Comment 171 for bug 445849

Revision history for this message
xtknight (xt-knight) wrote : Re: Highpitched Rattling like Sound with 5.1 Surround Configuration on Karmic Koala

Jerther: thank you for your work. You have made instructions from a user's perspective at least, something I didn't have the patience to do. I know my script was trash since the beginning, just had no patience to continue the project.

We knew the problem all along, but as I suspected, people weren't testing PULSE_NO_SIMD=1 correctly. When you try something...you've gotta...actually..try it.. (who would have thought). Besides, software configurations are full of entropy. We need to test things properly and using the scientific method.

I also had problems playing back files in Totem just starting today. They would stop after a short amount of time. I don't think it's because of my 4-line patch though, it just wouldn't make any sense. I've had so many PA problems regardless. The whole totem interface froze and the sound KEPT PLAYING, probably indicating it's not only the sound daemon. I always noticed my volume controls were different (just quieter) using 5.1 than headphones. Again my patch should not have touched any of that territory, although I did not write the assembly code I bypassed so I can never be 100% sure anyway.

You don't need to test the debdiff. I did it correctly...I am pretty sure. It's a pretty simple patch. I used the quilt patch system. A quick look inside it obviously indicates it's not going to do anything hugely bad at least. I don't want them to commit the 'comment patch' yet, anyway. I'd rather have them fix the SIMD. I want to know if a) they are willing to just remove SIMD optimizations for now, and if they are, then I will make a patch with deleted lines instead of commented lines. or b) this is just going to sit in a pipe until real PA people fix it.

If you're curious, installing debdiffs goes something like this (sometimes -p0 can be varied anywhere up to like -p5, what happens is the patch specifies a relative path and this specifies how many subdirs to ignore):
$ mkdir -p ~/compiles/pa
$ cd ~/compiles/pa
$ sudo apt-get build-dep pulseaudio
$ apt-get source pulseaudio
$ patch -p0 < /home/andy/Desktop/pulseaudio_fix.debdiff
$ cd pulseaudio-...
$ debuild
$ cd ..
$ sudo dpkg -i *.deb

Again I'm just writing this on the run, not 100% certain. Typically a debdiff is committed to a "proposed" archive where certain people that signed up for "proposed" packages test them. My debdiff comments the code out. It would be better to delete the code. Hell, it would be better to track down the bug in the assembly, but that's probably asking too much. I don't do x86 assembly (yet). And I've got no clue what's going on in that code. It's like a huge hairy bear. Sorry =)

Glad it worked for you. Let's rally and get a patch (and SRU--stable release update) into the official repositories. I'll try to do this when I get the time. It would be glad to at *least* hear an official word before we started all that, but hey who knows. Maybe that's what will prompt them.