ArchPPC32 has incorrect PC register number about pyvex HOT 6 CLOSED

Hi jeffball,

Sorry for the laggy response. I was at SECCON, and didn't have a chance to look into this. You've definitely triggered a "how did this ever work" moment. I'll look into this a bit further and get back to you.

from pyvex.

Can we have whatever sample binary you're working with that's little-endian 32 bit powerpc? As far as I know, code of this type has never been tested with angr, and might not actually exist?

from pyvex.

Ah, I should have mentioned, from my testing, it looks like this holds for
big-endian PPC (fauxware) as well.

I have a vague half-memory of running into this before for some reason, but
can't remember the details...
On Feb 2, 2016 8:51 AM, "Andrew Dutcher" [email protected] wrote:

Can we have whatever sample binary you're working with that's
little-endian 32 bit powerpc? As far as I know, code of this type has never
been tested with angr, and might not actually exist?

—
Reply to this email directly or view it on GitHub
#27 (comment).

from pyvex.

The vex source comments indicate that yes, it is in fact at that offset. I have no idea where this bug is introduced, and since this source has been copied around since before I was on this project it's impossible to tell when it was introduced, but I'll push a fix. :+)

from pyvex.

Fix pushed to the archinfo repository. Can you verify it resolves your issue?

from pyvex.

Yep, after upgrading archinfo from the archinfo github repository, this does fix the issue. Thanks!

from pyvex.