SOS A problem with NVS 4200M...

[Imbrius]

Hi

I installed " NVIDIA-Linux-x86_64-295.53.run " for my Thinkpad T420, but this caused a problem.

It needed to add " acpi=off " when I boot the kernel, otherwise it would display nothing on the screen...

I have no idea about this problem, and any suggestion is welcome.

FYI:

operation system: SL_6.2

card: NVS 4200M

$ uname -a
Linux localhost.localdomain 2.6.32-220.el6.x86_64 #1 SMP Sat Dec 10 17:04:11 CST 2011 x86_64 x86_64 x86_64 GNU/Linux

$ lspci | grep "VGA"
01:00.0 VGA compatible controller: nVidia Corporation Device 1057 (rev a1)


Best regards

Thanks.

[tux99]

Did you install bumblebee from the ElRepo 'testing' repository?

I think you need it since you have a Nvidia GPU with Optimus GPU switching tech.

I'm not the best candidate to help you here as I have never used a laptop with Nvidia Optimus myself, but there are other members on this forum who know more about this and might be able to help you.

[toracat]

Yes, indeed. The maintainer of ELRepo's bumblebee (Rob Mokkink) is active on the ELRepo mailing list. There is one session currently going on.

I suggest you post your question if you need assistance.

[Imbrius]

QUOTE (tux99 @ May 27 2012, 05:46 AM)

Did you install bumblebee from the ElRepo 'testing' repository? I think you need it since you have a Nvidia GPU with Optimus GPU switching tech. I'm not the best candidate to help you here as I have never used a laptop with Nvidia Optimus myself, but there are other members on this forum who know more about this and might be able to help you.

Thanks for your reply : )

I have already disabled the Optimus mode, and chosen the discrete graphics only in the BIOS.

If I close the acpi, the discrete graphics will work normally, so I think the key is the connection between the acpi and the driver. Is there anything wrong about my idea?


Best regards.

[tux99]

QUOTE (Imbrius @ May 27 2012, 12:26 AM)

I have already disabled the Optimus mode, and chosen the discrete graphics only in the BIOS. If I close the acpi, the discrete graphics will work normally,  so I think the key is the connection between the acpi and the driver. Is there anything wrong about my idea?

Nothing wrong with your idea, it makes perfect sense.

I have never had acpi problems (lucky me!) with Linux so far, so don't have the specific experience to help you debug this, but maybe the following article contains useful info for you:
http://lwn.net/Articles/237085/

Also, did you check that your BIOS is updated? Sometimes a newer BIOS can fix acpi issues.

Last thing: you might want to ask on the Nvidia Linux forum too, they might be more experienced with this issue there: http://www.nvnews.net/vbulletin/forumdisplay.php?forumid=14

[Imbrius]

QUOTE (tux99 @ May 27 2012, 07:17 AM)

QUOTE (Imbrius @ May 27 2012, 12:26 AM) I have already disabled the Optimus mode, and chosen the discrete graphics only in the BIOS. If I close the acpi, the discrete graphics will work normally,  so I think the key is the connection between the acpi and the driver. Is there anything wrong about my idea? Nothing wrong with your idea, it makes perfect sense. I have never had acpi problems (lucky me!) with Linux so far, so don't have the specific experience to help you debug this, but maybe the following article contains useful info for you: http://lwn.net/Articles/237085/ Also, did you check that your BIOS is updated? Sometimes a newer BIOS can fix acpi issues. Last thing: you might want to ask on the Nvidia Linux forum too, they might be more experienced with this issue there: http://www.nvnews.net/vbulletin/forumdisplay.php?forumid=14

I read the official document (http://us.download.nvidia.com/XFree86/Linux-x86_64/295.53/README/index.html), but still had no clear idea about this problem. I also started a new topic on the Nvidia Linux forum. It seems that no one met the same problem with me.

I will continue to think about this problem in spare time, if anything I found valuable, I will share it here.

Thanks for your help very much

[tux99]

QUOTE (Imbrius @ May 27 2012, 03:02 PM)

I also started a new topic on the Nvidia Linux forum. It seems that no one met the same problem with me.

You need to be patient, it's weekend and in some countries it's a long weekend (Monday is a bank holiday in many countries), therefore I wouldn't expect any useful replies before Monday or even Tuesday.

[Imbrius]

QUOTE (tux99 @ May 27 2012, 09:11 PM)

QUOTE (Imbrius @ May 27 2012, 03:02 PM) I also started a new topic on the Nvidia Linux forum. It seems that no one met the same problem with me. You need to be patient, it's weekend and in some countries it's a long weekend (Monday is a bank holiday in many countries), therefore I wouldn't expect any useful replies before Monday or even Tuesday.

Yes, I totally agree with what you said... But in my country it is 22:20 on sunday, and I had to work tomorrow morning... if I get the idea, I will share it with you immediately.

Thanks very much. : )

[Imbrius]

QUOTE (tux99 @ May 27 2012, 09:11 PM)

QUOTE (Imbrius @ May 27 2012, 03:02 PM) I also started a new topic on the Nvidia Linux forum. It seems that no one met the same problem with me. You need to be patient, it's weekend and in some countries it's a long weekend (Monday is a bank holiday in many countries), therefore I wouldn't expect any useful replies before Monday or even Tuesday.

Hi, a new discovery.

If I boot the kernel with "maxcpus=0" to disable the SMP activation entirely, the whole system could run normally.

Hmmm.....This caused a new problem, there is only one cpu in use. and its temperature is about 64.0°C, not a good news.

I must sleep now otherwise my boss will kill me today .

[Imbrius]

the forum has some problem, multiple post.

[Imbrius]

QUOTE (tux99 @ May 27 2012, 09:11 PM)

QUOTE (Imbrius @ May 27 2012, 03:02 PM) I also started a new topic on the Nvidia Linux forum. It seems that no one met the same problem with me. You need to be patient, it's weekend and in some countries it's a long weekend (Monday is a bank holiday in many countries), therefore I wouldn't expect any useful replies before Monday or even Tuesday.

I realized that I misunderstood your words for my bad English...

[Imbrius]

QUOTE (tux99 @ May 27 2012, 09:11 PM)

QUOTE (Imbrius @ May 27 2012, 03:02 PM) I also started a new topic on the Nvidia Linux forum. It seems that no one met the same problem with me. You need to be patient, it's weekend and in some countries it's a long weekend (Monday is a bank holiday in many countries), therefore I wouldn't expect any useful replies before Monday or even Tuesday.

Hi, tux99:

I have resolved this problem that was not perfectly by the following steps

1 get the edid and save it in a file, just named edid.bin, and move it to /etc/X11

# monitor-get-edid > edid.bin

# mv edid.bin /etc/X11

2 After Nvidia driver has installed, add 3 extra option name in the xorg.conf, like this:

Section "Device"
Identifier "Videocard0"
Driver "nvidia"
BusID "PCI:1:0:0"
Option "ConnectedMonitor" "DFP-0"
Option "CustomEDID" "DFP-0:/etc/X11/edid.bin"
EndSection

3 of course we must blacklist the nouveau.

4 close the optimus, and choose discrete graphics in the BIOS.

5 reboot

By these, I can simetimes start T420 with Nvidia driver working perfectly. But sometimes it also failed during the "starting udev" stage that the keyboard maybe out of action or any other things strange.

I realized that if I chose the discrete graphics and forbidden the Integrated graphics, the udev would failed to loading some driver that connected to the graphics. I have not got a clear idea.

If I can forbid loading the driver duing the "starting udev", I guess it will work good. Hmmm...Need to try more.

[tux99]

QUOTE (Imbrius @ Jun 4 2012, 07:42 PM)

By these, I can simetimes start T420 with Nvidia driver working perfectly. But sometimes it also failed during the  "starting  udev" stage that the keyboard maybe out of action or any other things strange. I realized that if I chose the discrete graphics and forbidden the Integrated graphics, the udev would failed to loading some driver that connected to the graphics. I have not got a clear idea. If I can forbid loading the driver duing the "starting udev", I guess it will work good. Hmmm...Need to try more.

That is weird, the behaviour should be consistent. I have the impression that the BIOS ACPI definitions of this laptop model are seriously broken. Are you sure there is no newer BIOS update available?

If I were you I would try contacting Lenovo customer support about this, you never know maybe they can provide you a fixed BIOS.

Also maybe your laptop has hardware problems, I just had a look at the reviews here and there appear to be quite a few people who claim the T420 is unreliable and died after only a few months: lenovo reviews

[Imbrius]

QUOTE (tux99 @ Jun 6 2012, 04:15 PM)

QUOTE (Imbrius @ Jun 4 2012, 07:42 PM) By these, I can simetimes start T420 with Nvidia driver working perfectly. But sometimes it also failed during the  "starting  udev" stage that the keyboard maybe out of action or any other things strange. I realized that if I chose the discrete graphics and forbidden the Integrated graphics, the udev would failed to loading some driver that connected to the graphics. I have not got a clear idea. If I can forbid loading the driver duing the "starting udev", I guess it will work good. Hmmm...Need to try more. That is weird, the behaviour should be consistent. I have the impression that the BIOS ACPI definitions of this laptop model are seriously broken. Are you sure there is no newer BIOS update available? If I were you I would try contacting Lenovo customer support about this, you never know maybe they can provide you a fixed BIOS. Also maybe your laptop has hardware problems, I just had a look at the reviews here and there appear to be quite a few people who claim the T420 is unreliable and died after only a few months: lenovo reviews

I checked the dmesg in detail, and found someting weird.

if I failed to boot, the message was like this:

dracut: Starting plymouth daemon
usb 1-1: new high speed USB device using ehci_hcd and address 2
dracut: rd_NO_DM: removing DM RAID activation
dracut: rd_NO_MD: removing MD RAID activation

and the boot hung up here.

on the contrary, the message was like this:

dracut: Starting plymouth daemon
dracut: rd_NO_DM: removing DM RAID activation
dracut: rd_NO_MD: removing MD RAID activation
ACPI: WMI: Mapper loaded
usb 1-1: new high speed USB device using ehci_hcd and address 2

This ran successfully

I really did not know why the positon of the red line was different, and it seems that this would infected the boot.

[Imbrius]

This page had the same problem:

https://bugzilla.redhat.com/show_bug.cgi?id=696513

Somebody said it may be a kernel problem. The kernel of Fedora 14 is 2.6.35 that is newer than sl 6.2.

Maybe I need to compile a 3.x kernel.

[Imbrius]

This page is more valuable. I will try to prevent loading the ehci_hcd during initrd in my free time.

http://www.thinkwiki.org/wiki/Problem_with_USB_2.0

[Imbrius]

I am back to the start

Now I am sure that it is an acpi bug. I will compile a newer kernel(2.6.39) following by this page(http://www.thinkwiki.org/wiki/Installing_Gentoo_on_a_ThinkPad_T420)

[tux99]

QUOTE (Imbrius @ Jun 7 2012, 10:02 PM)

I am back to the start  Now I am sure that it is an acpi bug. I will compile a newer kernel(2.6.39) following by this page(http://www.thinkwiki.org/wiki/Installing_Gentoo_on_a_ThinkPad_T420)

Why don't you try 3.4.0 from elrepo?

[Imbrius]

QUOTE (tux99 @ Jun 8 2012, 04:25 AM)

QUOTE (Imbrius @ Jun 7 2012, 10:02 PM) I am back to the start   Now I am sure that it is an acpi bug. I will compile a newer kernel(2.6.39) following by this page(http://www.thinkwiki.org/wiki/Installing_Gentoo_on_a_ThinkPad_T420) Why don't you try 3.4.0 from elrepo?

I tried it just now, and this caused another problem

I got an endless loop of errors when the kernel booted:


DRHD: handling fault status reg 2
DMAR: [DMA Read] Request device [0d:00.0] fault addr fffff000
DMAR: [fault reason 02] Present bit in context entry is clear

$ lspci | grep "0d:00.0"
0d:00.0 System peripheral: Ricoh Co Ltd Device e823 (rev 05)

It is out of my knowledge, and I will google it for more message

[Imbrius]

QUOTE (tux99 @ Jun 8 2012, 04:25 AM)

QUOTE (Imbrius @ Jun 7 2012, 10:02 PM) I am back to the start   Now I am sure that it is an acpi bug. I will compile a newer kernel(2.6.39) following by this page(http://www.thinkwiki.org/wiki/Installing_Gentoo_on_a_ThinkPad_T420) Why don't you try 3.4.0 from elrepo?

Hohohoho ~~~~~~

I got it with your help!

I installed the kernel 3.4.1-1(current) from elrepo, and booted it with "intel_iommu=off".

This works well !

Thanks so much for your help!

[tux99]

QUOTE (Imbrius @ Jun 8 2012, 08:13 PM)

I installed the kernel 3.4.1-1(current) from elrepo, and booted it  with "intel_iommu=off".  This  works well ! Thanks so much for your help!

You are welcome, but you did it almost all by yourself!