NotebookForums.com › Forums › General Notebook Discussions › Linux & Other OS's › The 5680/8890 Gentoo Thread
New Posts  All Forums:Forum Nav:

The 5680/8890 Gentoo Thread - Page 3

post #41 of 170
Thread Starter 
Great stuff, Xiphux! I'll give radeonfb another try I will disable the highmem pagefile thing and see what happens. As for the multimedia keys, I've tried lineakd too and it locked up the input as well due to the Fn keys, etc. What keyboard are you using under lineakd?

Mikhail

UPDATE: Couldnt get radeonfb to work no matter what I did with the video string.
UPDATE: Got rid of the pagefile thing and of APM which was useless since it was disabled anyways (dont even know why I compiled that into the kernel in the first place). All seems fine.
post #42 of 170
I created my own lineakd keyboard def:

[Sager]
brandname = "Sager"
modelname = "NP8887"
[KEYS]
email = 236
www = 178
user = 129
back = 144
forward = 153
play = 162
stop = 164
volumeup = 176
volumedown = 174
[END KEYS]
[END Sager]

but you probably want to use xev on your own buttons, just in case your keycodes happen to be different.

Do you see radeonfb anywhere in your dmesg output, or is it completely devoid of any radeonfb status messages?
post #43 of 170
Thread Starter 
Xiphux:

Radeonfb is completely ignored. When I use vesa, it shows up properly but there is absolutely no sign of radeonfb. Ive tried all possible combinations of radeon and radeonfb video strings - nothing.

As for the keys, there seems to be a problem, as I mentioned earlier, with the way they are programmed to respond (some extra keys either dont respond at all, produce errors in the kernel log, give no release code at all, or give the press and release codes in the wrong order). Can you check (with xev) if all the keys you use produce the proper set of calls? Also, can anyone with a 8890 do the same? It would be good to determine wether this call issue is 5680-only or not.

Thanks

Mikhail
post #44 of 170
Thread Starter 
Hey everyone.

Wanted to post yet another thing I learned the hard way: when you update your kernel (emerge a newer version that is), DON'T FORGET TO RE-EMERGE ATI-DRIVERS!!! Otherwise, DRI will stop working.

Mikhail
post #45 of 170
If radeonfb is completely nonexistent in dmesg, it's not even seeing your card. If you're willing to get... 'experimental' with your kernel sources, then you can try the following (can't guarantee anything, though):

You need sysfsutils installed. (you're going to need both 'lspci', which is standard, and 'lsbus' which is in sysfsutils)

First of all, do an 'lspci -v' and find your video card. Note its pci number (the very first number before the device name. For example, my M9 gives "01:00.0").

Now, do an 'lsbus pci'. It'll give a list of devices, with pci devices like the ones from lspci. (except that they're prefixed with some extra zeroes) Find the entry that matches the pci bus from lspci. So, for example, to match mine, I would dig out the entry that looks like this:

0000:01:00.0: 1002:4c66

Those last 4 digits are your card's pci id. Now, you want to edit the kernel include "/usr/src/linux[-whatever-version]/include/linux/pci_ids.h". You'll see a whole bunch of defines, so scan down until you find the radeon section. First, check if your card is there, but I doubt it is. (Your sig says you have an M10 and I don't see it listed in my sources at the moment)

Technically, you should add a new define, but that requires adding it into a bunch of other source files, too. As a really messy and temporary hack, it's easier to just "hijack" one of the other defines and change its pci id. The card may be recognized with an incorrect name, but the device should still be found according to the pci id. So anyway, find the define that most closely matches your card (in your case, probably the M9), and change its id. Using my id as an example again, I'd find some card's define, and change it from 0x???? to 0x4c66. Just append "0x" to the beginning of the four digits you got from lsbus.

Recompile and try that kernel. It _might_ work, I really can't say - but that's how I hacked up the sources to recognize my M9, way back before it was supported natively in the kernel sources.


Huh, that's weird. My multimedia buttons don't send any X events at all. Maybe it's because I'm using an experimental version of x. But lineak still works fine, I'm just using the same codes that I got a long time ago.

Well, since the ati driver is a kernel module, unless you're using experimental module versioning support in the kernel, you're going to have to recompile it for each new kernel you use.
post #46 of 170
Thread Starter 
Great post, Xiphux!

I dont really need the console framebuffer that much so I'm not gonna do what you suggested, but this is definitely great info for others that might have more of a need. To add to what you said about the recognition of the card, my lspci states an unknown ATI device, while the ATI driver in X properly reports it as an M10 (although non-Pro, but that makes no difference) in fireglcontrol[s]. It even reports the DAC frequency, AGP info, etc.

As for the buttons, are you saying that /usr/bin/X11/xev (might be slithly off) gives absolutely no response to any of your multimedia keys? Have you tried turning lineakd off beforehand (as it might catch all the events)? If there is still no response, does dmesg give you keyboard-related errors? This is really intersting; I'm really puzzled why the engineers designed the extra buttons the way they did (non-standard event sequences).

Mikhail
post #47 of 170
I'm not surprised that the ATI driver reports the card specs correctly. I'd certainly hope that ATI knows their own cards.

Actually, I don't use radeonfb either. While radeonfb is nice and fast, it also has some compatibility problems with bootsplash, which I'm also using. So I just use vesafb for simplicity's sake.

Yes, I turned off lineakd. Actually, I was just lazy; xev didn't give anything, but I could read all the keycodes with lineakd -v and the events with showkey -s. I don't have any problems; the one key that doesn't work for me is the so-called "user" button, the one closest to the status LEDs. That one doesn't send any events period; nothing to lineakd or showkey at all. Oh well.
post #48 of 170
If it makes you feel any better, mmarkin, the front panel buttons stopped working for me too after upgrading to the 2.6 kernel. I think it may have something to do with the new event interface, but xev doesn't show them doing anything.

I posted up here before:

http://notebookforums.com/showthread.php?t=7150
post #49 of 170
Thread Starter 
I guess thats just an issue we have to live with for a while...
post #50 of 170
I guess so, but it still sucks.
post #51 of 170
Thread Starter 
Yeah...

Well, time for an update!

As suggested in a different thread, I got the Synaptics driver (I thought having Synaptics compiled into the kernel was enough; apparently not, lol) off here: http://w1.894.telia.com/~u89404340/touchpad/index.html. I set it up all as per the manual and it worked great (the only thing that I did different, that may or may not break stuff, is I used "auto" for the Protocol setting in the XFree config file instead of "auto-dev" as suggested by the manual). Since it is no longer a generic mouse, I can finally tap-click with the touchpad in Linux. In addition, the cursor is finally stable (before it was really shacky and would jump when I put my finger onto or lifted it off the touchpad) and all the nifty touchpad "jestures" work. If you even sparingly use the touchpad, give this a try (otherwise you'll drive yourself insane trying to use the touchpad under generic settings when an external mouse isnt around ).

Good night.

Mikhail
post #52 of 170
Or you can just 'emerge synaptics' for the latest driver.
post #53 of 170
Thread Starter 
Quote:
Originally Posted by xiphux
Or you can just 'emerge synaptics' for the latest driver.
Unfortunately that one is one or two versions behind. If you get it off the site, its spankin' new. Plus, its just a module that you compile and copy. Thanks for pointing it out, though.

'Night

Mikhail

P.S. When setting up or using the Synaptics driver for X, make sure you're using a kernel with the Event Interface option enabled. Otherwise, the synaptics driver will not be able to load and will crash X (if youre like me and have x/kdm start up on boot, this would sometimes lead to a system hang). Careful!
post #54 of 170
Just copy the ebuild and merge.

cp /usr/portage/x11-misc/synaptics/synaptics-0.12.1.ebuild /usr/portage/x11-misc/synaptics/synaptics-0.12.3.ebuild
emerge synaptics
post #55 of 170
Thread Starter 
Isnt that pretty pointless, though, Xiphux?
post #56 of 170
Why?
post #57 of 170
Thread Starter 
By doing that, arent you just faking a version? Or am I missing something? Please explain.

Mikhail
post #58 of 170
No, the ebuild reads the version info from the ebuild's filename. They're written with that in mind; new versions come out all the time, and it would have to be a pain to rework an ebuild each time. Doing it this way makes it a lot easier to add new versions (ones that don't need a different process in the ebuild script, which doesn't happen too often). Take a look at the ebuild; the source url line reads:

SRC_URI="${HOMEPAGE}/files/${P}.tar.bz2"

${P} is a variable inherent to portage which takes the package name from the ebuild's name.

If you copy it over and emerge, you'll see it get synaptics-0.12.3.tar.bz2 and merge it (I just did it).
post #59 of 170
Thread Starter 
So, Xiphux, are you saying that the 12.1 ebuild actually contains the 12.3 driver?
post #60 of 170
No, I'm saying that it reads the version info from the filename and acts accordingly. So, for the 0.12.1 ebuild, it sees the filename is synaptics-0.12.1.ebuild, and fetches synaptics-0.12.1.tar.bz2. Likewise, if you rename (copy) the ebuild as synaptics-0.12.3.ebuild, it will see that filename, and download synaptics-0.12.3.tar.bz2. The ebuilds are written to be dynamic like that.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Linux & Other OS's
NotebookForums.com › Forums › General Notebook Discussions › Linux & Other OS's › The 5680/8890 Gentoo Thread