Re: latest merged ppc 32/64 NPTL port
Roland McGrath <[email protected]>
| Newsgroups | gmane.comp.lib.phil |
|---|---|
| Message-ID | <[email protected]> |
> OK, I'll double check. I did a make clean; make; make tests about 6 > hours ago and that worked. I'll update gcc and binutils from cvs and > redo the build from scratch to check. I realized that in my libc build I had not given --with-headers and so was using 2.4.19 headers. When I rebuilt from scratch using 2.5.64 headers, most of the failures went away. I am curious as to what the old headers could have been missing that would result in a miscompiled library rather than compile-time failures. With my new build, I see only two failures in nptl/ tests. Both tst-cond10 and tst-tls1 die with: Timed out: killed the child process but it exited 0 This could just be a timing thing I suppose, but those two fail consistently on repeated runs (at least a few). I haven't rerun the only top-level make check yet, but the first time I had failures only in nptl/ and rt/, and now there are no rt/ failures and nptl/ has only the two mentioned above. Thanks, Roland