[Slackbuilds-users] ladspa_sdk sandbox violation and potential segfault

[B Watson]

On 8/11/18, orbea at fredslev.dk <orbea at fredslev.dk> wrote:
> Hi,
>
> As reported in the "SBo scripts not building on current" thread at
> linuxquestions showed that the tests in the makefile can segfault.
>
> https://www.linuxquestions.org/questions/slackware-14/sbo-scripts-not-building-on-current-read-1st-post-pls-4175561999/page91.html#post5890532

The user's very next post says "A reboot made things work again". So
his system was in a screwed-up state. Probably any attempt to play
audio with any application would have caused a segfault.

Anyway, it's using sndfile-play to play the sound, so it would be
sndfile-play segfaulting: not a ladspa_sdk issue at all.

> Neither ponce or I could reproduce this, but I also noticed they are
> also sandbox violations which can be easily reproduced with:
>
> sandbox ./ladspa_sdk.SlackBuild

All I see are writes to /dev/snd/controlC0 (aka the alsa mixer device for
card 0). sndfile-play is linked to libasound, which opens the control
device for writing when it's initialized. Nothing to worry about. It's
not creating garbage files in the filesystem.

> I have attached a patch for the SlackBuild which should accomplish this.

I wasn't going to apply your patch since I don't consider the 2 issues
above to be real issues... but I thought of a reason to apply it: what if
the build host hasn't got a sound card at all? Either because it's got a
weird one that the kernel hasn't got a driver for, or because it's a VM
and whoever set it up thought "a test build system doesn't need sound"
so they didn't enable it...

So thanks for looking into this, and I'll patch it in time for the next
public update.