process listing only showing 2 processes (in ps and top) - XDA Assist

I've noticed on Android 7.0 with Nov 5th patch level, that "ps" and "top" are only showing two processes owned by me. I am running this on the on-device command line through the Terminal Emulator app. This is on a nexus 6. Fully stock. Not rooted. It used to show all running processes. Any idea how to get that back? Is this normal?
shamu:/ $ ps
USER PID PPID VSIZE RSS WCHAN PC NAME
u0_a104 31024 31006 3544 1184 sigsuspend b42a77ac S /system/bin/sh
u0_a104 31048 31024 4532 1264 0 ababb5f8 R ps
shamu:/ $ whoami
u0_a104
shamu:/ $

Yep, normal. Linux allows you to lock things away from normal users.

Marshmallow shows all processes, though. Is there no way to list all processes on the android command line anymore? This ps doesn't seem to take switches like regular Linux ps.

Here, read up about it; https://www.cyberciti.biz/faq/linux-...m-other-users/
Hint: its "2".

Thanks for providing that detailed Linux info. It let me Google the specifics on Android.
Here is a bug filed about this, which Google shot down.
https://code.google.com/p/android/is...tail?id=205565
And a discussion on Stack Overflow.
http://stackoverflow.com/questions/3...android-nougat
I find this change unfortunate since it makes simply running top to see which proc is hogging my CPU impossible.

Just do it as root.

Related

Hacking the Boxee Box (Violates GPL)

Well I'm looking for some talent to get the Boxee Box bootloaders unlocked so we can put custom firmware on it and/or XBMC. Boxee is based of of XBMC which is opened source yet the Boxee Box is closed and violates the GPL, see here for more info.
Not only that but they have lied again and again to there customers are now since it has reached EOL, and will only provide one more minor update before they put the maintenance tag on it. People requested they open the box but Boxee still refuses, the firmware still feels like a beta yet they think it's good enough. The firmware is a beta at best and it will become an expensive paperweight soon enough, actually it is for most of UK that have lost all online content. Before that happens to everyone i'm hoping one of you step up
The hardware is good just the software is utter crap, I'm sure if i can get some interested developers we soon will be free of Boxee's strong hold. I don't think there security is that good, we have some great talent here at xda and i'm sure it will be hacked rather easily. We already have root (see here) access its just a matter of time and the right people.
This is a company that has given the middle finger to every customer that has supported them, and i think its will be SWEET SWEET justice to have it hacked.
I have started a bounty thread in the Boxee forums so please do consider this. Hopefully i can get some interest and some donations for you guys.
UPDATE August 27th 2013
Thanks to the Samsung/Boxee buyout we are left with a broken product. I happy Sammy purchased them as it had so much potential and i'm sure we will see great future products. But for the Boxee Box its just junk now, but thanks to the hard work of "quarnster" he was able to get XBMC booting on Boxee. I ask if any dev's are willing to lend him a hand with the port and hardware acceleration.
Source
would love to this boxee box running custom content.....its been rooted in the past and being rooted again isn't going to magically generate interest.....i agree that it can be hacked and should have been by now i just don't think enough ppl have them...
I think all we really need is talented dev, because there s demand for it. I would for sure donate to prolong its life.
bump
Everybody go bump bump bump.
Aside from the (sometimes) lacking ability to shut down via remote, I have had no real issues with my Boxee Boxes (I have two).
I'm running the most current (non-beta) firmware, and use them to stream netflix as well as various media on my NAS. Although I'd love for this to be fully opened up, I see no real issues with the current performance. Don't get me wrong - I am all for unlocking and going fully custom, but these devices are working for me just fine.
I started off with XBMC (almost a decade ago) when I soldered my Xecuter 2.3b lite chip into my original xbox.
After that, I went a little crazy and equipped each room of my house with one (some chipped, some softmodded). But then HD took over, and the processors just couldn't hang... So I retired my clan of Microsoft hardware and picked up my first Boxee Box. It was fully functional right out of the box, and yes it has had its share of issues, but they seemed to always get worked out. I quickly started to prefer the Boxee layout to XBMC and still feel that way. Not to mention the RF/QWERTY remote (although back-lighting would have been nice). - Rii Mini works just fine for that.
Enough of the history rambling... I'd support this - especially since Boxee has turned its back from the Box, and moved on to a lesser device. I have no intention on picking up a BoxeeTV - the IR remote is the FIRST (of many) issues I have with it.
Just a heads up, there is already some movement on this front with Boxee+.
forums.boxee.tv/showthread.php?t=64648
boxeeplus.com
actually they are not violating the gpl
you can customize the "look" of boxee box - but cannot remake the firmware since it's signed....
if anyone is reading this - i believe the best method of approach is to access it's boot console (i believe it uses redboot) - method i have tried is direct ethernet connection and the boxee box is at 192.168.0.1 port 9000 - usually you access the console by hitting ctrl+c - i haven't tried ctrl+a though (i found ctrl+a within an iso file - for minicom) - the ctrl+alt+del seems to "freeze" or pause the boxee box.... and no i cannot check at the moment... lol
there is a 2-second window only for the redboot however, i can't find it
good luck
oh, and curious as this is:
from buildinfo:
build by root#bamboo-agent-13-ota on Fri, 02 Mar 2012 05:36:33 +0000
boxee 1.5.1.23735
ce_plumb dd2cb6584bd70454ebc4451d414f1ed77bb42b2f
and preflight.sh:
#!/bin/sh
# this is preflight.
# dump kernel
logger -t bxrec.preflight -s -p 0 "dd'ing kernel"
dd conv=sync if=/dev/spectra of=/tmp/kernel.img bs=256k skip=20 count=12
# hash it
logger -t bxrec.preflight -s -p 0 "hashing kernel"
kmdfive=$(md5sum /tmp/kernel.img | cut -f 1 -d " ")
logger -t bxrec.preflight -s -p 0 "kernel md5 hash is $kmdfive"
# if version 1 kernel, copy kernel.1 to kernel.img
# if version 2 kernel (hatung), copy kernel.2 to kernel.img
# else unknown kernel, use dumped version and note hash in mr. wong for support pickup
case $kmdfive in
"9f2ed11d55a3750019d019470212bf36") cp /tmp/upgrade_iso/kernel.1 /tmp/kernel.img;;
"a81c639b7c21d98fdde0b55d4e5356c1") logger -t bxrec.preflight -s -p 0 "doing nothing, this IS kernel.1";;
"d2ed7f9be38c0e0590a5680febd27502") cp /tmp/upgrade_iso/kernel.2 /tmp/kernel.img;;
"68777d9a8fede193ed434cf9a9cd3495") logger -t bxrec.preflight -s -p 0 "doing nothing, this IS kernel.2";;
*) echo $kmdfive >/data/.persistent/.mrwong/boxee.khash;;
esac
logger -t bxrec.preflight -s -p 0 "sync'ing"
sync;sync;sync
logger -t bxrec.preflight -s -p 0 "and we're outta here"
return
anyway... without accessing the box itself before it loads it's internal normal.img (ext2) and dlink_boxee_runtime.img (squashfs) - you are sol
the boxeeplus community is at least "hacking" - so does this mean boxee company is finally getting hacker interest?
Boxee won't open source the box supposedly due to concerns from content providers. 1st off they are closed-sourcing a product which is based off of the hard work of the open sourced XBMC community. What really kind of sucks is that people who invested in the box are pretty much being left out in the cold. I invested in the box because I thought Boxee had the potential to be at the forefront of innovation for media boxes (which in some regards it was). Boxee needs to take a page from Google's playbook. Open source the box and benefit. Or, stagnate innovation and die....its your move Boxee. The set-top media box market will only continue to get more and more crowded. Why not have an army of innovators who will make your current and future products better.
I absolutely agree with you. Desperate situation.
Coming from a fellow boxee user, it would be awesome to get this opened up. I'm tired of the audio drop outs. If it were hacked that issue could be eliminated and being able to really stretch boxee's legs to show the real power, would be awesome.

Replacement for craptacular adb shell

Hey everybody. I'm on windows and adb's shell just keeps on pissing me off. Mainly, how pressing tab actually inserts a tab rather than tab completion . I'd really like to do something about this.
Here's a list of the things i've tried to get rid of this super-annoying behaviour.
-rolled a cross-compiler toolchain (glibc-based) for my phone (HTC Magic) and cross-compiled bash (in a vm). I actually made this work even with dynamic linking, and can start bash from within an instance of "adb shell". Unfortunatly, this doesn't change the tab behaviour, which appears to be baked into adb rather than a problem with the default shell.
-I thought i'd try something fancy by doing...
adb shell mkfifo /fifo
adb shell "bash > /fifo < fifo 2> /fifo"
adb forward tcp:9001 dev:/fifo
and starting putty in raw mode, connecting to localhost:9001. Unfortunatly, all is what happens is i get an echo of what i type and if i press enter a few times putty spits out ": command not found". I can't see why this shouldn't work but I suspect i'm Doing It Wrong.
-Searching the forum revealed puttycyg, which is actually really handy but unfortunatly suffers from the same tab completion problem as using adb shell in cmd.exe
How do you guys do it? Am I onto something with the adb forward thing? If so, how do I make that work?
I have also compiled bash but I have tab completion though adb. I do however use Linux as my desktop OS and am connecting through Konsole. So you could just try to install KDE for Windows. I know that I've looked for good terminal applications for windows but found none. They're all **** compared to what's available on nix.
This is my simple bashrc:
Also it's full of aliases for busybox. By replacing built in ls and similar you even get nice colors But still no line editing for some reason. I figure I might just need a proper libreadline or similar for that.
If anybody's interested in how I got this working, i cross-compiled socat for the droid. Required configuring with --disable-termios and manually editing config.h to set the #define for CRDLY_SHIFT and a couple other *_SHIFT macros to some values i found in Config/linux-2.6.24.h (or something similar, posting from memory here)
Once socat is on the phone, I made it pipe data from a bash shell to a unix socket it is listening on:
socat unix-listen:/dev/adbsock,fork exec:"/bin/bash -li",pty,stderr,setsid,sigint
(command shamelessly copied from http://superuser.com/questions/12379...ld-propagation ).
Then, on the pc i run adb forward tcp:9001 localfilesystem:/tmp/adbsock
And if i connect putty to localhost:9001 in raw mode (disabling line editing and local echo) I get a fairly complete bash shell that you can not only use tab completion on but even vi works
Again, posting from memory here so if anyone wants to try this and is running into problems i can dig it up from my ~/.bash_history.
Any chance you can post your socat port?
The tab completion thing is driving me insane! I don't understand why the ADB shell in *nix is proper, but the windows version sucks so much hairy donkey balls.
I've been using Ansicon so I at least get color in my shell, but ADB still mangles tab, and I don't even want to talk about what it does to Nano. Ugh!
Oh wow, this takes me back. I don't even have my android phone anymore (cancelled the contract due to terribly crap network service, and (unfortunately) the iPhone was the best phone in its price range from my new carrier).
So I don't have any files to give to you...
But, from what I can remember, to make it work, I built a cross binutils and gcc loosely following the instructions from the LFS book, compiled glibc + a few other libs socat wanted with it, and then socat itself, and then bash. I then copied the cross libc.so.6, ld-linux.so.2 (or whatever it's called on ARM) and all the libs socat wanted to /lib on the phone, and the socat and bash binaries to /bin. It's convenient that android keeps its important stuff in /system because it means you can avoid screwing around with your toolchain to make it look for a dynamic linker in some location other than /lib.
From memory socat was the only thing I had problems compiling, and thankfully I've already posted roughly what i had to do to it to make it build (because there's no way I remember lol)
Sorry I can't be more helpful!

[WebTop] rDesktop 1.7.0 for WebTop

rDesktop 1.7.0
What is rDesktop?
rDesktop is a terminal services client for Linux - it will allow you to connect to a terminal services session of any Windows machine with Remote Desktop enabled.
This is a newer release of rDesktop than is available in the Debian repos.
PrerequisitesAccess to a terminal from WebTop
A small amount of Linux shell knowledge
UsageCopy the attached binary to somewhere on your device (I suggest /usr/bin, or somewhere else on your PATH)
Open a terminal
Enter the following:
Code:
rdesktop -g workarea -a16 -z -xm -P -0 <server>
You can use the following switches:-g workarea - sets the window size to the desktop size
-f - makes the session fullscreen (ie. on top of all current chrome)
-a16 - sets 16bit colour depth (valid values are 8, 15, 16 and 24)
-z - enables compression (leave this enabled!)
-xm - sets the 'experience' level to 'modem' (valid values are modem, broadband or LAN)
-0 - causes you to be logged in to the console session (same as the /admin switch for mstsc)
I've found this to be invaluable for getting onto my desktop and servers at work. Now to see if I can get rid of that bloody awful Avant Window Manager and just use FluxBox
I don't have a lapdock, but its compiled for arm/tegra?
Sent from my Motorola Atrix 4G on the network with the most backhaul, whatever that is. This post might have errors as I hate touchscreen keyboards.
I know rdesktop is only terminal based, but is there a sort of add-on to it that allows for a graphical representation of the screen, much like rdp clients in the Android market? that would be far more useful for my purposes, although this is a great step in that direction.
An easy solution is to simply create a shell script and then specify the server.
EG: ./rdp.sh <server>
Much prettier than having to type all the strings.
grdesktop is a graphical front end. It works on hacked webtops such as webtop2sd or the debian chroot mod.
Thanks! you just saved me $1000!
If I couldn't run some programs like matlab and mathematica, I would have had to but a laptop for classes.
Sent from my MB860 using XDA Premium App
Yes, it was compiled on the Atrix itself (that took a while...).
I use the chroot method on my atrix (I like keeping things seperate!), so this is probably only useful for those of us based on Debian (don't know what version is currently in the ubuntu repos)
The grdesktop package from the debian repos will work just fine with this binary.
I might even start a repo just for webtop... anyone think that would be useful?
That would be very useful, actually. Although I'm terrible at using Unix OSes and terminals in general haha.
this is awesome..almost as good as citrix xendesktop...who needs that though when this works so very well...one question did you get your sound working on the lapdock? when i try to enable sound via -r sound:local it says driver not found or smething like that..basically no matter what settings i use even via the gui tools or the cli..i get the same error and no sound....is there any way to get the sound to work?
If you're using the chroot method, it may be that you need to add some packages.
I haven't tried it myself, but I remember seeing a thread somewhere related to getting RythmBox working - this involved sorting out some missing audio-related stuff.
How is this better than the free 2X client for Android?

[TOOL/SCRIPT/RC2]FreeXplain-r / freex-BETTER than free & Why "free" RAM ISN'T WASTED!

See Post #2 for Installation/Requirements/Download.
Now before getting into my new tool, I'd like to explain why I felt the need to even bother making it.
First of all, if you've ever typed "free" in terminal, you'd see that it shows almost no free ram while any app you check with would show lots of free ram.
This what it looks like.
Now if you're like me, you'd probably do a google search to figure out what the hell it means lol
Because I had over 200 mb "free" when I took that screen shot but that shows only 40mb free and it would be nice if it actually made some sense
Some of you familiar with my V6 SuperCharger may have read this post of mine before: "Why 'Free RAM' Is NOT Wasted RAM!"
In there, I post some findings from my first hand experience and how it correlates with this article Linux Memory Consumption (which I had found afterwards)
Unfortunately, that page is now down but I had a saved copy in PDF format which I've attached.
In that pdf, it reads "The absolute free memory number is not only the value of MemFree, but the sum of the MemFree, Buffers and Cached fields."
There are many pages that can help you figure it out.
Here is a good one Check Linux Memory Usage Using Free Command
If you need to know exactly what buffers and cache actually are, click the button Another light hearted link is Help! Linux ate my RAM! That page has this table:
The middle yellow bar above is referring to the buffers (91 mb) and cache (764 mb) values below (855 mb total).869 mb is readily available and not just 13 mb!
So if you read that page, you'd know that 855 mb is NOT actually wasted, it's being used by linux to speed things up!
Basically, if you're not using the ram for apps, linux is using it for cache instead.
And when you need that ram for apps, you get to reclaim it instantly
Therefore...
================================================== ====
Ok that's all nice nice but why did I bother making a new tool?
Well, if you look at the first screen shot up top, you'll see that the android/busybox free command does NOT have the cached column.
Therefore, the free command on Android actually tells us a little something...
...but whole lotta nothing!
So, this is what I did about it...
Basically, I:
Added the cached colum
Removed the useless shared column (which is always 0 and is a waste of real estate)
Added the swap -/+ adjustment line (notice that 4076 kB is actually available, not just 16 kB)
Then total everything up nice nice
So here's a direct comparison of the 2 outputs:
I added colour to the free output for easy comparison.
But notice how the -/+ adjustment line is COMPLETELY different.
The free command shows only 44,644 kB as being freely available but in reality, the value is 208,352 kB... not even close! Simply because free doesn't utilize the cached value (which is found /proc/meminfo along with all the other data.)
The math works both downwards... (add same colours to get the bold ie green plus green = green bold)
...and sideways ie. freely available(208352) = free(40104) + buffers(4540) + cached(163708)
Now, not everybody like rainbows so there is an alternate colour scheme...
Of course, while I was at it (it took me nearly a month to perfect it), I added a whole bunch of options...
So let's see it actually explain something with the -x option...
And explaing even more with -r... ( it's kinda like a mini tutorial )
Since it has so many options, it would be a pita to remember all the options so that you can have your favourite output format all the time...
So it will actually remember your last command (but ignores -help and -install).
Just type "freex" and it automagically does what you did last time...
The output is different depending on whether swap is enabled...
I added the -o option from free which omits the -/+ adjustment line and added a -total only option...
Force -wide option when not using kilobytes (and using a smaller font)...
It's nice to see the percentage table along with kB and/or mb tables.
Also note that it automatically uses wide output when using kB.
Click the button for some gratuitous screen grabs...
Root is NOT required to use this script!
However, if not root, to actually install it for easy command line usage requires a little workaround.
Requirements:
1. An app that can run scripts lol
Fully compatible with Terminal Emulator and Script Manager/Script ManagerPro.
Other terminal apps probably won't display colours properly.
In fact Script Manager didn't either until I requested 256 colour support for it and #devwom updated it quite fast
So make sure to UPDATE Script Manager if that's your preference!
Tho it may not display all the colours, Terminal IDE will work too while installing BusyBox for itself.
2. BusyBox (for the grep and printf commands)
If you're rooted, then there is nothing else to do and go to the Install section.
If NOT rooted you have a few options.
a) download the attached busybox.zip and put it in the same place as the freex script. It is not a zip file. It's the busybox binary from Kbox2. I chose it for it's small size.
Just run freex as explained in the install section and freex will install "busybox.zip" for you as long as it's in the same location as the script.
b) install OneBox. It requires Terminal Emulator and automates the process to...
c) Install Kbox2
d) Easiest solution is probably BusyBox Non-Root but also for use with Terminal Emulator.
e) Use Terminal IDE (has its own BusyBox)
FreeXplain-r Installation:
(Pronounced "Free Explainer")
Note: Installation merely simplifies the execution of freex so that you can just type "freex" in your terminal app without having to type "sh /sdcard/freex*" all the time. Of course, if you're using a script runner like Script Manager, it doesn't matter too much since their purpose is to make it easy to run scripts.
1. Save attached script to sdcard. (No need to rename it from freex*.pdf)
2. Then just run the script!
If using Script Manager, no need for me to explain how to use that app.
If using a terminal app, do:No need for the -i switch it should install automatically.
Use the -i switch for forced installation/update/refresh.
If not running as root, it will copy to your terminal app's data folder and give you further instructions.
Notice that it says to run with "sh freex" in the help section?
That's because the $PATH statement hasn't yet been setup in the terminal app's settings.
If running as root, it will automatically install to /system/xbin and it's easy as pie.
Here is a forced refresh/reinstall.
Not running as root at first but the $PATH is setup so it doesn't give the additional help info to set that up.
Then as root, of course it installs to /system/xbin
Another masterpiece from the GoRM!
Even for a simple(but not so simple ) script you can explain this much. This really sums up what an amazing developer you are zep.
Sent from my GT-P7500 using Tapatalk
This is nice!! Explains all! I like rainbows of course..
Noted something.. when running (installing) the script saved to a folder which has a name with spaces, an error occurs (I'm using ScriptManager). "freex" didn't get installed to xbin properly (size=0). No issues otherwise.
i'll try it later, thanks man...
ok I give up... what's GoRM? IMMA N00B and forgot the URL for google.com
I don't bother if it don't work on all devices
Ah the scripter's life long nemesis... the evil space in a folder name.
I shall quash that empty soul in our next duel.
Btw you didn't update Script Manager... the rainbow is all wrong lol
Edit: Added a poll. For fun, I'm gonna view it as an IQ test. heh.
God of RAM Management
Hey zepp just wanted to confirm after running this script this is what my free ram should look like
sent from my DNA
Yup.. noted that. I meant rainbows in terminal emulator..
"watch freex" command lets me to enjoy the rainbows every 2 seconds :P
Ok any idea as to why I'm getting this
sent from my DNA

Stopping UT from killing apps

I finally found a way to not have my apps killed all the time!
I am using a device with 2 GB of ram (Nexus 7 2013) and I experienced frequently that apps are killed. Quite often it gets so bad that it's impossible to have both dekko and the browser open at the same time. It always felt like it was an out of memory situation. Today I stumbled over this remark of ShadowEO on his github page to configure the lowmemorykiller. In my experience, I need two other tweaks in addition. So in total I did the following three things:
Increase the swappiness
Checking with
Code:
dstat --top-oom --top-mem --vmstat -s 30
and
Code:
tail -f /var/log/syslog | grep sigkill
I see which app the out-of-memory killer (oom) considers for killing and I see when that app is being killed. Also, I see that even though there is swap space available, it isn't being used. In order to make the system actually use the available swap space I increased the swappiness from 0 to 70:
Code:
cat /proc/sys/vm/swappiness
0
echo 70 | sudo tee /proc/sys/vm/swappiness
Increase swapspace
Now the swap is being used, but it doesn't seem to make a whole lot of a difference with the killing, so I increase the size of the swapspace.
Out of the box, if I remember it correctly, I only had 32 MB of swap:
Code:
swapon -s
The swap file being /userdata/SWAP.img. I increased it to 2 GB following these tips:
Code:
sudo swapoff /userdata/SWAP.img
sudo rm /userdata/SWAP.img
sudo fallocate -l 2g /userdata/SWAP.img
sudo mkswap /userdata/SWAP.img
sudo chmod 0600 /userdata/SWAP.img
sudo swapon /userdata/SWAP.img
Configure the lowmemorykiller
Now a little more of the swap is being used, but it still doesn't stop the killing. Seems on Android there is something called lowmemorykiller, which on my Nexus 7 had this original configuration:
Code:
cat /sys/module/lowmemorykiller/parameters/minfree
1536,2048,4096,16384
Where 16384 pages (4k each) corresponds to 16384*4/1024 = 64 MB. So, seems that once the free memory drops below 64MB, the killing starts. Inspired by ShadowEO's tip, I set it:
Code:
echo "1536,2048,4096,5120,10240,15360" | sudo tee /sys/module/lowmemorykiller/parameters/minfree
which corresponds to 20 instead of 64 MB. (I also set the last two levels to 40 and 60 MB - don't know whether that is needed.)
With all three things in place, I can now have the browser, dekko, podbird, filemanager, unav, uradio playing music, terminal and gedit and libreoffice writer running from libertine all open at the same time. I'm almost running out of apps that I care to run in addition
At times there was some stuttering, but of course nobody forces you to have so many apps open at the same time. However, now you can choose what you want to close to relieve the memory pressure Swappiness and lowmemorykiller I need to repeat after every reboot. Need to find a solution to persist this.
I'd be curious to hear whether/how this works for other people!
I'm running this now for a couple of weeks without noticing any obvious problems caused by these settings (actually I'm running swappiness=100)
Just now I figured out another aspect: The browser does a very aggressive unloading of inactive tabs. The bugreport here suggests that this unloading is related to a percentage of free memory.
By doing
My expectation is that it should keep 300 megabytes of free memory (which of course is a waste). It seems to actually keep some 600+ mb free and after testing this for all of 5 minutes I can successfully listen to a youtube video playing in an inactive tab, while writing this post in another tab!
Ok that was easy. Put the two lines
into /etc/rc.local. I don't dare to put the min_free_kbytes setting in there just yet.
There is this OpenStore (I can't add link, google it 'OpenStore ubuntu') which have very nice app (UT Tweek tool) for tweeking the UT. It allows to prevent killing apps and it can be set per app, so no more annoying termination of i.e. terminal scripts when you switch the apps or youtube music kill. By this method you need no reentering of the code after every reboot and you have quite granular control over killing or not killing automatically the apps.
Yeah, I know it, and I use it, but to the best of my understanding that applies to a different kind of killing. I think the suspension itself doesn't actually terminate any process. It just freezes it. So with the UT tweak tool the process continues executing even when you switch to another app, which is indeed a good first step (yt playing in the background).
But there is another form of killing going on where UT seems to very aggressively try to keep some memory free and actually terminates a process. What is left is an a bit blurry picture of the last state of the app, that is shown in the task switcher and when you switch back. Then the app is restarted and its the job of the app to restore the state it was in previously.
My experience was that when I write an email and I'm switching to the browser to check something and back, the email app had been killed. It takes a few seconds to load and it doesnt perfectly restore the state (half written email, cursor position, selected text). Same the other way around. I'm typing a forum post but want to check that link I had in that email. Browser was killed, restarts, which takes a couple of seconds, and the halftyped posting is gone.
true, with the terminal I didnt have the problem of actually be killed so often. The script you had running has just been suspended. You can revive it by typing fg, or use uttt as you suggested.
So, are you saying, with the uttt settings, you have no trouble using email and browser at the same time? Switching back and forth? Switching btw tabs in the browser? How much RAM does your device have?
Hi!
Sadly in the aquaris E5 ubuntu edition(That I own) the SWAP is in a partition, so isn't so easy to expand it. I create a second swap in a file but isn't persistent after a reboot.
Interesting. Where did you create the swapfile? What exactly isnt persisted? Is it just not used anymore or does the file go away? How did you attempt to persist it? In rc.local?

Categories

Resources