effect parameter ranges not validated

Bug #1795234 reported by Fabzgy on 2018-09-30
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Mixxx
2.1
Critical
RJ Skerry-Ryan
2.2
Critical
RJ Skerry-Ryan

Bug Description

I was DJing yesterday with my normal setup and Mixxx produces noise realy loud above the track which appears to be playing in the background. The noise starts when I hit play of any file (just tried mp3 so far but I played these files already dozens of time). The two middle green/yellow/red columns are constantly colored.

I am running 2.1.4 (build 2.1 r6765) at Ubuntu 18.04.
Problems remains as well with 2.0.0 (build 1.12 r5772).

GDP log (for build 2.1 r6765) is attached.

Thanks for looking into this.

Fabzgy (fabzgy) wrote :
Fabzgy (fabzgy) on 2018-09-30
description: updated
Fabzgy (fabzgy) on 2018-09-30
description: updated
description: updated
RJ Skerry-Ryan (rryan) wrote :

What sound devices do you have enabled? Any inputs (aux or microphone) configured?

Does the sound appear in a recording?

Could you post a log file? (i.e. ~/.mixxx/mixxx.log right after shutting down mixxx when it happens)

Is this repeatable or intermittent?

Changed in mixxx:
importance: Undecided → High
importance: High → Critical
Fabzgy (fabzgy) wrote :

< What sound devices do you have enabled? Any inputs (aux or microphone) configured?
No mic or aux configured. It occurs with my internal audio device as well as with my usb device.

< Does the sound appear in a recording?
When I record the DJ set within Mixxx the noise is not recorded.

< Could you post a log file? (i.e. ~/.mixxx/mixxx.log right after shutting down mixxx when it happens)

See attachement.

< Is this repeatable or intermittent?

Happens all the time - can t play any sound ... it start immediately when I play a track. VLC (using Pulse is working just fine with the same audio device).

RJ Skerry-Ryan (rryan) wrote :

> When I record the DJ set within Mixxx the noise is not recorded.

Ah ok, so this isolates the problem a little bit. Since you can see the VU meters going crazy, and you can't hear it in the recording, I think this means it occurs sometime after the sidechain splits off from the master mix. That suggests either applying the balance, or the post-fader effects for the master output.

https://github.com/mixxxdj/mixxx/blob/1beb75e077801df4ff44dc336e68d100c980483a/src/engine/enginemaster.cpp#L672-L707

Can you post your ~/.mixxx/effects.xml? Are there any effects enabled on the master output whne this happens?

Fabzgy (fabzgy) wrote :

I do not have an efect active on purpouse. I attach my ~/.mixxx/effects.xml here in this comment

Fabzgy (fabzgy) wrote :

Now I had a crash of my entire system when mixxx started. Something was writing data as hell to my SSD but I can t figure what happened. /fabzgy/.mixxx/mixxx.log is empty.

I experienced this already once during a live radio session. Back then I was playing a track and my entire system froze (that has never happened before).

This gets really dirty right now. I really love mixxx but reliability is super important as a radio host as well as a DJ!

I would appreciate further help...

Fabzgy (fabzgy) on 2018-10-04
summary: - Mixxx just produces noises
+ Mixxx just produces noises & crashes the entire OS

This sounds bad, and I'm sorry you're experiencing such instability :(. I have some questions:

- Can you paste the output of "cat /proc/cpuinfo"

- What kernel are you running? Please paste the output of "uname -a" from a terminal.

- Could you please post your soundconfig.xml from ~/.mixxx/

- Do you see any Mixxx crash report in /var/crash ? If you do, please post it here (note that a crash report can contain personal information, like your terminal environment variables. If you'd prefer to email it, my email is <email address hidden>).

RJ Skerry-Ryan (rryan) wrote :

Also, can you try 2.2 beta? https://mixxx.org/download

Fabzgy (fabzgy) wrote :

here the kernel version:
uname -a
Linux fabzgy 4.15.0-36-generic #39-Ubuntu SMP Mon Sep 24 16:19:09 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux

Soundconfig:
<!DOCTYPE SoundManagerConfig>
<SoundManagerConfig force_network_clock="0" latency="5" deck_count="4" api="ALSA" samplerate="44100" sync_buffers="2">
 <SoundDevice name="0, HDA Intel PCH: ALC3202 Analog (hw:0,0)">
  <output channel="0" type="Master" index="0" channel_count="2"/>
 </SoundDevice>
</SoundManagerConfig>

cpuinfo is attached.

Fabzgy (fabzgy) wrote :

There is no crash report in /var/crash. Maybe because the crash was a couple of days ago already?

I ve just installed 2.2.0-beta (build 2.2 r6578) but the behavior remains. Same problem as with 2.1.4.
I will downgrade again because a stable program is more important for me then new features.

RJ Skerry-Ryan (rryan) wrote :

could you post your ~/.mixxx/mixxx.cfg

Also, on that note -- does the problem persist when you start with a new profile? (i.e. move ~/.mixxx to ~/.mixxx.bak)

Fabzgy (fabzgy) wrote :

A new profile "solves" the problem. At least the VU does not go crazy anymore.
But I am still willing to figure out where the problem is / was. Any chance of debugging this further?

RJ Skerry-Ryan (rryan) wrote :

The VU meter doesn't go crazy, and you don't get noise?

Could you please post your mixxx.cfg (or a zip/tar of the whole .mixxx foldre) from the buggy profile?

Fabzgy (fabzgy) wrote :

Here is my buggy .mixxx profile. I kept the analysis folder out because it is almost 4 GB big sic!). Is that how it is supposed to be?

RJ Skerry-Ryan (rryan) wrote :

Ah thanks a lot for the mixxx.cfg. I can reproduce this! (on macOS, 2.2 beta)

The graphic EQ effect parameters in your mixxx.cfg are way outside the valid ranges of -12 to 12. These parameter values aren't validated when we load them, and they cause the graphic EQ effect to go absolutely crazy on my machine, and as you say, cause really loud noises even when no tracks are playing.

EffectForGroup_[Master] org.mixxx.effects.graphiceq
EffectForGroup_[Master]_parameter1 0
EffectForGroup_[Master]_parameter2 1.75
EffectForGroup_[Master]_parameter3 1000
EffectForGroup_[Master]_parameter4 0
EffectForGroup_[Master]_parameter5 1.75
EffectForGroup_[Master]_parameter6 3000
EffectForGroup_[Master]_parameter7 0
EffectForGroup_[Master]_parameter8 0

Starting over with a new profile disabled the graphic EQ effect, and cleared the parameter values.

Now I'm wondering how those got there in the first place. Are they valid for the parametric EQ effect?

Changed in mixxx:
status: New → Confirmed
RJ Skerry-Ryan (rryan) wrote :

Ah, those are the default values for the parametric EQ effect. So that explains how they got there.

Be (be.ing) on 2018-11-03
summary: - Mixxx just produces noises & crashes the entire OS
+ parameter values from old master EQ effect used when switching master EQ
+ effects
summary: - parameter values from old master EQ effect used when switching master EQ
- effects
+ parameter values from old effect used when switching master EQ effects
RJ Skerry-Ryan (rryan) on 2018-11-03
Changed in mixxx:
assignee: nobody → RJ Skerry-Ryan (rryan)
summary: - parameter values from old effect used when switching master EQ effects
+ effect parameter ranges not validated
Fabzgy (fabzgy) wrote :

So this means you simply need to change the default values of the parametric EQ and validate the values before applying them?
Thats the solution?

RJ Skerry-Ryan (rryan) wrote :
Changed in mixxx:
status: Confirmed → In Progress
RJ Skerry-Ryan (rryan) wrote :

I haven't figured out how to get my mixxx.cfg into the same state yours was in (I reproduced by copy-pasting those lines into my mixxx.cfg) -- I've tried switching the parameters and settings in the EQ preferences in various different orders and can't produce it. Maybe we fixed the bug in the preferences that allowed you to get into that state at some point.

I found that we aren't actually validating effect parameter values though, so my fix to the validation code should prevent bad mixxx.cfg files from causing problems like this.

I'm not sure how your whole computer froze as a result of this bug though.. my guess is that was unrelated?

RJ Skerry-Ryan (rryan) on 2018-11-04
Changed in mixxx:
status: In Progress → Fix Committed
Fabzgy (fabzgy) wrote :

The freezing of the entire system was probably related to: https://askubuntu.com/questions/1036636/mitigation-for-gvfs-memory-leak

But this is unrelated I guess. It happened as well when mixxx was not running.

no longer affects: mixxx
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers