Fixes #17826
Updates #17302

Change-Id: I7c1ebd965e679e7169a97e62d27ae3ede2473aa1
Reviewed-on: https://go-review.googlesource.com/33152Reviewed-by: Ian Lance Taylor <iant@golang.org>

This change is an experimental implementation of asynchronous
cancelable I/O operations on Plan 9, which are required to
implement deadlines.

There are no asynchronous syscalls on Plan 9. I/O operations
are performed with blocking pread and pwrite syscalls.

Implementing deadlines in Go requires a way to interrupt
I/O operations.

It is possible to interrupt reads and writes on a TCP connection
by forcing the closure of the TCP connection. This approach
has been used successfully in CL 31390.

However, we can't implement deadlines with this method, since
we require to be able to reuse the connection after the timeout.

On Plan 9, I/O operations are interrupted when the process
receives a note. We can rely on this behavior to implement
a more generic approach.

When doing an I/O operation (read or write), we start the I/O in
its own process, then wait for the result asynchronously. The
process is able to handle the "hangup" note. When receiving the
"hangup" note, the currently running I/O operation is canceled
and the process returns.

This way, deadlines can be implemented by sending an "hangup"
note to the process running the blocking I/O operation, after
the expiration of a timer.

Fixes #11932.
Fixes #17498.

Change-Id: I414f72c7a9a4f9b8f9c09ed3b6c269f899d9b430
Reviewed-on: https://go-review.googlesource.com/31521Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

1e-9 has a 1 in the last place, causing some Duration calculations to
have unnecessary rounding errors.  1e9 does not, so use that instead.

Change-Id: I96334a2c47e7a014b532eb4b8a3ef9550e7ed057
Reviewed-on: https://go-review.googlesource.com/33116
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

Change-Id: I6e24d22eada190e9aa2adc161be7a753c8e5054b
Reviewed-on: https://go-review.googlesource.com/31514
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

Many linux distros distribute Chromium instead of Chrome.

Change-Id: I5474d94da28a7c79bdd7181f77472d4ce73bb225
Reviewed-on: https://go-review.googlesource.com/29293Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

The tree is inconsistent about single l vs double l in those
words in documentation, test messages, and one error value text.

	$ git grep -E '[Mm]arshall(|s|er|ers|ed|ing)' | wc -l
	      42
	$ git grep -E '[Mm]arshal(|s|er|ers|ed|ing)' | wc -l
	    1694

Make it consistently a single l, per earlier decisions. This means
contributors won't be confused by misleading precedence, and it helps
consistency.

Change the spelling in one error value text in newRawAttributes of
crypto/x509 package to be consistent.

This change was generated with:

	perl -i -npe 's,([Mm]arshal)l(|s|er|ers|ed|ing),$1$2,' $(git grep -l -E '[Mm]arshall' | grep -v AUTHORS | grep -v CONTRIBUTORS)

Updates #12431.
Follows https://golang.org/cl/14150.

Change-Id: I85d28a2d7692862ccb02d6a09f5d18538b6049a2
Reviewed-on: https://go-review.googlesource.com/33017
Run-TryBot: Minux Ma <minux@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

And rename it.

Fixes #17703

Change-Id: I73c82a9b3f96180699c6d33c069a666018eb30f9
Reviewed-on: https://go-review.googlesource.com/33149
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

pkg-config 0.24 adds support for quoting and escaping whitespace;
distros like CentOS 6 are still shipping pkg-config 0.23. Skip the test
there since there's no way to get whitespace into the pkg-config output.

Fixes #17846.

Change-Id: Ie4ea17e9b709372a20178b539498929754bcd51f
Reviewed-on: https://go-review.googlesource.com/33027
Run-TryBot: Quentin Smith <quentin@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

Fixes #17720

Change-Id: Ib95c230deef3934db729856c17908f8e5a1e2b7f
Reviewed-on: https://go-review.googlesource.com/33145
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

When a Go program crashes with GOTRACEBACK=crash, the OS creates a
core dump. Include the text-formatted output of some of the cause of
that crash in the core dump.

Output printed by the runtime before crashing is maintained in a
circular buffer to allow access to messages that may be printed
immediately before calling runtime.throw.

The stack traces printed by the runtime as it crashes are not stored.
The information required to recreate them should be included in the
core file.

Updates #16893

There are no tests covering the generation of core dumps; this change
has not added any.

This adds (reentrant) locking to runtime.gwrite, which may have an
undesired performance impact.

Change-Id: Ia2463be3c12429354d290bdec5f3c8d565d1a2c3
Reviewed-on: https://go-review.googlesource.com/32013
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

Users can implement the smtp.Auth interface and return zero bytes in
the "toServer []byte" return value from the Auth.Start method. People
apparently do this to implement the SMTP "LOGIN" method.

But we were then sending "AUTH LOGIN \r\n" to the server, which some
servers apparently choke on. So, trim it when the toServer value is
empty.

Fixes #17794

Change-Id: I83662dba9e0f61b1c5000396c096cf7110f78361
Reviewed-on: https://go-review.googlesource.com/33143
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

I don't have any way to test or reproduce this problem,
but the current code is clearly wrong for Windows.
Make it better.

As I said on #17165:

But the borrowing of M's and the profiling of M's by the CPU profiler
seem not synchronized enough. This code implements the CPU profiler
on Windows:

	func profileloop1(param uintptr) uint32 {
		stdcall2(_SetThreadPriority, currentThread, _THREAD_PRIORITY_HIGHEST)

		for {
			stdcall2(_WaitForSingleObject, profiletimer, _INFINITE)
			first := (*m)(atomic.Loadp(unsafe.Pointer(&allm)))
			for mp := first; mp != nil; mp = mp.alllink {
				thread := atomic.Loaduintptr(&mp.thread)
				// Do not profile threads blocked on Notes,
				// this includes idle worker threads,
				// idle timer thread, idle heap scavenger, etc.
				if thread == 0 || mp.profilehz == 0 || mp.blocked {
					continue
				}
				stdcall1(_SuspendThread, thread)
				if mp.profilehz != 0 && !mp.blocked {
					profilem(mp)
				}
				stdcall1(_ResumeThread, thread)
			}
		}
	}

	func profilem(mp *m) {
		var r *context
		rbuf := make([]byte, unsafe.Sizeof(*r)+15)

		tls := &mp.tls[0]
		gp := *((**g)(unsafe.Pointer(tls)))

		// align Context to 16 bytes
		r = (*context)(unsafe.Pointer((uintptr(unsafe.Pointer(&rbuf[15]))) &^ 15))
		r.contextflags = _CONTEXT_CONTROL
		stdcall2(_GetThreadContext, mp.thread, uintptr(unsafe.Pointer(r)))
		sigprof(r.ip(), r.sp(), 0, gp, mp)
	}

	func sigprof(pc, sp, lr uintptr, gp *g, mp *m) {
		if prof.hz == 0 {
			return
		}

		// Profiling runs concurrently with GC, so it must not allocate.
		mp.mallocing++

		... lots of code ...

		mp.mallocing--
	}

A borrowed M may migrate between threads. Between the
atomic.Loaduintptr(&mp.thread) and the SuspendThread, mp may have
moved to a new thread, so that it's in active use. In particular
it might be calling malloc, as in the crash stack trace. If so, the
mp.mallocing++ in sigprof would provoke the crash.

Those lines are trying to guard against allocation during sigprof.
But on Windows, mp is the thread being traced, not the current
thread. Those lines should really be using getg().m.mallocing, which
is the same on Unix but not on Windows. With that change, it's
possible the race on the actual thread is not a problem: the traceback
would get confused and eventually return an error, but that's fine.
The code expects that possibility.

Fixes #17165.

Change-Id: If6619731910d65ca4b1a6e7de761fa2518ef339e
Reviewed-on: https://go-review.googlesource.com/33132
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Note, most math functions are structured to use stubs, so that they can
be accelerated with assembly on any platform.
Sinh, cosh, and tanh were not structued with stubs, so this CL does
that. This set of routines was chosen as likely to produce good speedups
with assembly on any platform.

Technique used was minimax polynomial approximation using tables of
polynomial coefficients, with argument range reduction.
A table of scaling factors was also used for cosh and log10.

                     before       after      speedup
BenchmarkCos         22.1 ns/op   6.79 ns/op  3.25x
BenchmarkCosh       125   ns/op  11.7  ns/op 10.68x
BenchmarkLog10       48.4 ns/op  12.5  ns/op  3.87x
BenchmarkSin         22.2 ns/op   6.55 ns/op  3.39x
BenchmarkSinh       125   ns/op  14.2  ns/op  8.80x
BenchmarkTanh        65.0 ns/op  15.1  ns/op  4.30x

Accuracy was tested against a high precision
reference function to determine maximum error.
Approximately 4,000,000 points were tested for each function,
producing the following result.
Note: ulperr is error in "units in the last place"

       max
      ulperr
sin    1.43 (returns NaN beyond +-2^50)
cos    1.79 (returns NaN beyond +-2^50)
cosh   1.05
sinh   3.02
tanh   3.69
log10  1.75

Also includes a set of tests to test non-vector functions even
when SIMD is enabled

Change-Id: Icb45f14d00864ee19ed973d209c3af21e4df4edc
Reviewed-on: https://go-review.googlesource.com/32352
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Michael Munday <munday@ca.ibm.com>

If the Server's Shutdown (or SetKeepAlivesEnabled) method was called
while a connection was in a Handler, but after the headers had been
written, the connection was not later closed.

Fixes #9478
Updates #17754 (reverts that workaround)

Change-Id: I65324ab8217373fbb38e12e2b8bffd0a91806072
Reviewed-on: https://go-review.googlesource.com/33141Reviewed-by: Ian Lance Taylor <iant@golang.org>

CL 31462 made it possible to operate directly on reflect.Values
instead of always forcing a round trip to interface{} and back.
The round trip was losing addressability, which hurt users.

The round trip was also losing "interface-ness", which helped users.
That is, using reflect.ValueOf(v.Interface()) instead of v was doing
an implicit indirect any time v was itself an interface{} value: the result
was the reflect.Value for the underlying concrete value contained in the
interface, not the interface itself.

CL 31462 eliminated some "unnecessary" reflect.Value round trips
in order to preserve addressability, but in doing so it lost this implicit
indirection. This CL adds the indirection back.

It may help to compare the changes in this CL against funcs.go from CL 31462:
https://go-review.googlesource.com/#/c/31462/4/src/text/template/funcs.go

Everywhere CL 31462 changed 'v := reflect.ValueOf(x)' to 'v := x',
this CL changes 'v := x' to 'v := indirectInterface(x)'.

Fixes #17714.

Change-Id: I67cec4eb41fed1d56e1c19f12b0abbd0e59d35a2
Reviewed-on: https://go-review.googlesource.com/33139
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

Fixes #17600.

Change-Id: I7aa0eb0dd959da031b6039b51f07db668d4fb468
Reviewed-on: https://go-review.googlesource.com/33131
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Gudger <igudger@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Fixes #12745

Change-Id: Iebb7c97cb5b68dc080644d796a6ca1c120d41b26
Reviewed-on: https://go-review.googlesource.com/27950Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

TestInterruptWithPanic_h2 was added yesterday in
https://golang.org/cl/33099 and https://golang.org/cl/33103

Deflake it. The http2 server sends an error before logging.

Rather than reorder the http2 code to log before writing the RSTStream
frame, just loop for a bit waiting for the condition we're
expecting.

This goes from 2 in 500 flakes for me to unreproducible.

Change-Id: I062866a5977f50c820965aaf83882ddd7bf98f91
Reviewed-on: https://go-review.googlesource.com/33140
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

The restrictions were already being applied to the IP addresses
received from the host resolver. Apply the same restrictions to
literal IP addresses not passed to the host resolver.

For example, ResolveTCPAddr("tcp4", "[2001:db8::1]:http") used
to succeed and now does not (that's not an IPv4 address).

Perhaps a bit surprisingly,
ResolveTCPAddr("tcp4", "[::ffff:127.0.0.1]:http") succeeds,
behaving identically to ResolveTCPAddr("tcp4", "127.0.0.1:http"), and
ResolveTCPAddr("tcp6", "[::ffff:127.0.0.1]:http") fails,
behaving identically to ResolveTCPAddr("tcp6", "127.0.0.1:http").
Even so, it seems right to match (by reusing) the existing filtering
as applied to addresses resolved by the host C library.
If anyone can make a strong argument for changing the filtering
of IPv4-inside-IPv6 addresses, the fix can be applied to all
the code paths in a separate CL.

Fixes #14037.

Change-Id: I690dfdcbe93d730e11e00ea387fa7484cd524341
Reviewed-on: https://go-review.googlesource.com/32100
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

All the existing CPU profiler tests already parse the profile.
That should be sufficient indication that profiles can be parsed.

Fixes #17853.

Change-Id: Ie8a190e2ae4eef125c8eb0d4e8b7adac420abbdb
Reviewed-on: https://go-review.googlesource.com/33136
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

rsc's change golang.org/cl/32455 added a mechanism
that allows pprof to depend on gzip without introducing
an import cycle. This obsoletes the need for the gzip0
package, which was created solely to remove the need
for that dependency.

Change-Id: Ifa3b98faac9b251f909b84b4da54742046c4e3ad
Reviewed-on: https://go-review.googlesource.com/33137Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

Fix spelling of "original" and "occurred" in new gofmt docs. The same
misspelling of "occurred" was also present in crypto/tls, I fixed it there as
well.

Change-Id: I67b4f1c09bd1a2eb1844207d5514f08a9f525ff9
Reviewed-on: https://go-review.googlesource.com/33138Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

cmd/vet/all still doesn't run for mips/mipsle,
because the rest of the toolchain doesn't yet
fully support it.

Change-Id: I1a86b0edddbdcd5f43e752208508d99da7aabbb3
Reviewed-on: https://go-review.googlesource.com/33134
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

Fixes these complaints from vet:

cmd/compile/internal/gc/noder.go:32: cmd/compile/internal/syntax.Error composite literal uses unkeyed fields
cmd/compile/internal/gc/noder.go:1035: cmd/compile/internal/syntax.Error composite literal uses unkeyed fields
cmd/compile/internal/gc/noder.go:1051: cmd/compile/internal/syntax.Error composite literal uses unkeyed fields
cmd/compile/internal/syntax/parser_test.go:182: possible formatting directive in Error call
net/http/client_test.go:1334: possible formatting directive in Fatal call

Change-Id: I5f90ec30f3c106c7e66c92e2b6f8d3b4874fec66
Reviewed-on: https://go-review.googlesource.com/33133
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

Fixes #17788

Change-Id: I2f8a11321dc8f10bebbc8df90ba00ec65b9ee0fa
Reviewed-on: https://go-review.googlesource.com/32790Reviewed-by: Russ Cox <rsc@golang.org>

The 255-octet limit applies to wire format, not presentation format.

Fixes #17549

Change-Id: I2b5181c53fba32fea60178e0d8df9114aa992b55
Reviewed-on: https://go-review.googlesource.com/31722
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

Fixes #16042.

Change-Id: I0a28aa004246b7b0ffaaab457e077ad9035363c2
Reviewed-on: https://go-review.googlesource.com/31932Reviewed-by: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

Some commands generate binary outputs which are not human readable.
In interactive mode, there are no use-cases for such outputs.
Instead, the new code writes it to the temporary file on the $CWD and
shows the file name. So the user can use any program to display the
file outside interactive shell.

Fixes #17465

Change-Id: I5c479db26017607f7a28eafbff2385533e5c584e
Reviewed-on: https://go-review.googlesource.com/31123Reviewed-by: Russ Cox <rsc@golang.org>

It will just cause confusion later as the go tool will say
"warning: GOPATH set to GOROOT (%s) has no effect".
Better to just leave GOPATH unset and get that warning instead.

Change-Id: I78ff9e87fdf4bb0460f4f6d6ee76e1becaa3e7b0
Reviewed-on: https://go-review.googlesource.com/33105
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

Change-Id: I94e0f3e4bccd44a67934ddb4d5fc7da57bb8ac9f
Reviewed-on: https://go-review.googlesource.com/33112Reviewed-by: Russ Cox <rsc@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

On some systems the external linker defaults to PIE. On some systems
DT_TEXTREL does not work correctly. When both are true we have a bad
situation: any Go program built with the default buildmode (exe) that
uses external linking will fail to run. Fix this by passing -no-pie to
the external linker, if the option is supported.

Fixes #17847.

Change-Id: I9b5ff97825d8b7f494f96d29c4c04f72b53dbf4e
Reviewed-on: https://go-review.googlesource.com/33106
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Crawshaw <crawshaw@golang.org>

Fixes #16412

Change-Id: Idc65d2a62414a9b1573e6bd9f8601b52985b5dea
Reviewed-on: https://go-review.googlesource.com/33110Reviewed-by: Ian Lance Taylor <iant@golang.org>

The cgo tool used to simply ignore C type qualifiers. To avoid problems
when a C function expected a qualifier that was not present, cgo emitted
a cast to void* around all pointer arguments. Unfortunately, that broke
code that contains both a function declaration and a macro, when the
macro required the argument to have the right type. To fix this problem,
don't ignore qualifiers. They are easy enough to handle for the limited
set of cases that matter for cgo, in which we don't care about array or
function types.

Fixes #17537.

Change-Id: Ie2988d21db6ee016a3e99b07f53cfb0f1243a020
Reviewed-on: https://go-review.googlesource.com/33097
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

Whenever GOPATH is not defined in the environment, use $HOME/go
as its default value. For Windows systems use %USERPROFILE%/go
and $home/go for plan9.

The choice of these environment variables is based on what Docker
currently does. The os/user package is not used to avoid having
a cgo dependency.

Updates #17262. Documentation changes forthcoming.

Change-Id: I6368fbfbc5afda99d6e64c35c1980076fcf45344
Reviewed-on: https://go-review.googlesource.com/32019
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

Also:
- update performance improvements for CL 31275.

Change-Id: I2f2ec0a42b248643e76df8654e11bf0b01a5d030
Reviewed-on: https://go-review.googlesource.com/33114Reviewed-by: Robert Griesemer <gri@golang.org>

CL 32355 switched from using the output file as a
plugin prefix to the full package path. The linker dead code analysis
was not updated.

Updates #17821

Change-Id: I13fc45e0264b425d28524ec54c829e2c3e895b0b
Reviewed-on: https://go-review.googlesource.com/32916Reviewed-by: Ian Lance Taylor <iant@golang.org>

Fixes the build when CGO_ENABLED=0.

Change-Id: I7f3c67d61e156e69536558fda0a0a4b429b82bbd
Reviewed-on: https://go-review.googlesource.com/33104
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

This reverts the changes from https://golang.org/cl/33018: Instead
of writing the result of gofmt to a tmp file and then rename that
to the original (which doesn't preserve the original file's perm
bits, uid, gid, and possibly other properties because it is hard
to do in a platform-independent way - see #17869), use the original
code that simply overwrites the processed file if gofmt was able to
create a backup first. Upon success, the backup is removed, otherwise
it remains.

Fixes #17873.
For #8984.

Change-Id: Ifcf2bf1f84f730e6060f3517d63b45eb16215ae1
Reviewed-on: https://go-review.googlesource.com/33098
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

Updates http2 to x/net/http2 git rev 0e2717d for:

   http2: conditionally log stacks from panics in Server Handlers like net/http
   https://golang.org/cl/33102

Fixes #17790

Change-Id: Idd3f0c65540398d41b412a33f1d80de3f7f31409
Reviewed-on: https://go-review.googlesource.com/33103Reviewed-by: Joe Tsai <thebrokentoaster@gmail.com>

This test was only enabled by default today so it hasn't been hardened
by build.golang.org. Welcome to the ring, TestClientTimeout.

Change-Id: I1967f6c825699f13f6c659dc14d3c3c22b965272
Reviewed-on: https://go-review.googlesource.com/33101Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>