Nouveau fails with F17 Beta on GeForce 6150 (NV4E)

[Marcel Oliver]

Dear both,

thanks for your replies.  I have updated the bugzilla report with the
requested F16 log files using remote ssh.

However, I failed with both suggested methods to get boot logs off the
F17 beta life CD.

(a) 'nouveau.modeset=0' brings up plymouth, but when it finishes, it
    starts X with apparently the vesa driver, so working (poor) X and
    no nouveau debug output.  I did not find a way to disable
    graphical boot completely (and then change to X with the full
    nouveau driver).

(b) netconsole does not give me any output at the receiving end.  I
    have verified carefully that the receiving machine and the network
    is set up properly, and the source machine can send (on a
    functioning F16) UDP packets to the logger on the receiving end.
    However, with the netconsole on the kernel parameter line, nothing
    happens.  No idea how to debug this - I followed the instructions
    quite carefully and verified the MAC address as suggested as well.

In any case, I suspect that the problem is not F17 specific, but
rather tied to the 3.3 kernel, so that the F16 logs could possibly
suffice.

Could you let me know if you can work with the information provided,
or if there anything else I could try?

Regards,
Marcel

Konrad Rzeszutek Wilk writes:
 > On Thu, Apr 19, 2012 at 01:38:01PM +0200, Marcel Oliver wrote:
 > > Hi,
 > > 
 > > I am having worsening problems with the nouveau driver on GeForce 6150
 > > (NV4E), this is the onboard graphics on the ASUS M2NPV-VM.  So I tried
 > > F17 Beta (x86) yesterday to see what the status is.  Result: total
 > > failure - the screen goes to a very dark blue and the keyboard appears
 > > also dead.
 > > 
 > > This same problem already shows up on F16 with any of the 3.3 kernels
 > > (last working kernel is 3.2.10-3 without acceleration, acceleration
 > > never worked on this chip), see
 > > 
 > > https://bugzilla.redhat.com/show_bug.cgi?id=810490
 > > 
 > > Question:
 > > 
 > > - Should I file another bug report on F17 beta?
 > 
 > Can you just add the serial/dmesg output with drm.info=255 to the bug?
 > > 
 > > - If so, what are the chances that someone looks at it?
 > > 
 > > - Is there any easy way to extract the requested debug information
 > >   from the system in the failing configuration?  Ideally, I'd like to
 > >   write out to USB stick without installing the system, but since no
 > >   user interaction is possible, that seems too much to ask?
 > >   Alternatively, if I install on a scratch partition, is there a way
 > >   to instrument the startup code to write the necessary debug
 > >   information to disk for later inspection from a working system?
 > >   Networking would be a bit of a pain to bring up...
 > 
 > You could boot in text mode (so recovery) with 'nouveau.modeset=0'.
 > Set up your network, SSH in the box, and then 'exec /sbin/init 5'
 > to start X. Then from your SSH session run 'dmesg' to get that
 > output.
 > 
 > Oh, you should also have 'debug loglevel=8' on your Linux command
 > line.
 > 
 > > 
 > > Any comments appreciated, also an estimate of the chances of being
 > > able to use Fedora on this machine again...
 > > 
 > > Regards,
 > > Marcel
 > > -- 
 > > test mailing list
 > > test at lists.fedoraproject.org
 > > To unsubscribe:
 > > https://admin.fedoraproject.org/mailman/listinfo/test