> On Nov. 26, 2019, 1:59 p.m., Greg Mann wrote: > > 3rdparty/libprocess/src/ssl/socket_wrapper.cpp > > Lines 212-215 (patched) > > <https://reviews.apache.org/r/71664/diff/3/?file=2174350#file2174350line212> > > > > I wonder if openssl's internal calls to `BIO_flush()` will retry if the > > call fails... perhaps we could synchronously return zero here if there are > > pending writes, and return 1 if everything has been written?
No. If this call returns 0 (or anything less than 1), the SSL library will consider this a `SSL_ERROR_SYSCALL` and will refuse to make any further progress, thereby shutting down the socket. I haven't found any way around this, but I also have not seen any consequences from lying about the flush either. - Joseph ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/71664/#review218804 ----------------------------------------------------------- On Nov. 19, 2019, 12:16 p.m., Joseph Wu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/71664/ > ----------------------------------------------------------- > > (Updated Nov. 19, 2019, 12:16 p.m.) > > > Review request for mesos, Benno Evers, Benjamin Mahler, Greg Mann, and Till > Toenshoff. > > > Bugs: MESOS-10009 > https://issues.apache.org/jira/browse/MESOS-10009 > > > Repository: mesos > > > Description > ------- > > This implements the OpenSSL basic I/O abstraction based on the > libprocess event loop. This BIO wraps a socket and handles the > reading/writing, using io::read and io::write. > > This BIO can be passed into an SSL context to enable usage of > SSL translation functions like SSL_read and SSL_write. > > > Diffs > ----- > > 3rdparty/libprocess/src/ssl/socket_wrapper.cpp PRE-CREATION > > > Diff: https://reviews.apache.org/r/71664/diff/3/ > > > Testing > ------- > > cmake --build . --target process > > A tiny bit of testing next patch. > > > Thanks, > > Joseph Wu > >
