On Thu, 18 Sep 2014 15:36:39 -0500 Michael Roth <mdr...@linux.vnet.ibm.com> wrote:
> This series introduces visit_start_union and visit_end_union as a way > of allowing visitors to trigger generated code to bail out on visiting > union fields if the visitor implementation deems doing so to be unsafe. > > See patch 1 for the circumstances that cause the segfault in the > dealloc visitor. > > This is a spin-off of a patch submitted by Fam Zheng earlier. See the > thread for additional background on why we're taking this approach: > > http://thread.gmane.org/gmane.comp.emulators.qemu/296090 Applied to the qmp branch, thanks. PS: Michael, I'm assuming you were not planning to send a pull yourself. If that's not the case, please let me know. > > v3: > * fixed up commit msg in patch #2 (Eric) > * rebased on latest master > > v2: > * added comments to clarify true/false visit_start_union vs. errp > in dealloc visitor implementation (Eric) > * fixed a line that was >80 characters (Eric) > * fixed extra period in patch 2 commit msg (Eric) > * added comments to note reasons why the interface is useable for > dealloc visitor, but may require extra QAPI groundwork and > interface parameters to be useful for other types of visitors > * included Fam's qemu-iotest test case to exercise bug fix in a > user-triggerable scenario. > > include/qapi/visitor-impl.h | 2 ++ > include/qapi/visitor.h | 2 ++ > qapi/qapi-dealloc-visitor.c | 26 +++++++++++++++++++++++ > qapi/qapi-visit-core.c | 15 +++++++++++++ > scripts/qapi-visit.py | 6 ++++++ > tests/qapi-schema/qapi-schema-test.json | 10 +++++++++ > tests/qapi-schema/qapi-schema-test.out | 3 +++ > tests/qemu-iotests/087 | 17 +++++++++++++++ > tests/qemu-iotests/087.out | 13 ++++++++++++ > tests/test-qmp-input-strict.c | 17 +++++++++++++++ > 10 files changed, 111 insertions(+) >
