[Qemu-devel] qemu vga.c

[Qemu-devel] qemu vga.c

[Fabrice Bellard]

CVSROOT:	/cvsroot/qemu
Module name:	qemu
Branch: 	
Changes by:	Fabrice Bellard <fabrice.bellard@free.fr>	03/10/30 17:10:22

Modified files:
	.              : vga.c 

Log message:
	big endian fixes

CVSWeb URLs:
http://savannah.gnu.org/cgi-bin/viewcvs/qemu/qemu/vga.c.diff?tr1=1.6&tr2=1.7&r1=text&r2=text

Re: [Qemu-devel] qemu vga.c

[Chad Page]

	This fixed it on my iBook.  I can also see that freedos and netbsd
work now too...

	- Chad


On Thu, 30 Oct 2003, Fabrice Bellard wrote:

> CVSROOT:	/cvsroot/qemu
> Module name:	qemu
> Branch: 	
> Changes by:	Fabrice Bellard <fabrice.bellard@free.fr>	03/10/30 17:10:22
> 
> Modified files:
> 	.              : vga.c 
> 
> Log message:
> 	big endian fixes
> 
> CVSWeb URLs:
> http://savannah.gnu.org/cgi-bin/viewcvs/qemu/qemu/vga.c.diff?tr1=1.6&tr2=1.7&r1=text&r2=text
> 
> 
> 
> _______________________________________________
> Qemu-devel mailing list
> Qemu-devel@nongnu.org
> http://mail.nongnu.org/mailman/listinfo/qemu-devel
> 

Re: [Qemu-devel] qemu vga.c

[Leigh Dyer]

On Fri, 2003-10-31 at 10:27, Chad Page wrote:
> 	This fixed it on my iBook.  I can also see that freedos and netbsd
> work now too...
> 

It's all working here on my Powerbook too. I just noticed the VGA
directory on the freedos image - time to have a play :)

On an unrelated note, I've seen several segfaults when running my Debian
Woody system inside qemu, though I can't reproduce them because they
don't seem to have any obvious cause - sometimes the system doesn't make
it through the entire boot process, but other times it'll boot and run
for a good while before dying. This happens on both x86 and PPC hosts.
What's the best way to start looking in to these problems?

If anyone's interested in a copy of this disk image (compressed it's
probably about 100Mb), I'd be happy to make it available on the web.

Thanks
Leigh

> 	- Chad
> 
> 
> On Thu, 30 Oct 2003, Fabrice Bellard wrote:
> 
> > CVSROOT:	/cvsroot/qemu
> > Module name:	qemu
> > Branch: 	
> > Changes by:	Fabrice Bellard <fabrice.bellard@free.fr>	03/10/30 17:10:22
> > 
> > Modified files:
> > 	.              : vga.c 
> > 
> > Log message:
> > 	big endian fixes
> > 
> > CVSWeb URLs:
> > http://savannah.gnu.org/cgi-bin/viewcvs/qemu/qemu/vga.c.diff?tr1=1.6&tr2=1.7&r1=text&r2=text
> > 
> > 
> > 
> > _______________________________________________
> > Qemu-devel mailing list
> > Qemu-devel@nongnu.org
> > http://mail.nongnu.org/mailman/listinfo/qemu-devel
> > 
> 
> 
> 
> _______________________________________________
> Qemu-devel mailing list
> Qemu-devel@nongnu.org
> http://mail.nongnu.org/mailman/listinfo/qemu-devel
> 

Re: [Qemu-devel] qemu vga.c

[Stefano Marinelli]

On Fri, Oct 31, 2003 at 11:01:15AM +1100, Leigh Dyer wrote:
> If anyone's interested in a copy of this disk image (compressed it's
> probably about 100Mb), I'd be happy to make it available on the web.

I'm interested...thanks! :)

Stefano

[Qemu-devel] qemu vga.c

[Fabrice Bellard]

CVSROOT:	/cvsroot/qemu
Module name:	qemu
Branch: 	
Changes by:	Fabrice Bellard <fabrice.bellard@free.fr>	03/09/30 17:29:03

Modified files:
	.              : vga.c 

Log message:
	multiscan/doublescan fix (malc)

CVSWeb URLs:
http://savannah.gnu.org/cgi-bin/viewcvs/qemu/qemu/vga.c.diff?tr1=1.5&tr2=1.6&r1=text&r2=text

[Qemu-devel] qemu vga.c

[Fabrice Bellard]

CVSROOT:	/cvsroot/qemu
Module name:	qemu
Branch: 	
Changes by:	Fabrice Bellard <fabrice.bellard@free.fr>	03/08/11 16:33:04

Modified files:
	.              : vga.c 

Log message:
	mode X double scan fix (malc)

CVSWeb URLs:
http://savannah.gnu.org/cgi-bin/viewcvs/qemu/qemu/vga.c.diff?tr1=1.3&tr2=1.4&r1=text&r2=text

Re: [Qemu-devel] qemu vga.c

[Johan Rydberg]

Fabrice Bellard <fabrice.bellard@free.fr> wrote:

: Modified files:
: 	.              : vga.c 

I have not yet had the chanse of testing the VGA code,
so I wonder: how fast is it?  And does it look good? :)

-- 
Johan Rydberg, Free Software Developer, Sweden
http://rtmk.sf.net | http://www.nongnu.org/guss/

Playing Tom McCrae - Ghost Of A Shark

Re: [Qemu-devel] qemu vga.c

[Fabrice Bellard]

Johan Rydberg wrote:
> Fabrice Bellard <fabrice.bellard@free.fr> wrote:
> 
> : Modified files:
> : 	.              : vga.c 
> 
> I have not yet had the chanse of testing the VGA code,
> so I wonder: how fast is it?  And does it look good? :)

It is currently fast enough to have a usable PC emulator. Many DOS games 
are usable. It should be faster than the bochs emulation but slightly 
slower than the dosemu one. This VGA emulation code is more accurate 
than either bochs or dosemu (all VGA modes, including non standard ones 
(i.e. the ones not supported by int $0x10) should work), but of course 
more testing is needed.

There are some optimisations pending (VGA memory accesses could be 
faster, but I wanted first to have a simple code). The S3 specific code 
the SVGA emulation is not finished. It will be mainly used to launch X11 
or Windows in high resolution modes. I also have to modify the LGPL'ed 
vga bios to add S3 VESA support.

Fabrice.

Re: [Qemu-devel] qemu vga.c

[Johan Rydberg]

Fabrice Bellard <fabrice.bellard@free.fr> wrote:

: It is currently fast enough to have a usable PC emulator. Many DOS games 
: are usable. It should be faster than the bochs emulation but slightly 
: slower than the dosemu one. This VGA emulation code is more accurate 
: than either bochs or dosemu (all VGA modes, including non standard ones 
: (i.e. the ones not supported by int $0x10) should work), but of course 
: more testing is needed.

Sounds great.  Maybe you could show us deadly being some nice screenshots?

: There are some optimisations pending (VGA memory accesses could be 
: faster, but I wanted first to have a simple code). The S3 specific code 
: the SVGA emulation is not finished. It will be mainly used to launch X11 
: or Windows in high resolution modes. I also have to modify the LGPL'ed 
: vga bios to add S3 VESA support.

I saw that you trap all writings to the graphics memory.  Is this a
must, or can't you just store the data in a buffer and process it according
to the current mode when the display should be updated?

-- 
Johan Rydberg, Free Software Developer, Sweden
http://rtmk.sf.net | http://www.nongnu.org/guss/

Playing Tom McCrae - Line Of Fire

Re: [Qemu-devel] qemu vga.c

[Fabrice Bellard]

Johan Rydberg wrote:
> Fabrice Bellard <fabrice.bellard@free.fr> wrote:
> 
> : It is currently fast enough to have a usable PC emulator. Many DOS games 
> : are usable. It should be faster than the bochs emulation but slightly 
> : slower than the dosemu one. This VGA emulation code is more accurate 
> : than either bochs or dosemu (all VGA modes, including non standard ones 
> : (i.e. the ones not supported by int $0x10) should work), but of course 
> : more testing is needed.
> 
> Sounds great.  Maybe you could show us deadly being some nice screenshots?

If I have the time, the next release will really support VGA emulation 
so that you can actually try it. I'll try to put screenshots.

> : There are some optimisations pending (VGA memory accesses could be 
> : faster, but I wanted first to have a simple code). The S3 specific code 
> : the SVGA emulation is not finished. It will be mainly used to launch X11 
> : or Windows in high resolution modes. I also have to modify the LGPL'ed 
> : vga bios to add S3 VESA support.
> 
> I saw that you trap all writings to the graphics memory.  Is this a
> must, or can't you just store the data in a buffer and process it according
> to the current mode when the display should be updated?

Correct vga emulation requires that you trap all memory writes and read 
as in planar mode complex binary operations can be performed by the VGA 
card on the 4 planes simultaneously (when writing or reading one byte, 
the VGA card writes or reads 4 bytes simultaneously). Even in "chain4" 
mode which is the simplest memory mapping (linear VGA memory access), 
writes to some memory planes can be disabled. My upcoming optimizations 
will consist in using specialized memory access functions for each case, 
and by mapping memory directly when possible (it is possible only if 
chain4 mode is activated and if all 4 VGA planes are enabled).

Fabrice.