All this should be fixed now, and the configure output is now clear (it doesn't change its mind about pci_init/cleanup or pci_lookup_name without any obvious reason anymore).
FC7: checking for pci/pci.h... yes checking for pci_init in -lpci... no checking for pci_init in -lpci with -lz... yes checking for pci_lookup_name in -lpci... no checking for inet_ntoa in -lresolv... yes checking for pci_lookup_name in -lpci with -lresolv... yes RHEL5.6: checking for pci/pci.h... yes checking for pci_init in -lpci... yes checking for pci_lookup_name in -lpci... no checking for inet_ntoa in -lresolv... yes checking for pci_lookup_name in -lpci with -lresolv... yes RHEL5.3: checking for pci/pci.h... yes checking for pci_init in -lpci... yes checking for pci_lookup_name in -lpci... yes Christopher, it should work starting with trunk r3535. Brice Le 30/06/2011 07:50, Brice Goglin a écrit : > Le 29/06/2011 13:18, Brice Goglin a écrit : >> I don't think we finally fixed this. >> >> IIRC, we need either a way to bypass the cache, or always add -lresolv >> even if it's useless (or find another way to detect if lresolv is needed). > Redefining our own HWLOC_AC_CHECK_LIB_NO_CACHE looks possible. > > Otherwise, we could use something different from AC_CHECK_LIB for the > second check (AC_SEARCH_LIBS uses a different cache name). > Or even use AC_LINK_IFELSE/AC_TRY_LINK which never cache anything. > > Brice > > _______________________________________________ > hwloc-devel mailing list > hwloc-de...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/hwloc-devel