| hallyn | apw: just got the 4.4.112 update :) | 02:05 |
|---|---|---|
| hallyn | of course, now i'm watching the 4.4.113-rc1 :( now with minty retpoline flavor | 02:05 |
| blaze | how can I check the status of applied fixes (IBRS and IBPB)? | 08:38 |
| jdstrand | apw, bjf (cc sbeattie, tyhicks): I booted a Lenovo IdeaPad U460 into the new 4.13 linux-generic-hwe-16.04 kernel and it immediately reboots when selecting it from grub | 13:28 |
| jdstrand | apw, bjf (cc sbeattie, tyhicks): it can boot the 4.10 kernel fine | 13:29 |
| jdstrand | apw, bjf: because it won't boot, I can't ubuntu-bug it. can you advise on how to report? | 13:29 |
| apw | jdstrand, +filebug is your only real option | 13:31 |
| jdstrand | apw: what would be good to attach? | 13:31 |
| apw | jdstrand, and that is the -31 one yes ? | 13:31 |
| tyhicks | jdstrand: the kernel team will know better but in the meantime, you may want to try booting with the "nopti noibrs noibpb" kernel command line options to see if there's any different behavior | 13:31 |
| jdstrand | apw: meta is 4.13.0.31.51, kernel is 4.13.0-31.34~16.04.1 | 13:32 |
| apw | jdstrand, if you have nothing other than a blank screen then there is little to add, trying tyhicks options is the next step | 13:32 |
| apw | jdstrand, is this the first 3.13 you have tried on this ? | 13:32 |
| jdstrand | apw: it is the first 4.13, yes | 13:32 |
| apw | ok | 13:32 |
| jdstrand | the 4.10 was ok | 13:32 |
| apw | yep, it has none of the new fun applied | 13:32 |
| * jdstrand nods | 13:33 | |
| jdstrand | apw: is there a pre meltdown/spectre 4.13 kernel that would be useful to try? | 13:33 |
| tyhicks | jdstrand: Ubuntu-4.13.0-21.24 is the pre meltdown/spectre 4.13 kernel | 13:37 |
| jdstrand | ok, I'll try that to see if it is a 4.13 thing or a security patch thing | 13:37 |
| jdstrand | apw, bjf (cc sbeattie and tyhicks): ok, I filed https://bugs.launchpad.net/ubuntu/+source/linux-hwe/+bug/1744942 | 13:51 |
| ubot5 | Ubuntu bug 1744942 in linux-hwe (Ubuntu) "Lenovo IdeaPad U460 fails to boot with 4.13.0-31.34~16.04.1" [Undecided,Confirmed] | 13:51 |
| jdstrand | I'll now try different command lines and kernels | 13:51 |
| jdstrand | I think it is a Westmere/Arandale if that is helpful | 13:52 |
| jdstrand | (though you might want to check the bug to be sure) | 13:52 |
| jdstrand | I also marked the bug 'Confirmed' to avoid the bot | 13:53 |
| jdstrand | tyhicks: I don't see any 4.13.0-21 kernels in https://launchpad.net/ubuntu/+source/linux-hwe/+publishinghistory | 13:54 |
| jdstrand | tyhicks: are you saying just boot the artful kernel? | 13:55 |
| tyhicks | jdstrand: there was a 4.13.0-21 artful kernel so I just assumed there was a 4.13.0-21 linux-hwe kernel | 13:55 |
| jdstrand | seems so | 13:55 |
| jdstrand | that vintage is still 4.10 | 13:56 |
| apw | jdstrand, it would have been linux-hwe-edge back then | 13:56 |
| jdstrand | ah | 13:56 |
| tyhicks | oh, right... linux-hwe was bumped from 4.10 to 4.13 | 13:56 |
| jdstrand | also if it helps, it is an i386 install | 14:00 |
| apw | jdstrand, it makes it less worrying :) in the sense if it is broke you are in a smaller set | 14:00 |
| apw | jdstrand, also does the latest 4.4 boot for you | 14:09 |
| jdstrand | apw: I'll test that too | 14:09 |
| apw | jdstrand, ta | 14:10 |
| jdstrand | apw: honestly, I forgot I was using 4.10 on that machine so going back to 4.4 wouldn't be so bad | 14:10 |
| jdstrand | we'll see how things go | 14:10 |
| apw | jdstrand, and you finding an issue is always good | 14:11 |
| albert23 | jdstrand: does that system have more than 4GB ram? | 14:18 |
| albert23 | i have bisected a 32 bit boot failure with > 4GB ram and got 92a0f81... x86/cpu_entry_area: Move it out of the fixmap as first bad commit | 14:20 |
| jdstrand | albert23: no. 4G | 14:20 |
| albert23 | hmm, I thought the >4GB was important as it boots fine when pae is disabled in the kernel | 14:24 |
| apw | albert23, it might not matter how much ram you have just that you are or are not using pae | 14:39 |
| jdstrand | apw, bjf (cc tyhicks and sbeattie): https://bugs.launchpad.net/ubuntu/+source/linux-hwe/+bug/1744942/comments/1 | 14:54 |
| ubot5 | Ubuntu bug 1744942 in linux-hwe (Ubuntu) "Lenovo IdeaPad U460 fails to boot with 4.13.0-31.34~16.04.1" [Undecided,Confirmed] | 14:54 |
| jdstrand | apw, albert23: note that pae is listed in cpuinfo | 14:54 |
| jdstrand | apw: right now I'm in an up to date 4.4 kernel, and it seems to be working so I'll stay there for now | 14:55 |
| jdstrand | (this is not my machine. if you need me to test kernels, I can, but possibily not immediately) | 14:55 |
| albert23 | jdstrand: pae is also shown when pae is disabled in the kernel, and top shows 3547 MB ram instead of 8000 | 14:58 |
| jdstrand | it only has 4G, I promise | 14:59 |
| jdstrand | :) | 15:00 |
| apw | jdstrand, sounds like a plan, we may have some h/w showing similar issues, so we can investigate | 15:00 |
| tyhicks | bjf, apw: FYI, chrisccoulson hit an issue where he could not boot an artful VM with the new kernel that we determined to be related to the bad intel-microcode that was reverted yesterday | 15:45 |
| tyhicks | here's the bad combination of versions: | 15:46 |
| tyhicks | host kernel: 4.13.0-25 | 15:46 |
| tyhicks | host microcode: 3.20180108.0~ubuntu17.10.1 | 15:46 |
| tyhicks | guest kernel: 4.13.0-31 | 15:46 |
| tyhicks | bjf, apw: the guest kernel booted after updating the host kernel (4.13.0-31) OR applying the reverted microcode package (3.20180108.0+really20170707ubuntu17.10.1) in the host and rebooting | 15:48 |
| tyhicks | I'm glad we reverted the microcode package yesterday | 15:48 |
| tyhicks | I don't think there's anything for the kernel team to do here but I wanted to point out the combination of old host kernel, bad microcode, and new guest kernel could cause VMs to not boot | 15:49 |
| tyhicks | the solution is to apply the updates in the host (either the kernel or microcode but both are ideal) and reboot | 15:50 |
| apw | tyhicks, thanks, that is a good data point | 16:06 |
| * alkisg still hasn't had any reply for "kernel reboots on i5's", https://bugzilla.kernel.org/show_bug.cgi?id=198529 ... | 16:12 | |
| ubot5 | bugzilla.kernel.org bug 198529 in x86-64 "Reboot on kernel load due to 92a0f81d" [High,New] | 16:12 |
| tyhicks | apw: btw, here's the stack trace from the guest kernel that chrisccoulson saw: https://www.dropbox.com/s/rt6v13rifm7qt8z/Screenshot%20from%202018-01-23%2014-25-00.png | 16:22 |
| apw | tyhicks, i wonder if chrisccoulson reported a bug with that, with what he was doing to trigger it | 16:25 |
| tyhicks | apw: he didn't but he could if it would help you all | 16:26 |
| apw | tyhicks, all and any issues; we want to know as much as possible about htem | 16:26 |
| tyhicks | ack | 16:27 |
| apw | as any one may be impossible to fathom but that third one gives you insight into the set | 16:27 |
| tyhicks | chrisccoulson: please file a bug to help out the kernel team track the issue that you hit (https://wiki.ubuntu.com/Kernel/Bugs) | 16:28 |
| chrisccoulson | tyhicks, sure, will do | 16:34 |
| tyhicks | thanks! | 16:35 |
| albert23 | alkisg: that's the same commit i found bad too, on i3-4010U (intel nuc) | 16:51 |
| albert23 | that commit is known bad, see https://lkml.org/lkml/2017/12/23/121, but the fix 2 commits later still doesn't work for me | 16:52 |
| alkisg | albert23: on 32bit installation, or 64bit? | 17:04 |
| albert23 | alkisg: 32 bit | 17:05 |
| alkisg | Thanks, I verified this on a number of i5 processors, but not on any i3's until you mentioned that one | 17:08 |
| alkisg | Happens on: i5-3470, i5-4460, i5-4440, i5-6200U, and doesn't happen on i3-4170, i3-7100, Q6600, Q8300, i5-4590 | 17:12 |
| alkisg | albert23: do you think we need to do something more, except the bisection I did to pinpoin the commit, and the bugzilla report? | 17:13 |
| alkisg | I would have expected some reply until now... | 17:13 |
| albert23 | maybe mark is a regression | 17:14 |
| alkisg | Could you? And also comment that it affects you too? | 17:15 |
| albert23 | I think it's better i you do it. At least Intel graphics never liked "me too's" in bug reports | 17:16 |
| alkisg | I thought that more than 1 user need to confirm a bug in order to get in the "confirmed" state | 17:17 |
| alkisg | Otherwise it's just monologue... | 17:17 |
| alkisg | Anyways, I'll revert the schools to 4.4 kernel and leave it for them to solve it when they can | 17:17 |
| alkisg | albert23: btw, does it boot for you if you pass acpi=off? It worked on most of i5's, but a teacher reported it didn't work for him | 17:28 |
| * alkisg also notes that there are no instructions in the ubuntu wiki on how to revert to the non-hwe kernel series... in the past, there were... | 17:31 | |
| albert23 | alkisg: yes, it does boot for me too with acpi=off | 17:32 |
| alkisg | Nice | 17:32 |
| albert23 | hmm, apparently acpi=off also mean no PTI | 17:43 |
| albert23 | and also no PTI in a kvm guest and no PTI when the kernel is built without PAE | 17:50 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
