Commit Graph

33 Commits

Author SHA1 Message Date
Craig Tiller 27df2cf69c Pluck some low hanging concurrency fruit
Make the shutdown flag on servers be per thread to save contention on
the lock that must guard it.
2015-07-01 15:11:29 -07:00
Craig Tiller 51f938f169 Make async server use one CQ per server thread 2015-06-09 23:34:01 -07:00
Craig Tiller 250e96d941 Dont serialize new call requests under the server test lock 2015-06-08 09:51:43 -07:00
David Garcia Quintas c9516d4e28 Fixed memory leak in server_async 2015-06-02 13:41:48 -07:00
Craig Tiller 8ebcec484f clang-format 2015-05-21 13:55:22 -07:00
Craig Tiller aeea2f2203 Fix shutdown crash in async streaming test 2015-05-21 13:54:42 -07:00
Craig Tiller 5c8737d171 clang-format qps code - it was getting out of hand 2015-05-21 11:42:17 -07:00
Craig Tiller f9e6adf998 Completion queue binding for new requests API change
Move completion queue binding for new requests to the new request
request time, not server instantiation time.
2015-05-06 11:45:59 -07:00
Craig Tiller 66f37b477d Merge github.com:google/grpc into qps
Conflicts:
	Makefile
2015-04-10 08:28:31 -07:00
Nicolas "Pixel" Noble 0caebbfcfe Splitting gRPC service class codegen into its own set of files. 2015-04-09 23:08:51 +02:00
Craig Tiller 8221e40ac0 Indentation fixes, s/11/1/g 2015-04-09 12:23:21 -07:00
Vijay Pai 49673133b3 Remove unused header file for profiler 2015-03-31 12:39:51 -07:00
Vijay Pai 9032adcbef Fix conflicts 2015-03-26 10:32:52 -07:00
Vijay Pai 82dd80abc6 Elminate range-based fors and work around some other limitations in older compilers 2015-03-24 10:36:08 -07:00
Vijay Pai 1da729ad7d Merge branch 'qps-stream' of https://github.com/vjpai/grpc into vjpai-qps-stream 2015-03-23 12:52:56 -07:00
Vijay Pai 8ad32091e6 Make sure that nothing gets added to cq after shutdown 2015-03-23 12:44:28 -07:00
vjpai 46f65239cf Added streaming C++ tests for sync and sync cases 2015-03-23 10:10:27 -07:00
Jan Tattermusch 5e36d0d474 Merge pull request #1093 from nicolasnoble/windows
Various Windows fixes.
2015-03-19 10:29:47 -07:00
Nicolas Noble cfd6073a66 Various Windows fixes.
-) using dupenv_s instead of getenv_s and calling strdup ourselves.
-) few impossible-to-obtain if checks.
-) various signed/unsigned casting.
-) using time_t instead of time32_t
-) checking output of FormatMessage for failures.
-) don't redefine _WIN32_WINNT without undefining it first.
-) fixed msvc's interlocked casting.
-) renamed AddPort to AddListeningPort.
-) added protobuf's third_party includes to search path.
-) added a missing definition for inet_ntop in mingw32.
-) removed useless declarations.
2015-03-19 04:21:22 +01:00
Yang Gao 6baa9b67ad clang-format c++ code 2015-03-17 10:49:39 -07:00
Craig Tiller 68de8e91cc Merge github.com:grpc/grpc into credit 2015-03-05 15:45:46 -08:00
Craig Tiller a182bf12b0 clang-format 2015-03-04 13:54:39 -08:00
Craig Tiller d6479d6cc4 Async server works 2015-03-04 12:50:11 -08:00
Craig Tiller e7460e87d3 Merge github.com:grpc/grpc into credit 2015-03-03 13:48:40 -08:00
Vijay Pai acf6f318fc Better use of threads, avoid thread safety issues on destructor with
a proper join.

Also had been misusing EXPECT_EQ, as well as actually having an invalid
expectation on the ok field. Now it should be sane.
2015-03-02 15:13:39 -08:00
Craig Tiller 61691f92c3 Merge github.com:grpc/grpc into credit 2015-03-02 08:51:52 -08:00
vjpai 4e1e1bc28e Revert "Use typedefs to avoid triply-nested function templates"
This reverts commit 45b0bc4bec.

This revert is being done because the compilers on Travis don't
understand the C++11 template/using syntax.
2015-02-27 23:47:12 -08:00
vjpai 45b0bc4bec Use typedefs to avoid triply-nested function templates 2015-02-27 16:54:17 -08:00
vjpai 5b39f9a9fb refresh -> Reset 2015-02-27 09:33:00 -08:00
vjpai 6e2e64a8b4 Stop abusing operator() overloading 2015-02-27 09:28:28 -08:00
vjpai 3c11066217 override->GRPC_OVERRIDE 2015-02-27 07:17:16 -08:00
Vijay Pai 64ac47f389 clang-format all these files 2015-02-26 17:59:51 -08:00
vjpai dea740f329 New multithreaded async C++ tests. The server is architected the way
that it should be with multiple threads waiting on a single
completion queue.
The client currently uses a separate completion
queue per-thread, as trying to do a single unified queue was leading
to crashes for me. I need to figure that out.
2015-02-26 16:35:35 -08:00