Liquorix/Mint 13 ATI driver issues
Hello,I'm using the 3.7.0-9.dmz.1-liquorix-amd64 kernel on Mint 13 (Mate edition) because suspension to RAM wouldn't work with the default kernel:it works perfectly (and the whole process is much faster too,compared to other distros I've tried) with this kernel,but then I've hit kind of a brick wall,as I can't install the ATI driver in any way-apparently.
The traditional manual install (build the .deb packages with --buildpkg) fails on this kernel (and on Ubuntu mainline kernels,thinking of it) with every driver version I've tried (12.8,12.10,13.1),then I've tried the sgfxi route and I get loads of errors with every option I've tried so far,from "can't stop mdm" warnings to this message :: Code :: headerFile:
ERROR: (224) Your system is missing the required kernel header package: linux-headers-3.7.0-9.dmz.1-liquorix-amd64 and it is not currently available in your repos. Cannot continue. Please either update your kernel and kernel headers, or find and install kernel headers for 3.7.0-9.dmz.1-liquorix-amd64 You can override this test with -! 6 option. If install works, let maintainer know. but the weird thing is that there's no such header available when I search for it. Also,the output of sgfxi -L f says :: Code :: Kernels: 3.7 - 13.1 with patches, not yet in sgfxi
3.5, 3.6 - 12.10 probably works. 3.4 - Latest current. 12-6 only with patches. Working: 3.3 works with latest current. 3.0/3.2; 2.6.39 - use latest driver. which doesn't look very promising,does it? I've also tried the -B option but it says that there's no beta driver available. I've gathered from other threads that the Ubuntu/Mint way of packaging proprietary drivers and the unreliable support from ATI itself is the likely source of such issues,also being a total newbie at this kind of hacks (even more so using sgfxi,which I've installed yesterday) I may have done some obvious mistake. In short,can you give me some hint in the right direction to in install this ATI driver ? Is there a way to install a 3.3.x liquorix kernel (if such a thing exists,I can't see it in the repo)? If other info or the logs from this installation attempts are needed,please let me know and I'll post what I have. Thanks. Back to top |
|||||
Please install the current liquorix kernel and headers, there's not 'loads of errrors', there is you not installing the headers then a new liquorix version coming out, with its new headers. With that said, there's no guarantees at all with fglrx, I do not test it, and strongly recommend against trying to run it with current kernels. They do not support that, and I only marginally support it in sgfxi.
If you are running lmde, you can install an older liquorix, say, 3.6 version, and that will install the headers, the problems with newer kernels came about at 3.7.6 for video drivers. I have patches for 3.7 for fglrx but they are untested, no idea if they work, nobody has reported. I believe currently in debian smxi there are: 3.2, 3.4, 3.6 liquorix kernels available. I do not support, nor will I ever support, ubuntu, or ubunut mint, in smxi. however, thanks for the reminder, i do need to update that support message, I did add the patches, but I simply have received no word from users if they worked on 3.7. Back to top |
|||||
i've updated the -L f message to show that there are patches, but not yet shown to work or not work by users. Note this patch only is for 3.7 kernels, not 3.8.
I also updated sgfxi fglrx testing to show error with xorg 1.14 and kernel 3.8 if it wasn't already doing that. Back to top |
|||||
I completely missed that the installed headers didn't match the kernel version,thanks for pointing this out-I've now installed the latest versions of both.
Of course I meant that *I was* getting loads of errors,certainly not the the errors are actually in the script. Based on what you're saying above,he ATI driver won't probably work on this 3.7.x line of kernels anyways,right? If so,is there any way I can install the previous 3.6 version,or try to patch the current 3.7 ? Thanks. Back to top |
|||||
I already answered both of your questions. sgfxi will run a 3.7 kernel patch which may or may not not work, nobody using amd/ati has reported yet.
besides the debian only smxi archived liquorix kernel install options, damentz also currently has a 3.4 kernel in his 'past' repository. Back to top |
|||||
I've now added the "stable" Liquorix repo and installed the 3.4.0 kernel and headers,then I've run the sgfxi script and,after some struggling with the Mint mdm manager getting in the way and some warning messages about some possibly missing firmware,shutting down the computer and re-running the script,the 13.1 ATI driver is finally installed on my Mint 13 system.
Apparently everything works, only with the oddity that fglrx doesn't show up as installed when I search for it with dpkg or synaptic,probably a quirk specific to Mint that doesn't recognize manually installed drivers. Back to top |
|||||
fglrx not showing in menus is not a quirk, it's how package management systems work, since it's not a package, the system has no way to know about it, apt in this case. Actually it depends on the installer, for example, nvidia properly inserts its settings tool into the menus, but that's up to the installer programming, and given this is amd/ati, not a lot of hope for having it done right there. It can also be other things, like mint not doing it right, or it not being where you think it is, lots of options. You can always add it manually though in the menu, assuming your menu thing comes with a menu editor like kde does.
Can you be more specific about mdm getting in the way? I thought sgfxi sought that out and tried to kill it, its on the list of login / display managers to kill if detected. If it fails to do that, it would certainly explain a long time mint bug reported now and then about sgfxi not being able to terminate X, however your description adds the critical detail of why, and what, might be causing that issue. You can check this by doing: grep -n mdm /var/log/sgfxi/sgfxi.log then opening the file at that line number and seeing what sgfxi says about mdm. If you don't see mdm at all in sgfxi, it means that mdm isn't actually being detected, which would of course account for it not being shutdown. Slim had issues in the past with failing to shutdown as well, and it would certainly not surprise me in the slightest to see that mdm has similar bugs. Back to top |
|||||
What happened is that,when the script running in a console tried to shutdown the X server (I think),the X server took over again and the mdm login screen popped up:so I logged in,then switched again to the same console where sgfxi was still doing its thing,and took from there.
I'm afraid I can't be more specific here,as it's the first time that I've successfully used this script and installation hasn't maybe gone 100% smoothly,I have a feeling that some steps have been done more than once,also I've got some warnings about removing the radeon module and about some missing firmware:if I knew how to copy and paste from the console I could have saved that info. But below are the mdm occurencies in the sgfxi log,hope that is useful in some way :: Code :: Args: stop
default runlevel: current runlevel: 2 dmanCommand: stop mdm Function: x_is_running - Utility: Start xIsRunning: false Function: x_is_running - Utility: End -- Args: start default runlevel: current runlevel: 2 dmanCommand: start mdm Function: start_stop_x - Utility: End Function: start_x_question - Primary: End Function: x_start_stop_master - Primary: End Regarding the menu,ATI did it right.. it automagically adds a menu entry:what I meant is that I can't locate the actual fglrx driver with dpkg or synaptic,which is explained by your remark above. BTW,suspend/resume is working flawlessly so far with this kernel:unlike other workarounds I've had to do before to make it work in Ubuntu,here the system goes to sleep and awakes everytime (no blank screens or failures to resume),and it does this quickly too,and it also resumes from the keyboard,which I've never seen before.That's impressive,from my standpoint. Back to top |
|||||
ah, thank you, your description matches what I suspected, slightly, re mdm, it's got something to make it respawn if it gets killed, I believe. You've done a better and more complete job describing the event than anyone else, so pat yourself on the back, now I have an idea where to look, though I can't remember how to handle it when something is told to respawn.
I see, re the driver, that's correct, you will never see the driver per se listed in the package manager, and good to hear that amd/ati did it right, it's good to know what they do right and wrong. For seeing what driver, I always use inxi. I don't think debian/ubuntu list the driver in the menu either, even if you install it with their packages, it's not an application after all, it's a driver, like any other kernel driver, or the kernel, or any other core system library. Can you post your: /etc/inittab if present, the contents, that sometimes contains some respawning directives, but I'm reaching here, not sure. But this is almost certainly the source of the problem, if you could ask on mint forums how how you actually kill mdm without it restarting or whatever, and then let me know, maybe I can see about an added test or condition, not sure though. I have seen that behavior before myself, but it's been a while, and I can't remember what I did to deal with it, if anything. Back to top |
|||||
Thanks-on regard to inittab,there's no /etc/inittab on my system (or anywhere in the / directory,aside of
/usr/lib/upstart/migrate-inittab.pl) A quick search on the mint forums for an effective way to stop mdm returns -as expected- multiple options,from the obvious sudo service mdm stop (which didn't work) to the most credited :: Code :: sudo /etc/init.d/mdm stopBack to top |
|||||
All times are GMT - 8 Hours
|