Problem with i820 AGP patch

Problem with i820 AGP patch

[Daniele Venzano]

Your patch to add AGP support for i820 chipset doesn't work for me, if I
load agpgart module with agp_try_unsupported=1 I get:

-
Linux agpgart interface v0.99 (c) Jeff Hartmann
agpgart: Maximum main memory to use for agp memory: 262M
agpgart: Trying generic Intel routines for device id: 2501
						^^^^^^^^^^^^^
agpgart: AGP aperture is 256M @ 0xd0000000
-

That device id is different from the corresponding line in apg.h:174
#define PCI_DEVICE_ID_INTEL_820_0       0x2500

I tried to change that line in:
#define PCI_DEVICE_ID_INTEL_820_0       0x2501

And it worked! But I don't know why the id for your chip is different
from mine... 

My motherboard is an Asus P3C2000 with a P3 500 running kernel 2.4.14


I've also another problem (I think AGP related) with DRI/DRM ang XFree
4.1.0, but I'm still working on it.
If I try running some OpenGL app with more than 256Mb of RAM, the system hangs
(need hard reset, no SysReq works). This happens with 2.4.13 and 2.4.14,
before I hadn't tried since I had only 128Mb...)
Someone can please tell me if this problem has something to do with
having a "Maximum main memory to use for agp memory" value grater than
AGP aperture value ?


Regards,
-- 
------------------------------------------------------------
Daniele Venzano
Senior member of the Linux User Group Genova, Italy (LUGGe)
E-Mail: venza@iol.it
Web: http://digilander.iol.it/webvenza/
LUGGe: http://lugge.ziobudda.net

Re: Problem with i820 AGP patch

[lists]

  I have a similar problem with i860 - using 2.4.15-pre4.

  I get on a dell 530 dual p4 + radeon video:

Nov 14 00:31:40 flash kernel: Linux agpgart interface v0.99 (c) Jeff Hartmann
Nov 14 00:31:40 flash kernel: agpgart: Maximum main memory to use 
                for agp memory: 816M
Nov 14 00:31:40 flash kernel: agpgart: Unsupported Intel chipset (device id: 2531), 
                you might want to try agp_try_unsupported=1.


   agp.h has 850 id as 2530 and 860 as 8532.

   My 860 claims 2531 ... ...

   Regards

   Gene/

lspci -v:
00:00.0 Host bridge: Intel Corporation 82850 860 (Wombat) Chipset Host Bridge (MCH) (rev 04)
        Subsystem: Dell Computer Corporation: Unknown device 00d8
        Flags: bus master, fast devsel, latency 0
        Memory at e8000000 (32-bit, prefetchable) [size=128M]
        Capabilities: [a0] AGP version 2.0

00:01.0 PCI bridge: Intel Corporation 82850 850 (Tehama) Chipset AGP Bridge (rev 04) (prog-if 00 [Normal decode])
        Flags: bus master, 66Mhz, fast devsel, latency 64
        Bus: primary=00, secondary=01, subordinate=01, sec-latency=64
        I/O behind bridge: 0000e000-0000efff
        Memory behind bridge: ff900000-ffafffff
        Prefetchable memory behind bridge: f0000000-f7ffffff

00:02.0 PCI bridge: Intel Corporation 82860 860 (Wombat) Chipset AGP Bridge (rev 04) (prog-if 00 [Normal decode])

   etc.




On Wed, Nov 14, 2001 at 08:51:41PM +0100, Daniele Venzano wrote:
> Your patch to add AGP support for i820 chipset doesn't work for me, if I
> load agpgart module with agp_try_unsupported=1 I get:
> 
> -
> Linux agpgart interface v0.99 (c) Jeff Hartmann
> agpgart: Maximum main memory to use for agp memory: 262M
> agpgart: Trying generic Intel routines for device id: 2501
> 						^^^^^^^^^^^^^
> agpgart: AGP aperture is 256M @ 0xd0000000
> -
> 
> That device id is different from the corresponding line in apg.h:174
> #define PCI_DEVICE_ID_INTEL_820_0       0x2500
> 
> I tried to change that line in:
> #define PCI_DEVICE_ID_INTEL_820_0       0x2501
> 
> And it worked! But I don't know why the id for your chip is different
> from mine... 
> 
> My motherboard is an Asus P3C2000 with a P3 500 running kernel 2.4.14
> 
> 

Re: Problem with i820 AGP patch

[Marvin Justice]

I noticed this too with Tyan Thunder i860. Adding a new entry to agp.h, eg.

#ifndef PCI_DEVICE_ID_INTEL_860_DP_0
#define PCI_DEVICE_ID_INTEL_860_DP_0     0x2531
#endif

along with appropriate modifcations to agpgart_be.c:

	{ PCI_DEVICE_ID_INTEL_860_DP_0,
	        PCI_VENDOR_ID_INTEL,
	        INTEL_I860,
	        "Intel",
	        "i860",
	        intel_860_setup },

seems to work ok for me. On the other hand, the performance is pretty much 
the same as "agp_try_unsupported=1" :-). How does your /proc/interrupts look?

-- 
Marvin Justice
Software Developer
BOXX Technologies, Inc.
www.boxxtech.com
512-235-6318 (V)
512-835-0434 (F)

On Thursday 15 November 2001 12:27 am, lists@sapience.com wrote:
>   I have a similar problem with i860 - using 2.4.15-pre4.
>
>   I get on a dell 530 dual p4 + radeon video:
>
> Nov 14 00:31:40 flash kernel: Linux agpgart interface v0.99 (c) Jeff
> Hartmann Nov 14 00:31:40 flash kernel: agpgart: Maximum main memory to use
>                 for agp memory: 816M
> Nov 14 00:31:40 flash kernel: agpgart: Unsupported Intel chipset (device
> id: 2531), you might want to try agp_try_unsupported=1.
>
>
>    agp.h has 850 id as 2530 and 860 as 8532.
>
>    My 860 claims 2531 ... ...
>
>    Regards
>
>    Gene/
>
> lspci -v:
> 00:00.0 Host bridge: Intel Corporation 82850 860 (Wombat) Chipset Host
> Bridge (MCH) (rev 04) Subsystem: Dell Computer Corporation: Unknown device
> 00d8
>         Flags: bus master, fast devsel, latency 0
>         Memory at e8000000 (32-bit, prefetchable) [size=128M]
>         Capabilities: [a0] AGP version 2.0
>
> 00:01.0 PCI bridge: Intel Corporation 82850 850 (Tehama) Chipset AGP Bridge
> (rev 04) (prog-if 00 [Normal decode]) Flags: bus master, 66Mhz, fast
> devsel, latency 64
>         Bus: primary=00, secondary=01, subordinate=01, sec-latency=64
>         I/O behind bridge: 0000e000-0000efff
>         Memory behind bridge: ff900000-ffafffff
>         Prefetchable memory behind bridge: f0000000-f7ffffff
>
> 00:02.0 PCI bridge: Intel Corporation 82860 860 (Wombat) Chipset AGP Bridge
> (rev 04) (prog-if 00 [Normal decode])
>
>    etc.
>
> On Wed, Nov 14, 2001 at 08:51:41PM +0100, Daniele Venzano wrote:
> > Your patch to add AGP support for i820 chipset doesn't work for me, if I
> > load agpgart module with agp_try_unsupported=1 I get:
> >
> > -
> > Linux agpgart interface v0.99 (c) Jeff Hartmann
> > agpgart: Maximum main memory to use for agp memory: 262M
> > agpgart: Trying generic Intel routines for device id: 2501
> > 						^^^^^^^^^^^^^
> > agpgart: AGP aperture is 256M @ 0xd0000000
> > -
> >
> > That device id is different from the corresponding line in apg.h:174
> > #define PCI_DEVICE_ID_INTEL_820_0       0x2500
> >
> > I tried to change that line in:
> > #define PCI_DEVICE_ID_INTEL_820_0       0x2501
> >
> > And it worked! But I don't know why the id for your chip is different
> > from mine...
> >
> > My motherboard is an Asus P3C2000 with a P3 500 running kernel 2.4.14
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: Problem with i820 AGP patch

[Nicolas Aspert]

Daniele Venzano wrote:

> Your patch to add AGP support for i820 chipset doesn't work for me, if I
> load agpgart module with agp_try_unsupported=1 I get:
> 
> -
> Linux agpgart interface v0.99 (c) Jeff Hartmann
> agpgart: Maximum main memory to use for agp memory: 262M
> agpgart: Trying generic Intel routines for device id: 2501
> 						^^^^^^^^^^^^^
> agpgart: AGP aperture is 256M @ 0xd0000000
> -


Here is what I have in 'drivers/pci/pci.ids' :

         2500  82820 820 (Camino) Chipset Host Bridge (MCH)
                 1043 801c  P3C-2000 system chipset
         2501  82820 820 (Camino) Chipset Host Bridge (MCH)
                 1043 801c  P3C-2000 system chipset
         250b  82820 820 (Camino) Chipset Host Bridge
         250f  82820 820 (Camino) Chipset PCI to AGP Bridge

So I guess that 0x2500 and 0x2501 should be initialized in the same way, 
through the same routines. Unfortunately, I dont have such a machine to 
test... Can you send what 'lspci -ev' shows ?


> That device id is different from the corresponding line in apg.h:174
> #define PCI_DEVICE_ID_INTEL_820_0       0x2500
> 
> I tried to change that line in:
> #define PCI_DEVICE_ID_INTEL_820_0       0x2501
> 
> And it worked! But I don't know why the id for your chip is different
> from mine... 
> 


It worked, but in which way ? Did you test X+OpenGL apps to see whether 
it did the trick ? The fact that the module loads itself is not always 
sufficient to say that it works (believe me :-)


> My motherboard is an Asus P3C2000 with a P3 500 running kernel 2.4.14
> 
> 
> I've also another problem (I think AGP related) with DRI/DRM ang XFree
> 4.1.0, but I'm still working on it.
> If I try running some OpenGL app with more than 256Mb of RAM, the system hangs
> (need hard reset, no SysReq works). This happens with 2.4.13 and 2.4.14,
> before I hadn't tried since I had only 128Mb...)
> Someone can please tell me if this problem has something to do with
> having a "Maximum main memory to use for agp memory" value grater than
> AGP aperture value ?
> 

With your custom patch, or with the 'generic' stuff ?

Regards.
-- 
Nicolas Aspert      Signal Processing Laboratory (LTS)
Swiss Federal Institute of Technology (EPFL)

Re: Problem with i820 AGP patch

[Nicolas Aspert]

Marvin Justice wrote:


 On the other hand, the performance is pretty much 
> the same as "agp_try_unsupported=1" :-). How does your /proc/interrupts look?
> 
> 
This is not really surprising. If you look through 'agpgart_be.c', you can see that 

only the initialization is different. All the rest goes through the same generic routines.

There are already too many of these initializations in the file... but I 
think that R. Love is working on a cleaner version.

Regards

-- 
Nicolas Aspert      Signal Processing Laboratory (LTS)
Swiss Federal Institute of Technology (EPFL)

Re: Problem with i820 AGP patch

[Daniele Venzano]

On Thu, Nov 15, 2001 at 09:01:54AM +0100, Nicolas Aspert wrote:
> test... Can you send what 'lspci -ev' shows ?

Here it is, I think I should have thought before...
-ev didn't worked, I used -vn with a cut & paste for names (first line
of each entry)

00:00.0 Host bridge: Intel Corporation 82820 820 (Camino) Chipset Host Bridge (MCH) (rev 03)
00:00.0 Class 0600: 8086:2501 (rev 03)
	Subsystem: 1043:801c
	Flags: bus master, fast devsel, latency 0
	Memory at d0000000 (32-bit, prefetchable) [size=256M]
	Capabilities: <available only to root>

00:01.0 PCI bridge: Intel Corporation 82820 820 (Camino) Chipset PCI to AGP Bridge (rev 03) (prog-if 00 [Normal decode])
00:01.0 Class 0604: 8086:250f (rev 03)
	Flags: bus master, 66Mhz, fast devsel, latency 0
	Bus: primary=00, secondary=01, subordinate=01, sec-latency=64
	Memory behind bridge: cc000000-ccffffff
	Prefetchable memory behind bridge: cdf00000-cfffffff

00:1e.0 PCI bridge: Intel Corporation 82801AA PCI Bridge (rev 02) (prog-if 00 [Normal decode])
00:1e.0 Class 0604: 8086:2418 (rev 02)
	Flags: bus master, fast devsel, latency 0
	Bus: primary=00, secondary=02, subordinate=02, sec-latency=32
	I/O behind bridge: 0000c000-0000dfff
	Prefetchable memory behind bridge: cd000000-cdefffff

01:00.0 VGA compatible controller: Matrox Graphics, Inc. MGA G400 AGP (rev 04) (prog-if 00 [VGA])
01:00.0 Class 0300: 102b:0525 (rev 04)
	Subsystem: 102b:0378
	Flags: bus master, medium devsel, latency 64, IRQ 11
	Memory at ce000000 (32-bit, prefetchable) [size=32M]
	Memory at cc800000 (32-bit, non-prefetchable) [size=16K]
	Memory at cc000000 (32-bit, non-prefetchable) [size=8M]
	Expansion ROM at cdff0000 [disabled] [size=64K]
	Capabilities: <available only to root>

Other entries are all related to non-interesting (I think) devices (IDE, SMBus, audio, etc.)

> It worked, but in which way ? Did you test X+OpenGL apps to see whether 
> it did the trick ? The fact that the module loads itself is not always 
> sufficient to say that it works (believe me :-)
It loaded and glxgears worked correctly, I didn't tested other apps

> With your custom patch, or with the 'generic' stuff ?
Both (sigh!), the problem shows with or without your patch.
Now I'm tring your new patch, but I don't think something will change.

Regards.

-- 
-----------------------------------------------------
Daniele Venzano
Senior member of the Linux User Group Genova (LUGGe)
E-Mail: venza@iol.it
Web: http://digilander.iol.it/webvenza/
LUGGe: http://lugge.ziobudda.net

Re: Problem with i820 AGP patch [SOLVED]

[Daniele Venzano]

On Thu, Nov 15, 2001 at 08:11:46PM +0100, Daniele Venzano wrote:
> Both (sigh!), the problem shows with or without your patch.
> Now I'm tring your new patch, but I don't think something will change.

I was wrong, now (with 2.4.15-pre4 and Nicolas patch) everything is
working fine, the module agpgart loads finding my i820 chipset and my
OpenGL apps work correctly with more than 256Mb of RAM.

Thank you very much for your work.
Regards.

-- 
-----------------------------------------------------
Daniele Venzano
Senior member of the Linux User Group Genova (LUGGe)
E-Mail: venza@iol.it
Web: http://digilander.iol.it/webvenza/
LUGGe: http://lugge.ziobudda.net