----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/58512/#review172504 -----------------------------------------------------------
3rdparty/libprocess/src/decoder.hpp Lines 208 (patched) <https://reviews.apache.org/r/58512/#comment245597> hmm, these seem not related to the review i.e., avoiding the crash. Can you create a separate patch with these changes and mark this as dependent on it? Good catch! 3rdparty/libprocess/src/decoder.hpp Lines 715 (patched) <https://reviews.apache.org/r/58512/#comment245598> set `decoder->failed=true` here? If it's already set earlier, add an explicit invariant check here before this line? 3rdparty/libprocess/src/decoder.hpp Lines 1017 (patched) <https://reviews.apache.org/r/58512/#comment245600> Ditto as above. - Anand Mazumdar On April 19, 2017, 4:02 p.m., Anindya Sinha wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/58512/ > ----------------------------------------------------------- > > (Updated April 19, 2017, 4:02 p.m.) > > > Review request for mesos, Anand Mazumdar, James Peach, and Jiang Yan Xu. > > > Bugs: MESOS-7400 > https://issues.apache.org/jira/browse/MESOS-7400 > > > Repository: mesos > > > Description > ------- > > If the callback `on_headers_complete()` fails, the decoder's writer > object is `None()`. So, when the callback `on_message_complete()` is > called, we should not crash in that case. > Note that the `CHECK(decoder->writer)` is valid for the callback > `on_body()` is valid since this callback is called only on a success > in `on_headers_complete()` callback. > > > Diffs > ----- > > 3rdparty/libprocess/src/decoder.hpp > 4c779d42548958e610142438a57529ccb4478053 > 3rdparty/libprocess/src/tests/decoder_tests.cpp > c0efef571815752cc28e5a0dc7a07adc82bb31d5 > > > Diff: https://reviews.apache.org/r/58512/diff/2/ > > > Testing > ------- > > All tests passed. > > > Thanks, > > Anindya Sinha > >