* Benjamin Baier <program...@netzbasis.de> [2020-01-02 22:01:14 +0100]:
On Thu, 2 Jan 2020 18:56:14 +0100
Klemens Nanni <k...@openbsd.org> wrote:
On Thu, Jan 02, 2020 at 04:37:17PM +0100, Benjamin Baier wrote:
> "case CMD_SEND:" sets done=1 so ret will never be written to and
> the uninitialized value of ret is used to determine the return
> value of the function vmmaction.
Good catch:
$ doas vmctl start -b ~/vm/bsd.rd -m 128M test ; echo $?
vmctl: starting without disks
vmctl: starting without network interfaces
vmctl: started vm 4 successfully, tty /dev/ttyp2
0
$ doas vmctl send test >/dev/null ; echo $?
vmctl: sent vm test successfully
1
With your diff it exits zero.
I also just noticed that above example reproducibly causes vmd(8) to
exit:
Jan 2 18:53:57 eru vmd[55128]: startup
Jan 2 18:54:18 eru vmd[55128]: test: started vm 4 successfully, tty /dev/ttyp2
Jan 2 18:54:28 eru vmd[49983]: priv exiting, pid 49983
Jan 2 18:54:28 eru vmd[29885]: control exiting, pid 29885
I don't get vmd to exit, but got this in /var/log/messages with the above
example:
Jan 2 21:38:05 x220 vmd[86810]: control_dispatch_vmd: lost control connection:
fd 7
Hi Benjamin,
kn@ is running with vm.malloc_conf=SU. The crash is due to vm_remove
calling free on vm here,
https://github.com/openbsd/src/blob/926f477f07d3bba063ff6ee1ea9e0b7369ed8930/usr.sbin/vmd/vmm.c#L542
will work on a fix.
--
Pratik