| Kon- | Hi, I just noticed a 40% drop in performance between Bionic's 4.18 and 5.0 kernels in Mesa's RADV. Does the 5.0 kernel have all the same patches? How can I troubleshoot this? | 05:52 |
|---|---|---|
| sixwheeledbeast | Having trouble reporting my issue with ubuntu-bug linux "Problem in linux-image-4.15.0-55-generic" | 09:29 |
| sixwheeledbeast | Sorry I forgot i updated to disco | 10:13 |
| sixwheeledbeast | bug #1839707 | 10:13 |
| ubot5 | bug 1839707 in linux (Ubuntu) "System does not shutdown since upgrade from 16.04LTS" [Undecided,New] https://launchpad.net/bugs/1839707 | 10:13 |
| TJ- | sixwheeledbeast: in case it is related there were systemd bugs causing that which took some time to be solved; not sure if your system is affected by that because I cannot recall now whether those patches reached 18.04 so far | 10:41 |
| sixwheeledbeast | Ok I may try a devuan live image to rule that out when I get a minute. | 10:43 |
| TJ- | sixwheeledbeast: According to your bug report detail, the first sentence is wrong. It says upgrade was to 18.04 LTS but the detail shows "UpgradeStatus: Upgraded to disco on 2019-07-31 (9 days ago)" | 10:54 |
| sixwheeledbeast | Also I tried poweroff -f which causes the system to freeze, see my comments re 16.04 > 18.04 > 19.04 | 10:55 |
| TJ- | Because you seem to be suffering a kernel version regression I'd recommend altering the bug title to something like "System does not shutdown with any kernel >= v4.6.3" | 10:56 |
| TJ- | sixwheeledbeast: I'd also suggest trying this workaround for some ACPI issues that may be responsible: https://iam.tj/prototype/enhancements/Windows-acpi_osi.html | 10:58 |
| sixwheeledbeast | My plan was to stick to LTS released but will having issues I have tried many things to resolve. This includes many combinations of kernel flags including osi noapic nolapic etc | 10:59 |
| sixwheeledbeast | Above link was mentioned in #ubuntu-mate, I tried that last weekend | 11:00 |
| TJ- | sixwheeledbeast: you should mention that in your report, that acpi_osi workarounds didn't help | 11:04 |
| sixwheeledbeast | ok | 11:04 |
| TJ- | sixwheeledbeast: a major test you can do to find out if it is kernel/hardware rather than systemd/userspace, is - when the hang occurs - try the magic sysreq hotkeys the kernel listens for. Press and hold Alt+SysRq , release SysRq, then in turn press R E I U S B - at this point the system should reboot. | 11:08 |
| TJ- | sixwheeledbeast: those keys are the Magic SysRq commands, some of which may be disabled but the effect is to kill remaining tasks, sync RAM buffers to disk, and reboot | 11:09 |
| TJ- | sixwheeledbeast: if that works then it is highly likely the issue is due to some unit dependency issue in systemd | 11:10 |
| sixwheeledbeast | how I miss upstart ... | 11:11 |
| Kon- | What's the proper place to report regressions with the Bionic kernel? tbh Launchpad's layout is a little confusing | 11:11 |
| TJ- | !bug | Kon-: report against package 'linux' | 11:12 |
| ubot5 | Kon-: report against package 'linux': If you find a bug in Ubuntu or any of its official !flavors, please report it using the command « ubuntu-bug <package> » - See https://help.ubuntu.com/community/ReportingBugs for other ways to report bugs. | 11:12 |
| Kon- | Thanks | 11:13 |
| lotuspsychje | Kon-: also describe wich kernels you tested in your description ok | 11:13 |
| sixwheeledbeast | Kon-: Have you tried mainline 5.2? | 11:17 |
| sixwheeledbeast | REIUSB doesn't nothing at this point BTW | 11:19 |
| TJ- | sixwheeledbeast: there are quite a few shutdown related bugs in systemd, see https://bugs.launchpad.net/ubuntu/+source/systemd/?field.searchtext=shutdown&search=Search | 11:19 |
| TJ- | sixwheeledbeast: is magic sysrq doesn't cause a reboot that suggests an ACPI issue | 11:20 |
| TJ- | s/is/if/ grr | 11:20 |
| TJ- | sixwheeledbeast: there are 88 commits v4.6.2..v4.6.3 of which many are not applicable to the x86_64 architecture and several are ALSA which would be highly unlikely to be a cause so it should be relatively easy to bisect to find the problem | 11:24 |
| TJ- | sixwheeledbeast: I have it down to 54 possibles now | 11:28 |
| sixwheeledbeast | :) | 11:35 |
| TJ- | now 49 but many are network-related and unlikely to be used by the hardware listed in your bug report | 11:41 |
| TJ- | sixwheeledbeast: this is what I have: http://paste.ubuntu.com/p/ZSXQWhyGnx/ | 11:44 |
| TJ- | sixwheeledbeast: Am I correct in deducing you've got a RAID mirror there? | 11:46 |
| sixwheeledbeast | not that know of | 11:46 |
| TJ- | sixwheeledbeast: hmmm, i wonder why dm_mirror module was loaded then | 11:47 |
| sixwheeledbeast | it's a laptop with one spinning disk | 11:47 |
| TJ- | sixwheeledbeast: That's what I thought but that doesn't preclude using such things even on a single disk :) | 11:47 |
| TJ- | sixwheeledbeast: I asked because there are some known issues with some fake-RAID configs | 11:48 |
| sixwheeledbeast | ok | 11:48 |
| TJ- | sixwheeledbeast: does this happen immediately? As in, boot and shutdown without a user log-in, and/or boot, user-login, immediate shutdown (without running applications) | 11:49 |
| TJ- | sixwheeledbeast: I'm trying to deduce if some applications could trigger some state that causes this | 11:49 |
| lotuspsychje | TJ- sixwheeledbeast there's a 2007 bios update for his machine so it seems, maybe worth a shot? https://www.dell.com/support/home/uk/en/ukbsdt1/product-support/product/inspiron-1501/drivers | 11:51 |
| sixwheeledbeast | all of the above I will just confirm the first it's a little while since i tried | 11:51 |
| lotuspsychje | DMI: Dell Inc. Inspiron 1501 /0UW744, BIOS 2.6.1 08/23/2006 | 11:52 |
| sixwheeledbeast | Locutusofborg: I have tried that on the other laptop with the same issue, its only installable with a bootable Vista installation | 11:52 |
| sixwheeledbeast | oop sorry wrong highlight | 11:53 |
| sixwheeledbeast | lotuspsychje: ^ | 11:53 |
| lotuspsychje | sixwheeledbeast: just trying to widen your options | 11:54 |
| TJ- | sixwheeledbeast: one thing I've noticed that is suspicious is there is no ALSA sound device detected but there is one in in the lspci list "00:14.2 Audio device [0403]: Advanced Micro Devices, Inc. [AMD/ATI] SBx00 Azalia (Intel HDA) [1002:4383]" - that is worth following up | 11:54 |
| TJ- | sixwheeledbeast: "snd_hda_intel" has an alias for that device [1002:4383] but didn't load | 11:58 |
| sixwheeledbeast | Confirmed that shutting down before log in is the same as both booting and shutting down and booting using a browser and shutting down | 12:00 |
| TJ- | sixwheeledbeast: right, that's good, rules out a lot of possibilities | 12:01 |
| TJ- | sixwheeledbeast: Could you test a <= 4.6.2 kernel and check if the sound device is correctly handled? Specifically that snd_hda_intel and related modules are loaded with "lsmod | grep snd_hda" | 12:02 |
| TJ- | sixwheeledbeast: if it is loaded we likely have a candidate for the problem | 12:02 |
| sixwheeledbeast | BTW 5.0 runs like bag of spanners on this machine high system load and errors in dmesg, much more stable with 4.15 or even 5.2 mainline. I had to use that for apport to work tho. | 12:05 |
| sixwheeledbeast | TJ-: https://paste.ubuntu.com/p/2pSqmfPPKB/ | 12:09 |
| sixwheeledbeast | Will boot 4.6.2 if required? | 12:10 |
| TJ- | sixwheeledbeast: now I'm confused, that's got it loaded in 4.15 but it was missing in 5.0.0-21 - so maybe it can be discounted as the cause | 12:16 |
| TJ- | sixwheeledbeast: when you've tested all these kernel versions was that with the same userspace release (18.04 LTS) ? | 12:16 |
| sixwheeledbeast | TJ- yes, as i say 5.0 was pretty unstable OOTB from installing disco. All current testing is with disco using kernels from ukuu | 12:18 |
| sixwheeledbeast | Only since trying disco have I gone down the route of testing kernels, I used the standard ubuntu shipped versions in ubuntu releases before that | 12:21 |
| sixwheeledbeast | Thats where I found that => 4.6.3 started the issue as I knew 4.4 was fine on xenial 16.04LTS | 12:23 |
| TJ- | sixwheeledbeast: I can't find any 'smoking gun' commits that'd be obvious causes | 12:41 |
| TJ- | sixwheeledbeast: you could try disabling the GUI and adding verbose (systemd) logging in the hopes something useful is left on the console when it hangs | 12:43 |
| TJ- | sixwheeledbeast: that'd be "sudo systemctl set-default multi-user.target" and either via /etc/default/grub's GRUB_CMDLINE_LINUX or manually adding to kernel command-line at boot-time at the GRUB boot menu, "debug" and removing any "quiet splash". When booting to a console (no GUI) use "systemctl poweroff" or "systemctl reboot" | 12:44 |
| sixwheeledbeast | quiet splash has been off since i started with an issue | 12:45 |
| sixwheeledbeast | TJ-: To completely rule out systemd I have just tried a devuan ascii live image which is 4.9. This also freezes at shutdown | 13:44 |
| sixwheeledbeast | "[info] Will now Halt." "ACPI: Preparing to enter system sleep state S5" | 13:46 |
| TJ- | S5 !? | 13:48 |
| TJ- | hmmm, so there is something in the ACPI implementation in the PC's BIOS *or* some setting in BIOS setup that is causing this. Easiest is to check for any possible setup options you can change | 13:50 |
| sixwheeledbeast | There is nothing BIOS is minimal. | 13:51 |
| TJ- | right, most Dells are that way. Have you tested with and without AC power? Sometimes that can cause differences | 13:53 |
| sixwheeledbeast | without AC on shutdown yes its the same | 13:54 |
| TJ- | Must be a bug in the PC's BIOS ACPI... as the kernel progresses it tightens its implementation of the ACPI specifications and we do often see this causing failures on systems where previously it'd boot, but I didn't see anything obvious in the 4.6.2..4.6.3 commits thatd cause that | 13:56 |
| sixwheeledbeast | PhoenixBios Quickboot and switching USBs off is about all the "advanced" options | 13:57 |
| TJ- | :) my Dells have quite a few more options than that. I guess you'll have checked for a more recent BIOS ? | 13:59 |
| sixwheeledbeast | As above there it one version later you need a working Vista install to install and it didnt help on the other machine with the same issue | 14:01 |
| TJ- | OK, so I guess the ACPI errors in the dmesg are the obvious things to focus on. Can you check if you get the same exact ACPI errors in the 4.6.2 and 4.6.3 boots? if so that sort-of rules out the avenue too | 14:02 |
| sixwheeledbeast | The same ACPI message I had shutting down devuan? | 14:04 |
| TJ- | sixwheeledbeast: no, those ACPI errors in the dmesg | 14:23 |
| sixwheeledbeast | TJ- there is a diff let me pastebin | 14:24 |
| TJ- | starting at [ 0.393217] ACPI Error: AE_NOT_FOUND | 14:24 |
| sixwheeledbeast | ACPI: bus type USB registered | 14:25 |
| sixwheeledbeast | https://paste.ubuntu.com/p/Gmnb4FcXn5 | 14:27 |
| sixwheeledbeast | line 147 | 14:28 |
| sixwheeledbeast | actually its the same just a different order now looking at it | 14:29 |
| TJ- | sixwheeledbeast: the issue could be related to https://bugzilla.kernel.org/show_bug.cgi?id=199349 | 14:32 |
| ubot5 | bugzilla.kernel.org bug 199349 in Power-Off "System hangs on shutdown/reboot with newer kernel - Toshiba Satellite A210 (AMD Turion X2, ATI Radeon HD2600)" [High,New] | 14:32 |
| sixwheeledbeast | Look very similar | 14:41 |
| === fling_ is now known as fling | ||
| === jeremy31 is now known as j3r3my31 | ||
| === j3r3my31 is now known as jeremy31 | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!