On 26/04/2018 14:39, Paolo Bonzini wrote: > On 26/04/2018 14:11, Olaf Hering wrote: >> Am Thu, 26 Apr 2018 13:10:02 +0100 >> schrieb Daniel P. Berrangé <berra...@redhat.com>: >> >>> Presumably the min-glib docker job didn't enable the vfio feature, which >>> is what contained the bad code. >> >> It is Linux=y, every following commit would have failed too. >> I suggest to revisit that testcase. > > That's because Red Hat backported g_realloc_n into RHEL6 :) probably > because it's used for security reasons and they expected users to appear > in the next 10-ish years. > > https://bugzilla.redhat.com/show_bug.cgi?id=573966
Almost. It's because the g_*alloc_n was introduced in order to fix overflows in g_*new. When Red Hat fixed the overflows, that implicitly added those functions. Paolo
