Le 29/11/15 18:04, Stefano Zacchiroli a écrit :
On Sun, Nov 29, 2015 at 02:44:55PM +0100, Laurent Bigonville wrote:
Do you have a KMS driver loaded in the initramfs?
After discussion on IRC, I've done the following:
- added "i915" to /etc/initramfs-tools/modules
- update-initramfs
- got a warning stating some firmware might be missing
- apt-get install firmware-misc-nonfree
- update-initramfs again (verifying that the warning is now gone)
After the above, the boot works fine.
Comments/questions:
- it now works, YAY (and thanks!). But I conclude from the whole
experience that before now plymouth was somehow able to fallback to
askpass after self-detecting some sort of failure, while new versions
aren't any more. If that's the case, this might affect other users
Before the last update, plymouth was not enabled at all if splash was
not passed to the kernel cmdline, now the plymouth daemon is started
even if it's not passed, that probably explain the change of behavior.
- does the above mean that now non-free firmware is required to use
plymouth on i915 hardware? Or was the warning about something else
that is not strictly needed to make plymouth work? (sorry, I haven't
yet tried to re-remove the firmware and see what happens)
I'm not sure, I don't have a machine with an intel card to test, you
should test without them.
But usually that kind of firmware are used for 3D and video
decompression stuffs, I guess
I'm a bit surprised by the error message that I see in your 3rd picture, I
cannot reproduce this on my laptop even with plymouth completely disabled.
Did you modify anything in the initramfs scripts?
({/etc,/usr/share}/initramfs-tools/scripts)
I've had those messages since as long as I remember (with this laptop
having been bought in July 2014). According to debsums/etckeeper the
only relevant local change is:
- /etc/modprobe.d/dkms.conf, which is now empty, but used to contain
commented lines. It looks like it has been emptied by some postinst
script
That shouldn't be relevant.
You can also try to add plymouth.debug to the kernel cmdline (or
plymouth.debug=stream:/dev/kmsg)
I've done that, and took screenshots of all the debug messages before
the boot hangs. The 1st picture is the first page (obtained going back
up with Shift+PagUp); the 2nd picture is the 2nd page (overlapping with
the first one); the 3rd picture is the 2nd page, but in landscape rather
than portrait mode:
-http://upsilon.cc/~zack/stuff/IMG_20151129_174719.jpg
-http://upsilon.cc/~zack/stuff/IMG_20151129_174736.jpg
-http://upsilon.cc/~zack/stuff/IMG_20151129_174751.jpg
I cannot reproduce this without an intel card, could you please report
this upstream?
Cheers,
Laurent Bigonville
