SARPi kernel 5.14.x update and root filesystem detection

Exaga · 10-03-2021, 05:10 AM

If you weren't already aware, the SARPi installer and system pkgs are now using kernel 5.14.x version.

https://sarpi.penthux.net/index.php?p=downloads

When installing the kernel_sarpi* pkg, the root filesystem type is now detected and the /boot/cmdline.txt file 'rootfstype=' value is updated accordingly and automatically. This was an issue for users who did not select 'ext4' as their root fs type during Slackware ARM installation and thereafter needed to set this value manually in order to boot successfully. This has been fully tested with Slackware supported Linux filesystem types. Unsupported filesystem types have not been tested. Please keep that in mind.

A shoutout with many thanks to; doug713705 in the ##slackware channel on libera.chat for highlighting this problem and providing an idea towards a viable solution for detecting the root fs type, and also to Tadgy for his suggestions and advice on more efficient methods to do the job.

glorsplitz · 10-03-2021, 10:50 AM

Quote:

INIT: Id "s0" respawning too fast: disabled for 5 minutes

applied 5.14.9 to rpi4, saw above, not sure if this has happened before as I do headless, thanks.

Exaga · 10-03-2021, 02:25 PM

Quote:

Originally Posted by glorsplitz

applied 5.14.9 to rpi4, saw above, not sure if this has happened before as I do headless, thanks.

Install the sarpi-hacks*.txz pkg. This pkg contains the fix for the 'INIT: Id "s0" respawning too fast' error. I've not come across this issue after upgrading with the latest sarpi shizzle. Let me know if installing this pkg fixes it.

More details about the sarpi-hacks pkg can be found in the accompanying README: https://slackware.uk/sarpi/extra/sarpi-hacks.README

[EDIT] Just an afterthought, if sarpi pkgs are installed before Slackware ARM pkgs the '/etc/inittab' file might be overwritten with any settings that MoZes includes in the official pkgs. So, it best to upgrade with official Slackware ARM pkgs on the RPis before sarpi pkgs. This could be one reason why you are seeing this 'INIT: Id "s0" respawning too fast' error. Another reason could be if you didn't install the sarpi-hacks* pkg after upgrading Slackware ARM pkgs. It's a known issue on the RPis because they use /dev/ttyAMA0 and not /dev/ttyS0 serial port device.

glorsplitz · 10-03-2021, 04:49 PM

Thank you Exaga for promptly responding.

I have notes on how to apply what you provide when you update packages and I log what was applied and always follow my notes.

Quote:

/sbin/upgradepkg --reinstall --install-new /home/username/sarpi4-hacks-4.0-armv7l-1_slackcurrent_01Oct21_sp1.txz
-rw-r--r-- 1 root root 3496 Oct 3 17:30 /var/log/packages/sarpi4-hacks-4.0-armv7l-1_slackcurrent_01Oct21_sp1

I just did the above and rebooted.

and the culprit it seems, wasn't changed?

Quote:

-rw-r--r-- 1 root root 2439 Oct 1 16:48 /etc/inittab

/etc/inittab has

Quote:

s0:12345:respawn:/sbin/agetty --keep-baud 115200,38400,9600 ttyS1 vt100

on rpi3 there's

Quote:

s0:12345:respawn:/sbin/agetty --keep-baud 115200,38400,9600 ttyAMA0 vt100

I'm going to change it, am I missing something you said to do?

Exaga · 10-03-2021, 05:46 PM

Quote:

Originally Posted by glorsplitz

I'm going to change it, am I missing something you said to do?

First of all, thanks for bringing this problem of yours to me. The 'INIT: Id "s0" respawning too fast' error is a showstopper when logging in remotely because while it's apparent you cannot use the system. I make mistakes as much as the next guy and my work needs to be audited often. People like you keep me on my toes and that's a good thing!

All you need to do is install the sarpi-hacks pkg and it should modify the serial port entry line correctly. If you can copy and paste what that line is now in your /etc/inittab file then we can compare and maybe track down the cause of the problem. - SCRATCH THAT! see EDIT at the bottom...

I have checked and double-checked my RPis running kernel 5.14.9 and they all have the same setting in /etc/inittab file "serial lines" section:

Code:

# Local serial lines:
s0:12345:respawn:/sbin/agetty --keep-baud 115200,38400,9600 ttyAMA0 vt100
#s1:12345:respawn:/sbin/agetty -L ttyS0 9600 vt100
#s2:12345:respawn:/sbin/agetty -L ttyS1 9600 vt100

For the Raspberry Pi this is correct. The "s0" line is the one that should be set by installing the sarpi-hacks pkg. The question on my mind is why it's not working as designed or intended for you. That's what's bothering me most.

However, if you can manually edit this line so it's the same as above then your problem will be solved. In my /etc/inittab it's line: 60

----------------------

[EDIT] I can see what the problem is... your /etc/inittab entry:

Code:

s0:12345:respawn:/sbin/agetty --keep-baud 115200,38400,9600 ttyS1 vt100

Your serial port has been identified as /dev/ttyS1 - and it should be /dev/ttyS0 to be detected properly - I'm guessing your Raspberry Pi has something connected to it that has caused this to happen, like a HAT or another device that offers some serial functionality. If I'm incorrect then I'm at a loss to figure out the cause.

glorsplitz · 10-03-2021, 07:21 PM

Thank you again for following up Exaga.

I don't have anything connected using any serial port, which seems to be using the gpio.

I see /boot/cmdline.txt has console=tty1 but so does my rpi3 and its inittab has ttyAMA0.

If you're sure you dotted all your T's and crossed all your I's, maybe I've done something I don't know I did and you're not telling me what.

Maybe have me check on this or that?

Exaga · 10-04-2021, 03:10 AM

Quote:

Originally Posted by glorsplitz

If you're sure you dotted all your T's and crossed all your I's, maybe I've done something I don't know I did and you're not telling me what.

Maybe have me check on this or that?

The only thing you might like to test/check is...

(This is my RPi4 build-test machine - headless)

Code:

root@torq:~# cat /proc/device-tree/model && echo
Raspberry Pi 4 Model B Rev 1.2
root@torq:~# ls -l /dev/ttyS0 /dev/ttyS1 /dev/ttyAMA0
/bin/ls: cannot access '/dev/ttyS0': No such file or directory
/bin/ls: cannot access '/dev/ttyS1': No such file or directory
crw--w---- 1 root tty 204, 64 Oct  4 07:50 /dev/ttyAMA0
root@torq:~# lsof /dev/ttyAMA0
COMMAND PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
agetty  841 root    0u   CHR 204,64      0t0  130 /dev/ttyAMA0
agetty  841 root    1u   CHR 204,64      0t0  130 /dev/ttyAMA0
agetty  841 root    2u   CHR 204,64      0t0  130 /dev/ttyAMA0
root@torq:~# cat /proc/cmdline
coherent_pool=1M 8250.nr_uarts=0 snd_bcm2835.enable_compat_alsa=0 snd_bcm2835.enable_hdmi=1 bcm2708_fb.fbwidth=0 bcm2708_fb.fbheight=0 bcm2708_fb.fbswap=1 smsc95xx.macaddr=DC:A6:32:67:C4:2B vc_mem.mem_base=0x3eb00000 vc_mem.mem_size=0x3ff00000  dwc_otg.lpm_enable=0 console=tty1 nofont root=PARTUUID=662a1393-03 rootfstype=ext4 rootwait ro
root@torq:~#

(This is my RPi4 NTP/Apache server - headless)

Code:

root@kron:~#  cat /proc/device-tree/model && echo
Raspberry Pi 4 Model B Rev 1.4
root@kron:~# ls -l /dev/ttyS0 /dev/ttyS1 /dev/ttyAMA0
/bin/ls: cannot access '/dev/ttyS0': No such file or directory
/bin/ls: cannot access '/dev/ttyS1': No such file or directory
crw--w---- 1 root tty 204, 64 Sep 24 19:42 /dev/ttyAMA0
root@kron:~# lsof /dev/ttyAMA0
COMMAND PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
agetty  896 root    0u   CHR 204,64      0t0  130 /dev/ttyAMA0
agetty  896 root    1u   CHR 204,64      0t0  130 /dev/ttyAMA0
agetty  896 root    2u   CHR 204,64      0t0  130 /dev/ttyAMA0
root@kron:~# cat /proc/cmdline
coherent_pool=1M 8250.nr_uarts=0 snd_bcm2835.enable_compat_alsa=0 snd_bcm2835.enable_hdmi=1 bcm2708_fb.fbwidth=0 bcm2708_fb.fbheight=0 bcm2708_fb.fbswap=1 smsc95xx.macaddr=E4:5F:01:11:9B:EA vc_mem.mem_base=0x3eb00000 vc_mem.mem_size=0x3ff00000  dwc_otg.lpm_enable=0 console=tty1 nofont root=PARTUUID=35804286-03 rootfstype=ext4 rootwait ro
root@kron:~#

If the 'ls' and 'lsof' command output shows /dev/ttyS0 or /dev/ttyS1 then check that your /proc/cmdline does not contain 'console=ttyS0,115200' or 'console=ttyS1,115200', or something of that nature. If it does, I would say that getty has created a console on ttyS0 or ttyS1 - in that case change your /boot/cmdline.txt file console entry to 'console=tty1' and reboot.

You can also check dmesg for 'tty' entries...

Code:

root@torq:~# dmesg | grep tty
[    0.000000] Kernel command line: coherent_pool=1M 8250.nr_uarts=0 snd_bcm2835.enable_compat_alsa=0 snd_bcm2835.enable_hdmi=1 bcm2708_fb.fbwidth=0 bcm2708_fb.fbheight=0 bcm2708_fb.fbswap=1 smsc95xx.macaddr=DC:A6:32:67:C4:2B vc_mem.mem_base=0x3eb00000 vc_mem.mem_size=0x3ff00000  dwc_otg.lpm_enable=0 console=tty1 nofont root=PARTUUID=662a1393-03 rootfstype=ext4 rootwait ro
[    0.001783] printk: console [tty1] enabled
[    1.471287] fe201000.serial: ttyAMA0 at MMIO 0xfe201000 (irq = 36, base_baud = 0) is a PL011 rev2
root@torq:~#

Also, are you using Bluetooth at all? If your /proc/cmdline contains something like 'console=ttyS<number>,115200' or 'console=serial<number>,115200' then Bluetooth might be the cause of your problem. You can test this by editing your /boot/cmdline.txt file entry to 'console=tty1' and uncommenting (or adding) the line below in your /boot/config.txt file and rebooting.

Code:

dtoverlay=disable-bt

Turning off Bluetooth can solve a lot of issues with the serial port.

Exaga · 10-04-2021, 03:24 AM

Quote:

Originally Posted by glorsplitz

If you're sure you dotted all your T's and crossed all your I's, maybe I've done something I don't know I did and you're not telling me what.

All I know is that, in Slackware ARM 15.0 rc1 on my RPis, the console serial port is as it should be. If this problem is persisting for you then kindly paste the contents of your /boot/cmdline.txt and /boot/config.txt files and let's see what settings are in those files which may be affecting your system serial ports. Right now it's a mystery.

Exaga · 10-05-2021, 04:00 AM

Quote:

Originally Posted by glorsplitz

I see /boot/cmdline.txt has console=tty1 but so does my rpi3 and its inittab has ttyAMA0.

Did you manage to find out what caused the problem on your RPi4 and a solution for it yet, glorsplitz? I'm very interested to know how you're getting on with this issue.

glorsplitz · 10-05-2021, 07:23 PM

Exaga, thank you for following up. I'm very busy during the week working lots of hours, not much putzing around time.

Code:

cat /proc/device-tree/model && echo
Raspberry Pi 4 Model B Rev 1.4

ls -l /dev/ttyS0 /dev/ttyS1 /dev/ttyAMA0
/bin/ls: cannot access '/dev/ttyS0': No such file or directory
/bin/ls: cannot access '/dev/ttyS1': No such file or directory
crw--w---- 1 root tty 204, 64 Oct  3 17:57 /dev/ttyAMA0

lsof /dev/ttyAMA0
COMMAND  PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
agetty  1065 root    0u   CHR 204,64      0t0  131 /dev/ttyAMA0
agetty  1065 root    1u   CHR 204,64      0t0  131 /dev/ttyAMA0
agetty  1065 root    2u   CHR 204,64      0t0  131 /dev/ttyAMA0

cat /proc/cmdline
coherent_pool=1M 8250.nr_uarts=0 snd_bcm2835.enable_compat_alsa=0 snd_bcm2835.enable_hdmi=1 bcm2708_fb.fbwidth=1184 bcm2708_fb.fbheight=624 bcm2708_fb.fbswap=1 smsc95xx.macaddr=DC:A6:32:D5:98:9F vc_mem.mem_base=0x3ec00000 vc_mem.mem_size=0x40000000  dwc_otg.lpm_enable=0 console=tty1 nofont root=/dev/sda1 rootfstype=ext4 rootwait ro

cat /boot/cmdline.txt 
dwc_otg.lpm_enable=0 console=tty1 nofont root=/dev/sda1 rootfstype=ext4 rootwait ro

And speaking of putzing around, I setup my rpi4, got chromium, libreoffice, mpv, working and haven't done any other tweaking/add ons.

I've kept it up to date with MoZes updates and when you provide your updates.

drmozes · 10-06-2021, 05:15 AM

We encountered this on the most recent SlackChat episode - it should be ttyS1 for the RPI4, and a/sysvinit-scripts now sets it correctly within /etc/inittab.

If you do query this inteface, you should see the same:

Code:

 $ strings /sys/firmware/devicetree/base/chosen/stdout-path
serial1:115200n8

I have no console= settings on any ARM nor AArch64 Hardware Models, and serial and HDMI works on all. The settings are configured within the DTB.

Exaga · 10-06-2021, 06:37 AM

Quote:

Originally Posted by drmozes

We encountered this on the most recent SlackChat episode - it should be ttyS1 for the RPI4, and a/sysvinit-scripts now sets it correctly within /etc/inittab.

If you do query this inteface, you should see the same:

Code:

 $ strings /sys/firmware/devicetree/base/chosen/stdout-path
serial1:115200n8

I have no console= settings on any ARM nor AArch64 Hardware Models, and serial and HDMI works on all. The settings are configured within the DTB.

Thanks for the info Stu. Obviously this is something I need to look into and consider. When I perform this query the result is...

Code:

root@torq:~# strings /sys/firmware/devicetree/base/chosen/stdout-path
strings: '/sys/firmware/devicetree/base/chosen/stdout-path': No such file

I'm using the Raspberry Pi Github repo kernel with FDT and have 'console=tty1' in the kernel cmdline params...

Code:

root@torq:~# strings /sys/firmware/devicetree/base/chosen/bootargs | grep -o "console.*" /proc/cmdline | cut -d' ' -f1
console=tty1
root@torq:~# cat /proc/cmdline
coherent_pool=1M 8250.nr_uarts=0 snd_bcm2835.enable_compat_alsa=0 snd_bcm2835.enable_hdmi=1 bcm2708_fb.fbwidth=1824 bcm2708_fb.fbheight=984 bcm2708_fb.fbswap=1 smsc95xx.macaddr=DC:A6:32:67:C4:2B vc_mem.mem_base=0x3eb00000 vc_mem.mem_size=0x3ff00000  dwc_otg.lpm_enable=0 console=tty1 nofont root=PARTUUID=662a1393-03 rootfstype=ext4 rootwait ro

Which is taken directly from the kernel .config "CONFIG_CMDLINE=" params...

Code:

root@torq:~# grep "CONFIG_CMDLINE=" /root/linux/.config
CONFIG_CMDLINE="dwc_otg.lpm_enable=0 console=tty1 nofont root=/dev/mmcblk0p3 rootfstype=ext4 rootwait ro"

Are you using 'fdt addr ${fdt_addr} && fdt get value bootargs /chosen bootargs' or some variation of this in the /boot/boot.scr by any chance?

Much confusion for me over this serial port issue.

netcrawl · 10-06-2021, 05:05 PM

Hi Exaga,I'm getting the same notification about "s0" on my pi 4 after doing "slackpkg upgrade-all", but I'm still using kernel 5.10.59 and etc because of the HDMI audio issue. And I see the same result as you when I run the strings query.

glorsplitz · 10-06-2021, 07:13 PM

Quote:

Originally Posted by Exaga

Thanks for the info Stu. <-YES

Code:

root@torq:~# strings /sys/firmware/devicetree/base/chosen/stdout-path
strings: '/sys/firmware/devicetree/base/chosen/stdout-path': No such file

^SAME HERE^

Well this clears it all up

Seriously, cheers to all for following up!

Exaga · 10-07-2021, 03:32 AM

Quote:

Originally Posted by netcrawl

Hi Exaga,I'm getting the same notification about "s0" on my pi 4 after doing "slackpkg upgrade-all", but I'm still using kernel 5.10.59 and etc because of the HDMI audio issue. And I see the same result as you when I run the strings query.

Hi netcrawl. Thanks for letting me know. This is certainly puzzling.

I too used 'slackpkg upgrade-all' to install Slackware ARM 15.0 release candidate 1 on my Raspberry Pis - but I did not select to install the official Slackware ARM kernel, kernel_modules, kernel-headers, or kernel-source pkgs. Did you install any or all of these pkgs when you upgraded? The kernel-headers and source won't make any difference to the serial port but I'm thinking the kernel might. I wonder if that's what's made the difference here, and that's a guess at this moment in time. I am very aware of the 'INIT: Id "s0" respawning too fast: disabled for 5 minutes' error on the Raspberry Pi devices and it's been resolved by installing the sarpi-hacks pkg for goodness knows how many years. So something, somewhere, has changed recently to cause it to resurface.

Can I ask you, did this 'INIT: Id "s0" respawning too fast' appear right after you performed the 'slackpkg upgrade-all' or was it apparent before then?

I'll need to do some more investigating to see what's going on here. So far I've looked into and compared what the Raspberrry Pi OS settings are and found this...

Code:

root@raspberrypi:~# cat /boot/cmdline.txt
console=serial0,115200 console=tty1 root=PARTUUID=24f289c3-02 rootfstype=ext4 elevator=deadline fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles

So, 'console=serial0' is usually an alias of 'console=ttyS0'. However, I checked this out on Raspberry Pi OS and found...

Code:

root@raspberrypi:~# ls -l /dev/ttyS* /dev/serial* /dev/ttyAMA*
ls: cannot access '/dev/ttyS*': No such file or directory
lrwxrwxrwx 1 root root          7 Oct  7 08:44  /dev/serial1 -> ttyAMA0
crw-rw---- 1 root dialout 204, 64 Oct  7 08:44  /dev/ttyAMA0

Here I have '/dev/serial1' as a symlink to '/dev/ttyAMA0' and no '/dev/serial0' or '/dev/ttyS0' to be found.

Code:

root@raspberrypi:~# lsof /dev/serial1
COMMAND   PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
hciattach 525 root    3u   CHR 204,64      0t0  134 /dev/ttyAMA0

But then I saw this post from PhilE, Raspberry Pi Engineer who confirms that the default kernel command line includes console=serial0, which the firmware translates to ttyAMA0 or ttyS0 as necessary. That entire forum thread is well worth a read. PhilE suggests in a later post to add these to the /boot/config.txt file and says that the miniuart overlay uses ttyS0 for Bluetooth, freeing /dev/ttyAMA0, and he deliberately reversed the order of the UART overlays to demonstrate that ordering here makes no difference to the ttyAMA numbering.

Code:

enable_uart=1
dtoverlay=pi3-miniuart-bt
dtoverlay=uart5
dtoverlay=uart4
dtoverlay=uart3
dtoverlay=uart2

So you might want to test with these settings yourself. In the meantime, I'm looking into it but will say again that I have not seen the 'INIT: Id "s0" respawning too fast' error myself on any of my RPi devices for a long, long time (years), which includes running kernel 5.10.x, 5.13.x, and 5.14.x.

[EDIT] I also want to reiterate that whenever I use slackpkg to install any updated official pkgs I'll always install any new sarpi updated pkgs right after slackpkg has completed, but before rebooting the Raspberry Pi. Usually I'll 'upgradepkg --reinstall <pkg name>' the sarpi pkgs if/when any new official Slackware pkgs have been installed and there are no new sarpi pkgs. I'm especially mindful of the differences in the mainline and Raspberry Pi kernel sources. As MoZes develops and perfects his Slackware ARM/AArch64 kernel I expect there will be more clarity and less disparity for myself, as the RPi devices become officially supported by him.