[01:40] <jbicha> it looks like I need to re-release Ubuntu GNOME amd64, I didn't realize we needed linux-signed-generic for efi boot to work
[04:22] <slangasek> jbicha: you don't
[04:22] <slangasek> except on stgraber's hardware
[04:23] <slangasek> (this is an as-yet-undiagnosed bug in grub or shim; on reference hardware it boots unsigned linux just fine)
[04:25] <jbicha> ok, well we found a second case then, it's third hand though but I told them to file a bug
[04:25] <slangasek> ok
[04:25] <jbicha> what package should they file the bug against?
[04:26] <slangasek> mind you I'm not saying you can't reroll the images, I'm just saying it's not true that you need linux-signed-generic for efi boot
[04:26] <slangasek> package - grub2
[04:26] <jbicha> why does ubuntu amd64 have linux-signed-generic pre-installed though?
[04:27] <slangasek> furthermore, the only reported failures to date are with booting unsigned kernels under secure boot... is this user running SB?
[04:28] <slangasek> because linux-signed-generic lets us hand off to the kernel without calling exitbootservices first, which allows leveraging of some EFI quirk handling in the kernel early boot
[04:28] <slangasek> (e.g., the not-yet-landed fix for bug #1065263)
[04:28] <ubot2> Launchpad bug 1065263 in linux (Ubuntu) "wrong stride for efifb on some systems" [High,In progress] https://launchpad.net/bugs/1065263
[04:43] <ScottK> slangasek: I'm now convinced that multiarch in python3.3 is going to matter a lot more for migration than the actual changes in 3.3.
[04:45] <ScottK> Unfortunately I think I mostly understand the PyQt build system now.
[04:45] <ScottK> I had to waste a lot of brain cells on it in order to zero in on a two line fix.
[05:28] <micahg> ^^ rejected since I forgot the bug #
=== yofel_ is now known as yofel
=== doko_ is now known as doko
[16:57] <cafetiere> slangasek: any amd64 kernel lets you hand off without exiting boot services signed or not
[16:59] <cjwatson> but grub will not actually do so unless it's signed
[16:59] <cjwatson> so kind of moot
[16:59] <cjwatson> (well, I forget if it will allow that if SB is off)
[18:48] <slangasek> ScottK: multiarch> doesn't surprise me