=== RAOF1 is now known as RAOF === JanC is now known as Guest95676 === JanC_ is now known as JanC [11:39] 1.19.3 uploaded to ppa:canonical-x/x-staging [16:04] Hello, wondering if someone can give me a hint about the following. [16:04] In the VirtualBox Guest Additions we install a drm driver into /lib/modules which should get auto-loaded at boot time. A user is claiming that in a Debian Sid virtual machine, X.Org often starts before the module is loaded. [16:04] Is this expected? Not sure what else the user might have done of course. And if it is expected, should I be doing something to prevent it? [16:07] the driver probably isn't in the initrd image [16:07] ? [16:07] Kernel 4.8 broke 16bpp modes in virtualbox for me (I see two smaller broken images of the screen instead of one large one), while in 4.4 it worked, would that be related to the vbox code? Should i report that to the vbox bug tracker or to the kernel or to xorg? [16:09] tjaalton: don't think so. I tried putting it into initrd's on various systems (though it is only Debian that puts graphics drivers there), but it felt too fragile for me. For me it works without. Am I just lucky? [16:09] I assumed that the initrd graphics drivers were for the boot splash. [16:09] kms drivers [16:09] if you put it in /lib/modules it's a kms driver? [16:10] Exactly. [16:10] so it belongs there [16:10] Of course, I usually test Ubuntu which ships it by default. [16:10] ships what? [16:10] So that is probably why I am not noticing. [16:10] vboxvideo.ko [16:10] ah [16:10] And if it is in the initrd would that be expected to work? [16:11] I'd say so [16:11] otherwise X might start before the driver has a chance of loading [16:11] I guess [16:11] Oh dear, back to initrd fun then. It seems to go wrong in so many different ways when you try to add something from outside. [16:12] Actually, remembering, with Ubuntu one could just drop a file into some directory and ask to have the initrd rebuilt. [16:12] Thanks. [16:13] should be the same trickery [16:13] unless there's some ubuntu'ism [16:13] in initramfs-tools [16:14] Hopefully not. [16:15] you can check the diff to be sure :) [16:16] /usr/share/initramfs-tools/scripts/init-top/framebuffer [16:16] nope [16:16] /usr/share/initramfs-tools/hooks/framebuffer [16:16] copy_modules_dir kernel/drivers/gpu [16:17] so install a new hook [16:20] Ahem. That file exists in 0.125ubuntu9 but not in 0.127 from Sid. [16:20] Here's what I'm seeing with kernel 4.8: https://snag.gy/H3FeMW.jpg [16:21] alkisg: if you are using Guest Additions then probably yes. [16:21] michael-vb: ubuntu 16.04.2, I think it has some video module preinstalled? [16:21] I didn't install anything manually [16:23] Can't remember what Ubuntu 16.04 provides. Especially not .2. Could you pnopaste your Xorg.0.log from the guest for reference? [16:24] In any case, the Ubuntu maintainer for the VirtualBox drivers is not in this channel, so perhaps we should not be troubling other people with this. [16:25] We could switch to #vbox for now with it. [16:25] michael-vb: if it's a vbox issue, of course, I'll come there, thank you [16:27] tjaalton: so it looks like Debian (Sid at least) does not do kms drivers in the initrd. [16:51] michael-vb: then I don't understand the original bug [16:51] why it happens [16:51] Should Xorg wait for udev to finish loading drm drivers before starting? (It is udev, isn't it?) [16:52] whatever starts xorg [16:53] lightdm has [16:53] After=systemd-user-sessions.service getty@tty7.service plymouth-quit.service [16:54] And in Sid that is probably gdm. So I should probably be asking whoever takes care of that. [16:55] look at it's service file [16:57] Thanks. [16:57] Have to be moving, children to feed... [16:58] After=rc-local.service plymouth-start.service systemd-user-sessions.service [16:58] dunno what happens if there is no plymouth [16:59] hmm nothing