Commits · f5752848fde774c5c16c5e58f15558a253a03119 · go / golang

16 May, 2012 6 commits

pkg/runtime: Fix semasleep on Plan 9 · f5752848

Akshat Kumar authored May 16, 2012

With the timed semacquire patch
(kernel-tsemacquire) for Plan 9,
we can now properly do a timed
wait for the semaphore, in
semasleep.

R=golang-dev, rsc, rminnich, ality, r
CC=0intro, golang-dev, john, mirtchovski
https://golang.org/cl/6197046

f5752848

misc/chrome/gophertool: lower CL number's lower bound and fix input focus on Mac · 6dfe0162

Shenghou Ma authored May 16, 2012

As our CL number could be as small as 152046, changed CL number's lower bound
to 150000. Hopefully our issue count won't reach 150000 any time soon.
Chrome on Mac OS X has very strange behavior regarding the focus, although we
force the focus to the input box on load, the page still come up with focus on
"issue". Set the tabindex of the input box as a workaround.

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/6212055

6dfe0162

cmd/5c: re-enable regopt() · 70db4408

Shenghou Ma authored May 16, 2012

        After CL 6185047, ./all.bash passed.

benchmark                       old ns/op    new ns/op    delta
BenchmarkAppend                      5558         4894  -11.95%
BenchmarkAppendSpecialCase           5242         4572  -12.78%
BenchmarkSelectUncontended           3719         2821  -24.15%
BenchmarkSelectContended             3776         2832  -25.00%
BenchmarkSelectNonblock              1030         1089   +5.73%
BenchmarkChanUncontended              530          422  -20.38%
BenchmarkChanContended                534          444  -16.85%
BenchmarkChanSync                    1613         1492   -7.50%
BenchmarkChanProdCons0               1520         1351  -11.12%
BenchmarkChanProdCons10               785          668  -14.90%
BenchmarkChanProdCons100              564          473  -16.13%
BenchmarkChanProdConsWork0          11205        10337   -7.75%
BenchmarkChanProdConsWork10          9806         9567   -2.44%
BenchmarkChanProdConsWork100         9413         9398   -0.16%
BenchmarkChanCreation               11687         8715  -25.43%
BenchmarkChanSem                      553          453  -18.08%
BenchmarkCallClosure                   22           22   +0.44%
BenchmarkCallClosure1                  28           28   +0.71%
BenchmarkCallClosure2                2224         1668  -25.00%
BenchmarkCallClosure3                2217         1629  -26.52%
BenchmarkCallClosure4                2240         1684  -24.82%
BenchmarkComplex128DivNormal          930          912   -1.94%
BenchmarkComplex128DivNisNaN          862          866   +0.46%
BenchmarkComplex128DivDisNaN          849          852   +0.35%
BenchmarkComplex128DivNisInf          556          583   +4.86%
BenchmarkComplex128DivDisInf          522          512   -1.92%
BenchmarkConvT2E                      175          159   -9.14%
BenchmarkConvT2EBig                  2418         1823  -24.61%
BenchmarkConvT2I                      545          549   +0.73%
BenchmarkConvI2E                       35           32   -9.58%
BenchmarkConvI2I                      404          391   -3.22%
BenchmarkAssertE2T                     75           62  -16.25%
BenchmarkAssertE2TBig                  76           63  -16.80%
BenchmarkAssertE2I                    427          409   -4.22%
BenchmarkAssertI2T                     82           66  -20.29%
BenchmarkAssertI2I                    430          416   -3.26%
BenchmarkAssertI2E                     36           32  -12.50%
BenchmarkAssertE2E                     35           35   +0.57%
BenchmarkFinalizer                   3224         2941   -8.78%
BenchmarkFinalizerRun              117392        84772  -27.79%
BenchmarkStackGrowth                 5267         5930  +12.59%
BenchmarkSyscall                      191          167  -12.57%
BenchmarkSyscallWork                 9918         7713  -22.23%
BenchmarkIfaceCmp100                 1645         1652   +0.43%
BenchmarkIfaceCmpNil100              1433         1440   +0.49%

R=dave, rsc
CC=golang-dev
https://golang.org/cl/6202070

70db4408

codereview: support mercurial 2.2.1 · dbf6215d

Shenghou Ma authored May 16, 2012

        We explicitly use plainformatter to avoid the
        user's debug setting changing our behavior.
        Fixes #3603.

R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6201069

dbf6215d

log: fix typo in comment · c816d8c5
Benny Siegert authored May 16, 2012
```
R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/6195096
```
c816d8c5

doc/install: document minimum system requirements · ab708e35

Shenghou Ma authored May 16, 2012

        Partly copied from CL 5685076.

R=golang-dev, adg, r
CC=golang-dev
https://golang.org/cl/6112064

ab708e35

15 May, 2012 14 commits

test: add bug439, valid code that caused a gccgo crash · 6f6bbdf9
Ian Lance Taylor authored May 15, 2012
```
R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/6198075
```
6f6bbdf9
go/printer: don't print newlines for empty statements · 03e1d4bc
Robert Griesemer authored May 15, 2012
```
Fixes #3466.

gofmt -w src misc causes no changes.

R=rsc
CC=golang-dev
https://golang.org/cl/6206073
```
03e1d4bc

runtime: implement getcontext and sigprocmask for netbsd · efa67b2c

Joel Sing authored May 15, 2012

Implement getcontext and sigprocmask for NetBSD - these will soon be
used by the thread handling code.

Also fix netbsd/386 signal handling - there is no sigreturn, just
return so that we hit the trampoline.

R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6215049

efa67b2c

runtime: fix netbsd syscalls · 0b7bcb84

Joel Sing authored May 15, 2012

Use correct system calls/syscall numbers for NetBSD.

R=golang-dev, for.go.yong, rsc
CC=golang-dev
https://golang.org/cl/6200070

0b7bcb84

cmd/go: invoke godoc with import path when possible · 0c8f4694

Russ Cox authored May 15, 2012

Also add -n -x flags to doc, fmt, vet.
Also shorten unknown command error.

Fixes #3612.
Fixes #3613.

R=golang-dev, bradfitz, r
CC=golang-dev
https://golang.org/cl/6211053

0c8f4694

8l,5l: 2% faster · f47346c5

Jan Ziak authored May 15, 2012

R=golang-dev, for.go.yong
CC=golang-dev
https://golang.org/cl/6197080

f47346c5

cmd/gc: make append(nil, x) error more precise · fcc1f2ac

Russ Cox authored May 15, 2012

Before:
./x.go:6: first argument to append must be slice; have nil

After:
./x.go:6: first argument to append must be typed slice; have untyped nil

Fixes #3616.

R=ken2
CC=golang-dev
https://golang.org/cl/6209067

fcc1f2ac

runtime: fix netbsd signal handling · 4a5a5b20

Joel Sing authored May 15, 2012

Update/correct NetBSD signal handling - most of this is needed due to
the correctly generated runtime definitions.

R=golang-dev, m4dh4tt3r, rsc
CC=golang-dev
https://golang.org/cl/6195079

4a5a5b20

runtime: fix netbsd runtime defs · 8a9edcf7

Joel Sing authored May 15, 2012

Fix and regenerate runtime defs for NetBSD.

Whilst the mcontext struct can be handled across architectures,
the registers are provided as defines that index an array, rather
than as members of the struct. Since these are architecture
dependent, include them via a defs_netbsd_<arch>.go file.

R=golang-dev, m4dh4tt3r, rsc
CC=golang-dev
https://golang.org/cl/6190070

8a9edcf7

runtime: fix counting of free objects · 773685b4
Jan Ziak authored May 15, 2012
```
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6206056
```
773685b4

runtime: refactor helpgc functionality in preparation for parallel GC · 01826280

Dmitriy Vyukov authored May 15, 2012

Parallel GC needs to know in advance how many helper threads will be there.
Hopefully it's the last patch before I can tackle parallel sweep phase.
The benchmarks are unaffected.

R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6200064

01826280

os/syscall: correct netbsd types/names · b554fb91

Joel Sing authored May 15, 2012

Fix types/names so that these functions work with the generated
syscall/z* files.

R=golang-dev, m4dh4tt3r
CC=golang-dev
https://golang.org/cl/6201078

b554fb91

runtime: fix TLS handling for netbsd · b42ebb6f

Joel Sing authored May 15, 2012

Set the TLS base using the _lwp_setprivate() syscall, instead of via
sysarch(). NetBSD tracks the pointer passed to _lwp_setprivate() and
restores this value when restoring mcontext. If sysarch() is used
directly, restoring an mcontext trashes the FS/GS value, resulting
in a segfault when we next try to access the TLS.

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/6206062

b42ebb6f

mime/multipart: fix handling of empty parts without CRLF before next part · e393a829

Brad Fitzpatrick authored May 15, 2012

Empty parts can be either of the form:

a) "--separator\r\n", header (w/ trailing 2xCRLF), \r\n "--separator"...
or
b) "--separator\r\n", header (w/ trailing 2xCRLF), "--separator"...

We never handled case b).  In fact the RFC seems kinda vague about
it, but browsers seem to do a), and App Engine's synthetic POST
bodies after blob uploads is of form b).

So handle them both, and add a bunch of tests.

(I can't promise these are the last fixes to multipart, especially
considering its history, but I'm growing increasingly confident at
least, and I've never submitted a multipart CL with known bugs
outstanding, including this time.)

R=golang-dev, adg
CC=golang-dev
https://golang.org/cl/6212046

e393a829

14 May, 2012 7 commits

misc/dashboard/codereview: add sameer. · a90cbd74
David Symonds authored May 14, 2012
```
R=r
CC=golang-dev
https://golang.org/cl/6203076
```
a90cbd74

regexp/syntax: replace internal error on unexpected ) w/ ErrUnexpectedParen · bd13f6ff

Jan Mercl authored May 14, 2012

Unbalanced extra right parenthesis produced an internal error instead of
a more descriptive one.

Fixes #3406.

R=r, rsc
CC=golang-dev
https://golang.org/cl/6201063

bd13f6ff

syscall: regenerate z* files for netbsd · 2a47e044

Joel Sing authored May 14, 2012

Fix mkerrors.sh so that it works on NetBSD.

Remove directory mode bits from types - this already appears in errors.

Regenerate the z* files now that cgo is working.

R=golang-dev, m4dh4tt3r, r
CC=golang-dev
https://golang.org/cl/6201077

2a47e044

crypto/tls: fix decoding of certLen in certificateMsg.unmarshal · 99142f55
Michael Gehring authored May 14, 2012
```
certLen was decoded incorrectly if length > 2^16-1.

R=golang-dev, agl
CC=golang-dev
https://golang.org/cl/6197077
```
99142f55
A+C: add Michael Gehring (individual CLA) · 02d2b446
Adam Langley authored May 14, 2012
```
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/6188076
```
02d2b446

sync: use atomic.Store in Once.Do · 8c4c6c41

Dmitriy Vyukov authored May 14, 2012

No perf/semantic changes, merely improves code health.
There were several questions as to why Once.Do uses
atomic.CompareAndSwap to do a store.

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/6208057

8c4c6c41

misc/dashboard/codereview: set In-Reply-To header to properly thread mail. · 58bcec62
David Symonds authored May 14, 2012
```
R=adg
CC=golang-dev
https://golang.org/cl/6208051
```
58bcec62

13 May, 2012 1 commit
- A+C: Jan Ziak (individual CLA) · be15a8f2
  David Symonds authored May 13, 2012
```
R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/6120047
```
  be15a8f2
11 May, 2012 2 commits

runtime: revert MaxGcproc from 16 to 4 · eb0bc816

Dmitriy Vyukov authored May 11, 2012

The change accidentally come in with this revision:
https://code.google.com/p/go/source/detail?spec=svn345cbca96c5550f2e89bc727703301933802923c&r=14c38c23c819a17021b1808cf4a34ef3a1a17db5

R=golang-dev
CC=golang-dev
https://golang.org/cl/6195073

eb0bc816

runtime: add parallel for algorithm · 95643647

Dmitriy Vyukov authored May 11, 2012

This is factored out part of:
https://golang.org/cl/5279048/
(parallel GC)

R=bsiegert, mpimenov, rsc, minux.ma, r
CC=golang-dev
https://golang.org/cl/5986054

95643647

10 May, 2012 4 commits

runtime: fix complex division benchmarks · aa45e52e

Shenghou Ma authored May 10, 2012

we can't add the division result to n during iteration, because it might
turn n into NaN or Inf.

R=golang-dev, rsc, iant, iant
CC=golang-dev
https://golang.org/cl/6197045

aa45e52e

runtime: fix syscalls for netbsd · 5374ded1

Joel Sing authored May 10, 2012

Use correct syscall numbers and arguments for NetBSD.
Provide a trampoline for signal returns (using signal API 3).

R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6209048

5374ded1

runtime: fix netbsd sigaction struct · 38445ca0
Joel Sing authored May 10, 2012
```
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/6198063
```
38445ca0

syscall: repair netbsd z* files · a1bf5207

Joel Sing authored May 10, 2012

Manually repair/update z* files for netbsd - this allows Go to
compile again on 386.

R=golang-dev, rsc, minux.ma
CC=golang-dev
https://golang.org/cl/6194064

a1bf5207

09 May, 2012 4 commits

exp/locale/collate: Add maketables tool and generated tables. · 0355a717

Marcel van Lohuizen authored May 09, 2012

Also set maxContractLen automatically.
Note that the table size is much bigger than it needs to be.
Optimization is best done, though, when the language specific
tables are added.

R=r
CC=golang-dev
https://golang.org/cl/6167044

0355a717

cmd/go: fix remote help reference to "go help packages" · 25a8a8da

Francisco Souza authored May 09, 2012

go help remote used to reference "go help importpath", which has
changed to "go help packages".

Fixes #3598.

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/6201065

25a8a8da

go: fix the import path "./..." not matching ".". · c8332198

Nigel Tao authored May 09, 2012

Tested manually.

Fixes #3554.

Before:
$ cd $GOROOT/src/pkg
$ go list io
io
$ go list io/...
io
io/ioutil
$ cd $GOROOT/src/pkg/io
$ go list .
io
$ go list ./...
io/ioutil

After:
$ cd $GOROOT/src/pkg
$ go list io
io
$ go list io/...
io
io/ioutil
$ cd $GOROOT/src/pkg/io
$ go list .
io
$ go list ./...
io
io/ioutil
$ go list ././...
io
io/ioutil
$ go list ././.././io/...
io
io/ioutil
$ go list ../image
image
$ go list ../image/...
image
image/color
image/draw
image/gif
image/jpeg
image/png
$ go list ../.../template
html/template
text/template
$ cd $GOROOT/src/pkg
$ go list ./io
io
$ go list ./io/...
io
io/ioutil
$ go list ./.../pprof
net/http/pprof
runtime/pprof
$ go list ./compress
can't load package: package compress: no Go source files in /home/nigeltao/go/src/pkg/compress
$ go list ./compress/...
compress/bzip2
compress/flate
compress/gzip
compress/lzw
compress/zlib
$ cd $GOROOT/src/pkg/code.google.com
$ go list ./p/leveldb-go/...
code.google.com/p/leveldb-go/leveldb
code.google.com/p/leveldb-go/leveldb/crc
code.google.com/p/leveldb-go/leveldb/db
code.google.com/p/leveldb-go/leveldb/memdb
code.google.com/p/leveldb-go/leveldb/memfs
code.google.com/p/leveldb-go/leveldb/record
code.google.com/p/leveldb-go/leveldb/table
code.google.com/p/leveldb-go/manualtest/filelock
$ go list ./p/.../truetype
code.google.com/p/freetype-go/example/truetype
code.google.com/p/freetype-go/freetype/truetype
$ go list ./p/.../example
warning: "./p/.../example" matched no packages
$ go list ./p/.../example/...
code.google.com/p/freetype-go/example/freetype
code.google.com/p/freetype-go/example/gamma
code.google.com/p/freetype-go/example/raster
code.google.com/p/freetype-go/example/round
code.google.com/p/freetype-go/example/truetype
code.google.com/p/x-go-binding/example/imgview
code.google.com/p/x-go-binding/example/xgb

R=rsc
CC=golang-dev
https://golang.org/cl/6194056

c8332198

compress/testdata: change {e,pi}.txt from 10k to 100k digits. · 738e77aa

Nigel Tao authored May 09, 2012

These files change from exactly 10003 bytes long to 100003: a digit,
a '.', 100k digits, and a '\n'.

The magic constants in compress/flate/deflate_test.go change since
deflateInflateStringTests checks that the compressed form of e.txt
is not 'too large'. I'm not exactly sure how these numbers were
originally calculated (they were introduced in codereview 5554066
"make lazy matching work"); perhaps krasin@golang.org can comment.
My change was to increase the first one (no compression) to a tight
bound, and multiply all the others by 10.

Benchcmp numbers for compress/flate and compress/lzw below. LZW's
window size of 4096 is less than 10k, so shows no significant change.
Flate's window size is 32768, between 10k and 100k, and so the .*1e5
and .*1e6 benchmarks show a dramatic drop, since the compressed forms
are no longer a trivial forward copy of 10k digits repeated over and
over, but should now be more representative of real world usage.

compress/flate:
benchmark                            old MB/s     new MB/s  speedup
BenchmarkDecodeDigitsSpeed1e4           16.58        16.52    1.00x
BenchmarkDecodeDigitsSpeed1e5           68.09        18.10    0.27x
BenchmarkDecodeDigitsSpeed1e6          124.63        18.35    0.15x
BenchmarkDecodeDigitsDefault1e4         17.21        17.12    0.99x
BenchmarkDecodeDigitsDefault1e5        118.28        19.19    0.16x
BenchmarkDecodeDigitsDefault1e6        295.62        20.52    0.07x
BenchmarkDecodeDigitsCompress1e4        17.22        17.17    1.00x
BenchmarkDecodeDigitsCompress1e5       118.19        19.21    0.16x
BenchmarkDecodeDigitsCompress1e6       295.59        20.55    0.07x
BenchmarkEncodeDigitsSpeed1e4            8.18         8.19    1.00x
BenchmarkEncodeDigitsSpeed1e5           43.22        12.84    0.30x
BenchmarkEncodeDigitsSpeed1e6           80.76        13.48    0.17x
BenchmarkEncodeDigitsDefault1e4          6.29         6.19    0.98x
BenchmarkEncodeDigitsDefault1e5         31.63         3.60    0.11x
BenchmarkEncodeDigitsDefault1e6         52.97         3.24    0.06x
BenchmarkEncodeDigitsCompress1e4         6.20         6.19    1.00x
BenchmarkEncodeDigitsCompress1e5        31.59         3.59    0.11x
BenchmarkEncodeDigitsCompress1e6        53.18         3.25    0.06x

compress/lzw:
benchmark               old MB/s     new MB/s  speedup
BenchmarkDecoder1e4        21.99        22.09    1.00x
BenchmarkDecoder1e5        22.77        22.71    1.00x
BenchmarkDecoder1e6        22.90        22.90    1.00x
BenchmarkEncoder1e4        21.04        21.19    1.01x
BenchmarkEncoder1e5        22.06        22.06    1.00x
BenchmarkEncoder1e6        22.16        22.28    1.01x

R=rsc
CC=golang-dev, krasin
https://golang.org/cl/6207043

738e77aa

08 May, 2012 2 commits

compress/flate: benchmark some English text, not just the digits of e. · ffd0d02d

Nigel Tao authored May 08, 2012

The testdata/e.txt input is repeated on the longer benchmarks, but the
length of that data is less than flate's window size, so the numbers are
essentially measuring the performance of a trivial compression. A follow-up
CL will add more data to testdata/e.txt.

Sample output on my laptop (linux, amd64):

BenchmarkDecodeDigitsSpeed1e4 5000 603153 ns/op 16.58 MB/s
BenchmarkDecodeDigitsSpeed1e5 1000 1465602 ns/op 68.23 MB/s
BenchmarkDecodeDigitsSpeed1e6 200 8036050 ns/op 124.44 MB/s
BenchmarkDecodeDigitsDefault1e4 5000 581796 ns/op 17.19 MB/s
BenchmarkDecodeDigitsDefault1e5 2000 846653 ns/op 118.11 MB/s
BenchmarkDecodeDigitsDefault1e6 500 3385782 ns/op 295.35 MB/s
BenchmarkDecodeDigitsCompress1e4 5000 581180 ns/op 17.21 MB/s
BenchmarkDecodeDigitsCompress1e5 2000 846209 ns/op 118.17 MB/s
BenchmarkDecodeDigitsCompress1e6 500 3386174 ns/op 295.32 MB/s
BenchmarkDecodeTwainSpeed1e4 5000 643563 ns/op 15.54 MB/s
BenchmarkDecodeTwainSpeed1e5 500 5418408 ns/op 18.46 MB/s
BenchmarkDecodeTwainSpeed1e6 50 52277520 ns/op 19.13 MB/s
BenchmarkDecodeTwainDefault1e4 5000 583551 ns/op 17.14 MB/s
BenchmarkDecodeTwainDefault1e5 500 4443428 ns/op 22.51 MB/s
BenchmarkDecodeTwainDefault1e6 50 41862080 ns/op 23.89 MB/s
BenchmarkDecodeTwainCompress1e4 5000 583490 ns/op 17.14 MB/s
BenchmarkDecodeTwainCompress1e5 500 4426356 ns/op 22.59 MB/s
BenchmarkDecodeTwainCompress1e6 50 41657940 ns/op 24.01 MB/s
BenchmarkEncodeDigitsSpeed1e4 2000 1230907 ns/op 8.12 MB/s
BenchmarkEncodeDigitsSpeed1e5 1000 2319129 ns/op 43.12 MB/s
BenchmarkEncodeDigitsSpeed1e6 100 12378950 ns/op 80.78 MB/s
BenchmarkEncodeDigitsDefault1e4 1000 1597865 ns/op 6.26 MB/s
BenchmarkEncodeDigitsDefault1e5 500 3163458 ns/op 31.61 MB/s
BenchmarkEncodeDigitsDefault1e6 100 18770240 ns/op 53.28 MB/s
BenchmarkEncodeDigitsCompress1e4 1000 1603461 ns/op 6.24 MB/s
BenchmarkEncodeDigitsCompress1e5 500 3168766 ns/op 31.56 MB/s
BenchmarkEncodeDigitsCompress1e6 100 18855830 ns/op 53.03 MB/s
BenchmarkEncodeTwainSpeed1e4 1000 1338049 ns/op 7.47 MB/s
BenchmarkEncodeTwainSpeed1e5 500 7341622 ns/op 13.62 MB/s
BenchmarkEncodeTwainSpeed1e6 50 67484600 ns/op 14.82 MB/s
BenchmarkEncodeTwainDefault1e4 1000 1778399 ns/op 5.62 MB/s
BenchmarkEncodeTwainDefault1e5 100 23261810 ns/op 4.30 MB/s
BenchmarkEncodeTwainDefault1e6 10 243533600 ns/op 4.11 MB/s
BenchmarkEncodeTwainCompress1e4 1000 1795469 ns/op 5.57 MB/s
BenchmarkEncodeTwainCompress1e5 50 29447140 ns/op 3.40 MB/s
BenchmarkEncodeTwainCompress1e6 5 321686800 ns/op 3.11 MB/s
ok compress/flate 89.246s

R=rsc, r
CC=golang-dev
https://golang.org/cl/6195055

ffd0d02d

compress/testdata: remove Byte Order Mark from the Tom Sawyer data. · 32b85baa

Nigel Tao authored May 08, 2012

I'm not sure where the BOM came from, originally.
http://www.gutenberg.org/files/74/74.txt doesn't have it, although
a fresh download of that URL gives me "\r\n"s instead of plain "\n"s,
and the extra line "Character set encoding: ASCII". Maybe Project
Gutenberg has changed their server configuration since we added that
file to the Go repo.

Anyway, this change is just manually excising the BOM from the start
of the file, leaving pure ASCII.

R=r, bradfitz
CC=golang-dev, krasin, rsc
https://golang.org/cl/6197061

32b85baa