On 10/21/2016 12:01 PM, Pavel Hrdina wrote: > On Fri, Oct 21, 2016 at 09:01:42AM -0400, John Ferlan wrote: >> Rather than VIR_ALLOC of the virDomainChrDefPtr >> >> Signed-off-by: John Ferlan <jferlan@xxxxxxxxxx> >> --- >> src/vmx/vmx.c | 4 ++-- >> 1 file changed, 2 insertions(+), 2 deletions(-) >> >> diff --git a/src/vmx/vmx.c b/src/vmx/vmx.c >> index fc4347f..f61c4d6 100644 >> --- a/src/vmx/vmx.c >> +++ b/src/vmx/vmx.c >> @@ -2758,7 +2758,7 @@ virVMXParseSerial(virVMXContext *ctx, virConfPtr conf, int port, >> return -1; >> } >> >> - if (VIR_ALLOC(*def) < 0) >> + if (!(*def = virDomainChrDefNew(NULL))) >> return -1; >> >> (*def)->deviceType = VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL; >> @@ -2946,7 +2946,7 @@ virVMXParseParallel(virVMXContext *ctx, virConfPtr conf, int port, >> return -1; >> } >> >> - if (VIR_ALLOC(*def) < 0) >> + if (!(*def = virDomainChrDefNew(NULL))) >> return -1; >> >> (*def)->deviceType = VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL; >> -- > > You've missed those two occurrences: > src/conf/domain_conf.c:3876 if (VIR_ALLOC(def->consoles[0]) < 0) > src/conf/domain_conf.c:3891 if (VIR_ALLOC(chr) < 0) > Thanks - I added them... I tried scouring all virDomainChrDef[Ptr] - I probably just had the overwhelming power of assumption that domain_conf.c would have used virAllocChrDefNew properly. The *-aa-* were certainly one of those environments in the 5% from my cover... Those will have to be (NULL) calls though which I would think be correct since they're going to be implicitly added and they wouldn't be needing a TLS backend setup. Not clear if there was an implicit ACK here or not, so I'll wait to push since it's not ultra critical to get this in. John > I've found those because virt-aa-helper-test was crashing after applying the > next patch. Because this is a generic code for all drivers we need to pass > xmlopt to allocate the privateData. > > Pavel > -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list