make check failures

[Andrew Walrond]

Hi Richard,

Richard Henderson wrote:
 >
 > I am not mistaken about how it is *supposed* to work.
 > If this works for you, something is clearly broken.
 >

gcc 3.2.2 docs mention sse, but not sse2 in relation to -mfpmath
-mno-sse2 is however mentioned seperately.

I suggest you take this up with Ulrich, as he seems to know whats going on.

Just build glibc with -march=pentium4 -O2, then do a make check and 
you'll see this behaviour

Andrew Walrond