[Ltrace-devel] DWARF prototypes: handling symbol aliases

[Dima Kogan]

Thierry Fauck at linux.vnet.ibm.com <thierry at linux.vnet.ibm.com> writes:

> Dima,
>
> I tested your case and I got the following result on my ppc64le box, so
> I wonder if the problem you mention is not related to some
> implementation - let me explain with my 2 cases on ppc64el and ppc64
>
> ppc64 implemented a intermediate translation table ( OPD) when ppc64el
> is not any more.
>
> Don't know if that comfort your idea.
> Let me know if you need other tests to confirm your point

Hi Thierry.

The issue I'm seeing (aliased symbol names in libc) is specific to
particular builds of libc. It would appear that on the machine you're
looking at, there's a "nanosleep" symbol in the DWARF data of libc.
That's great. I'm using a fairly generic Debian/sid amd64 box, and
there's no "nanosleep" DWARF entry here. My setup is very common, so
it's very desirable for this particular case to work.

Thanks for testing

dima