This is a repro case for a bug where if the client is sending
full throttle and the receiver is not able to keep up, the
client reaches a case where it doesn't make forward progress
anymore.
Signed-off-by: Craig Tiller <craig.tiller@gmail.com>
run_tests.py will start a server (if it's not running, or if the running
port server mismatches the 'current' one) that serves ports to use for
tests. The server is left running after run_tests.py finishes, so that
in environments such as Mac and Windows where tests run unshielded from
each other, we don't start jumping on already used ports.
Depending on the presence of pkg-config, we're not necessarily doing the right thing in the Makefile. Zlib is a pkg-config dependency, but libm, libpthread, librt aren't.
Currently, if two threads call grpc_completion_queue_pluck on the same
completion queue for different tags, there is a 50% chance that we
deliver the completion wakeup to the wrong poller - forcing the correct
poller to wait until its polling times out before it can return an event
up to the application.
This change tweaks our polling interfaces so that we can indeed wake a
specific poller.
Nothing has been performance tuned yet. It's definitely sub-optimal in a
number of places. Wakeup file-descriptors should be recycled. We should
have a path that avoids calling poll() followed by epoll(). We can
probably live without it right at the second though.
This code will fail on Windows at least (I'll do that port when I'm in the office and have a Windows
machine).