=== tmpRAOF is now known as RAOF [01:19] hello i have a question: i want to apply pf patch and update my kernel to 3.19, how can i do that while ensuring the amd drivers still work? [01:20] when i upgraded the kernel i had to purge fglrx in tty as my disktop will not load === gerald is now known as Guest79364 [09:06] kees: i'm a bit confused about ARM_KERNMEM_PERMS (and DEBUG_RODATA) vs DEBUG_SET_MODULE_RONX [09:08] kees: if i read it correctly, they do the same thing (make all sections but text NX and RO) [09:09] kees: but the second seem to be geared towards modules only (IOW it seems to to the same dinamically on module loading etc) [09:09] kees: are these options complementary? can i turn on all of the or ARM_KERNMEM_PERMS (and DEBUG_RODATA) suffice to cover all the cases? [09:10] kees: and, last question, what happens when some legit code try to patch a text section? (e.g. JUMPLABELs)? does it handle that case too? === txspud|ORS is now known as txspud === work_alkisg is now known as alkisg === hallyn_ is now known as hallyn [16:43] Heya, there is a bug in a DKMS package for Vivid: https://bugs.launchpad.net/ubuntu/+source/sysdig/+bug/1419402 - I proposed a patch, how do we get from there? [16:43] Ubuntu bug 1419402 in sysdig (Ubuntu) "sysdig-dkms 0.1.87-1: sysdig kernel module failed to build" [High,Confirmed] [16:56] Zhenech, is that already uploaded to debian ? [16:57] apw, debian has no problem, as debian does not ship .19 in jessie [16:57] .19 is in experimental, and sysdig is fixed there too [16:57] (debian bug for experimental is linked in lp) [17:00] bah, rmadison is hating me for debian [17:02] Zhenech, ok, using the right machine it works much better [17:03] become a DD and use dak on some debian.org host ;-) [17:03] heh .. one day indeed [17:04] Zhenech, anyhow, first step is bringing it to my attention [17:04] Zhenech, and i'll have a look [17:04] check :) [17:04] I hoped subscribing kernel team would be good enough [17:09] Zhenech, we get a heck of a lot of bugs subbed to us, it is easy to miss them, cirtianly coming here gets our attentoin [17:52] bjf: I'm seeing a kernel test failure (when running in GCE): http://paste.ubuntu.com/10622256/ Is that an actual problem? [17:55] Odd_Bloke, which series? [17:56] bjf: trusty. [17:59] Odd_Bloke, i'm not seeing that particular error with my testing. sbeattie ^ ? [18:00] * sbeattie looks [18:01] yeah, that's a problem if true. I've not seen it in my testing, either. [18:02] Odd_Bloke: can you verify that wine and qemu-kvm-extras-static are not installed? [18:02] Checking. [18:04] Neither is installed at instance spin-up, let me wait for the test script to finish and I'll check then as well. [18:06] Does GCE run our kernel or google special sauce? [18:06] * sbeattie hasn't looked at GCE before. [18:06] Our kernel, I believe. [18:07] Neither installed after the test has run. [18:07] sbeattie: You should be able to get to ubuntu@130.211.100.128, if you want to have a poke around. [18:11] Odd_Bloke: this looks to be the culprit: [18:11] /etc/sysctl.d/99-gce.conf:# randomizes addresses of mmap base, heap, stack and VDSO page [18:11] /etc/sysctl.d/99-gce.conf:vm.mmap_min_addr = 0 [18:12] isn't that below the minimum we allow ? [18:12] kees: do you know why GCE would be doing that? [18:13] This is following the recommendation at https://cloud.google.com/compute/docs/tutorials/building-images#kernelbuild [18:13] Which has "CONFIG_DEFAULT_MMAP_MIN_ADDR=65536" and 'echo "vm.mmap_min_addr = 0" > /etc/sysctl.d/mmap_min_addr.conf' as equivalent recommendations. [18:14] Or maybe not equivalent, I don't know enough about sysctl to make that statement. [18:14] Would it make sense for that 0 to be a typo? [18:15] I would hope it's a typo. They're definitely not equivalent. [18:16] OK, I'll run it past someone at Google to try and work out what they mean. [18:17] Or at least, if not a typo, an insufficient explanation of how to disable it for the very few applications that break because it's set to non-zero. [18:20] Odd_Bloke: I'm logged out of your instance. [18:21] sbeattie: Ack; thanks for taking a look. [18:22] Odd_Bloke: sure thing, thanks for raising the issue. [18:38] sbeattie: Have filed an internal issue for it, and subscribed you; should be fixed soon. [18:39] Odd_Bloke: thanks! [18:46] sbeattie: omg, wtf! no, it should NOT be doing that! [18:46] sbeattie: where is that sysctl coming from? did google build that? [18:48] kees: I think the thing to get fixed is the MMAP_MIN_ADDR section in https://cloud.google.com/compute/docs/tutorials/building-images#kernelbuild [19:10] Hello! I'm trying to boot an Ubuntu Snappy image off a thumb drive. I seem to be getting some kernel panic and a reboot during boot. How do I dig deeper / work around it? [19:10] I can't clearly see the panic because it reboots... [19:10] And there isn't a grub menu so I can modify kernel parameters [19:10] Though I can edit the thumb drive manually, I'm less familiar with that bit [19:11] mterry, hmmm well i'd suggest taking pictures, and depending on the board does it have one of those header serial ports [19:11] either of those might get you the panic and a clue [19:11] what are you booting it on as well ? [19:11] apw, I'm just trying to boot on my laptop [19:12] hmmm [19:12] apw, I can't take a picture, because the screen disappears so quickly [19:12] I suppose I could take a movie and take a frame [19:12] a video might get it if you are lucky [19:12] deperate times and all that [19:12] * mterry tries out Ubuntu Touch's video capability [19:12] have you asked in #snappy in case someone else is seeing it ? [19:12] apw, I sent quick email to the snappy list, no replies yet [19:13] apw, I don't think this is commonly done yet in snappy world? I think most people just use beagleboards or kvm [19:14] mterry, right that would be my expectation too ... so i am not supprised its not perfect [19:14] apw, asked on #snappy in case I get lucky. Meanwhile will try to get a pic [19:14] apw, thanks for help so far! :) [19:21] apw, https://chinstrap.canonical.com/~mterry/snappy-panic.png -- snappy-devel just said "Aren't we using just the generic kernel pkg on x86? If so, you are missing all the modules from the generic-extra pkg, thus my bet is on a missing module" [19:22] Does that panic look like that makes sense? If so, I guess my next step is wondering how to get snappy to let me install those modules [19:23] so that looks like systemd exited [19:26] apw, yeah, says "attempted to kill init" [19:26] mterry: That's failing to pivot out of the initrd. [19:26] * mterry is super unfamiliar with debugging systemd [19:27] mterry: systemd isn't even in play yet. [19:27] ok [19:27] mterry: mount the img, download the linux-image-extra pkg corresponding to the kernel used in that img [19:27] mterry: and install it [19:27] mterry: Where's the image you're booting? [19:28] mterry: IOW, boot into ubuntu, plug the usb stick, chroot in it, install the pkg, etc [19:28] Also, how on earth is that taking 3 seconds to get out of the initrd? What evil is snappy perpetrating? [19:28] ppisati, well installing on snappy via apt, even using /usr/bin/apt directly, has been unsuccessful for me. read-only mounts and such. I guess I can just remount them all rw [19:29] mterry: yep, and apt is a symlink but you can find the original apt in /usr/local [19:29] infinity, http://cdimage.ubuntu.com/ubuntu-core/preview/ubuntu-core-alpha-02_amd64-virt.img is a qcow2 file that I'm booting [19:29] ppisati, I think you have that reveresd [19:32] infinity, I can't speak to the evils of snappy yet ;) [19:32] * ppisati just bought an xps 13 [19:32] lovely [19:33] ppisati, they are good laptops besides this oddity! :) I also tried booting on another Dell I have here, same result [19:34] mterry: ping lool, i know he did some work for snappy and real x86 hw [19:34] mterry: Right, so that boots fine in kvm. I'm sure it's failing to find your USB controller due to the fact that someone decided linux-image-virtual was good enough. [19:34] mterry: s/linux-image-virtual/linux-image-generic/ in the build scripts would likely make it happy for you. [19:35] Though the -virt in the filename sort if implies it's not meant for real hardware. [19:36] The tarball builds have the full kernel in them. [19:36] I'm guessing this "preview" is old? [19:36] infinity, maybe? It's still the one the snappy site recommends [19:37] Well, it's 3.18.0-9, so it's not new. [19:38] infinity, maybe I need to find out how to build my own new preview then, with -generic. Thanks for the pointers! [19:41] mterry: FWIW, snappy switched from -virtual to -generic on Jan 28, and that kernel is from Jan 12. [19:41] mterry: So, the image is just old and predates the fix. [19:41] mterry: Sort out how to get someone to make a fresh one, and you'll likely be set (or find instructions on how to turn the daily tarball into a bootable image) [19:42] * mterry hugs infinity for his detective work [20:26] infinity, nice thanks [21:23] sbeattie: yikes [21:23] sbeattie: I'll get them to fix that. [21:23] sbeattie: who did the sysctl file in the ubuntu image? was that someone else reading the "documentation" or was that generated by someone in google? [21:24] Odd_Bloke: do you know the answer to kees' question? === FreezingAlt is now known as FreezingCold [22:31] sbeattie: fire started, docs should be changing shortly, and when cansecwest is over, existing images will likely get reviewed. [23:29] kees: w00t, thanks!