Radeon 7000 in 9500 - which slot?

Radeon 7000 in 9500 - which slot?

[Stefan Jeglinski]

I have seen in Linux-PPC archives (mostly YDL archives) that a Radeon
7000 must live in the top slot of a 9500 to work in Linux. Any truth
to this rumor still? Indeed I have this card in the 4th slot, and I
get a crash into the debugger while booting 2.4.20pre5. Should I
provide a backtrace or just put it in the top slot?



Stefan Jeglinski

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Radeon 7000 in 9500 - which slot?

[Benjamin Herrenschmidt]

>
>I have seen in Linux-PPC archives (mostly YDL archives) that a Radeon
>7000 must live in the top slot of a 9500 to work in Linux. Any truth
>to this rumor still? Indeed I have this card in the 4th slot, and I
>get a crash into the debugger while booting 2.4.20pre5. Should I
>provide a backtrace or just put it in the top slot?

Yes, provide a backtrace. I suspect you need the radeon in a slot
where the low IO addresses are available for XFree to work, but
neither radeonfb nor offb should have this limitation, so the
crash you have is a bug.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Radeon 7000 in 9500 - which slot?

[Michel Dänzer]

On Don, 2002-10-10 at 11:51, Benjamin Herrenschmidt wrote:
>
> >
> >I have seen in Linux-PPC archives (mostly YDL archives) that a Radeon
> >7000 must live in the top slot of a 9500 to work in Linux. Any truth
> >to this rumor still? Indeed I have this card in the 4th slot, and I
> >get a crash into the debugger while booting 2.4.20pre5. Should I
> >provide a backtrace or just put it in the top slot?
>
> Yes, provide a backtrace. I suspect you need the radeon in a slot
> where the low IO addresses are available for XFree to work,

Shouldn't matter with Option "UseFBDev".


--
Earthling Michel Dänzer (MrCooper)/ Debian GNU/Linux (powerpc) developer
XFree86 and DRI project member   /  CS student, Free Software enthusiast


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Radeon 7000 in 9500 - which slot?

[Olaf Hering]

begin  On Thu, Oct 10, Benjamin Herrenschmidt wrote:

>
> >
> >I have seen in Linux-PPC archives (mostly YDL archives) that a Radeon
> >7000 must live in the top slot of a 9500 to work in Linux. Any truth
> >to this rumor still? Indeed I have this card in the 4th slot, and I
> >get a crash into the debugger while booting 2.4.20pre5. Should I
> >provide a backtrace or just put it in the top slot?
>
> Yes, provide a backtrace. I suspect you need the radeon in a slot
> where the low IO addresses are available for XFree to work, but
> neither radeonfb nor offb should have this limitation, so the
> crash you have is a bug.

we had this one before.

pci_device_to_OF_node() returns null and you pass this to get_property()


Gruss Olaf

--
 $ man Notes

BUGS
       You need http://www.sauerstoff-laden.de/ soon...

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Radeon 7000 in 9500 - which slot?

[Benjamin Herrenschmidt]

>we had this one before.
>
>pci_device_to_OF_node() returns null and you pass this to get_property()

Where ? Can you elaborate ? inside radeonfb ? pci_device_to_OF_node
should not return NULL, I though I fixed that...

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Radeon 7000 in 9500 - which slot?

[Olaf Hering]

begin  On Thu, Oct 10, Benjamin Herrenschmidt wrote:

> >we had this one before.
> >
> >pci_device_to_OF_node() returns null and you pass this to get_property()
>
> Where ? Can you elaborate ? inside radeonfb ? pci_device_to_OF_node
> should not return NULL, I though I fixed that...

Look in the first few lines of
arch/ppc/kernel/pci.c:pci_device_to_OF_node()
lots of return NULL; there ;)

OF doesnt init the card in some of the slots -> no OF properties.


Gruss Olaf

--
 $ man Notes

BUGS
       You need http://www.sauerstoff-laden.de/ soon...

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Radeon 7000 in 9500 - which slot?

[Benjamin Herrenschmidt]

>> Where ? Can you elaborate ? inside radeonfb ? pci_device_to_OF_node
>> should not return NULL, I though I fixed that...
>
>Look in the first few lines of
>arch/ppc/kernel/pci.c:pci_device_to_OF_node()
>lots of return NULL; there ;)
>
>OF doesnt init the card in some of the slots -> no OF properties.

Ah, you mean no OF properties at all ? ok, that's bad. I'll
make sure radeonfb is robust enough. The problem is that without
those properties, we won't know the card's clock, so we'll have
to guess it, that may not work.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Radeon 7000 in 9500 - which slot?

[Olaf Hering]

begin  On Fri, Oct 11, Benjamin Herrenschmidt wrote:

> >> Where ? Can you elaborate ? inside radeonfb ? pci_device_to_OF_node
> >> should not return NULL, I though I fixed that...
> >
> >Look in the first few lines of
> >arch/ppc/kernel/pci.c:pci_device_to_OF_node()
> >lots of return NULL; there ;)
> >
> >OF doesnt init the card in some of the slots -> no OF properties.
>
> Ah, you mean no OF properties at all ? ok, that's bad. I'll
> make sure radeonfb is robust enough. The problem is that without
> those properties, we won't know the card's clock, so we'll have
> to guess it, that may not work.

Yes, I did not say it works in this config, but it does not oops at
least ;)

Gruss Olaf

--
 $ man Notes

BUGS
       You need http://www.sauerstoff-laden.de/ soon...

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

part deux Re: Radeon 7000 in 9500 - which slot?

[Stefan Jeglinski]

Have moved the 7000 to the top slot on the 9500, 2 monitors attached.

When using 2.4.20-pre9-ben0, booting into RL3 (not troubleshooting X
at this point) and kernel arg video=radeon:1024x768-16@75, the
console boot briefly appears to start in accelerated mode, then the
monitor goes black and stays that way. However, I can tell from
listening to the disks that the boot proceeds normally.

Typing blindly to get the dmesg, we have:

	PCI: Enabling device 00:0d.0 (0086 -> 0087)
	radeonfb: ref_clk=2700, ref_div=12, xclk=18277 from OF
	radeon_get_moninfo: bios 4 scratch = 0
	Console: switching to colour frame buffer device 128x48
	radeonfb: ATI Radeon VE QY DDR SGRAM 32 MB
	radeonfb: DVI port no monitor connected
	radeonfb: CRT port no monitor connected

Am I wrong, or does it appear to think I have no monitors connected?
The main monitor's power LED indicates it has sync (green), while the
second monitor's power LED indicates not (yellow).

FWIW, lspci -vv yields

00:0d.0 VGA compatible controller: ATI Technologies Inc: Unknown
device 5159 (prog-if 00 [VGA])
	Subsystem: ATI Technologies Inc: Unknown device 0908
	Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop-
ParErr- Stepping+ SERR- FastB2B-
	Status: Cap+ 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort-
<TAbort- <MAbort- >SERR- <PERR-
	Latency: 8 min, 32 set, cache line size 08
	Interrupt: pin A routed to IRQ 23
	Region 0: Memory at 88000000 (32-bit, prefetchable) [size=128M]
	Region 1: I/O ports at 0400 [size=256]
	Region 2: Memory at 80910000 (32-bit, non-prefetchable) [size=64K]
	Expansion ROM at 80920000 [disabled] [size=128K]
	Capabilities: [50] Power Management version 2
		Flags: PMEClk- AuxPwr- DSI- D1+ D2+ PME-
		Status: D0 PME-Enable- DSel=0 DScale=0 PME-


Stefan Jeglinski


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: part deux Re: Radeon 7000 in 9500 - which slot?

[Benjamin Herrenschmidt]

>Have moved the 7000 to the top slot on the 9500, 2 monitors attached.
>
>When using 2.4.20-pre9-ben0, booting into RL3 (not troubleshooting X
>at this point) and kernel arg video=radeon:1024x768-16@75, the
>console boot briefly appears to start in accelerated mode, then the
>monitor goes black and stays that way. However, I can tell from
>listening to the disks that the boot proceeds normally.

What kind of monitors ? 2 CRTs ? 1 LCD and 1 CRT ?

The problem is apparently due to radeon_get_moninfo not finding
anything, can you eventually force rinfo->crtDisp_type to MT_CRT
and see if that helps ?

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: part deux Re: Radeon 7000 in 9500 - which slot?

[Olaf Hering]

begin  On Wed, Oct 16, Benjamin Herrenschmidt wrote:

>
> >Have moved the 7000 to the top slot on the 9500, 2 monitors attached.
> >
> >When using 2.4.20-pre9-ben0, booting into RL3 (not troubleshooting X
> >at this point) and kernel arg video=radeon:1024x768-16@75, the
> >console boot briefly appears to start in accelerated mode, then the
> >monitor goes black and stays that way. However, I can tell from
> >listening to the disks that the boot proceeds normally.
>
> What kind of monitors ? 2 CRTs ? 1 LCD and 1 CRT ?
>
> The problem is apparently due to radeon_get_moninfo not finding
> anything, can you eventually force rinfo->crtDisp_type to MT_CRT
> and see if that helps ?

I see something similar here.
It does work when booting with BootX App, but not with miboot.


Gruss Olaf

--
 $ man Notes

BUGS
       You need http://www.sauerstoff-laden.de/ soon...

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: part deux Re: Radeon 7000 in 9500 - which slot?

[Ani Joshi]

It is not detecting the heads properly because on many "mac" radeons, the
firmware doesn't initialize the SCRATCH registers properly which contain
information about display type.  There is no reliable method of probing
display type on ppc so I will end up adding a display type option to
radeonfb next version.


ani


On Wed, 16 Oct 2002, Benjamin Herrenschmidt wrote:

>
> >Have moved the 7000 to the top slot on the 9500, 2 monitors attached.
> >
> >When using 2.4.20-pre9-ben0, booting into RL3 (not troubleshooting X
> >at this point) and kernel arg video=radeon:1024x768-16@75, the
> >console boot briefly appears to start in accelerated mode, then the
> >monitor goes black and stays that way. However, I can tell from
> >listening to the disks that the boot proceeds normally.
>
> What kind of monitors ? 2 CRTs ? 1 LCD and 1 CRT ?
>
> The problem is apparently due to radeon_get_moninfo not finding
> anything, can you eventually force rinfo->crtDisp_type to MT_CRT
> and see if that helps ?
>
> Ben.
>
>
>


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: part deux Re: Radeon 7000 in 9500 - which slot?

[Stefan Jeglinski]

>It is not detecting the heads properly because on many "mac" radeons, the
>firmware doesn't initialize the SCRATCH registers properly which contain
>information about display type.  There is no reliable method of probing
>display type on ppc so I will end up adding a display type option to
>radeonfb next version.

Sounds like until you have a chance to do this, I'm outta luck on
accelerated video. Keep me in mind for beta testing, please :-) Can I
assume I will be able to use the card with offb as usual? What about
dual monitors with xinerama - should that still work, albeit
unaccelerated?

Odd... surely there's no way I'm the only one with Mac and a 7000,
eh? Wonder why I never ran across this issue in searching the
archives... the only citation I found was the issue about it being in
the top slot, but I had the impression that the correspondent had it
working... will go look that up again.


Stefan Jeglinski

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: part deux Re: Radeon 7000 in 9500 - which slot?

[Stefan Jeglinski]

>  > The problem is apparently due to radeon_get_moninfo not finding
>>  anything, can you eventually force rinfo->crtDisp_type to MT_CRT
>>  and see if that helps ?
>
>I see something similar here.
>It does work when booting with BootX App, but not with miboot.

I'm using BootX, so I would respectfully disagree :-)


Stefan Jeglinski

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: part deux Re: Radeon 7000 in 9500 - which slot?

[Benjamin Herrenschmidt]

>It is not detecting the heads properly because on many "mac" radeons, the
>firmware doesn't initialize the SCRATCH registers properly which contain
>information about display type.  There is no reliable method of probing
>display type on ppc so I will end up adding a display type option to
>radeonfb next version.

Well... It should be possible to get that info from OF properties
or is the OF driver broken enough to not export that either ?

How does the MacOS driver to probe displays ? Isn't it possible
to just try the various i2c channels until an EDID is found to
figure what's up ? The CRT port can definitely probe for the
presence of the display by measuring the load on the analog
outputs.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: part deux Re: Radeon 7000 in 9500 - which slot?

[Olaf Hering]

begin  On Wed, Oct 16, Stefan Jeglinski wrote:

>
> > > The problem is apparently due to radeon_get_moninfo not finding
> >> anything, can you eventually force rinfo->crtDisp_type to MT_CRT
> >> and see if that helps ?
> >
> >I see something similar here.
> >It does work when booting with BootX App, but not with miboot.
>
> I'm using BootX, so I would respectfully disagree :-)

I used the novideo driver checkbox once.
But with this checkbox disabled and 1024x768-75 in MacOS, all I get is a
stripe on the screen.
Using 640x480-120 in MacOS and it works also in Linux.
I must note that the XF 4.2 radeon driver can open the display
correctly, even with broken kernel framebuffer driver.


Gruss Olaf

--
 $ man Notes

BUGS
       You need http://www.sauerstoff-laden.de/ soon...

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/