Before I put this live as default, I'd like to see it verified as working.

This feature tests for and then will, if the tests pass, build a new kernel module instead of reinstalling the driver.

This test can be run in X, and it's quite easy to do.

First, make sure you have the latest driver installed. You might want to update sgfxi just to make sure it's got the required options:



Then after it's confirmed as latest, in X, do this:


This removes the current nvidia kernel module.

Don't worry, it's already loaded, so X will keep working.

Now let's test the autocheck/module build method:


<update: removed -! 8 from code, now is default>

-X lets you run sgfxi in X without complaining.

If all works as it does in my tests, you should get some output that looks about like this:



and that's it, your module is rebuilt to the new kernel.

If these tests go fine, I'm going to make this the default behavior for smxi starts sgfxi sessions, and default for sgfxi nvidia sessions as well.

These use the same logic as sgfxi -k (the kernel module build command that exits if it fails), but do a bit more in terms of letting you restart X if you are out of X and so on. Also, if the initial tests fail, sgfxi simply will reinstall your driver.

Make sure of course that you have the driver up to date in these tests, or the test will fail, and sgfxi will try to reinstall your driver in X, which doesn't work.

I may add in an X test there, actually I will, that's an obvious mistake you don't want people to get caught in.
Back to top

Just testet it with an 8600Card and the compililing went trough without problems. Will confirm after reboot

regards

Reiner
Back to top

Rebooted and all is ok. This was done on an sidux-system runnin the actual liquorix-kernel (2.6.33-0.dmz.1-liquorix-amd64 x86_64 (64 bit))

regards

Reiner
Back to top

Worked just fine on antiX-sid running 2.6.33-0.dmz.1-liquorix-686 kernel.
Back to top

Debian Lenny, nvidia fx5200 here

Good work, techAdmin! Nvidia is a pain without sgfxi.
Back to top

Ok, I'm going to remove the testing -! 8 flag now.

I would appreciate more tests using the same method as above, and pasting your output just to make sure it's working.

What this will mean is that if say you install a new kernel, you can simply build a new module with your existing nvidia driver.

sgfxi will run this test by default always for new kernels and all installs, if all that is required is a module build, not driver reinstall, it will run as you see here with the -! 8 option.

You can override the module build step and force a driver reinstall with the new -f option, which when used with nvidia will always install / reinstall the driver, which also removes all the kernel modules that were in your system.

To reinstall all kernel modules each time, do: sgfxi -! 40 [-o <driver version> - optional, only if you want to use a different default than current default offered by sgfxi]

You can also do only a module rebuild with -k (that ignores the -f option), if the module rebuild fails, it will exit with a more verbose error message, telling you why it happened more explicitly.

Since this is a very big change, I'd appreciate any testing people are willing to do.

Here's the kinds of tests you can run:

install a new kernel, in X. Then, assuming you have the latest nvidia driver installed (sgfxi will check all these things), run this command, say your new kernel is: 2.6.32-1-686

that will force only a module build
or:

will run the module check/build step and build module if it passes the test. If it fails the test, and if you are in X, it will exit with an error message, if you are not in X, it will install the driver and uninstall all current modules.

If you are in X, the -X option gets rid of the X check and lets you run the process. -k overrides the x check automatically.

A serious bug in these cases is if the module build tests pass with no error but the module build fails, then we'll need to collect data and see what went wrong with the process.
Back to top

I have also a squeeze install. It has lxde also kde3 installed to /opt (as described in a post elswhere on smxi forum) and GDM.

Kernel 2.6.32-0.dmz.1-liquorix-686, headers and module-assistant installed. Nvidia previously installed normally with smxi and working fine.

I got errors there :



then



It might be to do with something I misconfigured (it is my "experimental" install) but I can't see what
Back to top

Please confirm you are running the latest version of sgfxi: 4.13.0
sgfxi -v
There was a bug that returned a failed value when none was required.

Also, make sure you have: /lib/modules/$(uname -r)
and that in that directory you have the further path:
/lib/modules/$(uname -r)/kernel/drivers/video
then check in that directory for the nvidia.ko module.

All the tests I've done so far show that module file exists if the nvidia installer has built the module / installed the driver.

As you note, all the tests passed and it showed they failed, so that suggests that you may have gotten a buggy sgfxi.

If you try the test again, and it fails again, please use paste.debian.net and give the link to your paste for /var/log/sgfxi/sgfxi.log

<update>
Oh, I had to look at that several times:


It's much better to literally copy and paste commands than to type them, it avoids errors. To copy from gui to terminal, simply highlight command with mouse, then click in the terminal/console window, and use your center mouse button, click it, to paste it.

However, that still doesn't explain why the module exists test failed to show the existing module.
Back to top

Just installed 2.6.32-9.dmz.1-liquorix-686. Used smxi for that and nvidia driver, rebooted.

Same errors as before.

sgfxi 4.13.0

/lib/modules/$(uname -r)/kernel/drivers/video/nvidia.ko exists

sgfxi.log: paste.debian.net/62139/

It won't work out of X either. Nobody else is finding errors, nor me in Lenny, maybe the fault might is here although this system is apparently OK otherwise
Back to top

first, do not use -! 8, that is only for the initial testing phase, which is now completed.

Second, the module exists if you installed the driver with smxi after the kernel install, so there is no error, it's correct.

If you have REMOVED the module, and you get that error, then there is certainly an error, but your initial remove path you entered was wrong, incorrect, so of course no module could be removed to run the test, it was the wrong path.

So if you, from X, now, remove the module using the correct remove command, which will be clearly correct because the module will be removed, no file missing error occurs, then run:
sgfxi -X
(assuming you are in the current kernel, your new one), you should see the module rebuild, and everything should be as expected.

If you just installed the kernel this morning, and if sgfxi was at 4.13.x, then the module got built as expected, or should have. And so it exists. If it was before I uploaded 4.13, then smxi would have had sgfxi install the new driver, which would also create the module, but in that case, only for the new kernel.

I am suspecting simple user error here, or an incorrect assumption about what sgfxi is doing here.

But please keep the testing data coming, some of these errors may expose some poorly worded text in the script, which might cause confusion as well. There are lots of types of bugs here that are possible with such a big core change, and I'm sure they will get exposed over the coming weeks.
Back to top