Martin Michlmayr <[EMAIL PROTECTED]> writes:
> It's not as bizarre as it looks. See #386267 for a description of the
> problem. If you can come up with a better fix (i.e one for the glibc
> header), I'd appreciate it.
Aha.
Probably the reason why it was producing warnings (and one of the reasons
why I was confused) is that you don't mean:
WCOREDUMP((struct testset *)ts->status)
which passes a pointer to a struct testset into WCOREDUMP. You mean:
WCOREDUMP(((struct testset *) ts)->status)
since -> has a higher precedence than () casts in C.
My guess is that a compiler change uncovered this. I bet that before,
even if ts was a const struct testset *, an element "status" defined as an
int and referenced as ts->status was still an int, not a const int.
Whatever the problem is, though, it appears to be specific to AMD64. I
can't replicate it on x86 at all, even with the latest unstable compiler.
I'll look at this more from home this evening where I have an AMD64
system.
--
Russ Allbery ([EMAIL PROTECTED]) <http://www.eyrie.org/~eagle/>
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
