[00:30] <mamarley> It looks like the AMD64 build of the 4.5.4 mainline kernel failed: http://kernel.ubuntu.com/~kernel-ppa/mainline/v4.5.4-yakkety/BUILD.LOG.amd64
[02:29] <kees> we need a new x86_64 flavor that disables CONFIG_COMPAT. :)
[02:33] <mamarley> Another thing I have noticed is that the 4.6-rc mainline builds have CONFIG_GOLDFISH_BUS=y, which causes "genirq: Flags mismatch irq 4. 00000000 (serial) vs. 00000080 (goldfish_pdev_bus)" and breaks serial.
[08:52] <apw> mamarley, yep, nothing which builds in yakkety works right now, the compiler changed to enable pie by default
[10:09] <mamarley> apw: How does it decide what to build on which distribution?
[10:10] <apw> mamarley, it is nominally version based.  the nearest ubuntu distribution to the version it is trying to build
[10:11] <mamarley> Ah, OK.
[10:11] <apw> mamarley, more accuratly the configurations it uses are from the nearest ubuntu release tag it can find, and it builds in the same release that that configuration came from
[10:12] <mamarley> So this is the same bug that is preventing DKMS from compiling kernel modules on Yakkety?
[10:13] <apw> mamarley, yes the same one.  DKMS should start working with the next copy forward kernel from X, as we have fixed it there)
[10:13] <apw> mamarley, but that does not fix mainline builds whihc also do not yet have that patch applied
[10:14] <apw> changing compiler defaults is bloody confusing to the world
[10:14] <apw> it is possible i can turn it off from outside, i've not yet tried that
[10:17] <mamarley> apw: While I've got you, it also looks like CONFIG_GOLDFISH_BUS is enabled in the 4.6 kernel configuration.  I know I can't file a bug on it yet, but that driver attaches to the UART and breaks serial.
[10:18] <apw> mamarley, oh you can file a bug on it, indeed for something which breaks things a reason bug is always useful
[10:18] <mamarley> Ah, OK.  I will do that then.
[10:18] <apw> if you do it now and tell me the number i'll get ti changed and then any reruns we do will have it off (for the mainline builds)
[10:26] <mamarley> apw: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1580960
[10:42] <apw> well that is special we have CONFIG_X86_GOLDFISH off, and CONFIG_GOLDFISH which _depends_ on it and therefore cannot be y unless it is, is y
[10:42] <apw> which is kinda impossible in theory
[10:42] <apw> and yet is not impossible in practice apparenlty
[10:44] <mamarley> Yeah, it looked weird.  It definitely breaks serial though.
[10:46] <mamarley> I guess the configuration will just need to manually specify "n" for all the goldfish stuff then?
[10:48] <apw> yep fixing it isn't the issue, trying to work out why the thing has not self-healed like it should
[10:52] <apw> mamarley, ok pushed for when i get the other fixes in
[10:52] <mamarley> Thanks :)
[10:54] <apw> i suppose i could switch all yakkety builds back to xenial or something in the interim
[11:01] <mamarley> apw: Yeah, that would at least keep them working for now.
[11:01] <apw> and my inbox full of people moaning they are on the wrong version
[11:01] <apw> then agian it is going to be full of people moaning they don't work anyhow
[11:02] <mamarley> Silly people :/
[11:32]  * ricotz is happily using the 4.6 builds from kernel ppa with nvidia blob
[11:32] <mamarley> Yeah, the -wily ones build amd64 and work fine, unless you have a device attached to a built-in serial port.  That should be fixed for the next one though.
[11:34] <mamarley> And that should be this weekend, unless the switch from building on Yakkety to Xenial rebuilds the ones that have already been built.
[11:34] <mamarley> (Since 4,6 will be released officially this weekend, unless Linus changes his mind.)
[12:07] <apw> mamarley, i may have fixed the pie issue, am rebuilding a 4.6-rc7 to find out
[12:07] <mamarley> Thanks :)
[13:02] <apw> mamarley, and those rebuilds look better
[13:44] <mamarley> Yay!
[13:49] <mamarley> I really appreciate your quick response.
[13:50] <apw> mamarley, np, it heads of every increasing howls in my inbox :)