Re: Bug 648732 – Intel wireless broken on 11n for many users

[Ian Malone]

On 6 December 2011 21:40, Ian Malone <ibmalone at gmail.com> wrote:
> On 6 December 2011 18:23, Pete Travis <lists at petetravis.com> wrote:
>
>> On Dec 6, 2011 7:52 AM, "R. G. Newbury" <newbury at mandamus.org> wrote:
>>>
>>> On 12/05/2011 10:47 PM, Lawrence Graves<lgraves95 at gmail.com>
>>> > On 12/05/2011 01:32 PM, R. G. Newbury wrote:
>>> >> >  On 12/04/2011 09:44 PM, Lawrence Graves<lgraves95 at gmail.com>
>
>>> > After doing all the above from 1-18, it is still doing the same thing
>>> > and gave me the same screen as when I first start this journey.
>>>
>
>>>
>>> I didn't closely follow the beginning of the thread. Could you repeat
>>> what hardware you are using, and confirm the distro level (F15 I think)
>>> and kernel. I am interested in confirming the video chipset and the
>>> wireless chipset, since *something* weird is going on, involving those...
>>>
>
>> I've had some very frustrating problems with making GPUs work, learned a
>> little, and experienced an incredible amount of guidance and patience from
>> the Fedora community along the way.  I'll throw out a couple observations
>> I've made that may help.
>>
>
>> With the last two Fedora releases, installing kmod-nvidia has automagickally
>> ran dracut and appended my grub entry. Likewise, removing it has revoked
> After the last upgrade I had to do this manually.
>
>> those changes, as far as I can tell.  If you need to remove the nvidia
>> drivers for testing, I would suggest using this command:
>> yum remove $(rpm -qa | grep nvidia)
>> Try the section inside the parentheses by itself to see what's going on.
>> After you remove the packages, you will need to reboot for the changes to
>> take effect.
>>
>> It is also a good idea to install akmod-nvidia with kmod-nvidia.  The driver
>> needs to be built for a specific kernel, and when they get out of sync, this
>> will fill in the gaps.
>>
>> Experimenting with kernel parameters can be diagnostically useful, or
>> provide a workaround. To use them, when grub presents the menu of kernels to
>> boot, press 'e' to edit, then add them to the line that begins with 'linux.'
>> A few I've found useful when troubleshooting are:
>> 1 - boots to a fallback root console.  You have probably already found
>> something like this.
>> nomodeset - disables kernel modesetting, changing how the kernel works with
>> the GPU and it's driver.  I don't understand it well enough to offer a more
>> technical explanation, but I know its worth a try with and without it.
>>
>> acpi=off - disables advanced power management features.  Laptops are
>> especially prone to poor acpi implementations.  I have to use this to boot
>> with the nvidia drivers on my MSI laptop.  I also noticed that this disables
>> suspend and causes the power button to instantly drop power to the system,
>> so press it with caution with acpi disabled.
>>
>> One more thing - I can't recall if the default nouveau drivers didn't work
>> for you, or if you weren't happy with them.  At the very least, we should be
>> able to get vesa drivers working for you.
>>
>
> I assume the nouveau drivers are working, otherwise he wouldn't have
> been able to take those screenshots, and all the X logs I've seen from
> Lawrence so far show nouveau being loaded. Which brings me on to...
>
> Quick note on Xorg logs: if X isn't starting because of a driver
> problem the log will record the problem. However as I said, I haven't
> seen one yet which didn't indicate nouveau. I tried to indicate we
> need a log from the computer when nvidia is failing to load, but
> perhaps I should have gone into more detail. In F16 they get called
> /var/log/Xorg.N.log where N is a number starting at 0. And they get
> overwritten at each restart, which is why I was asking for a copy from
> when the system is failing to start X, by the time you've rebooted
> with the nouveau driver and logged back in it's gone.
> For similar reasons screenshots from a running X session aren't going
> to give much that can't simply be sent as copies of the text files.
> So either:
> We need a copy made during a session where X fails to start.
> We need a copy of one of the other attempts, if X tries a few times it
> will create a new one for each attempt, /var/log/Xorg.1.log to
> /var/log/Xorg.5.log, these may be left over when you start X again
> with nouveau.
> $ ls /var/log/Xorg.*.log -lhrt
> should give an indication of what logs are remaining and when they're from.
>
> A log from a failed start attempt will probably contain at least some
> indicators of what's going wrong, if not the exact problem. It's best
> to try and get this before starting to mess with aerials and things.
>
> Okay, with that said, checking the nvidia modules are installed:
> $ ls /lib/modules/$(uname -r)/extra/nvidia/ -l
> total 16400
> -rw-r--r--. 1 root root 16769688 Nov 23 14:39 nvidia.ko
> $ ls /usr/lib64/xorg/modules/extensions/nvidia -l
> total 8508
> lrwxrwxrwx. 1 root root      16 Nov 27 16:23 libglx.so -> libglx.so.290.10
> lrwxrwxrwx. 1 root root      16 Nov 27 16:23 libglx.so.1 -> libglx.so.290.10
> -rwxr-xr-x. 1 root root 8392712 Nov 17 02:06 libglx.so.290.10
> lrwxrwxrwx. 1 root root      23 Nov 27 16:23 libnvidia-wfb.so ->
> libnvidia-wfb.so.290.10
> lrwxrwxrwx. 1 root root      23 Nov 27 16:23 libnvidia-wfb.so.1 ->
> libnvidia-wfb.so.290.10
> -rwxr-xr-x. 1 root root  295416 Nov 18  2010 libnvidia-wfb.so.290.10
>
> $ modprobe nvidia
> (Normally we'd run this as root, but just want to see any errors)
>
> --
> imalone

Hi, Lawrence was able to send me some pictures of the X log (aside, I
realise it's tricky to work on the system when X wont start, simplest
way to get round this if you're able to get a command line is to take
copies of the file via the cp command. Anyway, back to the point...)

Typed out it looks like this (after the preamble):
(==) Log file: "/var/log/Xorg.0.log", Time: Tue Dec 6 15:17:51
(==) Using config file: "/etc/X11/xorg.conf"
(==) Using config directory: "/etc/X11/xorg.conf.d"
(==) Using system config directory: "/usr/share/X11/xorg.conf.d"
(EE) NVIDIA(0): Failed to get supported display device(s)
(EE) NVIDIA(0): No display devices found for this X screen.
(EE) Screen(s) found, but none have a usable configuration.

Fatal server error:
no screens found

...followed by X giving up and closing the log.

The machine is a Dell Inspiron 9400, with NVIDIA Go 7900 chipset, some
googling suggests (as in fact John Pilkington suggested a while back)
that this is an Nvidia 290 problem,
https://bbs.archlinux.org/viewtopic.php?pid=1022749 and reverting to
the 285 driver, i.e. downgrade to the packages:
kmod-nvidia-285.05.09 and xorg-x11-drv-nvidia-285

-- 
imalone