Comment 3 for bug 1162139

I've concluded that the best thing to do is to disable the i386 assembly optimisations altogether. This was discussed in the Debian bug last year, and gcc-4.4 was used as a workaround. I've asked them to revisit this again.

I am reluctant to suggest anything else, since given the current swing in interest towards MariaDB and Percona, I think it is prudent to keep our delta against Debian and upstream minimal. We could fix the i386 assembly so that upstream could accept it, but I don't think that's a useful use of time. But if anyone else wants to have a go at that, please go ahead.