[parisc-linux] Re: HP C180 video card

[parisc-linux] Re: HP C180 video card

[Grant Grundler]

On Sun, Jun 15, 2003 at 04:11:46PM +0000, Joel Soete wrote:
> I tried to use a "PGX64 alias Rage xl " pluged in the first slot of my 
> b180L. It seems to be well supported by kernel 2.4.21 (modeprobe 
> ati128fb; # modules were well loaded).

cool - that's a good first step. I don't know where to go from there.

> But no success to make XFree86 
> works (fbset /dev/fb0 not found)

Maybe it's as simple as "mknod /dev/fb0 ..." ?
if the driver loads without complaining, I'd expect fbset command
to work too.

> and even when I change console path to 
> graphics(1[a]) I never reach to obtain boot prompt on the screen (which 
> never seems to recieve init signal?).

PDC/IODC/STI won't know anything about the Rage XL card.
There's no way you can use that as a console.

> I do not understand, it seems to be a common pci gfx card? any idea?

I don't either really. Just guessing.

grant

RE: [parisc-linux] Re: HP C180 video card

[Joel Soete]

>>
>>On Sun, Jun 15, 2003 at 04:11:46PM +0000, Joel Soete wrote:
>> I tried to use a "PGX64 alias Rage xl " pluged in the first slot of my

>> b180L. It seems to be well supported by kernel 2.4.21 (modeprobe 
>> ati128fb; # modules were well loaded).

# lsmod
Module                  Size  Used by    Not tainted
aty128fb               20244   0  (unused)
fbcon-cfb32             5956   0  [aty128fb]
fbcon-cfb8              5244   0  [aty128fb]
fbcon-cfb24             5876   0  [aty128fb]
fbcon-cfb16             5828   0  [aty128fb]

>cool - that's a good first step. I don't know where to go from there.
>
>> But no success to make XFree86 
>> works (fbset /dev/fb0 not found)

>Maybe it's as simple as "mknod /dev/fb0 ..." ?

I do it (rm /dev/fb*; cd /dev ; MAKEDEV fb; fb* were well rebuild) but no
more success?

>> if the driver loads without complaining, I'd expect fbset command
>> to work too.

Hope so too.

Even "cat /proc/fb" don't show me anything?

>> and even when I change console path to 
>> graphics(1[a]) I never reach to obtain boot prompt on the screen (which

>> never seems to recieve init signal?).
>
>PDC/IODC/STI won't know anything about the Rage XL card.
>There's no way you can use that as a console.

Ok so I swicth to "path con serial_1"

>> I do not understand, it seems to be a common pci gfx card? any idea?

>I don't either really. Just guessing.

Just in case some more X(pert) read this mail, here is some more info:

root@fw02:/CVS/SRC.TMP# lspci
00:01.0 VGA compatible controller: ATI Technologies Inc Rage XL (rev 27)
00:02.0 Serial controller: Equinox Systems, Inc.: Unknown device 90f0 (rev
04)
...

/var/log/XFree86.0.log:
[...]
(WW) Open APM failed (/dev/apm_bios) (No such file or directory)
(II) PCI: PCI scan (all values are in hex)
(II) PCI: 00:01:0: chip 1002,4752 card 0000,0000 rev 27 class 03,00,00 hdr
00
(II) PCI: 00:02:0: chip 113f,90f0 card 0000,0000 rev 04 class 07,00,80 hdr
00
(II) PCI: 00:13:0: chip 1000,000f card 0000,0000 rev 04 class 01,00,00 hdr
00
(II) PCI: 00:14:0: chip 1011,0019 card 103c,104f rev 41 class 02,00,00 hdr
00
(II) PCI: End of PCI scan
(II) Host-to-PCI bridge:
(II) Bus 0: bridge is at (0:0:0), (-1,0,0), BCTRL: 0x00 (VGA_EN is cleared)
(II) Bus 0 I/O range:
        [0] -1 0        0x00000000 - 0x0000ffff (0x10000) IX[B]
(II) Bus 0 non-prefetchable memory range:
        [0] -1 0        0x00000000 - 0xffffffff (0x0) MX[B]
(II) Bus 0 prefetchable memory range:
        [0] -1 0        0x00000000 - 0xffffffff (0x0) MX[B]
(--) PCI:*(0:1:0) ATI Mach64 GR rev 39, Mem @ 0xf1000000/24, 0xf0fec000/12,
I/O @ 0xfd00/8, BIOS @ 0xf0f80000/17
(II) Addressable bus resource ranges are
        [0] -1 0        0x00000000 - 0xffffffff (0x0) MX[B]
        [1] -1 0        0x00000000 - 0x0000ffff (0x10000) IX[B]
[...]
(II) FBDEV: driver for framebuffer: fbdev
(II) VGA: Generic VGA driver (version 4.0) for chipsets: generic
(II) v4l driver for Video4Linux
(II) Primary Device is: PCI 00:01:0
(EE) open /dev/fb0: No such device
(EE) No devices detected.

Fatal server error:
no screens found


Thanks for your attention,
    Joel


---------------------------------
Découvrez les 6 clés et gagnez le Club Med à Vie avec Tiscali
http://www.tiscali.be/nl/subs/tiscali4life/default.asp?lang=fr

[parisc-linux] ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Joel Soete]

Hi Grant,

Grant Grundler wrote:

>On Sun, Jun 15, 2003 at 04:11:46PM +0000, Joel Soete wrote:
>  
>
>>I tried to use a "PGX64 alias Rage xl " pluged in the first slot of my 
>>b180L. It seems to be well supported by kernel 2.4.21 (modeprobe 
>>ati128fb; # modules were well loaded).
>>    
>>
>
>cool - that's a good first step. I don't know where to go from there.
>  
>
Hmm in fact I obtain better results with atyfb + "Mach64 CT/VT/GT/LT 
(incl. 3D RAGE) support (NEW)";
this time I obtain such message for kernel advising me that 'atyfb' 
modules is loaded and some other info collected from the device itself.

>>But no success to make XFree86 
>>works (fbset /dev/fb0 not found)
>>    
>>
I also reach to get (not yet changing) default setup with "fbset -fb 
/dev/fb1" :-)
But not yet find the way to setup X using /dev/fb1 in place of fb0 ;-( ?

Let see later because my very problem is there:
a) if I rmmod atyfb the system crashes
b) if I do not include "Generic sti fb" in the config of my kernel, when 
I insmod atyfb the system crashes also
c) if I try to use Equinox serial mux (8P) (an oem HP product) module 
(with our without your patch Grant) the system crashes also?

The system is a B180 model using Dino GSCPCI bridge and in the two cases 
(even if messages are a bit different) the TOC shows always a pb on this 
bus (the analyse of toc after the crash of insmod eqnx point out a pb in 
timer_? routine). And in fact I figure out that I never have to use this 
bridge for another card?
Is somebody else used it successfuly or could it be broken?

Thanks again for any help,
    Joel

[parisc-linux] ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Joel Soete]

Hi Grant,

Grant Grundler wrote:

>On Sun, Jun 15, 2003 at 04:11:46PM +0000, Joel Soete wrote:
>  
>
>>I tried to use a "PGX64 alias Rage xl " pluged in the first slot of my 
>>b180L. It seems to be well supported by kernel 2.4.21 (modeprobe 
>>ati128fb; # modules were well loaded).
>>    
>>
>
>cool - that's a good first step. I don't know where to go from there.
>  
>
Hmm in fact I obtain better results with atyfb + "Mach64 CT/VT/GT/LT 
(incl. 3D RAGE) support (NEW)";
this time I obtain such message for kernel advising me that 'atyfb' 
modules is loaded and some other info collected from the device itself.

>>But no success to make XFree86 
>>works (fbset /dev/fb0 not found)
>>    
>>
I also reach to get (not yet changing) default setup with "fbset -fb 
/dev/fb1" :-)
But not yet find the way to setup X using /dev/fb1 in place of fb0 ;-( ?

Let see later because my very problem is there:
a) if I rmmod atyfb the system crashes
b) if I do not include "Generic sti fb" in the config of my kernel, when 
I insmod atyfb the system crashes also
c) if I try to use Equinox serial mux (8P) (an oem HP product) module 
(with our without your patch Grant) the system crashes also?

The system is a B180 model using Dino GSCPCI bridge and in the two cases 
(even if messages are a bit different) the TOC shows always a pb on this 
bus (the analyse of toc after the crash of insmod eqnx point out a pb in 
timer_? routine). And in fact I figure out that I never have to use this 
bridge for another card?
Is somebody else used it successfuly or could it be broken?

Thanks again for any help,
    Joel

[parisc-linux] Re: ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Grant Grundler]

On Sat, Jun 21, 2003 at 02:24:41PM +0000, Joel Soete wrote:
...
> c) if I try to use Equinox serial mux (8P) (an oem HP product) module 
> (with our without your patch Grant) the system crashes also?

I haven't tested the Serial 8p PCI card under parisc yet.
To date, that card has always lived in a x86 host. I just
haven't had time to play with it under parisc - I've needed it
working to get other things done.


> The system is a B180 model using Dino GSCPCI bridge and in the two cases 
> (even if messages are a bit different) the TOC shows always a pb on this 
> bus (the analyse of toc after the crash of insmod eqnx point out a pb in 
> timer_? routine).

The symptom is very likely only similar and not identical.
I'd find it very odd if both cards crashed for the same reason.

AFAICT, eqnx driver doesn't call pci_enable_device().  The bug I'd be
looking for is a PCI "resource" (BAR) to be accessed before it was enabled.
Try adding a pci_enable_device() to eqnx_init() or eqx_pci_buspresent().
[ *gag* I'm surprised this code even works on x86. ]

PARISC "hard fails" when reading from disabled or non-existant PCI
address space while x86 "soft fails".  ie x86 gets garbage back on
a "bad" read while PARISC will crash.

The crash in the timer routine is interesting and would make it
interesting for me to move the equinox card to an a500.
Randolph, Paul (Bame), and I occcasionally try to track down the
timer problems that plague SMP under 64-bit parisc kernels.


> And in fact I figure out that I never have to use this 
> bridge for another card?

GSCtoPCI connects the PCI slot to the rest of the system.
Try "lspci" and see what you have connected.

> Is somebody else used it successfuly or could it be broken?

Dino worked fine before.
I ran B180 as my desktop for over a year (my desktop is now a C3000).

grant

Re: [parisc-linux] Re: ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Matthew Wilcox]

On Sun, Jun 22, 2003 at 11:04:50PM -0600, Grant Grundler wrote:
> AFAICT, eqnx driver doesn't call pci_enable_device().  The bug I'd be
> looking for is a PCI "resource" (BAR) to be accessed before it was enabled.
> Try adding a pci_enable_device() to eqnx_init() or eqx_pci_buspresent().
> [ *gag* I'm surprised this code even works on x86. ]

x86 typically doesn't need to call pci_enable_device, so people who've
never tried their driver on any other box may have missed it.

-- 
"It's not Hollywood.  War is real, war is primarily not about defeat or
victory, it is about death.  I've seen thousands and thousands of dead bodies.
Do you think I want to have an academic debate on this subject?" -- Robert Fisk

Re: [parisc-linux] Re: ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Joel Soete]

Matthew Wilcox wrote:

>On Sun, Jun 22, 2003 at 11:04:50PM -0600, Grant Grundler wrote:
>  
>
>>AFAICT, eqnx driver doesn't call pci_enable_device().  The bug I'd be
>>looking for is a PCI "resource" (BAR) to be accessed before it was enabled.
>>Try adding a pci_enable_device() to eqnx_init() or eqx_pci_buspresent().
>>[ *gag* I'm surprised this code even works on x86. ]
>>    
>>
>
>x86 typically doesn't need to call pci_enable_device, so people who've
>never tried their driver on any other box may have missed it.
>
>  
>
 From the very last relase 4.05 (dated April 7 iirc) it support also 
"Intel IA-64 platforms".
But I will have a look in more details and also post the toc analyse.

Thanks for help and attention,
    Joel

Re: [parisc-linux] Re: ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Grant Grundler]

On Mon, Jun 23, 2003 at 12:18:36PM +0100, Matthew Wilcox wrote:
> x86 typically doesn't need to call pci_enable_device, so people who've
> never tried their driver on any other box may have missed it.

Happen to know if that is true for hotplug support?
ie does x86 requires something like pci_enable_device() for hotplug to work?

BTW, my comment was actually intended for other parts of the code I
was digging though, not the just lack of pci_enable_device().

grant

Re: [parisc-linux] Re: ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Grant Grundler]

On Mon, Jun 23, 2003 at 04:00:08PM +0000, Joel Soete wrote:
> From the very last relase 4.05 (dated April 7 iirc) it support also 
> "Intel IA-64 platforms".

Those behave the same way as ia32 boxes.
Wonder if I can get a "sample card" to test an HP IA64 platform
(ZX1).... :^)


> But I will have a look in more details and also post the toc analyse.

yes, please do.

thanks,
grant

Re: [parisc-linux] Re: ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Grant Grundler]

On Mon, Jun 23, 2003 at 04:00:08PM +0000, Joel Soete wrote:
> From the very last relase 4.05 (dated April 7 iirc) it support also 
> "Intel IA-64 platforms".


Joel,
Could you post a URL or make the eqnx-4.05 linux driver available?
I didn't see it here:
ftp://guest:guest@www.equinox.com/library/legacy%20drivers/sst/

Last one I found was the 4.01 from last year...

thanks,
grant

Re: [parisc-linux] Re: ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Alan Cox]

On Maw, 2003-06-24 at 04:44, Grant Grundler wrote:
> On Mon, Jun 23, 2003 at 12:18:36PM +0100, Matthew Wilcox wrote:
> > x86 typically doesn't need to call pci_enable_device, so people who've
> > never tried their driver on any other box may have missed it.
> 
> Happen to know if that is true for hotplug support?
> ie does x86 requires something like pci_enable_device() for hotplug to work?

Yes

Re: [parisc-linux] Re: ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Joel Soete]

Hi Grant,

> BTW, my comment was actually intended for other parts of the code I
> was digging though, not the just lack of pci_enable_device().

Do you know where can I get more info on what have todo (naturaly only pci_enable_device()
doesn't help)?

Thanks,
    Joel



---------------------------------
Tiscali ADSL: 19,50 euros/mois...abonnez-vous sur www.tiscali.be

Re: [parisc-linux] Re: ATY RAGE XL progress or Dino gscpci bridge pb?[was: Re: HP C180 video card]

[Grant Grundler]

On Wed, Jun 25, 2003 at 11:43:51AM +0200, Joel Soete wrote:
> Do you know where can I get more info on what have todo
> (naturaly only pci_enable_device() doesn't help)?

Track down where in the code the MMIO read fails.
If it's crashing *after* pci_enable_device() is called,
then need to make sure the driver isn't directly reading
PCI BARs and is using struct resources in the pci_dev.

grant