Re: [PATCH] [gdb/testsuite] Fix gdb.base/inline-frame-cycle-unwind.exp for s390x (alternative)

[Tom de Vries <tdevries@suse.de>]

On 1/20/26 9:38 PM, Andrew Burgess wrote:
> Having spent some time now running and debugging this test, I think one
> neat thing about the way it was originally written, is that the Python
> script could be used stand-alone, so long as the test program was
> stopped in the right place you could source the Python script and it
> would do the right thing.
> 
> You proposed patch changes this slightly so that the .exp script is
> needed to setup some of the frame-id data.  That's absolutely fine, but
> can make the test harder to run outside DejaGNU; if the stack addresses
> change the user needs to figure out the updated values.  It's a really
> minor thing, but how about the patch below (still needs a commit
> message), but this does more of the work in the Python script again.

Hi Andrew,

I've taken your patch, and added a bit onto it:
- capture all available frame IDs: for frames #0 - #6.  Frame #6 was
   missing.
- print the captured frame IDs
- added a comment in TestUnwinder.__call__ that explains in detail what
   the normal behavior is, and how we're deviating from it to make the
   backtrace stop.
- added commit message

I've submitted this as a v2 ( 
https://sourceware.org/pipermail/gdb-patches/2026-January/224308.html ).

I hope my understanding of what the test-case is trying to do is correct 
now.

Thanks,
- Tom