Re: [PATCH 1/2] qdev: Separate implementations of qdev_get_machine() for user and system

[Thomas Huth <thuth@redhat.com>]

On 15/04/2021 21.07, Philippe Mathieu-Daudé wrote:
> On 4/15/21 6:56 PM, Greg Kurz wrote:
>> On Thu, 15 Apr 2021 18:45:45 +0200
>> Philippe Mathieu-Daudé <philmd@redhat.com> wrote:
>>
>>> On 4/15/21 3:30 PM, Greg Kurz wrote:
>>>> On Thu, 15 Apr 2021 14:39:55 +0200
>>>> Philippe Mathieu-Daudé <philmd@redhat.com> wrote:
>>>>
>>>>> On 4/9/21 6:03 PM, Greg Kurz wrote:
>>>>>> Despite its simple name and common usage of "getting a pointer to
>>>>>> the machine" in system-mode emulation, qdev_get_machine() has some
>>>>>> subtilities.
>>>>>>
>>>>>> First, it can be called when running user-mode emulation : this is
>>>>>> because user-mode partly relies on qdev to instantiate its CPU
>>>>>> model.
>>>>>>
>>>>>> Second, but not least, it has a side-effect : if it cannot find an
>>>>>> object at "/machine" in the QOM tree, it creates a dummy "container"
>>>>>> object and put it there. A simple check on the type returned by
>>>>>> qdev_get_machine() allows user-mode to run the common qdev code,
>>>>>> skipping the parts that only make sense for system-mode.
>>>>>>
>>>>>> This side-effect turns out to complicate the use of qdev_get_machine()
>>>>>> for the system-mode case though. Most notably, qdev_get_machine() must
>>>>>> not be called before the machine object is added to the QOM tree by
>>>>>> qemu_create_machine(), otherwise the existing dummy "container" object
>>>>>> would cause qemu_create_machine() to fail with something like :
>>>>>>
>>>>>> Unexpected error in object_property_try_add() at ../../qom/object.c:1223:
>>>>>> qemu-system-ppc64: attempt to add duplicate property 'machine' to
>>>>>>   object (type 'container')
>>>>>> Aborted (core dumped)
>>>>>>
>>>>>> This situation doesn't exist in the current code base, mostly because
>>>>>> of preventive fixing of some "latent bugs" in QEMU 4.0 (see 1a3ec8c1564
>>>>>> and e2fb3fbbf9c for details).
>>>>>>
>>>>>> A new kind of breakage was spotted very recently though :
>>>>>>
>>>>>> $ ./qemu-system-ppc64 -device power8_v2.0-spapr-cpu-core,help
>>>>>> /home/thuth/devel/qemu/include/hw/boards.h:24:
>>>>>>   MACHINE: Object 0x5635bd53af10 is not an instance of type machine
>>>>>> Aborted (core dumped)
>>>>>>
>>>>>> This comes from the change 3df261b6676b in QEMU 5.0. It unwillingly
>>>>>> added a new condition for qdev_get_machine() to be called too early,
>>>>>> breaking MACHINE(qdev_get_machine()) in generic cpu-core code this
>>>>>> time.
>>>>>>
>>>>>> In order to avoid further subtle breakages like this, change the
>>>>>> implentation of qdev_get_machine() to:
>>>>>> - keep the existing behaviour of creating the dummy "container"
>>>>>>    object for the user-mode case only ;
>>>>>> - abort() if the machine doesn't exist yet in the QOM tree for
>>>>>>    the system-mode case. This gives a precise hint to developpers
>>>>>>    that calling qdev_get_machine() too early is a programming bug.
>>>>>>
>>>>>> This is achieved with a new do_qdev_get_machine() function called
>>>>>> from qdev_get_machine(), with different implementations for system
>>>>>> and user mode.
>>>>>>
>>>>>> $ ./qemu-system-ppc64 -device power8_v2.0-spapr-cpu-core,help
>>>>>> qemu-system-ppc64: ../../hw/core/machine.c:1290:
>>>>>>   qdev_get_machine: Assertion `machine != NULL' failed.
>>>>>> Aborted (core dumped)
>>>>>>
>>>>>> Reported-by: Thomas Huth <thuth@redhat.com>
>>>>>> Signed-off-by: Greg Kurz <groug@kaod.org>
>>>>>> ---
>>>>>>   hw/core/machine.c        | 14 ++++++++++++++
>>>>>>   hw/core/qdev.c           |  2 +-
>>>>>>   include/hw/qdev-core.h   |  1 +
>>>>>>   stubs/meson.build        |  1 +
>>>>>>   stubs/qdev-get-machine.c | 11 +++++++++++
>>>>>>   5 files changed, 28 insertions(+), 1 deletion(-)
>>>>>>   create mode 100644 stubs/qdev-get-machine.c
>>>>> ...
>>>>>
>>>>>> diff --git a/stubs/meson.build b/stubs/meson.build
>>>>>> index be6f6d609e58..b99ee2b33e94 100644
>>>>>> --- a/stubs/meson.build
>>>>>> +++ b/stubs/meson.build
>>>>>> @@ -54,3 +54,4 @@ if have_system
>>>>>>   else
>>>>>>     stub_ss.add(files('qdev.c'))
>>>>>>   endif
>>>>>> +stub_ss.add(files('qdev-get-machine.c'))
>>>>>
>>>>> Adding this as a stub looks suspicious...
>>>>> Why not add it in to user_ss in hw/core/meson.build?
>>>>> Maybe name the new file hw/core/qdev-user.c?
>>>>>
>>>>
>>>> It turns out that this isn't specific to user-mode but rather
>>>> to any non-qemu-system-FOO binary built with qdev, e.g.
>>>> test-qdev-global-props :
>>>>
>>>> #0  do_qdev_get_machine () at ../../stubs/qdev-get-machine.c:10
>>>> #1  0x0000000100017938 in qdev_get_machine () at ../../hw/core/qdev.c:1134
>>>> #2  0x000000010001855c in device_set_realized (obj=0x100128b60, value=<optimized out>, errp=0x7fffffffd4e0) at ../../hw/core/qdev.c:745
>>>> #3  0x000000010001cc5c in property_set_bool (obj=0x100128b60, v=<optimized out>, name=<optimized out>, opaque=0x1000f33f0, errp=0x7fffffffd4e0) at ../../qom/object.c:2257
>>>> #4  0x0000000100020a9c in object_property_set (obj=0x100128b60, name=0x100093f78 "realized", v=0x100136d30, errp=0x1000e3af8 <error_fatal>) at ../../qom/object.c:1402
>>>> #5  0x000000010001c38c in object_property_set_qobject (obj=0x100128b60, name=0x100093f78 "realized", value=<optimized out>, errp=0x1000e3af8 <error_fatal>) at ../../qom/qom-qobject.c:28
>>>> #6  0x0000000100020e20 in object_property_set_bool (obj=0x100128b60, name=0x100093f78 "realized", value=<optimized out>, errp=0x1000e3af8 <error_fatal>) at ../../qom/object.c:1472
>>>> #7  0x000000010001612c in qdev_realize (dev=0x100128b60, bus=<optimized out>, errp=0x1000e3af8 <error_fatal>) at ../../hw/core/qdev.c:389
>>>> #8  0x000000010000fb10 in test_static_prop_subprocess () at /home/greg/Work/qemu/qemu-master/include/hw/qdev-core.h:17
>>>> #9  0x00007ffff7e95654 in g_test_run_suite_internal () from /lib64/libglib-2.0.so.0
>>>> #10 0x00007ffff7e954b8 in g_test_run_suite_internal () from /lib64/libglib-2.0.so.0
>>>> #11 0x00007ffff7e954b8 in g_test_run_suite_internal () from /lib64/libglib-2.0.so.0
>>>> #12 0x00007ffff7e954b8 in g_test_run_suite_internal () from /lib64/libglib-2.0.so.0
>>>> #13 0x00007ffff7e954b8 in g_test_run_suite_internal () from /lib64/libglib-2.0.so.0
>>>> #14 0x00007ffff7e959cc in g_test_run_suite () from /lib64/libglib-2.0.so.0
>>>> #15 0x00007ffff7e95a80 in g_test_run () from /lib64/libglib-2.0.so.0
>>>> #16 0x000000010000ecec in main (argc=<optimized out>, argv=<optimized out>) at ../../tests/unit/test-qdev-global-props.c:316
>>>>
>>>> Is there a meson thingy to handle this dependency ?
>>>
>>>    if not have_system
>>>      common_ss.add(files('qdev-machine-stubs.c'))
>>>    endif
>>>
>>> This is not pretty, but better than a generic stubs/qdev-get-machine.c
>>> IMO...
>>>
>>
>> Yeah it isn't that much different,
> 
> I'd expect symbols in stubs/ to be declared weak, while not with this
> approach, so we'd catch clash for incorrect configs. Maybe I'm wrong...

See https://gitlab.com/qemu-project/qemu/-/commit/3bc2f570ec9fc93 for the 
reason why symbols in stubs are *not* declared as weak anymore.

  Thomas