[libvirt] [PATCH] build: avoid type-punning in vbox

[Matthias Bolte]

Am 20. April 2012 08:53 schrieb Matthias Bolte <matthias.bolte at googlemail.com>:
> Am 20. April 2012 01:36 schrieb Eric Blake <eblake at redhat.com>:
>> Commit 78345c68 makes at least gcc 4.1.2 on RHEL 5 complain:
>>
>> cc1: warnings being treated as errors
>> In file included from vbox/vbox_V4_0.c:13:
>> vbox/vbox_tmpl.c: In function 'vboxDomainUndefineFlags':
>> vbox/vbox_tmpl.c:5298: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
>>
>> * src/vbox/vbox_tmpl.c (vboxDomainUndefineFlags): Use union to
>> avoid compiler warning.
>> ---
>>
>> Pushing this under the build-breaker rule.
>>
>>  src/vbox/vbox_tmpl.c |    7 +++++--
>>  1 files changed, 5 insertions(+), 2 deletions(-)
>>
>> diff --git a/src/vbox/vbox_tmpl.c b/src/vbox/vbox_tmpl.c
>> index be25828..57c18a4 100644
>> --- a/src/vbox/vbox_tmpl.c
>> +++ b/src/vbox/vbox_tmpl.c
>> @@ -5294,8 +5294,11 @@ vboxDomainUndefineFlags(virDomainPtr dom, unsigned int flags)
>>
>>         ((IMachine_Delete)machine->vtbl->Delete)(machine, &safeArray, &progress);
>>  # else
>> -        vboxArray array = VBOX_ARRAY_INITIALIZER;
>> -        machine->vtbl->Delete(machine, 0, (IMedium**)&array, &progress);
>> +        union {
>> +            vboxArray array;
>> +            IMedium *medium;
>> +        } u = { .array = VBOX_ARRAY_INITIALIZER };
>> +        machine->vtbl->Delete(machine, 0, &u.medium, &progress);
>>  # endif
>>         if (progress != NULL) {
>>             progress->vtbl->WaitForCompletion(progress, -1);
>
> Actually, NACK. As stated in the other mail, vboxArray is not castable
> to IMedium*. True, it silences the compiler any might even work by
> accident on XPCOM because VirtualBox might not touch the pointer
> beyond checking it for being non-NULL because the 0 tells it that it's
> an empty array. But still this is wrong and might crash on MSCOM.
>
> I might come up with a proper solution tomorrow.

Okay, looking at this in a bigger context shows that we already have a
special case in place for this call for MSCOM that doesn't use
vboxArray, because I didn't want to add a full vboxArray solution for
this single case.

XPCOM doesn't like being pass NULL for an array even if it's an empty
array, according to Jean-Baptiste (I can not reproduce the problem).
Anyway, I think the proper special case for this is to just a dummy
IMedium* array not involving vboxArray at all.

        IMedium *array[] = { NULL };
        machine->vtbl->Delete(machine, 0, array, &progress);

See this patch

https://www.redhat.com/archives/libvir-list/2012-April/msg01211.html

-- 
Matthias Bolte
http://photron.blogspot.com