[Ltrace-devel] current make check failures

Ian Wienand ianw at gelato.unsw.edu.au
Thu Oct 12 11:59:44 UTC 2006


On Wed, Oct 11, 2006 at 10:36:56PM -0700, Steve Fink wrote:
> Basically, on ia64 it seems like opd2addr() shouldn't do anything to the
> passed-in address. The test passes, at least, if you just have it return the
> address unmodified.

I get the feeling we've been here before.  I wrote up a comparison a
long time ago (and I thought this bug was fixed)

http://www.gelato.unsw.edu.au/IA64wiki/PPC64ABI

In summary, IIRC, on IA64 taking the address of an external function
leaves a relocation in the code that is fixed up by the dyanmic
linker.  So you don't need to do anything, as you identified.

On PPC, the code always refers to an entry in the .opd and a
relocation gets the dynamic linker to fix this entry in the .opd up,
hence the need to get the address in the .opd to break on.

The confusion is that both have the .opd section because on both you
have function descriptors.

It's been a while since I looked over this, any input is fine :) ...

-i
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/ltrace-devel/attachments/20061012/0b23b6f0/attachment.pgp


More information about the Ltrace-devel mailing list