[grsec] Nvidia libGL.so Problem

[pageexec at freemail.hu]

> Sep  6 18:21:18 amoeba PAX: execution attempt in: /usr/lib/opengl/nvidia/lib/libGL.so.1.0.6111, 2574c000-2575b000 00058000
> Sep  6 18:21:18 amoeba PAX: terminating task: /usr/X11R6/bin/glxinfo(glxinfo):7661, uid/euid: 9999/9999, PC: 25755afc, SP: 5bc95e2c
> Sep  6 18:21:18 amoeba PAX: bytes at PC: 65 a1 c8 ff ff ff ff a0 18 04 00 00 cc cc cc cc cc cc cc cc
> Sep  6 18:21:18 amoeba PAX: bytes at SP: 25daa80a 00000000 00000000 00000064 00000064 080662c8 0808ad58 269fc040 080ad400 080ad400 080ad430 080ad430 0000002c 0000002c 25974540 25974188 00000000 00000000 00000000 00008c09

this piece of code has something to do with thread local storage (TLS),
it retrieves some function pointer table pointer from there then jumps
through an entry in it. now, there's some investigation to do ;-). first,
could you make libGL.so available somewhere? second, you could in the
meantime provide the following information:

  readelf -e /usr/lib/opengl/nvidia/lib/libGL.so.1.0.6111

then also try paxctl/chpax -m on /usr/X11R6/bin/glxinfo and see if that
gets it to run (or at least die at a different place).

all the above will help establish where exactly that piece of code is,
whether it should be there (i.e., is it and does it really have to be
runtime generated, or it's just a toolchain/nvidia bug) and maybe, how
it can be fixed (in any case, it'll be nvidia's job).