/srv/irclogs.ubuntu.com/2012/04/23/#ubuntu-kernel.txt

=== kees_ is now known as kees
=== mthaddon` is now known as mthaddon
=== LetoTheII is now known as LetoThe2nd
=== smb` is now known as smb
* smb adjusts coffee level06:56
=== android is now known as jussi
ppisatimoin07:15
dileks_moin moin07:40
dileks_build_kernel-with-deb-pkg.sh: http://paste.debian.net/164589/07:40
dileks_might be interesting for people wanting to test upstream GIT kernels07:41
ohsixthere's already an autobuilder for them07:43
dileks_ohsix: can you give me more infos or a url?08:16
ohsixhttps://wiki.ubuntu.com/Kernel/MainlineBuilds08:17
ohsixthat article is pretty crusty, but it links to t he right places08:17
dileks_ohsix: thanks!08:28
dileks_http://kernel.ubuntu.com/~kernel-ppa/mainline/drm-intel-next/2012-04-22-precise/08:28
dileks_nice :-)08:28
dileks_http://kernel.ubuntu.com/~kernel-ppa/mainline/drm-intel-next/2012-04-22-precise/BUILT08:50
dileks_Release: lucid ? (VS. built for precise?)08:51
ohsixthe wiki article explains how they're built08:52
dileks_ohsix: what do you mean by "pretty crusty"? partially outdated?09:03
dileks_just fyi: CONFIG_FB_VESA is now boolean09:12
dileks_if not set =y ...09:12
dileks_config-3.4.0-997-generic:# CONFIG_FB_BOOT_VESA_SUPPORT is not set09:12
dileks_config-3.4.0-997-generic:# CONFIG_FB_VESA is not set09:12
dileks_this leads here to black screen (unity login screen should appear, after pressing return I get into unity login screen)09:14
dileks_config-3.4.0-rc4-2-generic:CONFIG_FB_BOOT_VESA_SUPPORT=y09:14
dileks_config-3.4.0-rc4-2-generic:CONFIG_FB_VESA=y09:14
dileks_(from my kernel)09:14
apwdileks_, indeed that is telling you the compile environment, the -precise tells you the config environment used09:19
dileks_OK, with release I understood (for) distribution (== precise)09:20
dileks_ohsix: the autobuild-service is really nice, but I tend to use my script for building custom-kernels09:21
dileks_not sure if I read here about it...09:25
dileks_Apr 23 11:08:20 fambox kernel: [    0.294037] pnp 00:0b: can't evaluate _CRS: 1231109:25
dileks_saw this with my 3.4-rc4 and drm-intel-netx_precise-current09:26
ohsixthe wiki article also says they're unsupported, but you are more likely to be able to file a bug with upstream with them09:28
dileks_...will do after a quick look into the sources (and find maintainers with get_maintainer.pl)09:31
ohsixwhat you pasted is a bios problem, i don't remember what method _CRS is, but since 'pnp' is there, and the machine still runs, probably not an important one09:32
dileks_yeah, saw some acpi error or even bugs09:33
dileks_[    0.241572] [Firmware Bug]: ACPI: BIOS _OSI(Linux) query ignored09:34
dileks_[    0.279385]  pci0000:00: ACPI _OSC request failed (AE_ERROR), returned control mask: 0x1d09:34
dileks_[    0.294032] ACPI Error: Invalid/unsupported resource descriptor: Type 0x00 (20120320/utresrc-650)09:34
apwbug #98705209:34
ubot2Launchpad bug 987052 in linux "WARNING: at /build/buildd/linux-3.2.0/kernel/softirq.c:159 local_bh_enable_ip+0x7a/0xa0()" [High,Confirmed] https://launchpad.net/bugs/98705209:34
ckingthe _OSC request failed occurs when bit 1 of the returned buffer from _OSC is set indicating "Platform Firmware was unable to process the request or query." - which is most likely to be a firmware issue09:43
ckingand the invalid/unsupported resource descriptor is another firmware bug, types 1, 2 or 3 are allowed and for some reason it is set as type 0 (invalid)09:50
dileks_https://lkml.org/lkml/2012/4/23/10509:50
ckingdileks_, so the "invalid/unsupported resource descriptor" message is due to extra checking added in commit e0fe0a8d4ed5474261d0ee1452f5d9ae77236958 which was part of the ACPI 5.0 updates09:54
ckingnothing new really, same old buggy firmware09:54
dileks_cking: thanks for your explanations!09:54
dileks_this is a samsung series-5 ultrabook09:55
ckingwhy doesn't that surprise me...09:55
dileks_currently, I even cant find the url for firmware-upgrades09:57
dileks_my bookmark is showing to nirvana09:57
dileks_lol, within a few weeks there is now A04 that why I could not find my A0110:16
dileks_http://www.samsung.com/de/support/model/NP530U3B-A04DE10:17
ckingthat's mad10:29
brendand_apw - during the Q cycle, are you guys going to be backporting the Q kernels to P via the SRU process or some other way?10:39
ppisatibrendand_: you mean, backports?10:39
=== brendand_ is now known as brendand
apwbrendand_, we will be doing LTS backport kernels, i am not sure it is clear how those will be delivered given the other backport plans10:40
apwbut they will exist10:40
brendandapw, so it will be a case of you have to opt-in?10:40
brendandapw, like you do with backports now10:40
apwbrendand, you always have to opt in to the backports as far as i know ... indeed10:40
brendandapw, see i thought this was different, as in the Q kernel would be officially supported10:41
brendandin P10:41
brendandand that you would get it when you do a normal upgrade10:42
apwbrendand, the backports kernels are supported, you still have to opt-in10:42
brendandapw, is there a difference then?10:43
apwbrendand, as far as i know there is not a difference in the model, you will be able to opt-in to the kernels and they will be supported, they may not be delivered in the exact same way10:43
* ppisati -> lunch11:43
Kanohi, could somebody backport efi stub to 3.2?11:49
Kanoalso it should be enabled for 3.3+11:49
Kanoi guess it is a simple backport11:51
dileks_cking: I could upgrade BIOS from 05XK to 06XK, but the pnp/_CRS warning is still there12:00
Kanohi tgardner ,did you test efi stub mode yet?12:01
tgardnerKano, afaik. I think some Dell machines have issues still.12:03
Kanodoes not really matter12:03
Kanobut you can enable that feature12:03
Kanofor 3.3+12:03
Kanofor those who know how to use it12:04
Kanoa backport would be best12:04
tgardnerKano, ain't gonna happen for 12.04. you'll have to wait for the first LTS backport kernel.12:05
Kanook, will try to backport on my own. but how about enabling it for mainline kernel?12:05
Kanoi always have to recompile it even for pure intel systems...12:05
tgardnerKano, I'll check with apw about it.  I'm just looking at Q to see if its enabled.12:06
tgardnerCONFIG_EFI_STUB=y12:06
Kanoyes12:06
Kanojust a tiny bit huger12:07
apwtgardner, i'll put it on my todo12:10
=== jdstrand_ is now known as jdstrand
=== dileks_ is now known as dileks
ckingdileks, that _CRS error code 12311 is 0x3017 -> AE_AML_INVALID_RESOURCE_TYPE - so once again it's buggy resource type info returned when evaluating the _CRS  :-/12:18
=== mjg59` is now known as mjg59
ckingdileks, not sure what device PNP0C02, it seems to map to  "General ID for reserving resources required by Plug and Play motherboard registers." so that's a bit nebulous 13:03
mjg59It's just resources that are used by the platform but not expected to be touched by the OS13:04
mjg59Although sometimes it covers PCI space13:04
mjg59The error's probably harmless13:05
mjg59Although it's probably worth checking to make sure that any valid resources returned in the same call are still parsed13:05
ckingmjg59, ok - so if the firmware is returning a bad _CRS with an invalid resource type it's just too bad, blame it on the firmware and complain to the vendor..13:12
mjg59cking: Yeah, but if it's returning a list of resources then we shouldn't ignore all of them just because one is invalid13:13
mjg59But I haven't checked whether the code does that13:13
Kanoapw: could you at least touch the missing include, that would already help13:13
Kanohttps://bugs.launchpad.net/ubuntu/+source/linux/+bug/94256913:15
ubot2Launchpad bug 942569 in linux "binary headers packages are missing include/generated/compile.h" [Low,In progress]13:15
Kanothe content does not really matter, a 0 size file is enough13:15
Kanowould fix nvidia dkms13:15
apwwe may as well fix nvidia to not include it if a 0 length file is sufficient13:28
apwi would prefer to fix the kernel package right if it is missing files13:28
Kanoapw: it happens when you use u packageing + debian nvidia dkms13:33
Kanothat way it wants that file13:33
Kanoi think one other module wanted it as well but can not remember13:33
Kanou nvidia dkms is not affected13:34
Kanobut that is not compatible with debian kernels13:34
Kanoi dislike to have to 2 differnet dkms.conf depending on the kernel used13:35
Kanodefinitely u packageing is broken. hotfix would be a touch13:35
Kanoapw: what script creates the config for the u mainline builds?13:42
Kanoyou updated 3.4rc4?13:43
Kanoah that has efi stub, nice13:44
Kanoa bit bad that you need to extract rdev from an old util-linux package13:46
Kanowhen you use it for efibootmgr you need it13:47
ckingmjg59, looks like pnpacpi_parse_allocated_resource() uses acpi_walk_resources() which bails out if pnpacpi_allocated_resource() hits an unknown resource type :-/13:48
mjg59cking: That may not be ideal13:55
ckingmjg59, one is kind of screwed at that point, an invalid resource type means one can't figure out what kind of resource it was, hence you can't determine where the next one is, so bailing out is probably the best option rather than trying to scan along the buffer and try and find the next valid resource type data structure. Hence if the firmware got it wrong, too bad14:01
mjg59Ah, true.14:03
mjg59A dump of the ACPI tables might be nice14:03
mjg59Just to see what iasl makes of it14:04
ckingdileks, can you run "sudo acpidump > acpidump.log" and then "pastebinit acpidump.log" and put the URL in the channel14:06
Kanoapw: can you rebuild 3.3.3 with efi stub14:25
Kano3.4rc3 rebuild has it14:25
tgardnerogasawara, I'm working on 3.2.16, OK ?14:37
tgardner'cause I see you've pushed a day-0 branch14:38
tgardneror at least I think it was you14:38
ogasawaratgardner: ack14:38
ogasawaratgardner: I'll do 3.4-rc4 then for Q, assuming you haven't?14:39
tgardnerogasawara, take it away14:39
* dileks has to install pkg which includes acpidump14:39
dileksKano: hi :-). drm-intel-next kernel for precise has no efi-stub activated14:40
Kanoit looks that efi stub needs 2 patches14:40
Kanomaybe 314:41
* ogasawara back in 2014:41
Kanowell the 3rd is only in 3.414:42
* dileks installs acpidump package14:42
dilekscking: does the log contain sensitive data?14:44
Kanohttp://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=history;f=arch/x86/boot/compressed14:44
Kanoshould be simple to use for 3.214:44
ckingdileks, just your ACPI tables - so no sweat over it14:44
dilekscking: http://paste.ubuntu.com/942586/14:46
ckingmjg59, ^^14:47
mjg59dileks: What kind of machine is this?14:47
mjg59raw text needs me to log into launchpad?14:47
dileksah the expert :-)14:47
mjg59Sigh.14:47
dilekssamsung series-5 ultrabook14:48
ckingoops, I forgot that14:48
mjg59No problem14:48
mjg59I need to find my password again for the plumbers CFP14:48
dilekshttp://www.samsung.com/au/consumer/pc-peripherals/notebook-pc/ultra-book/NP530U3B-A01AU-spec14:49
mjg59Ok, _CRS for PCI is dynamically generated14:55
mjg59So there's probably a bug in there14:55
ckingfor PNP0C01 _CRS is dodgy too if IGDS is false14:56
dileksso, its fixable in linux-kernel?14:56
mjg59Probably not trivially15:00
mjg59And if your machine works, probably not worth the effort15:00
* cking notes its one more thing to check before firmware gets released...15:04
tgardnerogasawara, I guess you can rename the Q repo now. Quantal Quetzal15:06
ogra_can you pronounce that please ? 15:07
ogra_:P15:07
ppisati:O15:07
ppisati‘Quantal Quetzal’15:07
mjg59Not to be confused with Quantel Paintbox15:08
* ogra_ grins15:08
ckingat least I can type it w/o too many errors15:13
jsalisburyhttp://www.youtube.com/watch?v=FO9gFAOAPFQ15:13
sforsheemjg59, I'm taking a look at radeon graphics on a macbook pro 8,2, efi boot. It seems the os x driver tries to somehow get the video rom address from EFI. Do you have any idea what it might be doing?15:17
mjg59sforshee: Nope. Only radeon one I have right now is an MBP 1,115:18
mjg59But I could believe it's in an EFI variable somewhere, or possibly one of the systab entries is magic15:18
sforsheethe machine does have a number of config tables we don't use15:18
mjg59sforshee: If there's any way you can get it accessible via ssh, I can play15:18
sforsheemjg59, sure. give me a few minutes15:19
tgardnerapw, are you OK with me rebooting gomeisa? I checked that 'sudo hdparm --fibmap /boot/grub/grub.cfg' is below the 32 bit LBA limit15:36
apwtgardner, go ahead15:43
apwtgardner, go ahead15:43
jsalisburyapw, Comment #20 in bug 922906 says they hit the oops with the test kernel.15:53
ubot2Launchpad bug 922906 in linux "Kernel Oops - BUG: unable to handle kernel NULL pointer dereference at 0000009c; EIP is at __ticket_spin_lock+0x8/0x30" [Medium,In progress] https://launchpad.net/bugs/92290615:53
tgardnerbjf, ayan, ppisati, jsalisbury, henrix: rebooting tangerine for kernel update15:56
jsalisburytgardner, ack15:57
henrixtgardner: ack15:57
bjftgardner: ack15:57
=== mkrufky1 is now known as mkrufky
tgardnerogasawara, is 3.2.0-23.36 gonna be the release kernel on the ISOs ?16:20
ogasawaratgardner: yep16:20
tgardnerogasawara, k16:20
ogasawaratgardner: but we'll do a day-0 upload.  I intend for that to contain all patches from master-next except the v3.2.15 and v3.2.16 stable patches (as I want those to bake and go through QA)16:21
ogasawaratgardner: I've coordinated with the release team to upload the day-0 kernel to precise-proposed tomorrow16:21
tgardnerogasawara, ok, was wondering a bit what you were planning for day 016:22
bjfogasawara: we can package and start precise through the cadence starting next monday16:22
ogasawaratgardner: I've pushed what I intend to be in our day-0 upload to the day-0 branch in precise16:22
ogasawarabjf: ack16:22
ogasawarabjf: after I upload our day-0 I'm thinking I'll hand off precise to you guys16:23
bjfherton, henrix, sconklin ^ precise is "ours" starting this Friday. We'll turn the crank on it next week just like all the others.16:24
sconklinWhee!16:24
henrixbjf: \o/16:24
ogasawarasuch enthusiasm, I like it :)16:25
mjg59sforshee: Well, there does seem to be a ReadEfiRom method in the driver16:25
mjg59sforshee: Let me try to work out what it actually does16:25
sforsheemjg59, yeah, I was looking at that16:28
sforsheeit makes some indirect function call, and I have no idea where it goes from there16:29
* cking drinks the btrfs koolaid16:40
tgardnercking, have you started to gag yet ?16:40
ckingheh, only when I lose all my data16:40
* ppisati -> gym/workout16:51
=== orated_ is now known as orated
dilekshmm17:20
dileks$ cat /etc/debian_version17:21
dilekswheezy/sid17:21
dileksis that intended? (ubuntu precise here)17:21
tgardnerdileks, you're asking in the wrong channel. thats likely a foundations issue17:22
dileksI am trying to add some improvements to scripts/package/builddeb (shipped with linux upstream)17:23
dilekstrying to determine distribution/codename17:23
tgardnerdileks, have you looked at /etc/lsb-release ?17:28
dilekshmm, I could check for it17:29
dileksdunno, if debian has such a file, too17:29
=== tgardner is now known as tgardner-lunch
shadeslayerI'm trying to package the stock ubuntu kernel with a couple of changes and patches, but when I try to set CONFIG_DRM_I915=y in debian.master/config/config.common.ubuntu ... it keeps getting reset to CONFIG_DRM_I915=m during the build17:41
shadeslayerany ideas how to fix?17:41
bjfshadeslayer: take a look at: https://wiki.ubuntu.com/Kernel/BuildYourOwnKernel17:43
* shadeslayer looks17:43
shadeslayerah ok17:44
shadeslayerbjf: btw patching works the same way as in standard debian packages right? ( put them in debian/patches )17:48
bjfshadeslayer: honestly i don't know, i've never done it that way17:51
shadeslayeroh, how have you done it?17:51
bjfshadeslayer: i modify a personal git repo to have the changes i want and i build/package that17:51
shadeslayerok17:51
shadeslayergrrr ... CONFIG_DRM_I915 depends on other stuff that is built as a module, and then that stuff depends on other stuff that is built as a module17:52
shadeslayerherp derp18:01
shadeslayerhttp://paste.kde.org/462242/18:01
shadeslayerand I didn't even change the power-pc config18:01
=== tgardner-lunch is now known as tgardner
Kanoapw: all needed commits to boot 3.2 with efi stub you find there: http://kanotix.com/files/fix/efi/linux-3.2-efi-stub-patcher.sh18:09
Kanojust tested and verified18:09
Kanothere was one patch the might be interesting but did not cleanly apply to boot a 32 bit kernel,but for 64 bit this is definitely enough18:10
=== arun__ is now known as arun_
shadeslayerAre there any junior jobs that I could look at?18:25
shadeslayersomething small to start off18:25
=== hggdh_ is now known as hggdh
oratedCould anyone brief me the difference between ubuntu-11.10-preinstalled-server-armel+omap.img and ubuntu-11.10-preinstalled-desktop-armel+omap.img ? What is the differnce in pacakges? I can see one is desktop, other server ...19:25
bjforated: you'd get a better answer in #ubuntu-arm probably19:33
oratedbjf: Okay, I'll try. Thanks19:34
* tgardner -> EOD20:32
=== lamont` is now known as lamont
miconfrinkHey, I've been looking all over for a comprehensive guide to building "Ubuntized"custom patched  kernels23:43
miconfrinkI'd like to build a Molnars Realtime Kernel with Ubuntu Patches as well23:44
miconfrinkI know I could just patch and make-kpkg 23:44
miconfrinkBut isn't there a cleaner way to segregate the patch from the vanilla source?23:45
miconfrinkAny help would be greatly appreciated :-)23:46
bjfmiconfrink: : take a look at: https://wiki.ubuntu.com/Kernel/BuildYourOwnKernel23:57
bjfmiconfrink: i'd suggest starting with a clone of the tree you are interested in23:57
bjfmiconfrink: you can then create a branch for all your patches if you want them kept separately23:57
miconfrinkbjf: this isn't a vanilla kernel23:58
bjfmiconfrink: you are starting with an ubuntu kernel, correct?23:58
miconfrinkIncorrect23:58
miconfrinkvanilla = from kernel.org23:58
miconfrinkIngo Molnar's patch is quite extensive23:59
miconfrinkwhat I'm trying to figure out is where to put the patch in the debian tree23:59

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!