Cannot boot to compiled kernel
danielhilst at gmail.com
Thu Nov 10 07:09:42 EST 2016
2016-11-10 2:48 GMT-02:00 Daniel Bokser <dan.bokser at gmail.com>:
>> On Nov 8, 2016, at 5:50 AM, Andrey Skvortsov <andrej.skvortzov at gmail.com> wrote:
>> On 16-11-07 23:26, Daniel Bokser wrote:
>>>> On Nov 3, 2016, at 11:22 AM, Matias Mucciolo <mmucciolo at suteba.org.ar> wrote:
>>> Thanks everyone! So I compiled keyboard, SATA, and BTRFS drivers in and was able to get further, but it looks like it had trouble starting up LXDE (it was repeatedly starting and stopping LXDE). Booting up in single user mode, I was able to get a prompt. I attempted startx but it failed. I looked and it looked like it was attempting to use the Nvidia modules, but they weren’t in the /lib/modules/4.9.0… directory (I have an Nvidia 670). It looks like they are stored as extra modules in the extramodules softlink in my default installation (now kernel 4.8).
>>> So I am gathering that many of the modules used in my default installation are not making it over to 4.9. When running make localmodconfig, I am noticing error messages related to Nvidia and virtual box. Some examples:
>>> Nvidia config not found!!
>> You have used proprietary driver from Nvidia. It is not included in
>> the kernel. I think this is the reason for your problem with LXDE.
>> The best way would be to use open source driver that is supported by
>> the mainline kernel - nouveau (config option DRM_NOUVEAU).
>> According to the nouveau site , your chip is supported. Try it.
>> 1. https://nouveau.freedesktop.org/wiki/FeatureMatrix/
>> Best regards,
>> Andrey Skvortsov
> Thanks! I tried to use nouveau. It seems that the kernel cannot load any dynamically linked module, including nouveau. I am kicked to the emergency console (after statically linking all the necessary drivers to ready my SSD). When I do a modprobe nouveau, I get a bunch of Unknown symbol errors similar to the errors I stated in my previous email. It errors out stating ERROR: could not insert nouveau: Invalid argument
> Some examples include:
> drm: no known symbol version for memcpy
> drm: Unknown symbol memcpy (err -22)
> drm: no known symbol version for __put_user_4
> drm: Unknown symbol __put_user_4 (err -22)
> It seems there is an underlying problem with loading the dynamically linked modules. lsmod reveals exactly 2 modules: crc16 and hid_generic. Honestly I am surprised there are any modules here.
> As was stated before, there might have been a problem compiling these modules. Has anyone encountered this issue before?
Usually this means that the modules you're trying to load wasn't
compiled to the running kernel. But it may be another problem. If you
have /proc/config.gz you may check if the CONFIG_* options that you
selected are really there, with something like `zcat /proc/config.gz |
"Do or do not. There is no try"
More information about the Kernelnewbies