> [ I would not call it "gcc 4.4.4", it is gcc-4.4.4-10.fc13 which is far from
> FSF GCC 4.4.4. ]
>

Gotcha!

> Anyway while it fixes the problem for me on Fedora 13 x86_64 (and even others)
> the testcase does not FAIL with FSF GDB HEAD (reproducing the problem from
> Dodji as is in the PR).
>

You are right. My bad; I should not have missed that. I fixed the test 
case by giving the new functions their own unique names.

I will commit the attached patch