This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: type problems in conform tests...
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: David Miller <davem at davemloft dot net>
- Cc: <libc-alpha at sourceware dot org>
- Date: Wed, 23 Apr 2014 15:39:44 +0000
- Subject: Re: type problems in conform tests...
- Authentication-results: sourceware.org; auth=none
- References: <20140421 dot 144641 dot 1442211345567521493 dot davem at davemloft dot net> <Pine dot LNX dot 4 dot 64 dot 1404222136360 dot 31754 at digraph dot polyomino dot org dot uk> <20140422 dot 174926 dot 617708372377562586 dot davem at davemloft dot net>
On Tue, 22 Apr 2014, David Miller wrote:
> For example si_band is "int" instead of "long int". Alpha has this
> problem too.
That may be harder - maybe an internal sigaction flag glibc passes to the
kernel to tell it to use a corrected-layout siginfo, together with some
symbol versioning? In the absence of that, any architecture-specific
XFAIL should at least give a detailed explanation of the ABI issues behind
the expected failures.
(Eventually we may want more fine-grained XFAILs for these tests, but
we're still a long way from simply having the tests reflect the standards
accurately and having the easy bugs in the headers fixed, and I think we
need to get to the point where the expected failures are genuinely hard to
fix before looking at more fine-grained expectations.)
--
Joseph S. Myers
joseph@codesourcery.com