[08:08]  * apw yawns
[10:49] <apw> bug #526045
[10:49] <ubot2> Launchpad bug 526045 in grub2 (Ubuntu) "no entry for device-mapper found" [Low,Fix released] https://launchpad.net/bugs/526045
[13:34] <NoNameYet_xnox> i guess that's what it's like to wait for merge window to open....
[14:26] <jsalisbury> ##
[14:26] <jsalisbury> ## No Kernel team meeting today.  Next meeting is April 29th, 2014
[14:26] <jsalisbury> ##
[14:27] <cking> thanks jsalisbury
[14:27] <jsalisbury> cking, np
[15:30] <bjf> jibel, can you verify bug #1088433 please?
[15:30] <ubot2> Launchpad bug 1088433 in linux (Ubuntu Quantal) "PERCPU: allocation failed when loading module kvm" [Medium,In progress] https://launchpad.net/bugs/1088433
[15:37] <jibel> bjf, I don't have access to these machines anymore.
[15:37] <bjf> jibel, ok, guess i'll revert that patch then
[15:38] <bjf> arges`, were are you able to verify ^
[15:39] <arges`> bjf: customer never responded to verify it unforutnatley
[17:35] <rtg> slangasek, has anyone developed a way to disable fast boot on UEFI systems ? There is a UI in Windows for it, but I'm not aware of a way to do it from Linux.
[18:06] <infinity> zequence: Were you going to do any regression testing on your lowlatency SRUs?
[18:06] <infinity> rtg: AFAIK, the way to do it is "boot Windows".
[18:07] <infinity> rtg: Given that only systems with Windows pre-installed have it enabled, that's not completely awful, though you need to remember to do it before you wipe the disk. :/
[18:21] <slangasek> rtg: we have a grub menu option that boots you back to the firmware
[18:21] <slangasek> rtg: called 'System setup'
[18:30] <antarus> slangasek: ohhhh you have commit access to linux-pam
[18:30] <antarus> slangasek: do you know if anyone actually implements new pam features?
[18:30] <antarus> slangasek: like..labels?
[18:30] <antarus> (instead of skip numbers?
[18:39] <rtg> slangasek, I'm not seeing that grub option. Only the installed kernel entries are in /boot/grub/grub.cfg. This is an upgrade from Quantal to Trusty, so that might have an impact.
[18:51] <slangasek> antarus: there are new features implemented from time to time, but they're generally new security features.  Has someone asked upstream for this new parser feature?
[18:51] <slangasek> rtg: the menu option comes from /etc/grub.d/30_uefi-firmware in grub-common; maybe have a look at why that script isn't doing the right thing on your system?
[18:52] <rtg> slangasek, ack
[19:00] <rtg> slangasek, well, I appear to have those files, and they are executable.
[19:00] <slangasek> rtg: "those files"?
[19:00] <rtg> /etc/grub.d/*
[19:01] <slangasek> rtg: sure; so what's wrong with your /sys/firmware/efi/vars ?
[19:02] <rtg> slangasek, bunch of stuff there too
[19:03] <rtg> hmm, OsIndicationsSupported doesn't exist
[19:03] <slangasek> well, then your firmware is buggy
[19:03] <slangasek> because that's the interface for disabling fastboot
[19:03] <slangasek> (and the same EFI variable that grub relies on setting in order to make its menu option work)
[19:03] <slangasek> can you file a bug against grub2 with your system details?
[19:04] <rtg> slangasek, ok, this is a tunnel mountain Tiano core reference platform. I've had it for a couple of years (I think). it came via manjo
[19:04] <slangasek> I'm not sure we can actually fix it, but we should at least document somewhere systems that have this problem
[19:04] <slangasek> hmm, and which version of the firmware do you have loaded?
[19:04] <slangasek> I guess if it's just a bad firmware load, no sense in reporting the bug against grub
[19:05] <slangasek> fwiw the strawberrymountain I have here has working OsIndicationsSupported
[19:05] <rtg> slangasek, ok, its not that big a deal. it is a legacy platform that never went into wide spread use.
[19:06] <manjo> rtg, I am guessing those systems were loaded with uefi 2.2
[19:06] <rtg> manjo, its the one you sent me eons ago
[19:06] <slangasek> yeah, I'm surprised fastboot is even implemented there - but it's implemented wrong
[19:06] <slangasek> fix is to upgrade the firmware
[19:06] <rtg> easier said then done (usually)
[19:07] <rtg> I'll research it
[19:07] <slangasek> if you can find the guide for it, the firmware upgrade should be fairly straightforward
[19:07] <slangasek> since the updates are themselves just EFI executables
[19:14] <antarus> slangasek: no, mostly just pissed on my end
[19:15] <antarus> slangasek: it looks like someone tried to make the parser do cool stuff, then gave up
[19:15] <antarus> (also I apaprently picked a bad channel to chat about this, sorry about that ;p)
[19:16] <slangasek> hmm, well, I don't think "cool" is the adjective I want attached to "parser for security-sensitive config files" :)
[19:16] <antarus> slangasek: at times I feel like pam is very much like openssl
[19:16] <slangasek> ouch
[19:16] <antarus> important, but barely functional
[19:17] <slangasek> the coding style is slightly more modern
[19:17] <antarus> slangasek: my hope is that a bunch of the 'session' stuff can move to logind or similar? it seems very bolted on ;)
[19:17] <slangasek> by at least a year or two
[19:17] <slangasek> um
[19:17] <antarus> no? ;)
[19:17] <slangasek> I fervently hope not
[19:18] <slangasek> logind is the best implementation there is of what it does; but it can keep its tentacles out of the pam sessions, kthx
[19:19] <antarus> so for example, we want to 'notify a user when someone that isn't them logs into their machine'
[19:19] <antarus> right now that is a pam module
[19:19] <antarus> but to me that is pretty terrible
[19:19] <antarus> I guess I wish there was a better session manager ;)
[20:11] <antarus> slangasek: https://fedorahosted.org/linux-pam/ticket/30 ;)
[20:11] <antarus> it doesn't seem terrible to implement at first glance
[20:52] <zequence> infinity: yea, sorry - will do it tomorrow
[20:54] <infinity> zequence: Ta,
[22:00] <antarus> slangasek: modern you say..
[22:25] <slangasek> antarus: I believe what I said was pam was at least a year or two years more modern than openssl ;)
[22:26] <antarus> I'm at least trying to write a patch! ;p
[22:27] <antarus> I was hoping to shoehorn labels masquerading as handlers, but that looks gross on initial implementation