Re: [RESEND PATCH V10 00/15] iothread: Support tracking and querying IOThread holders

On Tue, Jul 21, 2026 at 8:47 PM Markus Armbruster <[email protected]> wrote:
>
> With this series applied:
>
>     $ qemu-system-x86_64 -S -display none -chardev id=chr-mon0,backend=stdio -object monitor-hmp,id=mon0,chardev=chr-mon0 -chardev id=chr-mon1,backend=socket,path=test-hmp,server=on,wait=off -object monitor-qmp,id=mon1,chardev=chr-mon1
>     QEMU 11.0.90 monitor - type 'help' for more information
>     (qemu) q
>     **
>     ERROR:../iothread.c:99:iothread_unref: code should not be reached
>     Bail out! ERROR:../iothread.c:99:iothread_unref: code should not be reached
>     Aborted                    (core dumped) qemu-system-x86_64 -S -display none -chardev id=chr-mon0,backend=stdio -object monitor-hmp,id=mon0,chardev=chr-mon0 -chardev id=chr-mon1,backend=socket,path=test-hmp,server=on,wait=off -object monitor-qmp,id=mon1,chardev=chr-mon1
>
> Stack backtrace:
>
>     #0  0x00007f182873832c in __pthread_kill_implementation () at /lib64/libc.so.6
>     #1  0x00007f18286de15e in raise () at /lib64/libc.so.6
>     #2  0x00007f18286c56d0 in abort () at /lib64/libc.so.6
>     #3  0x00007f182ac53482 in g_assertion_message[cold] ()
>         at /lib64/libglib-2.0.so.0
>     #4  0x00007f182acc1447 in g_assertion_message_expr ()
>         at /lib64/libglib-2.0.so.0
>     #5  0x000055ffa06cb50e in iothread_unref
>         (iothread=0x55ffdb3bd740, holder=0x7ffc6152d8c0) at ../iothread.c:99
>     #6  0x000055ffa06cc13a in iothread_put_aio_context
>         (iothread=0x55ffdb3bd740, holder=0x7ffc6152d8c0) at ../iothread.c:444
>     #7  0x000055ffa07d9af6 in monitor_cleanup () at ../monitor/monitor.c:679
>     #8  0x000055ffa047ca8c in qemu_cleanup (status=0) at ../system/runstate.c:1046
>     #9  0x000055ffa07fccb7 in qemu_default_main (opaque=0x0) at ../system/main.c:51
>     #10 0x000055ffa07fcd64 in main (argc=12, argv=0x7ffc6152da98)
>         at ../system/main.c:93
>

Oh~ Thanks for the hmp and qmp monitor case test. I didn't trigger it
in the basic monitor test.
The root cause is monitor_cleanup unconditionally unregisters all monitors
if the global mon_iothread exists. I will fix it in next version.

Thanks
Chen