[libvirt] [PATCH] qemu: Fix compilation error when enum variable size differs from 'int'

[Peter Krempa]

On Mon, May 25, 2015 at 18:28:42 +0200, Michal Privoznik wrote:
> On 25.05.2015 17:47, Peter Krempa wrote:
> > On Mon, May 25, 2015 at 17:35:19 +0200, Michal Privoznik wrote:
> >> On 25.05.2015 17:18, Peter Krempa wrote:
> >>> Since commit bcd9a564b631aa virDomainNumatuneGetMode returns the value
> >>> via a pointer rather than in the return value. The change triggered
> >>> problems with platforms where the compiler decides to use a data type of
> >>> size different than integer at the point where we typecast it.
> >>>
> >>> Work around the issue by using an intermediate variable of the correct
> >>> type that gets casted back by the default typecasting rules.
> >>> ---
> >>>  src/qemu/qemu_driver.c | 10 ++++++----
> >>>  1 file changed, 6 insertions(+), 4 deletions(-)
> >>>
> >>> diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c
> >>> index aa0acde..2b9f125 100644
> >>> --- a/src/qemu/qemu_driver.c
> >>> +++ b/src/qemu/qemu_driver.c
> >>> @@ -10566,14 +10566,16 @@ qemuDomainGetNumaParameters(virDomainPtr dom,
> >>>          virMemoryParameterPtr param = &params[i];
> >>>
> >>>          switch (i) {
> >>> -        case 0: /* fill numa mode here */
> >>> +        case 0: {  /* fill numa mode here */
> >>> +            virDomainNumatuneMemMode tmp;
> >>> +            virDomainNumatuneGetMode(def->numa, -1, &tmp);
> >>> +
> >>>              if (virTypedParameterAssign(param, VIR_DOMAIN_NUMA_MODE,
> >>> -                                        VIR_TYPED_PARAM_INT, 0) < 0)
> >>> +                                        VIR_TYPED_PARAM_INT, tmp) < 0)
> >>>                  goto cleanup;
> >>>
> >>> -            virDomainNumatuneGetMode(def->numa, -1,
> >>> -                                     (virDomainNumatuneMemMode *) &param->value.i);
> >>>              break;
> >>> +        }
> >>>
> >>>          case 1: /* fill numa nodeset here */
> >>>              nodeset = virDomainNumatuneFormatNodeset(def->numa, NULL, -1);
> >>>
> >>
> >> I guess we saw the same coverity report since you're fixing the code I'm
> >> just looking at. But I think what is coverity trying to tell us is that
> >> in all other places virDomainNumatuneGetMode() is checked for return
> >> value, here it's not. Or did you really see problem  you're describing
> >> in the commit message?
> > 
> > The problem I'm trying to solve is failure to compile libvirt on CentOS
> > 5. The compilation error can be seen at:
> > 
> > https://ci.centos.org/view/libvirt-project/job/libvirt-daemon-build/systems=libvirt-centos-5/259/console
> > 
> > if you scroll down enough.
> > 
> 
> Ah, okay; So ACK. Lets see what will Coverity think after you push this.

Ah, I see what coverity might complain about. The value will not be
initialized if neither the specific nor the default mode is specified in
the XML. I'll send a v2 fixing both the coverity issue and the
compilation problem since this one would not fix coverity as it would be
possible to have @tmp uninitialized.

Peter
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://listman.redhat.com/archives/libvir-list/attachments/20150526/dc6a9150/attachment-0001.sig>