]> Cypherpunks repositories - gostls13.git/log
gostls13.git
6 years agomisc/cgo/test: add retry loop around pthread_create in TestSigprocmask
Ian Lance Taylor [Wed, 27 Jun 2018 00:23:07 +0000 (17:23 -0700)]
misc/cgo/test: add retry loop around pthread_create in TestSigprocmask

This is the same retry loop we use in _cgo_try_pthread_create in runtime/cgo.

Fixes #25078

Change-Id: I7ef4d4fc7fb89cbfb674c4f93cbdd7a033dd8983
Reviewed-on: https://go-review.googlesource.com/121096
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agomisc/cgo/test: skip Test18146 in short mode
Ian Lance Taylor [Wed, 27 Jun 2018 00:52:34 +0000 (17:52 -0700)]
misc/cgo/test: skip Test18146 in short mode

Fixes #21219

Change-Id: I1a2ec1afe06586ed33a3a855b77536490cac3a38
Reviewed-on: https://go-review.googlesource.com/121115
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/link: document limitation of -X
LE Manh Cuong [Tue, 26 Jun 2018 06:48:05 +0000 (13:48 +0700)]
cmd/link: document limitation of -X

Fixes #26042

Change-Id: Ica16f14a65c03659a19926852cca5e554c99baf1
Reviewed-on: https://go-review.googlesource.com/120935
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agomime/multipart: restore 1.9 handling of missing/empty form-data file name
Ian Lance Taylor [Tue, 26 Jun 2018 22:57:35 +0000 (15:57 -0700)]
mime/multipart: restore 1.9 handling of missing/empty form-data file name

Revert the code changes of CL 96975 and CL 70931, but keep the tests,
appropriately modified for the code changes. This restores the 1.9
handling of form-data entries with missing or empty file names.

Changing the handling of this simply confused existing programs for no
useful benefit. Go back to the old behavior.

Updates #19183
Fixes #24041

Change-Id: I4ebc32433911e6360b9fd79d8f63a6d884822e0e
Reviewed-on: https://go-review.googlesource.com/121055
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agonet: improve ExampleUDPConn_WriteTo docs
Mikio Hara [Sun, 24 Jun 2018 17:08:52 +0000 (02:08 +0900)]
net: improve ExampleUDPConn_WriteTo docs

Also updates comment on isConnected field of netFD for clarification.

Change-Id: Icb1b0332e3b4c7802eae00ddc26cd5ba54c82dc2
Reviewed-on: https://go-review.googlesource.com/120955
Run-TryBot: Mikio Hara <mikioh.mikioh@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/compile: improve escape analysis explanation go1.11beta1
David Chase [Tue, 26 Jun 2018 16:00:25 +0000 (12:00 -0400)]
cmd/compile: improve escape analysis explanation

No code changes, only revised comments in an attempt to make
escape analysis slightly less confusing.

Updates #23109.

Change-Id: I5ee6cea0946ced63f6210ac4484a088bcdd862fb
Reviewed-on: https://go-review.googlesource.com/121001
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
6 years agoast: refer to "embedded" rather than "anonymous" fields in
Rowan Marshall [Fri, 22 Jun 2018 17:36:17 +0000 (18:36 +0100)]
ast: refer to "embedded" rather than "anonymous" fields in

documentation.

Fixes #25684.

Change-Id: I9d0e47dff2446c27a3df88fb9ccfefef7419470b
Reviewed-on: https://go-review.googlesource.com/120556
Reviewed-by: Robert Griesemer <gri@golang.org>
6 years agogo/types: rename NewInterface2 to NewInterfaceType
Robert Griesemer [Mon, 25 Jun 2018 23:21:59 +0000 (16:21 -0700)]
go/types: rename NewInterface2 to NewInterfaceType

NewInterface2 was introduced with https://go-review.googlesource.com/114317
which fixed #25301. Changed the name to NewInterfaceType to better match
Go naming styles, per discussion with @josharian, @iant, et al.

Change-Id: Ifa4708a5efd4f708295b33c3d20fdc5812e1b4fc
Reviewed-on: https://go-review.googlesource.com/120875
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agosyscall/js: use stable references to JavaScript values
Richard Musiol [Sun, 24 Jun 2018 15:23:38 +0000 (17:23 +0200)]
syscall/js: use stable references to JavaScript values

This commit changes how JavaScript values are referenced by Go code.
After this change, a JavaScript value is always represented by the same
ref, even if passed multiple times from JavaScript to Go. This allows
Go's == operator to work as expected on js.Value (strict equality).
Additionally, the performance of some operations of the syscall/js
package got improved by saving additional roundtrips to JavaScript code.

Fixes #25802.

Change-Id: Ide6ffe66c6aa1caf5327a2d3ddbe48fe7c180461
Reviewed-on: https://go-review.googlesource.com/120561
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agosyscall: fix typo in RawSyscall6 comment
Xargin [Tue, 26 Jun 2018 03:33:20 +0000 (03:33 +0000)]
syscall: fix typo in RawSyscall6 comment

Change-Id: I589a60459a3b0811da77d46428eca57c741b1184
GitHub-Last-Rev: f9ab775b9d7f869916af6de9abb7c1d5f0005f1b
GitHub-Pull-Request: golang/go#26063
Reviewed-on: https://go-review.googlesource.com/120917
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/cgo: mention character restrictions on #cgo lines
Ian Lance Taylor [Tue, 26 Jun 2018 13:40:26 +0000 (06:40 -0700)]
cmd/cgo: mention character restrictions on #cgo lines

Fixes #26025

Change-Id: I43b3c9019ca8d2844e4d664c2090e506b2171077
Reviewed-on: https://go-review.googlesource.com/120975
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agotime: run genzabbrs.go to update zoneinfo_abbrs_windows.go
Alberto Donizetti [Sun, 24 Jun 2018 12:23:28 +0000 (14:23 +0200)]
time: run genzabbrs.go to update zoneinfo_abbrs_windows.go

The abbreviations list in zoneinfo_abbrs_windows.go was last updated
in June 2017, and it's currently outdated. Update it.

Change-Id: Ie2bf4268787f5aefe98ee110c2c279451e18fd97
Reviewed-on: https://go-review.googlesource.com/120559
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
6 years agocmd/compile: map delete should clear value always
Vladimir Kuzmin [Thu, 21 Jun 2018 05:19:56 +0000 (22:19 -0700)]
cmd/compile: map delete should clear value always

Map delete must clear value every time because
newly added map optimizations of compound-assignment
operators (CL #91557) rely on this behavior of map delete.

It slows down map delete operation for non-reference types:

name                   old time/op    new time/op    delta
MapDelete/Int32/100      23.9ns ± 2%    27.8ns ± 4%  +16.04%  (p=0.000 n=20+20)
MapDelete/Int32/1000     21.5ns ± 2%    25.2ns ± 2%  +17.06%  (p=0.000 n=20+19)
MapDelete/Int32/10000    24.2ns ± 6%    27.2ns ± 5%  +12.39%  (p=0.000 n=19+19)
MapDelete/Int64/100      24.2ns ± 4%    27.7ns ± 2%  +14.55%  (p=0.000 n=20+20)
MapDelete/Int64/1000     22.1ns ± 2%    24.8ns ± 2%  +12.36%  (p=0.000 n=10+20)

Fixes #25936

Change-Id: I8499b790cb5bb019938161b3e50f3243d9bbb79c
Reviewed-on: https://go-review.googlesource.com/120255
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Keith Randall <khr@golang.org>
6 years agocmd/dist: increase timeout scale to 3 for windows
Andrew [Mon, 25 Jun 2018 20:24:37 +0000 (20:24 +0000)]
cmd/dist: increase timeout scale to 3 for windows

cmd/go can sometimes take up to 400s on windows due
to various issues (disk I/O on builders being the
latest cause). Increase the timeout scale to account
for this.

Change-Id: I1fd4964472a70fb0f33cf6ed73298c034b9c1fb0
Reviewed-on: https://go-review.googlesource.com/120762
Run-TryBot: Andrew Bonventre <andybons@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/compile: avoid remainder in loopbce when increment=0
David Chase [Mon, 25 Jun 2018 19:26:45 +0000 (15:26 -0400)]
cmd/compile: avoid remainder in loopbce when increment=0

For non-unit increment, loopbce checks to see if the
increment evenly divides the difference between (constant)
loop start and end.  This test panics when the increment
is zero.

Fix: check for zero, if found, don't optimize the loop.

Also added missing copyright notice to loopbce.go.

Fixes #26043.

Change-Id: I5f460104879cacc94481949234c9ce8c519d6380
Reviewed-on: https://go-review.googlesource.com/120759
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/dist: enable more tests on js/wasm
Brad Fitzpatrick [Mon, 25 Jun 2018 19:05:25 +0000 (19:05 +0000)]
cmd/dist: enable more tests on js/wasm

Fixes #26050

Change-Id: I21697ac30eb3bc423263cba885bce5fea8f81d94
Reviewed-on: https://go-review.googlesource.com/120796
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Andrew Bonventre <andybons@golang.org>
6 years agonet/http: document and test behavior of ServeMux with ports
Guilherme Goncalves [Sat, 23 Jun 2018 22:32:26 +0000 (22:32 +0000)]
net/http: document and test behavior of ServeMux with ports

Beginning on Go 1.9, ServeMux has been dropping the port number from the Host
header and in the path pattern. This commit explicitly mentions the change in
behavior and adds a simple test case to ensure consistency.

Fixes #23351.

Change-Id: I0270c8bd96cda92c13ac6437cdf66c2807b3042d
Reviewed-on: https://go-review.googlesource.com/120557
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agoruntime/trace: reword the trace package summary
Jeremy Jackins [Mon, 25 Jun 2018 19:04:31 +0000 (19:04 +0000)]
runtime/trace: reword the trace package summary

Reword the package summary to be a little easier to read.

Change-Id: I84a9301a02e228b46165410a429548b3774762d5
GitHub-Last-Rev: 1342c7219f0ed06144d3bca9acd77790707e4b77
GitHub-Pull-Request: golang/go#26052
Reviewed-on: https://go-review.googlesource.com/120795
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agoos: have UserCacheDir return an error on failure
Andrew Bonventre [Mon, 25 Jun 2018 18:26:53 +0000 (14:26 -0400)]
os: have UserCacheDir return an error on failure

Previously, it would return an empty string if it
could not determine the user's cache directory.
Return an error instead.

Change-Id: I74f00b1ad3858efa3fe2700c599271ebfe5764b6
Reviewed-on: https://go-review.googlesource.com/120757
Run-TryBot: Andrew Bonventre <andybons@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agodoc/go1.11: add note about package versioning
Andrew Bonventre [Mon, 25 Jun 2018 17:48:59 +0000 (13:48 -0400)]
doc/go1.11: add note about package versioning

It will not be present in go1.11beta1 but will be present
in subsequent releases.

Change-Id: I298fb682945345bb4a34ec83802fd644f75bdd98
Reviewed-on: https://go-review.googlesource.com/120756
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agosyscall/js: turn constant package vars into functions
Richard Musiol [Mon, 25 Jun 2018 10:30:31 +0000 (12:30 +0200)]
syscall/js: turn constant package vars into functions

This is so the values can not be changed and the type is easy to see.

Requested on https://go-review.googlesource.com/c/go/+/120561.

Change-Id: If2ed48ca3ba8874074687bfb2375d2f5592e8e0d
Reviewed-on: https://go-review.googlesource.com/120564
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agosyscall.js: add Value.InstanceOf
Richard Musiol [Fri, 15 Jun 2018 08:45:04 +0000 (10:45 +0200)]
syscall.js: add Value.InstanceOf

Change-Id: Icf56188fdb2b8ce6789830a35608203fdb9a3df6
Reviewed-on: https://go-review.googlesource.com/120560
Reviewed-by: Paul Jolly <paul@myitcv.org.uk>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agoruntime: respect timeout in semasleep on Darwin
Nikhil Benesch [Sat, 23 Jun 2018 05:15:19 +0000 (01:15 -0400)]
runtime: respect timeout in semasleep on Darwin

semasleep on Darwin was refactored in https://golang.org/cl/118736 to
use the pthread_cond_timedwait function from libc. The new code
incorrectly assumed that pthread_cond_timedwait took a timeout relative
to the current time, when it in fact it takes a timeout specified in
absolute time. semasleep thus specified a timeout well in the past,
causing it to immediately exceed the timeout and spin hot. This was the
source of a large performance hit to CockroachDB (#26019).

Adjust semasleep to instead call pthread_cond_timedwait_relative_np,
which properly interprets its timeout parameter as relative to the
current time.

pthread_cond_timedwait_relative_np is non-portable, but using
pthread_cond_timedwait correctly would require two calls to
gettimeofday: one in the runtime package to convert the relative timeout
to absolute time, then another in the pthread library to convert back to
a relative offset [0], as the Darwin kernel expects a relative offset.

[0]: https://opensource.apple.com/source/libpthread/libpthread-301.30.1/src/pthread_cond.c.auto.html

Fix #26019.

Change-Id: I1a8c2429f79513b43d2b256365cd9166d235af8b
Reviewed-on: https://go-review.googlesource.com/120635
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agodoc/contribute: fix typo and reword a few sentences
Alberto Donizetti [Wed, 20 Jun 2018 17:23:24 +0000 (19:23 +0200)]
doc/contribute: fix typo and reword a few sentences

This change fixes a typo in doc/contribute.html (afects -> affects)
and rewords a few slightly akward sentences.

Change-Id: I6bfbacba0de29464fce134b0fdaf3898a97b8d57
Reviewed-on: https://go-review.googlesource.com/120105
Reviewed-by: Rob Pike <r@golang.org>
6 years agocmd/link: never coalesce type descriptors when dynamically linking Go
Michael Hudson-Doyle [Wed, 20 Jun 2018 22:31:57 +0000 (10:31 +1200)]
cmd/link: never coalesce type descriptors when dynamically linking Go

Add a test by making misc/cgo/testshared/src/trivial.go marginally less
trivial.

Fixes #25970.

Change-Id: I8815d0c56b8850fcdbf9b45f8406f37bd21b6865
Reviewed-on: https://go-review.googlesource.com/120235
Run-TryBot: Michael Hudson-Doyle <michael.hudson@canonical.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoapi: promote next to go1.11
Andrew Bonventre [Fri, 22 Jun 2018 18:25:08 +0000 (14:25 -0400)]
api: promote next to go1.11

Change-Id: Ib8fa0a12363993033201ff707c315f4030811f89
Reviewed-on: https://go-review.googlesource.com/120595
Run-TryBot: Andrew Bonventre <andybons@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/link: support DWARF compression on Darwin
Heschi Kreinick [Wed, 20 Jun 2018 20:45:25 +0000 (16:45 -0400)]
cmd/link: support DWARF compression on Darwin

We want to compress DWARF even on macOS, but the native toolchain isn't
going to understand it. Add a flag that can be used to disable
compression, then add Darwin to the whitelist used during internal
linking.

Unlike GNU ld, the Darwin linker doesn't have a handy linker flag to do
compression. But since we're already doing surgery to put the DWARF in
the output executable in the first place, compressing it at the same
time isn't unduly difficult. This does have the slightly odd effect of
compressing some Apple proprietary debug sections, which absolutely
nothing will understand. Leaving them uncompressed didn't make much
sense, though, since I doubt they're useful without (say) __debug_info.

Updates #11799

Change-Id: Ie00b0215c630a798c59d009a641e2d13f0e7ea01
Reviewed-on: https://go-review.googlesource.com/120155
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agosyscall, internal/poll: fix build on js/wasm
Brad Fitzpatrick [Fri, 22 Jun 2018 17:45:30 +0000 (17:45 +0000)]
syscall, internal/poll: fix build on js/wasm

Fixes #26014

Change-Id: I9d92414a9181c5d189e3e266666950656bf00406
Reviewed-on: https://go-review.googlesource.com/120576
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Andrew Bonventre <andybons@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/dist: skip non-std tests on js/wasm
Brad Fitzpatrick [Fri, 22 Jun 2018 17:36:12 +0000 (17:36 +0000)]
cmd/dist: skip non-std tests on js/wasm

After the std tests, most of the rest of the tests aren't applicable
to js/wasm. (anything with -cpu=>1, cgo, etc)

Skip them all for now. We can incrementally re-enable them over time
as the js/wasm port is fleshed out. But let's get the builder column
black again so we can enable trybots and keep it black.

Updates #26014
Updates #26015
Updates #18892

Change-Id: I8992ed3888f598fa42273ce8646a32d62ce45b1d
Reviewed-on: https://go-review.googlesource.com/120575
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Andrew Bonventre <andybons@golang.org>
6 years agocmd/compile: fix compile failure for lazily resolved shadowed types
Matthew Dempsky [Thu, 21 Jun 2018 23:12:17 +0000 (16:12 -0700)]
cmd/compile: fix compile failure for lazily resolved shadowed types

If expanding an inline function body required lazily expanding a
package-scoped type whose identifier was shadowed within the function
body, the lazy expansion would instead overwrite the local symbol
definition instead of the package-scoped symbol. This was due to
importsym using s.Def instead of s.PkgDef.

Unfortunately, this is yet another consequence of the current awkward
scope handling code.

Passes toolstash-check.

Fixes #25984.

Change-Id: Ia7033e1749a883e6e979c854d4b12b0b28083dd8
Reviewed-on: https://go-review.googlesource.com/120456
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
6 years agocmd/compile: convert uint32 to int32 in ARM constant folding rules
Cherry Zhang [Thu, 21 Jun 2018 14:59:19 +0000 (10:59 -0400)]
cmd/compile: convert uint32 to int32 in ARM constant folding rules

MOVWconst's AuxInt is Int32. SSA check complains if the AuxInt
does not fit in int32. Convert uint32 to int32 to make it happy.

The generated code is unchanged. MOVW only cares low 32 bits.

Passes "toolstash -cmp" std cmd for ARM.

Fixes #25993.

Change-Id: I2b6532c9c285ea6d89652505fb7c553f85a98864
Reviewed-on: https://go-review.googlesource.com/120335
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
6 years agodoc/go1.11: clarify GOOS/GOARCH pair of WebAssembly port
Dmitri Shuralyov [Wed, 20 Jun 2018 17:15:44 +0000 (13:15 -0400)]
doc/go1.11: clarify GOOS/GOARCH pair of WebAssembly port

It's more common to specify GOOS/GOARCH values in that order,
rather than the inverse. Fix the order.

Updates #18892.

Change-Id: I8551508599e019f6617dc007397b562c9926418d
Reviewed-on: https://go-review.googlesource.com/120057
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agosync: fix deficiency in RWMutex race annotations
Dmitry Vyukov [Fri, 22 Jun 2018 09:00:13 +0000 (11:00 +0200)]
sync: fix deficiency in RWMutex race annotations

Remove unnecessary race.Release annotation from Unlock.

For RWMutex we want to establish the following happens-before (HB) edges:
1. Between Unlock and the subsequent Lock.
2. Between Unlock and the subsequent RLock.
3. Between batch of RUnlock's and the subsequent Lock.

1 is provided by Release(&rw.readerSem) in Unlock and Acquire(&rw.readerSem) in Lock.
2 is provided by Release(&rw.readerSem) in Unlock and Acquire(&rw.readerSem) in RLock.
3 is provided by ReleaseMerge(&rw.writerSem) in RUnlock in Acquire(&rw.writerSem) in Lock,
since we want to establish HB between a batch of RUnlock's this uses ReleaseMerge instead of Release.

Release(&rw.writerSem) in Unlock is simply not needed.

FWIW this is also how C++ tsan handles mutexes, not a proof but at least something.
Making 2 implementations consistent also simplifies any kind of reasoning against both of them.

Since this only affects performance, a reasonable test is not possible.
Everything should just continue to work but slightly faster.

Credit for discovering this goes to Jamie Liu.

Change-Id: Ice37d29ecb7a5faed3f7781c38dd32c7469b2735
Reviewed-on: https://go-review.googlesource.com/120495
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoos, net: avoid races between dup, set-blocking-mode, and closing
Ian Lance Taylor [Wed, 20 Jun 2018 02:50:03 +0000 (19:50 -0700)]
os, net: avoid races between dup, set-blocking-mode, and closing

Fixes #24481
Fixes #24483

Change-Id: Id7da498425a440c91582aa5480c253ae7a9c932c
Reviewed-on: https://go-review.googlesource.com/119955
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agonet/http: fix test assertion
Thomas Bruyelle [Fri, 22 Jun 2018 09:16:40 +0000 (11:16 +0200)]
net/http: fix test assertion

Logf doesn't make the test fail, so the test was always OK.

Change-Id: I7c10ee74ff7e5d28cbd3a35e185093cb9f349470
Reviewed-on: https://go-review.googlesource.com/120496
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agoruntime: fix comments style typo
Zhou Peng [Fri, 22 Jun 2018 05:31:06 +0000 (05:31 +0000)]
runtime: fix comments style typo

Code comments should have a space between comments characters and
actual words.

Change-Id: I6274baf1fc09b37a32ec6c69ddbb8edca9eb5469
Reviewed-on: https://go-review.googlesource.com/120475
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agotext/template/parse: undo breaking API changes
Daniel Theophanes [Thu, 21 Jun 2018 17:41:26 +0000 (10:41 -0700)]
text/template/parse: undo breaking API changes

golang.org/cl/84480 altered the API for the parse package for
clarity and consistency. However, the changes also broke the
API for consumers of the package. This CL reverts the API
to the previous spelling, adding only a single new exported
symbol.

Fixes #25968

Change-Id: Ieb81054b61eeac7df3bc3864ef446df43c26b80f
Reviewed-on: https://go-review.googlesource.com/120355
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Rob Pike <r@golang.org>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agoruntime: avoid recursive panic on bad lock count
Austin Clements [Thu, 21 Jun 2018 20:56:13 +0000 (16:56 -0400)]
runtime: avoid recursive panic on bad lock count

Currently, if lock or unlock calls throw because the g.m.lock count is
corrupted, we're unlikely to get a stack trace because startpanic_m
will itself attempt to acquire a lock, causing a recursive failure.

Avoid this by forcing the g.m.locks count to a sane value if it's
currently bad.

This might be enough to get a stack trace from #25128.

Change-Id: I52d7bd4717ffae94a821f4249585f3eb6cd5aa41
Reviewed-on: https://go-review.googlesource.com/120416
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/compile: fix recursive inimport handling
Matthew Dempsky [Thu, 21 Jun 2018 22:35:32 +0000 (15:35 -0700)]
cmd/compile: fix recursive inimport handling

expandDecl can be called recursively, so it's not an appropriate place
to clean inimport. Instead, move this up to resolve, along with an
appropriate recursion check.

Passes toolstash-check.

Change-Id: I138d37b057dcc6525c780b4b3fbaa5e97f99655b
Reviewed-on: https://go-review.googlesource.com/120455
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
6 years agoflag: add a Value example
Terin Stock [Wed, 20 Jun 2018 21:00:04 +0000 (14:00 -0700)]
flag: add a Value example

Change-Id: I579cc9f4f8e5be5fd6447a99614797ab7bc53611
Reviewed-on: https://go-review.googlesource.com/120175
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agomisc/wasm: fix permissions on wasm_exec.js
Paul Jolly [Thu, 21 Jun 2018 17:30:04 +0000 (18:30 +0100)]
misc/wasm: fix permissions on wasm_exec.js

Currently wasm_exec.js is executable (0755) yet has no interpreter.
Indeed wasm_exec.js is only ever used as an argument to Node or loaded
via a <script> tag in a browser-loaded HTML file.  Hence the execute
mode bits are superfluous and simply serve to clutter your PATH if
$GOROOT/misc/wasm is on your PATH (as is required if you want to run go
test syscall/js).

Change-Id: I279e2457094f8a12b9bf380ad7f1a9f47b22fc96
Reviewed-on: https://go-review.googlesource.com/120435
Run-TryBot: Paul Jolly <paul@myitcv.org.uk>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agointernal/poll: use more fine-grained locking in Splice
Andrei Tudor Călin [Thu, 21 Jun 2018 17:11:34 +0000 (19:11 +0200)]
internal/poll: use more fine-grained locking in Splice

The previous code acquired a read lock on src and a write lock on
dst for the entire duration of Splice. This resulted in deadlock,
in a situation akin to the following:

Splice is blocking, waiting to read from src.

The caller tries to close dst from another goroutine, but Close on
dst blocks in runtime.semacquire, because Splice is still holding a
write lock on it, and the Close didn't unblock any I/O.

The caller cannot unblock the read side of Splice through other means,
because they are stuck waiting in dst.Close().

Use more fine-grained locking instead: acquire the read lock on src
just before trying to splice from the source socket to the pipe,
and acquire the write lock on dst just before trying to splice from
the pipe to the destination socket.

Fixes #25985

Change-Id: I264c91c7a69bb6c5e28610e2bd801244804cf86d
Reviewed-on: https://go-review.googlesource.com/120317
Run-TryBot: Aram Hăvărneanu <aram@mgk.ro>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/go: re-enable a couple of tests of gccgo
Ian Lance Taylor [Thu, 21 Jun 2018 19:01:54 +0000 (12:01 -0700)]
cmd/go: re-enable a couple of tests of gccgo

Updates #22472

Change-Id: I526d131f2ef8e0200f7a5634c75b31e0ee083f93
Reviewed-on: https://go-review.googlesource.com/120375
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/compile: improve atomic add intrinsics with ARMv8.1 new instruction
Wei Xiao [Fri, 3 Nov 2017 02:05:28 +0000 (02:05 +0000)]
cmd/compile: improve atomic add intrinsics with ARMv8.1 new instruction

ARMv8.1 has added new instruction (LDADDAL) for atomic memory operations. This
CL improves existing atomic add intrinsics with the new instruction. Since the
new instruction is only guaranteed to be present after ARMv8.1, we guard its
usage with a conditional on CPU feature.

Performance result on ARMv8.1 machine:
name        old time/op  new time/op  delta
Xadd-224    1.05µs ± 6%  0.02µs ± 4%  -98.06%  (p=0.000 n=10+8)
Xadd64-224  1.05µs ± 3%  0.02µs ±13%  -98.10%  (p=0.000 n=9+10)
[Geo mean]  1.05µs       0.02µs       -98.08%

Performance result on ARMv8.0 machine:
name        old time/op  new time/op  delta
Xadd-46      538ns ± 1%   541ns ± 1%  +0.62%  (p=0.000 n=9+9)
Xadd64-46    505ns ± 1%   508ns ± 0%  +0.48%  (p=0.003 n=9+8)
[Geo mean]   521ns        524ns       +0.55%

Change-Id: If4b5d8d0e2d6f84fe1492a4f5de0789910ad0ee9
Reviewed-on: https://go-review.googlesource.com/81877
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
6 years agonet/http: avoid deferred unlock in ServeMux.shouldRedirect
Emmanuel T Odeke [Mon, 21 May 2018 19:57:53 +0000 (12:57 -0700)]
net/http: avoid deferred unlock in ServeMux.shouldRedirect

CL 96575 introduced concurrency protection for
ServeMux.shouldRedirect with a read lock and deferred unlock.
However, the change produced a noticeable regression.
Instead add the suffix "RLocked" to the function name to
declare that we should hold the read lock as a pre-requisite
before calling it, hence avoiding the defer altogether.

Benchmarks:
name                  old time/op    new time/op    delta
ServeMux-8              63.3µs ± 0%    54.6µs ± 0%  -13.74%  (p=0.000 n=9+9)
ServeMux_SkipServe-8    41.4µs ± 2%    32.7µs ± 1%  -21.05%  (p=0.000 n=10+10)

name                  old alloc/op   new alloc/op   delta
ServeMux-8              17.3kB ± 0%    17.3kB ± 0%     ~     (all equal)
ServeMux_SkipServe-8     0.00B          0.00B          ~     (all equal)

name                  old allocs/op  new allocs/op  delta
ServeMux-8                 360 ± 0%       360 ± 0%     ~     (all equal)
ServeMux_SkipServe-8      0.00           0.00          ~     (all equal)

Updates #25383
Updates #25482

Change-Id: I2ffa4eafe165faa961ce23bd29b5653a89facbc2
Reviewed-on: https://go-review.googlesource.com/113996
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agosyscall: check faccessat flags parameter on Linux
Tobias Klauser [Wed, 20 Jun 2018 13:41:37 +0000 (15:41 +0200)]
syscall: check faccessat flags parameter on Linux

Port CL 119495 from golang.org/x/sys/unix to the syscall package.

Currently Linux faccessat(2) syscall implementation doesn't support the
flags parameter. As per the discussion in #25845, permit the same flags
as glibc [1].

[1] https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/faccessat.c;h=ea42b2303ff4b2d2d6548ea04376fb265f773436;hb=HEAD

Updates #25845

Change-Id: I132b33275a9cc72b3a97acea5482806c7f47d7f7
Reviewed-on: https://go-review.googlesource.com/120015
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agomime: change *.js mime type to application/javascript, not x-javascript
Brad Fitzpatrick [Wed, 20 Jun 2018 18:25:54 +0000 (18:25 +0000)]
mime: change *.js mime type to application/javascript, not x-javascript

We delayed doing this for 4 years for fear that it might break something,
but it was standardized (RFC 4329) 12 years ago, and the default in Debian
and other places is correct:

   $ cat /etc/mime.types  | grep js$
   application/javascript                          js

Time for us to change too.

I doubt there will be problems, but we'll see during the Go 1.11 beta.

Fixes #7498

Change-Id: Iba0bf8a6e707a64dd63317e1c0d6dd9a18634527
Reviewed-on: https://go-review.googlesource.com/120058
Reviewed-by: Andrew Bonventre <andybons@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/go/internal: add a note about GOCACHE=off
Agniva De Sarker [Tue, 19 Jun 2018 14:57:08 +0000 (20:27 +0530)]
cmd/go/internal: add a note about GOCACHE=off

Fixes #25928

Change-Id: I1401ecc54af26eeeee648bb8eeb5d2d3566fa60c
Reviewed-on: https://go-review.googlesource.com/119695
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agodoc: fix spelling of G Suite
Stephen Lewis [Wed, 20 Jun 2018 18:38:19 +0000 (19:38 +0100)]
doc: fix spelling of G Suite

'G Suite' seems to be the preferred spelling according to
https://gsuite.google.com/

Change-Id: Ica60938cf942594157bba84ef205e1cdcb8c1b08
Reviewed-on: https://go-review.googlesource.com/120132
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
6 years agodoc: change git clone url
Wayne Ashley Berry [Thu, 7 Jun 2018 19:54:08 +0000 (20:54 +0100)]
doc: change git clone url

The git clone url should be a Gerrit host and not Github, otherwise the
codereview command will fail.

git-codereview: failed to load Gerrit origin: git origin must be a Gerrit host, not GitHub: https://github.com/golang/go

Change-Id: I62f62c86ee6dce0720a844fc56340135dfae8405
Reviewed-on: https://go-review.googlesource.com/117178
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocrypto/tls: consolidate signatures handling in SKE and CV
Peter Wu [Wed, 22 Nov 2017 18:25:20 +0000 (18:25 +0000)]
crypto/tls: consolidate signatures handling in SKE and CV

ServerKeyExchange and CertificateVerify can share the same logic for
picking a signature algorithm (based on the certificate public key and
advertised algorithms), selecting a hash algorithm (depending on TLS
version) and signature verification.

Refactor the code to achieve code reuse, have common error checking
(especially for intersecting supported signature algorithms) and to
prepare for addition of new signature algorithms. Code should be easier
to read since version-dependent logic is concentrated at one place.

Change-Id: I978dec3815d28e33c3cfbc85f0c704b1894c25a3
Reviewed-on: https://go-review.googlesource.com/79735
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agointernal/poll: better panic message for lock overflow
Ian Lance Taylor [Wed, 20 Jun 2018 03:43:24 +0000 (20:43 -0700)]
internal/poll: better panic message for lock overflow

Instead of "inconsistent poll.fdMutex", panic with
"too many concurrent operations on a single file or socket (max 1048575)".

Fixes #25558

Change-Id: I5cad3633aa539fb6f48cca236c6656c86acfb663
Reviewed-on: https://go-review.googlesource.com/119956
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Dmitry Vyukov <dvyukov@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
6 years agomake.bash: don't pass GOOS and GOARCH to cmd/go when finding GOROOT_BOOTSTRAP
LE Manh Cuong [Wed, 20 Jun 2018 04:22:37 +0000 (11:22 +0700)]
make.bash: don't pass GOOS and GOARCH to cmd/go when finding GOROOT_BOOTSTRAP

Fixes #25962

Change-Id: I10d41713f6aef100d7b2c8c976f22d1c8ac376d5
Reviewed-on: https://go-review.googlesource.com/119937
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocrypto/rand: make documentation consistent between package comment and Reader
Brad Fitzpatrick [Tue, 19 Jun 2018 21:55:01 +0000 (21:55 +0000)]
crypto/rand: make documentation consistent between package comment and Reader

Updates #25959

Change-Id: I9ae64b216ab5807718db0db98b32de1dc5fa4bec
Reviewed-on: https://go-review.googlesource.com/119875
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
6 years agocmd/go: remove inadvertent comment for vgo
Agniva De Sarker [Wed, 20 Jun 2018 06:05:20 +0000 (11:35 +0530)]
cmd/go: remove inadvertent comment for vgo

This change was introduced while adding the dark copy
of golang.org/x/vgo in CL 118095.

While the comment made sense in a separate vgo repo, when it is
merged with the main repo, this should not remain.

Found while running mkalldocs.sh in CL 119695.

Change-Id: I112a4629c415032bd29e165ac1c27a0f3cabeede
Reviewed-on: https://go-review.googlesource.com/119938
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agogo/internal/gccgoimporter: read export data from archives
Ian Lance Taylor [Tue, 19 Jun 2018 23:53:18 +0000 (16:53 -0700)]
go/internal/gccgoimporter: read export data from archives

When used with the go tool, gccgo will normally generate archive files.
This change teaches the gccgoimporter package how to read the export
data from an archive.

This is needed by, for example, cmd/vet, when typechecking packages.

Change-Id: I21267949a7808cd81c0042af425c774a4ff7d82f
Reviewed-on: https://go-review.googlesource.com/119895
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
6 years agomisc/cgo/test: avoid duplicate definition with gccgo
Ian Lance Taylor [Tue, 19 Jun 2018 23:55:19 +0000 (16:55 -0700)]
misc/cgo/test: avoid duplicate definition with gccgo

Current versions of gccgo issue a duplicate definition error when both
a definition and an empty declaration occur. Use build tags to avoid
that case for the issue9400 subdirectory.

Change-Id: I18517af87bab05e9ca43f2f295459cf34347c317
Reviewed-on: https://go-review.googlesource.com/119896
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agoruntime: fix FreeBSDNumCPU test
David Carlier [Sun, 17 Jun 2018 17:27:13 +0000 (17:27 +0000)]
runtime: fix FreeBSDNumCPU test

num cpu unit test fixes for FreeBSD.
cpuset -g can possibly output more
data than expected.

Fixes #25924

Change-Id: Iec45a919df68648759331da7cd1fa3b9f3ca4241
GitHub-Last-Rev: 4cc275b519cda13189ec48b581ab9ce00cacd7f6
GitHub-Pull-Request: golang/go#25931
Reviewed-on: https://go-review.googlesource.com/119376
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agodebug/elf,macho,pe: support compressed DWARF
Heschi Kreinick [Tue, 19 Jun 2018 19:41:45 +0000 (15:41 -0400)]
debug/elf,macho,pe: support compressed DWARF

Since we're going to start compressing DWARF on Windows and maybe
Darwin, copy the ELF support for .zdebug sections to macho and pe. The
code is almost completely the same across the three.

While I was here I added support for compressed .debug_type sections,
which I presume were overlooked before.

Tests will come in a later CL once we can actually generate compressed
PE/Mach-O binaries, since there's no other good way to get test data.

Updates #25927, #11799

Change-Id: Ie920b6a16e9270bc3df214ce601a263837810376
Reviewed-on: https://go-review.googlesource.com/119815
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agocmd/link: enable DWARF compression on Windows
Heschi Kreinick [Tue, 19 Jun 2018 19:39:17 +0000 (15:39 -0400)]
cmd/link: enable DWARF compression on Windows

Simple follow-on to CL 118276. Everything worked except that the
compressed sections need to be aligned at PEFILEALIGN.

Fixes #25927
Updates #11799

Change-Id: Iec871defe30e3e66055d64a5ae77d5a7aca355f5
Reviewed-on: https://go-review.googlesource.com/119816
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agodoc: update "Mac OS X", "OS X" to macOS; bump up to 10.10
Dmitri Shuralyov [Tue, 19 Jun 2018 20:14:00 +0000 (16:14 -0400)]
doc: update "Mac OS X", "OS X" to macOS; bump up to 10.10

The name was "Mac OS X" during versions 10.0 to 10.7.
It was renamed to "OS X" starting from 10.8 until 10.11.
The current name is "macOS" starting with 10.12. [1]

Previous changes (e.g., CL 47252) updated "Mac OS X" to macOS
in some places, but not everywhere. This CL updates remaining
instances for consistency.

Only the pages that display current information were updated;
historical pages such as release notes for older Go releases,
past articles, blog posts, etc., were left in original form.

Rename the "#osx" anchor to "#macos" on /doc/install page,
along with the single reference to it on the same page.
Add an empty div with id="osx" to not break old links.

Update minimum macOS version from 10.8 to 10.10 per #23122.

[1]: https://en.wikipedia.org/wiki/macOS#History

Updates #23122.

Change-Id: I69fe4b85e83265b9d99f447e3cc5230dde094869
Reviewed-on: https://go-review.googlesource.com/119855
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocrypto: panic on illegal input and output overlap
Filippo Valsorda [Thu, 26 Apr 2018 21:35:01 +0000 (17:35 -0400)]
crypto: panic on illegal input and output overlap

Normalized all panic checks and added inexact aliasing panics across
Stream, Block, BlockMode and AEAD implementations.

Also, tweaked the aliasing docs of cipher.AEAD, as they did not account
for the append nature of the API.

Fixes #21624

Change-Id: I075c4415f59b3c06e3099bd9f76de6d12af086bf
Reviewed-on: https://go-review.googlesource.com/109697
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/compile: conditional on -race, disable inline of go:norace
David Chase [Fri, 15 Jun 2018 19:20:57 +0000 (15:20 -0400)]
cmd/compile: conditional on -race, disable inline of go:norace

Adds the appropriate check to inl.go.
Includes tests of both -race+go:norace and plain go:norace.

Fixes #24651.

Change-Id: Id806342430c20baf4679a985d12eea3b677092e0
Reviewed-on: https://go-review.googlesource.com/119195
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
6 years agocmd/compile: more accurate position for select case error message
Robert Griesemer [Tue, 19 Jun 2018 17:20:35 +0000 (10:20 -0700)]
cmd/compile: more accurate position for select case error message

Fixes #25958.

Change-Id: I1f4808a70c20334ecfc4eb1789f5389d94dcf00e
Reviewed-on: https://go-review.googlesource.com/119755
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoruntime: implement procyield properly for ppc64x
Carlos Eduardo Seo [Tue, 29 May 2018 19:00:38 +0000 (16:00 -0300)]
runtime: implement procyield properly for ppc64x

The procyield() function should yield the processor as in other
architectures. On ppc64x, this is achieved by setting the Program
Priority Register to 'low priority' prior to the spin loop, and
setting it back to 'medium-low priority' afterwards.

benchmark                          old ns/op     new ns/op     delta
BenchmarkMakeChan/Byte-8           87.7          86.6          -1.25%
BenchmarkMakeChan/Int-8            107           106           -0.93%
BenchmarkMakeChan/Ptr-8            201           204           +1.49%
BenchmarkMakeChan/Struct/0-8       78.2          79.7          +1.92%
BenchmarkMakeChan/Struct/32-8      196           200           +2.04%
BenchmarkMakeChan/Struct/40-8      236           230           -2.54%
BenchmarkChanNonblocking-8         8.64          8.85          +2.43%
BenchmarkChanUncontended-8         5577          5598          +0.38%
BenchmarkChanContended-8           66106         51529         -22.05%
BenchmarkChanSync-8                451           441           -2.22%
BenchmarkChanSyncWork-8            9155          9170          +0.16%
BenchmarkChanProdCons0-8           1585          1083          -31.67%
BenchmarkChanProdCons10-8          1094          838           -23.40%
BenchmarkChanProdCons100-8         831           657           -20.94%
BenchmarkChanProdConsWork0-8       1471          941           -36.03%
BenchmarkChanProdConsWork10-8      1033          721           -30.20%
BenchmarkChanProdConsWork100-8     730           511           -30.00%
BenchmarkChanCreation-8            135           128           -5.19%
BenchmarkChanSem-8                 602           463           -23.09%
BenchmarkChanPopular-8             3017466       2188441       -27.47%

Fixes #25625

Change-Id: Iacb1c888d3c066902152b8367500348fb631c5f9
Reviewed-on: https://go-review.googlesource.com/115376
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/compile: fix exporting of 'for' loops
Robert Griesemer [Mon, 18 Jun 2018 21:38:45 +0000 (14:38 -0700)]
cmd/compile: fix exporting of 'for' loops

The existing code for encoding 'for' loops in exported, inlineable
functions incorrectly assumed that the 'Right' field points to an
'expression' node. Adjusted the code to be able to handle any kind
of node. Made matching changes for the binary and indexed exporter.

This only shows up together with other pending compiler changes that
enable exporting of such functions which contain for loops.

No tests yet because we can't test this w/o those pending compiler
changes. Once those changes are in, this code will be tested implicitly.
However, the changes were tested manually together with the patches
described in the issue.

Fixes #25222.

Change-Id: I54babb87e5d665d2c1ef6116c1de1b8c50b1138e
Reviewed-on: https://go-review.googlesource.com/119595
Reviewed-by: David Chase <drchase@google.com>
6 years agoruntime: fix lldb test after DWARF compression
Heschi Kreinick [Mon, 18 Jun 2018 20:29:16 +0000 (16:29 -0400)]
runtime: fix lldb test after DWARF compression

Most (all?) released versions of lldb don't support compressed DWARF.
For now, skip the test if lldb can't find where to put the breakpoint.

This is the best I could think of -- there is no explicit error that I
can find that indicates it couldn't load the DWARF.

Fixes #25925.

Change-Id: Ib8fa486a04940cee5959ba7aab7bdbbaa3b2974e
Reviewed-on: https://go-review.googlesource.com/119535
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agoruntime: skip gdb tests on mips after DWARF compression
Heschi Kreinick [Mon, 18 Jun 2018 21:09:37 +0000 (17:09 -0400)]
runtime: skip gdb tests on mips after DWARF compression

DWARF compression appears to break GDB on mips for reasons unknown. Skip
the GDB tests there.

Fixes #25939.

Change-Id: Id76860d3a2ff8055999ac12ea891c37565bb6685
Reviewed-on: https://go-review.googlesource.com/119539
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agoruntime/pprof: fix incorrect assumption in TestMapping
Hana Kim [Mon, 18 Jun 2018 16:03:53 +0000 (12:03 -0400)]
runtime/pprof: fix incorrect assumption in TestMapping

TestMapping assumed that there was only one mapping entry corresponding
to /exe/main, but that is not always true.
This CL changes the test logic to examine whether all referenced mappings
are symbolized. Based on the result, the test determines whether the
corresponding mapping entries' HasFunctions fields to be true or false.

I initially attempted to create two mappings for referenced locations
(one for symbolized and another for unsymbolized) as described in the
TODO in proto.go as part of fixing this bug. But that change requires
non-trivial modification in the upstream profile package so I decided
to just fix the test for now.

Fixes #25891

Change-Id: Id27a5b07bb5b59e133755a0f863bf56c0a4f7f2b
Reviewed-on: https://go-review.googlesource.com/119455
Run-TryBot: Hyang-Ah Hana Kim <hyangah@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/cgo: avoid name confliction for C functions
Hiroshi Ioka [Sat, 6 Jan 2018 09:10:30 +0000 (18:10 +0900)]
cmd/cgo: avoid name confliction for C functions

Use more cryptic names for local variables inside C function wrappers.

Fixes #23356

Change-Id: Ia6a0218f27a13be14f589b1a0facc9683d22ff56
Reviewed-on: https://go-review.googlesource.com/86495
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/cover: fix off-by-one in test error message
Rob Pike [Mon, 18 Jun 2018 03:57:27 +0000 (13:57 +1000)]
cmd/cover: fix off-by-one in test error message

Drive-by after previous CL.

Change-Id: I87db65b65745a0d76500cce06ac276b0d7928404
Reviewed-on: https://go-review.googlesource.com/119395
Reviewed-by: Ralph Corderoy <ralph@inputplus.co.uk>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agodoc: update more stale pprof links
Mostyn Bramley-Moore [Sun, 17 Jun 2018 17:42:03 +0000 (19:42 +0200)]
doc: update more stale pprof links

Related to #25477.

Change-Id: I11261c6055b446ceca1b3acc538ab00fec4b47ca
Reviewed-on: https://go-review.googlesource.com/119321
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/dist: don't test Examples for js/wasm
Brad Fitzpatrick [Sun, 17 Jun 2018 20:30:27 +0000 (20:30 +0000)]
cmd/dist: don't test Examples for js/wasm

Fixes #25913

Change-Id: I4701ec94fa4b07211a8beed85c02ee5aa4fe3eb3
Reviewed-on: https://go-review.googlesource.com/119377
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Richard Musiol <neelance@gmail.com>
6 years agounicode: fix SpecialCase to follow its docs & respect explict no-op mappings
Brad Fitzpatrick [Thu, 7 Jun 2018 18:18:50 +0000 (18:18 +0000)]
unicode: fix SpecialCase to follow its docs & respect explict no-op mappings

If SpecialCase contains an explicit CaseRange with zero deltas,
respect those and don't fall back to the default behavior.

Fixes #25636

Change-Id: Ic554c6b3dd462b1b39c75194eec469b6ff4aa55b
Reviewed-on: https://go-review.googlesource.com/117155
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Marcel van Lohuizen <mpvl@golang.org>
6 years agocmd/compile: add more color choices to ssa.html
Yury Smolsky [Sat, 16 Jun 2018 13:02:48 +0000 (16:02 +0300)]
cmd/compile: add more color choices to ssa.html

It is possible to highlight up to 12 SSA values
and outlinte up to 11 SSA blocks.

Change-Id: Ifefb813ba07ee8856681552b77199f409e03d74d
Reviewed-on: https://go-review.googlesource.com/119318
Run-TryBot: Yury Smolsky <yury@smolsky.by>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
6 years agoruntime: move semaphore ops from system calls to libc calls on iOS
Elias Naur [Thu, 14 Jun 2018 08:23:10 +0000 (10:23 +0200)]
runtime: move semaphore ops from system calls to libc calls on iOS

Change-Id: I1c7a12497c47dd166cc41230d6e5e005edcbc848
Reviewed-on: https://go-review.googlesource.com/118819
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
6 years agoruntime: move semaphore ops from system calls to libc calls on Darwin
Keith Randall [Wed, 13 Jun 2018 21:39:36 +0000 (14:39 -0700)]
runtime: move semaphore ops from system calls to libc calls on Darwin

This CL removes the last of the direct system calls in the runtime package.
This is the last CL for 1.11.

Use libcCall instead of asmcgocall in a few places I accidentally used
the wrong one.

For 1.12, we need to think about whether/how the syscall package
should be moved over to libc.

Update #17490

Change-Id: I4f0bd9cd6023f662f2e29588266fdfae5233898f
Reviewed-on: https://go-review.googlesource.com/118736
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agonet/http: document how Hijack and Request.Context interact, take two
Brad Fitzpatrick [Fri, 15 Jun 2018 20:10:47 +0000 (20:10 +0000)]
net/http: document how Hijack and Request.Context interact, take two

Second try. The previous version (CL 115039 in git rev 3988863) wasn't
accurate.

Fixes #22347

Change-Id: I473165f308c730f50b14ba787cb215f7cb9ea364
Reviewed-on: https://go-review.googlesource.com/119235
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agonet/http: document that Handlers are resposible for validating Host headers
Brad Fitzpatrick [Tue, 29 May 2018 20:19:24 +0000 (20:19 +0000)]
net/http: document that Handlers are resposible for validating Host headers

Fixes #23993

Change-Id: I112415c894e8c680bfc17d53772275430e46794b
Reviewed-on: https://go-review.googlesource.com/115116
Reviewed-by: Tim Cooper <tim.cooper@layeh.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/link: compress DWARF sections in ELF binaries
Heschi Kreinick [Sun, 6 May 2018 01:49:40 +0000 (21:49 -0400)]
cmd/link: compress DWARF sections in ELF binaries

Forked from CL 111895.

The trickiest part of this is that the binary layout code (blk,
elfshbits, and various other things) assumes a constant offset between
symbols' and sections' file locations and their virtual addresses.
Compression, of course, breaks this constant offset. But we need to
assign virtual addresses to everything before compression in order to
resolve relocations before compression. As a result, compression needs
to re-compute the "address" of the DWARF sections and symbols based on
their compressed size. Luckily, these are at the end of the file, so
this doesn't perturb any other sections or symbols. (And there is, of
course, a surprising amount of code that assumes the DWARF segment
comes last, so what's one more place?)

Relevant benchmarks:
name        old time/op     new time/op     delta
StdCmd          10.3s ± 2%      10.8s ± 1%   +5.43%  (p=0.000 n=30+30)

name        old text-bytes  new text-bytes  delta
HelloSize       746kB ± 0%      746kB ± 0%     ~     (all equal)
CmdGoSize      8.41MB ± 0%     8.41MB ± 0%     ~     (all equal)
[Geo mean]     2.50MB          2.50MB        +0.00%

name        old data-bytes  new data-bytes  delta
HelloSize      10.6kB ± 0%     10.6kB ± 0%     ~     (all equal)
CmdGoSize       252kB ± 0%      252kB ± 0%     ~     (all equal)
[Geo mean]     51.5kB          51.5kB        +0.00%

name        old bss-bytes   new bss-bytes   delta
HelloSize       125kB ± 0%      125kB ± 0%     ~     (all equal)
CmdGoSize       145kB ± 0%      145kB ± 0%     ~     (all equal)
[Geo mean]      135kB           135kB        +0.00%

name        old exe-bytes   new exe-bytes   delta
HelloSize      1.60MB ± 0%     1.05MB ± 0%  -34.39%  (p=0.000 n=30+30)
CmdGoSize      16.5MB ± 0%     11.3MB ± 0%  -31.76%  (p=0.000 n=30+30)
[Geo mean]     5.14MB          3.44MB       -33.08%

Fixes #11799.
Updates #6853.

Change-Id: I64197afe4c01a237523a943088051ee056331c6f
Reviewed-on: https://go-review.googlesource.com/118276
Run-TryBot: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/link: compress debug sections in external linking mode
Heschi Kreinick [Mon, 7 May 2018 17:44:22 +0000 (13:44 -0400)]
cmd/link: compress debug sections in external linking mode

Forked from CL 111895.

For #11799.

Change-Id: Ie1346ac2c9122de494823b9058df3a0971e9dfe1
Reviewed-on: https://go-review.googlesource.com/118277
Run-TryBot: Heschi Kreinick <heschi@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agocmd/link: add buildid to wasm modules
Brad Fitzpatrick [Fri, 15 Jun 2018 19:13:51 +0000 (19:13 +0000)]
cmd/link: add buildid to wasm modules

Add Go buildids into a custom wasm section ("go.buildid", arbitrarily)
early in the wasm module, right after the magic & version.

Fixes #25910

Change-Id: If3f7cb267bf8c7beb6fa8d8b7a4829419720bbd8
Reviewed-on: https://go-review.googlesource.com/119175
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
6 years agocmd/go: add dark copy of golang.org/x/vgo
Russ Cox [Fri, 8 Jun 2018 17:59:17 +0000 (13:59 -0400)]
cmd/go: add dark copy of golang.org/x/vgo

This CL corresponds to golang.org/cl/118096 (7fbc8df48a7)
in the vgo repo.

It copies the bulk of the code from vgo back into the main repo,
but completely disabled - vgo.Init is a no-op and vgo.Enabled
returns false unconditionally.

The point of this CL is to make the two trees easier to diff and
to make future syncs smaller.

Change-Id: Ic34fd5ddd8272a70c5a3b3437b5169e967d0ed03
Reviewed-on: https://go-review.googlesource.com/118095
Reviewed-by: Bryan C. Mills <bcmills@google.com>
6 years agold/macho: add all missing load commands
Misty De Meo [Fri, 15 Jun 2018 18:15:50 +0000 (18:15 +0000)]
ld/macho: add all missing load commands

The ld/macho code currently understands a subset of the mach-o load
commands. I've encountered one of these in the wild in a Go-produced
binary, which tripped up the Go linker because its switch statement
expects its list of load commands to be exhaustive; the rest I've
added for the sake of completion.

The ruby-macho library is a good non-Darwin header resource for these:
https://github.com/homebrew/ruby-macho

Fixes #25908

Change-Id: Ib54c065d27e87d8726a9870df05a2bae24828b98
GitHub-Last-Rev: 655e3f488a4c1a89896a40edb0e1f715a5d3f734
GitHub-Pull-Request: golang/go#25906
Reviewed-on: https://go-review.googlesource.com/119115
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agomime: add .wasm for builtin
Yasuhiro Matsumoto [Fri, 15 Jun 2018 09:46:45 +0000 (18:46 +0900)]
mime: add .wasm for builtin

Change-Id: I9d9189ae7882164246abcc4b22d49c26d4a3f201
Reviewed-on: https://go-review.googlesource.com/118937
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/dist: run msan test only in testsanitizer on arm64
Fangming.Fang [Thu, 24 May 2018 08:36:00 +0000 (08:36 +0000)]
cmd/dist: run msan test only in testsanitizer on arm64

With latest gcc (7.3.0), misc/cgo/testsanitizer test will fail with reporting sigmentation
fault when running tsan test. On arm64, tsan is not supported currently and only msan test
can be run. So skip tsan test on arm64.

What needs to be pointed out is that msan test can be really run when setting clang
as c/c++ complier.

Fixes #25601

Change-Id: I6ab1a8d9edd243e2ee00ee40bc0abd6a0e6a125c
Reviewed-on: https://go-review.googlesource.com/114857
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/link: fix name section of WebAssembly binary
Richard Musiol [Thu, 14 Jun 2018 17:22:55 +0000 (19:22 +0200)]
cmd/link: fix name section of WebAssembly binary

Chrome and Node.js were not showing the names of WebAssembly
functions any more. This was due to the name section containing
names also for import functions, which is redundant.

Change-Id: I2f2b2d0b5bd7a59b34f108d2fd7b6ba2eb26f9c9
Reviewed-on: https://go-review.googlesource.com/118976
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agosyscall/js: improve import functions
Richard Musiol [Wed, 13 Jun 2018 13:43:54 +0000 (15:43 +0200)]
syscall/js: improve import functions

1. Make import functions not use the js.Value type directly,
but only the ref field. This gives more flexibility on the Go side
for the js.Value type, which is a preparation for adding
garbage collection of js.Value.

2. Turn import functions which are methods of js.Value into
package-level functions. This is necessary to make vet happy.

Change-Id: I69959bf1fbea0a0b99a552a1112ffcd0c024e9b8
Reviewed-on: https://go-review.googlesource.com/118656
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agonet/http: ensure that Listener.Close is called only once in Server.Serve
Brad Fitzpatrick [Thu, 12 Apr 2018 20:25:07 +0000 (20:25 +0000)]
net/http: ensure that Listener.Close is called only once in Server.Serve

Fixes #24803

Change-Id: I8b1e7c5a74018a0c333f8c38a7ec5f5827ab1606
Reviewed-on: https://go-review.googlesource.com/106715
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
6 years agocmd/link: separate virtual address layout from file layout
Austin Clements [Fri, 4 May 2018 18:55:31 +0000 (14:55 -0400)]
cmd/link: separate virtual address layout from file layout

Currently these two forms of layout are done in a single pass. This
makes it difficult to compress DWARF sections because that must be
done after relocations are applied, which must happen after virtual
address layout, but we can't layout the file until we've compressed
the DWARF sections.

Fix this by separating the two layout steps. In the process, we can
also unify the copy-pasted code in Link.address to compute file
offsets, which currently has some unnecessary variation.

Unlike the current file offset computation, which depends on virtual
addresses, the new computation only uses file offsets and sizes. This
will let us compress the file representation of a segment and create
the file layout based on its on-disk size rather than its original
in-memory size.

Tested by comparing the test binary for the "strings" package on all
supported GOOS/GOARCH combinations. All binaries are identical
(except, of course, their build IDs).

This is a second attempt at CL 111682.

For #11799.
Fixes #25863.

Change-Id: If09f28771bb4d78dd392fd58b8d7c9d5f22b0b9e
Reviewed-on: https://go-review.googlesource.com/118716
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Heschi Kreinick <heschi@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoall: enable vet/all for js/wasm and fix vet issues
Richard Musiol [Fri, 8 Jun 2018 16:12:57 +0000 (18:12 +0200)]
all: enable vet/all for js/wasm and fix vet issues

This commit enables vet/all for the js/wasm architecture. It got
skipped initially because the codebase did not fully compile yet
for js/wasm, which made vet/all fail.

startTimer and stopTimer are not needed in the syscall package.
Removed their assembly code since their Go stubs were already gone.

Change-Id: Icaeb6d903876e51ceb1edff7631f715a98c28696
Reviewed-on: https://go-review.googlesource.com/118657
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agomisc/wasm: add stub for fs.openSync for browsers
Richard Musiol [Thu, 14 Jun 2018 22:08:26 +0000 (00:08 +0200)]
misc/wasm: add stub for fs.openSync for browsers

This stub is necessary so the time package can fail to load
the timezone files in a nice way. It transitively makes the
log package work in browsers.

Change-Id: I4d360df82989d9b40cd31bb4508a6d057534443e
Reviewed-on: https://go-review.googlesource.com/118977
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agomath/big: handle negative exponents in Exp
Brian Kessler [Wed, 13 Jun 2018 16:04:49 +0000 (10:04 -0600)]
math/big: handle negative exponents in Exp

For modular exponentiation, negative exponents can be handled using
the following relation.

for y < 0: x**y mod m == (x**(-1))**|y| mod m

First compute ModInverse(x, m) and then compute the exponentiation
with the absolute value of the exponent.  Non-modular exponentiation
with a negative exponent still returns 1.

Fixes #25865

Change-Id: I2a35986a24794b48e549c8de935ac662d217d8a0
Reviewed-on: https://go-review.googlesource.com/118562
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
6 years agoruntime, sycall/js: add support for callbacks from JavaScript
Richard Musiol [Sat, 19 May 2018 22:56:36 +0000 (00:56 +0200)]
runtime, sycall/js: add support for callbacks from JavaScript

This commit adds support for JavaScript callbacks back into
WebAssembly. This is experimental API, just like the rest of the
syscall/js package. The time package now also uses this mechanism
to properly support timers without resorting to a busy loop.

JavaScript code can call into the same entry point multiple times.
The new RUN register is used to keep track of the program's
run state. Possible values are: starting, running, paused and exited.
If no goroutine is ready any more, the scheduler can put the
program into the "paused" state and the WebAssembly code will
stop running. When a callback occurs, the JavaScript code puts
the callback data into a queue and then calls into WebAssembly
to allow the Go code to continue running.

Updates #18892
Updates #25506

Change-Id: Ib8701cfa0536d10d69bd541c85b0e2a754eb54fb
Reviewed-on: https://go-review.googlesource.com/114197
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/cover: remove use of diff in cover_test.go
Rob Pike [Thu, 14 Jun 2018 01:09:28 +0000 (11:09 +1000)]
cmd/cover: remove use of diff in cover_test.go

It's non-portable, and the test isn't hard to write without diff.
It still produces helpful output in case of trouble:

--- FAIL: TestCoverHTML (0.75s)
    cover_test.go:325: line 4 differs: got:
         case &lt;-ch:<span class="cov0" title="0"></span>
        want:
         case &lt;-ch:<span class="cov0" xitle="0"></span>

This makes the test operating-system independent.

Change-Id: Iff35f00cb76ba89bc1b93db01c6f994e74341f4a
Reviewed-on: https://go-review.googlesource.com/118795
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/vet: simplify side effects func call logic
Daniel Martí [Mon, 11 Jun 2018 08:34:57 +0000 (09:34 +0100)]
cmd/vet: simplify side effects func call logic

Instead of first looking for values of unnamed signature type, first
treat the types and builtins. All the remaining cases will be what we're
after.

Change-Id: I328e22ae0be1cccaeb45ed4ddaa360233d447e7e
Reviewed-on: https://go-review.googlesource.com/117835
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Robert Griesemer <gri@golang.org>
6 years agomath/big: round x + (-x) to -0 for mode ToNegativeInf
Brian Kessler [Fri, 8 Jun 2018 17:48:18 +0000 (11:48 -0600)]
math/big: round x + (-x) to -0 for mode ToNegativeInf

Handling of sign bit as defined by IEEE 754-2008, section 6.3:

When the sum of two operands with opposite signs (or the difference of
two operands with like signs) is exactly zero, the sign of that sum (or
difference) shall be +0 in all rounding-direction attributes except
roundTowardNegative; under that attribute, the sign of an exact zero
sum (or difference) shall be −0. However, x+x = x−(−x) retains the same
sign as x even when x is zero.

This change handles the special case of Add/Sub resulting in exactly zero
when the rounding mode is ToNegativeInf setting the sign bit accordingly.

Fixes #25798

Change-Id: I4d0715fa3c3e4a3d8a4d7861dc1d6423c8b1c68c
Reviewed-on: https://go-review.googlesource.com/117495
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
6 years agocmd/compile: ensure that operand of ORETURN is not double-walked
David Chase [Tue, 12 Jun 2018 18:10:33 +0000 (14:10 -0400)]
cmd/compile: ensure that operand of ORETURN is not double-walked

Inlining of switch statements into a RETURNed expression
can sometimes lead to the switch being walked twice, which
results in a miscompiled switch statement. The bug depends
on:

1) multiple results
2) named results
3) a return statement whose expression includes a call to a
function containing a switch statement that is inlined.

It may also be significant that the default case of that
switch is a panic(), though that's not proven.

Rearranged the walk case for ORETURN so that double walks are
not possible.  Added a test, because this is so fiddly.
Added a check against double walks, verified that it fires
w/o other fix.

Fixes #25776.

Change-Id: I2d594351fa082632512ef989af67eb887059729b
Reviewed-on: https://go-review.googlesource.com/118318
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
6 years agomime/multipart: return error from NextPart if boundary is empty
Constantin Konstantinidis [Thu, 14 Jun 2018 09:51:57 +0000 (11:51 +0200)]
mime/multipart: return error from NextPart if boundary is empty

NewReader cannot return an error. This behaviour is kept.
NextPart returns EOF when boundary is empty.
RFC 2046 does not allow it. The fix is to return an error
on the call of NextPart.

Fixes #23170

Change-Id: I775afd3f93e8b56e6cb274bc5c9de362a18bcc3c
Reviewed-on: https://go-review.googlesource.com/118822
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>

6 years agocmd/go: follow convention for generated code comment
Tobias Klauser [Thu, 14 Jun 2018 07:06:01 +0000 (09:06 +0200)]
cmd/go: follow convention for generated code comment

Follow the convertion (https://golang.org/s/generatedcode) for generated
code in alldocs.go

Change-Id: I03985de20363ba8e09a5a624a931090ec8d196e6
Reviewed-on: https://go-review.googlesource.com/118816
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>