cmd/compile/internal/syntax: expose additional information for gc

gc needs access to line offsets for Nodes. It also needs access to the
end line offset for function bodies so it knows what line number to
use for things like implicit returns and defer executions.

Lastly, include an extra bool to distinguish between simple and full
slice expressions. This is redundant in valid parse trees, but needed
by gc for producing complete warnings in invalid inputs.

Change-Id: I64baf334a35c72336d26fa6755c67eb9d6f4e93c
Reviewed-on: https://go-review.googlesource.com/27196
Reviewed-by: Robert Griesemer <gri@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

log/syslog: document that Dial uses net.Dial parameters, add example

Fixes #16761

Change-Id: I709daa87926a31e5f8fd46a4c5ef69718ae349b1
Reviewed-on: https://go-review.googlesource.com/27209
Reviewed-by: Chris Broadfoot <cbro@golang.org>

os: reference LookupEnv from the Getenv docs

Fixes #16782

Change-Id: If54917bf5ca1588d8a6d443c3aa6e1d4ada6b620
Reviewed-on: https://go-review.googlesource.com/27322
Reviewed-by: anatoly techtonik <techtonik@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

cmd/vet: allow large shifts of constants

Large shifts of constants are frequently
used for fancy 32/64 bit detection.

This removes 14 false positives from the
standard library.

Updates #11041

Change-Id: Ib39346e5c161da04c38a6a3067932ef43bf74f2d
Reviewed-on: https://go-review.googlesource.com/27155
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Robert Griesemer <gri@golang.org>

crypto/tls: don't generate random ticket keys if already set.

If SetSessionTicketKeys was called on a fresh tls.Config, the configured
keys would be overridden with a random key by serverInit.

Fixes #15421.

Change-Id: I5d6cc81fc3e5de4dfa15eb614d102fb886150d1b
Reviewed-on: https://go-review.googlesource.com/27317
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

go/types: better doc string for Object.Parent and test

Fixes #14647.

Change-Id: Ib9012a9141e815f5b95f8ca2307e65ffc4587a5b
Reviewed-on: https://go-review.googlesource.com/27370
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

crypto/tls: fix comment typo.

This was pointed out in https://go-review.googlesource.com/#/c/27315/1
but I changed and uploaded the wrong branch. This actually makes the
fix.

Change-Id: Ib615b06c9141b914648b6abbeeb688c5ffa0d2e3
Reviewed-on: https://go-review.googlesource.com/27360
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

doc: upate go1.8.txt

Change-Id: I42597785be6121d8180520b3f7d8e936464f0048
Reviewed-on: https://go-review.googlesource.com/27361
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

crypto/tls: support AES-128-CBC cipher suites with SHA-256.

These were new with TLS 1.2 and, reportedly, some servers require it.
Since it's easy, this change adds suport for three flavours of
AES-128-CBC with SHA-256 MACs.

Other testdata/ files have to be updated because this changes the list
of cipher suites offered by default by the client.

Fixes #15487.

Change-Id: I1b14330c31eeda20185409a37072343552c3464f
Reviewed-on: https://go-review.googlesource.com/27315
Run-TryBot: Adam Langley <agl@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Jonathan Rudenberg <jonathan@titanous.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

doc: update go1.8.txt

Change-Id: I9a7654a6d623add8542a1c34ccc76ea136a9a7e3
Reviewed-on: https://go-review.googlesource.com/27359
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

cmd/dist: build cmd/compile/internal/syntax

Change-Id: Ie6dd2318e031be445c0b1ae65d4c78723d5a1167
Reviewed-on: https://go-review.googlesource.com/27197
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>

cmd/compile/internal/syntax: fast Go syntax trees, initial commit.

Syntax tree nodes, scanner, parser, basic printers.

Builds syntax trees for entire Go std lib at a rate of ~1.8M lines/s
in warmed up state (MacMini, 2.3 GHz Intel Core i7, 8GB RAM):

$ go test -run StdLib -fast
parsed 1074617 lines (2832 files) in 579.66364ms (1853863 lines/s)
allocated 282.212Mb (486.854Mb/s)
PASS

Change-Id: Ie26d9a7bf4e5ff07457aedfcc9b89f0eba72ae3f
Reviewed-on: https://go-review.googlesource.com/27195
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>

spec: undo spec date change introduced by prior commit

I accidentally included a modified spec in
https://go-review.googlesource.com/27290 .
Remove that change.

Change-Id: Icb62fe829072860e9eb74865d21e06f30efcfd26
Reviewed-on: https://go-review.googlesource.com/27357
Reviewed-by: Robert Griesemer <gri@golang.org>

go/types: fix scope extents for range and type switch variables

The changes match the existing compilers, and assume an adjusted
spec (per issue #16794).

Fixes #15686.

Change-Id: I72677ce75888c41a8f3c2963117a2f2d5501c42b
Reviewed-on: https://go-review.googlesource.com/27290
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

runtime: disable TestCgoCallbackGC on FreeBSD

The trybot flakes are a nuisance.

Updates #16396

Change-Id: I8202adb554391676ba82bca44d784c6a81bf2085
Reviewed-on: https://go-review.googlesource.com/27313
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

go/types: set Info.Types.Type for array composite literals of the form [...]T

Fixes #14092.

Change-Id: I00692f60a416348e38cab256b94fda07e334d258
Reviewed-on: https://go-review.googlesource.com/27316
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

unsafe: document use of &^ to round/align pointers

Follow-up to CL 27156

Change-Id: I4f1cfced2dced9c9fc8a05bbc00ec4229e85c5c9
Reviewed-on: https://go-review.googlesource.com/27314
Reviewed-by: Ian Lance Taylor <iant@golang.org>

cmd/compile: compare size in dead store elimination

Only remove stores that is shadowed by another store with same or
larger size. Normally we don't need this check because we did check
the types, but unsafe pointer casting can get around it.

Fixes #16769.

Change-Id: I3f7c6c57807b590a2f735007dec6c65a4fa01a34
Reviewed-on: https://go-review.googlesource.com/27320
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

cmd/compile: ppc64le working, not optimized enough

This time with the cherry-pick from the proper patch of
the old CL.

Stack size increased.
Corrected NaN-comparison glitches.
Marked g register as clobbered by calls.
Fixed shared libraries.

live_ssa.go still disabled because of differences.
Presumably turning on more optimization will fix
both the stack size and the live_ssa.go glitches.

Enhanced debugging output for shared libs test.

Rebased onto master.

Updates #16010.

Change-Id: I40864faf1ef32c118fb141b7ef8e854498e6b2c4
Reviewed-on: https://go-review.googlesource.com/27159
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

cmd/yacc: remove go tool yacc

It is no longer used by Go.

It's now moved to golang.org/x/tools/cmd/goyacc for anybody who needs it.

Fixes #11229

Change-Id: Ia431d5a380c7ff784a2050dee2f5bc8acee015da
Reviewed-on: https://go-review.googlesource.com/27325
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

cmd: generate DWARF for functions in compile instead of link.

This is a copy of golang.org/cl/22092 by Ryan Brown.

Here's his original comment:
On my machine this increases the average time for 'go build cmd/go' from
2.25s to 2.36s. I tried to measure compile and link separately but saw
no significant change.

Change-Id: If0d2b756d52a0d30d4eda526929c82794d89dd7b
Reviewed-on: https://go-review.googlesource.com/25311
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Crawshaw <crawshaw@golang.org>

cmd/vet: allow any printf verb with any interface

fmt treats interfaces as being transparent.
As a result, we cannot say with confidence
that any particular verb is wrong.

This fixes the following vet false positives
in the standard library:

database/sql/sql_test.go:210: arg dep for printf verb %p of wrong type: sql.finalCloser
fmt/fmt_test.go:1663: arg nil for printf verb %s of wrong type: untyped nil
go/ast/commentmap.go:328: arg node for printf verb %p of wrong type: ast.Node
net/http/transport_test.go:120: arg c for printf verb %p of wrong type: net.Conn
net/http/httptest/server.go:198: arg c for printf verb %p of wrong type: net.Conn
net/http/httputil/dump_test.go:258: arg body for printf verb %p of wrong type: io.Reader
reflect/set_test.go:81: arg x for printf verb %p of wrong type: io.Writer
reflect/set_test.go:141: arg bb for printf verb %p of wrong type: io.Reader

Updates #11041
Updates #16314

Change-Id: I76df01abb3c34a97b6960f551bed9c1c91377cfc
Reviewed-on: https://go-review.googlesource.com/27127
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

debug/pe: revert CL 22720

CL 22720 hid all recently added functionality for go1.7.
Make everything exported again, so we could use it now.

Updates #15345

Change-Id: Id8ccba7199422b554407ec14c343d2c28fbb8f72
Reviewed-on: https://go-review.googlesource.com/27212
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Crawshaw <crawshaw@golang.org>

crypto/x509: support PSS signatures.

Although the term “RSA” is almost synonymous with PKCS#1 v1.5, that
standard is quite flawed, cryptographically speaking. Bellare and
Rogaway fixed PKCS#1 v1.5 with OAEP (for encryption) and PSS (for
signatures) but they only see a fraction of the use of v1.5.

This change adds support for creating and verifying X.509 certificates
that use PSS signatures. Sadly, every possible dimension of flexibility
seems to have been reflected in the integration of X.509 and PSS
resulting in a huge amount of excess complexity. This change only
supports one “sane” configuration for each of SHA-{256, 384, 512}.
Hopefully this is sufficient because it saves a lot of complexity in the
code.

Although X.509 certificates with PSS signatures are rare, I'm inclined
to look favourably on them because they are sufficiently superior.

Fixes #15958.

Change-Id: I7282e0b68ad0177209f8b2add473b94aa5224c07
Reviewed-on: https://go-review.googlesource.com/24743
Run-TryBot: Adam Langley <agl@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

crypto/x509: require a NULL parameters for RSA public keys.

The RFC is clear that the Parameters in an AlgorithmIdentifer for an RSA
public key must be NULL. BoringSSL enforces this so we have strong
evidence that this is a widely compatible change.

Embarrassingly enough, the major source of violations of this is us. Go
used to get this correct in only one of two places. This was only fixed
in 2013 (with 4874bc9b). That's why lots of test certificates are
updated in this change.

Fixes #16166.

Change-Id: Ib9a4551349354c66e730d44eb8cee4ec402ea8ab
Reviewed-on: https://go-review.googlesource.com/27312
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

encoding/hex: change lookup table from string to array

name            old time/op  new time/op  delta
Encode/256-4     431ns ± 2%   391ns ± 2%   -9.36%  (p=0.000 n=8+8)
Encode/1024-4   1.68µs ± 0%  1.51µs ± 0%   -9.91%  (p=0.001 n=7+7)
Encode/4096-4   6.68µs ± 0%  6.03µs ± 1%   -9.69%  (p=0.000 n=8+8)
Encode/16384-4  27.0µs ± 1%  24.0µs ± 0%  -11.03%  (p=0.000 n=8+7)

Change-Id: I6994e02f77797349c4e188377d84f97dffe98399
Reviewed-on: https://go-review.googlesource.com/27254
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

go/types: enable disabled test for comma-ok expressions

This was fixed long ago but the test was not enabled.

For #8189.

Change-Id: Ia44ef752b6bf076f3e243d2d0db326a392a20193
Reviewed-on: https://go-review.googlesource.com/27310
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

crypto/hmac: don't test for length equality in Equal.

subtle.ConstantTimeCompare now tests the length of the inputs (although
it didn't when this code was written) so this test in crypto/hmac is now
superfluous.

Fixes #16336.

Change-Id: Ic02d8537e776fa1dd5694d3af07a28c4d840d14b
Reviewed-on: https://go-review.googlesource.com/27239
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

crypto/x509: return error for missing SerialNumber.

If the SerialNumber is nil in the template then the resulting panic is
rather deep in encoding/asn1 and it's not obvious what went wrong.

This change tests and returns a more helpful error in this case.

Fixes #16603.

Change-Id: Ib30d652555191eb78f705dff8d909e4b5808f9ca
Reviewed-on: https://go-review.googlesource.com/27238
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

runtime: fix the absolute URL to pprof tools

Change-Id: I82eaf5c14a5b8b9ec088409f946adf7b5fd5dbe3
Reviewed-on: https://go-review.googlesource.com/27311
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

text/template: remove unused Tree.parse return value

Fixes #13993

Change-Id: Ic61b2bcd9f4f71457d3a8581574633d505d5750e
Reviewed-on: https://go-review.googlesource.com/27240
Run-TryBot: Rob Pike <r@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

go/types: better error message for invalid fallthrough case

Now matches the gc compiler.

Fixes #15594.

Change-Id: I9f3942367bc0acf883c6216b8ca44820832f5fe3
Reviewed-on: https://go-review.googlesource.com/27241
TryBot-Result: Gobot Gobot <gobot@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

hash/crc32: improve the processing of the last bytes in the SSE4.2 code for AMD64

This commit improves the processing of the final few bytes in
castagnoliSSE42: instead of processing one byte at a time, we use all
versions of the CRC32 instruction to process 4 bytes, then 2, then 1.
The difference is only noticeable for small "odd" sized buffers.

We do the similar improvement for processing the first few bytes in
the case of unaligned buffer.

Fixing the test which was not actually verifying the results for
misaligned buffers (WriteString was creating an internal copy which
was aligned).

Adding benchmarks for length 15 (aligned and misaligned), results
below.

name                          old time/op    new time/op    delta
CastagnoliCrc15B-4              25.1ns ± 0%    22.1ns ± 1%  -12.14%
CastagnoliCrc15BMisaligned-4    25.2ns ± 0%    22.9ns ± 1%   -9.03%
CastagnoliCrc40B-4              23.1ns ± 0%    23.4ns ± 0%   +1.08%
CastagnoliCrc1KB-4               127ns ± 0%     128ns ± 0%   +1.18%
CastagnoliCrc4KB-4               462ns ± 0%     464ns ± 0%     ~
CastagnoliCrc32KB-4             3.58µs ± 0%    3.60µs ± 0%   +0.58%

name                          old speed      new speed      delta
CastagnoliCrc15B-4             597MB/s ± 0%   679MB/s ± 1%  +13.77%
CastagnoliCrc15BMisaligned-4   596MB/s ± 0%   655MB/s ± 1%   +9.94%
CastagnoliCrc40B-4            1.73GB/s ± 0%  1.71GB/s ± 0%   -1.14%
CastagnoliCrc1KB-4            8.01GB/s ± 0%  7.93GB/s ± 1%   -1.06%
CastagnoliCrc4KB-4            8.86GB/s ± 0%  8.83GB/s ± 0%     ~
CastagnoliCrc32KB-4           9.14GB/s ± 0%  9.09GB/s ± 0%   -0.58%

Change-Id: I499e37af2241d28e3e5d522bbab836c1a718430a
Reviewed-on: https://go-review.googlesource.com/24470
Reviewed-by: Keith Randall <khr@golang.org>

cmd/compile: CSE copied tuple selectors

In CSE if a tuple generator is CSE'd to a different block, its
selectors are copied to the same block. In this case, also CES
the copied selectors.

Test copied from Keith's CL 27202.

Fixes #16741.

Change-Id: I2fc8b9513d430f10d6104275cfff5fb75d3ef3d9
Reviewed-on: https://go-review.googlesource.com/27236
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Keith Randall <khr@golang.org>

cmd/vet: infer asm arch from build context

If we cannot infer the asm arch from the filename
or the build tags, assume that it is the
current build arch. Assembly files with no
restrictions ought to be usable on all arches.

Updates #11041

Change-Id: I0ae807dbbd5fb67ca21d0157fe180237a074113a
Reviewed-on: https://go-review.googlesource.com/27151
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

cmd/vet: allow ^& uintptr arithmetic

The unsafe.Pointer check allows adding to
and subtracting from uintptrs in order to do
arithmetic.

Some code needs to round uintptrs.
Allow &^ for that purpose.

Updates #11041

Change-Id: Ib90dd2954bb6c78427058271e13f2ce4c4af38fb
Reviewed-on: https://go-review.googlesource.com/27156
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

crypto/tls: set Conn.ConnectionState.ServerName unconditionally

Moves the state.ServerName assignment to outside the if
statement that checks for handshakeComplete.

Fixes #15571

Change-Id: I6c4131ddb16389aed1c410a975f9aa3b52816965
Reviewed-on: https://go-review.googlesource.com/22862
Run-TryBot: Adam Langley <agl@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Adam Langley <agl@golang.org>

cmd/compile: add more ARM64 optimizations

- Use machine instructions for uint64<->float conversions
- Do not enforce alignment on Zero/Move
ARM64 supports unaligned load/stores, but only aligned offset
or small offset can be encoded into instructions.
- Do combined loads

Change-Id: Iffca7dd0f13070b17b784861ce5a30af584680eb
Reviewed-on: https://go-review.googlesource.com/27086
Reviewed-by: David Chase <drchase@google.com>

math/big: avoid allocation in float.{Add, Sub} when there's no aliasing

name               old time/op    new time/op    delta
FloatAdd/10-4         116ns ± 1%      82ns ± 0%   -28.74%  (p=0.008 n=5+5)
FloatAdd/100-4        124ns ± 0%      86ns ± 1%   -30.34%  (p=0.016 n=4+5)
FloatAdd/1000-4       192ns ± 1%     123ns ± 0%   -35.94%  (p=0.008 n=5+5)
FloatAdd/10000-4      826ns ± 0%     438ns ± 0%   -46.99%  (p=0.000 n=4+5)
FloatAdd/100000-4    6.82µs ± 1%    3.36µs ± 0%   -50.74%  (p=0.008 n=5+5)
FloatSub/10-4         108ns ± 1%      77ns ± 1%   -29.06%  (p=0.008 n=5+5)
FloatSub/100-4        115ns ± 0%      79ns ± 0%   -31.48%  (p=0.029 n=4+4)
FloatSub/1000-4       168ns ± 0%      99ns ± 0%   -41.09%  (p=0.029 n=4+4)
FloatSub/10000-4      690ns ± 2%     288ns ± 1%   -58.24%  (p=0.008 n=5+5)
FloatSub/100000-4    5.37µs ± 1%    2.10µs ± 1%   -60.89%  (p=0.008 n=5+5)

name               old alloc/op   new alloc/op   delta
FloatAdd/10-4         48.0B ± 0%     0.0B ±NaN%  -100.00%  (p=0.008 n=5+5)
FloatAdd/100-4        64.0B ± 0%     0.0B ±NaN%  -100.00%  (p=0.008 n=5+5)
FloatAdd/1000-4        176B ± 0%       0B ±NaN%  -100.00%  (p=0.008 n=5+5)
FloatAdd/10000-4     1.41kB ± 0%   0.00kB ±NaN%  -100.00%  (p=0.008 n=5+5)
FloatAdd/100000-4    13.6kB ± 0%    0.0kB ±NaN%  -100.00%  (p=0.008 n=5+5)
FloatSub/10-4         48.0B ± 0%     0.0B ±NaN%  -100.00%  (p=0.008 n=5+5)
FloatSub/100-4        64.0B ± 0%     0.0B ±NaN%  -100.00%  (p=0.008 n=5+5)
FloatSub/1000-4        176B ± 0%       0B ±NaN%  -100.00%  (p=0.008 n=5+5)
FloatSub/10000-4     1.41kB ± 0%   0.00kB ±NaN%  -100.00%  (p=0.008 n=5+5)
FloatSub/100000-4    13.6kB ± 0%    0.0kB ±NaN%  -100.00%  (p=0.008 n=5+5)

Fixes #14868

Change-Id: Ia2b8b1a8ef0868288ecb25f812b17bd03ff40d1c
Reviewed-on: https://go-review.googlesource.com/23568
Reviewed-by: Robert Griesemer <gri@golang.org>

cmd/compile: add compiler phase timing

Timings is a simple data structure that collects times of labeled
Start/Stop events describing timed phases, which later can be written
to a file.

Adjacent phases with common label prefix are automatically collected
in a group together with the accumulated phase time.

Timing data can be appended to a file in benchmark data format
using the new -bench flag:

$ go build -gcflags="-bench=/dev/stdout" -o /dev/null go/types
commit: devel +8847c6b Mon Aug 15 17:51:53 2016 -0700
goos: darwin
goarch: amd64
BenchmarkCompile:go/types:fe:init              1       663292 ns/op      0.07 %
BenchmarkCompile:go/types:fe:loadsys           1      1337371 ns/op      0.14 %
BenchmarkCompile:go/types:fe:parse             1     47008869 ns/op      4.91 %    10824 lines    230254 lines/s
BenchmarkCompile:go/types:fe:typecheck:top1    1      2843343 ns/op      0.30 %
BenchmarkCompile:go/types:fe:typecheck:top2    1       447457 ns/op      0.05 %
BenchmarkCompile:go/types:fe:typecheck:func    1     15119595 ns/op      1.58 %      427 funcs     28241 funcs/s
BenchmarkCompile:go/types:fe:capturevars       1        56314 ns/op      0.01 %
BenchmarkCompile:go/types:fe:inlining          1      9805767 ns/op      1.02 %
BenchmarkCompile:go/types:fe:escapes           1     53598646 ns/op      5.60 %
BenchmarkCompile:go/types:fe:xclosures         1       199302 ns/op      0.02 %
BenchmarkCompile:go/types:fe:subtotal          1    131079956 ns/op     13.70 %
BenchmarkCompile:go/types:be:compilefuncs      1    692009428 ns/op     72.33 %      427 funcs       617 funcs/s
BenchmarkCompile:go/types:be:externaldcls      1        54591 ns/op      0.01 %
BenchmarkCompile:go/types:be:dumpobj           1    133478173 ns/op     13.95 %
BenchmarkCompile:go/types:be:subtotal          1    825542192 ns/op     86.29 %
BenchmarkCompile:go/types:unaccounted          1       106101 ns/op      0.01 %
BenchmarkCompile:go/types:total                1    956728249 ns/op    100.00 %

For #16169.

Change-Id: I93265fe0cb08e47cd413608d0824c5dd35ba7899
Reviewed-on: https://go-review.googlesource.com/24462
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

net: don't avoid resolving .local addresses

.local addresses are used by things like Kubernetes and Weave DNS; Go
should not avoid resolving them.

This is a partial revert of https://golang.org/cl/21328 which was too
strict of an interpretation of RFC 6762.

Fixes #16739

Change-Id: I349415b4eab5d61240dd18217bd95dc7d2105cd5
Reviewed-on: https://go-review.googlesource.com/27250
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/compile/internal/s390x: cleanup betypeinit

The Width{int,ptr,reg} assignments are no longer necessary since
golang.org/cl/21623. The other arch's betypeinit functions were
cleaned up, but apparently this one was missed.

Change-Id: I1c7f074d7864a561659c1f98aef604f57f285fd0
Reviewed-on: https://go-review.googlesource.com/27272
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

doc/go1.8: support "option ndots:0" in resolv.conf

Updates #15419.

Change-Id: If7c80adcb38b5731e337b2ae2d9d76fcf8513d8e
Reviewed-on: https://go-review.googlesource.com/27271
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

net: use libresolv rules for ndots range and validation

BIND libresolv allows values from 0 to 15.

For invalid values and negative numbers, 0 is used.
For numbers greater than 15, 15 is used.

Fixes #15419

Change-Id: I1009bc119c3e87919bcb55a80a35532e9fc3ba52
Reviewed-on: https://go-review.googlesource.com/24901
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

test: add test for CL 26831

Test nil check removal for access of PAUTOHEAP.

Change-Id: Id739a9cda7cd3ff173bdcccfedcad93ee90711ef
Reviewed-on: https://go-review.googlesource.com/27232
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

math/rand: Document origin of cooked pseudo-random numbers

The Source provided by math/rand relies on an array of cooked
pseudo-random 63bit integers for seeding. The origin of these
numbers is undocumented.

Add a standalone program in math/rand folder that generates
the 63bit integer array as well as a 64bit version supporting
extension of the Source to 64bit pseudo-random number
generation while maintaining the current sequence in the
lower 63bit.

The code is largely based on the initial implementation of the
random number generator in the go repository by Ken Thompson
(revision 399).

Change-Id: Ib4192aea8127595027116a0f5a7be53f11dc110b
Reviewed-on: https://go-review.googlesource.com/22230
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

runtime: fix check for vacuous page boundary rounding

sysUnused (e.g., madvise MADV_FREE) is only sensible to call on
physical page boundaries, so scavengelist rounds in the bounds of the
region being released to the nearest physical page boundaries.
However, if the region is smaller than a physical page and neither the
start nor end fall on a boundary, then rounding the start up to a page
boundary and the end down to a page boundary will result in end < start.
Currently, we only give up on the region if start == end, so if we
encounter end < start, we'll call madvise with a negative length and
the madvise will fail.

Issue #16644 gives a concrete example of this:

    start = 0x1285ac000
    end   = 0x1285ae000 (1 8K page)

This leads to the rounded values

    start = 0x1285b0000
    end   = 0x1285a0000

which leads to len = -65536.

Fix this by giving up on the region if end <= start, not just if
end == start.

Fixes #16644.

Change-Id: I8300db492dbadc82ac1ad878318b36bcb7c39524
Reviewed-on: https://go-review.googlesource.com/27230
Reviewed-by: Keith Randall <khr@golang.org>

syscall: mksyscall_windows.go: put path separator in suffix for matching GOROOT

fixes #16493

Change-Id: I86bec2f9bd7965449c43e94733791f7cb18c5c4c
Reviewed-on: https://go-review.googlesource.com/25165
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>

net: simplify internal dtoi and xtoi funcs

Callers pass strings sliced as necessary instead of giving
an offset.

Fixes #16350

Change-Id: I7ba896f6ff09e0fd0094ca6c5af5d9a81622f15e
Reviewed-on: https://go-review.googlesource.com/27206
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

go/build: introduce go1.8 build tag

Change-Id: Ib8855f8125970fc7ffb271635c28d31d310fcb5b
Reviewed-on: https://go-review.googlesource.com/27192
Run-TryBot: Joe Tsai <thebrokentoaster@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/compile: handle e == T comparison more efficiently

Instead of making a runtime call, compare types and values.

Change-Id: Id302083d5a6a5f18e04f36f304f3d290c46976ad
Reviewed-on: https://go-review.googlesource.com/26660
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

cmd/compile: inline x, ok := y.(T) where T is a scalar

When T is a scalar, there are no runtime calls
required, which makes this a clear win.

encoding/binary:
WriteInts-8                958ns ± 3%     864ns ± 2%   -9.80%  (p=0.000 n=15+15)

This also considerably shrinks a core fmt
routine:

Before: "".(*pp).printArg t=1 size=3952 args=0x20 locals=0xf0
After:  "".(*pp).printArg t=1 size=2624 args=0x20 locals=0x98

Unfortunately, I find it very hard to get stable
numbers out of the fmt benchmarks due to thermal scaling.

Change-Id: I1278006b030253bf8e48dc7631d18985cdaa143d
Reviewed-on: https://go-review.googlesource.com/26659
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Keith Randall <khr@golang.org>

cmd/compile: remove conditional code dealing with two export formats

This removes some scaffolding introduced pre-1.7, introduced to
fix an export format bug, and to minimize conflicts with older
formats. The currently deployed and recognized format is "v1",
so don't worry about other versions. This is a step towards a
better scheme for internal export format versioning.

For #16244.

Change-Id: Ic7cf99dd2a24ad5484cc54aed44fa09332c2cf72
Reviewed-on: https://go-review.googlesource.com/27205
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Robert Griesemer <gri@golang.org>

cmd/compile: remove encoding of safemode bit from export data

Removes the encoding of this bit which was ignored but left behind
for 1.7 to minimize pre-1.7 export format changes. See the issue
for more details.

Fixes #15772.

Change-Id: I46cd7a66ad4c6003b78c64295cf3bda503ebf2dd
Reviewed-on: https://go-review.googlesource.com/27201
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

text/template: fix Parse when called twice with empty text

Fixes #16156

Change-Id: I6989db4fd392583a2d490339cefc525b07c11b90
Reviewed-on: https://go-review.googlesource.com/24380
Reviewed-by: Andrew Gerrand <adg@golang.org>
Run-TryBot: Andrew Gerrand <adg@golang.org>

cmd/internal/obj: reduce per-architecture opcode space

s390x took up the last available chunk of int16 opcodes.
There are RISC-V and sparc64 ports in progress out of tree,
and there will likely be other architectures.
Reduce the opcode space to allow more architectures to
fit without increasing to int32.

This is the smallest power of two that accomodates all
existing architectures. All else being equal, smaller is
better--smaller numbers are easier to generate immediates
for and easier on the eyes when debugging.

Change-Id: I4d0824b28913892fbd0579d3f90bea34e44c8946
Reviewed-on: https://go-review.googlesource.com/24223
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Minux Ma <minux@golang.org>

cmd/compile: remove support for textual export format

Fixes #15323.

Change-Id: I50e996e6fde6b24327cb45dd84da31deef4dcc56
Reviewed-on: https://go-review.googlesource.com/27171
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

runtime: fix map iterator concurrent map check

We should check whether there is a concurrent writer at the
start of every mapiternext, not just in mapaccessK (which is
only called during certain map growth situations).

Tests turned off by default because they are inherently flaky.

Fixes #16278

Change-Id: I8b72cab1b8c59d1923bec6fa3eabc932e4e91542
Reviewed-on: https://go-review.googlesource.com/24749
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>

test: remove unused variable

ssaMain is no longer needed.

Change-Id: I0b77f0bcd482329d73018bd80a6e068e622e191b
Reviewed-on: https://go-review.googlesource.com/27190
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/compile: use shorter versions of zero-extend ops

Only need to zero-extend to 32 bits and we get the top
32 bits zeroed for free.

Only the WQ change actually generates different code.
The assembler did this optimization for us in the other two cases.
But we might as well do it during SSA so -S output more closely
matches the actual generated instructions.

Change-Id: I3e4ac50dc4da124014d4e31c86e9fc539d94f7fd
Reviewed-on: https://go-review.googlesource.com/23711
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>

cmd/internal/obj: add opcode space safety check

This CL adds a safety mechanism
for changing the number of opcodes
available per architecture.

A subsequent CL will actually make the change.

Change-Id: I6332ed5514f2f153c54d11b7da0cc8a6be1c8066
Reviewed-on: https://go-review.googlesource.com/24222
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

doc/go1.7.html: fix name of tls.Config.NextProtos

Updates #16737

Change-Id: Ia51fc9b06df43b7c6f7136e90b40362263c20081
Reviewed-on: https://go-review.googlesource.com/27126
Reviewed-by: Chris Broadfoot <cbro@golang.org>

regexp: add some tests that were fixed in #12980

Also includes a minor golint cleanup in the tests.

Change-Id: I8c0fc81479e635e7cca18d5c48c28b654afa59d8
Reviewed-on: https://go-review.googlesource.com/25380
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/go: document -v flag for get

Fixes #16719.

Change-Id: I20550628814e3454f17d6f8ae8b66cce17f09859
Reviewed-on: https://go-review.googlesource.com/27118
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

cmd/compile: allow unsafe.Pointer(nil) as static data

Fixes #16306

Change-Id: If8e2f411fe9a5a5c198f10765fee7261ba8feaf2
Reviewed-on: https://go-review.googlesource.com/24836
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>

cmd/compile: accept literals in samesafeexpr

This only triggers a few places in the stdlib,
but it helps a lot when it does.

Before:

runtime.mapassign1 t=1 size=2400 args=0x20 locals=0xe0

After:

runtime.mapassign1 t=1 size=2352 args=0x20 locals=0xd8

name           old time/op  new time/op  delta
MapPop100-8    19.8µs ±11%  18.4µs ± 9%  -7.16%  (p=0.000 n=20+19)
MapPop1000-8    367µs ±17%   335µs ±11%  -8.63%  (p=0.000 n=19+19)
MapPop10000-8  7.29ms ±15%  6.86ms ±12%  -5.84%  (p=0.020 n=20+20)

Change-Id: I9faf32f95a6ba6a6d5d0818eab32cc271e01d10a
Reviewed-on: https://go-review.googlesource.com/26666
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

cmd/compile: fix uint<->float conversion on 386

The frontend rewriting lowers them to runtime calls on 386. It
matches explicitly uint32, but missed uint.

Fixes #16738.

Change-Id: Iece7a45edf74615baca052a53273c208f057636d
Reviewed-on: https://go-review.googlesource.com/27085
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/compile: add size hint to map literal allocations

Might as well tell the runtime how large the map is going to be.
This avoids grow work and allocations while the map is being built.

Will wait for 1.8.

Fixes #15880
Fixes #16279

Change-Id: I377e3e5ec1e2e76ea2a50cc00810adda20ad0e79
Reviewed-on: https://go-review.googlesource.com/23558
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>

cmd/internal/obj/x86: minor code cleanup

Update #16415

Change-Id: I83e0966931ada2f1ed02304685bb45effdd71268
Reviewed-on: https://go-review.googlesource.com/26665
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

cmd/compile: optimize bool to int conversion

This CL teaches SSA to recognize code of the form

// b is a boolean value, i is an int of some flavor
if b {
i = 1
} else {
i = 0
}

and use b's underlying 0/1 representation for i
instead of generating jumps.

Unfortunately, it does not work on the obvious code:

func bool2int(b bool) int {
if b {
return 1
}
return 0
}

This is left for future work.
Note that the existing phiopt optimizations also don't work for:

func neg(b bool) bool {
if b {
return false
}
return true
}

In the meantime, runtime authors and the like can use:

func bool2int(b bool) int {
var i int
if b {
i = 1
} else {
i = 0
}
return i
}

This compiles to:

"".bool2int t=1 size=16 args=0x10 locals=0x0
0x0000 00000 (x.go:25) TEXT "".bool2int(SB), $0-16
0x0000 00000 (x.go:25) FUNCDATA $0, gclocals·23e8278e2b69a3a75fa59b23c49ed6ad(SB)
0x0000 00000 (x.go:25) FUNCDATA $1, gclocals·33cdeccccebe80329f1fdbee7f5874cb(SB)
0x0000 00000 (x.go:32) MOVBLZX "".b+8(FP), AX
0x0005 00005 (x.go:32) MOVBQZX AL, AX
0x0008 00008 (x.go:32) MOVQ AX, "".~r1+16(FP)
0x000d 00013 (x.go:32) RET

The extraneous MOVBQZX is #15300.

This optimization also helps range and slice.
The compiler must protect against pointers pointing
to the end of a slice/string. It does this by increasing
a pointer by either 0 or 1 * elemsize, based on a condition.
This CL optimizes away a jump in that code.

This CL triggers 382 times while compiling the standard library.

Updating code to utilize this optimization is left for future CLs.

Updates #6011

Change-Id: Ia7c1185f8aa223c543f91a3cd6d4a2a09c691c70
Reviewed-on: https://go-review.googlesource.com/22711
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

cmd/compile: don't crash when exporting self-recursive interfaces

For #16369.

Change-Id: I4c9f5a66b95558adcc1bcface164b9b2b4382d2f
Reviewed-on: https://go-review.googlesource.com/24979
Reviewed-by: Alan Donovan <adonovan@google.com>

reflect: clear tflag on new types

Fixes #16722

Change-Id: I50a0e69d3e79d13bc1860cd983267c3db087a4b8
Reviewed-on: https://go-review.googlesource.com/27119
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/compile: don't fold >32bit constants into a MULQ

Don't fold constant factors into a multiply
beyond the capacity of a MULQ instruction (32 bits).

Fixes #16733

Change-Id: Idc213c6cb06f7c94008a8cf9e60a9e77d085fd89
Reviewed-on: https://go-review.googlesource.com/27160
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/compile: inline _, ok = i.(T)

We already inlined

_, ok = e.(T)
_, ok = i.(E)
_, ok = e.(E)

The only ok-only variants not inlined are now

_, ok = i.(I)
_, ok = e.(I)

These call getitab, so are non-trivial.

Change-Id: Ie45fd8933ee179a679b92ce925079b94cff0ee12
Reviewed-on: https://go-review.googlesource.com/26658
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

cmd/internal/obj: convert Aconv to a stringer

Now that assembler opcodes have their own type, they can have a true
stringer, rather than explicit calls to Aconv, which makes for nicer
format strings.

Change-Id: Ic77f5f8ac38b4e519dcaa08c93e7b732226f7bfe
Reviewed-on: https://go-review.googlesource.com/25045
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>

cmd/compile/internal/big: re-vendor

Pick up a bunch of changes and fixes.

Change-Id: If4101f7185d433a4c89096bc786ee5de8eeabac0
Reviewed-on: https://go-review.googlesource.com/27123
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

runtime: move printing of extra newline

No functional changes, makes vet happy.

Updates #11041

Change-Id: I59f3aba46d19b86d605508978652d76a1fe7ac7b
Reviewed-on: https://go-review.googlesource.com/27125
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

net/http: use keyed composite literal

Makes vet happy.

Updates #11041

Change-Id: I23ca413c03ff387359440af8114786cd7880a048
Reviewed-on: https://go-review.googlesource.com/27124
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

crypto/cipher, math/big: fix example names

Fixes (legit) vet warnings.
Fix some verb tenses while we're here.

Updates #11041

Change-Id: I27e995f55b38f4cf584e97a67b8545e8247e83d6
Reviewed-on: https://go-review.googlesource.com/27122
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: David Crawshaw <crawshaw@golang.org>

crypto/tls: fix WriteTo method signature

Give *recordingConn the correct WriteTo signature
to be an io.WriterTo. This makes vet happy.
It also means that it'll report errors,
which were previously being ignored.

Updates #11041

Change-Id: I13f171407d63f4b62427679bff362eb74faddca5
Reviewed-on: https://go-review.googlesource.com/27121
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

net: change t.Error to t.Errorf

Caught by vet.

Updates #11041

Change-Id: I4dbb2eeaf633eea5976074840064edc2349e01d8
Reviewed-on: https://go-review.googlesource.com/27120
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: David Crawshaw <crawshaw@golang.org>

container/list: silence vet warnings

container/list/list_test.go:274: self-assignment of e1 to e1
container/list/list_test.go:274: self-assignment of e4 to e4
container/list/list_test.go:282: self-assignment of e1 to e1
container/list/list_test.go:286: self-assignment of e1 to e1
container/list/list_test.go:286: self-assignment of e4 to e4

Updates #11041

Change-Id: Ibd90cf6a924e93497908f437b814c3fc82937f4a
Reviewed-on: https://go-review.googlesource.com/27114
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

cmd/compile: remove nil check in accessing PAUTOHEAP variable

CL 23393 introduces PAUTOHEAP, and access of PAUTOHEAP variable is
rewritten to indirection of a PAUTO variable. Mark this variable
non-nil, so this indirection does not introduce extra nil checks.

Change-Id: I31853eed5e60238b6c5bc0546e2e9ab340dcddd9
Reviewed-on: https://go-review.googlesource.com/26831
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

all: fix "result not used" vet warnings

For tests, assign to _.
For benchmarks, assign to a sink.

Updates #11041

Change-Id: I87c5543245c7bc74dceb38902f4551768dd37948
Reviewed-on: https://go-review.googlesource.com/27116
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/compile: fix bad generated format strings in test

We were generating format strings containing
a lone %. Vet legitimately complains:

cmd/compile/internal/gc/constFold_test.go:339: unrecognized printf verb ' '

The fix doesn't make for very readable code,
but it is simple and obviously correct.

Updates #11041

Change-Id: I90bd2d1d140887f5229752a279f7e46921472fbb
Reviewed-on: https://go-review.googlesource.com/27115
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

syscall: unify unix/amd64 asm implementations

Updates #11041

Change-Id: I77e5ca0b61ffc530ee46848721a177867c81d548
Reviewed-on: https://go-review.googlesource.com/25116
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

syscall: split out unix Syscall9 asm support

This is preliminary work to unifying them.
Aside from Syscall9, all are identical.
Syscall9 has a netbsd/openbsd variant
and a dragonfly/freebsd variant.

Updates #11041

Change-Id: Ia5ce95d5e9115d4c0492d5e53aa7a4316deafd1f
Reviewed-on: https://go-review.googlesource.com/25115
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

syscall: superficial cleanup of amd64 unix assembly

This is preliminary work to unifying them.

Updates #11041

Change-Id: Ibe83da3d626f1da9e8888e26cedd3af2152b42e6
Reviewed-on: https://go-review.googlesource.com/25114
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

syscall: fix dragonfly/amd64 assembly argument sizes

This is preliminary work to unifying the
unix amd64 assembly implementations,
which is preliminary work to making the
assembly vet-friendly.

Updates #11041

Change-Id: Ic64985124f8fb86cc08898be2ec7fca972ced4ca
Reviewed-on: https://go-review.googlesource.com/25113
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

syscall: unify unix 386 implementations

They were identical.

This will allow us to do the TODO at the top
of the file only once.

Updates #11041

Change-Id: I07aaca27ae46b66b65780082988bdc7546ed534b
Reviewed-on: https://go-review.googlesource.com/25112
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/compile: disable Duff's device on darwin/arm64

Darwin linker does not support BR26 reloc with non-zero addend.

Fixes #16724.

Change-Id: I1b5b4dc7159141bde3e273490f435c08c583afaf
Reviewed-on: https://go-review.googlesource.com/27081
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dave Cheney <dave@cheney.net>

net/http: make Transport retry non-idempotent requests if no bytes written

If the server failed on us before we even tried to write any bytes,
it's safe to retry the request on a new connection, regardless of the
HTTP method/idempotence.

Fixes #15723

Change-Id: I25360f82aac530d12d2b3eef02c43ced86e62906
Reviewed-on: https://go-review.googlesource.com/27117
Reviewed-by: Andrew Gerrand <adg@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

cmd/compile: reduce garbage from autolabel

Follow-up to CL 26661

Change-Id: I67c58d17313094675cf0f30ce50d486818ae0dcb
Reviewed-on: https://go-review.googlesource.com/27113
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

archive/tar: preallocate slice from paxHeaders

Preallocate keys slice with the length of paxHeaders map
to prevent slice growth with append operations.

Change-Id: Ic9a927c4eaa775690a4ef912d61dd06f38e11510
Reviewed-on: https://go-review.googlesource.com/23782
Reviewed-by: Joe Tsai <thebrokentoaster@gmail.com>
Run-TryBot: Joe Tsai <thebrokentoaster@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

crypto/tls: gofmt -w -s

Change-Id: Iedf9000e3bb1fa73b4c3669eae846e85f1f5fdfe
Reviewed-on: https://go-review.googlesource.com/24489
Run-TryBot: Mikio Hara <mikioh.mikioh@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

net/http: deflake BenchmarkClient and its use of a fixed port for testing

Let the kernel pick a port for testing, and have the server in the
child process tell the parent (benchmarking) process the port that
was selected.

Fixes flakes like seen in https://golang.org/cl/27050 (and previously)

Change-Id: Ia2b705dc4152f70e0a5725015bdae09984d09d53
Reviewed-on: https://go-review.googlesource.com/27051
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

go/token: Fix race in FileSet.PositionFor.

Methods of FileSet are documented to be safe for concurrent use by
multiple goroutines, so FileSet is protected by a mutex and all its
methods use it to prevent concurrent mutations. All methods of File that
mutate the respective FileSet, including AddLine, do also lock its
mutex, but that does not help when PositionFor is invoked concurrently
and reads without synchronization what AddLine mutates.

The change adds acquiring a RLock around the racy call of File.position
and the respective test.

Fixes #16548

Change-Id: Iecaaa02630b2532cb29ab555376633ee862315dd
Reviewed-on: https://go-review.googlesource.com/25345
Reviewed-by: Robert Griesemer <gri@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

bytes: add examples

`bytes` and `strings` are pretty similar to each other, this commit
brings `strings` examples to its counter-part.

Partially addresses #16360

Change-Id: I551320eaa78be9df69012035f1c3333f500e04c9
Reviewed-on: https://go-review.googlesource.com/25062
Reviewed-by: Andrew Gerrand <adg@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

doc: add 1.7 to golang.org/project

Change-Id: Ib17f6643efd49e2bca188c4faa505f79832d18b1
Reviewed-on: https://go-review.googlesource.com/27110
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

os/exec: fix nit found by vet

Change-Id: I8085ed43d63215237a4871cc1e44257132a7f5de
Reviewed-on: https://go-review.googlesource.com/27130
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>