LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Slackware (https://www.linuxquestions.org/questions/slackware-14/)
-   -   Nvidia 304.xx (https://www.linuxquestions.org/questions/slackware-14/nvidia-304-xx-4175458722/)

Woodsman 04-18-2013 11:23 PM

Nvidia 304.xx
 
To anybody using the Nvidia 304.xx series:

I was using 304.64 but updated to 304.88. I don't experience the black screen of death that 304.84 is famous for. :)

brianL 04-19-2013 05:03 AM

No problems with the 310.40 here.

m23 04-19-2013 06:34 AM

Quote:

Originally Posted by Woodsman (Post 4934522)
To anybody using the Nvidia 304.xx series

Quote:

Originally Posted by brianL (Post 4934636)
No problems with the 310.40 here.

:)

304.88 works fine here.

cynwulf 04-19-2013 10:00 AM

304.88 fixes the blank virtual terminals/console problem in 304.84.

Woodsman 04-19-2013 01:53 PM

Quote:

No problems with the 310.40 here.
People using Nvidia chipsets older than series 7 (6150, etc.) cannot use the 310 series. :)

TobiSGD 04-19-2013 04:07 PM

Quote:

Originally Posted by Woodsman (Post 4934866)
People using Nvidia chipsets older than series 8 (6150, etc.) cannot use the 310 series. :)

Fixed that for you.

astrogeek 04-20-2013 04:24 AM

I now have 304.88 running on a 6150 chipset with the 3.8.8 kernel - at last!

I had to patch the kernel/nv-mmap.c file to not fail on VM_RESERVED, but it then built and installed and no longer failed on the ABI version. This is on a fully -current system and Pat's reverting to the earlier Xorg seems to have fixed the ABI problem (if I understand that correctly).

[EDIT]
Forgot to mention - on this system I had to turn off the vblank option. With it I got 60FPS, without ~2400.
[/EDIT]

Loslobo 04-20-2013 09:55 PM

Question for Woodman. Did you use the older SlackBuild version for the 304 series? Or rolled your own?

There's an old 6150 ADP video card gathering dust within an nvidia AMD2 chipset based PC. I'd like to put SlackWare 14 on. And the nvidia I have on file is the 304.51 using the older slackbuild. Would it work just updating the version number in that older slackbuild?

Quote:

Originally Posted by Woodsman (Post 4934522)
To anybody using the Nvidia 304.xx series:

I was using 304.64 but updated to 304.88. I don't experience the black screen of death that 304.84 is famous for. :)


Woodsman 04-20-2013 10:08 PM

I used the build scripts for 13.37 (302.17) from slackbuilds.org. I changed the version and build numbers.

Note: The driver script has an incorrect sym link command:

AS IS (line 154):

ln -sf libnvcuvidv.so.$VERSION libnvcuvid.so

CHANGE TO:

ln -sf libnvcuvid.so.$VERSION libnvcuvid.so

TobiSGD 04-21-2013 07:14 AM

Quote:

Originally Posted by astrogeek (Post 4935185)
[EDIT]
Forgot to mention - on this system I had to turn off the vblank option. With it I got 60FPS, without ~2400.
[/EDIT]

You will get 60 FPS (or whatever refresh rate you use) with any driver on any system with the vblank option on. That is exactly what this option is for, it syncs the FPS rate with the refresh rate to avoid tearing. If I were you I would turn that option on again, I really hate tearing.

Drakeo 04-21-2013 07:28 AM

Quote:

Originally Posted by Woodsman (Post 4934866)
People using Nvidia chipsets older than series 7 (6150, etc.) cannot use the 310 series. :)

Thank you Woodsman that is a big deal to me and grateful. my little private custom distro of puppy linux thanks you from the bottom of my heart.

digger95 04-22-2013 03:10 PM

Question: I have the Nvidia GeForce 6150SE nForce 430 chipset and everything seems to be working fine out of the box with the generic 3.2.29 kernel and stock nouveau drivers on a fresh install of Slackware 14. At least nothing is crashing and I'm able to watch movies and such without problems. Is there any advantage to installing the proprietary driver that I may not be aware of? Thanks in advance for any input.

Philip Lacroix 04-22-2013 05:51 PM

I'm using version 304.64 on Slackware 14 and it works fine (7600 GS).

Version 310.32 (from the slackbuild) didn't work at all, and log messages pointed me to 304.64, saying the card was supported up to that version. I had to fix several links and libraries, and I guess it would have been easier if I had modified the slackbuild for 13.37 instead.

Quote:

Originally Posted by digger95 (Post 4936737)
Is there any advantage to installing the proprietary driver that I may not be aware of?

If you really need a good 3D support, the proprietary driver seem to be (so far) a better choice than nouveau, otherwise the latter is a better choice because it's free. :)

http://nouveau.freedesktop.org/wiki/
http://nouveau.freedesktop.org/wiki/HardwareStatus

Kind regards,

Philip

digger95 04-22-2013 07:11 PM

Thanks Philip, I suppose I was just looking for a good reason to load the Nvidia driver, but if everything is working well as is, I guess I will skip it.

Philip Lacroix 04-22-2013 08:27 PM

Quote:

Originally Posted by digger95 (Post 4936849)
Thanks Philip, I suppose I was just looking for a good reason to load the Nvidia driver, but if everything is working well as is, I guess I will skip it.

You're welcome. As far as I know, the fact that nouveau is part of the official Slackware release is a very strong reason for using it, if it does the job: I guess this is the safest, crash-proof solution.

Philip

digger95 05-27-2013 12:14 PM

I went ahead and installed the Nvidia 304.88 driver for my 6150SE and everything worked perfectly thanks to SlackBuilds (and a special shout out to Edward Koenig). I didn't even have to add/edit any config files afterwards since everything just worked. As Woodsman outlined above I used the kernel and driver build scripts from 13.37 and simply changed the version number. The only minor hassle was locating the 304.88 source files for nvidia-installer, nvidia-settings and nvidia-xconfig but some noodling around on the Nvidia website found all of those.

Woodsman 05-27-2013 01:43 PM

Quote:

The only minor hassle was locating the 304.88 source files for nvidia-installer, nvidia-settings and nvidia-xconfig but some noodling around on the Nvidia website found all of those.
ftp://download.nvidia.com/XFree86/

:)

digger95 05-27-2013 01:47 PM

Quote:

Originally Posted by Woodsman (Post 4960024)

Nice... and thanks! That will be very helpful for other noobs like myself.

digger95 05-29-2013 11:42 AM

Can I assume that this is normal slackpkg behavior after blacklisting nouveau and that I can just ignore it?

Quote:

bash-4.2# slackpkg upgrade-all

Checking local integrity... DONE
You have a broken /var/log/packages - with two versions of the same package.
The list of packages duplicated in your machine are shown below, but don't
worry about this list - when you select your action, slackpkg will show a
better list:

xf86-video-nouveau-blacklist-noarch-1
xf86-video-nouveau-git_20120323_efd7a52-x86_64-1

You can (B)lacklist, (R)emove, or (I)gnore these packages.
Select your action (B/R/I): I


Okay - slackpkg won't do anything now, but please, do something to fix it.

brianL 05-29-2013 01:31 PM

Put this in /etc/slackpkg/blacklist to solve that problem:
Code:

# Nouveau driver
xf86-video-nouveau


Bertman123 06-22-2013 04:00 PM

just installed nvidia 304.88 from nvidia's webiste now web browser crashes.
 
I have the 6150se graphics card on my tower and just installed the 304.88 graphics driver from nvidia's website. It seemed to go well, however now I can not open a web browser. As soon as I open firefox or seamonkey they immediately crash. Any ideas as to what is causing this?

Update: I upgraded to current with the 3.9.5 kernel and when I rebooted and tried to open firefox my system froze. I rebooted again and then firefox seemed to work. I wonder if it's going to freeze up occasionally?

ReaperX7 06-22-2013 07:18 PM

Woodsman you should probably submit a SlackBuild package to SlackBuilds.org to add another Legacy tier for Nvidia chipsets. A lot of laptops and workstations still use the 6x00/7x00 series graphics processors.

astrogeek 08-06-2013 09:14 PM

I am probably a latecomer to 3.10.5, but just updated to -current on an AMD/nvidia box.

Hit the nvidia kernel update problems with 304.88, quickly solved by Willy Sudiarto Raharjo's notes here.

So 304.88 survives yet another round...

Woodsman 08-06-2013 09:55 PM

Nice to hear. After today's current change log announcement, and this thread, I wondered how well 304.88 would survive. I haven't done any testing in current in a long while but expect to do so soon. I have no desire to buy a new motherboard. Nonetheless, I always hold my breath a wee bit when current gets closer to release because I don't know what third party packages will break. :)

willysr 08-06-2013 11:09 PM

I tends to install the latest kernel and find an appropriate patch for NVidia and VMWare and post it on my blog just in case someone else might need it...
GIad it help when Pat decided to take 3.10.5 into Current

Nh3xus 08-11-2013 10:40 AM

Thanks for your work willysr.

Your advice did the trick for me regarding the 304.88 drivers and the 3.10.5 kernel provided by Pat. :)

I just have one question :

You said on your blog to apply the patch with the command "apply".

On my -Current box, when logged as root, the system claimed that the command "apply" was not found.

So I did use the command "patch" with the same arguments as yours.

Is the command "patch" just a symlink to the apply one ? Why couldn't I execute the command "apply" even if the man file was readable ?

Woodsman 08-11-2013 11:55 AM

Long term, seems one of two events need to occur?

1. The nvidia folks update 304.88 to 304.89.

2. When Slackware 14.1 is released officially, the slackbuilds.org 304.88 maintainer incorporates the ArchLinux patch into the build script.

In the short term, adventurous 3.10.x users need to slip the ArchLinux patch into the slackbuilds.org build script?

Does this make sense?

willysr 08-11-2013 05:27 PM

Quote:

Originally Posted by Nh3xus (Post 5007285)
Thanks for your work willysr.

Your advice did the trick for me regarding the 304.88 drivers and the 3.10.5 kernel provided by Pat. :)

I just have one question :

You said on your blog to apply the patch with the command "apply".

On my -Current box, when logged as root, the system claimed that the command "apply" was not found.

So I did use the command "patch" with the same arguments as yours.

Is the command "patch" just a symlink to the apply one ? Why couldn't I execute the command "apply" even if the man file was readable ?

Sorry,my mistake
It should be patch
Fixed now :)

willysr 08-11-2013 05:30 PM

Quote:

Originally Posted by Woodsman (Post 5007311)
Long term, seems one of two events need to occur?

1. The nvidia folks update 304.88 to 304.89.

2. When Slackware 14.1 is released officially, the slackbuilds.org 304.88 maintainer incorporates the ArchLinux patch into the build script.

In the short term, adventurous 3.10.x users need to slip the ArchLinux patch into the slackbuilds.org build script?

Does this make sense?

Since 304.88 is considered legacy, they will only receive security fixes and for new kernel support, i don't really hoped too much for this, but still it's possible if they wanted.

astrogeek 08-11-2013 06:40 PM

Quote:

Originally Posted by willysr (Post 5007432)
Sorry,my mistake
It should be patch
Fixed now :)

HA! I noted that, and just used patch instead, but wondered if it was a typo or just another thing missing in my brain pan...

While here let me say THANKS for all your contributions - they are appreciated!

Woodsman 08-11-2013 06:53 PM

Quote:

Since 304.88 is considered legacy, they will only receive security fixes and for new kernel support, i don't really hoped too much for this, but still it's possible if they wanted.
I have little hope too. That means the slackbuilds.org script should be updated to add the patch when the site is updated for 14.1?

willysr 08-12-2013 06:49 AM

That depends on the maintainer if he has the hardware to install slackware 14.1

willysr 08-12-2013 06:01 PM

A new release of 304.108, but no new kernel support added

Quote:

Implemented workarounds for two Adobe Flash bugs by applying libvdpau commit ca9e637c61e80145f0625a590c91429db67d0a40 to the version of libvdpau shipped with the NVIDIA driver.
Fixed a bug in nvidia-settings that could cause the wrong resolution to be set in basic mode for setups based on one display per X screen.
Added /usr/lib/modprobe.d to the list of directories where nvidia-installer may optionally install a generated modprobe configuration file to attempt to disable Nouveau.

TobiSGD 08-12-2013 07:47 PM

Nvidia produces their Linux drivers in the first place for their workstation users, which mostly run RHEL on their machines. Seeing that RHEL 7 will be based on Fedora 19, which shipped with a 3.9.5 kernel, it is not very likely that there will be an update for newer kernels in the near future for the legacy drivers, unless Red Hat decides to ship a newer kernel.

m23 08-13-2013 02:01 PM

Quote:

Originally Posted by willysr (Post 5008119)
A new release of 304.108, but no new kernel support added

It does have kernel 3.10 support actually. See: https://devtalk.nvidia.com/default/t...iver-releases/.
I'm using it now with kernel 3.10.6.

m23

Olek 09-02-2013 12:32 PM

How You made it?
 
The patch not work.

I have Nvidia source 304.88 32-bit driver.
After patching it i got:
ERROR: Unable to build the NVIDIA kernel module.

I have current kernel 3.10.9-smp 32-bit

I tried patching several times.
From http://bitc.bme.emory.edu/~lzhou/blogs/?p=379 i downloaded patched files to.

Always error.
I hate Nvidia.

willysr 09-02-2013 01:54 PM

did you install the kernel-source package as well?
i have mine working nicely with 3.10.9

cynwulf 09-02-2013 03:01 PM

The patch does work, I'm running the blob at the moment on a 3.10 kernel

Code:

$ uname -r
3.10.9-smp

Code:

# lspci -k | grep -A2 VGA
01:00.0 VGA compatible controller: nVidia Corporation G73 [GeForce 7300 GT] (rev a2)
        Subsystem: eVga.com. Corp. Device a501
        Kernel driver in use: nvidia

Anyway, did you try the new release?
http://us.download.nvidia.com/XFree8...86-304.108.run

//edit: Now running 304.108 - kernel module builds ok on 3.10

Olek 09-03-2013 04:17 AM

Yes :) 304.108 works fine.

Thanks!

It's funny so Nvidia.com still recommends 304.88 for GeForce 7 Series on Linux 32-bit.

willysr 09-03-2013 06:48 AM

I have NVidia 304.108 working with Linux Kernel 3.11 (only need one single patch). It's available on my SlackHacks if anyone is interested on building Linux Kernel 3.11

GazL 09-03-2013 10:31 AM

The above patch also works for nvidia-325.15 on 3.11.

I was planning to stay with 3.10.y this time, but O_TMPFILE looks really useful.

Woodsman 09-05-2013 12:12 PM

Does the 304.108 release still need a patch for the 3.10.x kernel (which will be the kernel version released with 14.1)?

TobiSGD 09-05-2013 12:53 PM

Quote:

Originally Posted by Woodsman (Post 5022443)
Does the 304.108 release still need a patch for the 3.10.x kernel (which will be the kernel version released with 14.1)?

See post #35.

dwightpaige79 10-16-2013 05:14 PM

Quote:

Originally Posted by willysr (Post 5020831)
I have NVidia 304.108 working with Linux Kernel 3.11 (only need one single patch). It's available if anyone is interested on building Linux Kernel 3.11


I would love to make this work in OpenMandriva 2013 Beta but I keep getting one of 2 results, This if I click on the patch and select to "Save As':

Code:

# chmod a+x NVIDIA-Linux-x86_64-304.108.run

# exit
exit

$ ./NVIDIA-Linux-x86_64-304.108.run --extract-only
Creating directory NVIDIA-Linux-x86_64-304.108                                                                                                                             
Verifying archive integrity... OK
Uncompressing NVIDIA Accelerated Graphics Driver for Linux-x86_64 304.108................................................................................................................................................................

$ cd NVIDIA-Linux-x86_64-304.108/

NVIDIA-Linux-x86_64-304.108]$ patch -p1 < /home/dwight/Downloads/nvidia-3.11.patch
patch: **** Only garbage was found in the patch input.

And this if I copy and paste the patch to a file:

Code:

NVIDIA-Linux-x86_64-304.108]$ patch -p1 < /home/dwight/Downloads/nvidia-3.11.patch
patching file kernel/nv-linux.h
patch unexpectedly ends in middle of line
Hunk #1 FAILED at 957.
1 out of 1 hunk FAILED -- saving rejects to file kernel/nv-linux.h.rej

What am I doing wrong? Am I downloading the patch incorrectly?

Edit: Or is this patch only for x86 and I'm using x86_64?

willysr 10-16-2013 06:40 PM

Use the raw version on my github repository

ReaperX7 10-16-2013 08:19 PM

You also should be logged in as root or be running in superuser mode.

willysr 10-16-2013 09:12 PM

Try this one: nvidia-3.11.patch

dwightpaige79 10-17-2013 11:27 AM

Quote:

Originally Posted by willysr (Post 5047159)
Use the raw version on my github repository

Thanks that worked.

kingbeowulf 10-17-2013 11:39 PM

For those of you who still use older Nvidia cards and need the Nvidia driver if the nouveau driver is still a bit "weak," you should ping the Nvidia SBo script maintainer. He still maintains legacy173.xx, and legacy304.xx in addition to the latest long lived branch. Since he no longer has functional cards with older GPUs, it difficult to fully test those for script defects. Fighting to get buggy 13.37 or earlier scripts to work is a bit silly.

Woodsman 10-18-2013 12:48 PM

I still use the 304.xx sources. I have not yet tested in 14.1 but I will be doing that when I find time to start testing 14.1 (soon I hope!). I can't maintain the SBo build package but I will be among those who will continue testing and using simply out of need. Thus I will post comments and questions to the forum as needed.

Currently I'm using 304.108 on two 14.0 systems. No problems encountered but the 14.0 kernel does not require special attention or patching of the 304.xx sources. :)


All times are GMT -5. The time now is 12:51 AM.