This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH roland/bz14743] BZ#14743: Move clock_* symbols from librt to libc.


On Monday 22 October 2012 12:46:08 Roland McGrath wrote:
> > > If nobody cries foul, I'll commit this on Tuesday or Wednesday.
> > 
> > ia64 rt tests are now segfaulting with this change :(
> > 
> > i'll have to debug further
> 
> One quick path to diagnosis: toggle the HAVE_IFUNC conditional in
> clock-compat.c and see if it works then.

yeah, the tests no longer crash after that

configure define HAVE_IFUNC to 1, and libc_cv_asm_gnu_indirect_function=yes, but 
ia64 doesn't have ifunc support.

i guess this gets back to the ifunc configure tests not being sufficient -- it 
tests for the gnu_indirect_function directive, but that is a common directive, 
and the backend isn't involved.  so we end up returning true for all ports 
that are running binutils-2.20 or newer.  which is everyone now since we 
started requiring at least 2.20 :).

perhaps extend the code like so:
	.type foo,%gnu_indirect_function
	foo:
	.data
	.long foo
then compile it and look at the `readelf -r` output and make sure there is a 
relocation there
-mike

Attachment: signature.asc
Description: This is a digitally signed message part.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]