core: speed up Status code and message parsing

[carl-mastrangelo]

This introduces the idea of a "Trusted" Ascii Marshaller, which is
known to always produce valid ASCII byte arrays. This saves a
surprising amount of garbage, since String conversion involves
creating a new java.lang.StringCoding, and a sun.nio.cs.US_ASCII.

There are other types that can be converted (notably
Http2ClientStream's :status marshaller, which is particularly
wasteful).

Before:

Benchmark                              Mode     Cnt     Score    Error  Units
StatusBenchmark.codeDecode           sample  641278    88.889 ±  9.673  ns/op
StatusBenchmark.codeEncode           sample  430800    73.014 ±  1.444  ns/op
StatusBenchmark.messageDecodeEscape  sample  433467   441.078 ± 58.373  ns/op
StatusBenchmark.messageDecodePlain   sample  676526   268.620 ±  7.849  ns/op
StatusBenchmark.messageEncodeEscape  sample  547350  1211.243 ± 29.907  ns/op
StatusBenchmark.messageEncodePlain   sample  419318   223.263 ±  9.673  ns/op

After:

Benchmark                              Mode     Cnt    Score    Error  Units
StatusBenchmark.codeDecode           sample  442241   48.310 ±  2.409  ns/op
StatusBenchmark.codeEncode           sample  622026   35.475 ±  0.642  ns/op
StatusBenchmark.messageDecodeEscape  sample  595572  312.407 ± 15.870  ns/op
StatusBenchmark.messageDecodePlain   sample  565581   99.090 ±  8.799  ns/op
StatusBenchmark.messageEncodeEscape  sample  479147  201.422 ± 10.765  ns/op
StatusBenchmark.messageEncodePlain   sample  560957   94.722 ±  1.187  ns/op

Also fixes #2237

cc: @ejona86 for the fairly complex parsing logic.

[grpc-kokoro]

Thanks for your pull request. The automated tests will run as soon as one of the admins verifies this change is ok to test.

[buchgr]

@grpc-jenkins ok to test

[carl-mastrangelo]

Updated the code to show how to share this across the io.grpc.internal boundary. The status line parser also shows up in the allocation profiler.

[carl-mastrangelo]

One last thing: The difference here is noticeable:

Before:
Benchmark                         (direct)  (transport)    Mode      Cnt       Score     Error  Units
TransportBenchmark.unaryCall1024      true        NETTY  sample  3205966  155710.268 ± 149.278  ns/op

After:
Benchmark                         (direct)  (transport)    Mode      Cnt       Score     Error  Units
TransportBenchmark.unaryCall1024      true        NETTY  sample  3385015  147474.794 ± 128.733  ns/op

I included the full numbers in the second commit message, but this better. The 99% latency goes down by 25us!

[buchgr]

@carl-mastrangelo wow! very nice! 6% faster! I ll have a look at it tomorrow morning . it's almost midnight here 😴 :)

[buchgr]

looking now ... sorry for the delay ...

[carl-mastrangelo]

FYI Our internal visibility rules center mostly around files, so restricted actions can be put in a file that can be protected in ways the java type system can't. Open source doesn't have this, so they just have to read the comments.

[buchgr]

why not put this in the internal package?

[carl-mastrangelo]

Because this has to be in the public package as a bridge between internal and public. It accesses the package protected methods in io.grpc.

[buchgr]

I thought that was only necessary if it accessed public methods in io.grpc?

[carl-mastrangelo]

It does. The Metadata.Key.of is package protected and has a special implementation for TrustedAsciiKey (which pairs with TrustedAsciiMarshaller in this class). io.grpc.internal needs access to this package because of HTTP_STATUS_MARSHALLER (as well as others).

Finally, all the shortcutting work between transport headers and Metadata are going to be don in this class.

Also, all the other

[buchgr]

LGTM, once remaining comments are addressed and build passes. Good work @carl-mastrangelo !

[carl-mastrangelo]

Fixed the fallthrough

[carl-mastrangelo]

Thanks for taking the time @buchgr !