Fatdog64-802/801/800 Final [21 May 2019]

[Sage]

re. above discussions of Bluetooth and, separately, wifi mice and k/b s, inter alia. Might seem like a good idea in certain circumstances, but the remote items all must have a battery inside and experience shows that those tend to peg-out at the most inconvenient times. Best advice may be to avoid?!

[rufwoof]

@rufwoof: busybox bin sizes.

Thanks for the interesting size comparison. The huge static size comes because you're building a static binary with glibc. You will get a much smaller figure if you use uclibc or musl libc - but they present another set of problems (certain applets won't build etc).

In the usual Fatdog case, we don't compress the initrd at all, so the size difference does show up. The reason why we don't compress it is because we carry the initrd during uptime of the system (unlike standard Puppy, which removes its initrd once the main basesfs has been booted up); hence the entire uncompressed initrd will stay in memory. Thus, we have to make sure that the uncompressed initrd must remain the smallest size as possible - because it is an extra baggage.

Thanks James. Learning. Rebuilt the kernel with the default config and without libc inbuilt the vmlinuz is a much more comparable size.

# uname -a
Linux bulldog64 4.14.153 #11 SMP Tue Nov 12 12:02:28 UTC 2019 x86_64 GNU/Linux
# free -m
total used
Mem: 3404 22
busybox 1.31.1 ... Fixes for dc and wget (+ others)
BusyBox v1.31.1 (2019-11-12 11:05:20 UTC) multi-call binary.

Opted to build for 'performance' rather than 'size' and used lz4 compression instead of xz ... yielding a <21MB vmlinuz (versus <12MB). Get's from BIOS to wifi net connected cli extremely quickly

[SFR]

bin sizes ... 2,729,960 bytes (for 1.31.1) versus fatdog's 861,696 bytes

If it comes for the size and the max. amount of applets, I've got the best results with Buildroot, which uses uClibc-ng.

Code: Select all

# ./busybox --list | wc -l
412
# 
# du -k busybox 
1128	busybox
# 

It's a 32bit binary, though, 64bit one would be slightly bigger.

If you'd like to fiddle with it, here's a quick start:

0. Download and extract Buildroot https://buildroot.org/download.html (I used "Latest stable release: 2019.08.2")

1. Download the attached tarball and extract it to ./package/busybox/
It bumps the BB version from 1.31.0 to 1.31.1, adds a few Fatdog patches and contains a config file with practically everything enabled

2. unset LD_LIBRARY_PATH PKG_CONFIG_PATH CPLUS_INCLUDE_PATH

3. make menuconfig
(to choose the arch and optionally other things to build)

4. FORCE_UNSAFE_CONFIGURE=1 make
(if you're running as root)

It's gonna take a while, since it builds the toolchain first.
When it's finished, you'll find the BB binary in ./output/target/bin/.

Greetings!

[jake29]

I'm updating bluez to 5.52.

The current build of bluez-alsa doesn't support aptx, because we don't have the underlying aptx library in it. In theory it would still work but it would fallback to using other codec.

I just updated bluez-alsa build to v 2.0.0 but I would withhold further update as it has a sudden growth spurts after being relatively dormant for months; so I will wait until it stabilises a bit.

Thanks James. I've already looked into bluez-alsa aptx support, and it does exist experimentally (and the impression I get is that it will be implemented officially sooner rather than later.) I have compiled this custom version and have it up and running - but have not tested it to date. Please see the 'Contributed Packages' thread post below:

http://murga-linux.com/puppy/viewtopic.php?p=1041032&search_id=1689710725#1041032

Also, regarding gnome-keyring - I've had no success with it so far. Please note, the specific app is launched using --no-sandbox. Is that a contributing factor?

/var/log/messages: (after reboot)

Code: Select all

Nov 12 21:39:22 fatdog64-ccb user.notice kernel: Initialise system trusted keyrings

/var/log/messages: (after running the gnome-keyring shortcuts in /etc/xdg/autostart)

Code: Select all

Nov 13 17:19:47 fatdog64-ccb auth.notice gnome-keyring-daemon[4822]: couldn't access control socket: /tmp/runtime-root/keyring/control: No such file or directory

@Sage - I have been using Microsoft wireless mice for years, and cannot recall any peg-outs other than when the battery went flat (red light warning some time prior). Bluetooth mice do seem a little more challenging however, but then I guess I like a challenge.

[rufwoof]

If it comes for the size and the max. amount of applets, I've got the best results with Buildroot, which uses uClibc-ng.

Thanks SFR

[rufwoof]

Both /etc/wpa_supplicant.conf and /proc/cmdline are readable by 'others', which if they contain wifi ssid and password ... not good (a browser with a flaw even though running as spot could enable remote reading of your wifi ssid and password).

[jake29]

Hi all. I was hoping to upgrade the Intel WiFi card currently in my laptop - Intel AC 7265 - not just to improve WiFi connectivity, but to also benefit from Bluetooth 5.0.

Pricing for older models is almost the same as for the latest 2019 card 'AX200' - however I noticed that official support page states that Kernel 5.1+ is required.

https://www.intel.com/content/www/us/en/support/articles/000005511/network-and-io/wireless-networking.html

Can anyone comment on the back-porting of WiFi adapter drivers? Could the driver be made work with the current Kernel 4.19? Is an upgrade to Kernel 5.1 planned any time soon?

EDIT: I followed a back-porting tutorial here and build: intel_ax200_backport-x86_64-802.txz

[jamesbond]

Thanks rufwoof, fixed for future release.
Thanks jake29 for the module.

[don570]

I have been reading about bluetooth drivers and I found this software
that claims to be universal driver for linux...
https://github.com/torvalds/linux/blob/ ... th/btusb.c
__________________________________

[don570]

I've been using pclock for a countdown timer but the warning sound isn't loud enough
so I replaced the warning sound with 2barks by modifying pclock script...

Code: Select all

echo '/usr/share/audio/2barks.au' > $WORKDIR/COUNTDOWN_ACTION

____________________________________________

[don570]

I suggest that instructions for auto-mounting ntfs Windows partition be included in /usr/share/ docs
________________________________________________

[jake29]

I have been reading about bluetooth drivers and I found this software
that claims to be universal driver for linux...
https://github.com/torvalds/linux/blob/ ... th/btusb.c
__________________________________

I would say Bluetooth support is working very well currently in Fatdog64-802, at least in regards to the mouse I have tested. My only problem is that both mouse and my Bluetooth chip are v4.1. I was motivated to buy a bluetooth mouse for the benefit of what I hoped would be improved tracking and distance. Unfortunately, this has not been entirely the case.

I have now ordered a replacement Bluetooth 5.0 mouse and an Intel AX200 WiFi / Bluetooth card which supports v5.0.

[belham2]

Hi Fatdog,

Is there a problem with the "Save as multisession" function of Fatdog?

For example:

1) a pristine boot of a frugal folder of Fatdog-802

2) Boots up great (using initrd-nano)

3) Make changes to Fatdog, like setting Time and Format, disabling and enabling services (using Fatdog64 Service Manager), then using "Manage desktop startup programs" to set some things.

4) Then I want to test multi-save session function: I click reboot, go thru the process, creating new passwd and then clicking "Save as multisession"

5) Fatdog creates the appropriate multisave-base.sfs and mutlisave-save.sfs as it shuts down & reboots itself


6) Fatdog re-boots back up, goes to the same pristine desktop, then I....

7) go into Control Panel, System tab, click on "System SFS Loader". Tell SFS Loader where those two SFSes (above) are, in the same sda1 partition as the main Fatdog folder, and then load the two SFSes. It says they load and are loaded (and manually checking, clicking on the SFSes, shows all the changes I had made to the SFS are in these two SFSes that had been previously created)

Ok, great, but not one change (inside the multisave.SFSes) that i had made specfically to Fatdog's settings is loaded, i.e. no clock changes, no services that had been set and started (before the multi-session save was enacted), nothing. Other stuff like browser changes and stuff is loaded, but not other things.

What am I doing wrong?

I would like Fatdog to boot up pristine every time, and then I will manually load whichever previous-session-reated "multisession.SFS" file I choose. But, darn, this is just not working.

Fatdog is only loading some of the changes from the Multisession.SFS files (when loading then after the next prisitine bootup). Pointedly, none of the services changes, toolbar changes, etc, etc are loaded when the base and save-SFSes are loaded. Very frustrating; I got so frustrated had to walk away for awhile as I was getting angry at myself for being too stupid to understand what is going on . (Also, please know, I tried this repeatedly with Fatdog booting on different USB EXT4-formatted sticks, thinking maybe a stick was bad, but it's not the USB sticks).

Is it possible I have my menu.lst entry all wrong for this to work (pristine Fatdog boots every time, no savefile whatsoever, and loading previous-boots-created multisession.SFSes only)? Here is the menu.lst entry I use when Fatdog64-802 boots up:

Code: Select all

"title Fatdog64-802"
   uuid ########-####-#####-##########
   kernel /Fatdog64-802/vmlinuz rootfstype=ramfs mergeinitrd1=local:/Fatdog64-802/initrd waitdev=3
   initrd /Fatdog64-802/initrd-nano

It boots great, and fast, but like mentioned before, many of the changes that the 'Save as Multisession" SFSes should have saved do not load after I manually choose them at the next pristine boot.

Thank you for any help regarding this.[/code]

[rufwoof]

In multi-session save mode, you set Fatdog to load all of the prior saves automatically at next reboot, there's no need to manually set/load them by-hand, you just add the appropriate 'savefile' kernel boot parameter.

You can't be selective, as they're sequential save sfs's, so missing one out in the middle would mess things up. You can however opt to not load the last N saves. For readability I've shortened down my actual uuid in the following, but that's the kernel boot paramater I use for multi-session. The last two ":" don't have to be included, but I leave them as that way its a reminder of where the 'skip' N saves value is

savefile=direct:multi:uuid:5d::

i.e. if I change that to

savefile=direct:multi:uuid:5d::2

then it wont load the last two saves.

Or you could just manually delete the last (most recent) save sfs

When you have say 10+ saves and want to consolidate them, I boot (so all the saves are loaded into ram), manually delete all of the multisession save files (sfs's) - backing them up first just in case, and then click the desktop 'save' icon ... and its back down to just two save sfs's.

My full menu.lst entry looks like
title FatDog
root (hd0,0)
kernel /fatdog-vmlinuz pkeys=uk waitdev=5 basesfs=ram:uuid:5df8f89e-33d5-4720-b3f2-9c9030a718bd:/fd64.sfs savefile=direct:multi:uuid:5df8f89e-33d5-4720-b3f2-9c9030a718bd::
initrd /fatdog-initrd.xz

Note that in my case I originally clicked on the initrd to open it, dragged out the fd64.sfs from that, and then closed the initrd again (by clicking the repack-initrd file) so as to have the main sfs (fd64.sfs) outside of initrd. Note also that I've renamed the vmlinuz and initrd (and I've also opted to xz compress the initrd) ... i.e. not they're not the default/standard names

The appeal of multisession for me is that I install it all to and boot/run from usb stick (bootloader etc.), and all saves are written back to that usb. But where I have Event Manager's Save Session Interval set to zero so it only ever saves on demand, and the usb can be removed between times (i.e. is physically isolated after having booted up the system). Mostly I'm very frugal with saves i.e. tend to just boot, use, shutdown without saving, and I'm careful as to what saves are added (boot clean, make changes, save ... without doing anything else so that the system remains 'clean'). I keep my bookmarks in a text file, and have tilda drop down terminal installed, where one of its tabs has that bookmarks text file open (and where hovering-over and clicking a link opens up that url in chrome). Chrome is about the only reason I run 'save' i.e. boot, use the control panel to update to the latest version of chrome, save.

Another appeal is that Fatdog only improves with age, each later release builds upon the last release. With Puppy it often seemed one step forward two steps back i.e. some past improvements were suddenly lost.

Yet another appeal of Fatdog for me is that its alsa equaliser works. i.e. running alsamixer presents the usual alsa mixer, but running alsamixer -D equal ... presents a working equaliser. And, at least for me, its alsa supports multiple sound sources being played simultaneously.

[rufwoof]

If you do use a usb for your entire fatdog system, just use the HDD for data, then noteworthy is that Fatdog includes encfs ... i.e. create a top level folder called DATA (or whatever) and store files/folders under that, and use encfs to encrypt that DATA folder (in rox, right click on the folder and its one of the options there). That way if your laptop is lost, your data is at least encrypted. I also do similar for the ~/.ssh files stored on usb - but using a more intensive encryption method, so if my usb is lost/stolen then my ssh keys are strongly encrypted. Whilst I could use the standard ssh encrypt key I prefer a stronger encryption than the 'built-in' encryption.

Booting to ram from usb using multi-session saves, where that usb is removed once the system has booted, combined with being careful/frugal with 'saves', along with data (and ssh keys) being encrypted, and provided you also keep good backups of your data (to physically disconnected media) ... makes Fatdog one of the best systems around IMO.

[jake29]

Hi all. I am getting the following error when a streaming app drops it's connection:

Code: Select all

(shadow-preprod:11390): libnotify-WARNING **: 23:02:51.651: Failed to connect to proxy

Does this indicate libnotify plays a role in the crash or is only reporting it?

I have tried to pkgbuild libnotify 0.7.8 just in case, however it now seems to rely on MESON.

[SFR]

@Jake29: There's v0.7.7 is in the repo, but it requires GTK3.
It might be the culprit indeed, though: https://github.com/meetfranz/franz/issues/592

Greetings!

[jake29]

@Jake29: There's v0.7.7 is in the repo, but it requires GTK3.
It might be the culprit indeed, though: https://github.com/meetfranz/franz/issues/592

Thanks for the reply SFR. I have v0.7.7 installed already, that's why I looked into upgrading. I have had this problem for many months, talking with the app developers has not been productive (mainly because issue does not occur in Ubuntu). Could you assist with building libnotify v0.7.8? I have tried my best on a few occasions now.

[SFR]

Sure, but it turned out that 0.7.8 requires meson 0.47+ to build, we have 0.46.1.
But ok, the attached contains both meson-0.52.0 and libnotify-0.7.8 recipes.

Anyway, is there an option to disable notifications in that app you're using?
It seems that it helped the OP in the previously linked GitHub issue.

Greetings!

[jake29]

Sure, but it turned out that 0.7.8 requires meson 0.47+ to build, we have 0.46.1.
But ok, the attached contains both meson-0.52.0 and libnotify-0.7.8 recipes.

Anyway, is there an option to disable notifications in that app you're using?
It seems that it helped the OP in the previously linked GitHub issue.

Greetings!

Thanks SFR for this. I was able to build and install the updated libnotify. Unfortunately, there has been no change so far regarding crashing. There are zero in-app options related to notifications. I think I will need to boot into Ubuntu and observe notification behaviors. It could well be the source of the issue.

The app will only crash in windowed-mode; full-screen it is perfectly stable. This could lead to the solution if I can convince the developers to allow suppressing of notifications or include a safe-mode startup command.