Tim Haines

Confirmed. Thanks for the detailed report! [tau.rhel8.ppc64.klargest.tar.gz](https://github.com/dyninst/dyninst/files/12788090/tau.rhel8.ppc64.klargest.tar.gz) ---- ### Backtrace ```terminal Thread 1 "tau_run" received signal SIGSEGV, Segmentation fault. 0x00007ffff7e8d5f0 in Dyninst::Relocation::Instrumenter::handleCondDirExits (this=, trace=0x13cd65b0, cfg=0x11d26150, exit=) at /home/users/thaines/workspace/typhon/tau2/dyninst-12.3.0-working/dyninstAPI/src/Relocation/Transformers/../CFG/RelocTarget.h:110 110 Target(RelocBlock...

@bigtrak At that time, we didn't have a particularly good reason to remove it the code wasn't causing any trouble. With the new build system, that's no longer the case....

@bigtrak Did you get a chance to take a look at this? In particular, we should try re-doing Will's analysis with a version of Dyninst that doesn't include #1249.

I have a suspicion as to the problem. `DW_AT_call_line` is the line where the function is _called_ (hence, the `call` part). Could you check to see if that function is...

Hmm. I'll take a deeper look. Is it ok to copy that GPU binary off of ufront?

@mwkrentel I think I might need the full source. Now that Sunspot is back up, could you pull a copy to ufront and I'll grab it from there?

Oh, I thought the header files were part of John's code. I have a copy of `vector-add-buffers.cpp` now. I'd recommend deleting that tarball off of ufront- I'm not entirely sure...

The crux of the issue is that there are two CUs for the same file. ``` Compilation Unit @ offset 0x0: Length: 0xf41d (32-bit) Version: 4 Abbrev Offset: 0x0 Pointer...

@mwkrentel @jmellorcrummey The last fix for this is in #1581. You'll also need https://gitlab.com/hpctoolkit/hpctoolkit/-/merge_requests/940 before you can use the new module interface this requires. The line info for the functions...

I couldn't say for certain what the reason is, but it is true that Dyninst does not adjust the DWARF of rewritten binaries. As far as I know, this is...