This Time Self-Hosted
dark mode light mode Search

Strict aliasing part 2

So, after a quite old blog post some three months ago, today strict aliasing breakages are coming into topic again. This time suggested by Luca (lu_zero) on gentoo-dev mailing list.

Now that I can see it’s not just something I care alone, I decided it was about time I started looking to fix some things. I fixed for first xine-lib, directly upstream, so that -fno-strict-aliasing is used only where actually needed, and when possible the warnings are fixed.
After that, I also fixed libfaad2 and libdts, that I encountered in xine-lib itself too, and both are dead upstream, so I don’t have to hunt down where to submit the patches.

Now I’m going to fix kdm, both Gentoo and upstream, and then I’ll try to see what else I maintain to fix them.

The reason why I want to fix all of this is that I do have a PPC, that is one of the architectures that have problems with strict aliasing, and I do care about users that are on different arches so that they can have a good experience with the software I maintain 🙂

Until I have another contract, I’ll be watching and fixing as much stuff as I can about this, but also about a few other questions that are still lingering around. I also want to try to review and commit some patches to xine-lib, both before and after 1.1.2, so that it can improve. Unfortunately some things I’d like to do about that will require me quite a lot of time (like implementing wavpack support that might be useful for some people), mainly to learn the whole framework, and I don’t know if that will be appreciated too much, not as much as fixing generic bugs, say.

Anyway, now I’ll return to my log analysis so that I can find what I need to fix next.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.