* Tooling to remove redundant grpc_core:: namespaces
These references tend to show up in our C++ code after C modules get
converted. Many get caught in review, many get missed.
* use it
* clang-format
* Check if memory owner available prior to polling it
The transport may drop the memory owner during its destruction sequence
* tcp_fix
* Revert "Revert "New resource quota integration (#27643)" (#28014)"
This reverts commit 0ea2c37263.
* clang-format
* fix-path
* fix
The urgent argument is a platform-specific flag that leaked into the (ideally) platform-independent HTTP/2 transport layer. In an effort to clean up the cross-platform API surface, it would be helpful if we can remove this argument from the TCP Read api without losing the performance optimization that was introduced along with it (see #18240).
TCP_INQ is a socket option we added to Linux to report pending bytes
on the socket as a control message.
Using TCP_INQ we can accurately decide whether to continue read or not.
Add an urgent parameter, when we do not want to wait for EPOLLIN.
This commit improves the latency of 1 RPC unary (minimal benchmark)
significantly:
Before:
l_50: 61.3584984733
l_90: 94.8328711277
l_99: 126.211351174
l_999: 158.722406029
After:
l_50: 51.3546011488 (-16%)
l_90: 72.3420731581 (-23%)
l_99: 103.280218974 (-18%)
l_999: 130.905689996 (-17%)
Passing grpc_slice by value and/or returning it can be very costly,
introducing many extra instructions to push the structure to the
stack and poping it.
This CL, wherever possible, changes grpc_slice to be passed by
value.
On a local benchmark, I obserse 4-7% improvements in latency and QPS.
There are still copies to the slice_ref vtable which @arjunroy
is fixing as part of his major effort to use grpc_core::RefCount
for slices and devirtualizing them.
src/core. exec_ctx is now a thread_local pointer of type ExecCtx instead of
grpc_exec_ctx which is initialized whenever ExecCtx is instantiated. ExecCtx
also keeps track of the previous exec_ctx so that nesting of exec_ctx is
allowed. This means that there is only one exec_ctx being used at any
time. Also, grpc_exec_ctx_finish is called in the destructor of the
object, and the previous exec_ctx is restored to avoid breaking current
functionality. The code still explicitly calls grpc_exec_ctx_finish
because removing all such instances causes the code to break.
Craig Tiller
Esun Kim
AJ Heller
Mark D. Roth
Yash Tibrewal
Vijay Pai
Soheil Hassas Yeganeh
Jan Tattermusch
Alexander Polcyn
Noah Eisen
Muxi Yan
ncteisen
Jan Tattermusch