Page 3 of 4
Re: Wiki update
Posted: Wed Jan 02, 2019 5:30 pm
by raboof
lilith wrote:If I don't have any entry in fstab it is still used as it is in the kernel by default? Can this be checked somehow?
You can check with "mount | grep shm" to see if /dev/shm is available. I'm not sure if JACK tells you whether it is actually using it, but if it's available then it should.
For me "ls -alFh /dev/shm" shows a /dev/shm/jack-1000 "directory", that could be a good hint
.
Re: Wiki update
Posted: Wed Jan 02, 2019 5:48 pm
by khz
With the shm entry in the fstab he shows me this.
Code: Select all
~ $ ls -alFh /dev/shm
insgesamt 79M
drwxrwxrwt 2 root root 240 2. Jan 20:24 ./
drwxr-xr-x 18 root root 4,2K 2. Jan 10:40 ../
-rw-r--r-- 1 daw daw 79M 2. Jan 19:06 jack-1000-0
-rw-r--r-- 1 daw daw 1,2K 2. Jan 19:06 jack-1000-1
srwxr-xr-x 1 daw daw 0 2. Jan 19:06 jack_default_1000_0=
-rwxr-xr-x 1 daw daw 12 2. Jan 19:06 jack_sem.1000_default_a2j*
-rwxr-xr-x 1 daw daw 12 2. Jan 19:06 jack_sem.1000_default_dbusapi*
-rwxr-xr-x 1 daw daw 12 2. Jan 19:06 jack_sem.1000_default_freewheel*
-rwxr-xr-x 1 daw daw 12 2. Jan 19:06 jack_sem.1000_default_qjackctl*
-rwxr-xr-x 1 daw daw 12 2. Jan 19:07 jack_sem.1000_default_REAPER*
-rwxr-xr-x 1 daw daw 12 2. Jan 19:06 jack_sem.1000_default_system*
-rw-rw-rw- 1 daw daw 70K 2. Jan 10:41 jack-shm-registry
<EDIT>
Code: Select all
~ $ du -h /dev/shm/*
79M /dev/shm/jack-1000-0
4,0K /dev/shm/jack-1000-1
0 /dev/shm/jack_default_1000_0
4,0K /dev/shm/jack_sem.1000_default_a2j
4,0K /dev/shm/jack_sem.1000_default_dbusapi
4,0K /dev/shm/jack_sem.1000_default_freewheel
4,0K /dev/shm/jack_sem.1000_default_qjackctl
4,0K /dev/shm/jack_sem.1000_default_REAPER
4,0K /dev/shm/jack_sem.1000_default_system
72K /dev/shm/jack-shm-registry
Code: Select all
~ $ df -h
shm 7,4G 155M 7,2G 3% /dev/shm
tmpfs 7,4G 12K 7,4G 1% /tmp
https://www.cyberciti.biz/tips/what-is- ... usage.html
https://superuser.com/a/1030777</EDIT>
What without the entry?
Re: Wiki update
Posted: Wed Jan 02, 2019 7:48 pm
by Jack Winter
I'm not sure I understand, AFAIK this has nothing to do with gcc, swap space or something like that.
JACK needs to communicate between the client libs and the server and does this via iirc named pipes. This takes at the most a few megabytes. Probably there are some other apps using shm too, but I've never seen anything take up much space in /dev/shm. Sending it out over the network and back would probably be a bad idea..
Re: Wiki update
Posted: Wed Jan 02, 2019 7:50 pm
by Jack Winter
lilith wrote:I'll take it out completely. As stated somewhere else this is also better for SSD on the long-term.
I don't understand how it could have anything to do with your SSD.
Re: Wiki update
Posted: Wed Jan 02, 2019 7:51 pm
by Jack Winter
lilith wrote:If I don't have any entry in fstab it is still used as it is in the kernel by default? Can this be checked somehow?
Yes, have a look with "ls /dev/shm"
Re: Wiki update
Posted: Wed Jan 02, 2019 7:56 pm
by Jack Winter
khz wrote:With the shm entry in the fstab he shows me this.
Code: Select all
~ $ ls -alFh /dev/shm
insgesamt 79M
What without the entry?[/quote]
Wow, that's quite a lot more than I see, still only 80MB..
Try without the fstab entry and I don't think it will change anything at all.
Re: Wiki update
Posted: Wed Jan 02, 2019 8:19 pm
by raboof
Jack Winter wrote:lilith wrote:I'll take it out completely. As stated somewhere else this is also better for SSD on the long-term.
I don't understand how it could have anything to do with your SSD.
I suspect what lilith means is if you use RAM-backed storage instead of Disk-backed storage, that might reduce wear and tear on your SSD. So using tmpfs/SHM for transient data instead of your SSD is probably a good idea.
Re: Wiki update
Posted: Wed Jan 02, 2019 8:21 pm
by khz
Jack Winter wrote:Try without the fstab entry and I don't think it will change anything at all.
Without fstab entry I get the same.
Code: Select all
~ $ ls -alFh /dev/shm
insgesamt 79M
drwxrwxrwt 2 root root 240 2. Jan 21:07 ./
drwxr-xr-x 18 root root 4,2K 2. Jan 21:03 ../
-rw-r--r-- 1 daw daw 79M 2. Jan 21:03 jack-1000-0
-rw-r--r-- 1 daw daw 1,2K 2. Jan 21:03 jack-1000-1
srwxr-xr-x 1 daw daw 0 2. Jan 21:03 jack_default_1000_0=
-rwxr-xr-x 1 daw daw 12 2. Jan 21:03 jack_sem.1000_default_a2j*
-rwxr-xr-x 1 daw daw 12 2. Jan 21:07 jack_sem.1000_default_ardour*
-rwxr-xr-x 1 daw daw 12 2. Jan 21:03 jack_sem.1000_default_dbusapi*
-rwxr-xr-x 1 daw daw 12 2. Jan 21:03 jack_sem.1000_default_freewheel*
-rwxr-xr-x 1 daw daw 12 2. Jan 21:03 jack_sem.1000_default_qjackctl*
-rwxr-xr-x 1 daw daw 12 2. Jan 21:03 jack_sem.1000_default_system*
-rw-rw-rw- 1 daw daw 70K 2. Jan 21:03 jack-shm-registry
Code: Select all
~ $ du -h /dev/shm/*
79M /dev/shm/jack-1000-0
4,0K /dev/shm/jack-1000-1
0 /dev/shm/jack_default_1000_0
4,0K /dev/shm/jack_sem.1000_default_a2j
4,0K /dev/shm/jack_sem.1000_default_ardour
4,0K /dev/shm/jack_sem.1000_default_dbusapi
4,0K /dev/shm/jack_sem.1000_default_freewheel
4,0K /dev/shm/jack_sem.1000_default_qjackctl
4,0K /dev/shm/jack_sem.1000_default_system
72K /dev/shm/jack-shm-registry
Code: Select all
~ $ df -h
shm 5,9G 102M 5,8G 2% /dev/shm
What I have written is not necessary, yes.
Re: Wiki update
Posted: Wed Jan 02, 2019 8:27 pm
by khz
raboof wrote:if you use RAM-backed storage instead of Disk-backed storage, that might reduce wear and tear on your SSD.
Yes. shm == RAM
Code: Select all
┌───────────┬──────────────┬────────────────┐
│ /dev/shm │ always tmpfs │ Linux specific │
├───────────┼──────────────┼────────────────┤
│ /tmp │ can be tmpfs │ FHS 1.0 │
├───────────┼──────────────┼────────────────┤
│ /var/tmp │ never tmpfs │ FHS 1.0 │
└───────────┴──────────────┴────────────────┘
?
Re: Wiki update
Posted: Wed Jan 02, 2019 8:36 pm
by Jack Winter
raboof wrote:Jack Winter wrote:I suspect what lilith means is if you use RAM-backed storage instead of Disk-backed storage, that might reduce wear and tear on your SSD. So using tmpfs/SHM for transient data instead of your SSD is probably a good idea.
Yes, but /dev/shm is a tmpfs even if you haven't mounted it so in fstab, it's automatic nowdays, and as all tmpfs by default it gets a maximum of half your ram. I just saw that steam also uses a few MB on it.
Or maybe it's systemd that mounts it automatically, and I've been misled by my system
I haven't looked at this for years, back then I concluded that it was not an issue that really concerned me. Possibly after so many discussions on the subject I ought to read the kernel documentation again to refresh my memory
Still the beauty of using Linux is that we can all do as we want, and I suppose that there is no right way to do anything. Just trying to bring over the POV that this probably is not an issue for most people and that explicitly mounting it doesn't really offer any advantage than changing the max size available. I really don't know many apps that use it, and personally I've never seen it taking up much ram at all. YMMV and all that.
Re: Wiki update
Posted: Wed Jan 02, 2019 8:55 pm
by Jack Winter
khz wrote:Code: Select all
~ $ df -h
shm 5,9G 102M 5,8G 2% /dev/shm
In fact you made it bigger with the entry in fstab
Re: Wiki update
Posted: Wed Jan 02, 2019 9:33 pm
by khz
Jack processes in shm (RAM) is great.
Would be worth a mention in the WIKI,
shm is included by default.
Re: Wiki update
Posted: Fri Mar 15, 2019 3:48 pm
by khz
@
https://wiki.linuxaudio.org/wiki/start? ... irqbalance
@realTimeConfigQuickScan script
add the entry "check if irqbalance is switched off"?
Disabling the irqbalance daemon >>
https://access.redhat.com/documentation ... ss_binding
Re: Wiki update
Posted: Fri Oct 04, 2019 1:20 pm
by khz
When I read the
Reaper thread I asked myself why the
wiki still mentions "hpet".
"tsc" is "only" mentioned in the FA
Q: What about timers and timing ?.
Is there a reason for this?
So far I have always entered the hpet entries from the wiki.
But both
Code: Select all
cat /sys/devices/system/clocksource/clocksource0/current_clocksource
and
show me "tsc"
https://en.wikipedia.org/wiki/Time_Stamp_Counter.
Re: Wiki update
Posted: Sun Jan 12, 2020 8:55 am
by khz
If I understood it correctly we could not edit the wiki or anything else, there was some problem with the wiki.
Is the problem now solved by (*1)?
(*1)
https://lists.linuxaudio.org/archives/l ... 02771.html
[LAA] A new year, a fresh start
Dear all,
First of all a happy 2020 (even though it's way past Epiphany)! So
what's the fresh start about? Well, registration of the linuxaudio.org
domain has been handed over to yours truly so if you need any DNS
changes regarding any linuxaudio.org subdomains then I can take care of
that now directly. Thanks Daniel James of 64 Studio for taking care of
the domain for so long!
Other part of the fresh start is that I migrated the web server to a new
cloud region which is faster and more performant. I just changed all the
DNS records so if you spot anything weird that just worked before then
please let me know! The mail server will follow this weekend so if mail
seems a bit flaky the coming weekend then you know what might cause it.
Best regards,
Jeremy