gdbstub debug of multi-cluster machines is undocumented and confusing

[Martin Schönstedt]

Working with Zephyr RTOS, running a multi core sample on mps2_an521 works fine. Both cpus start.
Trying to debug with options -s -S the second core fails to boot.

Posted with explanation also at: https://github.com/zephyrproject-rtos/zephyr/issues/33635

[Martin Schönstedt (m-schoenstedt)]

there was no bug, it was my fault. How do I delete this

[Peter Maydell (pmaydell)]

In general, you don't need to delete bugs that turn out to be user error, or edit the description/title; just mark them as 'invalid', perhaps with a comment about what turned out to be the cause. That leaves the trail of what was going on for future readers who might be going down the same path as you.

There are actually a couple of things we should do here in upstream QEMU:

* we should document the process for using the debugstub with multi-cluster board models like the mps2-an521
* we should check whether we are doing the right/most appropriate thing when the user connects to the debug stub and is only attaching to one 'inferior' -- it sounds from your report like the un-attached inferior is left permanently in the 'stopped' state. Maybe that's what the gdb protocol requires, but it seems a bit unhelpful.

I'm going to update the bug status/text accordingly.

[Martin Schönstedt (m-schoenstedt)]

Thanks for the answer,

indeed the second cluster of the board has been halted when I was starting gdb the "normal" way - not adding the second inferior. In my own research I did not find out about these inferiors, so I was wondering why "info threads" did only show one cpu. Maybe gdb could inform the user about unattached inferiors when using "info threads"

[Peter Maydell (pmaydell)]

Could you provide an example QEMU command line and guest image file which you were having this problem with, please? That would save me figuring out how to build zephyr binaries :-)

[Martin Schönstedt (m-schoenstedt)]

qemu-system-arm -cpu cortex-m33 -machine mps2-an521 -nographic -m 16 -vga none -net none -chardev stdio,id=con,mux=on -serial chardev:con -mon chardev=con,mode=readline -icount shift=7,align=off,sleep=off -rtc clock=vm -device loader,file=<remote_dir>/zephyr.elf -s -S -kernel <master_dir>/zephyr.elf

i've included both .elf files in a zip. That should be enough for running the sample

[Peter Maydell (pmaydell)]

There's now a new docs section on debugging multicore (including multi-cluster) machines:

https://qemu-project.gitlab.io/qemu/system/gdb.html#debugging-multicore-machines

[Thomas Huth (th-huth)]

Is there still anything to do here or could we close the ticket now?

[Martin Schönstedt (m-schoenstedt)]

It can be closed. The added documentation is very helpful.

[Thomas Huth (th-huth)]

Ok, thanks, so I'm closing this ticket now.