HQPlayer Linux Desktop and HQplayer embedded

[shadowlight]

@Miska

It has been awhile since I looked at the hqplayerd log file, but I am seeing messages similar to this.  Am I clipping?  This is with Redbook upconverted to pc 352.  My max fixed volume is set to 0, with startup set for -3db

 

2018/01/29 19:28:11 Limiting for channel 0 gain -0.029578 dB
  2018/01/29 19:28:11 Limiting for channel 1 gain -0.0850421 dB
  2018/01/29 19:28:12 Limiting for channel 0 gain -0.0825723 dB
  2018/01/29 19:28:12 Limiting for channel 1 gain -0.330939 dB
  2018/01/29 19:28:13 Limiting for channel 0 gain -0.499887 dB
  2018/01/29 19:28:13 Limiting for channel 1 gain -0.367265 dB
  2018/01/29 19:28:13 Limiting for channel 1 gain -0.434788 dB
  2018/01/29 19:28:13 Limiting for channel 0 gain -0.977933 dB
  2018/01/29 19:28:14 Limiting for channel 1 gain -0.770779 dB
  2018/01/29 19:28:15 Limiting for channel 1 gain -1.1094 dB
  2018/01/29 19:28:21 Limiting for channel 0 gain -0.992944 dB
 

[shadowlight]

On 1/29/2018 at 7:35 PM, shadowlight said:

@Miska

It has been awhile since I looked at the hqplayerd log file, but I am seeing messages similar to this.  Am I clipping?  This is with Redbook upconverted to pc 352.  My max fixed volume is set to 0, with startup set for -3db

 

2018/01/29 19:28:11 Limiting for channel 0 gain -0.029578 dB
  2018/01/29 19:28:11 Limiting for channel 1 gain -0.0850421 dB
  2018/01/29 19:28:12 Limiting for channel 0 gain -0.0825723 dB
  2018/01/29 19:28:12 Limiting for channel 1 gain -0.330939 dB
  2018/01/29 19:28:13 Limiting for channel 0 gain -0.499887 dB
  2018/01/29 19:28:13 Limiting for channel 1 gain -0.367265 dB
  2018/01/29 19:28:13 Limiting for channel 1 gain -0.434788 dB
  2018/01/29 19:28:13 Limiting for channel 0 gain -0.977933 dB
  2018/01/29 19:28:14 Limiting for channel 1 gain -0.770779 dB
  2018/01/29 19:28:15 Limiting for channel 1 gain -1.1094 dB
  2018/01/29 19:28:21 Limiting for channel 0 gain -0.992944 dB
 

@Miska, anything for me to be concerned with?

 

Can other folks share their volume settings and if you are seeing similar messages in your log file?

[shadowlight]

The limiting messages are only occurring when I am playing PCM.  If I change the setting from auto to see the message goes away.

[Miska]

On 1/29/2018 at 11:08 AM, Mogos said:

Thank you for info. I have never been using Linux but it is not a problem to learn it :). Do you use the Ubuntu version?

 

Yes, in the RME configuration. While on other machine with TEAC NT-503 I use Debian Stretch.

 

[Miska]

On 1/30/2018 at 2:35 AM, shadowlight said:

@Miska

It has been awhile since I looked at the hqplayerd log file, but I am seeing messages similar to this.  Am I clipping?  This is with Redbook upconverted to pc 352.  My max fixed volume is set to 0, with startup set for -3db

 

If you have fixed volume setting, the default startup value doesn't really matter. So essentially you have volume set to "0" and that's why you are getting overs and limiting. If you want variable volume, remove the fixed volume setting from config file. If you want fixed volume, set it to some value you like.

[Miska]

On 2/1/2018 at 1:10 PM, shadowlight said:

The limiting messages are only occurring when I am playing PCM.  If I change the setting from auto to see the message goes away.

 

If you have Direct SDM enabled and decide to output SDM from PCM sources, then you have the fixed -2.5 dB gain. With "auto" mode you get PCM->PCM and DSD->DSD playback.

 

[shadowlight]

On 2/6/2018 at 5:22 AM, Miska said:

 

If you have Direct SDM enabled and decide to output SDM from PCM sources, then you have the fixed -2.5 dB gain. With "auto" mode you get PCM->PCM and DSD->DSD playback.

 

Thanks, I also noticed that the newer version of BubbleUPnP has changed how volume is controlled.  I will have to play around it.

 

I am running into a new situation with the latest version of beta.  Any time I make a change and apply the settings it seems to hang for a long time.  Even doing a systemctl stop or restart takes couple of minutes.  I have confirmed that the network connectivity does not have any issues.  I have a ssh session and a ping session going at the same time and both systems are responsive.  

 

The kernel version that I am using is the latest custom kernel that you have released for Debian and the Debian system is up to date.

 

[shadowlight]

I reinstalled Debian and HQPe to see if my installation was corrupt.  Now I am seeing the following error message and I am no getting the ability to select the NAA.

 

"clPlayerDaemon::WebConfiguration(): clNetMiniEngine::Disco(): clSocket::Recv(): recv(): Connection refused"

[Miska]

5 hours ago, shadowlight said:

I reinstalled Debian and HQPe to see if my installation was corrupt.  Now I am seeing the following error message and I am no getting the ability to select the NAA.

 

"clPlayerDaemon::WebConfiguration(): clNetMiniEngine::Disco(): clSocket::Recv(): recv(): Connection refused"

 

Hmmh, that is strange, looks like some network problem. Are you starting hqplayerd with the provided systemd service file? Is the network configured through /etc/network/interfaces file? Does the network interface properly show IPv4 and IPv6 IP's? Could you check the NAA side too?

 

This error is likely happening at the point when HQP is talking to the NAA to fetch information about the available devices. Something goes wrong between the two at this point.

[shadowlight]

6 hours ago, Miska said:

 

Hmmh, that is strange, looks like some network problem. Are you starting hqplayerd with the provided systemd service file? Is the network configured through /etc/network/interfaces file? Does the network interface properly show IPv4 and IPv6 IP's? Could you check the NAA side too?

 

This error is likely happening at the point when HQP is talking to the NAA to fetch information about the available devices. Something goes wrong between the two at this point.

 

I do not believe it was a network problem, since I was able to access the web interface and was connected via ssh.  In /tmp/hqplayerd.log file I was also seeing NAA discovery message, from two different NAA.  One is running on Win10 Enterprise and another is running under Debian.  Right before I left for work today, I restarted the system and I was no longer gettting the error message but the hanging problem after making changes via web interface is back.  So I am back to the original hanging issues.  The services is starting based on the provided systemd service file.

 

Here is what I have tried so far:

swapped network port on the switch, changed the ethernet cable, installed Debian on a brand new ssd drive, the only thing that i have not changed is the PC and network interface built-in on the motherboard (ThinkServer TS140 with Xeon processor).

 

To confirm that network interface was not an issue I ran two seperate tcpdump commands one checking for traffic on port 8088 and another ssh port.  I was seeing plenty of traffic going back and forth on ssh port.

 

I can start and stop the services after the reboot with out any hanging from commandline, but as soon as I make a change via web interface both the reload that is suppose to happens hangs and similar hanging issues happen via the commandline stop/start command.

[Miska]

What does NAA side log say when that happens? This happens after the initial discovery message...

 

I don't have other ideas right now. I cannot reproduce the problem on my installation...

 

If you could email me log snippets of this happening from both HQP and NAA sides.

 

Does the problem go away if you disable the built-in Rygel UPnP (assuming you are using it)?

[Miska]

Regarding volume control, check out the included rmeaio example for volume control that is fixed to -3 dBFS.

[shadowlight]

18 hours ago, Miska said:

What does NAA side log say when that happens? This happens after the initial discovery message...

 

I don't have other ideas right now. I cannot reproduce the problem on my installation...

 

If you could email me log snippets of this happening from both HQP and NAA sides.

 

Does the problem go away if you disable the built-in Rygel UPnP (assuming you are using it)?

Ok, found the problem.  I had two instances of NAA running on the Win10 system, not sure how that happened but I was able to recreate the hanging problem everything I ran two instances of NAA.

 

Are segfault messages like below in syslog file are normal when you make changes via the web interface?

 

Feb  9 11:10:30 rhythm kernel: [  623.964193] hqplayerd[827]: segfault at 7f19955be010 ip 000055e36c232709 sp 00007ffd108a54c0 error 4 in hqplayerd[55e36c1b3000+459000]
Feb  9 11:10:30 rhythm systemd[1]: hqplayerd.service: Main process exited, code=killed, status=11/SEGV
Feb  9 11:10:30 rhythm systemd[1]: hqplayerd.service: Unit entered failed state.
Feb  9 11:10:30 rhythm systemd[1]: hqplayerd.service: Failed with result 'signal'.
Feb  9 11:10:32 rhythm systemd[1]: hqplayerd.service: Service hold-off time over, scheduling restart.
Feb  9 11:10:32 rhythm systemd[1]: Stopped HQPlayer Embedded daemon.
Feb  9 11:10:32 rhythm systemd[1]: Started HQPlayer Embedded daemon.

 

[cat6man]

On 2/8/2018 at 3:34 AM, Miska said:

 

Are you starting hqplayerd with the provided systemd service file?

 

Hi Miska,

 

I'm running beta19 but starting the program the way I always have, opening 2 command windows, typing "hqplayerd" in one and "rygel" in the other.

Can you explain the systemd service file and what changed?  My old method seems to work fine.

 

Also, my system has been very stable since the latest bubbleupnp updates.  In particular, in addition to turning off the local renderer as I always have, I have hidden both other renderers (only sees hqp4) and other libraries (3 others are found on my network).

Previously I would find that the renderer had changed to local renderer and sometimes the library would also change.

Now that there are no other choices for bubbleupnp to pick, it seems much more solid.

[Miska]

On 2/9/2018 at 6:15 PM, shadowlight said:

Are segfault messages like below in syslog file are normal when you make changes via the web interface?

 

No, segfaults are never normal. I did some review of the code and found refcounting issue in the Rygel handling code and fixed that for the next release... Hopefully that'll get rid of such.

 

[Miska]

4 hours ago, cat6man said:

I'm running beta19 but starting the program the way I always have, opening 2 command windows, typing "hqplayerd" in one and "rygel" in the other.

Can you explain the systemd service file and what changed?  My old method seems to work fine.

 

Service file has always been there, just to help up starting things up at boot on headless setup.

 

But in most recent setup you don't need to run rygel separately anymore, instead it can run within HQPlayer. You can also continue to use your current method instead.

 

Just check with "sudo systemctl status hqplayerd" that it is not started at boot when using your current method. It should say the service is "disabled".

 

[Miska]

b21 is out and should be now close to first non-beta release.

 

Now there's configuration pages for multichannel speaker setup and convolution engine. Library file has been relocated to $HOME/.hqplayer or $HOME depending on user/system scope. In addition, network interface for UPnP can be now defined as "auto", meaning it will be bound to all physical interfaces found at the startup time.

 

On Ubuntu build, CUDA support is now enabled.

 

[shadowlight]

On 2/21/2018 at 4:56 PM, Miska said:

b21 is out and should be now close to first non-beta release.

 

Now there's configuration pages for multichannel speaker setup and convolution engine. Library file has been relocated to $HOME/.hqplayer or $HOME depending on user/system scope. In addition, network interface for UPnP can be now defined as "auto", meaning it will be bound to all physical interfaces found at the startup time.

 

On Ubuntu build, CUDA support is now enabled.

 

 

Nice.  I will have to try it out over the weekend.  I will not be able to test the CUDA support yet.

[cat6man]

any info on beta22?

[Miska]

9 hours ago, cat6man said:

any info on beta22?

 

Ales found some bugs on the control API, it fixes those...

[ppy]

Completely disabled IPV6 support through grub - GRUB_CMDLINE_LINUX_DEFAULT="ipv6.disable=1 quiet"

HQPlayer v4.0.0b22 now does not start - [hqplayerd] : clSocket::Create(): socket(): Address family not supported by protocol

[Miska]

7 hours ago, ppy said:

Completely disabled IPV6 support through grub - GRUB_CMDLINE_LINUX_DEFAULT="ipv6.disable=1 quiet"

HQPlayer v4.0.0b22 now does not start - [hqplayerd] : clSocket::Create(): socket(): Address family not supported by protocol

 

Yes, at the moment control API uses IPv6 always, through the IPv4-in-IPv6 support where entire IPv4 address space is mapped into one IPv6 subnet.

 

Do you have some particular reason to disable IPv6? I personally prefer IPv6 over IPv4...

 

[ppy]

7 hours ago, Miska said:

Do you have some particular reason to disable IPv6? I personally prefer IPv6 over IPv4...

 

Yes, there are problems with connecting HQPDcontrol. I was hoping that by disabling IPv6, port 4321 would be opened in IPv4. Android and HQPlayer are in the same IPv4 address space but in physically different networks (Wi-Fi 2.4 and 5Ghz)

[Miska]

14 hours ago, ppy said:

Yes, there are problems with connecting HQPDcontrol. I was hoping that by disabling IPv6, port 4321 would be opened in IPv4. Android and HQPlayer are in the same IPv4 address space but in physically different networks (Wi-Fi 2.4 and 5Ghz)

 

It is working for me and Ales, so likely it is something else. Some WiFi routers have a setting to prevent WiFi-to-WiFi connections, you could check if this is the case for you.

 

Anyway, I added environment variable to control IPv6 support and built b23 with this change. Now if you set environment HQPLAYERD_IPV6=0 then control interface binds only to IPv4. Note that the NAA support side has a separate configuration item for IPv6 support in the configuration file. But that applies only if you use network backend.

 

[ppy]

8 hours ago, Miska said:

Some WiFi routers have a setting to prevent WiFi-to-WiFi connections, you could check if this is the case for you.

Communication over IPv4 WiFi-to-WiFi is working. But I have a very complex network infrastructure. Combined home network with corporate. I have to disable IPv6 support on all network routers and devices.

8 hours ago, Miska said:

Anyway, I added environment variable to control IPv6 support and built b23 with this change. Now if you set environment HQPLAYERD_IPV6=0 then control interface binds only to IPv4.

Thank you! I hope this option is useful to someone else.