Bug 215918
|
Description
Daniel Calcoen
2022-04-29 09:14:59 UTC
my hardware : laptop Asus N751JK Network controller: Qualcomm Atheros AR9462 Wireless Network Adapter (rev 01) Subsystem: AzureWave AR9462 Wireless Network Adapter I was runing Ubuntu 20.10, 21.04 up to Ubuntu 21.10 with Kernel 5.13.0 without any problem. When migrated to Ubuntu 22.04 which installs Kernel 5.15 then the problems started. I upgraded and tested unsuccessfully the following Kernels : 5.15.28, 5.15.36, 5.16.14, 5.16.20, 5.17.4, 5.17.5 I disabled the power save as explained at https://askubuntu.com/questions/1403773/22-04-wifi-keeps-disconnecting-for-a-few-seconds-frequently/1404471#1404471 without improvement in any of the Kernels tested. Also the "intel_iommu=off" mentioned at https://groups.google.com/g/linux.debian.kernel/c/RFpPIp0cncA/m/v9ELDxLICgAJ Bug#994590 didn't help. I downgraded and currently running without any problem using the Kernel 5.13.19-051319-generic. There is no problem in my airpoint (router) nor my internet access, I have other different computers and devices working correctly on wifi. In addition my N751JK works correctly when using the Ethernet adapter for the physical cable. Note: I use Ubuntu Mainline Kernel Installer to install/remove the kernels https://github.com/bkw777/mainline Ubuntu Mainline Kernel Installer fetchs from kernel.ubuntu.com I'm not one of the maintainers that needs to take care of this, but FWIW: it afaics would be really helpful to check if this was introduced between 5.13 (some would call it 5.13.0) and 5.15 or between 5.15 and 5.15.28 Created attachment 300848 [details]
installation of kernel 5.17.5 log, show compiling switches used
installation of kernel 5.17.5 log, show compiling switches used https://bugzilla.kernel.org/attachment.cgi?id=300848 result of sudo dbus-send --system --print-reply --dest=fi.w1.wpa_supplicant1 \ /fi/w1/wpa_supplicant1 \ org.freedesktop.DBus.Properties.Set \ string:fi.w1.wpa_supplicant1 \ string:DebugLevel \ variant:string:"msgdump" >wpa_supplicant.log method return time=1651226558.254675 sender=:1.5 -> destination=:1.146 serial=166 reply_serial=2 method return time=1651226649.835958 sender=:1.5 -> destination=:1.149 serial=195 reply_serial=2 Created attachment 300849 [details]
dmesg for Kernel 5.17.5
the problem appears under heavy load for the wifi, i.e. when downloading a big file over the net at the dmesg or Kernel 5.17.5 https://bugzilla.kernel.org/attachment.cgi?id=300849 1)boot 2)I downloaded a 80Mb file, problem appears 3)turn off Wi-FI 4)turn on WI-FI the problem persist (without heavy load over the WI-FI) (In reply to The Linux kernel's regression tracker (Thorsten Leemhuis) from comment #4) > I'm not one of the maintainers that needs to take care of this, but FWIW: it > afaics would be really helpful to check if this was introduced between 5.13 > (some would call it 5.13.0) and 5.15 or between 5.15 and 5.15.28 Which is your advice ? should I start testing from 5.15.0 onward, marking if the problem is present ? which should be the log information you wish to have ? these days I have the time and the will to help, so just ask :) (In reply to Daniel Calcoen from comment #10) > > Which is your advice ? Check if 5.15.0 is affected, as we already known that 5.13.y is working and 5.15.28 is not. Depending on the outcome a git bisection between 5.13 and 5.15 or 5.15 and 5.15.28 would be helpful, unless the maintainer has an idea what might be wrong here. Created attachment 300851 [details]
dmesg for Kernel 5.15.5
Created attachment 300852 [details]
installation of kernel 5.15.5 log, show compiling switches used
the kernel 5.15.5 has the problem result for the wpa_supplicant.log method return time=1651239579.077019 sender=:1.4 -> destination=:1.164 serial=342 reply_serial=2 dmesg for Kernel 5.15.5 https://bugzilla.kernel.org/attachment.cgi?id=300851 installation of kernel 5.15.5 log, show compiling switches used https://bugzilla.kernel.org/attachment.cgi?id=300852 Created attachment 300853 [details]
dmesg for Kernel 5.15.0
Created attachment 300854 [details]
installation of kernel 5.15.0 log, show compiling switches used
the Kernel 5.15.0 works so far ! result for the wpa_supplicant.log method return time=1651238101.785291 sender=:1.1 -> destination=:1.144 serial=52 reply_serial=2 dmesg for Kernel 5.15.0 https://bugzilla.kernel.org/attachment.cgi?id=300853 installation of kernel 5.15.0 log, show compiling switches used https://bugzilla.kernel.org/attachment.cgi?id=300854 Created attachment 300855 [details]
dmesg for Kernel 5.15.4
Created attachment 300856 [details]
installation of kernel 5.15.4 log, show compiling switches used
the Kernel 5.15.4 also works for both 5.15.0 and 5.15.4 I also checked to stop the WI-FI and start again and both succeed, for the other cases doing that normally make the problem worse for completeness result for the wpa_supplicant.log method return time=1651241954.655624 sender=:1.10 -> destination=:1.163 serial=106 reply_serial=2 dmesg for Kernel 5.15.4 https://bugzilla.kernel.org/attachment.cgi?id=300855 installation of kernel 5.15.4 log, show compiling switches used https://bugzilla.kernel.org/attachment.cgi?id=300856 seems to be something between 5.15.4 and 5.15.5 that is also propagates to 5.16.x and 5.17.x as suggested I'll continue at Ubuntu distribution place https://bugs.launchpad.net/ubuntu/+source/kernel-package/+bug/1970965 this ticket can be closed Created attachment 300857 [details]
screen shot advising for using bugzilla.kernel.org
the screen shot form the collect tool advises to publish in this site as I'm using a mainline kernel https://bugzilla.kernel.org/attachment.cgi?id=300857 FWIW, all the Ubuntuisms here make it hard to follow for me. (In reply to Daniel Calcoen from comment #21) > seems to be something between 5.15.4 and 5.15.5 that is also propagates to > 5.16.x and 5.17.x So just to understand things properly: your problem occurs with vanilla 5.15 and started to happen with 5.15.5; it's also still present in the lastet kernel? Could you maybe give 5.18-rc5 a quick try? I use Ubuntu Mainline Kernel Installer to install/remove the kernels https://github.com/bkw777/mainline which fetches the files from kernel.ubuntu.com actually this app list, the newest kernel shown is the 5.17.5 I'll try 5.18-rc5 this night (just give me a bit of time because I need to install it manually :( ) another person has confirmed the same behavior at https://bugs.launchpad.net/ubuntu/+source/kernel-package/+bug/1970965 with not identical but similar hardware > actually at this app list the newest kernel shown is the 5.17.5
>
> I'll try 5.18-rc5 this night (just give me a bit of time because I need to
> install it manually :( )
my bad, I un-hide the RCs and the 5.18.0-rc5 is there
I'll test it in an hour and let you know
Created attachment 300891 [details]
installation of kernel 5.18.0rc5 log, show compiling switches used (unsuccessful)
installation of kernel 5.18.0rc5 log, show compiling switches used
unsuccessful installation
Couldn't install the 5.18.0rc5 at the 1st try unset ARCH; [ ! -h /usr/bin/cc ] && export CC=/usr/bin/gcc; env NV_VERBOSE=1 'make' -j8 NV_EXCLUDE_BUILD_MODULES='' KERNEL_UNAME=5.18.0-051800rc5-generic IGNORE_XEN_PRESENCE=1 IGNORE_CC_MISMATCH=1 SYSSRC=/lib/modules/5.18.0-051800rc5-generic/build LD=/usr/bin/ld.bfd modules........(bad exit status: 2) ERROR (dkms apport): kernel package linux-headers-5.18.0-051800rc5-generic is not supported Error! Bad return status for module build on kernel: 5.18.0-051800rc5-generic (x86_64) Consult /var/lib/dkms/nvidia/510.60.02/build/make.log for more information. Created attachment 300892 [details]
NVIDIA make log
NVIDIA make log
I'm sorry I need to solve the errors with the NVIDIA or desinstall it before testing the 5.18.0rc5 (In reply to Daniel Calcoen from comment #32) > desinstall it before testing the 5.18.0rc5 Do that, most kernel developers don't care about reports with nvidia's driver loaded, as it might be the reason why something malfunctions I desinstalled the NVIDIA drivers and remove all NVIDIA packages and files running with xorg driver in x11 (just to confirm) Created attachment 300893 [details]
installation of kernel 5.18.0rc5 log, show compiling switches used
installation of kernel 5.18.0rc5 log, show compiling switches used
this time the installation of 5.18.0rc5 went ok (also no compilation of the modules as NVIDIA is not anymore there) Created attachment 300894 [details]
dmesg for Kernel 5.18.0rc5
dmesg for Kernel 5.18.0rc5
the problem is still there with 5.18.0.rc5 I'm back on Linux ubuntu 5.15.4-051504-generic #202111211331 SMP Sun Nov 21 13:53:10 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux this time without NVIDIA and again everything works fine Few tips how to solve this: Stop using Ubuntu built kernels and Ubuntu specific tools. This is kernel.org bugzilla, we only support kernels retrieved from kernel.org which users have built themselves. We don't know what distros do so we can't support distro kernels. Don't use out of tree drivers like nvidia or virtualbox. Same comment as above, we don't know what they do. For testing use the linux-stable tree and build the kernel yourself: git clone https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git If I'm understanding correctly v5.15.4 works and v5.15.5 is broken. First verify that v5.15.4 really works without problems by checking out the release and building the kernel yourself: git checkout v5.15.4 Then do the same for v5.15.5 and make sure you see the bug: git checkout v5.15.5 If you are 100% sure (this is important!) that v5.15.4 works and v5.15.5 is broken you can start a bisect search to find the commit which broke this: git bisect start git bisect bad v5.15.5 git bisect good v5.15.4 Again you have to be 100% sure before marking a commit good or bad. One mistake and the bisect is useless. After you have found the commit which broke this you can verify that by reverting the commit: git checkout v5.15.15 git revert <commitid> More info: https://www.kernel.org/doc/html/latest/admin-guide/bug-bisect.html (In reply to Kalle Valo from comment #40) > Again you have to be 100% sure before marking a commit good or bad. One > mistake and the bisect is useless. After you have found the commit which > broke this you can verify that by reverting the commit: > > git checkout v5.15.15 > git revert <commitid> Oops, a typo. This should be: git checkout v5.15.5 git revert <commitid> Understood, I'll proceed following your advice. In parallel, while testing each kernel, I wish to profit and collect more information related to the wi-fi, which are the logs related to wi-fi modules/driver/function that I can produce ? (because the output obtained from dmesg didn't provide too much help in this case, or I don't know how to interpret it correctly...) Unfortunately with my limited knowledge I was not able to bisect the Ubuntu Kernel version but I installed then Mainline Kernel 5.17.11 and the problem has disappeared https://askubuntu.com/questions/1411565/which-repository-should-i-use-to-build-ubuntu-kernel-and-how-should-i-build-to-b |