/FRESTORE (these instructions are emulated)
since CPUFlag is set to "68040" on these machines.

Bye,
Christian

-- 
  / Coding on PowerPC and proud of it
\/ http://www.uni-mainz.de/~bauec002/

Re: [B2-devel] Re: Any use of MMU emulation?

[Christian B. <cb...@st...>]

Hi!

On Mon, Jan 21, 2002 at 02:21:59PM +0100, Gwenole Beauchesne wrote:
> 1) It enables the use of virtual memory.
> 
> My opinion is that if a user wants X MB of virtual Mac memory, where X >
> physical memory available to the host system, he can already do so.

Exactly (except for AmigaOS where CPU/MMU emulation isn't used anyway).

> 2) It enables other systems to run under B2.
> 
> This is a false problem. B2 is "only" a Macintosh emulator and doesn't
> emulate much of a Mac hardware.

Specifically, it is designed to run MacOS on some nonexistant virtual 68k-ish
hardware. :-)

Bye,
Christian

-- 
  / Coding on PowerPC and proud of it
\/ http://www.uni-mainz.de/~bauec002/

[B2-devel] Re: Any use of MMU emulation?

[Gwenole B. <gb...@di...>]

Hi,

I was wondering if B2 would benefit from MMU emulation. Some users want it
because:

1) It enables the use of virtual memory.

My opinion is that if a user wants X MB of virtual Mac memory, where X >
physical memory available to the host system, he can already do so. The
host system will handle much better extra memory than letting MacOS
swapping to the hardfile.

2) It enables other systems to run under B2.

This is a false problem. B2 is "only" a Macintosh emulator and doesn't
emulate much of a Mac hardware. So, if B2 were to have MMU emulation,
Linux or other systems will probably not work anyway.

Other ideas?

There are currently two MMU implementations available on top of UAE:

1) One located at Bernd's website for UAE:
<http://www.freiburg.linux.de/~uae/bin/patches/mmu/uae-0.8.20.2-mmu.diff.gz>

2) The other included in ARAnyM (Atari Running on Any Machine):
<http://aranym.sophics.cz/index.html>

The latter is the first ever project I am aware of that is based on
Basilisk II. ;-)

Bye,
Gwenole.

Re: [B2-devel] Use of ExpandMem lowmem?

[<gb...@di...>]

[Oops, that one didn't get onto the list]

Hi,

> Well, it seems to call HandleZone() with a handle it gets from the
> ExpandMem area but which is uninitialized.

Well, I could explore a little more in the train: actually, I disabled
FPUType that was forced to 1 in main.cpp when CPUType=4. Disabling FPU
emulation in 68040 mode (i.e. 68LC040) lead to no crash at all with APD.

The only parts that differ in the FPU core between 68040 and the rest
are FSAVE and FRESTORE instructions. Probably state frames are mis
popped/pushed to the stack.

Christian, in the series of system patches, does MacOS actually know
that in 68040 (UAE) mode it has to generate state frames of that type
when the FPU is enabled? Hadn't you a similar problem for 68040/68060
native mode on AmigaOS, NetBSD?

Bye,
Gwenole.

Re: [B2-devel] buggy display on the JIT version of B2

[Hetz B. H. <he...@wi...>]

On Friday 18 January 2002 19:57, Gwenole Beauchesne wrote:
> Hi,
>
> I will be short.
>
> > That makes it compiles - but with the JIT enabled - I get the bad emu
> > screen like in the screenshot that I sent with the previous email..
>
> Huh, the video code is the same between the JIT and non-JIT version of
> BasiliskII. Have you tried with the "jit" prefs item set to either "false"
> or "true"? Same result?

No,

JIT disabled (from the menu) - everything looks ok.

JIT enabled - graphics problems...

I got once another problem while testing - the emu crashed (JIT enabled) with 
the following error:

<JIT compiler> : actual translation cache size : 8192 KB at 0x49564000
<JIT compiler> : gather statistics on translation time
Illegal instruction: 0009 at 00000720
do_handle_screen_fault: unhandled address 0x62661a78 [IP=0x80dbb61]
D0: 07f19f8a D1: 07ef0003 D2: 07ea99a0 D3: 00000001
D4: 00000020 D5: 00000008 D6: 00000003 D7: 07f19f8a
A0: 22152a78 A1: 07ed5d6c A2: 07eb6a04 A3: 00155ea0
A4: 00035d7a A5: 07f19f8a A6: 07ef6d98 A7: 07ef6d78
USP=00000000 ISP=07ef6d78 MSP=00000000 VBR=00000000
T=00 S=1 M=0 X=1 N=0 Z=0 V=0 C=0 IMASK=0
FP0: nan FP1: nan FP2: nan FP3: nan
FP4: nan FP5: nan FP6: nan FP7: nan
N=0 Z=0 I=0 NAN=0
00030d8a: 2650 b9ed fbbe 56c3 4403 MOVEA.L (A0),A3
next PC: 00030d8c

Hetz

Re: [B2-devel] buggy display on the JIT version of B2

[Gwenole B. <gb...@di...>]

Hi,

I will be short.

> That makes it compiles - but with the JIT enabled - I get the bad emu screen
> like in the screenshot that I sent with the previous email..

Huh, the video code is the same between the JIT and non-JIT version of
BasiliskII. Have you tried with the "jit" prefs item set to either "false"
or "true"? Same result?

Looking at the picture makes me definitely think of a problem in the
blitters, especially the residual bytes to copy. I will check
<video_blit.h>. This is weird as buffers to be blitters are now a multiple
of <size of a line> bytes.

Bye,
Gwenole (hopefully catching his train in time)

Re: [B2-devel] buggy display on the JIT version of B2

[Hetz B. H. <he...@wi...>]

That makes it compiles - but with the JIT enabled - I get the bad emu screen 
like in the screenshot that I sent with the previous email..

Hetz

On Friday 18 January 2002 18:49, Gwenole Beauchesne wrote:
> On Fri, 18 Jan 2002, Hetz Ben Hamo wrote:
> > However, running your "devconfig" script which compiles brings an error
> > while compiling (on GCC 2.96 with your JIT version of B2 from today):
>
> Sorry, the appended patch should do the trick or configure with
> --enable-jit-debug as well (provided you have cxmon available).
>
> Bye,
> Gwenole.

Re: [B2-devel] buggy display on the JIT version of B2

[Gwenole B. <gb...@di...>]

On Fri, 18 Jan 2002, Hetz Ben Hamo wrote:

> However, running your "devconfig" script which compiles brings an error while
> compiling (on GCC 2.96 with your JIT version of B2 from today):

Sorry, the appended patch should do the trick or configure with
--enable-jit-debug as well (provided you have cxmon available).

Bye,
Gwenole.

Re: [B2-devel] buggy display on the JIT version of B2

[Gwenole B. <gb...@di...>]

On Fri, 18 Jan 2002, Hetz Ben Hamo wrote:

> This one:
> http://gwenole.beauchesne.online.fr/basilisk2/files/BasiliskII-0.9.JIT-3.i586.rpm
> - is not available at your web pages (error 404)

Yes, I removed it because it was miscompiled, and I hadn't had the time to
rebuild it.

> If you decide to release a version - then please point me to an src-rpm file
> and I'll issue an RPM built on RH 7.2.

You have two options:

1) Grab the SRPM I provided for Mandrake Linux but clean the specfile a
little. i.e. don't use Patch0, possiblity redefine DebugHWBases and
JITDebug to 0, s/5mdk/5/ for release tag, etc, remove BuildRequires:
dependencies that probably don't exist in RH-7.2.

2) Use the specfile from the B2/JIT tarball. Make sure however that you
define CONF_OPT_FLAGS="--enable-jit-compiler" prior issuing the "rpm"
command or simply add this flag to configure in the %build part.
Otherwise, the JIT compiler won't be enabled as wasn't it enabled in the
release -3 on my website. ;-)

Bye,
Gwenole.

Re: [B2-devel] buggy display on the JIT version of B2

[Christian B. <cb...@st...>]

Hi!

On Fri, Jan 18, 2002 at 06:13:52PM +0200, Hetz Ben Hamo wrote:
> Another (a bit offtopic question) - there is this launcher of apps which on 
> my emu session is staying in the middle of the screen - how can I move it 
> down to the bottom of the screen? (OS 7.5.3)

The Control Strip? Cmd-Click (or was it Shift-Click?) on the handle, then
you can drag it down.

Bye,
Christian

-- 
  / Coding on PowerPC and proud of it
\/ http://www.uni-mainz.de/~bauec002/

Re: [B2-devel] buggy display on the JIT version of B2

[Hetz B. H. <he...@wi...>]

> > I don't have the priviledge of a branch but the latest snapshot is here:
> > <http://gwenole.beauchesne.online.fr/basilisk2/files/>
> > File: BasiliskII_jit_src_15012002.tar.bz2

Ok, I checked that out and the problem goes away..

However, running your "devconfig" script which compiles brings an error while 
compiling (on GCC 2.96 with your JIT version of B2 from today):

c++ -I../include -I. -I../uae_cpu -DHAVE_CONFIG_H  -DOS_linux -DCPU_i386 
-DDIRECT_ADDRESSING -fno-strength-reduce -DSAHF_SETO_PROFITABLE 
-DUNALIGNED_PROFITABLE -DREGPARAM="__attribute__((regparm(3)))" 
-DX86_ASSEMBLY -DOPTIMIZED_FLAGS -DUSE_JIT -DUSE_JIT_FPU -DFPU_IEEE 
-D_REENTRANT -DDATADIR=\"/usr/local/share/BasiliskII\"   -O2 
-I/usr/X11R6/include -I/usr/include/gtk-1.2 -I/usr/include/glib-1.2 
-I/usr/lib/glib/include -I/usr/X11R6/include -fomit-frame-pointer 
-fno-merge-constants -c ../uae_cpu/compiler/compemu_support.cpp -o 
obj/compemu_support.o
../uae_cpu/compiler/compemu_support.cpp: In function `void compile_block 
(cpu_history *, int)':
../uae_cpu/compiler/compemu_support.cpp:6325: `last_regs_pc_p' undeclared 
(first use this function)
../uae_cpu/compiler/compemu_support.cpp:6325: (Each undeclared identifier is 
reported only once for each function it appears in.)
../uae_cpu/compiler/compemu_support.cpp:6326: `last_compiled_block_addr' 
undeclared (first use this function)
../uae_cpu/compiler/compemu_support.cpp:6565: `struct blockinfo_t' has no 
member named `direct_handler_size'
make: *** [obj/compemu_support.o] Error 1

Another (a bit offtopic question) - there is this launcher of apps which on 
my emu session is staying in the middle of the screen - how can I move it 
down to the bottom of the screen? (OS 7.5.3)

Thanks,
Hetz

Re: [B2-devel] Use of ExpandMem lowmem?

[Christian B. <cb...@st...>]

Hi!

On Fri, Jan 18, 2002 at 12:17:58PM +0100, Gwenole Beauchesne wrote:
> Any ideas?

Well, it seems to call HandleZone() with a handle it gets from the
ExpandMem area but which is uninitialized.

> PS: Christian, it would be neat if cxmon could resolve the address of low
> mem global by name, I currently get:

Excellent idea. I've modified the cxmon in CVS to predefine all Low Mem
globals as variables.

Bye,
Christian

-- 
  / Coding on PowerPC and proud of it
\/ http://www.uni-mainz.de/~bauec002/

Re: [B2-devel] buggy display on the JIT version of B2

[Hetz B. H. <he...@wi...>]

On Friday 18 January 2002 17:31, Gwenole Beauchesne wrote:
> On Fri, 18 Jan 2002, Hetz Ben Hamo wrote:
> > I tried the RPM version of the B2-JIT (Redhat 7.2) - it works ok, but the
> > mac screen looks very bad...
>
> Which version exactly?

From.. .february 2001 - thats the only one which is actually available for 
redhat - 
http://gwenole.beauchesne.online.fr/basilisk2/files/BasiliskII-0.9.JIT-2.i386.rpm

This one: 
http://gwenole.beauchesne.online.fr/basilisk2/files/BasiliskII-0.9.JIT-3.i586.rpm 
- is not available at your web pages (error 404)

> > Do you have a CVS branch which can be checked out and be tested?
>
> I don't have the priviledge of a branch but the latest snapshot is here:
> <http://gwenole.beauchesne.online.fr/basilisk2/files/>
> File: BasiliskII_jit_src_15012002.tar.bz2

Ok, checking it out..

> I have RPMs for Mandrake Linux in contribs/ mirror by the way. ;-)

If you decide to release a version - then please point me to an src-rpm file 
and I'll issue an RPM built on RH 7.2.

Hetz

Re: [B2-devel] buggy display on the JIT version of B2

[Gwenole B. <gb...@di...>]

On Fri, 18 Jan 2002, Hetz Ben Hamo wrote:

> I tried the RPM version of the B2-JIT (Redhat 7.2) - it works ok, but the mac
> screen looks very bad...

Which version exactly?

> Take a look at the following screenshot:
> http://witch.dyndns.org/images/bad-emu.png

Sounds a problem in the blitters.

> Do you have a CVS branch which can be checked out and be tested?

I don't have the priviledge of a branch but the latest snapshot is here:
<http://gwenole.beauchesne.online.fr/basilisk2/files/>
File: BasiliskII_jit_src_15012002.tar.bz2

I have RPMs for Mandrake Linux in contribs/ mirror by the way. ;-)

Bye,
Gwenole.

Re: [B2-devel] forgot something

[Christian B. <cb...@st...>]

Hi!

On Fri, Jan 18, 2002 at 05:12:13PM +0200, Hetz Ben Hamo wrote:
> Yes, go figure - stupid Apple dealer ;)

Maybe they're not familiar with the old equipment any more... :)

Bye,
Christian

-- 
  / Coding on PowerPC and proud of it
\/ http://www.uni-mainz.de/~bauec002/

[B2-devel] buggy display on the JIT version of B2

[Hetz B. H. <he...@wi...>]

Hi Gwenole, others..

I tried the RPM version of the B2-JIT (Redhat 7.2) - it works ok, but the mac 
screen looks very bad...

Take a look at the following screenshot: 
http://witch.dyndns.org/images/bad-emu.png

Do you have a CVS branch which can be checked out and be tested?

Thanks,
Hetz

Re: [B2-devel] forgot something

[Hetz B. H. <he...@wi...>]

Yes, go figure - stupid Apple dealer ;)

Hetz

On Friday 18 January 2002 17:09, Christian Bauer wrote:
> Hi!
>
> On Thu, Jan 17, 2002 at 07:17:57PM +0200, Hetz Ben Hamo wrote:
> > I was talking to our local Apple dealer for purchasing Mac OS 8.1 CD and
> > he told me that Quadra 900 cannot be installed with OS 8.1, only 7.6.1
>
> According to my specs list, 8.1 does run on the Quadra 900.
>
> Bye,
> Christian

Re: [B2-devel] forgot something

[Christian B. <cb...@st...>]

Hi!

On Thu, Jan 17, 2002 at 07:17:57PM +0200, Hetz Ben Hamo wrote:
> I was talking to our local Apple dealer for purchasing Mac OS 8.1 CD and he 
> told me that Quadra 900 cannot be installed with OS 8.1, only 7.6.1

According to my specs list, 8.1 does run on the Quadra 900.

Bye,
Christian

-- 
  / Coding on PowerPC and proud of it
\/ http://www.uni-mainz.de/~bauec002/

Re: [B2-devel] Status report

[Hetz B. H. <he...@wi...>]

On Friday 18 January 2002 16:43, Christian Bauer wrote:
> Hi!
>
> On Thu, Jan 17, 2002 at 07:09:15PM +0200, Hetz Ben Hamo wrote:
> > In Quadra Mode I get:
> > [...]
> > Illegal instruction: 000b at 0003cb54
>
> Which version of MacOS were you trying here?

7.5.3 - I thought that the B2 can be used with that on Quadra..

>
> > Now - may I ask if Gwenole or Christian can add a simple checking of
> > which ROM the user got?
>
> I don't know whether this is easily possible. All 32-bit clean Mac II
> series ROMs are "universal" to a certain degree.

I understand, but doing this check will prevent others who download their 
roms from the net to get those errors - look at ths SF forums to see the 
errors about it - and thats only occures when someone tried to use MacII rom 
in Quadra mode.

Hetz

Re: [B2-devel] Status report

[Christian B. <cb...@st...>]

Hi!

On Thu, Jan 17, 2002 at 07:09:15PM +0200, Hetz Ben Hamo wrote:
> In Quadra Mode I get:
> [...]
> Illegal instruction: 000b at 0003cb54

Which version of MacOS were you trying here?

> Now - may I ask if Gwenole or Christian can add a simple checking of which 
> ROM the user got?

I don't know whether this is easily possible. All 32-bit clean Mac II series
ROMs are "universal" to a certain degree.

Bye,
Christian

-- 
  / Coding on PowerPC and proud of it
\/ http://www.uni-mainz.de/~bauec002/