headers
Michael | 1 Apr 2005 02:01
Picon
Favicon

Re: XF86 on macppc, issue 8472854 ;)


Hello,

> hey, -current runs on my G4 again -- after it stopped working early 
> october last year, but that's a different story...
Any idea what caused it?

> i'm running -current now and have XFree 4.4.0 installed (4.5.0's make 
> failed -- no time to figure it out yet.)
Please try 4.5 ( but make sure to keep the 4.4 binaries ) - it has 
serious problems here and I'd like to now if it's just me. ( it will 
freeze after a while, enter an endless loop in libpthread. ssh and so 
on will still work su the process can be killed )

> however, X ran on my G4 with the XF86Config file applied (see below), 
> but now i got every time i try to fire up X my screen out of sync -- 
> my monitor (22" CRT, really not crap ;) says 'Horizontal 260kHz, 
> Vertical 208Hz' (blabla).
Ouch. I got similar nonsense when hacking my Voodoo3 half a year ago, 
but that's hardly related ( and still works just fine )

> the important dmesg lines:
>
> (...)
> ofb0 at pci0 dev 16 function 0: ATI Technologies product 0x4966
> ofb0: 1600 x 1200, 8bpp
Ok, that's some sort of Radeon.

> http://snom.dyndns.org/XF86Config
Looks ok.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Michael | 1 Apr 2005 02:11
Picon
Favicon

Re: XF86 on macppc, issue 8472854 ;)


Hello,

> no, i'm not running michael's kernel because i've spent the last 
> days/weeks figuring out (no, that was more the part of tim kelly -- 
> thanks, tim!) why -current didn't run on my G4 and G3 since early 
> october last year... now we know that it's an MCHK issue which gets 
> (re)solved when highly optimizing the kernel (altivec, mcpu/mtune, and 
> O3).
I'll make a new one after the S900 is done compiling KDE 3.4.
Chris - was it you who had a Voodoo5 in a Mac somewhere? I remember 
someone here had one, but I have no idea who. Turns out the voodoo3-5 
have nearly identical blitters ( at least from a driver's point of view 
) so I changed my driver to attach to voodoo4 and 5 too, it would be 
nice to know if this really works.

> furthermore, i don't have michael's X sets installed -- didn't he pull 
> up his patches into -current yet?
I'll make new ones too.

Maybe I've got an idea what's wrong with the Radeons - I still think 
it's PCI IO related, so my comment about the voodoo3 in the other mail 
may not be that far off the mark. So the next kernel will probably 
enable a few things like sparse IO and so on for Radeons so the Xserver 
can actually access legacy VGA registers ( on the voodoo3 it works 
because these registers are actually in a PCI resource )

have fun
Michael
(Continue reading)

Permalink | Reply | 1 comment |
headers
Tim Kelly | 1 Apr 2005 02:35

Re: XF86 on macppc, issue 8472854 ;)

On Thu, 31 Mar 2005 23:18:24 +0200
Timo Schoeler <wanker4freedom <at> web.de> wrote:

> no, i'm not running michael's kernel because i've spent the last 
> days/weeks figuring out (no, that was more the part of tim kelly -- 
> thanks, tim!) why -current didn't run on my G4 and G3 since early 
> october last year...

Just a bit of clarification - Nathan's patch to preserve external
interrupt enable fixed Timo's original problem, but during diagnosing
this optimizations were turned off. It took a little while for Timo and
I to realize this and attempt the kernel with optimizations on, which
fixed the MHCK problem that remained after Nathan's patch. We don't
know why turning optimizations off introduces an MHCK exception.

tim
Permalink | Reply | 0 comments |
headers
Chris Tribo | 1 Apr 2005 03:45
Favicon

Re: XF86 on macppc, issue 8472854 ;)


On Mar 31, 2005, at 7:11 PM, Michael wrote:

> Hello,
>
>> no, i'm not running michael's kernel because i've spent the last 
>> days/weeks figuring out (no, that was more the part of tim kelly -- 
>> thanks, tim!) why -current didn't run on my G4 and G3 since early 
>> october last year... now we know that it's an MCHK issue which gets 
>> (re)solved when highly optimizing the kernel (altivec, mcpu/mtune, 
>> and O3).
> I'll make a new one after the S900 is done compiling KDE 3.4.
> Chris - was it you who had a Voodoo5 in a Mac somewhere? I remember 
> someone here had one, but I have no idea who. Turns out the voodoo3-5 
> have nearly identical blitters ( at least from a driver's point of 
> view ) so I changed my driver to attach to voodoo4 and 5 too, it would 
> be nice to know if this really works.
>

Yes that's me. The problem at the moment is that I need classic MacOS 
to flash the card back to the Mac ROM, it was serving duty as my 
official Unreal Tournament Glide video card in my PC for a while.
Permalink | Reply | 0 comments |
headers
Miles Nordin | 2 Apr 2005 22:12

Re: the 8600 lives!

>>>>> "m" == Michael  <macallan18 <at> earthlink.net> writes:

     m> http://macallam.homeunix.org:6704/stuff/BSD/

k., I got the X server and kernel to test if it helps with chips 65550
support.

I can't boot the kernel because there's a bug with the power
management driver on my PB2400, so as soon as it attaches, the machine
powers off, blinks the green Sleep light, and reboots.

Could I trouble you to post the source patches, either patches to go
with each new kernel/xserver you release, or else a ``latest patches''
diff if that's easier?  Optimistically, having the source code
available might help get your work reviewed and committed, too.  It
seems like quite a few cards are affected.
Permalink | Reply | 5 comments |
headers
Michael | 2 Apr 2005 22:42
Picon
Favicon

Re: the 8600 lives!

Hello,

>      m> http://macallam.homeunix.org:6704/stuff/BSD/
> 
> k., I got the X server and kernel to test if it helps with chips 65550
> support.
> 
> I can't boot the kernel because there's a bug with the power
> management driver on my PB2400, so as soon as it attaches, the machine
> powers off, blinks the green Sleep light, and reboots.
> 
> Could I trouble you to post the source patches, either patches to go
> with each new kernel/xserver you release, or else a ``latest patches''
> diff if that's easier?  Optimistically, having the source code
> available might help get your work reviewed and committed, too.  It
> seems like quite a few cards are affected.

The kernel changes are short and simple ( can't post a real patch because my console stuff differs a lot from
the official source )
Basically you need to change src/sys/arch/macppc/dev/ofb.c and ofbvar.h like this like this:

add 	
	bus_space_tag_t sc_iot;		/* parent's IO tag - we need to allow mmap()ing this space. */
to struct ofb_softc in ofbvar.h

than add 
	/* store parent's IO tag to allow mmap()ing IO space */
	sc->sc_iot=pa->pa_iot;
to ofbattach() in ofb.h and finally do this to ofb_mmap():
paddr_t
(Continue reading)

Permalink | Reply | 1 comment |
headers
zeta.orionis | 2 Apr 2005 23:35

modem support

The models page (http://www.netbsd.org/Ports/macppc/models.html) has a 
lot of "The internal modem is not supported yet" notes. Are these 
modems going to be supported at some date? When?

The specific model I'm looking at is the revision B 300MHz iBook.
Permalink | Reply | 1 comment |
headers
Michael | 2 Apr 2005 23:54
Picon
Favicon

Re: modem support

Hello,

> The models page (http://www.netbsd.org/Ports/macppc/models.html) has a 
> lot of "The internal modem is not supported yet" notes. Are these 
> modems going to be supported at some date? When?
> 
> The specific model I'm looking at is the revision B 300MHz iBook.

Find documentation and someone who has the right combination of time, knowledge and hardware. The newer
internal modems ( like in my iBook G4 ) are braindead anyway so the main CPU needs to do all the signal
wrangling - not quite trivial.
The older modems may just be connected to some serial port, these should work.

have fun
Michael
Permalink | Reply | 0 comments |
headers
Niels S. Eliasen | 3 Apr 2005 00:57
Favicon

Re: the 8600 lives!

Hi Miles
Please correct me if im wrong but I do believe that your power-off 
problem has nothing to do with the "exquisite" corrections that Michael 
has done for X11(where would we be without them!)..

I had very much the same problem, but Allen Briggs did wonders in 
correcting the power-off problems for my Powerbook 3500(and his own 
3400)...
Try to get a kernel from Allen... I think it should make quite a 
difference for you with getting the Powerbook working..

Den 2/4-2005, kl. 22.12, skrev Miles Nordin:

>>>>>> "m" == Michael  <macallan18 <at> earthlink.net> writes:
>
>      m> http://macallam.homeunix.org:6704/stuff/BSD/
>
> k., I got the X server and kernel to test if it helps with chips 65550
> support.
>
> I can't boot the kernel because there's a bug with the power
> management driver on my PB2400, so as soon as it attaches, the machine
> powers off, blinks the green Sleep light, and reboots.
>
> Could I trouble you to post the source patches, either patches to go
> with each new kernel/xserver you release, or else a ``latest patches''
> diff if that's easier?  Optimistically, having the source code
> available might help get your work reviewed and committed, too.  It
> seems like quite a few cards are affected.
>
(Continue reading)

Permalink | Reply | 2 comments |
headers
Michael | 3 Apr 2005 01:47
Picon
Favicon

Re: the 8600 lives!

Hello,

> Please correct me if im wrong but I do believe that your power-off 
> problem has nothing to do with the "exquisite" corrections that Michael 
> has done for X11(where would we be without them!)..

> I had very much the same problem, but Allen Briggs did wonders in 
> correcting the power-off problems for my Powerbook 3500(and his own 
> 3400)...
Yes, apparently the problem is still present in the kernel Miles got - here's a more recent one:
http://macallan.homeunix.org:6704/stuff/BSD/netbsd_GENERIC_voodoo5.bz2
It contains some additional console drivers ( for ATI mach64 and Voodoo3 with some experimental Voodoo5
support ) - this one should also contain Allen's fixes, provided they were in -current yesterday.

have fun
Michael.
Permalink | Reply | 0 comments |

Gmane