[Bug gdb/26819] RISC-V: internal-error: int finish_step_over(execution_control_state*): Assertion

["jmatyas at codasip dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=26819

Jan Matyas <jmatyas at codasip dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|FIXED                       |---
             Status|RESOLVED                    |REOPENED

--- Comment #13 from Jan Matyas <jmatyas at codasip dot com> ---
Hello,

I have finally managed to re-test the updated code. I am sorry to report back
that the issue appears not yet fully resolved. Re-running the GDB test
scenarios in which this issue was originally encountered (that is:
https://github.com/riscv/riscv-tests/tree/master/debug) now triggers a
different assert. 


(1)
In many of the above test scenarios, it is the following assert that gets
triggered immediately after connecting GDB to the remote server, i.e. OpenOCD. 
OpenOCD is at that point already connected to the underlying target, which is
the Spike RISC-V ISA simulator:

(gdb) target extended-remote localhost:44553
Remote debugging using localhost:44553
warning: No executable has been specified and target does not support
determining executable automatically.  Try using the "file" command.
../../../binutils-gdb/gdb/remote.c:7820: internal-error: ptid_t
remote_target::select_thread_for_ambiguous_stop_reply(const
target_waitstatus*): Assertion `first_resumed_thread != nullptr' failed.

A full log from one such test run is attached -
20210116-101128-spike32-CheckMisa.log. It shows the steps to reproduce the
issue - most importantly the cmdline arguments to start Spike and OpenOCD.


(2)
There was one test run in which the original assert was encountered:

../../../binutils-gdb/gdb/infrun.c:5614: internal-error: int
finish_step_over(execution_control_state*): Assertion
`ecs->event_thread->control.trap_expected' failed.

It is recorded in file: 20210116-101308-spike32_2-ProgramSwWatchpoint.log



The revisions of Spike ISA sim, OpenOCD and gdb-binutils that I used were in
all cases the latest HEADs as of now:

- binutils-gdb: 68b007788a (Fri Jan 15 18:07:07 2021 -0800)
- Spike: 35d50bc (Thu Jan 14 09:43:04 2021 -0800)
- OpenOCD: ed37337b0


Please let me know if any further information is needed for reproduction or if
you'd like me to re-test any other scenario.

Best regards,
Jan

-- 
You are receiving this mail because:
You are on the CC list for the bug.