Comment 19 for bug 1082742

Revision history for this message
Matt Sealey (mwsealey) wrote :

For the edification of the reporters I can confirm this doesn't just happen on Chromebook but on an Efika MX running kernel 3.7, using the imx-drm staging driver. Xorg seems to crash a hell of a lot too, but I can't tell if it's related. Using xserver-xorg-video-modesetting or -fbdev makes no difference (it will explode the same way), and trying to get Plymouth out of the way makes no difference.

Xorg, modesetting driver/libkms and Plymouth using the same seems to get ARM systems no matter what they are into a complete tizzy - indicative of something far, far more serious floating around (maybe plymouth or xorg is being miscompiled/misconfigured for armhf?) than just some esoteric Chromebook bug.

At some point Xorg will segfault and what you see behind it is this Plymouth message about ply_terminal_open. The console= lines on the kernel command line make absolutely no difference here - my cmdline is "console=tty1 root=/dev/mmcblk0p3 rootfstype=ext4 rootwait rw quiet splash".

I did notice that no matter what I do with mkinitramfs or update-initramfs with "-v" option. it seems like no plymouth themes or configurations are being copied into the ramdisk which seems quite odd to me. A lot of X11 libraries are thrown in, a lot of Plymouth support libs, and a file ubuntu-text.so but the actual splash images, logos are annoyingly missing if you correctly enable initramfs-tools/conf.d/splash FRAMEBUFFER=y. It doesn't seem like the plymouth hooks are doing the right thing at all.. they DO seem to do the exact right thing under a much older Ubuntu (Maverick), though, where themes and images are put into the ramdisk.

BTW boot.log is also showing some kind of mountall: Event failed before console_setup runs and then it says Starting userspace splash daemon. However, what event failed is a mystery; all the boot.log entries say OK and the system continues to boot after also saying "unexpectedly disconnected from boot status daemon"...

Has anyone figured out anything regarding this bug or anything related as to why Plymouth has so many problems here on ARM with a DRM/KMS framebuffer?