XenialDog 64bit (Ubuntu 'Xenial Xerus' LTS, 64-bit)

A home for all kinds of Puppy related projects
Message
Author
Lassar
Posts: 235
Joined: Tue 08 Jul 2014, 20:01

#286 Post by Lassar »

Installed icewm, copied the /root/.icewm directory from my kodipup.

Edited .xsession putting in icewm-session.

Copied the /etc/xdg/menus from kodipup to xenialdog /etc/xdg/menus.

Copied the /etc/xdg/templates to xenialdog /etc/xdg/templates.

Went to the command prompt and type startx.

Icewm came up but the menu was all wrong.

Is there anyway to use fixmeus from puppy linux in xenialdog?

I really like the way puppy linux does the icewm menus.

wiak
Posts: 2040
Joined: Tue 11 Dec 2007, 05:12
Location: not Bulgaria

#287 Post by wiak »

Hi Fred,

Just noticed something a bit weird with pulseaudio - not sure if it is working correctly for me in pristine frugal XenialDog64 install after pulseaudio --start. When I open alsamixer, it is still showing all the alsa controls. Repeating the same steps in XenialDog32 when I open alsamixer it just has singe pulseaudio volume (which is more what I expected). Is pulseaudio working okay in XenialDog64? - I don't understand the difference with alsamixer between XenialDog32 and XenialDog64. Any ideas?

wiak

EDIT: 'pactl list' command seems to suggest pulseaudio is working - I just can't understand how alsamixer is different between xenialdog32 (just shows a pulseaudio volume control) and xenialdog64 (shows all the usual alsa volume controls). Whether there is a problem with pulseaudio in xenial64 I'm still not sure... by the way we don't need 'killall pulseaudio' to kill the pulseaudio daemon, there is a command: pulseaudio -k

wiak
Posts: 2040
Joined: Tue 11 Dec 2007, 05:12
Location: not Bulgaria

#288 Post by wiak »

It's not that I'm actually so interested in pulseaudio by the way, though I do like using it (so I become more familiar with its facitlies really). Main thing I'm trying out just now is bluetooth connectivity and having some issues which I thought might be pulseaudio related. I have mainly been trying the package "blueman" but thereafter been playing with rcsrn51's bt4stretch (but in xenialdog64).

Oh, btw, part of my memory has returned - we have had several issues/discussions about pulseaudio, Fred, and I remember now:

Code: Select all

alsamixer -D pulse
I keep promising myself to take better notes since so many discussions, so many things discovered, but thus also so many things forgotten...

I suspect overall that my pulseaudio is all working on my xenialdog64 system - not bluetooth quite as yet - having problems with pin string being requested and me not knowing what the pin is, but bluetooth working from Windows 7 on same system (easily...) so I have to get to the bottom of this.

wiak

wiak
Posts: 2040
Joined: Tue 11 Dec 2007, 05:12
Location: not Bulgaria

#289 Post by wiak »

On a separate topic (not just xenialdog specific) here is a piece of information that some may find useful/interesting (others will probably just know it already). I've been comparing 64-bit Firefox version 57.0.4 with Google-Chrome 63.0.3239.108 (Official Builds) on my xenialdog64 system (though the results are not distribution specific). People claim Chrome uses more RAM than Firefox but frankly, according to my tests, there is very little difference (nothing practically relevant for me).

The reason I have been testing these is that sometimes I get browser 'freeze' - unresponsive mouse etc (in fact my whole system freezes and I have to reboot usually...). I wondered if it was a particular OS distribution issue or if it was browser-related only. Anyway, I have definitely discovered it is wholly browser eating up RAM that is causing the problem since I now know how to repeat the problem reliably (i.e. it is not an intermittent problem due to OS or physical memory fault on my system, or even a browser memory leak... it is simply the modern browser using so much memory).

Test I use is to open multiple tabs, but not just any tabs. Simple web pages don't use much memory per tab... But complex pages, like that from Gmail, uses somewhere around 150 MB to 200 MB per tab opened! So what I do in my tests is to open a terminal and run top (to watch what happens to free memory, memory used, and memory available) and then start opening gmail tabs (right clicking on input, send, trash links etc till I have multiple tabs opened all containing gmail-related pages). On my 2 GB RAM system I find with both these versions of Firefox and Chrome that my system keeps running on 7 or maybe just 8 such tabs but as soon as I try to open one more... the 'top' reported 'memory free' and 'memory available' just get too small to allow that one more 150 to 200 MB RAM required for that extra tab. If I open that one-too-many tabs then the system absolutely reliably freezes - Firefox being no better in this test than Chrome - same result...

Solution is to use less RAM hungry browser (latest Opera gave pretty much exact same result sorry...) or add more RAM to your system or if you don't have that create a bit breathing space by:

Code: Select all

apt-get update && apt-get install zram-config
need to reboot to get that zram swap then available

i.e. with zram swap space thus created I can open twelve or more of these gmail RAM-hungry tab pages and the system response time is generally unaffected since the zram doesn't get touched unless many tabs are already opened... Facebook is another RAM-hungry page of course, as are any google search pages with 'Images' clicked (i.e. pages full of images). No doubt some Flash or Javascript webpages are sometimes just as bad or worse in the amount of physical RAM allocated to them (note it really is RAM allocated - nothing to do with browser cache or system cache - clearing these does not help... increasing RAM is the answer to stop system freeze here).

wiak

User avatar
fredx181
Posts: 4448
Joined: Wed 11 Dec 2013, 12:37
Location: holland

#290 Post by fredx181 »

Hi wiak, thanks for the info.
To reply on your earlier post:
EDIT: I can't add /mnt/home as a bookmark in pcmanfm it seems...
Not by drag and drop (probably because /mnt/home is symlink) but you can, by going inside /mnt/home and add bookmark from top menu (or press Ctrl+D)

@Lassar
Sorry, can't help with Icewm, seems like you did all the right things but looks like Puppy menu setup is very different from Xenialdog.

Fred

wiak
Posts: 2040
Joined: Tue 11 Dec 2007, 05:12
Location: not Bulgaria

Failed to initialize glamor - need xserver-xorg-video-intel

#291 Post by wiak »

Some further info in case anyone has machine that has the following X graphics problem:

Tried installing XenialDog64 on old emachines 355 pav70 10.1in display (1024x600) netbook which uses Intel Atom N570 processor (mine has 2GB RAM).

X wouldn't start because of modesetting driver error:
(EE) modeset(0): Failed to initialize glamor at ScreenInit() time.
Fixed by installing xserver-xorg-video-intel (which has dependency libxvmc1 if you are installing manually with dpkg -i rather than apt).

I have six old laptops at home, which my family use, all using this same xenialdog64 configuration. Attached screenshot gives an idea of the apps I have installed. Note that I prefer to mount drives from pcmanfm and don't use desktop-drive-icons. I have Firefox, Chrome, and WPS Office installed on /mnt/home as basically portable apps and just symlink to them from desktop so can use between other installed distributions. Desktop uses Rox pinboard and tint2 bottom panel for menu/volume/clock etc. I use porteus boot and my 01-filesystem.squashfs is LZ4 compressed and of size 784MB total. I keep my changes folder empty (i.e. I don't save on shutdown, though if I want something added I do a quick-remaster). Finally, my preference is to have conky disabled at startup (but the toggle button is there for if I want it).

wiak
Attachments
xenial_screenshot.jpg
My XenialDog64 system screenshot
(40.54 KiB) Downloaded 513 times

User avatar
rcrsn51
Posts: 13096
Joined: Tue 05 Sep 2006, 13:50
Location: Stratford, Ontario

Re: Failed to initialize glamor - need xserver-xorg-video-intel

#292 Post by rcrsn51 »

wiak wrote:Fixed by installing xserver-xorg-video-intel (which has dependency libxvmc1.
I now add this package to my mklive builds. With all my non-Atom machines, the kernel can find a decent video driver OOTB.

But the Atom machines really need it.

backi
Posts: 1922
Joined: Sun 27 Feb 2011, 22:00
Location: GERMANY

#293 Post by backi »

Hi wiak !

Regarding having little Ram but keeping System more fluent using Zram ........good Tip .......but not so much commonly known .

Me using Zram on a Toshiba Satellite 3000 with 1 gig Ram ........gives me a much better Performance than without .

" Increased Performance In Linux With zRam (Virtual Swap Compressed in RAM)

Author: Andrew | Posted: October 02, 2011 | Updated: June 13, 2013

Tweet

While trying to optimize the elementary OS performance, Sergey Davidoff stumbled upon a project called compcache that creates a RAM based block device which acts as a swap disk, but is compressed and stored in memory instead of swap disk (which is slow), allowing very fast I/O and increasing the amount of memory available before the system starts swapping to disk. compcache was later re-written under the name zRam and is now integrated into the Linux kernel.

I decided to give it a try, and the result on my desktop with a quad-core CPU and 2Gb of RAM was fantastic: instead of freezing after running out of RAM, the system worked like nothing happened. I didn't notice any difference at all. It looked just like adding more RAM! Surprisingly, I got almost the same results on a 6-year-old laptop with Pentium M and 1Gb of RAM! So, I've improved the script to automatically adapt to the amount of memory in the system and automatically scale across several CPUs or CPU cores, packaged it in .deb and uploaded to PPA.

- Sergey Davidoff


Sergey also mentions that the only thing that prevents this from being enabled by default in elementary OS Luna for now is the existence of Atom netbooks with fast SSDs for which he doesn't know if this would be useful or not.


This is especially useful for netbooks, old computers (or computers that don't have too much RAM), virtualization or embedded devices but of course, you can use it on any computer. "

There is some discussion going on if you have 2 Gigs ram or more ,it would be contra productive .

But there is no reason not to experiment with Zram .

If zram-config is installed via Synaptic.....it can be enabled or disabled on demand.

Here are 2 scripts to enable or disable zram on Demand :

Enable :

#!/bin/bash
swapon /dev/zram0

Disable :

#!/bin/bash
swapoff /dev/zram0

Regards !

User avatar
fredx181
Posts: 4448
Joined: Wed 11 Dec 2013, 12:37
Location: holland

#294 Post by fredx181 »

Thanks all for the info.
Regarding missing package "xserver-xorg-video-intel", it's added now to mklive-stretch and to all config files:
http://murga-linux.com/puppy/viewtopic. ... 225#980225

Fred

wiak
Posts: 2040
Joined: Tue 11 Dec 2007, 05:12
Location: not Bulgaria

#295 Post by wiak »

Just a small addition to what I wrote about xserver-xorg-video-intel in my previous post above. I note that with that intel driver installed, glamor is not longer used. It can be as the following shows, but as I say, I have no idea if that is an advantage or a disadvantage to force use of glamor also. What is Glamor - explained via link below too:

https://www.freedesktop.org/wiki/Software/Glamor/
The xf86-video-modesetting driver uses glamor by default but other drivers have support for it (xf86-video-ati and xf86-video-intel)
...
Why Glamor?
Basically, the biggest two advantages of Glamor are:

Graphics device has powerful 3D capability. To use 3D function to accelerate 2D rendering is possible and many drivers already do so
...
How to Enable Glamor on Intel? [wiak: i.e. when using xserver-xorg-video-intel]
...
Make sure you have the xorg configure file named glamor.conf at conf/glamor.conf under the directory /usr/share/X11/xorg.conf.d or /etc/X11/xorg.conf.d/.

[wiak: I used file called 20-glamor.conf containing the following:]

Section "Module"
Load "dri2"
Load "glamoregl"
EndSection
Section "Device"
Identifier "intel"
Driver "intel"
Option "AccelMethod" "glamor"
EndSection
...

After you finish all the above steps, then you can try to start x with glamor enabled DDX. To make sure it's glamor running, you can refer to Xorg.0.log, and check that Glamor is enabled if you can find the log like: intel(0): Use GLAMOR acceleration.
I didn't actually find any measurable advantage to using glamor. Indeed, vblank_mode=0 glxgears, indicated a very slightly lower frame rate (not higher), though I doubt that means much and maybe glamor would help if I ran 3D apps (which I don't...). There may be other advantages (I know little to nothing about this...)

wiak

wiak
Posts: 2040
Joined: Tue 11 Dec 2007, 05:12
Location: not Bulgaria

#296 Post by wiak »

Hi all, I have just discovered another good reason why installing xserver-xorg-video-intel might be a good idea: if you want to use Palemoon 64bit on Intel video-based machines (at least I found that this applies to mine... the info may also be relevant to XenialPup systems, where I kind of remember Palemoon was provided by default?):

http://murga-linux.com/puppy/viewtopic. ... 406#980406

EDIT:

A couple of posts down from that link I also provide some results of trying Palemoon on slacko64 system (which includes the intel driver by default). I discover there that renaming the intel driver and rebooting (so it isn't found) doesn't stop Palemoon working correctly so I reason that it is actually one of the dependencies of xserver-xorg-video-intel that is actually needed, not the xorg intel driver itself. Might be libdrm-intel1, though I'm guessing at this stage since haven't tried. That may also be the case with the Atom CPU machines - I'll test later (though easy solution remains that of installing xserver-xorg-video-intel which of course pulls in its dependencies....

EDIT2: No, the Atom-based cpu machines definitely need that intel-drv.so. As I say, Palemoon, on the otherhand, works correctly without crashing on going to facebook.com when xserver-xorg-video-intel is installed but doesn't actually need intel-drv.so itself - not on Puppy Slacko64 anyway, though could be needed on XenialDog cos of a timing issue I suppose... (too much rebooting involved trying to find the key dependency or exact solution, sorry, but installing xserver-xorg-video-intel certainly seems to be a working fix for that Palemoon using facebook issue).

wiak
Last edited by wiak on Tue 16 Jan 2018, 21:03, edited 5 times in total.

The Flying Cat
Posts: 40
Joined: Mon 10 Oct 2016, 12:23

#297 Post by The Flying Cat »

Is there a relatively painless way of installing another kernel? Or change kernel upgrade script to use another metapackage, say HWE from LTS releases or custom from PPA like liquorix? Maybe via integrating config files.
Was kinda curious about testing https://liquorix.net/, seems like it might be fast:)

dancytron
Posts: 1519
Joined: Wed 18 Jul 2012, 19:20

#298 Post by dancytron »

wiak wrote:Hi all, I have just discovered another good reason why installing xserver-xorg-video-intel might be a good idea: if you want to use Palemoon 64bit on Intel video-based machines (at least I found that this applies to mine... the info may also be relevant to XenialPup systems, where I kind of remember Palemoon was provided by default?):

http://murga-linux.com/puppy/viewtopic. ... 406#980406

EDIT:

A couple of posts down from that link I also provide some results of trying Palemoon on slacko64 system (which includes the intel driver by default). I discover there that renaming the intel driver and rebooting (so it isn't found) doesn't stop Palemoon working correctly so I reason that it is actually one of the dependencies of xserver-xorg-video-intel that is actually needed, not the xorg intel driver itself. Might be libdrm-intel1, though I'm guessing at this stage since haven't tried. That may also be the case with the Atom CPU machines - I'll test later (though easy solution remains that of installing xserver-xorg-video-intel which of course pulls in its dependencies....

EDIT2: No, the Atom-based cpu machines definitely need that intel-drv.so. As I say, Palemoon, on the otherhand, works correctly without crashing on going to facebook.com when xserver-xorg-video-intel is installed but doesn't actually need intel-drv.so itself - not on Puppy Slacko64 anyway, though could be needed on XenialDog cos of a timing issue I suppose... (too much rebooting involved trying to find the key dependency or exact solution, sorry, but installing xserver-xorg-video-intel certainly seems to be a working fix for that Palemoon using facebook issue).

wiak
On my desktop (Intel integrated G30 graphics) DD Stretch (Jessie was the same), everything is fine without xserver-xorg-video-intel until you install mesa-dri. Once you do that, it will crash trying to start x until you install it. On an old version of XenialDog 64, it had mesa-dri installed by default and no xserver-xorg-video-intel. It crashed before x until I installed it.

On Palemoon/facebook mystery, look in the settings and see if there is a switch to turn hardware acceleration on and off. In Firefox 57 it is kind of hidden, under preferences you uncheck "Use recommended performance settings" and then the specific settings appear. Maybe even hidden under about:config? I am guessing that maybe it is turned off by default on the one that worked and turned on where it didnt' work.

wiak
Posts: 2040
Joined: Tue 11 Dec 2007, 05:12
Location: not Bulgaria

#299 Post by wiak »

@dancytron: you have some interesting theories there.

Regarding XenialDog64 not managing to reach X (i.e. crashing out with error to console on booting). I have this problem on my machine, but solved it with the following "sleep 7" change to /etc/profile:

Code: Select all

if [ -z "${DISPLAY}" ] && [ $(tty) = /dev/tty1 ]
then
sleep 7 #wiak: was previously sleep 3 but failed to then start X
startx
fi
Clearly that is a timing issue so perhaps when you installed xserver-xorg-video-intel that had the effect on your machine of changing the timing sufficiently for your machine to successfully boot into X. But that would be a hit and miss machine-dependent affair possibly, hence using a large enough sleep value in /etc/profile to make sure plenty of time for earlier processes to complete before attempting to start X may be more reliable between machines. Having said that, the extra 'delay' involved in using xserver-xorg-video-intel may well also allow me to reduce the sleep value in /etc/profile on my own machine too - I haven't tried it but will out of curiousity soon... ;-)
(EDIT: Yes, installing xserver-xorg-video-intel did indeed allow me to reduce that /etc/profile sleep value back to 3 (maybe even lower, I haven't tried). It may be that xserver...intel has some kind of reliable 'wait' mechanism built into it of course or it may just be fortunate enough to add sufficient delay - impossible to say with certainty - either way, this also suggests that it is a good idea to install xserver-xorg-video-intel on such Intel graphics based machines...)

Regarding Palemoon. Yes, you can disable it using hardware acceleration via a checkbox in its advanced preferences dialog. Unfortunately that doesn't help with latest Palemoon 27.7.0 on my Intel graphics based laptop. Taking out libgl1-mesa-dri via apt-get isn't so easy (since removes lots/most of X-related components); I did try disabling it via renaming a few libs but that had no effect on Palemoon issue - would still crash out of X if given facebook.com url. But, as I said, installing xserver-xorg-video-intel always fixed the problem.

Nevertheless, I think it is a pity Palemoon requires that remedy; neither Chrome nor Firefox have any problem without xserver-xorg-video-intel and users who come across the problem with Palemoon should not expect such an issue (and may well not use Palemoon because they think it thus unstable). The Palemoon developers should look into why it happens and find a way round it since Chrome and Firefox manage... I suppose I should notify them of the issue and the 'fix'... but googling suggests that such Palemoon crashes seem to have been around for a long time.

wiak

stemsee

#300 Post by stemsee »

@The Flying Cat

I used to install liquorix kernels, generally to to build a fast kernel one configures 1000hz and add the realtime patch fromhttps://www.kernel.org/pub/linux/kernel ... t10.tar.gz latest kernel with realtime patch. Much of the realtime patches are already incorporated into the kernel.

You can use SUKK to build authentic debian kernel package and initrd, to get a realtime system up and running will require system tuning and optimisations (rtirq) and a framework for elevating and being given cpu priority over other processes consistently, not just a faster kernel. The system will need to be optimised for a purpose too, such as audio, video or compiling etc. The liquorice kernel config from its kernel source package could be used to roll your own with SUKK.http://murga-linux.com/puppy/viewtopic.php?t=96723 But I am fairly sure the liquorice kernel will NOT come with aufs compiled in, and some aufs patches may be difficult to apply on the liquorice kernel sources. But overlay should be ok. Maybe DebianDog could use an overlay init like Gyro has written for puppy.

Studio1337 has a realtime kernel. Ubuntu studio repo probably has realtime kernels available too. Though they usually use a regular kernel for running live cd with aufs built-in and a realtime kernel for full install.

I am not an expert, so do some research, and get advice from Fredx et al ...
regards stemsee

User avatar
fredx181
Posts: 4448
Joined: Wed 11 Dec 2013, 12:37
Location: holland

#301 Post by fredx181 »

The Flying Cat wrote:Is there a relatively painless way of installing another kernel? Or change kernel upgrade script to use another metapackage, say HWE from LTS releases or custom from PPA like liquorix? Maybe via integrating config files.
Was kinda curious about testing https://liquorix.net/, seems like it might be fast:)
Indeed, as stemsee wrote, the liquorix kernel will be problematic because it doesn't have aufs (porteus-boot needs it)
For testing, I just installed 4.14 kernel using "new-kernel" script, attached (NOT fake .gz, extract and run it) from here and works fine:
https://launchpad.net/~canonical-kernel ... hive-extra
Direct links to amd64 packages (you need also the linux-image-extra package (it contains aufs):
https://launchpad.net/~canonical-kernel ... _amd64.deb
https://launchpad.net/~canonical-kernel ... _amd64.deb
Just download the deb packages, run the new-kernel script and select them both (using yad file-selector).
You can also search for other linux-image .deb packages, but as said, the linux-image-extra-.....-generic... is required.

Fred
Attachments
new-kernel.tar.gz
Extract, run "new-kernel", and select .deb packages
(3.61 KiB) Downloaded 166 times

The Flying Cat
Posts: 40
Joined: Mon 10 Oct 2016, 12:23

#302 Post by The Flying Cat »

Yep, new 4.14.0-16 kernel as expected have first wave of meltdown/spectre patches

Code: Select all

root@live:~# cat /proc/cpuinfo | grep bugs
bugs		: cpu_meltdown
As an intel cpu owner I expected to get 0-10% performance drop (so liquorix could've been nice offset :P )
But so far I have seen none

Code: Select all

4.14.0-13
systemd-analyze
Startup finished in 5.495s (kernel) + 5.015s (userspace) = 10.510s
4.14.0-16
root@live:~# systemd-analyze
Startup finished in 5.263s (kernel) + 4.126s (userspace) = 9.390s
glmark2 also shows no performance drop, actually also small increase probably due better firmware (I heard they did a good job on opensource drivers)
I guess I'll fiddle a bit more with this

Forgot to mention: this is on Bionic Dog (really nice one btw)

wiak
Posts: 2040
Joined: Tue 11 Dec 2007, 05:12
Location: not Bulgaria

#303 Post by wiak »

backi wrote: Here are 2 scripts to enable or disable zram on Demand :

Enable :

#!/bin/bash
swapon /dev/zram0

Disable :

#!/bin/bash
swapoff /dev/zram0

Regards !
Hi backi,

Yes, its useful to be able to enable or disable the zram swap when wished. I should point out that sometimes there is more than one zram swap created. Possible it is one per cpu core. Certainly on my intel core II duo machine the swap automatically provided is combination of /dev/ram0 and /dev/ram1. One way of disabling both at once is simply to disable all swap with:

Code: Select all

swapoff -a
Cheers, wiak

backi
Posts: 1922
Joined: Sun 27 Feb 2011, 22:00
Location: GERMANY

#304 Post by backi »

Hi wiak !
I should point out that sometimes there is more than one zram swap created. Possible it is one per cpu core.
Yes thats`s the case .

Just one observation ---swapoff -a also disables swap-partition .

Zram is a cool thing for low-ram machines !

Regards ! :)

wiak
Posts: 2040
Joined: Tue 11 Dec 2007, 05:12
Location: not Bulgaria

#305 Post by wiak »

backi wrote: Just one observation ---swapoff -a also disables swap-partition .
Yes, I don't use a swap partition. But if you have a swap partition also does the system 'know' to use zram first before using slow swap partition?

wiak

Post Reply