Gforce4 Ti 4600 and malfunctioning Frambuffer console

[phlashback]

I have been running this graphics card since around july 2002, and have been quite happy with it. I have run many different kernel version all with out problems. Untill last night.......

My specs:
Motherboard: Tyan S2462
Proc: 2x AMD Athlon MP 1900+
Memory: 2x256MB Registerd DDR, 1x512MB Registerd DDR all PC2100
Linux Distro: Gentoo Linux, based on the 1.4 profile (gcc 3.2.2)
Linux kernel(s): 2.4.20-gentoo patches, 2.4.20 ck4 patches, 2.5.66, 2.5.66-mm3

I baught the 512 MB DDR on saturday morning , and installed it in the system along side the 2x256MB DDR, and booted my pc directly into memtest86 so that I could verify working memory befor trusting data to it :angel:

I stopped testing monday morning (2:00am central) with out any problems. rebooted the computer into the 2.5.66-mm3 kernel (I have had no problems with it) and instead of getting my 1280x1024 console, all I got was a blank screen, I logged in blindly once my disk access stopped, and try to load X... I worked. So I rebooted in to 2.5.66, same thing, as well as the oher kernels.

Yes I have highmem enabled in my kernel as I was planning on purchasing memory for a couple months (ok... so I am lazy).

I removed the 2x256DDR and the console loads as it should w/ the new memory, I put 1 of the 256MB, (then the other) modules back, and it boots normaly. w/768 MB ram. After rarranging the memory again I booted w/ all modules installed no console.

I borrowd a freinds ATI Radeon 7500, and it works, but not my gforce4...

Kernel command line: root=/dev/hda3 vga=794 video=vesa:1280x1024@70

if I remove vga to end of command line I get the standard console.

Here is the error That dmesg has logged for runnig w/full comand line:
vesafb: abort, cannot ioremap video memory 0x8000000 @ 0xf0000000

Tonight I am going to borrow some ram from work, and try it, and borrow a friends Gforce4 ti 4200, as well as his Gforce2 (from a dell) and see if it reporduces.

any insight would be much aprciated.

Thanks
phlashback:afro2:

[bwkaz]

As I said in the other thread that you posted in with this same problem:

Originally posted by bwkaz
That really sounds like a highmem issue...

Which highmem do you have enabled (4GB or 64GB)? Maybe try the other one?

Alternatively, there's something in the README about ioremap -- did you use IGNORE_CC_MISMATCH when you compiled the drivers? That can cause ioremap() to fail. But I see now that you're using a 2.5 kernel. Why? They are experimental, and have issues almost by definition.

Also, does it work better if you use the latest 2.4 kernel instead?

[phlashback]

Thanks for your reply.... I posted a new thread because as I started reading through the forums I came across a post that said you should make a new thread with your quest tion so that is what I did....:angel2:

Any way I currently have 4GB enabled... I will try the other things you have mentiond, but I am not currently infront of the problemd child:rolleyes:

FYI: I keep up w/ the kernel releases, and put them through tests, and for me at least the 2.5 series since 2.5.64 has been rock solid on my box w/ nvidia drivers. To be honist, when I came about this problem I reverted back to the 2.4.20 kernel for debugging;) as it is better uderstood by the masses ie not beta

also I will try compiling the drivers w/ IGNORE_CC_MISMATCH, but i dont see how this could be the problem as vesafb fails which is in kernel and befor nvidia drivers are ever touched (the nvidia drivers work, and load x i just have to do it blindly)...

I have completed testing multiple memory configs (swiped memory from other machines at work) and placed my card into another board that had 1GB of memory (using my hd to boot) and had the same thing.

I am just for the hell of it going to test on 2 other nvidia cards just to see, and rule out the Graphics card being the problem hopefully..... if not... The more I look at it the more i belive that it is a kernel issue so trying the 64GB higmem will be done.... also I guess I better take a harder look at the readme, Im dyslexic, and sometimes I miss a whole lot due to the symbols on the page removing themselves for the paper and implanting on other objects..... Yes I am sober when this happens belive it or not.

Anyways thanx a million for replying, and I will keep you posted.

phlashback :afro2:

[bwkaz]

Originally posted by phlashback
also I will try compiling the drivers w/ IGNORE_CC_MISMATCH I meant to not use IGNORE_CC_MISMATCH. Using it can cause problems with ioremap().

but i dont see how this could be the problem as vesafb fails which is in kernel and befor nvidia drivers are ever touched (the nvidia drivers work, and load x i just have to do it blindly)... Ah, you're right, it doesn't even seem to be an nVidia driver issue, more of a VESA issue. Are there any similar bugreports on lkml about it? Since the driver in question is, after all, part of the Linux kernel? ;)

Do keep us posted.

[phlashback]

Going to start looking through the lkml after breakfast :)

I tried using the 64GB highmem... to no avail, and after I read the readme from nvidia... I figured that the IGNORE_CC_MISMATCH was the problem due to the fact that I compiled it all myself:D

After testing a couple of cards this is what I have found:

ATI Radeon 7500 -- Works as expected
Gforce2 GTS 32MB DDR (Dell) -- Works as expected
Gforce4 Ti4200 (VisionTek) -- No Framebuffer
Gforce4 Ti4600 (VisonTek) -- No Framebuffer

I can confirm that w/ only 768 all cards work (sorry dont have the ram to test between 1024 and 768 :mad:
as I posted earlier this is only w 1024MB RAM (also now confirmed 1280MB and 1536MB)

There is at least one other person in Gentoo land with the same problem (apparently he figured something out with an older kernel) I am waiting for a response from them.

BTW I am positive now that it is the vesafb, and not the video card (given that X works, and I have found one other person)

Any info I get I will be sure to pass on.

Thanks again
phlashback:afro2:

<<EDIT>>
I Just found something, and am dying to try it.. but I was just called into work..

lkml--Some one with the same problem ;) (http://lkml.org/archive/2003/3/27/119/index.html)

just throghing it out, and will test the hacks later

[bwkaz]

And for people having the same problem, and not wanting to wade through the fairly horribly-designed mail archive (sorry, but I really don't like it ;)), the "solution" message (or at least, what worked for that guy) is here:

http://lkml.org/archive/2003/3/27/258/index.html

[phlashback]

Argh... not enough coffee. I copied the link out of the wrong browser window (to many for my own good that close to waking up)..

Thanks for putting right one in, and also for reminding me about the lkml (and yess I agree that there layout is something not to desirable :rolleyes:)

Now for the long awaited update (drum roll please)....................

I have my vesa console back:D, and fonts that are worth looking at. So far only tried the 2.4.20-gentoo (heavaly patched) and 2.4.20-gaming (ck4 based), downloading new development kernel now (just when I wasnt looking... updates galore).

Thanks for your time, and happy hacking
phlashback:afro2:

Just to prove it ;)
screenshot (http://www.copymat.net/phlashback/screenshot.png)