1 |
On Sat, 2011-12-31 at 14:59 -0200, Claudio Roberto França Pereira wrote: |
2 |
> On Sat, Dec 31, 2011 at 13:53, Jeff Cranmer <jeff@××××××××××××××.com> wrote: |
3 |
> > I'm attempting to bring up a new system. Processor is AMD Phenom 1055, |
4 |
> > running on a Sabertooth 990FX motherboard. Graphics card is an NVIDIA |
5 |
> > GEForce GTX550Ti. |
6 |
> > |
7 |
> > If I start X with twm, the xserver connection is made, and TWM comes up |
8 |
> > correctly with three Xterm windows. |
9 |
> > |
10 |
> > I note that I get two errors reported, |
11 |
> > (EE) NVIDIA: Failed to load module "dri2" (Module does not exist, 0) |
12 |
> > /etc/X11/xinit/xinitrc: line 62: xclock: command not found |
13 |
> |
14 |
> This is normal behavior. Don't know why, but nVidia drivers doesn't |
15 |
> use the dri2 module, I think it has it's own implementation or |
16 |
> something, you can safely ignore that, even if your 20-nvidia.conf |
17 |
> disables (or tries to) dri2. |
18 |
> The xclock error is also ok, you probably have twm emerged, but not |
19 |
> xclock. The default xinitrc (in /etc/X11/xinit/xinitrc) tries to run |
20 |
> twm, 3 xterms and xclock. |
21 |
> |
22 |
OK, thanks - that explains that problem. |
23 |
> > |
24 |
> > The first error is definitely on startup, but the latter could have |
25 |
> > appeared on exit. |
26 |
> > |
27 |
> > I have a file 20-nvidia.conf in the /usr/share/X11/xorg.conf.d directory |
28 |
> > with the following information |
29 |
> > |
30 |
> > Section "Device" |
31 |
> > Identifier "Device0" |
32 |
> > Driver "nvidia" |
33 |
> > VendorName "NVIDIA Corporation" |
34 |
> > BoardName "GeForce GTX550 Ti" |
35 |
> > BusID "PCI:5:0:0" |
36 |
> > Screen 0 |
37 |
> > EndSection |
38 |
> > |
39 |
> > Section "Module" |
40 |
> > Disable "dri" |
41 |
> > Disable "dri2" |
42 |
> > Load "glx" |
43 |
> > EndSection |
44 |
> > |
45 |
> > If I add exec startkde to the top of the /etc/X11/xinit/xinitrc file, |
46 |
> > then startx fails to start. I get the same dri2 error as above, but |
47 |
> > also get |
48 |
> > |
49 |
> > startkde: Starting up... |
50 |
> > Connecting to deprecated signal |
51 |
> > QDBusConnectionInterface::serviceOwnerChanged(QString,QString,QString) |
52 |
> > kded(13865): Communication Problem with "kded", it probably crashed. |
53 |
> > Error message was: "org.freedesktop.DBus.Error.NoReply" : " "Message did |
54 |
> > not receive a reply (timeout by message bus)" " |
55 |
> |
56 |
> You should not edit this file, you should create a ~/.xinitrc and add |
57 |
> startkde to that. Try again without messing with the default system |
58 |
> xinitrc and using your own. You need to let the system xinitrc run, it |
59 |
> will call yours afterwards. |
60 |
> |
61 |
> |
62 |
Thanks - I made this change, but it didn't fix the problem. |
63 |
|
64 |
> > I verified using /etc/init.d/dbus status that the dbus service is |
65 |
> > running. Same for consolekit. |
66 |
> > |
67 |
> > /var/log/kdm.log shows one error, repeated for each startup attempt. |
68 |
> > KCrash: Appication 'kdmgreet' crashing.... |
69 |
> > KCrash: Attempting to start /usr/lib64/kde4/libexec/drkonqi directly |
70 |
> > Server Terminated successfully (0). Closing log file. |
71 |
> > |
72 |
|
73 |
After a little bit of further investigation, I discovered that the |
74 |
kdm.log file was full of old error messages. Deleting this log file did |
75 |
not result in a replacement file when I retried startx |
76 |
|
77 |
dmesg, however, was reporting some 'invalid opcode' errors. After |
78 |
finding a 2-page pdf on the correct compile flags to use with and |
79 |
amdfam10 processors (attached), I had enabled a couple of extra compile |
80 |
flags, -mabm and -msse4a, as well as setting -march=amdfam10. |
81 |
|
82 |
I've deleted the -mabm and -msse4a compile flag options from make.conf |
83 |
and I'm now running a full recompile (emerge -Dav system, emerge -Dav |
84 |
world). If that doesn't work, I'll try changing the architecture flag |
85 |
to -march=amd64 and recompile once more. |
86 |
|
87 |
Thanks |
88 |
|
89 |
Jeff |