This series is a follow-up to David's "Better handling of machine specific per-cpu information" v3 patchset. It addresses issues mentioned at:
https://lists.nongnu.org/archive/html/qemu-ppc/2018-06/msg00514.html This series fixes potential crashes and leaks in case of error during CPU hotplug. It hence assumes that the machine_data pointer is set a bit earlier as suggested at the above URL, for CPU hotplug to work. Patch 1 is just code cleanup. Patch 2 fixes a 2.12 regression. Patch 3 fixes a long standing issue. It could possibly be fixed in current master but this would require to rebase David's series on top of it. Not sure it's worth the pain. Patch 4 and 5 are some more cleanup. -- Greg --- Greg Kurz (5): spapr_cpu_core: convert last snprintf() to g_strdup_printf() spapr_cpu_core: fix potential leak in spapr_cpu_core_realize() spapr_cpu_core: add missing rollback on realization path spapr_cpu_core: introduce spapr_create_vcpu() spapr_cpu_core: simplify spapr_cpu_core_realize() hw/ppc/spapr_cpu_core.c | 93 ++++++++++++++++++++++++++++++----------------- 1 file changed, 60 insertions(+), 33 deletions(-)
