On a Tuesday in 2020, John Ferlan wrote:
[...]
Coverity notes...
> @@ -1074,20 +1070,17 @@ qemuProcessHandleGraphics(qemuMonitorPtr mon G_GNUC_UNUSED,
> virDomainEventGraphicsSubjectPtr subject = NULL;
> size_t i;
>
> - if (VIR_ALLOC(localAddr) < 0)
> - goto error;
> + localAddr = g_new0(virDomainEventGraphicsAddress, 1);
> localAddr->family = localFamily;
> localAddr->service = g_strdup(localService);
> localAddr->node = g_strdup(localNode);
>
> - if (VIR_ALLOC(remoteAddr) < 0)
> - goto error;
> + remoteAddr = g_new0(virDomainEventGraphicsAddress, 1);
> remoteAddr->family = remoteFamily;
> remoteAddr->service = g_strdup(remoteService);
> remoteAddr->node = g_strdup(remoteNode);
>
> - if (VIR_ALLOC(subject) < 0)
> - goto error;
> + subject = g_new0(virDomainEventGraphicsSubject, 1);
> if (x509dname) {
> if (VIR_REALLOC_N(subject->identities, subject->nidentity+1) < 0)
> goto error;
There's no way to error: now w/o @localAddr, @remoteAddr, & @subject
being allocated, thus there's no need to check whether they're non-null
before accessing.
Actually, there's no way to error at all - VIR_REALLOC_N aborts on OOM.
I don't think it's worth deleting in the meantime.
Jano
John
[...]