Re: [DNG] any gotchas in a new install?

[Brad Campbell via Dng]

On 19/6/22 01:57, o1bigtenor via Dng wrote:
> Greetings
> 
> Am having entirely too much fun (doa psu followed by the mobo) but
> hoping to be very soon installing Devuan testing (daedalus - - - -
> sic!) on a AMD 5800X cpu with a Radeon 570X gpu.
> 
> When I've been looking for info there seems to be lots of chatter
> about how the Radeon gpus aren't installing or working well on Debian
> testing.
> 
> Any tips from those that have assembled a system in the last year?

Chimaera on a Ryzen 5600G.

Worked out of the box. I did install current linux-firmware from git, but only 
because I always do. Goes like a stung cat.

Whilst it's not a dedicated card, it's certainly a Radeon GPU.

Brad.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Shutdown/halt versus WiFi and NFS

[Brad Campbell via Dng]

On 28/5/22 18:44, d...@d404.nl wrote:
> On 28-05-2022 10:23, Brad Campbell via Dng wrote:
>> On 24/9/20 03:55, Michael S. Keller via Dng wrote:
>>> My desktop is running Chimaera, and I saw this with Beowulf, but didn't 
>>> spend much time on it then.
>>>
>>> My network connection is via WiFi, and I have permanent NFS mounts in 
>>> place. I run SysV init.
>>>
>>> During halt or shutdown via init scripts, NetworkManager is terminated 
>>> before the NFS unmount, which brings down the active NIC, and usually the 
>>> unmount hangs forever, so I have to do a hard reset or power-off.
>>>
>>> After futzing with it for a while, trying to find a more elegant solution, 
>>> I ended up just renaming K01network-manager and K02sendsigs in rc0.d and 
>>> rc6.d. Now shutdown and reboot run reliably.
>>>
>>> Before that, I tried renaming K01network-manager to K06network-manager, to 
>>> place it after the NFS unmount, but it ran earlier anyway.
>>>
>>> I also tried shielding NetworkManager from sendsigs, and I think it would 
>>> have worked if I could make K0.network-manager run later, but that was 
>>> about the point I gave up and took a virtual hammer to the issue.
>> 2 years on and I keep bumping up against this issue on my laptop. Truth be 
>> told I've been bumping up against it for a long while, but an enforced break 
>> saw me have time to look into it.
>>
>> I've tried shuffling dependencies in the scripts, but can't seem to sort it 
>> out.
>>
>> Mine are transient nfs mounts, but if I forget to unmount one before 
>> rebooting it hangs and has to be power cycled.
>>
>> $ cat /etc/devuan_version
>> chimaera
>>
>> Love to hear any suggestions.
>>
>> Regards,
>> Brad
>> ___
>> Dng mailing list
>> Dng@lists.dyne.org
>> https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
> 
> For my Raspberry Pi's to boot from nfs and to shutdown normally I had to add 
> the variable
> 
> persistent
> 
> to /etc/dhcpcd.conf.
> 
> Hope this helps.
> 

Thanks Nick,

Yes, I saw your response back in 2020.

I've been trying to figure out how that applies to network manager, or how I 
could make a similar outcome.
The problem here is network-manager gets killed early in the process. Same 
result but a different cause.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Shutdown/halt versus WiFi and NFS

[Brad Campbell via Dng]

On 24/9/20 03:55, Michael S. Keller via Dng wrote:
> My desktop is running Chimaera, and I saw this with Beowulf, but didn't spend 
> much time on it then.
> 
> My network connection is via WiFi, and I have permanent NFS mounts in place. 
> I run SysV init.
> 
> During halt or shutdown via init scripts, NetworkManager is terminated before 
> the NFS unmount, which brings down the active NIC, and usually the unmount 
> hangs forever, so I have to do a hard reset or power-off.
> 
> After futzing with it for a while, trying to find a more elegant solution, I 
> ended up just renaming K01network-manager and K02sendsigs in rc0.d and rc6.d. 
> Now shutdown and reboot run reliably.
> 
> Before that, I tried renaming K01network-manager to K06network-manager, to 
> place it after the NFS unmount, but it ran earlier anyway.
> 
> I also tried shielding NetworkManager from sendsigs, and I think it would 
> have worked if I could make K0.network-manager run later, but that was about 
> the point I gave up and took a virtual hammer to the issue.

2 years on and I keep bumping up against this issue on my laptop. Truth be told 
I've been bumping up against it for a long while, but an enforced break saw me 
have time to look into it.

I've tried shuffling dependencies in the scripts, but can't seem to sort it out.

Mine are transient nfs mounts, but if I forget to unmount one before rebooting 
it hangs and has to be power cycled.

$ cat /etc/devuan_version 
chimaera

Love to hear any suggestions.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Beowulf to Chimaera update breaks suspend on laptop

[Brad Campbell via Dng]

On 7/4/22 20:58, Brad Campbell via Dng wrote:
> G'day,
> 
> I could use a bit of advice if anyone has the relevant experience.
> 
> My laptop is running Devuan Beowulf currently, but this suspend config goes 
> back at least 10 years.
> It suspends / hibernates using pm-utils with the uswsusp back-end onto a 
> dmycrypted swap partition. The kernel is vanilla with a cut-down config and 
> locally compiled.
> 
> This works flawlessly and has done since I set it up. The GUI is xfce4 and 
> when I hit the power button xfce4-power-manager does all the right things 
> (which is hit up pm-utils and get out of its way).
> 
> Last night I upgraded to Chimaera. This installed elogind and tries to use 
> that to pull the relevant kernel levers to suspend. Unfortunately on my 
> system, while it suspends most times, it comes back about 1 in 10 and then 
> usually the nvme is broken and it dies in a ball of flames. I spent the best 
> part of a day experimenting with the in-kernel suspend mechanisms and I can't 
> seem to make it work, while the trusty old uswsusp userspace suspend/resume 
> works first time every time.
> 
> So, I'm asking for either :
> - Experience in making the in-kernel mechanisms work; or preferably
> - How to remove elogind from an xfce4 Chimaera install and make 
> xfce4-power-manager use pm-utils like it used to.
> 
> Does anyone have any ideas?
> 

Just to follow this up, it appears to be related to power management. When 
unplugged pm-utils pulls a pile of levers in the kernel to reduce power 
consumption. This does reduce on-battery consumption significantly, but it also 
breaks resume on a number of pcie devices. In the past I had issues with both 
brcmfmac and the xhci driver, so I was unloading those pre-suspend and 
re-loading on resume. I also had to serialize the device suspend as async 
seemed to lock up.

By inserting some magic into the suspend hook to undo all the power management 
magic, it would appear I no longer need to unload any modules and suspend, 
hibernate and hybrid suspend all work and resume as-was.

I don't want to speak too soon, but so far I have 24 hybrid cycles on the 
machine and it's still resuming ok.

To be specific, pre-suspend I'm using :

echo 0 > /sys/power/pm_async
echo default > /sys/module/pcie_aspm/parameters/policy 
/usr/lib/pm-utils/power.d/pci_devices false

Async device freeze off and then undo all the power saving modes.

Once I'm convinced it's reliable, I'll start peeling these commands back one by 
one to see which one(s) cause the issues, but at the moment it's looking happy.

Now, why this all works on Beowulf and doesn't on Chimaera is entirely a 
mystery give the pm-utils, uswsusp and kernel config is identical is beyond me, 
but I figure as both pm-utils and uswsusp are effectively dead I'm going to 
have to use the in-kernel stuff at some point.

My current hook script in /lib/elogind/system-sleep/Hooks.sh :

#!/bin/bash
do_wakeup () {
for i in LID0 XHC1 ; do
if [ -z "`cat /proc/acpi/wakeup | grep $i | grep disabled`" ] ; 
then
echo $i > /proc/acpi/wakeup
fi;
done;
}

do_suspend() {
echo N > /sys/module/printk/parameters/console_suspend
echo 0 > /sys/power/image_size
echo 0 > /sys/power/pm_async
echo default > /sys/module/pcie_aspm/parameters/policy 
/usr/lib/pm-utils/power.d/pci_devices false
do_wakeup
/etc/init.d/openvpn stop
xscreensaver-command -lock
sync
}

do_resume() {
for i in gpe66 gpe4E ; do
echo disable > /sys/firmware/acpi/interrupts/$i 2>/dev/null
done;
do_wakeup
rfkill block bluetooth
echo 120 > /sys/class/leds/smc::kbd_backlight/brightness
xscreensaver-command -deactivate
if [ "`cat /sys/class/power_supply/ADP1/online`" -eq 1 ] ; then
   pm-powersave false
else
   pm-powersave true
fi 

}
PID=`pgrep xfce4-session`
USER=`ps -p $PID -o ruser=`
export DISPLAY=:0.0
export XAUTHORITY=/home/$USER/.Xauthority

case $1/$2 in
pre/*)
do_suspend
;;
post/*)
do_resume
;;
esac


Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Beowulf to Chimaera update breaks suspend on laptop

[Brad Campbell via Dng]

On 8/4/22 07:45, Adrian Zaugg wrote:
> Hi Brad
> 
> In der Nachricht vom Thursday, 7 April 2022 14:58:36 CEST schrieb Brad 
> Campbell via Dng:
>> So, I'm asking for either :
>> - Experience in making the in-kernel mechanisms work; or preferably
> Updating the BIOS of your Box to the newest version might help here.
> 

G'day Adrian,

It's a MacbookPro with the current EFI.

Further digging seems to point to some interaction with pm-utils and pcie power 
management because the pm-utils suspend is flaky on Chimaera also (more 
reliable than in-kernel still).

The frustrating thing is with the exact same kernel and exact same pm-utils 
config and hooks it's reliable on Beowulf.

Two steps forward, one step back. 

Thanks for the input.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Beowulf to Chimaera update breaks suspend on laptop

[Brad Campbell via Dng]

On 8/4/22 08:15, Ralph Ronnquist wrote:
> afaik on chimaera xfce4 relies on elogind for handling "suspend on lid
> close", and this is something you may disable by a slight edit of
> /etc/elogind/logind.conf, to have the three assignments:
> HandleLidSwitch=ignore
> HandleLidSwitchExternalPower=ignore
> HandleLidSwitchDocked=ignore
> 
> Hopefully that will let your previous suspend solution work.
> 

G'day Ralph,

No, that didn't work. xfce4-power-manager had already taken over the monitoring 
and management and I couldn't
figure out how to stop it using elogind as the back-end even with all the 
suspend/hibernate tasks disabled
in elogind.

As pm-utils hasn't been updated in years, and xfce4 recently dropped support 
for it entirely (removed the interface code)
I figure I'm going to have to make elogind work at some point, so it might as 
well be now.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

[DNG] Beowulf to Chimaera update breaks suspend on laptop

[Brad Campbell via Dng]

G'day,

I could use a bit of advice if anyone has the relevant experience.

My laptop is running Devuan Beowulf currently, but this suspend config goes 
back at least 10 years.
It suspends / hibernates using pm-utils with the uswsusp back-end onto a 
dmycrypted swap partition. The kernel is vanilla with a cut-down config and 
locally compiled.

This works flawlessly and has done since I set it up. The GUI is xfce4 and when 
I hit the power button xfce4-power-manager does all the right things (which is 
hit up pm-utils and get out of its way).

Last night I upgraded to Chimaera. This installed elogind and tries to use that 
to pull the relevant kernel levers to suspend. Unfortunately on my system, 
while it suspends most times, it comes back about 1 in 10 and then usually the 
nvme is broken and it dies in a ball of flames. I spent the best part of a day 
experimenting with the in-kernel suspend mechanisms and I can't seem to make it 
work, while the trusty old uswsusp userspace suspend/resume works first time 
every time.

So, I'm asking for either :
- Experience in making the in-kernel mechanisms work; or preferably
- How to remove elogind from an xfce4 Chimaera install and make 
xfce4-power-manager use pm-utils like it used to.

Does anyone have any ideas?

Thankfully I always backup the root partition before an upgrade, so a rollback 
to a working system was only a quick rsync away. I'd like to upgrade (and I've 
upgraded all my desktops) but working suspend is a not-negotiable on a laptop.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Installing on a Raspberry or a Banana?

[Brad Campbell via Dng]

On 8/2/22 5:17 am, Antony Stone wrote:
> On Monday 07 February 2022 at 22:14:32, d...@d404.nl wrote:
> 
>> On 07-02-2022 22:03, Antony Stone wrote:
>>>
>>> I want to install Devuan on a Raspberry Pi Zero W, and also on a Banana
>>> Pi R1.
> 
>> Last time I needed an RPi image I used this link
>> https://arm-files.devuan.org/
> 
> Oh!
> 
> Thanks - I didn't spot this linked from any of the "download" or "how to 
> install" pages I looked at.

Just a +1 thanks. I've done several recent ASCII installs to then progressively 
(slowly) upgrade because I thought someone had stopped producing images.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] [OT] Jörg Schilling has passed away

[Brad Campbell via Dng]

On 12/10/21 04:16, Steve Litt wrote:

hal said on Mon, 11 Oct 2021 10:59:26 -0500


Just thought I'd mention it. There was a time when Jörg's software was
the only way to burn a CD on Solaris. I was grateful for his
contributions as it saved me a lot of time moving things to tape. RIP
Jörg, and thanks for all the fish. https://lwn.net/Articles/872489/rss


To this day I use cdrecord, and am grateful.



^^^ What he said. When wodim came along, I just compiled cdrecord from 
source.


Bollocks to Cancer..

Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] random sudden stops

[Brad Campbell via Dng]

On 26/8/21 8:10 am, Hendrik Boom wrote:
> For the past few months my home server (running an ascii installation 
> physically moved from another computer) has been suddenly stopping all 
> processing about once a month. apparently at random.  It seems to stop 
> instantly, leaving power on and becoming completely responsive to ping,
> existing ssh connexions and use of the physical keyboard.
> 
> The system log, after a reboot, shows nothing unusual except of course 
> that there are no log entries for a shut-down.
> 
> Can anyone provide ideas about tracking this down?
> 
> It could of course be a random rare intermittent hardware error.

Sounds like the perfect application for netconsole.

I have a raspberry pi that runs some stuff, on that I installed udplogger : 
https://lwn.net/Articles/571589/
Run with : /usr/local/bin/udplogger port= dir=/root/udplogs/

I have a number of machines set up with netconsole on the command line, or 
loaded after boot. There are easier ways to do this, but for whatever reason 
this is what I use (I honestly don't recall) :

DEST=192.168.24.218
mount none -t configfs /sys/kernel/config
mkdir /sys/kernel/config/netconsole/target1
pushd /sys/kernel/config/netconsole/target1
echo 192.168.24.1 > local_ip
echo $DEST > remote_ip
echo br0 > dev_name
arping -c1 $DEST | grep -o ..:..:..:..:..:.. > remote_mac
echo 1 > enabled
popd

Or on the kernel command line  :
netconsole=@192.168.24.187/eth0,@192.168.42.218/ab:cd:ef:12:34:56

That way I pretty much always get the oops that never makes it to disk.

2021-07-09 11:19:14 192.168.24.187: [1076324.113147] Kernel panic - not 
syncing: stack-protector: Kernel stack is corrupted in: 
radeon_dp_needs_link_train+0x69/0x70 [radeon]
2021-07-09 11:19:14 192.168.24.187: [1076324.113163] CPU: 4 PID: 4109 Comm: 
kworker/4:1 Not tainted 5.12.10+ #11
2021-07-09 11:19:14 192.168.24.187: [1076324.113170] Hardware name: Apple 
Inc. iMac12,2/Mac-XX, BIOS 87.0.0.0.0 06/14/2019
2021-07-09 11:19:14 192.168.24.187: [1076324.113174] Workqueue: events 
radeon_dp_work_func [radeon]
2021-07-09 11:19:14 192.168.24.187: [1076324.113229] Call Trace:
2021-07-09 11:19:14 192.168.24.187: [1076324.113232]  dump_stack+0x64/0x7c
2021-07-09 11:19:14 192.168.24.187: [1076324.113237]  panic+0xf6/0x280
2021-07-09 11:19:14 192.168.24.187: [1076324.113241]  ? 
radeon_dp_needs_link_train+0x69/0x70 [radeon]
2021-07-09 11:19:14 192.168.24.187: [1076324.113267]  
__stack_chk_fail+0x10/0x10
2021-07-09 11:19:14 192.168.24.187: [1076324.113271]  
radeon_dp_needs_link_train+0x69/0x70 [radeon]
2021-07-09 11:19:14 192.168.24.187: [1076324.113297]  
radeon_connector_hotplug+0xa8/0xe0 [radeon]
2021-07-09 11:19:14 192.168.24.187: [1076324.113315]  
radeon_dp_work_func+0x28/0x40 [radeon]
2021-07-09 11:19:14 192.168.24.187: [1076324.113335]  
process_one_work+0x1c4/0x310
2021-07-09 11:19:14 192.168.24.187: [1076324.113339]  
worker_thread+0x240/0x3c0
2021-07-09 11:19:14 192.168.24.187: [1076324.113341]  ? 
wq_update_unbound_numa+0x10/0x10
2021-07-09 11:19:14 192.168.24.187: [1076324.113344]  kthread+0x10a/0x120
2021-07-09 11:19:14 192.168.24.187: [1076324.113346]  ? 
kthread_park+0x80/0x80
2021-07-09 11:19:14 192.168.24.187: [1076324.113348]  
ret_from_fork+0x1f/0x30
2021-07-09 11:19:14 192.168.24.187: [1076324.113391] Kernel Offset: disabled
2021-07-09 11:19:14 192.168.24.187: [1076324.113393] Rebooting in 10 
seconds..
2021-07-09 11:19:24 192.168.24.187: [1076334.114131] ACPI MEMORY or I/O 
RESET_REG.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Err; hardware!

[Brad Campbell via Dng]

On 9/8/21 8:13 am, terryc wrote:
> On Sat, 7 Aug 2021 18:36:19 +0800
> Brad Campbell via Dng  wrote:
> 
>>
>> I honestly can't see who in their right mind would use a base system
>> of that age,
> 
> 
> My 2c, hardware?
> 
> If it has lasted that long, then it must be reliable and its free, why
> not. 
> 
> anyway, back to regular discussions.
> 

I can understand hardware (one of my Colo boxes is a 2008 Mac Mini), but this 
discussion is talking about a new "bare metal hypervisor" based on a Debian 
Wheezy installation.
Keeping legacy systems running? sure, but a new install?? I can't think of any 
system with hardware virtualisation that ran Wheezy and won't run Beowulf.

Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] ..a viable basis for Devuan as a hypervisor?, was: libvirt package without X11 and DBus

[Brad Campbell via Dng]

On 7/8/21 4:25 am, Arnt Karlsen wrote:
> <4f39fd8f-64a5-4488-6640-668d3ceec...@fnarfbargle.com>:
> 
>> I'm still running my self-compiled libvirt because I've progressively
>> upgraded from Debian Wheezy 
> 
> ..dangit, you started precisely where we|Devuan _should_ have started.  
> Too lazy to document it too? ;oD

The scripts are the documentation.

Now, there's a pile of trial and error in finding the latest version of each 
package that would build on wheezy, then there's also the odd base packaged 
that had been upgraded from testing now and then (and I don't have records of 
those).

I honestly can't see who in their right mind would use a base system of that 
age, but hey here's the scripts I used to build libvirt, spice and the rest of 
the stuff.
These were built in a VM, then /usr/local/libvirt was rsynced to a 
staging/testing machine. if it passed all the tests I'd copy it across to the 
production server.

Build script for spice and deps :

download opus-1.3.1 and clone spice and spice gtk from git.

#!/bin/sh
set -e
export PREFIX=/usr/local/libvirt/
export LD_LIBRARY_PATH=$PREFIX/lib
export PKG_CONFIG_PATH=$PREFIX/lib/pkgconfig:$PREFIX/share/pkgconfig

#cd celt-0.5.1.3
cd opus-*
./configure --prefix=/usr/local/libvirt
make -j3
make install
cd ..
cd spice
./go
cd ../spice-gtk
./go



Build script for libvirt and all deps :

#!/bin/bash

export PREFIX=/usr/local/libvirt/
export LD_LIBRARY_PATH=$PREFIX/lib
#export XDG_DATA_DIRS=$PREFIX/share
export PKG_CONFIG_PATH=$PREFIX/lib/pkgconfig:$PREFIX/share/pkgconfig
export LDFLAGS="-I $PREFIX"
export AUTOMAKE_OPTIONS = subdir-objects
CPUS=`egrep '^processor*' /proc/cpuinfo  | wc -l`
CPUS=$((CPUS+1))
set -e 

build () {
if [ -e "$1" ] ; then
echo $1 Already built
else
echo Building $1
tar -xvf archives/$1.tar.*
pushd $1
shift 1
if [ "$1" = 'nopar' ] ; then
shift
else
par='-j'$CPUS
fi;
./configure --prefix=$PREFIX $1
make $par
make install
popd
fi;
}

build-python () {
if [ -e "$1" ] ; then
echo $1 Already built
else
echo Building $1
tar -xvf archives/$1.tar.*
pushd $1
python setup.py build 
python setup.py install --prefix=$PREFIX
popd
fi;
}

build-python2 () {
if [ -e "$1" ] ; then
echo $1 Already built
else
echo Building $1
tar -xvf archives/$1.tar.*
pushd $1
python setup.py configure --prefix=$PREFIX
python setup.py install --prefix=$PREFIX
popd
fi;
}


#for i in archives/*.tar.* ; do tar -xvf $i ; done
#patch -p0 -R < archives/viewer.patch

pushd ../gtk
#./go
popd 

build augeas-*
build libnl-*
build netcf-*
build usbredir*
build vte-*
build lxc* "--disable-bash"
build libvirt-3* "--without-capng"
build-python libvirt-python-*
build libvirt-glib-* "--enable-introspection=no --disable-vala"
build gtk-vnc-* "--with-gtk=2.0"

if [ ! -e spice-done ] ; then
pushd ../spice/
./go
popd
mkdir spice-done
fi;

build virt-viewer-* "--with-gtk=2.0 --disable-update-mimedb --with-yajl"
#build-python2 virt-manager-*
build virt-manager-* 
build-python virtinst-*


Some of these versions might be a bit too new as I was flirting with gtk3 and 
newer deps. I never got that to work, so I'm sticking with what I have.

brad@debian64:~/build/libvirt$ ls archives/
augeas-1.7.0.tar.gz  gtk-vnc-0.6.0.tar.xz   
libvirt-glib-2.0.0.tar.gznew
virt-bootstrap-1.1.1.tar.gz  vte-0.29.1.tar.xz
backup   libnl-3.2.25.tar.gz
libvirt-python-5.6.0.tar.gz  oldvirtinst-0.600.4.tar.gz
config   libsoup-2.40.3.tar.xz  lxc-1.0.5.tar.gz
 usbredir_0.7.orig.tar.bz2  virt-manager-1.5.1.tar.gz
gobject-introspection-0.10.8.tar.gz  libvirt-5.6.0.tar.xz   netcf-0.2.8.tar.gz  
 viewer.patch   virt-viewer-8.0.tar.gz

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] ..a viable basis for Devuan as a hypervisor?, was: libvirt package without X11 and DBus

[Brad Campbell via Dng]

On 7/8/21 5:26 am, Arnt Karlsen wrote:
> On Fri, 6 Aug 2021 20:20:15 +0800, Brad wrote in message 
> :
> 
>> On 6/8/21 5:12 pm, Andrzej Peszynski wrote:
>>>
>>>
>>> On 06.08.2021 06:25, Brad Campbell via Dng wrote:  
>>>>
>>>> Why do you even need/want libvirt? I have several machines which
>>>> run qemu guests just using simple bash scripts to bring them up
>>>> (and all the bash script is there for is to hold the command line
>>>> parameters). I like libvirt and virt-manager for configuring and
>>>> customising the guests, but at the end of the day all that is is a
>>>> fancy front end to qemu. 
>>>   
>>>> If you are stripping the guts out of libvirt, why use it in the
>>>> first place? 
>>> Brad, thanks a million! Learning is fun especially for a "apt
>>> install" man as I am. I am looking now at how I can simplify all
>>> this (may be stripping parts of QEMU too?), to keep running, and
>>> handle my configurations and resources binding. In the end, all
>>> what I need is executing in isolated ring the ELF of dozen of (not
>>> trusted) proxies, servers and libraries + resources balancing +
>>> isolated filesystems + sockets.
>>>
>>> From the other side, I think that the Type 1 hypervisor for desktop
>>> is also interesting thing, It's very tempting to have windowed
>>> multimachine with realtime switch capability.  
> 
> ..I get the idea that Andrzej and I are looking for Brad's kinda bare
> metal hypervisor Devuan install?  
> We might come up with minimal net-install size install image as an 
> alternative to https://en.wikipedia.org/wiki/Qubes_OS ,  only without
> systemd and based on Devuan.

I'm not quite sure how my description of a self-compiled (ancient) libvirt 
install qualifies as "bare metal".

> ..in Debian and Devuan we often have package conflict that means hold
> back upgrades or ditch good software we'd like to keep, those conflicts
> disappears when we can contain each of those old or new things in e.g. 
> a vm.
> 
>> I mostly do that and have done for over 20 years now. 
>> My desktop is essentially a moderately powered thin client (currently
>> a 2011 iMac27 with 2 27" thunderbolt displays running Beowulf). On
>> the server side :
> 
> ..running on that same 2011 vintage iMac27?

Yes, 3 27" displays managed by xmonad. Enough screen realestate to go around.

>> - A Dual head windows 8 VM for Autodesk products & MS Office.
>> - A Dual head 
> 
> ..meaning 2 27" physical displays on top of each client's desk?

Meaning I have 2 27" heads on the VM that display on 2 of my 3 27" displays on 
the desktop.
Being able to use more than one display on a windows VM revolutionized my VM 
use.

>> windows 10 VM for newer Autodesk products that won't
>> run on Windows 8.
>> - A Debian wheezy / xfce VM for a specific older version of
>> Openoffice.
> 
> ..I have some old hw running Debian Sarge thru Wheezy, can I simply 
> yank out those old disks and run them off a vm each?  
> Junk's still useful as guinea pig rigs. ;o)

Might need a little tweak here or there, but sure. I've run VMs off dedicated 
drives. Right now the CCTV server has 4 8TB drives passed straight through.
 
>> - A Windows 8 VM with a CCTV server.
> 
> ..is that running better than e.g. motion or zoneminder, or just 
> based on company policy?

Part of what I do is work with numerous CCTV systems. I've never found anything 
open source that was worth a second look.

>> - A Devuan Ascii / XFCE VM for Peer to peer.
>> - A Devuan Ascii headless VM for Cacti, HLI to the HVAC system and
>> some general development and plumbing
>> - Numerous VMs with specific build configurations for embedded
>> software.
>>
>> The other thing VMs are good for is tying up Scam call-center
>> workers. "My windows is full of viruses you say? And you can help me
>> with that? Brilliant, just let me sit down and start my computer
>> up" (spins up a fresh clean windows VM I prepared earlier)
> 
> ..ooo, and with a nice juicy tarpit too? :o)

Yep, useful for honeypots also.
 
>> Win4lin, bochs and sheepshaver were brilliant, then qemu came and
>> conquered. I can't imagine ever running Windows on the bare metal
>> again.
>>
>> Between spice and rdp, there's not much you can't do.
>>
>> Brad
> 
> ..I get the impression you have what we're looking for here.
> Your 'dpkg --get-selections' would get us started, you also 
> h

Re: [DNG] ..a viable basis for Devuan as a hypervisor?, was: libvirt package without X11 and DBus

[Brad Campbell via Dng]

On 6/8/21 5:12 pm, Andrzej Peszynski wrote:
> 
> 
> On 06.08.2021 06:25, Brad Campbell via Dng wrote:
>>
>> Why do you even need/want libvirt? I have several machines which run qemu 
>> guests just using simple bash scripts to bring them up (and all the bash 
>> script is there for is to hold the command line parameters). I like libvirt 
>> and virt-manager for configuring and customising the guests, but at the end 
>> of the day all that is is a fancy front end to qemu.
>>
> 
>> If you are stripping the guts out of libvirt, why use it in the first place?
>>
> Brad, thanks a million! Learning is fun especially for a "apt install" man as 
> I am.
> I am looking now at how I can simplify all this (may be stripping parts of 
> QEMU too?), to keep running, and handle my configurations and resources 
> binding.
> In the end, all what I need is executing in isolated ring the ELF of dozen of 
> (not trusted) proxies, servers and libraries + resources balancing + isolated 
> filesystems + sockets.
> 
> From the other side, I think that the Type 1 hypervisor for desktop is also 
> interesting thing, It's very tempting to have windowed multimachine with 
> realtime switch capability.

I mostly do that and have done for over 20 years now. 
My desktop is essentially a moderately powered thin client (currently a 2011 
iMac27 with 2 27" thunderbolt displays running Beowulf).
On the server side :
- A Dual head windows 8 VM for Autodesk products & MS Office.
- A Dual head windows 10 VM for newer Autodesk products that won't run on 
Windows 8.
- A Debian wheezy / xfce VM for a specific older version of Openoffice.
- A Windows 8 VM with a CCTV server.
- A Devuan Ascii / XFCE VM for Peer to peer.
- A Devuan Ascii headless VM for Cacti, HLI to the HVAC system and some general 
development and plumbing
- Numerous VMs with specific build configurations for embedded software.

The other thing VMs are good for is tying up Scam call-center workers. "My 
windows is full of viruses you say? And you can help me with that? Brilliant, 
just let me sit down and start my computer up" (spins up a fresh clean windows 
VM I prepared earlier)

Win4lin, bochs and sheepshaver were brilliant, then qemu came and conquered.
I can't imagine ever running Windows on the bare metal again.

Between spice and rdp, there's not much you can't do.

Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] ..a viable basis for Devuan as a hypervisor?, was: libvirt package without X11 and DBus

[Brad Campbell via Dng]

On 6/8/21 1:04 am, AP wrote:



On 05.08.2021 12:37, Arnt Karlsen wrote:



..any of you guys wanting to package what you have running?
To me, this sounds like a viable basis for the bare metal
hypervisor idea in the "[DNG] Devuan as a hypervisor?" thread.


I like the idea of this, if the DEVUAN type1 hypervisor can be avaiable as iso 
- I would use it myself, especially in headless configuration - no X11, no 
DBus, no hot plug, no productivity tools in headless variant.
I could then maintain several pre-installed headless guests, e.g. with horde 
webmail, or letsencrypt ssl/postfix/ldap/dove/bind, web hosting, video/audio 
rtsp/hosting/transcoding, or anything else.

Do I understand correctly, that having configuration variant with QEMU /X11/vnc 
is required, since desktop guests configurations will be supported?


Why do you even need/want libvirt? I have several machines which run qemu 
guests just using simple bash scripts to bring them up (and all the bash script 
is there for is to hold the command line parameters). I like libvirt and 
virt-manager for configuring and customising the guests, but at the end of the 
day all that is is a fancy front end to qemu.

#!/bin/bash
PATH=$PATH:/home/brad/bin
qemu -enable-kvm\
 -m 4096\
 -vga std\
 -vnc :1\
 -no-hpet\
 -global VGA.vgamem_mb=32\
 -rtc base=localtime\
 -usbdevice tablet\
 -cpu host,kvm=off\
 -net tap,ifname=tap0,script=/home/brad/bin/wintap \
 -net nic,model=e1000,macaddr=52:54:00:12:34:58 \
 -device ahci,id=ahci_cont \
 -device ide-drive,bus=ahci_cont.0,drive=HDD \
 -drive id=HDD,if=none,file=Win8.Debugger.qcow2,format=qcow2\
 -device nec-usb-xhci,id=xhci\
 -readconfig /home/brad/qemu/docs/config/ich9-ehci-uhci.cfg \
 -serial tcp::9090,server,nowait \
 -cdrom /media/src/isos/GRMWDK_EN_7600_1.ISO \
 -device usb-host,bus=xhci.0,vendorid=0x0403,productid=0x6001 \
 -device usb-host,bus=xhci.0,vendorid=0x05ac,productid=0x1261 \
 -device usb-host,bus=xhci.0,vendorid=0x04d8,productid=0xe11c \
 -device usb-host,bus=xhci.0,vendorid=0x1a86,productid=0x7523 \

If you are stripping the guts out of libvirt, why use it in the first place?

Brad
--
An expert is a person who has found out by his own painful
experience all the mistakes that one can make in a very
narrow field. - Niels Bohr
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] ..a viable basis for Devuan as a hypervisor?, was: libvirt package without X11 and DBus

[Brad Campbell via Dng]

On 5/8/21 6:37 pm, Arnt Karlsen wrote:
> On Thu, 5 Aug 2021 11:13:06 +0800, Brad wrote in message 
> <184151f6-16e3-f59c-1d07-47394f30f...@fnarfbargle.com>:
> 
>> On 5/8/21 4:40 am, AP wrote:
>>> Hi everyone,
>>>
>>> first I thank all DEVUAN people for the pure pleasure of running my
>>> system (since ASCII 2018) without a bloatware.
>>>
>>> This is my first message and I am sorry, that my search did not
>>> give me the answer about:
>>>
>>> maintenance of the libvirt package without X11 and DBus
>>>
>>> Question: is there a way to get the package for libvirt + QEMU/KVM
>>> for headless VMs - when no X, no DBus needed? 
>>
>> You can always compile it yourself. I've just checked and I'm still
>> running a self-compiled v4 on my main box. Certainly in the v5.6.0
>> code I just looked at there's an option to disable dbus.
>>
>>  From memory I stopped upgrading when they migrated away from
>> autoconf and make. The redhat-isms were just making it too hard to
>> build on older stable debian-based systems, so I stuck with V4. These
>> days it'd be easy enough to use the packaged versions I suppose.
>>
>> Brad
> 
> ..any of you guys wanting to package what you have running?
> To me, this sounds like a viable basis for the bare metal 
> hypervisor idea in the "[DNG] Devuan as a hypervisor?" thread.
> 

Nope. I did it because I needed to if I wanted the bits "I needed" from libvirt 
on a Debian version that was pre-jessie.
I still built with dbus, I just had to also build the right version of dbus.

If I was starting from scratch now, I'd install Devuan Beowulf and apt-get 
install libvirt. I don't get hung up on dependencies and I'm too lazy to want 
to expunge dbus just because I don't understand it and I'm paranoid (I do and 
I'm not). I'm still running my self-compiled libvirt because I've progressively 
upgraded from Debian Wheezy (which is what I compiled it on) to Devuan 
Jessie->Ascii->Beowulf and it hasn't broken. Because it hasn't broken, I 
haven't fixed it. It still lives (with all its dependencies) in 
/usr/local/libvirt.

Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] libvirt package without X11 and DBus

[Brad Campbell via Dng]

On 5/8/21 4:40 am, AP wrote:

Hi everyone,

first I thank all DEVUAN people for the pure pleasure of running my system 
(since ASCII 2018) without a bloatware.

This is my first message and I am sorry, that my search did not give me the 
answer about:

maintenance of the libvirt package without X11 and DBus

Question: is there a way to get the package for libvirt + QEMU/KVM for headless 
VMs - when no X, no DBus needed?



You can always compile it yourself. I've just checked and I'm still running a 
self-compiled v4 on my main box. Certainly in the v5.6.0 code I just looked at 
there's an option to disable dbus.

From memory I stopped upgrading when they migrated away from autoconf and make. 
The redhat-isms were just making it too hard to build on older stable 
debian-based systems, so I stuck with V4. These days it'd be easy enough to use 
the packaged versions I suppose.

Brad
--
An expert is a person who has found out by his own painful
experience all the mistakes that one can make in a very
narrow field. - Niels Bohr
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Ascii to Beowulf upgrade borked with eudev

[Brad Campbell via Dng]

On 3/8/21 1:32 pm, Brad Campbell via Dng wrote:

On 11/5/21 12:57 pm, Brad Campbell via Dng wrote:

G'day all,

I use a self-compiled kernel (v5.10) at the moment.
An Ascii to Beowulf upgrade died early on because the eudev preinst script 
isn't correctly parsing /proc/kallsyms.

Preparing to unpack .../25-eudev_3.2.9-8~beowulf1_amd64.deb ...
Since release 198, udev requires support for the following features in
the running kernel:

- inotify(2)    (CONFIG_INOTIFY_USER)
- signalfd(2)   (CONFIG_SIGNALFD)
- accept4(2)
- open_by_handle_at(2)  (CONFIG_FHANDLE)
- timerfd_create(2) (CONFIG_TIMERFD)
- epoll_create(2)   (CONFIG_EPOLL)
dpkg: error processing archive 
/tmp/apt-dpkg-install-PJCPSM/25-eudev_3.2.9-8~beowulf1_amd64.deb (--unpack):
  new eudev package pre-installation script subprocess returned error exit 
status 1



Just got stung again by this one, this time upgrading an ascii 64 bit rpi3 
image to beowulf while using a 5.10 kernel provided by the rpi foundation.

This time I just unpacked the deb, commented out the "exit 1" and repacked the 
deb, then a bit of manual installation to get the dependencies updated.




After another re-install ascii and upgrade to beowulf to verify I can confirm 
that if you remove or rename /run/udev prior to the dist-upgrade the check gets 
disabled. It's pretty obvious in the pre-inst file, but as it only ever caught 
me in the middle of an upgrade when I was more interested in getting things 
running rather than finessing a work-around I never really looked too hard.

In my case I ran the dist-upgrade until it bombed out, rm -r /run/udev and then 
ran the upgrade again and this time it ran to completion.
A manual /etc/init.d/eudev restart afterwards re-created the directory and we 
are off to the races.

I can't see why I wouldn't just zap /run/udev before the upgrade.

Brad
--
An expert is a person who has found out by his own painful
experience all the mistakes that one can make in a very
narrow field. - Niels Bohr
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Ascii to Beowulf upgrade borked with eudev

[Brad Campbell via Dng]

On 11/5/21 12:57 pm, Brad Campbell via Dng wrote:
> G'day all,
> 
> I use a self-compiled kernel (v5.10) at the moment.
> An Ascii to Beowulf upgrade died early on because the eudev preinst script 
> isn't correctly parsing /proc/kallsyms.
> 
> Preparing to unpack .../25-eudev_3.2.9-8~beowulf1_amd64.deb ...
> Since release 198, udev requires support for the following features in
> the running kernel:
> 
> - inotify(2)    (CONFIG_INOTIFY_USER)
> - signalfd(2)   (CONFIG_SIGNALFD)
> - accept4(2)
> - open_by_handle_at(2)  (CONFIG_FHANDLE)
> - timerfd_create(2) (CONFIG_TIMERFD)
> - epoll_create(2)   (CONFIG_EPOLL)
> dpkg: error processing archive 
> /tmp/apt-dpkg-install-PJCPSM/25-eudev_3.2.9-8~beowulf1_amd64.deb (--unpack):
>  new eudev package pre-installation script subprocess returned error exit 
> status 1
>

Just got stung again by this one, this time upgrading an ascii 64 bit rpi3 
image to beowulf while using a 5.10 kernel provided by the rpi foundation.

This time I just unpacked the deb, commented out the "exit 1" and repacked the 
deb, then a bit of manual installation to get the dependencies updated.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Beowulf; Increasing the number of boot options under grub. How To?

[Brad Campbell via Dng]

On 17/5/21 9:31 pm, terryc wrote:
> I am hoping there is a simple config/number change in the grub
> config files, but I can not find it in the doco I've read.
> 
> In the past, the list of kernel images just grew each time you
> loaded/updated the linux-image files. But the current system
> automatically trims it to just two, with normal and recovery versions of
> each.
>
I'm pretty sure you want to enable : GRUB_DISABLE_SUBMENU in /etc/default/grub 
before doing an update-grub.
Backup your grub.cfg, make the change and then diff with the new one to see if 
it does what you want before rebooting.

I can't verify that locally anymore as I've moved all my machines to boot with 
rEFInd, but I used grub for years.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Ascii to Beowulf upgrade borked with eudev

[Brad Campbell via Dng]

On 11/5/21 2:36 pm, Didier Kryn wrote:

Le 11/05/2021 à 06:57, Brad Campbell via Dng a écrit :

G'day all,

I use a self-compiled kernel (v5.10) at the moment.
An Ascii to Beowulf upgrade died early on because the eudev preinst
script isn't correctly parsing /proc/kallsyms.

Preparing to unpack .../25-eudev_3.2.9-8~beowulf1_amd64.deb ...
Since release 198, udev requires support for the following features in
the running kernel:

- inotify(2)    (CONFIG_INOTIFY_USER)
- signalfd(2)   (CONFIG_SIGNALFD)
- accept4(2)
- open_by_handle_at(2)  (CONFIG_FHANDLE)
- timerfd_create(2) (CONFIG_TIMERFD)
- epoll_create(2)   (CONFIG_EPOLL)
dpkg: error processing archive
/tmp/apt-dpkg-install-PJCPSM/25-eudev_3.2.9-8~beowulf1_amd64.deb
(--unpack):
  new eudev package pre-installation script subprocess returned error
exit status 1

root@bkmac:/tmp/temp# needed_symbols='inotify_init signalfd accept4
open_by_handle_at timerfd_create epoll_create'

root@bkmac:/tmp/temp# for symbol in $needed_symbols; do echo $symbol ;
egrep "^[a-fA-F0-9]+ T \.?sys_${symbol}$" /proc/kallsyms ; done
inotify_init
signalfd
accept4
open_by_handle_at
timerfd_create
epoll_create

root@bkmac:/tmp/temp# for symbol in $needed_symbols; do echo $symbol ;
egrep "^[a-fA-F0-9]+ T __x64.sys_${symbol}$" /proc/kallsyms ; done
inotify_init
811ac000 T __x64_sys_inotify_init
signalfd
811af380 T __x64_sys_signalfd
accept4
813df4a0 T __x64_sys_accept4
open_by_handle_at
811d3410 T __x64_sys_open_by_handle_at
timerfd_create
811b00c0 T __x64_sys_timerfd_create
epoll_create
811ad9b0 T __x64_sys_epoll_create

I unpacked, patched out the check and re-packed the deb, then manually
installed it.
Unfortunately for me (yay) the upgrade then proceeded to break
horribly with an apt-get -f install resulting it in wanting to
uninstall X and everything associated with it, requiring severe manual
intervention to get a functioning system again.

root@bkmac:/var/cache/apt/archives# apt-get -u dist-upgrade
Reading package lists... Done
Building dependency tree
Reading state information... Done
You might want to run 'apt --fix-broken install' to correct these.
The following packages have unmet dependencies:
  cups-daemon : Depends: libcups2 (= 2.2.10-6+deb10u4) but
2.2.1-8+deb9u6 is installed
  cups-ipp-utils : Depends: libcups2 (= 2.2.10-6+deb10u4) but
2.2.1-8+deb9u6 is installed
  libcupsimage2 : Depends: libcups2 (= 2.2.10-6+deb10u4) but
2.2.1-8+deb9u6 is installed
  libeudev-dev : Depends: libeudev1 (= 3.2.2-13) but 3.2.9-8~beowulf1
is installed
  libreoffice-java-common : Depends: libreoffice-common (=
1:6.1.5-3+deb10u7) but 1:5.2.7-1+deb9u11 is installed
  libreoffice-style-colibre : Depends: libreoffice-common (=
1:6.1.5-3+deb10u7) but 1:5.2.7-1+deb9u11 is installed
  libusb-1.0-0-dev : Depends: libusb-1.0-0 (= 2:1.0.22-2) but
2:1.0.21-1 is installed
  python3-distutils : Depends: python3 (>= 3.7.1-1~) but 3.5.3-1 is
installed
  python3-lib2to3 : Depends: python3 (>= 3.7.1-1~) but 3.5.3-1 is
installed
E: Unmet dependencies. Try 'apt --fix-broken install' with no packages
(or specify a solution).

I got there, but there was a lot of manual dpkg -i from the cache
directory to get things back on track.

Just a headsup.

Brad



     Hello Brad.

     I'd bet you disabled - or failed to enable - one of the mentionned
kernel features during the config phase of your kernel build. So it
would just be a matter of carefully configuring your kernel build, not
reusing an old config file.



G'day Didier,

I did neither. Here's the relevant bit again where I modified the regex to 
match the output from the v5 kernel indicating that all required symbols are 
present and accounted for :

root@bkmac:/tmp/temp# needed_symbols='inotify_init signalfd accept4 
open_by_handle_at timerfd_create epoll_create'
root@bkmac:/tmp/temp# for symbol in $needed_symbols; do echo $symbol ; egrep 
"^[a-fA-F0-9]+ T __x64.sys_${symbol}$" /proc/kallsyms ; done
inotify_init
811ac000 T __x64_sys_inotify_init
signalfd
811af380 T __x64_sys_signalfd
accept4
813df4a0 T __x64_sys_accept4
open_by_handle_at
811d3410 T __x64_sys_open_by_handle_at
timerfd_create
811b00c0 T __x64_sys_timerfd_create
epoll_create
811ad9b0 T __x64_sys_epoll_create

Somewhere between the kernel the eudev package was built for and the kernel I'm 
running, the output of /proc/kallsyms changed. It now contains the architecture 
prepended to the symbol name, and the eudev preinst script can't parse it. 
eudev runs perfectly well on the kernel.

For example :
brad@bkmac:~$ grep sys_open_by_handle_at /proc/kallsyms
 T __x64_sys_open_by_handle_at
 T __ia32_sys_open_by_handle_at
 T __ia32_compat_sys_open_by_handle_at

It matches what I found in this bug : 
https://bugs.devuan.o

[DNG] Ascii to Beowulf upgrade borked with eudev

[Brad Campbell via Dng]

G'day all,

I use a self-compiled kernel (v5.10) at the moment.
An Ascii to Beowulf upgrade died early on because the eudev preinst script 
isn't correctly parsing /proc/kallsyms.

Preparing to unpack .../25-eudev_3.2.9-8~beowulf1_amd64.deb ...
Since release 198, udev requires support for the following features in
the running kernel:

- inotify(2)(CONFIG_INOTIFY_USER)
- signalfd(2)   (CONFIG_SIGNALFD)
- accept4(2)
- open_by_handle_at(2)  (CONFIG_FHANDLE)
- timerfd_create(2) (CONFIG_TIMERFD)
- epoll_create(2)   (CONFIG_EPOLL)
dpkg: error processing archive 
/tmp/apt-dpkg-install-PJCPSM/25-eudev_3.2.9-8~beowulf1_amd64.deb (--unpack):
 new eudev package pre-installation script subprocess returned error exit 
status 1

root@bkmac:/tmp/temp# needed_symbols='inotify_init signalfd accept4 
open_by_handle_at timerfd_create epoll_create'

root@bkmac:/tmp/temp# for symbol in $needed_symbols; do echo $symbol ; egrep 
"^[a-fA-F0-9]+ T \.?sys_${symbol}$" /proc/kallsyms ; done
inotify_init
signalfd
accept4
open_by_handle_at
timerfd_create
epoll_create

root@bkmac:/tmp/temp# for symbol in $needed_symbols; do echo $symbol ; egrep 
"^[a-fA-F0-9]+ T __x64.sys_${symbol}$" /proc/kallsyms ; done
inotify_init
811ac000 T __x64_sys_inotify_init
signalfd
811af380 T __x64_sys_signalfd
accept4
813df4a0 T __x64_sys_accept4
open_by_handle_at
811d3410 T __x64_sys_open_by_handle_at
timerfd_create
811b00c0 T __x64_sys_timerfd_create
epoll_create
811ad9b0 T __x64_sys_epoll_create

I unpacked, patched out the check and re-packed the deb, then manually 
installed it.
Unfortunately for me (yay) the upgrade then proceeded to break horribly with an 
apt-get -f install resulting it in wanting to uninstall X and everything 
associated with it, requiring severe manual intervention to get a functioning 
system again.

root@bkmac:/var/cache/apt/archives# apt-get -u dist-upgrade
Reading package lists... Done
Building dependency tree
Reading state information... Done
You might want to run 'apt --fix-broken install' to correct these.
The following packages have unmet dependencies:
 cups-daemon : Depends: libcups2 (= 2.2.10-6+deb10u4) but 2.2.1-8+deb9u6 is 
installed
 cups-ipp-utils : Depends: libcups2 (= 2.2.10-6+deb10u4) but 2.2.1-8+deb9u6 is 
installed
 libcupsimage2 : Depends: libcups2 (= 2.2.10-6+deb10u4) but 2.2.1-8+deb9u6 is 
installed
 libeudev-dev : Depends: libeudev1 (= 3.2.2-13) but 3.2.9-8~beowulf1 is 
installed
 libreoffice-java-common : Depends: libreoffice-common (= 1:6.1.5-3+deb10u7) 
but 1:5.2.7-1+deb9u11 is installed
 libreoffice-style-colibre : Depends: libreoffice-common (= 1:6.1.5-3+deb10u7) 
but 1:5.2.7-1+deb9u11 is installed
 libusb-1.0-0-dev : Depends: libusb-1.0-0 (= 2:1.0.22-2) but 2:1.0.21-1 is 
installed
 python3-distutils : Depends: python3 (>= 3.7.1-1~) but 3.5.3-1 is installed
 python3-lib2to3 : Depends: python3 (>= 3.7.1-1~) but 3.5.3-1 is installed
E: Unmet dependencies. Try 'apt --fix-broken install' with no packages (or 
specify a solution).

I got there, but there was a lot of manual dpkg -i from the cache directory to 
get things back on track.

Just a headsup.

Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] exim4 packages 4.92-8+deb10u6 in beowulf-security?

[Brad Campbell via Dng]

On 8/5/21 11:13 am, Gastón via Dng wrote:
> On Sat, May 08, 2021 at 10:20:23AM +0800, Brad Campbell via Dng wrote:
>> On 7/5/21 2:21 pm, Thomas Besser via Dng wrote:
>>> Am 06.05.21 um 20:53 schrieb Ludovic Bellière:
>>>> You mean this[1] package?
>>>>
>>>> [1]: 
>>>> https://pkginfo.devuan.org/cgi-bin/package-query.html?c=package&q=exim4=4.92-8+deb10u6
>>>
>>> It's a platform independent ('all') package.
>>>
>>> Look at 
>>> https://pkginfo.devuan.org/cgi-bin/policy-query.html?c=package&q=exim4-base 
>>> and you will see, that beowulf-security for e.g. amd64 stays on 
>>> "4.92-8+deb10u5"
>>>
>>> After my mail yesterday I got reply on IRC, that there seems to be a 
>>> problem in amprolla with that and a dev will look into this as soon as 
>>> possible.
>>>
>>
>> So I can see this in beowulf-proposed-updates now. Thanks for pushing this 
>> through.
>>
>> Showing that while I've been using Debian based distributions for 25 years 
>> I'm still a bit behind on anything other than install/upgrade/remove/purge. 
>> I couldn't get it to install automatically after adding proposed-updates to 
>> my sources.list, but I manually installed it as follows :
>>
>> apt-get -u upgrade exim4-base=4.92-8+deb10u6 exim4=4.92-8+deb10u6 
>> exim4-daemon-heavy=4.92-8+deb10u6 exim4-config=4.92-8+deb10u6
>>
>> Is there a better way to be able to install the required packages without 
>> having to individually name each one and the version?
> 
> yes, you can try to do it like this:
> 
> apt-get install exim4 -t beowulf-proposed-updates

Much obliged. I had to call out the exim4 packages on the server or it wasn't 
going to upgrade the server :

root@xxx:~# apt-get install exim4 -t beowulf-proposed-updates
Reading package lists... Done
Building dependency tree   
Reading state information... Done
The following additional packages will be installed:
  exim4-base
Suggested packages:
  exim4-doc-html | exim4-doc-info eximon4 mail-reader spf-tools-perl swaks
Recommended packages:
  mailx
The following packages will be upgraded:
  exim4 exim4-base
2 upgraded, 0 newly installed, 0 to remove and 4 not upgraded.

so on this machine I had to do this :

root@xxx:~# apt-get install exim4 exim4-base exim4-config exim4-daemon-light -t 
beowulf-proposed-updates
Reading package lists... Done
Building dependency tree   
Reading state information... Done
Suggested packages:
  exim4-doc-html | exim4-doc-info eximon4 mail-reader spf-tools-perl swaks
Recommended packages:
  mailx
The following packages will be upgraded:
  exim4 exim4-base exim4-config exim4-daemon-light
4 upgraded, 0 newly installed, 0 to remove and 2 not upgraded.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] exim4 packages 4.92-8+deb10u6 in beowulf-security?

[Brad Campbell via Dng]

On 7/5/21 2:21 pm, Thomas Besser via Dng wrote:
> Am 06.05.21 um 20:53 schrieb Ludovic Bellière:
>> You mean this[1] package?
>>
>> [1]: 
>> https://pkginfo.devuan.org/cgi-bin/package-query.html?c=package&q=exim4=4.92-8+deb10u6
> 
> It's a platform independent ('all') package.
> 
> Look at 
> https://pkginfo.devuan.org/cgi-bin/policy-query.html?c=package&q=exim4-base 
> and you will see, that beowulf-security for e.g. amd64 stays on 
> "4.92-8+deb10u5"
> 
> After my mail yesterday I got reply on IRC, that there seems to be a problem 
> in amprolla with that and a dev will look into this as soon as possible.
> 

So I can see this in beowulf-proposed-updates now. Thanks for pushing this 
through.

Showing that while I've been using Debian based distributions for 25 years I'm 
still a bit behind on anything other than install/upgrade/remove/purge. I 
couldn't get it to install automatically after adding proposed-updates to my 
sources.list, but I manually installed it as follows :

apt-get -u upgrade exim4-base=4.92-8+deb10u6 exim4=4.92-8+deb10u6 
exim4-daemon-heavy=4.92-8+deb10u6 exim4-config=4.92-8+deb10u6

Is there a better way to be able to install the required packages without 
having to individually name each one and the version?

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Odd issue with busybox dc in beowulf

[Brad Campbell via Dng]

On 29/4/21 8:14 pm, tito via Dng wrote:

> Hi,
> by looking at the latest git code:
> 
> static const struct op operators[] ALIGN_PTR = {
> #if ENABLE_FEATURE_DC_LIBM
>   {"^",   power},
> //{"exp", power},
> //{"pow", power},
> #endif
>   {"%",   mod},
> //{"mod", mod},
>   // logic ops are not standard, remove?
>   {"and", and},
>   {"or",  or},
>   {"not", not},
>   {"xor", eor},
>   {"+",   add},
> //{"add", add},
>   {"-",   sub},
> //{"sub", sub},
>   {"*",   mul},
> //{"mul", mul},
>   {"/",   divide},
> //{"div", divide},
>   {"p", print_no_pop},
>   {"f", print_stack_no_pop},
>   {"o", set_output_base},
> };
> 
> it seems to me that mod, add, sub, mul, div are disabled
> and only %, +, -, *,  / are supported.
> Cannot say if simply uncommenting  them restores
> the previous functionality, could be worth a try.
> Eventually if it works a patch for making them optional
> (CONFIG_DC_LONG_OPS or the like) could be sent
> to the list.
> 
> Hope this helps, 

It did. Thanks to all that replied.

This bit of code runs in the initramfs and is used to unlock ATA locked drives 
using keys from the network.
If it fails, the machine can't unlock the drives. This prompted me to take the 
time to re-write the arithmetic 
in a POSIX compatible shell to take dc out of the equation altogether. I can't 
risk it breaking again down the track.

I still never got to the bottom of it printing the top of the stack as a 
character rather than a number.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

[DNG] Odd issue with busybox dc in beowulf

[Brad Campbell via Dng]

G'day All,

I've upgraded a staging server from Jessie to Beowulf and find a script in my 
initramfs is now broken, tracking it down it is a huge change in behaviour in 
the busybox version of dc and I can't find any reference to what I'm missing. 
Has anyone bumped up against this? I've tried this on both the arm and x64 
versions and the behaviour is identical, so it's not an arm thing specifically.

On jessie :
brad@srv:~$ busybox dc
2
2
add
p
4

On beowulf :
root@rpi31:~# busybox dc
2
2
add
p


What you can't see at the bottom of the last example is the rendering of the 
0x02 character in the shell.

The input processor also appears to have changed.

brad@srv:~$ busybox dc
0x127
p
295

root@rpi31:~# busybox dc
0x127
p
127

Installed packages.

root@rpi31:~# apt-cache showpkg busybox
Package: busybox
Versions:
1:1.30.1-4 
(/var/lib/apt/lists/deb.devuan.org_merged_dists_beowulf_main_binary-armhf_Packages)
 (/var/lib/dpkg/status)
 Description Language:
 File: 
/var/lib/apt/lists/deb.devuan.org_merged_dists_beowulf_main_binary-armhf_Packages
  MD5: b7707908219c331294f3f9e8d926a9dc
 Description Language: en
 File: 
/var/lib/apt/lists/deb.devuan.org_merged_dists_beowulf_main_i18n_Translation-en
  MD5: b7707908219c331294f3f9e8d926a9dc


Reverse Depends:
  initramfs-tools-core,busybox 1:1.22.0-17~
  zfs-initramfs,busybox
  udhcpd,busybox 1:1.30.1
  udhcpc,busybox 1:1.30.1
  open-iscsi,busybox
  open-infrastructure-system-boot,busybox
  live-boot-initramfs-tools,busybox
  initramfs-tools-core,busybox 1:1.22.0-17~
  busybox-syslogd,busybox 1:1.30.1
  dropbear-initramfs,busybox
  cryptsetup-initramfs,busybox
  bootcd,busybox
  busybox-static,busybox
  busybox-static,busybox
Dependencies:
1:1.30.1-4 - libc6 (2 2.28) busybox-static (0 (null)) initramfs-tools (3 0.99) 
busybox-static (0 (null))
Provides:
1:1.30.1-4 -
Reverse Provides:
busybox-static 1:1.30.1-4 (= )


brad@srv:~$ apt-cache showpkg busybox
Package: busybox
Versions:
1:1.22.0-9+deb8u4 (/var/lib/dpkg/status)
 Description Language:
 File: 
/var/lib/apt/lists/archive.devuan.org_merged_dists_jessie_main_binary-amd64_Packages
  MD5: b7707908219c331294f3f9e8d926a9dc
 Description Language: en
 File: 
/var/lib/apt/lists/archive.devuan.org_merged_dists_jessie_main_i18n_Translation-en
  MD5: b7707908219c331294f3f9e8d926a9dc

1:1.22.0-9+deb8u1 
(/var/lib/apt/lists/archive.devuan.org_merged_dists_jessie_main_binary-amd64_Packages)
 Description Language:
 File: 
/var/lib/apt/lists/archive.devuan.org_merged_dists_jessie_main_binary-amd64_Packages
  MD5: b7707908219c331294f3f9e8d926a9dc
 Description Language: en
 File: 
/var/lib/apt/lists/archive.devuan.org_merged_dists_jessie_main_i18n_Translation-en
  MD5: b7707908219c331294f3f9e8d926a9dc


Reverse Depends:
  udhcpd,busybox 1:1.22.0
  udhcpc,busybox 1:1.22.0
  live-boot-initramfs-tools,busybox
  initramfs-tools,busybox 1:1.01-3
  initramfs-tools,busybox 1:1.01-3
  cryptsetup,busybox
  busybox-syslogd,busybox 1:1.22.0
  busybox-static,busybox
  busybox-static,busybox
  bootcd,busybox
Dependencies:
1:1.22.0-9+deb8u4 - libc6 (2 2.16) busybox-static (0 (null)) initramfs-tools (3 
0.99) busybox-static (0 (null))
1:1.22.0-9+deb8u1 - libc6 (2 2.16) busybox-static (0 (null)) initramfs-tools (3 
0.99) busybox-static (0 (null))
Provides:
1:1.22.0-9+deb8u4 -
1:1.22.0-9+deb8u1 -
Reverse Provides:
busybox-static 1:1.22.0-9+deb8u1

Am I doing something dumb?

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Your system is not supported by certbot-auto anymore.

[Brad Campbell via Dng]

On 8/12/20 5:02 pm, Martin Steigerwald wrote:


I am still using dehydrated. It is a simple shell script which just
depends on curl, openssl and ca-certificates. There is an additional
package for apache2 support, which just contains the site configuration
for the web challenge thing, and one for DNS challenge.

I think there is an alternative to it, called acme.sh. I never looked
into it.


I use acme.sh with dns challenges. Does what it says on the tin and 
didn't need any complex deps.


For the rest of the thread :
 +1 mythic-beasts for co-lo and vps hosting and uf.r4l.com for domain 
registration since 2006. I run my own dns servers though.


I don't particularly like letsencrypt, but for installing certs to keep 
stuff happy it does the job for now. Most of my stuff uses self-signed 
where I can get away with it.


Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Can this drive be saved?

[Brad Campbell via Dng]

On 10/9/20 2:04 pm, Simon Walter wrote:

On 2020-09-09 15:53, Brad Campbell via Dng wrote:


It really doesn't. It'll mark a sector as "pending" (as in, I can't read
from it so I'll mark it for later).


What does the OS get at this point? Is that a short read error?


Yep. I have some old drives here I keep around for testing :

current smart excerpt :


196 Reallocated_Event_Count 0x0032   200   200   000Old_age   Always   
-   0
197 Current_Pending_Sector  0x0032   200   001   000Old_age   Always   
-   1
198 Offline_Uncorrectable   0x0030   200   200   000Old_age   Offline  
-   0



SMART Self-test log structure revision number 1
Num  Test_DescriptionStatus  Remaining  LifeTime(hours)  
LBA_of_first_error
# 1  Short offline   Completed: read failure   80% 59372 
1099436031

root@test:~/smartdir# dd if=/dev/sdm bs=512 count=1 skip=1099436031
dd: error reading ‘/dev/sdm’: Input/output error
0+0 records in
0+0 records out
0 bytes (0 B) copied, 5.56285 s, 0.0 kB/s

root@test:~/smartdir# dmesg | tail -n11
[  122.928898] sd 1:0:7:0: [sdm] tag#30 UNKNOWN(0x2003) Result: hostbyte=0x00 
driverbyte=0x08 cmd_age=2s
[  122.928974] sd 1:0:7:0: [sdm] tag#30 Sense Key : 0x3 [current]
[  122.929035] sd 1:0:7:0: [sdm] tag#30 ASC=0x11 ASCQ=0x0
[  122.929094] sd 1:0:7:0: [sdm] tag#30 CDB: opcode=0x28 28 00 41 88 0f f8 00 
00 08 00
[  122.929159] blk_update_request: critical medium error, dev sdm, sector 
1099436024 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 0
[  125.712897] sd 1:0:7:0: [sdm] tag#8 UNKNOWN(0x2003) Result: hostbyte=0x00 
driverbyte=0x08 cmd_age=2s
[  125.712975] sd 1:0:7:0: [sdm] tag#8 Sense Key : 0x3 [current]
[  125.713034] sd 1:0:7:0: [sdm] tag#8 ASC=0x11 ASCQ=0x0
[  125.713092] sd 1:0:7:0: [sdm] tag#8 CDB: opcode=0x28 28 00 41 88 0f f8 00 00 
08 00
[  125.713156] blk_update_request: critical medium error, dev sdm, sector 
1099436024 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0
[  125.713229] Buffer I/O error on dev sdm, logical block 137429503, async page 
read

This drive just grows defects. I can zap that sector and in a couple of days 
it'll come up with another one.

Oddly enough, it's either unreliable electronics or flat out lying as I've 
zapped plenty of pending sectors on this disk :

  5 Reallocated_Sector_Ct   0x0033   200   200   140Pre-fail  Always   
-   0



I hope I didn't make it sound like that. SMART does not include
prediction. It's just data, and it needs be interpreted. I should have
probably said that I have never had a drive fail without being warned by
my monitoring system (which includes logging SMART data).


Right. Sorry, it sounded like you were relying on the SMART (Good/Bad) 
prediction which is notoriously terrible.
I've had drives that were practically in their death throes which SMART was quite happy 
to report as "Good".

On 9/9/20 11:00 pm, Hendrik Boom wrote:


Let me wonder how the drive knows a sector is bad when it writes to it.
Does it read it back as a check?



All I know is the drive will push a media error up the stack when there is a 
problem writing. I've never really thought about *how* it knows it's a bum 
write. I suppose a read-back is the only way I can think of achieving it.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Can this drive be saved?

[Brad Campbell via Dng]

On 5/9/20 10:38 pm, Simon Walter wrote:

On 9/5/20 12:50 PM, Gregory Nowak wrote:

On Sat, Sep 05, 2020 at 12:26:21PM +0900, Simon Walter wrote:

Reallocation, to my knowledge, should happen in the background. It's
*possible* that the reallocation event and the FS corruption are unrelated.


My understanding is that the drive won't attempt to reallocate a
sector until that sector is written to. So, if the e2fsck -f did try
to write to that sector, the drive did reallocate it in the
background. I do stand to be corrected as always.



Interesting. I think reallocation also happens as part of SMART self checks and 
reads.


It really doesn't. It'll mark a sector as "pending" (as in, I can't read from 
it so I'll mark it for later).

You can try reading from the sector repeatedly and if you get a good read, 
write it back. Spinrite does that and it has about an even-odds success rate. 
It's old, clunky, slow and it's capabilities are very over-sold however. It's 
really a one trick pony.

A drive will absolutely not re-allocate a sector until it is written to. Then 
it will first try and re-write the sector in case the error is due to something 
like a power loss or other transient. Failing a good write it'll then 
reallocate and write the data to another (transparent) location.

The issue with a SMART long test (or any SMART media test in fact) is it'll abort on the 
first bad sector. If you really want to read every sector and mark *all* the duds as 
pending then you need something like dd conv=noerror with a blocksize of whatever the 
smallest sector the disk supports (4k for most SATA spinners). That'll chew through the 
sectors and the drive will flag every bad read as "pending".

I have drives that have > 70,000h on them with one or two reallocated sectors. 
I've also had drives grow them at a rapid rate. SMART isn't all that good at 
actually predicting pending failure. Analysis of the raw data on a large 
population of the same disk is better, but the best is to avoid the worry with 
up-to-date backups.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Can this drive be saved?

[Brad Campbell via Dng]

On 30/8/20 8:19 pm, Hendrik Boom wrote:

There's also the badblocks program, which can be set up either to do a
nondestructive read-only test for bad blocks, or a more throrough
destructive test, where it writes every block and later checks tht it
can read it correctly again, using a variety of bit patterns.

That might make the hard drive reassign the bad blocks on its own.

If the hard drive doesn't do that, 


 then it's ready for the bin.

If the drive can't reallocate sectors it's due to a catastrophic failure and 
means it's out of spare sectors.
Drives come from the factory with "lots" of spare sectors, so if it's out then 
the drive has both feet, 8 fingers and a thumb in the grave.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Can this drive be saved?

[Brad Campbell via Dng]

On 5/9/20 10:38 pm, Simon Walter wrote:

On 9/5/20 12:50 PM, Gregory Nowak wrote:

On Sat, Sep 05, 2020 at 12:26:21PM +0900, Simon Walter wrote:

Reallocation, to my knowledge, should happen in the background. It's
*possible* that the reallocation event and the FS corruption are unrelated.


My understanding is that the drive won't attempt to reallocate a
sector until that sector is written to. So, if the e2fsck -f did try
to write to that sector, the drive did reallocate it in the
background. I do stand to be corrected as always.



Interesting. I think reallocation also happens as part of SMART self checks and 
reads.


It really doesn't. It'll mark a sector as "pending" (as in, I can't read from 
it so I'll mark it for later).

You can try reading from the sector repeatedly and if you get a good read, 
write it back. Spinrite does that and it has about an even-odds success rate. 
It's old, clunky, slow and it's capabilities are very over-sold however. It's 
really a one trick pony.

A drive will absolutely not re-allocate a sector until it is written to. Then 
it will first try and re-write the sector in case the error is due to something 
like a power loss or other transient. Failing a good write it'll then 
reallocate and write the data to another (transparent) location.

The issue with a SMART long test (or any SMART media test in fact) is it'll abort on the 
first bad sector. If you really want to read every sector and mark *all* the duds as 
pending then you need something like dd conv=noerror with a blocksize of whatever the 
smallest sector the disk supports (4k for most SATA spinners). That'll chew through the 
sectors and the drive will flag every bad read as "pending".

I have drives that have > 70,000h on them with one or two reallocated sectors. 
I've also had drives grow them at a rapid rate. SMART isn't all that good at 
actually predicting pending failure. Analysis of the raw data on a large 
population of the same disk is better, but the best is to avoid the worry with 
up-to-date backups.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Raspberry Pi 4 (Brad Campbell)

[Brad Campbell via Dng]

On 6/7/19 21:57, . via Dng wrote:
I tried what you've described --- put a fresh copy of 
"devuan_ascii_2.0.0_arm64_raspi3.img" on an 8GB microSD card, then 
copied the parts of Raspbian Buster over; but I end up with an error 
"/sbin/init exists, but is not executable (error -8)", possibly 
suggesting that there is a mismatch between 32-bit boot code and 64-bit 
/sbin/init.


Did you make any other changes?



No, but mine is running Jessie not ASCII.

root@rpi31:~# cat /etc/devuan_version
jessie
root@rpi31:~# file /lib/arm-linux-gnueabihf/ld-2.19.so
/lib/arm-linux-gnueabihf/ld-2.19.so: ELF 32-bit LSB shared object, ARM, 
EABI5 version 1 (SYSV), dynamically linked, 
BuildID[sha1]=b40c3daf67d7bfe1c1ab9b7686326406622ca5d9, stripped

root@rpi31:~# file /bin/bash
/bin/bash: ELF 32-bit LSB executable, ARM, EABI5 version 1 (SYSV), 
dynamically linked, interpreter /lib/ld-linux-armhf.so.3, for GNU/Linux 
2.6.32, BuildID[sha1]=90132c73f3c9d4dacfda87128daa35cb6618f4e7, stripped

root@rpi31:~# file /sbin/init
/sbin/init: ELF 32-bit LSB executable, ARM, EABI5 version 1 (SYSV), 
dynamically linked, interpreter /lib/ld-linux-armhf.so.3, for GNU/Linux 
2.6.32, BuildID[sha1]=aeb9f58211f7ae4d12612d6b992fcbbb652460e2, stripped


So that could well be the issue.

I have another one arriving next week, so I'll have a play with that and 
an ASCII image.


Regards,
--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

[DNG] Raspberry Pi 4

[Brad Campbell via Dng]

G'day all,

I have an old RPI3 running Jessie that I wanted to replace with a 4 
(wanted USB3 & GbE)


Not wanting to re-configure or re-install I downloaded the latest 
Raspbian for the Pi4 (Buster) to use for parts.


I replaced the boot partition contents straight from the Buster image. 
Had to edit cmdline text to alter the root parameter "root=/dev/mmcblk0p2".


Then I simply rsynced /lib/modules and /lib/firmware across to the root 
partition (adding not replacing).


It booted right up with USB, net and console functional and I was back 
in business on the new hardware.


I don't use the GPU except for initial bringup, nor wlan or bluetooth. 
So I can't vouch for those, although udev certainly detects the wlan.


The config.txt lines to disable wlan & bt still work also :

dtoverlay=pi3-disable-bt
dtoverlay=pi3-disable-wifi

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] PostScript vs PCL

[Brad Campbell]

On 20/03/18 14:27, Tom wrote:

Thanks for the replies.  I'm currently using the default ColorMFP PPD
driver for a Toshiba e-Studio 3555C printer/copier, and it has a habit
of taking 5+ minutes to print a single page.  Would I get better
results with using a generic PS or PCL driver?  What is the
recommended generic driver for PS or PCL these days?

I use whatever comes with CUPS.

I have 2 fussy printers here, both with default CUPS PS drivers. A HP 
LJ2200 and a CLJ5500. I've had issues with both when printing overly 
complex PS, and I've had it happen often enough that I have a separate 
additional print queue set up for each.


The 2200 has a queue using foomatic/pxlmono and the 5500 has a queue 
using foomatic/pxlcolor.


Both of these drivers have limitations in the printer features available 
and therefore only see use when the printer is unable to handle the PS 
print job. Rendering is faster on the server however and on some old 
huge raster PDF files I had the displeasure of working with the printer 
was taking 5-15 minutes to print a single page. Using the pxl driver 
that came down to as fast as the old JetDirect card could shovel data 
into the printer.


Setting up multiple drivers is a neat way to A/B test and work around 
any real oddities in the printers RIP implementations.


I still have nightmares over getting Windows clients to print to a Canon 
colour copier with external Unix RIP using Samba and lpd as an 
intermediary back in the 90's.


Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Preserving rsync: was systemd local privilege escalation (CVE-2017-18078)

[Brad Campbell]

On 08/02/18 02:02, Arnt Karlsen wrote:


..ok, I'll bite, I have rsync.git and rsync-web.git cloned,
how do I clone rsync-patches.git?



git clone git://git.samba.org/rsync-patches.git


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Preserving rsync: was systemd local privilege escalation (CVE-2017-18078)

[Brad Campbell]

On 07/02/18 07:10, Steve Litt wrote:


According to https://git.samba.org/?p=rsync.git;a=shortlog , it looks
to me like poettering has never committed to rsync, so that's good
news. LOL, I had to take some flack in order to get that URL from the
Samba list, but it's worth it.


I'm not trying to be rude Steve, but that post to the SAMBA mailing list 
kinda makes you come off like a raving lunatic. I thought they let you 
off relatively lightly.


A quick google for the RSYNC repository, and a quick browse of the tree 
would have answered your question without even bothering the SAMBA list. 
Not only that, but a git clone would give you a complete and unmolested 
archive up to the minute.



Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Help with Spectre and Meltdown

[Brad Campbell]

On 17/01/18 23:00, jacksprat wrote:
Thanks for the replies.  When I run the "spectre-meltdown-checker.sh" 
script [github.com/speed47 ] I see that even 
using a recent kernel [4.15-rc8] only Meltdown is covered.


The two mitigations for Spectre [IBRS or kernel compiled with 
"retpoline" option with a retpoline-aware compiler] are harder for me to 
achieve.  The latter requires a retpoline-aware version of gcc - did 
anyone try to make one?  Even then, it seems I need "reptoline-aware" 
versions of things like Firefox.




If my understanding is correct, IBRS is just effectively a microcode 
implementation providing the same outcome as a retpoline (but 
theoretically slightly faster). In both cases you still need the 
compiler and resulting compiled binaries to support the feature. Just 
applying the microcode to enable the IBRS instructions isn't going to do 
anything without the supporting code.


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Help with Spectre and Meltdown

[Brad Campbell]

On 18/01/18 08:28, Hendrik Boom wrote:

On Wed, Jan 17, 2018 at 09:25:19PM +0800, Tom Cassidy wrote:

You can install the intel-microcode package. AMD processors have a similar 
amd-microcode package.

https://packages.debian.org/intel-microcode

It looks like the updated microcode with the latest fixes is currently in 
Debian testing so I guess you could grab it from there directly and install 
manually if required.

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=886806


Has anyone revealed how this microcode avoids the Spectre problem?
Does it, for example, disable memory fetch from proteted memory?


There is quite a bit of information out there but it's scattered fairly 
thinly. Note this is my current understanding only.


The microcode doesn't "avoid" the Spectre problem by and in itself. A 
large part of the fix is implementing extra instructions to inhibit 
branch prediction under certain circumstances. It is then up to the OS 
to manage that. The bit people are stuggling with right now is that when 
the microcode is applied, extra processor feature flags appear so the 
kernel will need to re-scan the processor flags after an update and deal 
with the consequential fallout.


That won't itself inhibit the ability for userspace processes to access 
other process memory. The chief mitigation for that at the moment is the 
reduction in timer resolution in Javascript interpreters. That still 
doesn't prevent malware using it as a vector, but it does make it a 
*lot* harder for a rouge bit of Javascript served as an in-page add from 
scrounging through the browser process space for your internet banking 
credentials.


What the microcode fix does do is allow some further mitigation to be 
applied from the OS perspective, although for people with Haswell and 
Broadwell variants it'd be wise not to install the microcode until Intel 
get it fixed.


The most important issue for Spectre is ensuring your browser is up to 
date (and by up to date I mean the absolute latest version of Firefox). 
If you use one of the "free" forks, make sure the mitigation has been 
ported across or disable Javascript full stop.


If you use an Intel processor then make sure you have the OS Meltdown 
mitigation applied (ie KPTI).


Again, my current understanding based on lots of reading and the 
occasional discussion with people who *do* know better.


Corrections welcome. Hysterical ranting about the closed source nature 
of microcode >> /dev/null

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Backup plans: was Which is free, which is open source, et al.

[Brad Campbell]

On 14/01/18 06:30, KatolaZ wrote:

The bet ingredient for a successful "Primary Plan" is to assume that
there is no backup plan, an act accordingly ;)


Quite on the contrary. Having a well formulated and tested backup plan 
means you won't need it.


That applies as well to plans as it does to backups (although my well 
formulated and tested backup plan came in handy when I accidentally 
started 2 instances of the same VM on the same backing file last year. 
That ended ok, but the bit in the middle wasn't much fun).


It's those without a plan B that often have need for it.

I don't buy insurance for the "just in case". I buy it because those 
that have it don't generally need it. Murphy is a bit of a swine that way.


I'm a Debian escapee. My future is Devuan. My plan B is some form of 
BSD. I have it, but because I have it I know I won't need it.


Brad
--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Debian Devs using OSx? was Devuan in the German Wikipedia

[Brad Campbell]

On 21/12/17 05:13, Arnt Gulbrandsen wrote:

Steve Litt writes:

Is there evidence somewhere that Debian DDs use OS/x?


It's so common among developers in general that it would be very 
surprising if zero debianites do it.


Macbooks are almost a standard among developers, certainly a majority of 
the last hundred developers I've worked with used macbooks, perhaps even 
more than 90.


I love my Macbook. It is by far the best laptop I've ever owned (and 
I've had a few since my original Bondwell Z80 CP/M unit).


I've also been using a Mac on the desktop exclusively now since 2006.

Nowhere does that imply I'm not running a Debian based derivative on 
them. I just happen to be in love with the hardware.


OSX is neat and all, and I have a soft spot for the old System 7, but on 
the whole Mac hardware runs Linux well enough that I really don't have 
to compromise. I get the OS I want with the hardware I want. Win!


Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

[DNG] Upgrade from Debian Wheezy

[Brad Campbell]

G'day All,

I have a couple of production machines running Debian 7.11. One of those 
has a twin I use for staging and I routinely rsync the prod box across 
to ensure I'm testing in as accurate a replica as I can get.


Upgrade time is coming, and I'm migrating to Devuan Jessie.

This environment is a bit of a bitza with home-made backports of things 
like e2fsprogs, mdadm and plenty of supporting libraries plus custom 
grub configs, kernels and the usual gumph that is the result of a dist 
upgrade from Squeeze, and as-required piecemeal patchwork to keep things 
stable over the last 6 years or so.


I'm happy to say the staging box went through a relatively seamless 
upgrade with no apparent fallout. Nice. Better still, I can lose all the 
custom compiled stuff and go back to a standard install.


I'll be re-doing the upgrade a few times to ensure I catch all the 
little config file niggles that will need tweaking, but I was fairly 
blown away with how easy it was, given my experience "upgrading" to 
Debian Jessie during the late testing phase often resulted in wreckage 
and an unbootable box.


I've been used to Debian upgrades being seamless over the last 20 years 
or so, and they have been until the wreck that was Jessie, so it's nice 
to go back to a stable base that won't move stuff around under me, 
re-name all my network adapters and just fail to boot far enough into a 
system to allow me to recover when I do something dumb.


Two thumbs up to the faceless volunteers that make all this machinery 
happen.


Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Linus can no longer trust "init"

[Brad Campbell]

On 13/07/17 00:56, Dragan FOSS wrote:

On 07/12/2017 07:18 AM, Brad Campbell wrote:

Did I miss something


Yes.. ;)
When I wrote "recommends", and your interpretation of this word is
"mandating", I have no choice but to think that your reading system is,
to put it mildly, funny :)


Nope. Not at all. I don't see that postgres is "recommending" systemd at 
all. All they are saying is "if you want to use it with systemd, we have 
this option to make it easier for you". No recommendation in that and 
they are certainly not mandating.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Linus can no longer trust "init"

[Brad Campbell]

On 12/07/17 02:35, Dragan FOSS wrote:

On 07/11/2017 05:28 PM, Enrico Weigelt, metux IT consult wrote:

make it crystal clear: fuck off the
upstream.


Do you want to say that devuan users do not need any upstream that
recommends systemd?
For example, Postgresql?
***
With PostgreSQL 9.6 or newer, it is recommended to build with
--with-systemd and use the unit file shown in the documentation...
***
https://wiki.postgresql.org/wiki/Systemd


I read that page but my interpretation is different to yours.

I read it as "If you want to run PG 9.6 or newer with systemd, use this 
option and the unit file we include. If you want to run an older version 
the recommended approach is to write a unit file using Type=forking and 
use pg_ctl for ExecStart and ExecStop."


There is absolutely nothing there that is mandating systemd in any way, 
shape or form. They've just included code to make it behave better with 
systemd if that is how you choose to build it.


Did I miss something or are you fearmongering?

Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] A nice write-up.

[Brad Campbell]

On 23/04/17 17:33, Renaud (Ron) OLGIATI wrote:

www.theregister.co.uk/2017/04/22/devuan_1_0_0_released/




The comments are more entertaining.


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] FF pulseaudio hard dependency is here

[Brad Campbell]

On 14/03/17 19:19, m wrote:

hi there, do u know privoxy? its a filtering proxy :-)

m.



+10 for privoxy. Does a great job of ad-removal for me and one instance 
does the whole house :)


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Won't boot after dist-upgrade

[Brad Campbell]

On 20/12/16 06:08, marc wrote:

Hi devuanfanboy,

Thanks but no, that dind't work.


I have also had an interesting time getting the devuan beta to
run with a newer kernel.

In my case I have forgone the entire initrd, and built the newest
stable kernel from kernel.org with the necessary drivers built in.
In some kernel configurations the system boots into a blank screen.
In others X doesn't come up.

I suspect (but can't confirm) that it has something to do with how
the framebuffer or related kernel graphics drivers are initialised.

I think there was some sort of changeover between the old style
framebuffers and simpleframebuffer ? Please regard this as a
rumor, and somebody correct me if it is obviously wrong - as I
haven't gotten around to looking at it more. For the record: I was
doing this on an intel graphics chipset.


Just as a test, try CONFIG_DRM_FBDEV_EMULATION=y

I had a hells own time upgrading an older system to a newer kernel and 
not getting any console until I enabled that legacy option.


It might help, then again maybe not. Your post just triggered me to look 
back through my upgrade notes. I think that came in around 4.4 (ish).


Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] I ask Devuan to remove all defaming emails.

[Brad Campbell]

On 30/11/16 15:35, Edward Bartolo wrote:

Sir,

I am asking the Devuan Administration to delete and destroy all
defaming emails on DNG as they infringe upon the (Personal) Data
Protection Act. In those threads, I never approved my abilities or
their lack to be discussed, yet, abusive respondents, felt they could
infringe upon my basic legal rights. Reading some threads, a reader
invariably gets the message that I am a person of very low
intellectual abilities. Other emails give further health related
information that I refrain from disclosing, especially on a public
place. This is in direct violation of the Personal Data Protection Act
since it was done publicly without my approval. Please note I am an EU
citizen which gives certain legal rights. I am considering of getting
legal assistance about this matter. I suffer of NO mental ilnesses
irrespective of what has been illegally posted on DNG regarding my
mental health.

I repeat to write that the distribution has no excuse to refuse to
satisfy my request. Defamation damages a person's reputation and is a
criminal offence.


Dude, you posted on a public mailing list. It's archived in a million 
places (including my and many others private archives). As I tell 
everyone with a facebook account, if you don't want it accessible to the 
public forever, don't post it.


Best of luck.

Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] What does Devuan expect from a boot-loader?

[Brad Campbell]

On 01/10/16 12:44, Edward Bartolo wrote:

As far as I
am concerned, as long as a boot-loader can read GPT and MSDOS disks
plus extN partitions, that  should be enough.



That's nice. You have a beautifully simple use case.

*I* need a bootloader that can sort through various partition/RAID 
configurations and find it's config no matter where the BIOS puts the 
disk, if it's plugged into different hardware or if it's booted from a 
floppy/USB/CDROM and load the kernel/initramfs from ext or FAT 
partitions. It also needs to be able to skip over disks that are too 
dead to read, but alive enough to take up BIOS slots. Ohter people need 
to be able to boot from xfs, jffs, reiserfs, btrfs.. the list goes on.


My use case is different from yours, and what works for you won't work 
for me (having said that what works for me will work for you, it's just 
a lot larger and more complex to configure).


Don't go trying to re-invent the wheel or simplify things because *you* 
don't need the complex case. Some of us do and that's why there are 
multiple bootloaders available.


If you think the common bootloaders are becoming bloated, do some 
reading on LAB (Linux As a Bootloader) and see what bloat really means. 
Yes, that was a special case for embeded hardware, but none the less the 
point stands.




___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Kernel error requesting UEFI upgrade.

[Brad Campbell]

Please tell me you didn't submit a bug report to the freedesktop project 
without popping onto the mailing list to ask for some friendly advice 
and/or guidance because that is in no way what I suggested let alone 
implied.


On 20/09/16 17:53, Edward Bartolo wrote:

Hi,

I submitted a bug report as suggested.

Thanks.

On 20/09/2016, Brad Campbell  wrote:

On 19/09/16 22:53, Edward Bartolo wrote:

Hi,

If DNG administrators consider this email not appropriate for the
list, please delete it.

A kernel error is requesting me to upgrade my HP Probook 4540s UEFI.
The error involves the i915 graphics driver. On another laptop, the
one on which I coded simple-netaid, i915 failed to turn on the
backlight for a whole year. However, at that time I found a workaround
using a kernel parameter. I wish I find another kernel parameter that
prevents the current error. Since DNG is read by many who may have
used kernel parameters to override sloppy firmware implementations, I
am posting here for some help.

This is really the wrong place. You want to be reaching out to the i915
developers at : dri-de...@lists.freedesktop.org

It's not insisting you upgrade UEFI, it's attempting to say "Your BIOS
has mis-programmed the GPU. Maybe if you are lucky your vendor might
have a newer BIOS that does not do dumb stuff, so try upgrading".

My overall experience with this has been that in general the Windows
drivers don't care about the issue or else the BIOS would not have made
release, so your chances of a viable update are slim to none. Approach
the guys who actually write the driver and give them some useful
information to work with, and you might be pleasantly surprised with a
workaround or fix.

Brad
--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.





--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Kernel error requesting UEFI upgrade.

[Brad Campbell]

On 19/09/16 22:53, Edward Bartolo wrote:

Hi,

If DNG administrators consider this email not appropriate for the
list, please delete it.

A kernel error is requesting me to upgrade my HP Probook 4540s UEFI.
The error involves the i915 graphics driver. On another laptop, the
one on which I coded simple-netaid, i915 failed to turn on the
backlight for a whole year. However, at that time I found a workaround
using a kernel parameter. I wish I find another kernel parameter that
prevents the current error. Since DNG is read by many who may have
used kernel parameters to override sloppy firmware implementations, I
am posting here for some help.


This is really the wrong place. You want to be reaching out to the i915 
developers at : dri-de...@lists.freedesktop.org


It's not insisting you upgrade UEFI, it's attempting to say "Your BIOS 
has mis-programmed the GPU. Maybe if you are lucky your vendor might 
have a newer BIOS that does not do dumb stuff, so try upgrading".


My overall experience with this has been that in general the Windows 
drivers don't care about the issue or else the BIOS would not have made 
release, so your chances of a viable update are slim to none. Approach 
the guys who actually write the driver and give them some useful 
information to work with, and you might be pleasantly surprised with a 
workaround or fix.


Brad
--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] tor and systemd?

[Brad Campbell]

On 02/09/16 00:20, Arnt Karlsen wrote:


..we honk that nice big red "No-Shit???" horn now? ;o)
https://www.torproject.org/docs/faq.html.en#Backdoor

..to the Tor people: systemd made its way into Debian on banana
republic politics, not on tech merit.  Devuan tries to become a
non-systemd fork, and not a death trap to the next Ed Snowdon.

..since I didn't subscribe to your mail lists, I'm totally fine
with you guys rejecting my post there, just kick libsystemd0 off
any Tor packages until Edward Snowdon tells me in person here in
Norway, he is the man behind SystemD.



Please tell me you didn't send that to the tor list.

For the love of all that is holy stop trying to make the Devuan project 
look like tinfoil-hat doomsday preppers.


If you don't like tor pulling in systemd, then *fix* it. Don't go 
hysterically screaming about the sky falling and dropping names like 
"Edward Snowd*e*n" without good reason. If you are going to drop names, 
spell them correctly. Not doing so makes you look even more of a ranting 
lunatic.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] eudev [was: vdev]

[Brad Campbell]

On 24/08/16 13:57, Steve Litt wrote:

On Wed, 24 Aug 2016 11:37:53 +0800
Brad Campbell  wrote:


On 24/08/16 11:13, Steve Litt wrote:

On Tue, 23 Aug 2016 21:47:41 -0400
Clarke Sideroad  wrote:


I think kdbus is dead due to the bad press, but I believe there is
bus1 coming along to replace that.
https://github.com/bus1/bus1
http://www.bus1.org/

Some familiar names, but possibly not directly part of
systemd

Clarke



DANGER Will Robinson. From the COPYING document:

===
COPYRIGHT: (ordered alphabetically)
Copyright (C) 2014-2015 Red Hat, Inc.
AUTHORS: (ordered alphabetically)
David Herrmann 
Tom Gundersen 
===

And from Wikipedia's systemd page:

===
Original author(s)  
Lennart Poettering, Kay Sievers, Harald
Hoyer, Daniel Mack, Tom Gundersen and David Herrmann
===

These saboteurs just won't quit. It's our job to get out the word so
bus1 fares no better than kdbus, because Lennart bragged about his
plans when he gets the kernel to enforce use of systemd.


I'm not worried. Mantra from get-go has been "Don't break userspace".
If there is a valid use-case for a feature there will be plenty of
opposition to it's removal.


[snip]


If bus1 really has technical merit, can demonstrate it solves real
problems and has all its shortcomings addressed there is no reason it
shouldn't be integrated into the kernel. They can't then just go and
remove netlink to spite non-systemd users. It has an existing
userspace and other use cases.


Assuming by "they" you mean the Lennart and the Redhats, they already
have an established pattern and practice of breaking user space. If you
mean the kernel developers, they won't be the ones breaking userspace,
but a kernel-included bus1 will act very much like the firmware chips
they put into toner cartridges just so you won't buy competing toner.


I'm not entirely sure you understand what I mean by "break userspace". 
It is entirely in the context of the kernel and its interface with 
userspace and absolutely nothing to do with userspace itself. It means 
they can't just go and rip bits out of the kernel that mean *our* 
userspace won't run on it. I don't care what they do with *their* userspace.




We're way past the point of thinking the world is a technocracy.

Edbarx said it best: "attempting to remove systemd from SID is more
like attempting to remove the DNA from living cells expecting them not
to die."

That sounds very much like breaking userspace to me.


No, again you have the wrong end of the "userspace". You refer to 
distributions, and I don't care what those distributions do, what they 
break or which init they use. What I care passionately about is ensuring 
that stuff that runs right now continues to run on newer kernels. Oddly 
enough, history has shown that's generally what Linus appears to care 
about also.


It takes *years* of notice and warning for features to be marked 
deprecated, and then years for them to be removed. *If* during those 
years we discover that our device manager is going to cease to function, 
we have several years to figure out a solution and get it implemented 
and tested. That's a BIG *IF*.


Don't Panic.

--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] eudev [was: vdev]

[Brad Campbell]

On 24/08/16 11:13, Steve Litt wrote:

On Tue, 23 Aug 2016 21:47:41 -0400
Clarke Sideroad  wrote:


I think kdbus is dead due to the bad press, but I believe there is
bus1 coming along to replace that.
https://github.com/bus1/bus1
http://www.bus1.org/

Some familiar names, but possibly not directly part of systemd

Clarke



DANGER Will Robinson. From the COPYING document:

===
COPYRIGHT: (ordered alphabetically)
Copyright (C) 2014-2015 Red Hat, Inc.
AUTHORS: (ordered alphabetically)
David Herrmann 
Tom Gundersen 
===

And from Wikipedia's systemd page:

===
Original author(s)  
Lennart Poettering, Kay Sievers, Harald
Hoyer, Daniel Mack, Tom Gundersen and David Herrmann
===

These saboteurs just won't quit. It's our job to get out the word so
bus1 fares no better than kdbus, because Lennart bragged about his
plans when he gets the kernel to enforce use of systemd.


I'm not worried. Mantra from get-go has been "Don't break userspace". If 
there is a valid use-case for a feature there will be plenty of 
opposition to it's removal.



What is the best way of getting the word out?


I wouldn't worry. You are not giving anyone involved in kernel 
development enough credit if you honestly believe this will fly under 
the radar and people won't notice.


Banging drums and putting forth objections based on some names and 
conjecture will simply get you roundly ridiculed and then ignored by 
those that actually matter. kdbus was not rejected on politics, it was 
rejected on technical merit quite validly by those who care. If bus1 
hasn't rectified _all_ of those objections and can demonstrate a real 
requirement then it won't get past the gate.


The mistake with kdbus was it was a shit design with the sole purpose of 
papering over existing shit design in dbus user space, and because the 
systemd folk have Greg KH on board they assumed they'd just be able to 
slip more shit into the kernel without question. People who actually 
knew better stepped up and nak'd it on technical grounds.


If bus1 really has technical merit, can demonstrate it solves real 
problems and has all its shortcomings addressed there is no reason it 
shouldn't be integrated into the kernel. They can't then just go and 
remove netlink to spite non-systemd users. It has an existing userspace 
and other use cases.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] OT: true read-only disk

[Brad Campbell]

On 19/08/16 00:08, Rob Owens wrote:


I don't know the answer to your read-only question.  But having done
some data recovery in the past, I've found that attaching the drive via
USB and sitting the drive in the freezer during recovery can help in
situations like this.  Also, besides ddrescue you can also take a look
at photorec (part of the testdisk package).



+2 for the freezer. I've used a long USB and power lead, put the drive 
in a ziplock bag with some silica gel, pulled as much air as possible 
out and sealed it up with duct tape. Done this on about 3 drives over 
the years and had great results. Only on drives where they exhibited 
thermal issues though.


I've never done a filesystem recovery on a dying drive for fear of 
further damage. I _always_ image using ddrescue and deal with the 
filesystem issues on the clone.


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] lilo development has ended

[Brad Campbell]

On 16/08/16 21:54, Rainer Weikusat wrote:


Apart from that, current LILO versions already do seem to support this
and even if they didn't, adding a "Try 0x81. In case it doesn't exist,
try 0x80" option to the code wouldn't be terribly complicated. Just
requires "another learning curve" ...


I have zero desire to *try* anything as I have a solution that "just 
works" (and has for 11 years). But don't worry, I won't take your 
insinuation that I didn't read the manuals or faq or source, and didn't 
spend hours exploring every possible option before moving to another 
solution personally.


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] lilo development has ended

[Brad Campbell]

On 16/08/16 11:50, Rick Moen wrote:

Quoting Brad Campbell (lists2...@fnarfbargle.com):


Actually, this exact reason is why I moved from Lilo to Grub a few
moons ago.

It happens *when* one of the primary OS drives dies in your server
and you get a reboot before you have a chance to fix the array.

Example (because this is what triggered the move for me). You have a
RAID1 with your root and boot on it. LILO is installed on those
disks, and you can tell the BIOS to boot from either and it works
just fine. One of those drives goes titsup and before you've had a
chance to deal with it (because it's in an office 13,000km away) you
get a power outage that exceeds the runtime of the UPS. The system
cleanly shuts down and never comes back up because LILO does not
cope with the fact the BIOS has re-ordered the drives.


What a pity you never read the tip in the Boot+Root+Raid+LILO HOWTO that
I referred to upthread:

  # BIOS=line -- if your bios is smart enough (most are not) to detect
  that that the first disk is missing or failed and will automatically
  boot from the second disk, then bios=81 would be the appropriate
  entry here. This is more common with SCSI bios than IDE bios. I
  simply plan on relocating the drive so it will replace the dead
  drive C: in the event of failure of the primary boot drive.

http://www.tldp.org/HOWTO/Boot+Root+Raid+LILO-3.html

That _seems_ to address the exact situation you speak of.  I haven't
tested this, but I might eventually get around to doing a test.
(My systems currently use GRUB 0.9x, and my present intention is to
migrate not back to LILO but rather to extlinux.)


G'day Rick,

Details are sketchy now as I made this change in 2005 after 9 good years 
with LILO. I did try very hard to make it work, and it may have been an 
issue with the BIOS ultimately. I actually had a hard copy of that 
particular howto next to my console as I was constantly working with 
LILO and RAID trying to get something that was reliable across the 
plethora of nasty hardware I had to support, so if there was a tip in 
there that might have applied I certainly would have tried it more than 
once. Thankfully I've managed to repress most of those memories.


The fact remains once bitten, twice shy and it was *much* easier to get 
Grub to work reliably than LILO in this particular instance. After 
spending the couple of hours I needed to come up to speed on Grub, I 
just saw no point persisting with LILO anymore. Under the circumstances 
it was a more reliable solution. I've used Grub now on all x86 based 
machines since 2005 and I've never bumped up against an issue that has 
taken more than 5 minutes to resolve, and certainly nothing that wasn't 
a fat finger issue rather than hardware doing funky stuff.


Horses for courses, I just find Grub reliable enough that I can't see a 
point in moving away from it. If you read the docs it really does work, 
even against pathological BIOS implementations that make LILO fragile.


I'm not rejecting your experience, I'm just saying *I* found a case 
would reliably break LILO, and after learning how to actually configure 
Grub I've never bumped up against anything similar. That's just me however.


I even migrated from Grub 0.9x to Grub2 and I've had nothing but 
unicorns and rainbows there too. Again, another learning curve, but 
nothing a couple of hours didn't sort out. I seem to be the only person 
alive that actually gets along with Grub, but that's ok because it works 
for *me* and that's what matters.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] lilo development has ended

[Brad Campbell]

On 16/08/16 00:09, Rick Moen wrote:

Quoting Adam Borowski (kilob...@angband.pl):






Or, the first disk gets assigned a different position.


Which happens when exactly?  Because you're screwing around with
swapping in and out different HBAs?  Well, if you're doing that, see



Actually, this exact reason is why I moved from Lilo to Grub a few moons 
ago.


It happens *when* one of the primary OS drives dies in your server and 
you get a reboot before you have a chance to fix the array.


Example (because this is what triggered the move for me). You have a 
RAID1 with your root and boot on it. LILO is installed on those disks, 
and you can tell the BIOS to boot from either and it works just fine. 
One of those drives goes titsup and before you've had a chance to deal 
with it (because it's in an office 13,000km away) you get a power outage 
that exceeds the runtime of the UPS. The system cleanly shuts down and 
never comes back up because LILO does not cope with the fact the BIOS 
has re-ordered the drives.


Grub copes with this just fine. Slip the drive into any slot and the 
bios will search for the first disk with a viable boot sector and loads 
the grub 1st stage. Grub searches for its second stage, finds it and 
you're good to go.


Of course there might be user error in there, but I spend an *awful* lot 
of time trying to make Lilo work in what were fairly simple failure 
scenarios before giving up and moving to Grub.


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] SystemD's brownie points over non-systemd OSs.

[Brad Campbell]

On 09/08/16 06:03, Go Linux wrote:


I posted a link to this response at the FDN link posted above.  This was the 
response from the author of the howto:

http://forums.debian.net/viewtopic.php?p=621972#p621972

He concludes:

". . . I would prefer to use the systemd-supplied components (systemd-boot, 
systemd-networkd, systemd-resolved, etc) wherever available as I believe this offers a 
more cohesive, UNIX-like working environment."

That would depend on your definition of a 'UNIX-like working environment' IMO.


In that context I believe it is along the lines of an environment where 
the same basic programming mistakes and logical fallacies are propagated 
over the maximum amount of system components.


Thankfully there are still alternatives.

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] KVM/ksm: was SystemD's brownie points over non-systemd OSs.

[Brad Campbell]

On 08/08/16 14:17, Steve Litt wrote:

On Mon, 8 Aug 2016 14:00:48 +0800
Brad Campbell  wrote:


Just for the record, so does KVM if you turn on ksm. It actually
works very well for a free solution.


I use Qemu all the time, with hardware assist. How would I turn on ksm?



KSM=/sys/kernel/mm/ksm
echo 500 > $KSM/sleep_millisecs
echo 1000 > $KSM/pages_to_scan
echo 1 > $KSM/run

I do it by putting this in a script that gets run at bootup. My numbers 
are relatively arbitrary and are designed to work slowly with a low load 
on the system. Redhat published a daemon to do it called ksmtuned, but 
for my use case it just made things more complex for no real gain.

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] SystemD's brownie points over non-systemd OSs.

[Brad Campbell]

On 08/08/16 13:44, Aldemir Akpinar wrote:


Sun, Aug 07, 2016 at 06:31:10PM +0200, Edward Bartolo wrote:

On the other hand, full-machine virtualization costs you the max of
assigned
memory to that system, at all time.


Going off-topic but I just wanted to correct this statement, when you're
using full machine virtualization in the worst case it will use all the
memory assigned to the server (unless you're using xen). Products like
vmware or hyper-v uses in memory deduplication and swapping the virtual
machine files to the disk to reduce the memory usage.


Just for the record, so does KVM if you turn on ksm. It actually works 
very well for a free solution.


This is a tiny array of linux & windows VMs (5 linux / 3 windows). ksm 
effectively saves half the memory footprint.


root@srv:~# ./ksmstat2
Shared memory is 488 MB
Saved memory is 6557 MB
Shared pages usage ratio is 13.43
Unshared pages usage ratio is .95

Sure, not as efficient as any of the containers, but then the containers 
can't run a blackberry BES or any of the other horrid windows software 
you sometimes have to run as part of day to day business life catering 
for others.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] SystemD's brownie points over non-systemd OSs.

[Brad Campbell]

On 08/08/16 00:31, Edward Bartolo wrote:

Hi All,

http://forums.debian.net/viewtopic.php?f=16&p=621842#p621842

Excuse me for the topic title. But the above link at first looked like
some inherent advantage in using SystemD. However, after a little
reflection, a couple of minutes, it seems there are actually no extra
brownie-points in using SystemD.

This *appears* like an advantage SystemD users have over non-systemd
users. But if Devuan allows the use of virtualisation software, the
same can be achieved without requiring any sort of benediction
SystemD.



Unless I'm mistaken, it looks like someone re-invented a more 
complicated chroot.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Ugly, ugly news

[Brad Campbell]

On 26/07/16 10:27, Simon Walter wrote:


Is that really the case? Did the Debian leadership do a poll to find out
what their users wanted and who were their typical users?
Desktop/personal vs. server/professional? Did they consult their package
survey stats?

I wasn't participating in those discussions. Maybe I should have. For
that, my friends, I am sorry. I am sorry to open a can of worms of
perhaps what is a sensitive issue with those of you who were involved in
the discussion and the exodus.



My gut feeling is that discussion is a waste of time. You won't 
influence the politics, you won't actually learn anything from the 
discussion and you'll divert time and energy away from moving in the 
forward direction.



The reason I want to know about that is because if most Debian users are
desktop users, perhaps it's for the better that they went with systemd
as the default. It gives more credence to Devuan as the professional
choice.


I'd rather just keep doing what we're doing and working on a credible 
alternative. The results will speak for themselves rather than more 
rattling of empty tins.


Let's keep our eyes on the ball.


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] F1 and special usernames on the login screen

[Brad Campbell]

On 19/07/16 10:43, Rick Moen wrote:

Quoting Brad Campbell (lists2...@fnarfbargle.com):


This is one I find interesting. I've never used an operating system
where it was required to know root credentials to halt or reboot the
machine from the login screen.


Remember, Unix is a multiuser operating system, and also one supporting
both local and remote users, who would be annoyed by someone deciding to
cut them off.



Rick I completely understand that sentiment, and none of my servers have 
a GUI on them. I just "assumed" (yeah, my mistake) that display managers 
were used only on single user desktop machines.


Windows is a special case, well just because Windows has always been a 
special case and it has taken Microsoft some 20 years to come to terms 
with a GUI-less server.


Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] F1 and special usernames on the login screen

[Brad Campbell]

On 19/07/16 00:37, Steve Litt wrote:

Hi all,



SPECIAL USERNAMES

The login screen has no controls other than the input field. So how do
you reboot, halt or go to a console from the login screen? The answer
is, you use a special username.

For instance, to reboot the computer from the login screen, type in the
username "reboot" (without the quotes), then when asked for the
password put the root password, and it reboots. If you want to halt,
use the username "halt". To go to a console, use "console" without any
password. A little console appears, asking you for username and
password. When you exit the console via the exit command or Ctrl+d on
the console, you go back to the login screen.


This is one I find interesting. I've never used an operating system 
where it was required to know root credentials to halt or reboot the 
machine from the login screen. Certainly if the machine is logged in but 
locked, unlocking is required first but on any of the other OS I use I 
can simply shutdown or reboot unauthenticated.


A good example is my desktop. It's an iMac and it triple boots OSX, 
Windows & Linux. Now, it only gets rebooted every year or so for a 
kernel update or if I need to spark up OSX to test something specific.
If I reboot the machine and am not there to hold the magic keys required 
for it to boot to linux, it winds up at the OSX login screen. From there 
I can simply click reboot and away we go. Why should I need to know the 
root login to do that?


As for root logins on my machines. They all have 16 character randomly 
generated passwords that are all but impossible to remember. The only 
time I actually need a root password is if something catastrophic 
happens and I need to access the rescue console. Everything else is 
managed with sudo. When I do need the root password I can look it up in 
my password wallet. I've had to do that precisely 3 times in the last 6 
years.


So all that rambling comes back to "why do I need to know my root 
password to halt or reboot the machine from the login screen?"


Regards,
Brad



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Need for documentation (Steve Litt)

[Brad Campbell]

On 16/07/16 00:24, Hendrik Boom wrote:

On Fri, Jul 15, 2016 at 02:25:01AM -0400, Steve Litt wrote:


Cool! Like F1, these four things should be discoverable from text on the
login screen.


Or else put this on the login screen, along with the F1 stuff:

1) Hold down Alt and SysRq and type "reisub" to reboot, or
2) Alt-SysRq plus "reisuo" to shutdown.


That's a little like cracking a walnut with a sledge hammer. Let's just 
pull the machine out from under the OS and hope it didn't need or want 
to do anything prior to shutting down (like oh, I don't know, cleanly 
shutting down some RAID and ensuring drive buffers are flushed before a 
hard power off?). There's more to a controlled shutdown than a sync and 
remount.


Additionally, I can't find a SysRq key on my Mac keyboard. Ancient 
legacy "PC" based systems are not the only ones running modern Linux 
distributions.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Mini init script written in Perl boots.

[Brad Campbell]

On 20/06/16 00:40, Steve Litt wrote:

On Sun, 19 Jun 2016 17:18:48 +0200
Edward Bartolo  wrote:


So, I joined Devuan to contribute code. But, I am NOT appreciated, and
to be sincere, sometimes I even think about leaving the project
altogether. I am being denigrated just because I make a conscious
effort to write simple code.


Whoaa, Edward,

The words "I am not appreciated" are just not rational. The fact is, a
lot of people appreciate you and your efforts, some don't. Please don't
let your emotions change your perception of the true facts. Just keep
plugging away at your code, accept help from those who cooperate, and
ignore those who issue personal accusations. I'd suggest you keep all
emotional content out of your emails and stick to tech. All the drama
in the world won't change the content of an integer.


Yeah, wot 'e said.

I think the crux of the problem here Edward is a bit of 
miscommunication. It was quite obvious to me that what you were out to 
do was build a basic init system as a bit of an educational exercise. To 
others it must have appeared that you were serious about writing an init 
replacement, and their criticism (to me) appeared to be on that basis.


Personally I enjoy watching your process when you get into something. We 
all learn differently and you appear to learn well in an 
iterative/collaborative process that is verbose and visible. It's 
different to the way *I* learn, but the world would be a boring place if 
we were all the same.


Now, having said that I think the way you do what you do is actually 
quite helpful as it makes the process transparent enough that you 
provide some education to others as you go. No matter the ultimate 
outcome, there is value in the process.


I spent some years writing cross platform tools for a uController that 
only had windows tools. Ultimately the continual abuse I got for not 
open-sourcing those tools and the process of managing expectations in a 
forum style of environment drove me to a breakdown. The thing I've 
subsequently learned is I'm not suited to that because I don't have a 
thick enough hide not to take the criticism personally. I did learn 
however that empty vessels make the most noise; and a text only 
environment coupled with non-native english speakers often means gross 
misunderstandings are the norm anytime it strays outside of the purely 
technical.


On a personal level, keep doing what you are doing because I find it an 
enjoyable read. Don't take it personally because it's not important, and 
do it because you enjoy it. If you don't love every minute of the 
process, find something else to do.


Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Custom OS initiator. In need of some hints...

[Brad Campbell]

On 13/06/16 16:19, Tomasz Torcz wrote:

On Mon, Jun 13, 2016 at 08:16:44AM +0200, Edward Bartolo wrote:

Hi,

This is a question about implementing a minimal OS initiator.  If
DNG's admins/mods deem this email is misplaced please delete it or
ignore it.


  If you really want to create yet another init, I can recommend this
simple, step by step explanation of Linux' init responsibilities:

  https://felipec.wordpress.com/2013/11/04/init/



That was a great read and a lot simpler than getting into the busybox 
source :)


I've not particularly bought into the whole systemd politics, I just
don't like it on a technical level. Having said that the two links on 
the first line of that article made for horrifying reading.


If that is the position taken by the systemd author, there is no wonder 
it's such a three legged goat.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Custom OS initiator. In need of some hints...

[Brad Campbell]

On 13/06/16 15:29, KatolaZ wrote:


If I can provide my 2 cents to the discusion, before writing a new
init you should have studied and understood very well one of the
existing ones, and what should happen behind the scenes from the
moment your kernel is decompressed to the appearance of a login
terminal. I would recommend to start with sysvinit or the standard
rc-based init of *BSD. By studying I mean looking into the code of the
process that is called "init" and into the code of all the processes
directly started by init (e.g., /etc/init.d/rc, in the case of
sysvinit, getty/agetty, etc...), and being sure to have understood
what is in there.


Concur. If you wanted a simple and easy to follow example, grab the 
busybox init source. It does everything you need and is small enough 
that you can digest it relatively easily.


Most other init systems tend to rapidly devolve into a nest of vipers 
and are therefore harder to follow.


Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] USB file transfer to your Android device

[Brad Campbell]

On 13/05/16 09:35, Steve Litt wrote:

The following document looked pretty good, but the author kept pulling
unexplained things out of every orifice, so it was impossible to know
how the commands applied to me:

https://pve.proxmox.com/wiki/USB_Devices_in_Virtual_Machines

USB passthrough is a pretty complex thing because I have a Void host, a
Devuan guest, by necessity I have a USB hub. There are so many
variables that, without finding a person who really knows how this
stuff works, I could spend days.


Hey Steve,

USB passthrough can be either complex or simple. I kinda like it the 
simple way.


Here's how I pass iOS devices through to a Windows VM on my desktop and 
laptop.


 I add these three lines to my qemu command line script

 -usb\
 -device nec-usb-xhci,id=xhci\
 -device usb-host,bus=xhci.0,vendorid=0x05ac,productid=0x1261 \

The first tells qemu we want the usb stack. The second creates a virtual 
xhci controller and associated root hub, and the third assigns any 
device plugged into the host with the VID/PID pair through to the VM.


With that, when I plug my iOS device into the host, qemu claims it and 
pokes it through to the VM. VID/PID can be just pulled from lsusb when 
you plug the device into your host.


There are other ways whereby you can attache busses and hubs to the qemu 
instance, but that seems pretty complex when you just want to pass a 
single device through (or multiple devices even).


eg :
 -usb\
 -device nec-usb-xhci,id=xhci\
 -device usb-host,bus=xhci.0,vendorid=0x05ac,productid=0x1261 \
 -device usb-host,bus=xhci.0,vendorid=0x05ac,productid=0x12ab \
 -device usb-host,bus=xhci.0,vendorid=0x05ac,productid=0x12aa \
 -device usb-host,bus=xhci.0,vendorid=0x05ac,productid=0x1281 \


That passes through my iPad, my Wife's iPad, my iPod and a mates iPod if 
any or all of them are connected to the host simultaneously.


Now this works with pretty much anything I've tried from iPads to HASP 
dongles, digital cameras. You get the idea.


When I pass through my Blackberry for a firmware update, it uses one PID 
for normal mode and another PID for the bootloader, so I need both of 
those. Come to think of it, so do the iOS devices when you are firmware 
updating them. Just claim both and you are good to go.


Normally I'd just pass the VID through  and wildcard the PID, but since 
my desktop and laptop are Apple hardware I end up claiming the accessory 
devices as well (like camera, sound card and keyboard), so I use both 
VID and PID.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] No DNS: OVH under DDOS

[Brad Campbell]

On 06/05/16 17:41, hellekin wrote:

As reported on IRC just a minute ago:

<   Guy-> is it just me, or is there a DNS problem with devuan.org?
 Guy-: http://travaux.ovh.com/?do=details&id=17875
 dns anycast.me ( the CDN where devuan.org is hosted for
DNS ) is under dDoS

 i will take care to add other secondary DNS in next few
days to avoid issues like this one in future



I'm sure there are plenty of other offers, but I run a little set of DNS 
servers for some work related stuff. I'd be happy to add a secondary or 
two if it's needed (or if it'd help).


(Posted to the list because I have no idea whom to send it to otherwise).
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] OpenRC and Devuan

[Brad Campbell]

On 03/05/16 17:24, Didier Kryn wrote:

Le 03/05/2016 08:51, Mitt Green a écrit :

The current init system is old. Ancient.
We should all agree on it. Devuan is looking
for a new init system that is not systemd and my
personal choice for this task from now on is
Gentoo's OpenRC.

‎
Unix is old. Ancient. We should all agree on it.
Devuan is looking for a new base system that
is not Unix and my personal choice for this
task from now is Microsoft's Windows.



 Debian-potato was systemd-free. OK it's old now, but still less
than Unix. Why not still use it? No need for Devuan.


I am still using it in several locations, but for net connected systems 
security updates are nice.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] OpenRC and Devuan

[Brad Campbell]

On 03/05/16 07:19, parazyd wrote:

The current init system is old. Ancient. We should all agree on it.



Devuan is looking for a new init system that is not systemd and my
personal choice for this task from now on is Gentoo's OpenRC.


Without getting specific, I just want to flag this as the sort of 
"change for changes sake" that I moved to Devuan to get away from.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Skype: Re: What do we want for ascii ?

[Brad Campbell]

On 22/04/16 15:56, Svante Signell wrote:

On Fri, 2016-04-22 at 15:26 +0800, Brad Campbell wrote:

On 22/04/16 12:51, Joel Roth wrote:



Skype for Linux, OTOH, *is* stale (and last time I checked was
tricky to install) AIUI intentionally so due to Microsoft's
special love for Linux.



brad@bkmac:~$ apt-cache search apluse


Typo here, but the search result was still empty.



Funny how it does not matter how many times you stare at something, but 
you still miss it. Seems to get worse as I get older.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] What do we want for ascii ?

[Brad Campbell]

On 22/04/16 12:51, Joel Roth wrote:


Skype for Linux, OTOH, *is* stale (and last time I checked was
tricky to install) AIUI intentionally so due to Microsoft's
special love for Linux.



It's certainly behind the other os clients (windows/osx/iOS) but 
installing it on a several year old Mint/DE (so a very old Debian 
Jessie/testing base) was as hard as dpkg -i skype.deb.


I got sound/video with no futzing about. I can conference call my 
colleagues without resorting to a Windows VM or iPad, and show my inlaws 
their grandkids. Does the job. It's also more bandwidth friendly and 
tolerant of grungy links than facetime.


brad@bkmac:~$ apt-cache search apluse
brad@bkmac:~$ dpkg -S apulse
dpkg-query: no path found matching pattern *apulse*
brad@bkmac:~$ dpkg -l skype
Desired=Unknown/Install/Remove/Purge/Hold
| 
Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend

|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name  Version   Architecture 
 Description

+++-=-=-=-===
ii  skype 4.3.0.37-1i386 
  Wherever you are, wherever they are

brad@bkmac:~$

Yes, I don't use it for confidential stuff, nor client sensitive comms, 
but it serves its purpose admirably, and until a decent open alternative 
becomes available it'll continue to do the job.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Devuan Web A11y

[Brad Campbell]

On 20/04/16 15:21, Noel Torres wrote:


Trond Arild Ydersbond  escribió:


I agree. And IMHO, it is not only the docs, but the whole approach to
information handling where we need to improve on Debian.  How do we
cater for tens of thousands or more users, spread over the whole
spectrum of computing needs and with widely differing skills? We have
to prepare for becoming hugely successful :-)


And even to Debian users. If we develop a good documentation system,
people will start telling other people to search our docs, and
mouth-to-ear we'll receive more visits and can explain to them why we
are here.


As a personal preference when I'm in a hurry, I've always liked the Arch 
documentation.


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Systemd criciticized for malfunctioning OS

[Brad Campbell]

On 18/04/16 23:33, Steve Litt wrote:


By the way, if he were initting with Runit this would be so trivial
to troubleshoot, because with Runit there's no place you can't put your
voltage probe.


There! Right there. That quote : "because with Runit there's no place 
you can't put your voltage probe.". That's the gold I've been looking 
for. It perfectly sums up the problem. Fault finding is orders of 
magnitude easier with instrumentation, and with existing init systems 
you can *easily* instrument them, step them or completely manually 
iterate the process. Systemd makes that considerably harder* and makes 
the process more opaque.


* I won't say impossible because there may be some way of doing it that 
I've not found (not that I've looked very hard because systemd puts the 
bar at a level I have little inclination to try and get over).



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] suspend and hybernate

[Brad Campbell]

On 13/04/16 09:27, Steve Litt wrote:

Hi all,

What's the difference between suspend and hybernate?

How can I achieve each condition from the command prompt?

If I achieve each condition from the command prompt, how do I "wake up"
the computer when I'm ready to use it again?


In practical terms :

Hibernate saves state to disk and powers the machine off. Resume is 
effected by powering up the machine and booting the kernel you were 
using. Either the kernel or with the help of the initramfs will find the 
suspended state and restore it.


Suspend quiesces the machine, powers down accessories it can power down 
and put the machine into an ultra low power mode with state preserved in 
RAM. The machine will have one or more wakeup devices active and powered 
on and hitting one of those will bring the machine back.


Personally I use a hybrid known as s2both. It queisces the machine, 
saves state to disk and then suspends to ram. This has the disadvantage 
that it takes time to write out to disk, but has the advantage that when 
I inevitably forget to plug the machine in, or it runs out of battery 
because I've suspended it with 5% battery left and that goes in a day, 
when I plug it back in and boot it up it resumes from disk.


All of these options live in pm-utils and can be configured up from 
there. I have suspend scripts that unload the wireless driver module 
(broadcoms wl driver is not reliable on resume), configures my wakeup 
sources so only the power button brings it back (because the Macbook 
sometimes issues spurious usb keyboard/mouse wake events that turn your 
laptop into a backpack heater), and sets up other niceties that make 
things just nicer in general.


I also suspend to encrypted swap, so if I need to resume from disk the 
initramfs is required to unlock the swap prior to starting the resume 
process.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] useradd defaults

[Brad Campbell]

On 04/04/16 16:19, KatolaZ wrote:


(unfortunately Linux does not run on microcontrollers, yet, mainly due
to the general lack of some form of underlying MMU in the vast
majority of microcontrollers...).


http://www.uclinux.org/index.html

It has been around for years :) I ran it for quite a while on some m68k 
processors before I moved to embedded Intel boards.


Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] X forwarding over SSH over ADSL

[Brad Campbell]

On 13/03/16 20:36, Brad Campbell wrote:


I've snipped the remainder of your un-informed rant.



Sorry, that was un-called for.

Never the less, tightvnc works a hell of a lot better over super low 
bandwidth links than I've ever managed to get out of RDP or straight 
remote X. Plus, none of the compressing/caching X proxies have ever 
worked right, whereas spinning up an x11vnc instance and using tightvnc 
as a viewer has always been a quick and very efficient way of getting a 
graphical console on low bandwidth without needing any configuration on 
the remote end.




___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] X forwarding over SSH over ADSL

[Brad Campbell]

On 13/03/16 00:48, Rainer Weikusat wrote:

Brad Campbell  writes:

On 08/03/16 04:41, Simon Hobson wrote:


VNC is lousy over anything but a very fast link, it's just a remote
framebuffer - anything painted to the screen is bit copied to the
client which is bandwidth intensive.


Whereas tightvnc works quite well over almost anything, and if you are
willing to drop to a small colour pallete


Considering that tightvnc is Windows-only software it seems a bit out of


Que? Better not tell any of my linux boxes that have been using it as 
both a server and client for many more years than I care to remember.


brad@srv:~$ apt-cache search tightvnc | grep -i tightvnc
tightvnc-java - TightVNC java applet and command line program
ssvnc - Enhanced TightVNC viewer with SSL/SSH tunnel helper
tightvncserver - virtual network computing server software
xtightvncviewer - virtual network computing client software for X

I've snipped the remainder of your un-informed rant.

Brad.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] X forwarding over SSH over ADSL (Was: Claywand dosplay bananager)

[Brad Campbell]

On 08/03/16 04:41, Simon Hobson wrote:


VNC is lousy over anything but a very fast link, it's just a remote framebuffer 
- anything painted to the screen is bit copied to the client which is bandwidth 
intensive.


Whereas tightvnc works quite well over almost anything, and if you are 
willing to drop to a small colour pallete (with the -bgr233 option) 
it'll almost work on RS485 over wet string.


My personal experience over the last  years is TightVNC > RDP > 
VNC, but YMMV.


Brad.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] OT: Assembly resources

[Brad Campbell]

On 28/02/16 17:49, Mitt Green wrote:

Hi,

I believe, here are some people that know assembly, I'd like to know
what resources would ye recommend that teach it. Preferably AT&T syntax
using gas.

I do my coding in C, but always wondered about something low level.

Also, I have a book The Art  of Assembly Language, that focuses
on Windows, and has explanations of algebra and other maths.
So, maths resources would be great too.


I did it the other way around. Started with a binary and a disassembler 
and figured out how it worked, then learned to patch it up to check my 
understanding was correct.


I started with the 6502 and a dead tree of the Apple ][ ROM source a 
couple of moons ago. That was so much more pleasant than x86 assembly, 
but that method works just as well.


These days I only get into PC assembler when correcting broken software, 
but I still write a lot of uController code in asm.


There is a free version of IDA available. It's the ducks nuts as far as 
disassemblers go.


Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Debian archives (Was:systemd==bad)

[Brad Campbell]

On 23/02/16 16:10, Simon Hobson wrote:

Brad Campbell  wrote:




Interestingly the package files are still all present, so after an
apt-get update ; apt-get install foo I just put the package names
it can't download into google and they turn up in odd corners of
the net.


Yes that's a problem. It would be nice if, instead of just dropping
everything, they kept a snapshot of the latest versions for that.


It really would make life so much easier, even if it was one server with 
no mirrors. They're not exactly going to be high traffic. Back in the 
days of 33k modems I used to mirror my own Debian distributions locally 
for fast net installs, maybe I need to start doing that again. Storage 
is a lot cheaper now!



Which leads to a question that's been swimming around my head since
it was mentioned here. Does the LTS repo have a complete snapshot of
Squeeze (and Wheezy when it's live) as of takeover day, or just
anything updated ? If they keep the whole lot online then that makes
building new boxen a lot easier. I was thinking that, without a full
repo, bringing up a new Wheezy box would be tricky unless the base
image I already have contained every package I might need in future.


A single source (with an updates repo for security as required) is 
always a better option for a new install.



BTW - I was running updates last week, all the Wheezy systems needed
a new glibc but all the older ones are older than the DNS bug :-) I
realised I still have one Etch system running - and that's a system I
knocked up "as a temporary measure" to replace an ailing mail server
running on NT4. Some of the systems I updated needed a reboot (new
kernel), which is a shame as several were either close to or over a
year of uptime.


I recently rebooted a co-lo that had over 1130 days on it. Locked down 
to buggery, no user accessible shells and thus no CVE's that required a 
kernel upgrade. Don't fix it if it ain't broken. It does have the fixed 
glibc however.


Heck, my WD Mybook had 420 days on it until my mother-in-law plugged her 
hair dryer into the UPS power point last week and popped the breaker.


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] systemd==bad

[Brad Campbell]

On 22/02/16 22:52, Simon Hobson wrote:


But then I still have Squeeze and Lenny systems running (they aren't broken 
...) - don't think I have anything older than that !



I just bumped up against a problem with a squeeze system. It's ppc, and 
everyone has dropped the non-x86/x64 archives. That made it hard to 
install tcpdump to do some investigative work.


Interestingly the package files are still all present, so after an 
apt-get update ; apt-get install foo I just put the package names it 
can't download into google and they turn up in odd corners of the net.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Bad UEFI: was Systemd at work: rm -rf EFI

[Brad Campbell]

On 05/02/16 15:30, Edward Bartolo wrote:

Hi,

The argument of those who support protecting the hardware against a
probable breakage are logically sound: I support them.



I see it simpler than that.

I've always believed that best practice was if you don't need it mounted 
rw, then don't. For example all my machines (deskop, laptop & servers) 
always have /boot as a separate small partition. It's always kept 
mounted ro unless I need to update the kernel or bootloader.


I have a couple of EFI machines (Macs) and I sure as heck don't keep any 
of the EFI firmware directories even mounted, let alone mounted rw.


I see it as another "Well it makes our life easier, we don't see a 
problem with it and if you nuke your system it's really not our fault 
even though we did it behind your back without warning".


Which is why I'm here after 20 years of exclusive Debian use. I tried 
systemd during the Jessie freeze. It broke during the upgrade and I got 
lost trying to figure out how to fix it. I can fix a broken SysV system 
in my sleep, and I'm old enough not to want to learn new stuff just 
because someone thinks it's a good idea.


Brad.


___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Purchasing a new computer/laptop

[Brad Campbell]

On 28/01/16 11:15, Arnt Karlsen wrote:

On Thu, 28 Jan 2016 10:05:08 +0800, Brad wrote in message
<56a97754.7050...@fnarfbargle.com>:


I'm running an early 2015 Macbook Pro. I have OSX, Windows 7 & Linux
installed. I live in Linux, but on the rare occassions I need to boot
into OSX or Win 7 I simply s2disk. That way when I re-boot into linux
it just wakes up where it was rather than losing any context.

Aside from the non-functioning camera, the only other hardware issue
I routinely see is the USB SD reader mostly disappears off the bus
after a system sleep.


..doesn't your early 2015 Macbook have enough opmh to run all
these OS'es off e.g. https://www.qubes-os.org/  (aside from
their systemd issue)?


Yeah it does, but occasionally when doing things like firmware updates 
to peripherals it's just a shitload easier to run the native OS on the 
bare metal rather than fart around with USB re-direction and hardware 
passthrough. For those reasons, I keep a windows and OSX install on the 
bare metal.


Sure, I *can* do it in Linux via a VM, but I'm old enough now to not 
want to do time consuming shit like that just because I can. Time is 
money. Suspend linux, boot windows, update device, resume Linux.. job done.


I've snipped the remainder of your reply because personally I could not 
make head nor tail of what you were on about and it looked like a 
steaming pile of misdirected political 

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Purchasing a new computer/laptop

[Brad Campbell]

On 27/01/16 23:57, Simon Hobson wrote:

Wim  wrote:


I still have my previous model, I suppose I ought to try a native install on it 
- and perhaps see if I can get OS X running as a VM.

I would prefer dual booting personally, since running OSX in a VM isn't always 
perfect. Fi, access to external hardware over USB, like audio interfaces, 
doesn't work properly. Harddisks and the usual stuff like HID devices just 
works.


I don't like dual boot. It's OK for something where all OSs are only used 
intermittently so it's no problem to shut down and boot into something else. 
It's a real PITA when the main OS is used all day long, and you've a gazilion 
web pages open, and need access to your mail while booted in the other OS, and 
...
Obviously that's a matter of personal preference.


I'm running an early 2015 Macbook Pro. I have OSX, Windows 7 & Linux 
installed. I live in Linux, but on the rare occassions I need to boot 
into OSX or Win 7 I simply s2disk. That way when I re-boot into linux it 
just wakes up where it was rather than losing any context.


Aside from the non-functioning camera, the only other hardware issue I 
routinely see is the USB SD reader mostly disappears off the bus after a 
system sleep.



--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Slackware now uses PulseAudio...

[Brad Campbell]

On 18/01/16 10:00, Simon Wise wrote:


installing it is one step ... but you will not get sound from your
desktop apps to your speakers without setting things up to achieve that.
Normally using Jack you will have a number of other apps in the chain
... say processing some audio path, or a gui to set the mix, or a matrix
of some sort that connects all the elements in your audio chain.

Many find the everything-in-one-app approach easier (and learn to match
their ambitions to the workflows and plugins available) but jack offers
the advantages of linking lots of small apps (each built along the unix
line of making a program do one thing and do it well) and allowing the
user to arrange the ones they prefer in the configuration they want.
Which takes a little work on the part of the user to set it all up.


My point was using Jack for 'desktop audio' is a bit like booting your 
PC by keying in the bootloader using front panel toggle switches. It's 
just not the right tool for the job. Now, if you want to multi-track a 
couple of channels of audio including playback and live monitoring, and 
even patch some FX into the chain, then pulse just ain't going to cut 
the mustard.




___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Slackware now uses PulseAudio...

[Brad Campbell]

On 18/01/16 02:23, Steve Litt wrote:


In all fairness, I've found few softwares as difficult to install and
get right as Jack. In fact, of the five times I've tried to install it
on various distros, I've succeeded zero times.

So I'd settle for Pulse (or ALSA or OSS) over Jack simply because I can
actually get those installed.



Jack is one of those interesting cases. That you could not get it to 
work indicates you don't need it. If you needed it then you'd figure it 
out. If you need Jack it's because any of the other 'sound systems' are 
useless to you. Real time (ie multi-track studio work) is one of those 
instances (actually it's the only one I can think of).


Interestingly, installing Jack for me on Debian systems was a matter of 
download, compile, install and run. No frustration required. Probably 
because the hardware and drivers I was using was the sort of stuff Jack 
was written for.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Giving Devuan sans-initramfs capabilities

[Brad Campbell]

On 02/01/16 02:18, Rainer Weikusat wrote:

Steve Litt  writes:

[...]



For a real deployment, this is usually just humbug and can be replaced
with a kernel containing the drivers necessary for mounting a root
filesystem.


That's nice, until you want to do something like an encrypted root, or 
encrypted swap with suspend/resume. That's pretty hard without an initramfs.


And before you say that's a special case, it's pretty standard for any 
laptop in a sensitive environment. In fact we have servers configured 
that way.


Brad
--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [DNG] Jeep Cherokee hacked

[Brad Campbell]

On 22/07/15 10:15, James Powell wrote:

http://www.foxbusiness.com/technology/2015/07/21/hackers-discover-way-to-remotely-control-jeeps/?cmpid=cmty_twitter_fb

It appears, from the looks of things, Ashley Madison isn't the only one
getting hacked.

I'm wondering how this is happening so rampantly recently, but I'm
trying hard not to point any fingers.


Good, stop all the completely unfounded and uninformed hysterical 
speculation. It has absolutely nothing to do with Devuan or this list.



Does anyone know what OS is being used by these vehicles?


They are achieving it via an on-line firmware update and downloading a 
completely new firmware to the entertainment unit. OS is irrelevant.



___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [Dng] How to bust into a broken Qemu VM?

[Brad Campbell]

On 17/05/15 21:32, Steve Litt wrote:

On Sat, 16 May 2015 17:12:04 +0800
Brad Campbell  wrote:


On 16/05/15 07:37, Steve Litt wrote:


When you accidentally bork a Qemu VM such that it won't boot to a
virtual terminal, how do you bust back in. I doubt System Rescue CD
would help, unless you can boot from the "cdrom" but somehow also
access the existing "hard disk" borked VM image.

So how do you bust back into a borked VM?


With systemrescuecd.
qemu -hda my.img -cdrom /path/to/sysresccd.iso -boot d

Easy peasy.

Brad


Thanks Brad,

Your method both worked and was dead bang easy. Yesterday I performed
this procedure about 10 times in 2 hours. Note that System Rescue CD
requires -vga std in order to have anything resembling a decent image.
And of course I used --enable-kvm so I wouldn't get gray hair waiting
for it to boot.


While I'm here there is another qemu trick I find handy.
I have a requirement to run bespoke software that relies on various 
versions of windows (specifically 2003 & 7) and I need a _fast_ machine 
to run them. I procured a nice fast SSD and partitioned it up, and I 
pass the *partitions* to qemu.. so


qemu -hda /dev/sdb1

This makes the guest think its running in a disk of $partition size. 
This can then be partitioned in the guest. So a partition inside a 
partition. I can access these from the host with kpartx, but it prevents 
windows from seeing anything else on the machine (most of all the other 
version of windows next door) but leaves me with the performance of a 
raw partition and native IO.


I have loads of little qemu guest files with everything from dos through 
to windows & macos. When I need to rescue anything I can just run up 
that VM with the hard disk or partition passed through. I use 
systemrescuecd a _lot_ though.


When testing new OS variants, I use a qcow2 guest file, and snapshot it 
before making changes. That makes it super quick and easy to roll back.


Regards,
Brad
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [Dng] How to bust into a broken Qemu VM?

[Brad Campbell]

On 16/05/15 15:37, Peter Maloney wrote:

On 05/16/2015 02:00 AM, Adam Borowski wrote:

On Fri, May 15, 2015 at 07:37:57PM -0400, Steve Litt wrote:

When you accidentally bork a Qemu VM such that it won't boot to a
virtual terminal, how do you bust back in. I doubt System Rescue CD
would help, unless you can boot from the "cdrom" but somehow also
access the existing "hard disk" borked VM image.

Well... you use the exact same arguments as you used for installing the
system in the first place.  That is, give qemu both the hard disk image and
the CD.  Then -boot d to boot from the first attached CD.


Or you can have more fun and mount it on the host...


losetup /dev/loop0 disk1.img


With the caveat that only works for RAW format images though.


Brad
--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [Dng] How to bust into a broken Qemu VM?

[Brad Campbell]

On 16/05/15 07:37, Steve Litt wrote:


When you accidentally bork a Qemu VM such that it won't boot to a
virtual terminal, how do you bust back in. I doubt System Rescue CD
would help, unless you can boot from the "cdrom" but somehow also
access the existing "hard disk" borked VM image.

So how do you bust back into a borked VM?


With systemrescuecd.
qemu -hda my.img -cdrom /path/to/sysresccd.iso -boot d

Easy peasy.

Brad
--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.

--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [Dng] About (k)dbus in LKML

[Brad Campbell]

On 28/04/15 21:00, Alex 'AdUser' Z wrote:

https://news.ycombinator.com/item?id=9450806

Hot discussion about merging kdbus in kernel.

TL;DR: The people who talk about how kdbus improves performance are just
full of sh*t. (c) Linus


It is certainly a deep and wide ranging thread. When looked at from the 
10,000 foot view it does look like the dbus/systemd people have painted 
themselves into a corner and are trying to push an ill-conceived and 
incomplete hack into the kernel to try and work around it. It seems to 
have very little tangible benefit and lots of warts all over the place.


Hopefully it does not get legs and people spend some time coming up with 
a well thought out alternative solution that works for more than systemd 
lock in. I don't hold a _lot_ of hope however.


Regardless, Devuan is a step in the right direction.

Regards,
Brad
--
Dolphins are so intelligent that within a few weeks they can
train Americans to stand at the edge of the pool and throw them
fish.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng

Re: [Dng] dev-list

[Brad Campbell]

On 09/04/15 23:56, Laurent Bercot wrote:

On 09/04/2015 10:37, Jaromil wrote:

a -Dev list is there already, just not public and invite only.


  That's really a shame, because I would love to have access to that list -
even read-only. Isn't it possible to open subscriptions while keeping
posts moderated ? (posts from devs would be auto-approved, of course)



This +1. I don't need to post, and in fact it's probably better if I 
don't, but I actually *really* want to see what is going on under the 
bonnet. I joined the dng list for probably the same reasons everyone 
else did. I want to *use* what is produced and even possibly contribute 
(bug reports and patches I suppose are my limits). I do understand the 
technical components, and I don't want to ask user level questions.


I will say I'm completely disillusioned by the quantity of pontificating 
and bike-shedding that seems to dominate this list and I'd love to read 
a list with actual technical substance.


Don't lock us out. Moderate us, or create a read-only gateway, but don't 
isolate us. Please.


Brad.

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng