[opensuse] Intel_AGP / AGPgart / P965 / G965 / ASUS P5B Deluxe Mix-up

Randall R Schulz

9 Dec 2006 9 Dec '06

23:29

Hi, As some of you may have already noticed, owners of ASUS P5B Deluxe mainboards with Intel P965 (_not_ G965) chipsets and > 3GB of RAM and the BIOS set to enable remapping of the I/O memory space to enable access to the RAM beyond 3GB, the intel_agp and / or agpgart kernel module (or related software) spuriously detects a 965G and immediately aftward, during start-up, does something that causes the system to lock up hard. I'm trying to learn enough about the relevant software and hardware to rectify this situation. The work-around, suggested by Marcus Camen, of disabling memory remapping in the BIOS works, and I can boot and run the 10.2 final kernel, but it comes at the cost of losing 1 GB of my installed RAM. Naturally, I need to rectify this. This is all unfamiliar to me. Where might I learn what I'll need to know to regain the full use of my hardware? Thanks. Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Show replies by date

Joe Morris (NTM)

10 Dec 10 Dec

00:16

Randall R Schulz wrote:

...

I'm trying to learn enough about the relevant software and hardware to rectify this situation. The work-around, suggested by Marcus Camen, of disabling memory remapping in the BIOS works, and I can boot and run the 10.2 final kernel, but it comes at the cost of losing 1 GB of my installed RAM. Naturally, I need to rectify this.

This is all unfamiliar to me. Where might I learn what I'll need to know to regain the full use of my hardware?

i don't know that hardware in particular, but assuming that is an x86_64 arch, some searching in the amd64 list for the past few years will reveal that this is a BIOS problem, and the only way to get use of your memory (has to do IIRC with PCI memory remapping and a hole used for allocation for PCI cards) back is a corrected BIOS. So, make sure you have the latest BIOS update. Otherwise, you might complain to ASUS, but don't hold your breath there. -- Joe Morris Registered Linux user 231871 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Randall R Schulz

00:34

On Saturday 09 December 2006 16:16, Joe Morris (NTM) wrote:

...

Randall R Schulz wrote: ...

i don't know that hardware in particular, but assuming that is an x86_64 arch,

Not. Intel Pentium, Conroe, LGA775, Core 2 Duo. etc. http://www.asus.com/products.aspx?l1=3&l2=11&l3=307&model=1295&modelmenu=1

...

...

-- Joe Morris

RRS -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Bernhard Walle

01:42

* Randall R Schulz [2006-12-10 01:34]:

...

On Saturday 09 December 2006 16:16, Joe Morris (NTM) wrote:

...
Randall R Schulz wrote: ...

i don't know that hardware in particular, but assuming that is an x86_64 arch,

Not. Intel Pentium, Conroe, LGA775, Core 2 Duo. etc.

And which architecture is Core 2 Duo? :) Regards, Bernhard

Randall R Schulz

00:36

On Saturday 09 December 2006 16:16, Joe Morris (NTM) wrote:

...

Randall R Schulz wrote: ... So, make sure you have the latest BIOS update. Otherwise, you might complain to ASUS, but don't hold your breath there.

This is not about a hardware problem or an out-of-date BIOS. It's basically a kernel software problem introduced between 10.2's RC1 kernel, which worked fine, and that used in the GM release.

...

-- Joe Morris

RRS -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Felix Miata

00:51

On 2006/12/10 08:16 (GMT+0800) Joe Morris (NTM) apparently typed:

...

this is a BIOS problem, and the only way to get use of your memory (has to do IIRC with PCI memory remapping and a hole used for allocation for PCI cards) back is a corrected BIOS. So, make sure you have the latest BIOS update. Otherwise, you might complain to ASUS, but don't hold your breath there.

I gave up on Asus years ago: http://www.mozillaquest.com/Linux04/Asus_Sucks_Story-01.html -- "Let your conversation be always full of grace." Colossians 4:6 NIV Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://mrmazda.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Randall R Schulz

01:49

On Saturday 09 December 2006 16:51, Felix Miata wrote:

...

...

I gave up on Asus years ago:

Blah, blah, blah. I have one. The problem did not exist from 10.2 alpha5 onward until the GM release. It's not a hardware problem. RRS -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Bernhard Walle

02:00

* Randall R Schulz [2006-12-10 02:49]:

...

On Saturday 09 December 2006 16:51, Felix Miata wrote:

...
...

I gave up on Asus years ago:

Blah, blah, blah.

I have one. The problem did not exist from 10.2 alpha5 onward until the GM release.

So which was the last Beta / RC that worked? Can you enter a Bugzilla. I don't see an entry for openSUSE for this bug. Did I miss one? Regards, Bernhard

M Harris

06:20

...

I have one. The problem did not exist from 10.2 alpha5 onward until the GM release.

It's not a hardware problem. I agree with *you* Schulz, but as you pointed out earlier, MBs are relatively inexpensive... if it were me, seriously, I'd dump the ASUS board... I mean, why fight the problem. You can install a new board in that thing first thing next week and have it up and running before anyone at bugzilla has poured

On Saturday 09 December 2006 19:49, Randall R Schulz wrote: their first cup of coffee. I kept searching after we spoke earlier about this and I am finding scads of hits that are not making the ASUS P5B look too good... several are regarding the IDE chip, and several are specifically regarding the memory. Sounds like the board really isn't worth the trouble you're having with it. -- Kind regards, M Harris <>< -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

John Pierce

16:00

...

...
It's not a hardware problem. I agree with *you* Schulz, but as you pointed out earlier, MBs are relatively inexpensive... if it were me, seriously, I'd dump the ASUS board... I mean,

Well it does not always seem to get fixed from the mobo makers. I had a biostar with a VIA chipset and an athlon XP 2000. There is a problem in the dma that conflicts with the ivtv driver and would cause spontaneous reboots. Even the latest bios updates never solved it. In the end, the only solution was to get another mobo with a different chipset. I wish they would have fixed the problem, but they never did. Same OS 10.1, ivtv driver compiled from source, pvr 500, soundcard, video card, and network card and different mobo and all work properly. John -- Registered Linux User 263680, get counted at http://counter.li.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Ryouga Hibiki

01:49

Interesting note, a bit peculiar because in openSuSE webpage many mainboards that work well with Linux are Asus: http://en.opensuse.org/HCL/Main_Boards#ATX In Linux-Hardware.org happens the same: http://www.linux-hardware.org/hardware/liste2.php?hersteller=asus&geraet=&kategorie=Mainboard not the same with MSI: http://www.linux-hardware.org/hardware/liste2.php?hersteller=MSI&geraet=&kategorie=Mainboard Nope, I'm not looking for fight I just would like to have a good mainboard that will have zero problems with Linux, any recomendations are welcome. =) Carlos A. PS: Is it like, most Linux users test Asus Mainboards? are those mainboards so common? Here the most common are MSI and PCChips =/ PS2: 29.5% of DVD ISO yah for me ^^; 3 more days =P~

...

From: Felix Miata To: opensuse@opensuse.org Subject: Re: [opensuse] Intel_AGP / AGPgart / P965 / G965 / ASUS P5B Deluxe Mix-up Date: Sat, 09 Dec 2006 19:51:56 -0500

On 2006/12/10 08:16 (GMT+0800) Joe Morris (NTM) apparently typed:

...
this is a BIOS problem, and the only way to get use of your memory (has to do IIRC with PCI memory remapping and a hole used for allocation for PCI cards) back is a corrected BIOS. So, make sure you have the latest BIOS update. Otherwise, you might complain to ASUS, but don't hold your breath there.

I gave up on Asus years ago: http://www.mozillaquest.com/Linux04/Asus_Sucks_Story-01.html -- "Let your conversation be always full of grace." Colossians 4:6 NIV

Team OS/2 ** Reg. Linux User #211409

Felix Miata *** http://mrmazda.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

_________________________________________________________________ FREE pop-up blocking with the new MSN Toolbar - get it now! http://toolbar.msn.click-url.com/go/onm00200415ave/direct/01/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Felix Miata

02:19

On 2006/12/10 01:49 (GMT) Ryouga Hibiki apparently typed:

...

From: Felix Miata Sat, 09 Dec 2006 19:51:56 -0500

...

...
I gave up on Asus years ago: http://www.mozillaquest.com/Linux04/Asus_Sucks_Story-01.html

...

Interesting note, a bit peculiar because in openSuSE webpage many mainboards that work well with Linux are Asus: http://en.opensuse.org/HCL/Main_Boards#ATX In Linux-Hardware.org happens the same: http://www.linux-hardware.org/hardware/liste2.php?hersteller=asus&geraet=&kategorie=Mainboard not the same with MSI: http://www.linux-hardware.org/hardware/liste2.php?hersteller=MSI&geraet=&kategorie=Mainboard

...

Nope, I'm not looking for fight I just would like to have a good mainboard that will have zero problems with Linux, any recomendations are welcome. =)

I've bought new boards made Asus, Biostar, PC Chips, Tyan, AOpen, Intel, FIC, Abit, Soyo, & Giga-Byte according to total feature set, only one portion of which is manufacturer reputation. Counting those acquired used, I've used boards from virtually every manufacturer. Everyone makes some bad ones. Sometimes the only problem is a BIOS defect, which ought to be easy to fix, but sometimes never happens. ATM my 2 24/7 systems are based on Intel 478 & 775 chips on PC Chips and Abit boards with Intel chipsets.

...

PS: Is it like, most Linux users test Asus Mainboards? are those mainboards so common? Here the most common are MSI and PCChips =/ PS2: 29.5% of DVD ISO yah for me ^^; 3 more days =P~

The Asus URL I provided is not about product engineering quality. The problem is about the sufficiency of manufacturer support for those in the unfortunate circumstance of needing it while not being a windoz user. -- "Let your conversation be always full of grace." Colossians 4:6 NIV Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://mrmazda.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Randall R Schulz

04:32

On Saturday 09 December 2006 18:19, Felix Miata wrote:

...

...

The Asus URL I provided is not about product engineering quality. The problem is about the sufficiency of manufacturer support for those in the unfortunate circumstance of needing it while not being a windoz user.

And what sort of support does one really ever need from the manufacturer? I've never had a need to contact a mainboard vendor or get anything from them that could not be gotten from the Internet. I've had SuperMicro, MSI, Intel and ASUS mainboards (and Macs). All I've ever needed post-installation is new BIOS firmware, and that's always available via the Internet. (And I've never needed to contact Apple for Mac support, either, and I've had more of them than I've had Intel PCs running Linux or Windows.) As far as I'm concerned, the vendor support issue is a red herring. And lastly, my question was about the "agpgart" and "intel_agp" modules, which I still need to understand so I can deal with the problem in the latest 10.2 kernel. E.g., if I have a nVidia board (and no mainboard graphics), do I need the agpgart and / or intel_agp kernel modules at all? And if not, how can I suppress them so I can reenable memory mapping and regain use of my system's RAM above 3GB? Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Felix Miata

04:49

On 2006/12/09 20:32 (GMT-0800) Randall R Schulz apparently typed:

...

And what sort of support does one really ever need from the manufacturer? I've never had a need to contact a mainboard vendor or get anything from them that could not be gotten from the Internet. I've had SuperMicro, MSI, Intel and ASUS mainboards (and Macs). All I've ever needed post-installation is new BIOS firmware, and that's always available via the Internet.

For some, good luck never seems to end.

...

As far as I'm concerned, the vendor support issue is a red herring.

If someday you are confronted with a product for which the only recourse is a warranty claim perhaps you will understand the support issue. Could be the difference for you between RC1 and final is telling you something about your hardware rather than about your OS. -- "Let your conversation be always full of grace." Colossians 4:6 NIV Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://mrmazda.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Randall R Schulz

05:32

On Saturday 09 December 2006 20:49, Felix Miata wrote:

...

...

...
As far as I'm concerned, the vendor support issue is a red herring.

If someday you are confronted with a product for which the only recourse is a warranty claim perhaps you will understand the support issue.

Here's one way to look at it: I make well over $100 per hour. Mainboards cost maybe a couple hundred dollars. So it's rarely worth my time to try to get anything from a vendor. (Not to mention the fact that I absolutely hate the telephone.) If I don't choose well in purchasing a mainboard, then at least I can learn something more about how to choose next time.

...

Could be the difference for you between RC1 and final is telling you something about your hardware rather than about your OS.

What is the logic in that? Why are you trying to make this a hardware problem, when the hardware works just fine? There clearly was a regression in the kernel and / or a module. If you don't have any information that is actually an answer to the question I asked, don't feel compelled to reply to my posts. Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Felix Miata

05:53

On 2006/12/09 21:32 (GMT-0800) Randall R Schulz apparently typed:

...

On Saturday 09 December 2006 20:49, Felix Miata wrote:

...

...
Could be the difference for you between RC1 and final is telling you something about your hardware rather than about your OS.

...

What is the logic in that? Why are you trying to make this a hardware

I like to think hardware manufacturers who aren't openly against open source software do their own testing with Linux, to ensure they aren't doing anything to create situations like the one you're in. Support isn't necessarily only what you think it is.

...

problem, when the hardware works just fine? There clearly was a regression in the kernel and / or a module.

I'm not making anything. I read the whole thread. It does clearly seem to be a regression. It's just that things that seem to be clear are sometimes just an illusion of clarity. Maybe it worked because of some previously needed kludge found to be no longer necessary by those in power to so decide, without benefit of facts in your possession only. Maybe the right place to ask, and time to expect a helpful reply, lie after the weekend, and on the factory list or bugzilla. -- "Let your conversation be always full of grace." Colossians 4:6 NIV Team OS/2 ** Reg. Linux User #211409 Felix Miata *** http://mrmazda.no-ip.com/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Randall R Schulz

06:43

On Saturday 09 December 2006 21:53, Felix Miata wrote:

...

On 2006/12/09 21:32 (GMT-0800) Randall R Schulz apparently typed:

...
On Saturday 09 December 2006 20:49, Felix Miata wrote:

...
Could be the difference for you between RC1 and final is telling you something about your hardware rather than about your OS.

What is the logic in that? Why are you trying to make this a hardware

I like to think hardware manufacturers who aren't openly against open source software do their own testing with Linux,

Yes. "Like to think." And thus it shall be so.

...

to ensure they aren't doing anything to create situations like the one you're in.

Still lacking in logic, since I've had the board for a couple of months, but the failure didn't occur until a kernel released just a few days ago. How could the hardware vendor's testing have anything to do with that regression? Here's something that makes just as much sense: Windows users who buy ASUS hardware are supporting the product volumes that make it possible for us to get affordable hardware we can use to run Linux. We're benefitting from Windows users. Anyway, what is it with this vendetta against certain mainboard vendors? Besides, if you don't want to help me, then fine. Leave it be.

...

Support isn't necessarily only what you think it is.

When I say "support," I mean what I mean when I say "support."E Randall "Humpty Dumpty" Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Thomas Hertweck

11:11

Randall R Schulz wrote:

...

[...] There clearly was a regression in the kernel and / or a module.

Did you check the SUSE kernel changelog? That should tell you the differences between the RC1 kernel that worked and the GM kernel that fails. Cheers, Th. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Bernhard Walle

12:47

* Thomas Hertweck [2006-12-10 12:11]:

...

Randall R Schulz wrote:

...
[...] There clearly was a regression in the kernel and / or a module.

Did you check the SUSE kernel changelog? That should tell you the differences between the RC1 kernel that worked and the GM kernel that fails.

Well, we added - patches.drivers/agp-dma32: Allocate AGP pages with GFP_DMA32 by default. - patches.drivers/965-drm: drm: Add support for Intel i965G chipsets.. - patches.drivers/965-agp: Intel 965 Express support.. - patches.drivers/965-agp-suspend: Add suspend callback for i965. Does it work with vanilla 2.6.19? All these patches are from 2.6.19 (I think). Regards, Bernhard

Randall R Schulz

15:24

Bernhard, On Sunday 10 December 2006 04:47, Bernhard Walle wrote:

...

...

Well, we added

- patches.drivers/agp-dma32: Allocate AGP pages with GFP_DMA32 by default. - patches.drivers/965-drm: drm: Add support for Intel i965G chipsets.. - patches.drivers/965-agp: Intel 965 Express support.. - patches.drivers/965-agp-suspend: Add suspend callback for i965.

Perhaps you know if there is a way, short of recompiling the kernel, to suppress the agpgart-related modules? Can you answer these questions: I have a nVidia board (and no mainboard graphics), do I need the agpgart and / or intel_agp kernel modules at all? And if not, how can I suppress them so I can reenable memory mapping and regain use of my system's RAM above 3GB? The facts I have about symptoms are: 1) The start-up logging to the primary console (as in ALT-F1) shows these lines: -==- Starting udevd Linux agpgart interface v0.101 (c) Dave Jones agpgart: Detected an Intel 965G Chipset. -==- At this point, the system hangs hard, requiring a power cycle or a press of the reset button to get out. 2) The BIOS has feature that allows remapping of PCI I/O addresses out of the 3GB - 4GB region so RAM in that region may be accessed. Disabling that option alleviates the symptom (and renders 1 GB of my RAM inaccessible, of course). 3) The kernels supplied with alpha5, beta1, beta2 and RC1 did not have this problem. The GM release's kernel does. The hardware is: - Intel Core 2 Duo E6700 CPU - Intel P965 (_not_ G965) chipset - ICH8R Other hardware information is available here: http://www.asus.com/products.aspx?l1=3&l2=11&l3=307&model=1295&modelmenu=1.

...

Does it work with vanilla 2.6.19? All these patches are from 2.6.19 (I think).

I don't know. I don't know how to find out.

...

Regards, Bernhard

Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Bernhard Walle

15:41

* Randall R Schulz [2006-12-10 16:24]:

...

On Sunday 10 December 2006 04:47, Bernhard Walle wrote:

...
...

Well, we added

- patches.drivers/agp-dma32: Allocate AGP pages with GFP_DMA32 by default. - patches.drivers/965-drm: drm: Add support for Intel i965G chipsets.. - patches.drivers/965-agp: Intel 965 Express support.. - patches.drivers/965-agp-suspend: Add suspend callback for i965.

Perhaps you know if there is a way, short of recompiling the kernel, to suppress the agpgart-related modules? Can you answer these questions:

You mean just prevent the modules from beiing loaded automatically? Yes, of course, just enter blacklist intel_agp in /etc/modprobe.conf.local file. And any other module you want to supress from loading.

...

The facts I have about symptoms are:

Is there a report in Bugzilla? If no, please enter one. Thanks. I don't know the answer now without investigation.

...

...
Does it work with vanilla 2.6.19? All these patches are from 2.6.19 (I think).

I don't know. I don't know how to find out.

Ok, if you don't know how to compile and install a kernel, it's nothing for you. Currently there's no RPM package for 2.6.19. Regards, Bernhard

Randall R Schulz

15:58

Bernhard, Thanks! On Sunday 10 December 2006 07:41, Bernhard Walle wrote:

...

* Randall R Schulz [2006-12-10 16:24]:

...
...

Perhaps you know if there is a way, short of recompiling the kernel, to suppress the agpgart-related modules? Can you answer these questions:

You mean just prevent the modules from beiing loaded automatically? Yes, of course, just enter

blacklist intel_agp

Excellent. That's almost too easy, but it seems to have done the trick. Clearly I don't need the intel_agp, and that seems like the most likely culprit for the symptom I'm seeing. Does the agpgart module or any of the others you mentioned provide any advantages for a system with an nVidia card (GeForce 7300)? Does it interact with or mutually exclude the nVidia proprietary driver (which I'm not yet running, but did install during pre-release testind and will probably install again at some point)?

...

in /etc/modprobe.conf.local file. And any other module you want to supress from loading.

...
The facts I have about symptoms are:

Is there a report in Bugzilla? If no, please enter one. Thanks. I don't know the answer now without investigation.

I'll file one.

...

...
...
Does it work with vanilla 2.6.19? All these patches are from 2.6.19 (I think).

I don't know. I don't know how to find out.

Ok, if you don't know how to compile and install a kernel, it's nothing for you. Currently there's no RPM package for 2.6.19.

I've done it, but my understanding is that SuSE kernels are significantly patched w.r.t. the stock Linux kernels.

...

Regards, Bernhard

Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Bernhard Walle

20:00

* Randall R Schulz [2006-12-10 16:58]:

...

...
...
...
Does it work with vanilla 2.6.19? All these patches are from 2.6.19 (I think).

I don't know. I don't know how to find out.

Ok, if you don't know how to compile and install a kernel, it's nothing for you. Currently there's no RPM package for 2.6.19.

I've done it, but my understanding is that SuSE kernels are significantly patched w.r.t. the stock Linux kernels.

Not as heavily any more. There's Xen and there's KDB and there are a few other fixes/enhancements. I'm running a plain 2.6.19 here because of testing purposes with openSUSE 10.2 without problems. Regards, Bernhard

Philipp Thomas

12 Dec 12 Dec

22:20

On Sun, 10 Dec 2006 07:58:30 -0800, Randall R Schulz wrote:

...

Excellent. That's almost too easy, but it seems to have done the trick.

You should still report it in bugzilla.novell.com to alert our kernel developers to the problem.

...

Clearly I don't need the intel_agp, and that seems like the most likely culprit for the symptom I'm seeing.

What makes you think so?

...

Does the agpgart module or any of the others you mentioned provide any advantages for a system with an nVidia card (GeForce 7300)?

See http://en.wikipedia.org/wiki/Graphics_Address_Remapping_Table for what a GART is and why you possibly need it.

...

interact with or mutually exclude the nVidia proprietary driver (which I'm not yet running, but did install during pre-release testind and will probably install again at some point)?

I do not know whther the proprietary nvidia driver needs GART support from the kernel. Philipp -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Randall R Schulz

22:47

On Tuesday 12 December 2006 14:20, Philipp Thomas wrote:

...

On Sun, 10 Dec 2006 07:58:30 -0800, Randall R Schulz wrote:

...
Excellent. That's almost too easy, but it seems to have done the trick.

You should still report it in bugzilla.novell.com to alert our kernel developers to the problem.

...
Clearly I don't need the intel_agp, and that seems like the most likely culprit for the symptom I'm seeing.

What makes you think so?

Because, as I've mentioned several times, my system incorporates a P965 chipset, not a G965. There is no mainboard graphics, just an nVidia card. The intel_agp module spuriously detects a "965G" (Intel's specifications and product literature all put the 'P' or 'G' before the "965").

...

...

Philipp

Randall Schuzl -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Bernhard Walle

15 Dec 15 Dec

21:25

Hello, * Randall R Schulz [2006-12-12 23:47]:

...

On Tuesday 12 December 2006 14:20, Philipp Thomas wrote:

...
On Sun, 10 Dec 2006 07:58:30 -0800, Randall R Schulz wrote:

...
Excellent. That's almost too easy, but it seems to have done the trick.

You should still report it in bugzilla.novell.com to alert our kernel developers to the problem.

...
Clearly I don't need the intel_agp, and that seems like the most likely culprit for the symptom I'm seeing.

What makes you think so?

Because, as I've mentioned several times, my system incorporates a P965 chipset, not a G965. There is no mainboard graphics, just an nVidia card. The intel_agp module spuriously detects a "965G" (Intel's specifications and product literature all put the 'P' or 'G' before the "965").

File out a bugreport, please. Include me (bwalle@novell.com) as CC, but don't assign it to me. Thanks. Regards, Bernhard

Randall R Schulz

18 Dec 18 Dec

16:18

Bernhard, On Friday 15 December 2006 13:25, Bernhard Walle wrote:

...

...

File out a bugreport, please. Include me (bwalle@novell.com) as CC, but don't assign it to me. Thanks.

Done: - Bug #229365 https://bugzilla.novell.com/show_bug.cgi?id=229365 -==--==--==--==--==--==--==--==--==--==--==--==--==--==--==--==- During start-up processing the following lines are printed to the console: -==- Starting udevd Linux agpgart interface v0.101 (c) Dave Jones agpgart: Detected an Intel 965G Chipset. -==- Immediately following this, the system completely hung / locked up and the only way out is the reset button or power cycling. Hardware is ASUS P5B with Core 2 Duo chip, Intel P965 / ICH8R chipset with 4 GB of RAM, an nVidia GeForce 7300 videao and an LSI Logic SCSI interface installed. There is no Intel graphics. Mainboard's vendor information is at http://www.asus.com/products.aspx?l1=3&l2=11&l3=307&model=1295&modelmenu=1. Work-arounds include: 1) Disabling the BIOS setting that allows software to remap I/O addresses so the upper-most 1 GB of RAM may be addressed -or- 2) Adding "blacklist intel_agp" to "/etc/modprobe.conf.local" I ran all publicly announced releases from 10.2 alpha5 through RC1 without experiencing this symptom. It appeared only upon installing and booting the final release. The problem does not occur for the system / kernel used by the installer itself. -==--==--==--==--==--==--==--==--==--==--==--==--==--==--==--==-

...

Regards, Bernhard

Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

6343

Age (days ago)

6352

Last active (days ago)

List overview

Download

26 comments

9 participants

participants (9)

Bernhard Walle
Felix Miata
Joe Morris (NTM)
John Pierce
M Harris
Philipp Thomas
Randall R Schulz
Ryouga Hibiki
Thomas Hertweck