Solved Message starting FreeBSD

SirDice · Dec 2, 2022

Waldin said:
grep -r procstat /etc /usr/local/etc /var/cron
/etc/mtree/BSD.tests.dist: procstat

Ok, it's fine in that file. The error you keep seeing is the result of something trying to execute procstat(1) with some bad parameters. We just need to figure out where this happens. It doesn't seem to be in any of the usual suspects; the rc(8) files.

_al · Dec 2, 2022

Waldin said:

Good Morning. For some time now, during the FreeBSD 13.1 system load, I have seen the following message:

Code:

Waiting 3s for the default route interface: usage: procstat [--libxo] [-h] [-M core] [-N system] [-w interval] command
                [pid ... | core ...]
       procstat [--libxo] -a [-h] [-M core] [-N system]  [-w interval] command
       procstat [--libxo] [-h] [-M core] [-N system] [-w interval]
                [-S | -b | -c | -e | -f [-C] | -i [-n] | -j [-n] | -k [-k] |
                 -l | -r [-H] | -s | -t | -v | -x] [pid ... | core ...]
       procstat [--libxo] -a [-h] [-M core] [-N system] [-w interval]
                [-S | -b | -c | -e | -f [-C] | -i [-n] | -j [-n] | -k [-k] |
                 -l | -r [-H] | -s | -t | -v | -x]
       procstat [--libxo] -L [-h] [-M core] [-N system] core ...
Available commands:
       argument(s)
       auxv
       basic
       binary
       [cpuset | cs]
       credential(s)
       environment
       [fd(s) | file(s)] [-C]
       kstack [-v]
       pargs
                 -l | -r [-H] | -s | -t | -v | -x]
       procstat [--libxo] -L [-h] [-M core] [-N system] core ...
Available commands:
       argument(s)
       auxv
       basic
       binary
       [cpuset | cs]
       credential(s)
       environment
       [fd(s) | file(s)] [-C]
       kstack [-v]
       pargs
       penv
       ptlwpinfo
       pwdx
       rlimit
       rusage [-Ht]
       sigfastblock
       signal(s) [-n]
       thread(s)
       tsignal(s) [-n]
       vm

I don't know where to look to make it go away Any ideas? Thank you very much.

I also got the same message. It disappeared after removing the automount package.

SirDice · Dec 2, 2022

vermaden 's sysutils/automount?

_al · Dec 2, 2022

Hmm, I honestly did not know that this was the development of the esteemed vermaden.
And yes, I meant automount(8)

SirDice · Dec 2, 2022

_al said:
I honestly did not know that this was the development of the esteemed @vermaden.

Yep, it is. https://github.com/vermaden/automount

_al said:
And yes, I meant automount(8)

Wrong man page

After a bit of digging, procstat(1) is used exactly once in /usr/local/sbin/automount.

Code:

      for I in ${GROUP_USERS}
      do
        [ "${I}" = "root" ] && continue
        DISPLAY_ID=$( procstat pargs $( pgrep Xorg ) | grep "^argv\[1\]:" | awk '{print $NF}' )
        if [ -z "${DISPLAY_ID}" ]
        then
          continue
        fi
        __log "${DEV}: starting '${FM}' file manager"
        su -l "${I}" -c "env DISPLAY=${DISPLAY_ID} ${FM} ${MNT} &" 1> /dev/null 2>&1
      done

If pgrep Xorg doesn't return anything procstat pargs is executed with a missing argument. Which results in this error:

Code:

root@molly:~ # procstat pargs
usage: procstat [--libxo] [-h] [-M core] [-N system] [-w interval] command
                [pid ... | core ...]
       procstat [--libxo] -a [-h] [-M core] [-N system]  [-w interval] command
       procstat [--libxo] [-h] [-M core] [-N system] [-w interval]
                [-S | -b | -c | -e | -f [-C] | -i [-n] | -j [-n] | -k [-k] |
                 -l | -r [-H] | -s | -t | -v | -x] [pid ... | core ...]
       procstat [--libxo] -a [-h] [-M core] [-N system] [-w interval]
                [-S | -b | -c | -e | -f [-C] | -i [-n] | -j [-n] | -k [-k] |
                 -l | -r [-H] | -s | -t | -v | -x]
       procstat [--libxo] -L [-h] [-M core] [-N system] core ...
Available commands:
       advlock(s)
       argument(s)
       auxv
       basic
       binary
       [cpuset | cs]
       credential(s)
       environment
       [fd(s) | file(s)] [-C]
       kstack [-v]
       pargs
       penv
       ptlwpinfo
       pwdx
       rlimit
       rusage [-Ht]
       sigfastblock
       signal(s) [-n]
       thread(s)
       tsignal(s) [-n]
       vm

Which is exactly the error we've been looking for. I guess you get to see it when the system boots and Xorg hasn't started yet. That happens later, or not at all, depending on how you're starting Xorg. /usr/local/sbin/automount is triggered when devd(8) detects certain devices.

_al · Dec 2, 2022

As for me, I start X with startx. And as for the situation with automount, I discovered it by accident. I have a Broadcom LSI SAS 9300-8i on my system and the HDDs on it show up as 'da0...da1...' in FreeBSD. When the system boots, automount tries to mount them in /media/daX... So I removed it. After that, this message disappeared.

Perhaps I did something wrong, and incorrectly configured to work with LSI SAS's HDDs.

SirDice · Dec 2, 2022

_al said:
I discovered it by accident.

That's how a lot of bugs are discovered

_al said:
Perhaps I did something wrong, and incorrectly configured to work with LSI SAS's HDDs.

No, that seems "by design", not because of an incorrect configuration. Automount looks for da(4) devices, as that's usually a USB stick or external drive. Most desktop machines typically use ada(4) drives internally, so there's no "confusion" between internal and external drives.

smithi · Dec 2, 2022

Waldin said:
Yes.

cat /etc/rc.local
#!/bin/sh
usbconfig -d ugen2.3 power_save

Webcam, fair enough.

Waldin said:
apm -e enable
apm -d enable
apm -h enable

These are only not a problem because they don't do anything at all, because of some clever tidying up maybe 12? years ago when there were still a few APM BIOS laptops made. I had one between ~2003 and 2019.

ACPI precludes APM, and amd64 mode requires ACPI.

APM functions are now just emulated with corresponding ACPI functions, eg apm -z will suspend exactly as acpiconf -s3 will, apm derives battery info and state from where acpiconf -i0 does.

If not convinced, see
% less `which zzz` which shows that presence of sysctl machdep.apm_suspend_delay indicates real APM support, despite what
'apm -e enable' suggests.

Were these trying to solve some problem?

_al · Dec 2, 2022

SirDice said:
Automount looks for da(4) devices, as that's usually a USB stick or external drive.

It means that such a message is displayed when automount cannot mount the device. For example, an external (dongle) usb modem...

Waldin · Dec 2, 2022

_al said:
I also got the same message. It disappeared after removing the automount package.

Oh! Finally the culprit of the error appeared. I guess it will be a matter of updating the version of automount of Vermaden. This program is essential for me and I cannot uninstall it

Waldin · Dec 2, 2022

smithi said:
Webcam, fair enough.

These are only not a problem because they don't do anything at all, because of some clever tidying up maybe 12? years ago when there were still a few APM BIOS laptops made. I had one between ~2003 and 2019.

ACPI precludes APM, and amd64 mode requires ACPI.

APM functions are now just emulated with corresponding ACPI functions, eg apm -z will suspend exactly as acpiconf -s3 will, apm derives battery info and state from where acpiconf -i0 does.

If not convinced, see
% less `which zzz` which shows that presence of sysctl machdep.apm_suspend_delay indicates real APM support, despite what
'apm -e enable' suggests.

Were these trying to solve some problem?

Some time ago I searched for information to improve the power consumption of the laptop and I just saw usbconfig -d ugen2.3 power_save for the webcam, and apm, but if that doesn't help, I have no problem removing it.

Waldin · Dec 2, 2022

SirDice said:

Yep, it is. https://github.com/vermaden/automount

Wrong man page

After a bit of digging, procstat(1) is used exactly once in /usr/local/sbin/automount.

Code:

      for I in ${GROUP_USERS}
      do
        [ "${I}" = "root" ] && continue
        DISPLAY_ID=$( procstat pargs $( pgrep Xorg ) | grep "^argv\[1\]:" | awk '{print $NF}' )
        if [ -z "${DISPLAY_ID}" ]
        then
          continue
        fi
        __log "${DEV}: starting '${FM}' file manager"
        su -l "${I}" -c "env DISPLAY=${DISPLAY_ID} ${FM} ${MNT} &" 1> /dev/null 2>&1
      done

If pgrep Xorg doesn't return anything procstat pargs is executed with a missing argument. Which results in this error:

Code:

root@molly:~ # procstat pargs
usage: procstat [--libxo] [-h] [-M core] [-N system] [-w interval] command
                [pid ... | core ...]
       procstat [--libxo] -a [-h] [-M core] [-N system]  [-w interval] command
       procstat [--libxo] [-h] [-M core] [-N system] [-w interval]
                [-S | -b | -c | -e | -f [-C] | -i [-n] | -j [-n] | -k [-k] |
                 -l | -r [-H] | -s | -t | -v | -x] [pid ... | core ...]
       procstat [--libxo] -a [-h] [-M core] [-N system] [-w interval]
                [-S | -b | -c | -e | -f [-C] | -i [-n] | -j [-n] | -k [-k] |
                 -l | -r [-H] | -s | -t | -v | -x]
       procstat [--libxo] -L [-h] [-M core] [-N system] core ...
Available commands:
       advlock(s)
       argument(s)
       auxv
       basic
       binary
       [cpuset | cs]
       credential(s)
       environment
       [fd(s) | file(s)] [-C]
       kstack [-v]
       pargs
       penv
       ptlwpinfo
       pwdx
       rlimit
       rusage [-Ht]
       sigfastblock
       signal(s) [-n]
       thread(s)
       tsignal(s) [-n]
       vm

Which is exactly the error we've been looking for. I guess you get to see it when the system boots and Xorg hasn't started yet. That happens later, or not at all, depending on how you're starting Xorg. /usr/local/sbin/automount is triggered when devd(8) detects certain devices.

If I'm not mistaken, on my computer Xorg starts automatically, through the Lightweight Display Manager, using the Mate desktop. I don't know if I can somehow make Xorg start differently so I don't see the procstat message, but hey, knowing that it's caused by automount, if necessary, I ignore it.

_al · Dec 3, 2022

SirDice said:
Wrong man page

Yes, I see. Thank you. In addition to github link you pointed, I found this - - https://vermaden.wordpress.com/2018/10/11/freebsd-desktop-part-17-automount-removable-media/

smithi · Dec 4, 2022

Waldin said:
Some time ago I searched for information to improve the power consumption of the laptop and I just saw usbconfig -d ugen2.3 power_save for the webcam,

That's likely worth up to 200mA. I use power_off and will turn it back on as needed.

I also set power_off on the fingerprint reader (100mA) and the big win of ~500 mA mW turning bluetooth off.

NB if you suspend, you need to also add your usbconfig power_{save,off} commands to the end of /etc/rc.resume or power is turned back on.

Waldin said:
and apm, but if that doesn't help, I have no problem removing it.

It won't help. Tuning your powerd{,xx} settings might.

vermaden · Dec 5, 2022

SirDice said:

Yep, it is. https://github.com/vermaden/automount

Wrong man page

After a bit of digging, procstat(1) is used exactly once in /usr/local/sbin/automount.

Code:

      for I in ${GROUP_USERS}
      do
        [ "${I}" = "root" ] && continue
        DISPLAY_ID=$( procstat pargs $( pgrep Xorg ) | grep "^argv\[1\]:" | awk '{print $NF}' )
        if [ -z "${DISPLAY_ID}" ]
        then
          continue
        fi
        __log "${DEV}: starting '${FM}' file manager"
        su -l "${I}" -c "env DISPLAY=${DISPLAY_ID} ${FM} ${MNT} &" 1> /dev/null 2>&1
      done

If pgrep Xorg doesn't return anything procstat pargs is executed with a missing argument. Which results in this error:

Code:

root@molly:~ # procstat pargs
usage: procstat [--libxo] [-h] [-M core] [-N system] [-w interval] command
                [pid ... | core ...]
       procstat [--libxo] -a [-h] [-M core] [-N system]  [-w interval] command
       procstat [--libxo] [-h] [-M core] [-N system] [-w interval]
                [-S | -b | -c | -e | -f [-C] | -i [-n] | -j [-n] | -k [-k] |
                 -l | -r [-H] | -s | -t | -v | -x] [pid ... | core ...]
       procstat [--libxo] -a [-h] [-M core] [-N system] [-w interval]
                [-S | -b | -c | -e | -f [-C] | -i [-n] | -j [-n] | -k [-k] |
                 -l | -r [-H] | -s | -t | -v | -x]
       procstat [--libxo] -L [-h] [-M core] [-N system] core ...
Available commands:
       advlock(s)
       argument(s)
       auxv
       basic
       binary
       [cpuset | cs]
       credential(s)
       environment
       [fd(s) | file(s)] [-C]
       kstack [-v]
       pargs
       penv
       ptlwpinfo
       pwdx
       rlimit
       rusage [-Ht]
       sigfastblock
       signal(s) [-n]
       thread(s)
       tsignal(s) [-n]
       vm

Which is exactly the error we've been looking for. I guess you get to see it when the system boots and Xorg hasn't started yet. That happens later, or not at all, depending on how you're starting Xorg. /usr/local/sbin/automount is triggered when devd(8) detects certain devices.

This seems to be in the old version in automount(8) daemon.

This is the newer code that does continue (in for loop) when there is no Xorg process.

Code:

      GROUP_USERS=$( pw group show ${MNT_GROUP} | sed -e 's|.*:||' -e 's|,| |g' )
      for I in ${GROUP_USERS}
      do
        [ "${I}" = "root" ] && continue
        XORG_PID=$( pgrep Xorg )
        [ "${XORG_PID}" = "" ] && continue
        DISPLAY_ID=$( procstat pargs ${XORG_PID} | grep "^argv\[1\]:" | awk '{print $NF}' )
        [ -z "${DISPLAY_ID}" ] && continue
        __log "${DEV}: starting '${FM}' file manager"
        su -l "${I}" -c "env DISPLAY=${DISPLAY_ID} ${FM} ${MNT} &" 1> /dev/null 2>&1
      done

Hope that helps.

Waldin · Dec 6, 2022

Thank you very much for your help vermaden.
It work fine

Solved Message starting FreeBSD

SirDice

Administrator

_al

SirDice

Administrator

_al

SirDice

Administrator

_al

SirDice

Administrator

smithi

_al

Waldin

Waldin

Waldin

_al

smithi

vermaden

Waldin