Skip to content

Disable architecture specific optimisations on ARM,

Daniel Schürmann requested to merge daschuer/libmad:deb1108248 into master

This i a backport of an Ubuntu patch:

As there is a bug in this codepath which causes segfaults, and the assembler is very old (likely bitrotted). (LP: #989846)

My assumption is that the compilers are smart enough these days to do the implemented vectorization atomatically. This patch does not bypass the ASM code for the fixed point math which is still faster than the compiler generated equivalent.

Without this patch Mixxx on armhf targets which is using libmad crashes right away when loading an MP3

This has been tested on my PPA:daschuer/mixxx

The pipeline fault is unrelated.

Merge request reports