monitor: Fix find_device_state() for IDs containing slashespull-monitor-2021-11-10
commit1bf4d3294bd48b702530b131e3344860495425fd
authorMarkus Armbruster <armbru@redhat.com>
Tue, 19 Oct 2021 08:57:11 +0000 (19 10:57 +0200)
committerMarkus Armbruster <armbru@redhat.com>
Wed, 10 Nov 2021 05:14:51 +0000 (10 06:14 +0100)
tree96d79ed8d3bcf780ce4df4e61552b616b81437c9
parent0a70bcf18caf7a61d480f8448723c15209d128ef
monitor: Fix find_device_state() for IDs containing slashes

Recent commit 6952026120 "monitor: Tidy up find_device_state()"
assumed the function's argument is "the device's ID or QOM path" (as
documented for device_del).  It's actually either an absolute QOM
path, or a QOM path relative to /machine/peripheral/.  Such a relative
path is a device ID when it doesn't contain a slash.  When it does,
the function now always fails.  Broke iotest 200, which uses relative
path "vda/virtio-backend".

It fails because object_resolve_path_component() resolves just one
component, not a relative path.

The obvious function to resolve relative paths is
object_resolve_path().  It picks a parent automatically.  Too much
magic, we want to specify the parent.  Create new
object_resolve_path_at() for that, and use it in find_device_state().

Reported-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20211019085711.86377-1-armbru@redhat.com>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
include/qom/object.h
qom/object.c
softmmu/qdev-monitor.c