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

["andrew.burgess at embecosm dot com" <sourceware-bugzilla@sourceware.org>]

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

Andrew Burgess <andrew.burgess at embecosm dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |andrew.burgess at embecosm dot com

--- Comment #1 from Andrew Burgess <andrew.burgess at embecosm dot com> ---
Jan,

Thank you for including such clear steps to reproduce, they were very helpful.

I tried following the steps but could not reproduce this issue, however, thanks
to your great instructions I was able to see where I diverge from your
behaviour.

When you start GDB, you report this:

 (gdb) info threads
   Id   Target Id                     Frame 
 * 1    Thread 1 (Name: Hart 0, RV64) 0x00000012123408b0 in stack_bottom ()
   2    Thread 2 (Name: Hart 1, RV64) 0x000000121234089c in stack_bottom ()
 (gdb) thread 2

However, when I start GDB, I see this:

 (gdb) info threads
   Id   Target Id                     Frame 
 * 1    Thread 1 (Name: Hart 0, RV64) 0x0000001212340840 in stack_bottom ()

I did try following the rest of your steps using thread 1, but the bug didn't
reproduce.

The fact that I only see a single thread doesn't seem too surprising given that
in my openocd startup I see this:

 Warn : `-rtos riscv` is deprecated and will be removed at the end of 2020!
Please
 Warn : change your configuration to use `-rtos hwthread` instead. To do that,
you
 Warn : will have to explicitly list every hart in the system as a separate
target. See
 Warn :
https://github.com/riscv/riscv-tests/blob/ec6537fc4a527ca88be2f045e01c460e640ab9c5/debug/targets/SiFive/HiFiveUnleashed.cfg#L11
 Warn : for an example.

And:

 Info : Examined RISC-V core; found 1 harts
 Info :  hart 0: XLEN=64, misa=0x8000000000141129

Could you confirm the exact git SHA you are using for spike and openocd please.
 Or any other suggestions for what might be different between our environments?

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