GentooPlayer

[luisma]

Hi Antonello, I'm sorry as I didn't explained properly

 

I have a NUC with 2 interfaces a wireless and a wired interface (hence multihomed) wireless has 192.168.95.225 with default gateway 192.168.95.1 and connected to my network.

 

Wired interface is directly connected to an endpoint, NUC IP 192.168.96.225, endpoint ip 192.168.96.227.

 

NUC runs HQPe and endpoint runs NAA on GP as well.

 

The NUC can't find the NAA if kernel version is higher than 5.2.21

 

The issue happens indeed because of the 2 network interfaces on the NUC, I'm sure with just one network and one interface everything will work properly, I have tried this many times with GP, AL and the results are consistent.

 

Why can't I create a single network? Because I would need to open the walls and run a cable from the room upstairs to the living room and it is concrete walls and there is no path, or direct way to it. why I don't use powerline adapters? Because I tried and it doesn't work properly. Why I am using wireless with no issues? Because I happen to be a partner reseller of  ruckus technologies possibly the best commercial access points in the planet and it works flawless in my house.

 

I can keep using the 4.19 kernel but at some point will be discarded and it will break my system.

 

[stefano_mbp]

@luisma maybe a silly question: is your netmask 255.255.0.0 over all your network?

[luisma]

nope, just a /24 on each segment

[stefano_mbp]

I could mistake but with /24 your two subsets... subnets.... are segregated, they do not communicate with each other

[stefano_mbp]

More sure now .... I am not used to /xx notation ...

 

/24 means all the addresses inside a single subnet, that’s to say

255 addresses in 192.168.95

255 addresses in 192.168.96

 

but the two subnets do not communicate with each other, you need to define a different netmask and it should be 255.255.0.0, in this way your two segment will communicate each other

[luisma]

That is correct and that's the idea. A host (NUC in this case) which has a leg on each segment and should be able to communicate transparently, why does it works with a 3 month old kernel and stopped is what I would like to know. I am not very familiar with the way HQPe interacts with multicast, Miska a few times have explained but ultimately he is been always against multihoming, and it is understandable as it complicates things, I would try to avoid it if I could.

 

Now running this

GentooPlayer-64 ~ # uname -r
4.19.94-rt-rt38-x86_64-GP-RT-ST
 

Discovery works just fine

2020/04/10 16:55:01 NAA output network Audio IPv6 support enabled
  2020/04/10 16:55:01 Automatic rate family enabled
  2020/04/10 16:55:01 NAA output discovery from 0.0.0.0
  2020/04/10 16:55:02 NAA output discovery from ::
& 2020/04/10 16:55:02 NAA output discovered network audio: name='GentooPlayer-64' version='Signalyst Network Audio Daemon 4.1.1'  @[fe80::1fb6:d09:6a12:efa3%eth0]:43210
& 2020/04/10 16:55:02 NAA output discovered network audio: name='GentooPlayer-64' version='Signalyst Network Audio Daemon 4.1.1'  @[fe80::1fb6:d09:6a12:efa3%eth0]:43210
  2020/04/10 16:55:03 NAA output connect to [fe80::1fb6:d09:6a12:efa3%eth0]:43210 [ipv6]
  2020/04/10 16:55:03 NAA output network endpoint: Intel HDMI/DP LPE Audio: Intel HDMI/DP LPE Audi (hw:CARD=Audio,DEV=0)
  2020/04/10 16:55:03 NAA output network endpoint: Intel HDMI/DP LPE Audio: Intel HDMI/DP LPE Audi (hw:CARD=Audio,DEV=1)
  2020/04/10 16:55:03 NAA output network endpoint: Pre Box S2 Digital: USB Audio (hw:CARD=Digital,DEV=0)
  2020/04/10 16:55:03 NAA output discovered 1 Network Audio Adapters
+ 2020/04/10 16:55:03 NAA output connect to [fe80::1fb6:d09:6a12:efa3%eth0]:43210 [ipv6]

 

Then I do an kerneluspw switching to 5.4.13-200208-RT from 4.19.94-200302-RT

Then selectkernel, 14 which is ST and supports wireless

 

GentooPlayer-64 ~ # uname -r
5.4.13-rt-rt7-x86_64-GP-RT-ST

 

Now the NAA is not visible and not discovered

2020/04/10 17:00:17 Enable UPnP Renderer functionality for interface: auto
  2020/04/10 17:00:17 Adding interface 'eth0' for UPnP Renderer
  2020/04/10 17:00:17 Adding interface 'wlan0' for UPnP Renderer
% 2020/04/10 17:00:17 ...configuration reloaded
  2020/04/10 17:00:17 NAA output network Audio IPv6 support enabled
  2020/04/10 17:00:17 NAA output discovery from 0.0.0.0
  2020/04/10 17:00:17 asoundlib version: 1.2.1.2
  2020/04/10 17:00:17  network Audio IPv6 support enabled
  2020/04/10 17:00:17  discovery from 0.0.0.0
  2020/04/10 17:00:18 NAA output discovery from ::
  2020/04/10 17:00:18  discovery from ::
  2020/04/10 17:00:19 NAA output discovered 0 Network Audio Adapters

Just let me try something else, the renderer functionality is set to "auto", I can

"nano /etc/hqplayer/hqplayerd.xml"

and change "<upnp interface="auto">" to "<upnp interface="eth0">"

save and restart

 

Even pre setting the eth0 it won't discover the NAA

2020/04/10 17:07:27 Enable UPnP Renderer functionality for interface: eth0
% 2020/04/10 17:07:27 ...configuration reloaded
  2020/04/10 17:07:28 NAA output network Audio IPv6 support enabled
  2020/04/10 17:07:28 NAA output discovery from 0.0.0.0
  2020/04/10 17:07:29 NAA output discovery from ::
  2020/04/10 17:07:30 NAA output discovered 0 Network Audio Adapters
  2020/04/10 17:07:30 asoundlib version: 1.2.1.2
  2020/04/10 17:07:30  network Audio IPv6 support enabled
  2020/04/10 17:07:30  discovery from 0.0.0.0
  2020/04/10 17:07:31 NAA output network Audio IPv6 support enabled
  2020/04/10 17:07:31 NAA output discovery from 0.0.0.0
  2020/04/10 17:07:31  discovery from ::
  2020/04/10 17:07:32 NAA output discovery from ::
  2020/04/10 17:07:33 NAA output discovered 0 Network Audio Adapters

 

Really puzzling

 

[luisma]

16 minutes ago, stefano_mbp said:

but the two subnets do not communicate with each other, you need to define a different netmask and it should be 255.255.0.0, in this way your two segment will communicate each other

Thank you, I know you are trying to help, but I think that is not correct, if I define that netmask 255.255.0.0 everything will flow through a single adapter, most likely the one with the default route associated in this case the wireless, the second adapter won't be used.

Another interesting thing is that IPv6 is being used based on the logs for multicast discovery not IPv4 unicast, which is great as there is no router on that segment. Switching to unicast unchecking the IPv6 option in HQPe's interface won't work either as it is expected as there is no router with arp resolution to direct HQPe to the NAA now you see IPv6 disabled and no IPv6 addresses being used

 

2020/04/10 17:17:02 Enable UPnP Renderer functionality for interface: auto
  2020/04/10 17:17:02 Adding interface 'eth0' for UPnP Renderer
  2020/04/10 17:17:02 Adding interface 'wlan0' for UPnP Renderer
% 2020/04/10 17:17:02 ...configuration reloaded
  2020/04/10 17:17:02 NAA output network Audio IPv6 support disabled
  2020/04/10 17:17:02 NAA output discovery from 0.0.0.0
  2020/04/10 17:17:03 NAA output discovered 0 Network Audio Adapters
  2020/04/10 17:17:03 asoundlib version: 1.2.1.2

The ethernet has multicast enabled

2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 94:c6:91:1b:02:5f brd ff:ff:ff:ff:ff:ff
    inet 192.168.96.225/24 brd 192.168.96.255 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::96c6:91ff:fe1b:25f/64 scope link
       valid_lft forever preferred_lft forever

 

But something in the newer kernel is OFF and I suspect it is related to multicast

 

You guys are devs and know more than me on the kernel details, I used to played with monolithic kernels back in 1995 with Slackware (the first distribution I used when I learned Linux), used to run an ISP at that time, sendmail, freshmeat.org to download and compile the packages etc. but I haven't touch Linux in a while, not for these purposes anyway.

 

[stefano_mbp]

.... network management is sometimes a hard topic ... the nuc having the two legs on different subnets can be misleading, all the network things must know where to send a packet and if the netmask is too narrow the packets are lost.

cannot say why it was working before  but this is what the network management theory says.

 

try it ... we all have plenty of time in these days ....

[luisma]

Or could be something off on the IPv6 implementation as well.

[luisma]

ok let me try it, it goes against everything I have learned but you are correct, also you have been kind trying to help so I owe you that

[stefano_mbp]

Just now, luisma said:

Or could be something off on the IPv6 implementation as well.

Cannot say anything about ipv6, sorry

[luisma]

My existing config

GentooPlayer-64 ~ # cat /etc/conf.d/net
modules_wlan0="!iw !iwconfig !wpa_supplicant"
config_wlan0="192.168.95.225/24"
routes_wlan0="default via 192.168.95.1"
dns_servers_wlan0="1.1.1.1 8.8.8.8"
config_eth0="192.168.96.225/24"

 

The new config

GentooPlayer-64 ~ # cat /etc/conf.d/net
modules_wlan0="!iw !iwconfig !wpa_supplicant"
config_wlan0="192.168.95.225/16"
routes_wlan0="default via 192.168.95.1"
dns_servers_wlan0="1.1.1.1 8.8.8.8"
config_eth0="192.168.96.225/24"

 

Most likely this is going to get broken, rebooting now

 

 

[luisma]

As expected the NUC cannot see the NAA because of the netmask conflict

GentooPlayer-64 ~ # ping 192.168.96.227
PING 192.168.96.227 (192.168.96.227) 56(84) bytes of data.
From 192.168.96.225 icmp_seq=1 Destination Host Unreachable
From 192.168.96.225 icmp_seq=2 Destination Host Unreachable
From 192.168.96.225 icmp_seq=3 Destination Host Unreachable
^C
--- 192.168.96.227 ping statistics ---
4 packets transmitted, 0 received, +3 errors, 100% packet loss, time 88ms
pipe 4

 

and still cannot find the NAA

  2020/04/10 17:30:22 NAA output discovered 0 Network Audio Adapters
  2020/04/10 17:30:23 NAA output network Audio IPv6 support enabled

The solution to this could be bridging and I have a config to try:

nano /etc/conf.d/net

bridge_br0="wlan0 eth0"

# Bridge static config
config_br0="192.168.95.225/24"
routes_br0="default via 192.168.95.1"

bridge_forward_delay_br0=0
bridge_hello_time_br0=1000

 

I would have to change the IP address on the NAA as well and I am not sure you could bridge wlan and eth but let's try that

 

[stefano_mbp]

So it seems this could be a NAA constraint... cannot help you on this, sorry

[luisma]

Guess I have to wait, something going on with the github rep

/opt/.gentooplayer/function/fkernelsw.sh: line 62: cd: too many arguments
--2020-04-10 17:41:37--  https://github.com/antonellocaroli/GP-x86-Kernel/releases/downl/GP-x86-Kernel-RT-5.4.13-200208.tar.xz
Resolving github.com... 140.82.112.4
Connecting to github.com|140.82.112.4|:443... connected.
HTTP request sent, awaiting response... 404 Not Found
2020-04-10 17:41:37 ERROR 404: Not Found.

tar: GP-x86-Kernel-RT-5.4.13-200208.tar.xz: Cannot open: No such file or directory
tar: Error is not recoverable: exiting now

[luisma]

3 minutes ago, stefano_mbp said:

So it seems this could be a NAA constraint... cannot help you on this, sorry

No worries, thank you though.

 

[Aberrant-Decoder]

i was hoping some charitable soul would help me configure my irq settings.

 

Goal is to optimise sq, i have no issues, clicks or pops or anything.  My issue is that even though I think i'm setting the irq for ethernet, usb, to separate cpu affinities, it doesn't look to be respected.

 

So firstly, is IRQ assignment only respected in RealTime kernels?  I've tried both types anyway, but I tend to see ethernet for example used on multiple cpus.

 

Is there a way to 'factory restore' all settings relating to irq?

 

I've tried the web interface, to choose Profile 1, which should separate these activities onto separate cpus, but it doesnt. So then i use the console commands (but it also warns not too if you've set a profile).

 

I could use a hand !.

 

[antonellocaroli]

1 minute ago, Aberrant-Decoder said:

i was hoping some charitable soul would help me configure my irq settings.

 

Goal is to optimise sq, i have no issues, clicks or pops or anything.  My issue is that even though I think i'm setting the irq for ethernet, usb, to separate cpu affinities, it doesn't look to be respected.

 

So firstly, is IRQ assignment only respected in RealTime kernels?  I've tried both types anyway, but I tend to see ethernet for example used on multiple cpus.

 

Is there a way to 'factory restore' all settings relating to irq?

 

I've tried the web interface, to choose Profile 1, which should separate these activities onto separate cpus, but it doesnt. So then i use the console commands (but it also warns not too if you've set a profile).

 

I could use a hand !.

 

What system are you on?

[antonellocaroli]

5 minutes ago, luisma said:

Guess I have to wait, something going on with the github rep

/opt/.gentooplayer/function/fkernelsw.sh: line 62: cd: too many arguments
--2020-04-10 17:41:37--  https://github.com/antonellocaroli/GP-x86-Kernel/releases/downl/GP-x86-Kernel-RT-5.4.13-200208.tar.xz
Resolving github.com... 140.82.112.4
Connecting to github.com|140.82.112.4|:443... connected.
HTTP request sent, awaiting response... 404 Not Found
2020-04-10 17:41:37 ERROR 404: Not Found.

tar: GP-x86-Kernel-RT-5.4.13-200208.tar.xz: Cannot open: No such file or directory
tar: Error is not recoverable: exiting now

[Aberrant-Decoder]

Just now, antonellocaroli said:

What system are you on?

5.4.13-rt-rt7-x86_64-GP-RT-ST

 

and also tried on FP-ST

 

my issue with the other kernels is that the audio card drivers are not installed.  As i prefer to play via toslink, but i have the option of going via usb to mscaler too.

[antonellocaroli]

16 minutes ago, Aberrant-Decoder said:

 

 

I've tried the web interface, to choose Profile 1, which should separate these activities onto separate cpus, but it doesnt. So then i use

 

how can you say it doesn't?

if you check with

watch -n1 "cat /proc/interrupts"

you will see that you have some interrupts on some cpus that increase, there are some if other cpus but you will see that they do not increase, they are the ones before the script starts...

profile 6 is the default

[luisma]

Hi Antonello, did not understand your post.

 

I am running an RT kernel

GentooPlayer-64 ~ # uname -r
5.4.13-rt-rt7-x86_64-GP-RT-ST

 

But I can't switch to the 4.19 RT now

 [1] 5.4.13-200208-RT
 [2] 5.5.0-200208
 [3] 4.19.94-200302-RT

 â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€

 [4] Main Menu
 [0] Exit

 â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€

Choose your operation:

 [0 - 4]:3

/opt/.gentooplayer/function/fkernelsw.sh: line 62: cd: too many arguments
--2020-04-10 18:04:16--  https://github.com/antonellocaroli/GP-x86-Kernel/releases/downl/GP-x86-Kernel-RT-5.4.13-200208.tar.xz
Resolving github.com... 140.82.114.3
Connecting to github.com|140.82.114.3|:443... connected.
HTTP request sent, awaiting response... 404 Not Found
2020-04-10 18:04:16 ERROR 404: Not Found.

tar: GP-x86-Kernel-RT-5.4.13-200208.tar.xz: Cannot open: No such file or directory
tar: Error is not recoverable: exiting now
Generating grub configuration file ...
Found background: /boot/grub/gp2.png
Found linux image: /boot/kernel-genkernel-x86_64-5.1.0-pf7-GP-ST-AU-RAVENNA
Found initrd image: /boot/initramfs-genkernel-x86_64-5.1.0-pf7-GP-ST-AU-RAVENNA
Found linux image: /boot/kernel-genkernel-x86_64-5.0.21-rt-rt16-GP-RT-RAVENNA
Found initrd image: /boot/initramfs-genkernel-x86_64-5.0.21-rt-rt16-GP-RT-RAVENNA
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-ST
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-ST.img
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-MIN-GP1-CORE
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-MIN-GP1-CORE.img
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-MIN-GP1
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-MIN-GP1.img
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-MIN-EUP-CORE
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-MIN-EUP-CORE.img
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-MIN-EUP
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-MIN-EUP.img
done
wait...5min

 

 

 

[Aberrant-Decoder]

Just now, antonellocaroli said:

how can you say it doesn't?

if you check with cat /proc/interrupts you will see that you have some interrupts on some cpus that increase, there are some if other cpus but you will see that they do not increase, they are the ones before the script starts...

profile 6 is the default

ok, thanks for helping me to understand a bit better.

so, i removed my custom irq changes, i ran profile 1, after boot, it seems to be saying only cpu 1 is in use.

 

now why is that, whenever profile 1 is supposed to be splitting the activities over the 4 default cores?

 

 

, 

 

i had saw an earlier post that says Profile 1 will split the activities, but maybe this is not right, maybe profile 1 just sets the system up into a good default state, then i must assign the cpu affinities myself from this point?

 

[antonellocaroli]

5 minutes ago, Aberrant-Decoder said:

ok, thanks for helping me to understand a bit better.

so, i removed my custom irq changes, i ran profile 1, after boot, it seems to be saying only cpu 1 is in use.

 

now why is that, whenever profile 1 is supposed to be splitting the activities over the 4 default cores?

 

 

, 

control it with htop....

and you should also update the scripts with gp-update

[antonellocaroli]

16 minutes ago, luisma said:

Hi Antonello, did not understand your post.

 

I am running an RT kernel

GentooPlayer-64 ~ # uname -r
5.4.13-rt-rt7-x86_64-GP-RT-ST

 

But I can't switch to the 4.19 RT now

 [1] 5.4.13-200208-RT
 [2] 5.5.0-200208
 [3] 4.19.94-200302-RT

 â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€

 [4] Main Menu
 [0] Exit

 â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€â”€

Choose your operation:

 [0 - 4]:3

/opt/.gentooplayer/function/fkernelsw.sh: line 62: cd: too many arguments
--2020-04-10 18:04:16--  https://github.com/antonellocaroli/GP-x86-Kernel/releases/downl/GP-x86-Kernel-RT-5.4.13-200208.tar.xz
Resolving github.com... 140.82.114.3
Connecting to github.com|140.82.114.3|:443... connected.
HTTP request sent, awaiting response... 404 Not Found
2020-04-10 18:04:16 ERROR 404: Not Found.

tar: GP-x86-Kernel-RT-5.4.13-200208.tar.xz: Cannot open: No such file or directory
tar: Error is not recoverable: exiting now
Generating grub configuration file ...
Found background: /boot/grub/gp2.png
Found linux image: /boot/kernel-genkernel-x86_64-5.1.0-pf7-GP-ST-AU-RAVENNA
Found initrd image: /boot/initramfs-genkernel-x86_64-5.1.0-pf7-GP-ST-AU-RAVENNA
Found linux image: /boot/kernel-genkernel-x86_64-5.0.21-rt-rt16-GP-RT-RAVENNA
Found initrd image: /boot/initramfs-genkernel-x86_64-5.0.21-rt-rt16-GP-RT-RAVENNA
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-ST
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-ST.img
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-MIN-GP1-CORE
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-MIN-GP1-CORE.img
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-MIN-GP1
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-MIN-GP1.img
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-MIN-EUP-CORE
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-MIN-EUP-CORE.img
Found linux image: /boot/vmlinuz-5.5.0-pf2-x86_64-PF-MIN-EUP
Found initrd image: /boot/initramfs-5.5.0-pf2-x86_64-PF-MIN-EUP.img
done
wait...5min

 

 

 

gp-update

and use webinterface