On Wed, Jul 13, 2011 at 08:27:13AM -0400, Jeff Squyres wrote:
> On Jul 12, 2011, at 1:37 PM, Steve Kargl wrote:
> > (many lines removed)
> > checking prefix for function in .type... @
> > checking if .size is needed... yes
> > checking if .align directive takes logarithmic value... no
> > configure: error: No atomic primitives available for amd64-unknown-freebsd9.0
> Hmm; this is quite odd. This worked in v1.4, but didn't work in trunk?
> There are a bunch of changes to our configure assembly tests between v1.4, but I don't see any that should affect AMD vs. Intel. Weird.
> I wonder if this has to do with versions of config.* scripts. What does config/config.guess report from the trunk tarball, and what does it report from the v1.4 tarball?
After I run configure in openmpi-1.4.4rc2 and trunk, I see
diff -u openmpi-1.4.4rc2/config/config.sub ompi/config/config.sub | more
@@ -354,73 +292,56 @@
| a29k-* \
| alpha-* | alphaev[4-8]-* | alphaev56-* | alphaev6-* \
| alpha64-* | alpha64ev[4-8]-* | alpha64ev56-* | alpha64ev6-* \
- | alphapca5-* | alpha64pca5-* | arc-* \
+ | alphapca5-* | alpha64pca5-* | amd64-* | arc-* \
@@ -458,9 +376,6 @@
- basic_machine=x86_64-`echo $basic_machine | sed 's/^[^-]*-//'`
So, it appears that the amd64-* rule has changed, but where to undo
the "fix" I do not know.