=== ivan_ is now known as ivan
[05:29] <Mehul>   I am following https://help.ubuntu.com/lts/serverguide/kernel-crash-dump.html to enable kernel crash dump for Ubuntu-16.04. I want to automate entire configuration process through shell script, but as mentioned in the above doc, when "apt install kdump-tools" is run, it prompts a dialog asking "Should kdump-tools be enabled by default?". Is there any way to answer any prompted question in "YES" through script?
=== JanC is now known as Guest1067
=== JanC_ is now known as JanC
[07:39] <apw> Mehul (N,BFTL), I am sure you can, that is a debconf prompt, and i believe you can pre-answer them as you feel fit.  you might ask on #ubuntu-devel they might well know better than me.
[09:40] <oSoMoN> re- that libreoffice unit test failure: it’s not just the build that fails, the actual application crashes on x86, see bug #1702165 (and https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=865303)
[09:40] <ubot5> Debian bug 865303 in src:linux "libreoffice: Libreoffice Java features crash with Linux 3.16.43-2+deb8u1" [Serious,Open]
[09:40] <ubot5> bug 1702165 in libreoffice (Ubuntu) "Libreoffice writter crashes when JDK is enabled" [High,Confirmed] https://launchpad.net/bugs/1702165
[09:56] <smb> The problem is that the tests we got so far include one that does a simple jsvc run. And I checked this morning on a Xenial 32bit installed VM with the 4.4.0-83 kernel and at least that does work ok. Which puts us back at: what are exactly these java incantations/tests trying to do.
[10:00] <LocutusOfBorg> but since the regression seems to be in the kernel, can't we forcebad tests for it?
[10:00] <LocutusOfBorg> a lot of stuff will migrate if we exclude that bad test
[11:00] <cascardo> I can try running a libreoffice install... that should take less time than the build itself
[11:24] <apw> LocutusOfBorg, as we are saying it is likely the kernel is now "right" and anythign which does not work is broken, this might be the wrong approach
=== cachio is now known as cachio_afk
[16:12] <oSoMoN> smb: that libreoffice crash is happening there: https://cgit.freedesktop.org/libreoffice/core/tree/jvmfwk/plugins/sunmajor/pluginlib/sunjavaplugin.cxx?h=libreoffice-5-3-1#n831. The way LO instantiates a JVM is clearly much more complex than the simple jsvc run
[16:16] <smb> oSoMoN, Been watching the Debian side and they found a suspicious piece of code in jdk which might end up trying to use exactly that bigger area of code which is now forbidden (specifically for 32bit)
[16:16] <smb> I mean that bigger guard area
[16:17] <smb> http://hg.openjdk.java.net/jdk8/jdk8/hotspot/file/tip/src/os_cpu/linux_x86/vm/os_linux_x86.cpp#l881
[16:19] <oSoMoN> aha
[16:21] <smb> cascardo, were you looking deeper into that? I was kinda distracted by trying to cope with netplan and my not that simple network setup
[17:05] <cascardo> smb: yes, I am looking at that, trying to write a reproducer that would help us understand what is happening
[17:06] <cascardo> if not, I have managed to run libreoffice on i386 and get a segfault
=== cachio_afk is now known as cachio
[17:23] <apw> bwh has produced some more proposed patches: https://marc.info/?l=linux-kernel&m=149927397912305&w=2