Moving Linux kernel to 4.19 [Already moved for many months now (in january 2020)]

[jimbojones]

This dwc-otg driver problem is back with 4.19.y, https://github.com/raspberrypi/linux/issues/2450

It does not appear that this commit has made it into 4.19 https://github.com/raspberrypi/linux/co ... 56fa68d57d

It does not apply cleanly, but I think I merged it properly,rebuilt the kernel, and it does resolve the issue.

That’s interesting. I think that’s exactly what I was experiencing (and mentioned in my above post) with the arm64 rpi-4.19.y branch on my 3B+. The kernel panics where intermittent and sometimes the device could restart ok, mostly not.

Should you/I report this on github as well for the 4.19.y branch?

[paul433]

jimbojones,

Normally during these kernel upgrades, the rpi folks monitor this thread, I don't think we need to create an issue, unless there is something new to report. One thing to verify your issue, put "pause_on_oops=30" without the quotes on the kernel command line. Verify you seeing "Bad mode in data abort handler detected" If you are not seeing that oops, then your problem is something else.

Paul

[dom]

This dwc-otg driver problem is back with 4.19.y, https://github.com/raspberrypi/linux/issues/2450

It does not appear that this commit has made it into 4.19 https://github.com/raspberrypi/linux/co ... 56fa68d57d

It does not apply cleanly, but I think I merged it properly,rebuilt the kernel, and it does resolve the issue.

Thanks for report. I've updated to 4.19 to include the missing commit and rebuilt rpi-update kernel. Can you confirm if problem is fixed?

[paul433]

This dwc-otg driver problem is back with 4.19.y, https://github.com/raspberrypi/linux/issues/2450

Thanks for report. I've updated to 4.19 to include the missing commit and rebuilt rpi-update kernel. Can you confirm if problem is fixed?

This mostly occurs with custom kernels using initramfs, but yes, I'll rebuild my kernels to validate.

One other thing I've found.... lirc_rpi does not appear to have been ported to 4.19 yet. /driver/staging/media/lirc/lirc_rpi.ko The main lirc module has moved from staging to media/rc

[dom]

One other thing I've found.... lirc_rpi does not appear to have been ported to 4.19 yet. /driver/staging/media/lirc/lirc_rpi.ko The main lirc module has moved from staging to media/rc

I believe lirc_rpi is no longer recommended and gpio-ir should be used instead. Some info here:
viewtopic.php?t=205415

[paul433]

Thanks Dom, I missed that in my searches.

Should probably remove the old overlay. https://github.com/raspberrypi/linux/bl ... verlay.dts

[paul433]

This dwc-otg driver problem is back with 4.19.y, https://github.com/raspberrypi/linux/issues/2450

Thanks for report. I've updated to 4.19 to include the missing commit and rebuilt rpi-update kernel. Can you confirm if problem is fixed?

Built from latest git, and everything seems okay.

[dom]

Built from latest git, and everything seems okay.

Thanks for testing.

[cjan]

my rootfs is on SSD, after 4.19.3 R/W become halt about 1 or 2 sec.

[dom]

my rootfs is on SSD, after 4.19.3 R/W become halt about 1 or 2 sec.

Does reverting to 4.19.2 resolve the issue?
Do you get a kernel panic when it halts? Can you post the output?

[cjan]

my rootfs is on SSD, after 4.19.3 R/W become halt about 1 or 2 sec.

Does reverting to 4.19.2 resolve the issue?
Do you get a kernel panic when it halts? Can you post the output?

not 100% recall 4.19.2 was OK or not? and back to 4.14.
it just freeze 1 or 2 sec and still alive.
and found this, don't know does it cause problem or not?
https://www.phoronix.com/scan.php?page= ... Corruption

[dom]

and found this, don't know does it cause problem or not?
https://www.phoronix.com/scan.php?page= ... Corruption

I've been reading around this and I believe it only affects users with no scheduler set.
I believe the default raspbian image uses the deadline scheduler and so shouldn't be affected by bug.

But I've cherry-picked the upstream fix for this into 4.19.7 and pushed it to rpi-update.

[jimbojones]

Having finally managed to attach a display to my 3B+, I think that the kernel panics I’ve been seeing on startup are to do with the lan78xx driver, probably something related to this https://github.com/raspberrypi/linux/issues/2608.

I can’t reproduce the bug reported there exactly once the pi has successfully booted, but I am experiencing a kernel panic during startup in about 3/5 of all startups, and the messages all have to do with lan78xx. Have those patches previously applied to the 4.14 branch been applied to the the 4.19 branch?

[dom]

I can’t reproduce the bug reported there exactly once the pi has successfully booted, but I am experiencing a kernel panic during startup in about 3/5 of all startups, and the messages all have to do with lan78xx. Have those patches previously applied to the 4.14 branch been applied to the the 4.19 branch?

The patch that fixed that github issue is present in 4.19 tree.

[jimbojones]

I can’t reproduce the bug reported there exactly once the pi has successfully booted, but I am experiencing a kernel panic during startup in about 3/5 of all startups, and the messages all have to do with lan78xx. Have those patches previously applied to the 4.14 branch been applied to the the 4.19 branch?

The patch that fixed that github issue is present in 4.19 tree.

Interesting. Well I don’t know then. I am experiencing kernel panics during boot on 4.19 that I do not on 4.14.
How does one upload the output from dmesg during/after a kernel panic? I have taken pictures of the messages that happen during the panic, and as far as I can tell they are because of a problem with the lan78xx driver, but I could well be wrong. I can upload some of the pictures of these messages if you think this would be helpful, but obviously the best thing to do though would be to upload copies of the logs themselves.

[incognitum]

How does one upload the output from dmesg during/after a kernel panic?

On a kernel panic everything stops, and no logs will be written to storage.

If you have a serial console cable (like: https://www.aliexpress.com/item/USB-To- ... 99599.html ), you may be able to capture the console output on your normal computer.
But if not, photos do are most practical.

[allfox]

Greetings. I'd like to report that I meet a bug in the apt-get 4.14 kernel: https://bugzilla.kernel.org/show_bug.cgi?id=201483

The fix is in https://www.spinics.net/lists/netdev/msg530560.html and already applied into the mainline kernel.

Currently, a rpi-update 4.14 kernel would fix the problem.

[letifosiferrari]

Hi all,

First of all, happy new year!

A DVB driver for the newest (2018) Astrometa DVB-T2 USB dongle has recently been merged to Linus' tree. The commit is https://github.com/torvalds/linux/commi ... 9706c593f4.

Is there any chance that this would be backported to the Raspberry Pi 4.19 (and 4.14) branch?

Thank you!

[Dygear]

Thanks mainly to Marek Wyborski (Robotrovsky) who wrote the bulk of the code and Sean O'Brien of Google's Chrome OS team for pushing it into the Linux Kernel HID driver, there is now Apple Magic Trackpad 2 in tree support. While the pressure support was removed it still does things like two finger scroll, and I think two finger tap produces a right click, and that's quite a bit more useable than the current setup. It's currently in the 4.20 release of the kernel. Is there any chance of a back port for this into the Raspberry Pi kernel? It would be massively more helpful for me as I use these keyboards and mouse pretty much exclusively.

HID: magicmouse: add support for Apple Magic Trackpad 2 (9d7b186)

Code: Select all

pi@raspberrypi:~ $ git clone https://github.com/robotrovsky/Linux-Magic-Trackpad-2-Driver.git
Cloning into 'Linux-Magic-Trackpad-2-Driver'...
remote: Enumerating objects: 265, done.
remote: Total 265 (delta 0), reused 0 (delta 0), pack-reused 265
Receiving objects: 100% (265/265), 73.04 KiB | 0 bytes/s, done.
Resolving deltas: 100% (70/70), done.
pi@raspberrypi:~ $ cd Linux-Magic-Trackpad-2-Driver/
pi@raspberrypi:~/Linux-Magic-Trackpad-2-Driver $ cd scripts/
pi@raspberrypi:~/Linux-Magic-Trackpad-2-Driver/scripts $ chmod +x install.sh 
pi@raspberrypi:~/Linux-Magic-Trackpad-2-Driver/scripts $ sudo apt install raspberrypi-kernel-headers
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following NEW packages will be installed:
  raspberrypi-kernel-headers
0 upgraded, 1 newly installed, 0 to remove and 0 not upgraded.
Need to get 16.1 MB of archives.
After this operation, 104 MB of additional disk space will be used.
Get:1 http://archive.raspberrypi.org/debian stretch/main armhf raspberrypi-kernel-headers armhf 1.20181112-1 [16.1 MB]
Fetched 16.1 MB in 2s (6,405 kB/s)                     
Selecting previously unselected package raspberrypi-kernel-headers.
(Reading database ... 86802 files and directories currently installed.)
Preparing to unpack .../raspberrypi-kernel-headers_1.20181112-1_armhf.deb ...
Unpacking raspberrypi-kernel-headers (1.20181112-1) ...
Setting up raspberrypi-kernel-headers (1.20181112-1) ...
Error: Timeout was reached
pi@raspberrypi:~/Linux-Magic-Trackpad-2-Driver/scripts $ sudo apt install dkms
Reading package lists... Done
Building dependency tree       
Reading state information... Done
Suggested packages:
  python3-apport menu
The following NEW packages will be installed:
  dkms
0 upgraded, 1 newly installed, 0 to remove and 0 not upgraded.
Need to get 74.8 kB of archives.
After this operation, 289 kB of additional disk space will be used.
Get:1 http://raspbian-us.ngc292.space/raspbian stretch/main armhf dkms all 2.3-2 [74.8 kB]
Fetched 74.8 kB in 0s (75.0 kB/s)
Selecting previously unselected package dkms.
(Reading database ... 119233 files and directories currently installed.)
Preparing to unpack .../archives/dkms_2.3-2_all.deb ...
Unpacking dkms (2.3-2) ...
Setting up dkms (2.3-2) ...
Processing triggers for man-db (2.7.6.1-2) ...
pi@raspberrypi:~/Linux-Magic-Trackpad-2-Driver/scripts $ sudo ./install.sh 
+ dkms_name=hid-magicmouse-dkms
+ dkms_version=4.18+magictrackpad2
+++ dirname ./install.sh
++ cd .
++ pwd
+ DIR=/home/pi/Linux-Magic-Trackpad-2-Driver/scripts
+ dkms status -m hid-magicmouse-dkms -v 4.18+magictrackpad2
+ egrep '(added|built|installed)'
+ dkms add /home/pi/Linux-Magic-Trackpad-2-Driver/scripts/../linux/drivers/hid

Creating symlink /var/lib/dkms/hid-magicmouse-dkms/4.18+magictrackpad2/source ->
                 /usr/src/hid-magicmouse-dkms-4.18+magictrackpad2

DKMS: add completed.
+ egrep '(built|installed)'
+ dkms status -m hid-magicmouse-dkms -v 4.18+magictrackpad2
+ dkms build hid-magicmouse-dkms/4.18+magictrackpad2
Error! echo
Your kernel headers for kernel 4.19.13-v7+ cannot be found at
/lib/modules/4.19.13-v7+/build or /lib/modules/4.19.13-v7+/source.
pi@raspberrypi:~/Linux-Magic-Trackpad-2-Driver/scripts $ uname -a
Linux raspberrypi 4.19.13-v7+ #1186 SMP Tue Jan 1 11:32:58 GMT 2019 armv7l GNU/Linux

Looks like I can't side load the module, as I don't have the kernel source code for this version of the kernel. Can I just ...

Code: Select all

$ cd /lib/modules/4.19.14-v7+/source/
$ git clone https://github.com/raspberrypi/linux.git

[dom]

In general we are not keen on backporting commits for new devices not supported by current kernel.
That is really a decision for the upstream kernel maintainers.

It's not something we can test and sometimes new drivers rely on other new kernel features so it may not be as simple as a single cherry pick.

Best bet would be to compile your own kernel, cherry pick the desired commits and test that it works with your hardware.
If it does then a github pull request can be made where including this can be discussed.

[Dygear]

I happy to do that, I just don't know how to do that. lol.

[dom]

I happy to do that, I just don't know how to do that. lol.

First step build your own kernel: https://www.raspberrypi.org/documentati ... uilding.md

[fruitoftheloom]

I happy to do that, I just don't know how to do that. lol.

First step build your own kernel: https://www.raspberrypi.org/documentati ... uilding.md

https://www.kernel.org

4.19.15 Linux Kernel now states Long Term 13/01/2019

[kozman]

I happy to do that, I just don't know how to do that. lol.

First step build your own kernel: https://www.raspberrypi.org/documentati ... uilding.md

https://www.kernel.org

4.19.15 Linux Kernel now states Long Term 13/01/2019

Yup, and that's a good thing since Debian is using the 4.19 LTS kernel for Buster. I've been playing with 4.19 on a Pi 3 and it is noticeably snappier than even later 4.14.x kernels. Can't wait to see how well a version Raspbian built off a final version of Buster runs. I hope a lot of what's held the Pi back will be helped by Buster.

[Zofiel]

Hi team,

Do you recommend to install kernel 4.19 now? Is It stable?

Can I back to kernel 4.14 simply running a rpi-update without BRANCH=next?

thanks in advance