Commits · b548eee3d96fc0b6e962a243b28121e1f37ad792 · go / golang

13 Feb, 2017 7 commits

cmd/compile: fix load-combining rules · b548eee3

Keith Randall authored Feb 13, 2017

CL 33632 reorders args of commutative ops in order to make
CSE for commutative ops more robust.  Unfortunately, that
broke the load-combining rules which depend on a certain ordering
of OR ops' arguments.

Introduce some additional rules that order OR ops' arguments
consistently so that the load-combining rules fire.

Note: there's also something else wrong with the s390x rules.
I've filed #19059 for that.

Fixes #18946

Change-Id: I0a5447196bd88a55ccee683c69a57b943a9972e1
Reviewed-on: https://go-review.googlesource.com/36911
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>

b548eee3

cmd/trace: document the final step to use pprof-like profiles · 76b4b8c7

Jaana Burcu Dogan authored Feb 11, 2017

The tutorial ends without mentioning how to use the generated
pprof-like profile with the pprof tool. This may be very trivial
for users who are already very familiar with the Go tools, but
for the newcomers, it saves a lot of time to finalize the tutorial
with an example of `go tool pprof` invocation.

Change-Id: Idf034eb4bfb9672ef10190e66fcbf873e8f08f6a
Reviewed-on: https://go-review.googlesource.com/36803Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>

76b4b8c7

cmd/compile: cull some dead arch-specific Ops · c5fed5bb

Josh Bleecher Snyder authored Feb 13, 2017

Change-Id: Iee7daa5b91b7896ce857321e307f2ee47b7f095f
Reviewed-on: https://go-review.googlesource.com/36906
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

c5fed5bb

cmd/compile: optimize non-empty-interface type conversions · 5a75d6a0

Keith Randall authored Jan 04, 2017

When doing i.(T) for non-empty-interface i and concrete type T,
there's no need to read the type out of the itab. Just compare the
itab to the itab we expect for that interface/type pair.

Also optimize type switches by putting the type hash of the
concrete type in the itab. That way we don't need to load the
type pointer out of the itab.

Update #18492

Change-Id: I49e280a21e5687e771db5b8a56b685291ac168ce
Reviewed-on: https://go-review.googlesource.com/34810
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: David Chase <drchase@google.com>

5a75d6a0

cmd/compile/internal/parser: don't crash after unexpected token · ee2f5faf

Robert Griesemer authored Feb 13, 2017

Added missing nil-check. We will get rid of the gcCompat corrections
shortly but it's still worthwhile having the new test case added.

Fixes #19056.

Change-Id: I35bd938a4d789058da15724e34c05e5e631ecad0
Reviewed-on: https://go-review.googlesource.com/36908
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

ee2f5faf

runtime: add Frames example · 8da91a62

Josh Bleecher Snyder authored Feb 11, 2017

Based on sample code from iant.

Fixes #18788.

Change-Id: I6bb33ed05af2538fbde42ddcac629280ef7c00a6
Reviewed-on: https://go-review.googlesource.com/36892
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

8da91a62

net: use bytes.Equal instead of bytesEqual · 39fcf8bf

Erik Dubbelboer authored Feb 09, 2017

bytes.Equal is written in assembly and is slightly faster than the
current Go bytesEqual from the net package.

benchcmp:
benchmark                 old ns/op     new ns/op     delta
BenchmarkIPCompare4-8     7.74          7.01          -9.43%
BenchmarkIPCompare6-8     8.47          6.86          -19.01%

Change-Id: I2a7ad35867489b46f0943aef5776a2fe1b46e2df
Reviewed-on: https://go-review.googlesource.com/36850Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

39fcf8bf

12 Feb, 2017 3 commits

path/filepath: add test for directory junction walk · 61bf0d1c

Alex Brainman authored Feb 12, 2017

For #10424.

Change-Id: Ie4e87503b0ed04f65d2444652bd1db647d3529f4
Reviewed-on: https://go-review.googlesource.com/36851Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

61bf0d1c

runtime: use two-level list for semaphore address search in semaRoot · 45c6f59e

Russ Cox authored Feb 10, 2017

If there are many goroutines contending for two different locks
and both locks hash to the same semaRoot, the scans to find the
goroutines for a particular lock can end up being O(n), making
n lock acquisitions quadratic.

As long as only one actively-used lock hashes to each semaRoot
there's no problem, since the list operations in that case are O(1).
But when the second actively-used lock hits the same semaRoot,
then scans for entries with for a given lock have to scan over the
entries for the other lock.

Fix this problem by changing the semaRoot to hold only one sudog
per unique address. In the running example, this drops the length of
that list from O(n) to 2. Then attach other goroutines waiting on the
same address to a separate list headed by the sudog in the semaRoot list.
Those "same address list" operations are still O(1), so now the
example from above works much better.

There is still an assumption here that in real programs you don't have
many many goroutines queueing up on many many distinct addresses.
If we end up with that problem, we can replace the top-level list with
a treap.

Fixes #17953.

Change-Id: I78c5b1a5053845275ab31686038aa4f6db5720b2
Reviewed-on: https://go-review.googlesource.com/36792
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

45c6f59e

image/png: reduce memory allocs encoding images by reusing buffers · 93a18acf

Cezar Sa Espinola authored Dec 08, 2016

This change allows greatly reducing memory allocations with a slightly
performance improvement as well.

Instances of (*png).Encoder can have a optional BufferPool attached to
them. This allows reusing temporary buffers used when encoding a new
image. This buffers include instances to zlib.Writer and bufio.Writer.

Also, buffers for current and previous rows are saved in the encoder
instance and reused as long as their cap() is enough to fit the current
image row.

A new benchmark was added to demonstrate the performance improvement
when setting a BufferPool to an Encoder instance:

$ go test -bench BenchmarkEncodeGray -benchmem
BenchmarkEncodeGray-4                 	    1000	   2349584 ns/op	 130.75 MB/s	  852230 B/op	      32 allocs/op
BenchmarkEncodeGrayWithBufferPool-4   	    1000	   2241650 ns/op	 137.04 MB/s	     900 B/op	       3 allocs/op

Change-Id: I4488201ae53cb2ad010c68c1e0118ee12beae14e
Reviewed-on: https://go-review.googlesource.com/34150Reviewed-by: Nigel Tao <nigeltao@golang.org>
Run-TryBot: Nigel Tao <nigeltao@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

93a18acf

11 Feb, 2017 9 commits

cmd/internal/obj/x86: add comments to wrapper prologue insertion · 5030bfdf

Josh Bleecher Snyder authored Feb 10, 2017

Make the comments a bit clearer and more accurate,
in anticipation of updating the code.

Change-Id: I1111e6c3405a8688fcd29b809a48a762ff41edaa
Reviewed-on: https://go-review.googlesource.com/36833Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

5030bfdf

cmd/compile: make panicwrap argument-free · 2c91bb4c

Josh Bleecher Snyder authored Feb 11, 2017

When code defines a method on T,
the compiler generates a corresponding wrapper method on *T.
The first thing the wrapper does is check whether
the pointer is nil and if so, call panicwrap.
This is done to provide a useful error message.

The existing implementation gets its information
from arguments set up by the compiler.
However, with some trouble, this information can
be extracted from the name of the wrapper method itself.

Removing the arguments to panicwrap simplifies and
shrinks the wrapper method.
It also means that the call to panicwrap does not
require any stack space.
This enables a further optimization on amd64/x86,
which is to skip the function prologue if nothing
else in the method requires stack space.
This is frequently the case in simple, hot methods,
such as Less and Swap in sort.Interface implementations.

Fixes #19040.

Benchmarks for package sort on amd64:

name                  old time/op  new time/op  delta
SearchWrappers-8       104ns ± 1%   104ns ± 1%    ~     (p=0.286 n=27+27)
SortString1K-8         128µs ± 1%   128µs ± 1%  -0.44%  (p=0.004 n=30+30)
SortString1K_Slice-8   118µs ± 2%   117µs ± 1%    ~     (p=0.106 n=30+30)
StableString1K-8      18.6µs ± 1%  18.6µs ± 1%    ~     (p=0.446 n=28+26)
SortInt1K-8           65.9µs ± 1%  60.7µs ± 1%  -7.96%  (p=0.000 n=28+30)
StableInt1K-8         75.3µs ± 2%  72.8µs ± 1%  -3.41%  (p=0.000 n=30+30)
StableInt1K_Slice-8   57.7µs ± 1%  57.7µs ± 1%    ~     (p=0.515 n=30+30)
SortInt64K-8          6.28ms ± 1%  6.01ms ± 1%  -4.19%  (p=0.000 n=28+28)
SortInt64K_Slice-8    5.04ms ± 1%  5.04ms ± 1%    ~     (p=0.927 n=28+27)
StableInt64K-8        6.65ms ± 1%  6.38ms ± 1%  -3.97%  (p=0.000 n=26+30)
Sort1e2-8             37.9µs ± 1%  37.2µs ± 1%  -1.89%  (p=0.000 n=29+27)
Stable1e2-8           77.0µs ± 1%  74.7µs ± 1%  -3.06%  (p=0.000 n=27+30)
Sort1e4-8             8.21ms ± 2%  7.98ms ± 1%  -2.77%  (p=0.000 n=29+30)
Stable1e4-8           24.8ms ± 1%  24.3ms ± 1%  -2.31%  (p=0.000 n=28+30)
Sort1e6-8              1.27s ± 4%   1.22s ± 1%  -3.42%  (p=0.000 n=30+29)
Stable1e6-8            5.06s ± 1%   4.92s ± 1%  -2.77%  (p=0.000 n=25+29)
[Geo mean]             731µs        714µs       -2.29%

Before/after assembly for sort.(*intPairs).Less follows.
It can be optimized further, but that's for a follow-up CL.

Before:

"".(*intPairs).Less t=1 size=214 args=0x20 locals=0x38
	0x0000 00000 (<autogenerated>:1)	TEXT	"".(*intPairs).Less(SB), $56-32
	0x0000 00000 (<autogenerated>:1)	MOVQ	(TLS), CX
	0x0009 00009 (<autogenerated>:1)	CMPQ	SP, 16(CX)
	0x000d 00013 (<autogenerated>:1)	JLS	204
	0x0013 00019 (<autogenerated>:1)	SUBQ	$56, SP
	0x0017 00023 (<autogenerated>:1)	MOVQ	BP, 48(SP)
	0x001c 00028 (<autogenerated>:1)	LEAQ	48(SP), BP
	0x0021 00033 (<autogenerated>:1)	MOVQ	32(CX), BX
	0x0025 00037 (<autogenerated>:1)	TESTQ	BX, BX
	0x0028 00040 (<autogenerated>:1)	JEQ	55
	0x002a 00042 (<autogenerated>:1)	LEAQ	64(SP), DI
	0x002f 00047 (<autogenerated>:1)	CMPQ	(BX), DI
	0x0032 00050 (<autogenerated>:1)	JNE	55
	0x0034 00052 (<autogenerated>:1)	MOVQ	SP, (BX)
	0x0037 00055 (<autogenerated>:1)	NOP
	0x0037 00055 (<autogenerated>:1)	FUNCDATA	$0, gclocals·4032f753396f2012ad1784f398b170f4(SB)
	0x0037 00055 (<autogenerated>:1)	FUNCDATA	$1, gclocals·69c1753bd5f81501d95132d08af04464(SB)
	0x0037 00055 (<autogenerated>:1)	MOVQ	""..this+64(FP), AX
	0x003c 00060 (<autogenerated>:1)	TESTQ	AX, AX
	0x003f 00063 (<autogenerated>:1)	JEQ	$0, 135
	0x0041 00065 (<autogenerated>:1)	MOVQ	(AX), CX
	0x0044 00068 (<autogenerated>:1)	MOVQ	8(AX), AX
	0x0048 00072 (<autogenerated>:1)	MOVQ	"".i+72(FP), DX
	0x004d 00077 (<autogenerated>:1)	CMPQ	DX, AX
	0x0050 00080 (<autogenerated>:1)	JCC	$0, 128
	0x0052 00082 (<autogenerated>:1)	SHLQ	$4, DX
	0x0056 00086 (<autogenerated>:1)	MOVQ	(CX)(DX*1), DX
	0x005a 00090 (<autogenerated>:1)	MOVQ	"".j+80(FP), BX
	0x005f 00095 (<autogenerated>:1)	CMPQ	BX, AX
	0x0062 00098 (<autogenerated>:1)	JCC	$0, 128
	0x0064 00100 (<autogenerated>:1)	SHLQ	$4, BX
	0x0068 00104 (<autogenerated>:1)	MOVQ	(CX)(BX*1), AX
	0x006c 00108 (<autogenerated>:1)	CMPQ	DX, AX
	0x006f 00111 (<autogenerated>:1)	SETLT	AL
	0x0072 00114 (<autogenerated>:1)	MOVB	AL, "".~r2+88(FP)
	0x0076 00118 (<autogenerated>:1)	MOVQ	48(SP), BP
	0x007b 00123 (<autogenerated>:1)	ADDQ	$56, SP
	0x007f 00127 (<autogenerated>:1)	RET
	0x0080 00128 (<autogenerated>:1)	PCDATA	$0, $1
	0x0080 00128 (<autogenerated>:1)	CALL	runtime.panicindex(SB)
	0x0085 00133 (<autogenerated>:1)	UNDEF
	0x0087 00135 (<autogenerated>:1)	LEAQ	go.string."sort_test"(SB), AX
	0x008e 00142 (<autogenerated>:1)	MOVQ	AX, (SP)
	0x0092 00146 (<autogenerated>:1)	MOVQ	$9, 8(SP)
	0x009b 00155 (<autogenerated>:1)	LEAQ	go.string."intPairs"(SB), AX
	0x00a2 00162 (<autogenerated>:1)	MOVQ	AX, 16(SP)
	0x00a7 00167 (<autogenerated>:1)	MOVQ	$8, 24(SP)
	0x00b0 00176 (<autogenerated>:1)	LEAQ	go.string."Less"(SB), AX
	0x00b7 00183 (<autogenerated>:1)	MOVQ	AX, 32(SP)
	0x00bc 00188 (<autogenerated>:1)	MOVQ	$4, 40(SP)
	0x00c5 00197 (<autogenerated>:1)	PCDATA	$0, $1
	0x00c5 00197 (<autogenerated>:1)	CALL	runtime.panicwrap(SB)
	0x00ca 00202 (<autogenerated>:1)	UNDEF
	0x00cc 00204 (<autogenerated>:1)	NOP
	0x00cc 00204 (<autogenerated>:1)	PCDATA	$0, $-1
	0x00cc 00204 (<autogenerated>:1)	CALL	runtime.morestack_noctxt(SB)
	0x00d1 00209 (<autogenerated>:1)	JMP	0

After:

"".(*intPairs).Swap t=1 size=147 args=0x18 locals=0x8
	0x0000 00000 (<autogenerated>:1)	TEXT	"".(*intPairs).Swap(SB), $8-24
	0x0000 00000 (<autogenerated>:1)	MOVQ	(TLS), CX
	0x0009 00009 (<autogenerated>:1)	SUBQ	$8, SP
	0x000d 00013 (<autogenerated>:1)	MOVQ	BP, (SP)
	0x0011 00017 (<autogenerated>:1)	LEAQ	(SP), BP
	0x0015 00021 (<autogenerated>:1)	MOVQ	32(CX), BX
	0x0019 00025 (<autogenerated>:1)	TESTQ	BX, BX
	0x001c 00028 (<autogenerated>:1)	JEQ	43
	0x001e 00030 (<autogenerated>:1)	LEAQ	16(SP), DI
	0x0023 00035 (<autogenerated>:1)	CMPQ	(BX), DI
	0x0026 00038 (<autogenerated>:1)	JNE	43
	0x0028 00040 (<autogenerated>:1)	MOVQ	SP, (BX)
	0x002b 00043 (<autogenerated>:1)	NOP
	0x002b 00043 (<autogenerated>:1)	FUNCDATA	$0, gclocals·e6397a44f8e1b6e77d0f200b4fba5269(SB)
	0x002b 00043 (<autogenerated>:1)	FUNCDATA	$1, gclocals·69c1753bd5f81501d95132d08af04464(SB)
	0x002b 00043 (<autogenerated>:1)	MOVQ	""..this+16(FP), AX
	0x0030 00048 (<autogenerated>:1)	TESTQ	AX, AX
	0x0033 00051 (<autogenerated>:1)	JEQ	$0, 140
	0x0035 00053 (<autogenerated>:1)	MOVQ	(AX), CX
	0x0038 00056 (<autogenerated>:1)	MOVQ	8(AX), AX
	0x003c 00060 (<autogenerated>:1)	MOVQ	"".i+24(FP), DX
	0x0041 00065 (<autogenerated>:1)	CMPQ	DX, AX
	0x0044 00068 (<autogenerated>:1)	JCC	$0, 133
	0x0046 00070 (<autogenerated>:1)	SHLQ	$4, DX
	0x004a 00074 (<autogenerated>:1)	MOVQ	8(CX)(DX*1), BX
	0x004f 00079 (<autogenerated>:1)	MOVQ	(CX)(DX*1), SI
	0x0053 00083 (<autogenerated>:1)	MOVQ	"".j+32(FP), DI
	0x0058 00088 (<autogenerated>:1)	CMPQ	DI, AX
	0x005b 00091 (<autogenerated>:1)	JCC	$0, 133
	0x005d 00093 (<autogenerated>:1)	SHLQ	$4, DI
	0x0061 00097 (<autogenerated>:1)	MOVQ	8(CX)(DI*1), AX
	0x0066 00102 (<autogenerated>:1)	MOVQ	(CX)(DI*1), R8
	0x006a 00106 (<autogenerated>:1)	MOVQ	R8, (CX)(DX*1)
	0x006e 00110 (<autogenerated>:1)	MOVQ	AX, 8(CX)(DX*1)
	0x0073 00115 (<autogenerated>:1)	MOVQ	SI, (CX)(DI*1)
	0x0077 00119 (<autogenerated>:1)	MOVQ	BX, 8(CX)(DI*1)
	0x007c 00124 (<autogenerated>:1)	MOVQ	(SP), BP
	0x0080 00128 (<autogenerated>:1)	ADDQ	$8, SP
	0x0084 00132 (<autogenerated>:1)	RET
	0x0085 00133 (<autogenerated>:1)	PCDATA	$0, $1
	0x0085 00133 (<autogenerated>:1)	CALL	runtime.panicindex(SB)
	0x008a 00138 (<autogenerated>:1)	UNDEF
	0x008c 00140 (<autogenerated>:1)	PCDATA	$0, $1
	0x008c 00140 (<autogenerated>:1)	CALL	runtime.panicwrap(SB)
	0x0091 00145 (<autogenerated>:1)	UNDEF

Change-Id: I15bb8435f0690badb868799f313ed8817335efd3
Reviewed-on: https://go-review.googlesource.com/36809
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

2c91bb4c

testing: fix copy/paste in docs · 61e963e9

Josh Bleecher Snyder authored Feb 11, 2017

Follow-up to CL 36791.

Change-Id: I1c4831e5dfe90c205782e970ada7faff8a009daa
Reviewed-on: https://go-review.googlesource.com/36890Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

61e963e9

cmd/compile: Ensure left-to-right assignment · 1cde87b3

Dhananjay Nakrani authored Dec 24, 2016

Add temporaries to reorder the assignment for OAS2XXX nodes.
This makes orderstmt(), rewrite
  a, b, c = ...
as
  tmp1, tmp2, tmp3 = ...
  a, b, c = tmp1, tmp2, tmp3
and
  a, ok = ...
as
  t1, t2 = ...
  a  = t1
  ok = t2

Fixes #13433.

Change-Id: Id0f5956e3a254d0a6f4b89b5f7b0e055b1f0e21f
Reviewed-on: https://go-review.googlesource.com/34713
Run-TryBot: Dhananjay Nakrani <dhananjayn@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

1cde87b3

doc: improve issue template and contribution guidelines · e0d50a58

Paul Jolly authored Dec 09, 2016

Encourage people towards the various help forums as a first port of
call. Better sign-posting will reduce the incidence or questions being
asked in the issue tracker that should otherwise be handled elsewhere,
thereby keeping the issue tracker email traffic more focussed.

Change-Id: I13b2e498d88be010fca421067ae6fb579a46d6b7
Reviewed-on: https://go-review.googlesource.com/34250Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

e0d50a58

strings: make parameters names less confusing · 272ec231

Alberto Donizetti authored Feb 11, 2017

Using 'sep' as parameter name for strings functions that take a
separator argument is fine, but for functions like Index or Count that
look for a substring it's better to use 'substr' (like Contains
already does).

Fixes #19039

Change-Id: Idd557409c8fea64ce830ab0e3fec37d3d56a79f0
Reviewed-on: https://go-review.googlesource.com/36874
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

272ec231

path/filepath: fix TestWinSplitListTestsAreValid on some systems · 2ceeb7b0

Remi Gillig authored Feb 11, 2017

The environment variables used in those tests override the default
OS ones. However, one of them (SystemRoot) seems to be required on
some Windows systems for invoking cmd.exe properly.

This fixes #4930 and #6568.

Change-Id: I23dfb67c1de86020711a3b59513f6adcbba12561
Reviewed-on: https://go-review.googlesource.com/36873Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

2ceeb7b0

cmd/nm: extend help text to document the flags · 60d7d247

Jaana Burcu Dogan authored Feb 10, 2017

Change-Id: Ia2852666ef44e7ef0bba2360e92caccc83fd0e5c
Reviewed-on: https://go-review.googlesource.com/36796Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

60d7d247

testing: only call ReadMemStats if necessary when benchmarking · 948b21a3

Josh Bleecher Snyder authored Feb 10, 2017

When running benchmarks with -cpuprofile,
the entire process gets profiled,
and ReadMemStats is surprisingly expensive.
Running the sort benchmarks right now with
-cpuprofile shows almost half of all execution
time in ReadMemStats.

Since ReadMemStats is not required if the benchmark
does not need allocation stats, simply skip it.
This will make cpu profiles nicer to read
and significantly speed up the process of running benchmarks.
It might also make sense to toggle cpu profiling
on/off as we begin/end individual benchmarks,
but that wouldn't get us the time savings of
skipping ReadMemStats, so this CL is useful in itself.

Change-Id: I425197b1ee11be4bc91d22b929e2caf648ebd7c5
Reviewed-on: https://go-review.googlesource.com/36791
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

948b21a3

10 Feb, 2017 21 commits

image/color: tweak the YCbCr to RGBA conversion formula again. · 5157039f

Nigel Tao authored Feb 10, 2017

The 0x10101 magic constant is a little more principled than 0x10100, as
the rounding adjustment now spans the complete range [0, 0xffff] instead
of [0, 0xff00].

Consider this round-tripping code:

y, cb, cr := color.RGBToYCbCr(r0, g0, b0)
r1, g1, b1 := color.YCbCrToRGB(y, cb, cr)

Due to rounding errors both ways, we often but not always get a perfect
round trip (where r0 == r1 && g0 == g1 && b0 == b1). This is true both
before and after this commit. In some cases we got luckier, in others we
got unluckier.

For example, before this commit, (180, 135, 164) doesn't round trip
perfectly (it's off by 1) but (180, 135, 165) does. After this commit,
both cases are reversed: the former does and the latter doesn't (again
off by 1). Over all possible (r, g, b) triples, there doesn't seem to be
a big change for better or worse.

There is some history in these CLs:

image/color: tweak the YCbCr to RGBA conversion formula.
https://go-review.googlesource.com/#/c/12220/2/src/image/color/ycbcr.go

image/color: have YCbCr.RGBA work in 16-bit color, per the Color
interface.
https://go-review.googlesource.com/#/c/8073/2/src/image/color/ycbcr.go

Change-Id: Ib25ba7039f49feab2a9d1a4141b86db17db7b3e1
Reviewed-on: https://go-review.googlesource.com/36732
Run-TryBot: Nigel Tao <nigeltao@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

5157039f

cmd/compile: eliminate OASWB · bdb9b945

Matthew Dempsky authored Dec 19, 2016

Instead we can just call needwritebarrier when constructing the SSA
representation.

Change-Id: I6fefaad49daada9cdb3050f112889e49dca0047b
Reviewed-on: https://go-review.googlesource.com/34566Reviewed-by: Cherry Zhang <cherryyz@google.com>

bdb9b945

cmd/go: copy FFLAGS from build.Package · e7ec06e0

Ian Lance Taylor authored Feb 07, 2017

Fixes #18975.

Change-Id: I60dfb299233ecfed4b2da93750ea84e7921f1fbb
Reviewed-on: https://go-review.googlesource.com/36482
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

e7ec06e0

expvar: make BenchmarkAdd{Same,Different} comparable to 1.8 · aa153879

Bryan C. Mills authored Feb 10, 2017

bradfitz noted in change 36717 that the new behavior was no longer
comparable with the old.  This change restores comparable behavior
for -cpu=1.

BenchmarkMapAddSame                 909           909           +0.00%
BenchmarkMapAddSame-6               1309          262           -79.98%
BenchmarkMapAddDifferent            2856          3030          +6.09%
BenchmarkMapAddDifferent-6          3803          581           -84.72%

updates #18177

Change-Id: Ifaff5a1f48be92002d86c296220313b7efdc81d6
Reviewed-on: https://go-review.googlesource.com/36723Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

aa153879

cmd/go: use Bitbucket v2 REST API when determining VCS · 275e1fdb

Mark Adams authored Feb 03, 2017

The existing implementation uses v1.0 of Bitbucket's REST API. The newer
version 2.0 of Bitbucket's REST API provides the same information but
with support for partial responses allowing the client to request only
the response fields that are relevant to their usage of the API
resulting in a much smaller payload size.

The partial response functionality in the Bitbucket API is documented here:
https://developer.atlassian.com/bitbucket/api/2/reference/meta/partial-response

The v2.0 of the Bitbucket repositories API is documented here:
https://developer.atlassian.com/bitbucket/api/2/reference/resource/repositories/%7Busername%7D/%7Brepo_slug%7D#get

Fixes #18919

Change-Id: I319947d5c51adc241cfe3a2228a667cc43fb1f56
Reviewed-on: https://go-review.googlesource.com/36219
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

275e1fdb

go/types: unsafe.Pointer is not an alias · a896869a

Alan Donovan authored Feb 09, 2017

Change-Id: Ieb0808caa24c9a5e599084183ba5ee8a6536f7d8
Reviewed-on: https://go-review.googlesource.com/36622Reviewed-by: Robert Griesemer <gri@golang.org>

a896869a

runtime: implement fastrand in go · d03c1248

Sokolov Yura authored Jan 05, 2017

So it could be inlined.

Using bit-tricks it could be implemented without condition
(improved trick version by Minux Ma).

Simple benchmark shows it is faster on i386 and x86_64, though
I don't know will it be faster on other architectures?

benchmark                       old ns/op     new ns/op     delta
BenchmarkFastrand-3             2.79          1.48          -46.95%
BenchmarkFastrandHashiter-3     25.9          24.9          -3.86%

Change-Id: Ie2eb6d0f598c0bb5fac7f6ad0f8b5e3eddaa361b
Reviewed-on: https://go-review.googlesource.com/34782Reviewed-by: Minux Ma <minux@golang.org>
Run-TryBot: Minux Ma <minux@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

d03c1248

runtime/debug: don't run a GC when setting SetGCPercent negative · 9f75ecd5

Brad Fitzpatrick authored Dec 06, 2016

If the user is calling SetGCPercent(-1), they intend to disable GC.
They probably don't intend to run one. If they do, they can call
runtime.GC themselves.

Change-Id: I40ef40dfc7e15193df9ff26159cd30e56b666f73
Reviewed-on: https://go-review.googlesource.com/34013
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>

9f75ecd5

cmd/trace: Record mark assists in execution traces · 2a74b9e8

Heschi Kreinick authored Jan 31, 2017

During the mark phase of garbage collection, goroutines that allocate
may be recruited to assist. This change creates trace events for mark
assists and displays them similarly to sweep assists in the trace
viewer.

Mark assists are different than sweeps in that they can be preempted, so
displaying them in the trace viewer is a little tricky -- we may need to
synthesize multiple slices for one mark assist. This could have been
done in the parser instead, but I thought it might be preferable to keep
the parser as true to the event stream as possible.

Change-Id: I381dcb1027a187a354b1858537851fa68a620ea7
Reviewed-on: https://go-review.googlesource.com/36015
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>

2a74b9e8

cmd/compile/internal/gc: unexport or remove global functions · 249aca5d

Hajime Hoshi authored Feb 10, 2017

Change-Id: Ib2109ab773fbf2a35188300cf91a54735f75fc7c
Reviewed-on: https://go-review.googlesource.com/36736Reviewed-by: Matthew Dempsky <mdempsky@google.com>

249aca5d

expvar: parallelize BenchmarkMapAdd{Same,Different} · 39651bb5

Bryan C. Mills authored Feb 09, 2017

The other expvar tests are already parallelized, and this will help to
measure the impact of potential implementations for #18177.

updates #18177

Change-Id: I0f4f1a16a0285556cbcc8339855b6459af412675
Reviewed-on: https://go-review.googlesource.com/36717Reviewed-by: Russ Cox <rsc@golang.org>

39651bb5

reflect: clear ptrToThis in Ptr when allocating result on heap · 066ac428

Michael Hudson-Doyle authored Feb 10, 2017

Otherwise, calling PtrTo on the result will fail.

Fixes #19003

Change-Id: I8d7d1981a5d0417d5aee52740469d71e90734963
Reviewed-on: https://go-review.googlesource.com/36731
Run-TryBot: Michael Hudson-Doyle <michael.hudson@canonical.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

066ac428

sort: optimize average calculation in symMerge and doPivot. · 78e6abd2

David R. Jenni authored Feb 08, 2017

Change code of the form `i + (j-i)/2` to `int(uint(i+j) >> 1)`.

The optimized average calculation uses fewer instructions to calculate
the average without overflowing at the addition.

Analogous to https://golang.org/cl/36332.

name                 old time/op  new time/op  delta
StableString1K-4     49.6µs ± 3%  49.1µs ± 8%    ~     (p=0.659 n=16+19)
StableInt1K-4         160µs ±10%   148µs ± 5%  -7.29%  (p=0.000 n=20+16)
StableInt1K_Slice-4   139µs ± 4%   136µs ± 3%  -2.52%  (p=0.000 n=20+16)
StableInt64K-4       8.84ms ± 6%  8.57ms ± 5%  -3.07%  (p=0.001 n=20+19)
Stable1e2-4           162µs ±19%   147µs ±16%  -8.79%  (p=0.002 n=20+20)
Stable1e4-4          31.0ms ± 5%  30.6ms ± 5%    ~     (p=0.221 n=20+20)
Stable1e6-4           6.37s ± 3%   6.27s ± 2%  -1.67%  (p=0.000 n=19+20)

Change-Id: I1cea0bcb9ace8ef7e48b8fab772e41b4b2170da9
Reviewed-on: https://go-review.googlesource.com/36570
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

78e6abd2

crypto/sha512: Add AVX2 version for AMD64 · 6ac8ccf4

Ilya Tocar authored Feb 08, 2017

name          old time/op    new time/op     delta
Hash8Bytes-6     913ns ± 0%      667ns ± 0%  -26.91%  (p=0.000 n=10+10)
Hash1K-6        6.58µs ± 0%     4.23µs ± 0%  -35.69%  (p=0.000 n=10+9)
Hash8K-6        45.9µs ± 0%     28.1µs ± 0%  -38.93%  (p=0.000 n=10+10)

name          old speed      new speed       delta
Hash8Bytes-6  8.76MB/s ± 0%  11.99MB/s ± 0%  +36.87%  (p=0.000 n=10+8)
Hash1K-6       156MB/s ± 0%    242MB/s ± 0%  +55.49%  (p=0.000 n=10+9)
Hash8K-6       178MB/s ± 0%    292MB/s ± 0%  +63.74%  (p=0.000 n=10+10)

Change-Id: Ic9211d68b02935b2195995f264ec57d6bc36f713
Reviewed-on: https://go-review.googlesource.com/36630
Run-TryBot: Ilya Tocar <ilya.tocar@intel.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

6ac8ccf4

crypto/x509: ignore CN if SAN extension present. · 630e93ed

Adam Langley authored Feb 09, 2017

The code previously tested only whether DNS-name SANs were present in a
certificate which is only approximately correct. In fact, /any/ SAN
extension, including one with no DNS names, should cause the CN to be
ignored.

Change-Id: I3d9824918975be6d4817e7cbb48ed1b0c5a2fc8b
Reviewed-on: https://go-review.googlesource.com/36696
Run-TryBot: Adam Langley <agl@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

630e93ed

cmd/{asm,internal/obj/s390x}, math: remove emulated float instructions · a5246168

Michael Munday authored Nov 17, 2016

The s390x port was based on the ppc64 port and, because of the way the
port was done, inherited some instructions from it. ppc64 supports
3-operand (4-operand for FMADD etc.) floating point instructions
but s390x doesn't (the destination register is always an input) and
so these were emulated.

There is a bug in the emulation of FMADD whereby if the destination
register is also a source for the multiplication it will be
clobbered. This doesn't break any assembly code in the std lib but
could affect future work.

To fix this I have gone through the floating point instructions and
removed all unnecessary 3-/4-operand emulation. The compiler doesn't
need it and assembly writers don't need it, it's just a source of
bugs.

I've also deleted the FNMADD family of emulated instructions. They
aren't used anywhere.

Change-Id: Ic07cedcf141a6a3b43a0c84895460f6cfbf56c04
Reviewed-on: https://go-review.googlesource.com/33350
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

a5246168

math: check overflow in amd64 Exp implementation · f44e5870

Alberto Donizetti authored Feb 03, 2017

Unlike the pure go implementation used by every other architecture,
the amd64 asm implementation of Exp does not fail early if the
argument is known to overflow. Make it fail early.

Cost of the check is < 1ns (on an old Sandy Bridge machine):

name   old time/op  new time/op  delta
Exp-4  18.3ns ± 1%  18.7ns ± 1%  +2.08%  (p=0.000 n=18+20)

Fixes #14932
Fixes #18912

Change-Id: I04b3f9b4ee853822cbdc97feade726fbe2907289
Reviewed-on: https://go-review.googlesource.com/36271
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com>
Reviewed-by: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

f44e5870

encoding/json: clarify documention for Unmarshal into a pointer. · 4fafc843

Tuo Shan authored Feb 07, 2017

Fixes #18730.

Change-Id: If3ef28e62f7e449d4c8dc1dfd78f7d6f5a87ed26
Reviewed-on: https://go-review.googlesource.com/36478Reviewed-by: Russ Cox <rsc@golang.org>

4fafc843

database/sql: ensure driverConns are closed if not returned to pool · a335c344

Daniel Theophanes authored Feb 09, 2017

Previously if a connection was requested but timed out during the
request and when acquiring the db.Lock the connection request
is fulfilled and the request is unable to be returned to the
connection pool, then then driver connection would not be closed.

No tests were added or modified because I was unable to determine
how to trigger this situation without something invasive.

Change-Id: I9d4dc680e3fdcf63d79d212174a5b8b313f363f1
Reviewed-on: https://go-review.googlesource.com/36641Reviewed-by: Russ Cox <rsc@golang.org>

a335c344

runtime/pprof: merge internal/protopprof into pprof package · 9a754439

Russ Cox authored Feb 09, 2017

These are very tightly coupled, and internal/protopprof is small.
There's no point to having a separate package.

Change-Id: I2c8aa49c9e18a7128657bf2b05323860151b5606
Reviewed-on: https://go-review.googlesource.com/36711
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

9a754439

image: fix the overlap check in Rectangle.Intersect. · 3a209281

Nigel Tao authored Feb 10, 2017

This is a re-roll of a previous commit,
a855da29, which was rolled back in
14347ee4.

It was rolled back because it broke a unit test in image/gif. The
image/gif code was fixed by 9ef65dbe
"image/gif: fix frame-inside-image bounds checking".

The original commit message:

image: fix the overlap check in Rectangle.Intersect.

The doc comment for Rectangle.Intersect clearly states, "If the two
rectangles do not overlap then the zero rectangle will be returned."
Prior to this fix, calling Intersect on adjacent but non-overlapping
rectangles would return an empty but non-zero rectangle.

The fix essentially changes
if r.Min.X > r.Max.X || r.Min.Y > r.Max.Y { etc }
to
if r.Min.X >= r.Max.X || r.Min.Y >= r.Max.Y { etc }
(note that the > signs have become >= signs), but changing that line to:
if r.Empty() { etc }
seems clearer (and equivalent).

Change-Id: I2e3af1f1686064a573b2e513b39246fe60c03631
Reviewed-on: https://go-review.googlesource.com/36734Reviewed-by: Rob Pike <r@golang.org>
Run-TryBot: Nigel Tao <nigeltao@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

3a209281