]> Cypherpunks repositories - gostls13.git/log
gostls13.git
5 years agocmd/go: strip trailing slash from versioned arguments
Jay Conrod [Tue, 10 Sep 2019 20:06:12 +0000 (16:06 -0400)]
cmd/go: strip trailing slash from versioned arguments

'go get' accepts arguments of the form path@version, and it passes
them through search.CleanPatterns before querying proxies. With this
change, CleanPatterns preserves text after '@' and will strip trailing
slashes from the patn.

Previously, we did not strip trailing slashes when a version was
present, which caused proxy base URL validation to fail. Module paths
that end with ".go" (for example, github.com/nats-io/nats.go) use
trailing slashes to prevent 'go build' and other commands from
interpreting packages as source file names, so this caused unnecessary
problems for them.

Updates #32483

Change-Id: Id3730c52089e52f1cac446617c20132a3021a808
Reviewed-on: https://go-review.googlesource.com/c/go/+/194600
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocmd/go/internal/modconv: use modules to examine instead of using only direct source...
Aofei Sheng [Wed, 21 Aug 2019 21:23:41 +0000 (05:23 +0800)]
cmd/go/internal/modconv: use modules to examine instead of using only direct source control entries

Since modules now support parsing multiple forms of versions (including
commit hash and source control tag), I think modconv.ConvertLegacyConfig
no longer needs modfetch.ImportRepoRev. So I suggest that we use modules
to convert legacy config instead of using VCS directly. By doing this,
we can make the module proxy participate in the conversion process and
benefit from it (such as speeding up "go mod init" or breaking through
the firewall).

And since modconv.ConvertLegacyConfig is the only caller of
modfetch.ImportRepoRev, I think modfetch.ImportRepoRev can be removed.

Fixes #33767

Change-Id: Ic79b14fa805ed297ca1735a8498cfed2a5ddeec2
Reviewed-on: https://go-review.googlesource.com/c/go/+/191218
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agoencoding/json: encode nil encoding.TextMarshaler instance as "null"
William Poussier [Wed, 11 Sep 2019 15:33:25 +0000 (15:33 +0000)]
encoding/json: encode nil encoding.TextMarshaler instance as "null"

Fixes #34235.

Change-Id: Ia3795fd18860530fa6a4b171545f525e784ffdcb
GitHub-Last-Rev: 1a319c452857818f7aaf22ef46823b43ca9b2276
GitHub-Pull-Request: golang/go#34238
Reviewed-on: https://go-review.googlesource.com/c/go/+/194642
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/internal/obj/riscv: fix up instruction groupings
Joel Sing [Sun, 8 Sep 2019 17:17:43 +0000 (03:17 +1000)]
cmd/internal/obj/riscv: fix up instruction groupings

Some of the instructions were incorrectly grouped - untangle this and
separate the RV64I instructions, which are under separate sections of
the RISC-V specification.

Change-Id: I232962ab4054bf0b4745887506f51e74ea73f73d
Reviewed-on: https://go-review.googlesource.com/c/go/+/194238
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/go/internal/modload: propagate errors from Query for missing imports
Bryan C. Mills [Thu, 8 Aug 2019 21:08:19 +0000 (17:08 -0400)]
cmd/go/internal/modload: propagate errors from Query for missing imports

Updates #30748
Updates #28459

Change-Id: I1c34b3dae0bf9361dba0dae66bb868901ecafe29
Reviewed-on: https://go-review.googlesource.com/c/go/+/189780
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go/internal/get: propagate server errors if no go-import tags are found
Bryan C. Mills [Thu, 8 Aug 2019 21:06:16 +0000 (17:06 -0400)]
cmd/go/internal/get: propagate server errors if no go-import tags are found

Updates #30748

Change-Id: Ic93c68c1c4b2728f383edfdb06371ecc79a6f7b1
Reviewed-on: https://go-review.googlesource.com/c/go/+/189779
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go/internal/get: propagate parse errors in parseMetaGoImports
Bryan C. Mills [Mon, 5 Aug 2019 18:39:48 +0000 (14:39 -0400)]
cmd/go/internal/get: propagate parse errors in parseMetaGoImports

The signature of parseMetaGoImports implies that it can return an error,
but it has not done so since CL 119675. Restore the missing error check,
and remove the named return-values to avoid reintroducing this bug in the
future.

Updates #30748
Updates #21291

Change-Id: Iab19ade5b1c23c282f3c385a55ed277465526515
Reviewed-on: https://go-review.googlesource.com/c/go/+/189778
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go/internal/get: make the implementation of charsetReader match its comment
Bryan C. Mills [Mon, 5 Aug 2019 18:32:46 +0000 (14:32 -0400)]
cmd/go/internal/get: make the implementation of charsetReader match its comment

The doc comment for charsetReader claims that it supports UTF-8,
but in practice it does not: instead, it is never invoked for UTF-8.

We could update the comment to clarify that fact, but it seems simpler
to change the implementation to match the comment.

Change-Id: I39b11395ccef3feff96480b9294e8f2a232728dd
Reviewed-on: https://go-review.googlesource.com/c/go/+/189777
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go/internal/web: log complete URLs
Bryan C. Mills [Wed, 4 Sep 2019 16:43:50 +0000 (12:43 -0400)]
cmd/go/internal/web: log complete URLs

Incoming URLs may omit the scheme to indicate “either HTTP or HTTPS”.
For such URLs, log the scheme actually used instead of leaving it out.

(This issue was noticed while triaging #34075.)

Updates #34075

Change-Id: I39e5ca83543dd780258d41d5c2c4ba907cd20e5c
Reviewed-on: https://go-review.googlesource.com/c/go/+/193262
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go/internal/work: use pie link mode when using MSAN on arm64
fanzha02 [Thu, 8 Aug 2019 01:21:46 +0000 (01:21 +0000)]
cmd/go/internal/work: use pie link mode when using MSAN on arm64

Currently, when running the "CC=clang go run -msan misc/cgo/
testsanitizers/testdata/msan.go" command on arm64, it will
report an error and the error is reported by llvm/compiler-rt/
lib/msan and it is "Make sure to compile with -fPIE and to link
with -pie".

This CL fixes this issue, using PIE link mode when using MSAN
on arm64.

This CL also updates the related document and go build help message.

Fixes #33712

Change-Id: I0cc9d95f3fa264d6c042c27a40ccbb82826922fb
Reviewed-on: https://go-review.googlesource.com/c/go/+/190482
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agocmd/go/internal/modfetch/codehost: treat nonexistent repositories as “not found”
Bryan C. Mills [Thu, 5 Sep 2019 18:27:27 +0000 (14:27 -0400)]
cmd/go/internal/modfetch/codehost: treat nonexistent repositories as “not found”

If a go-import directive refers to a nonexistent repository, today we
treat that as an error fetching a module that actually exists.
That makes the HTTP server responsible for determining which
repositories do or do not exist, which may in general depend on
the user's separately-stored credentials, and imposes significant
complexity on such a server, which can otherwise be very simple.

Instead, check the repository URL and/or error message to try to
determine whether the repository exists at all. If the repo does not
exist, treat its absence as a “not found” error — as if the server had
not returned it in the first place.

Updates #34094

Change-Id: I142619ff43b96d0de428cdd0b01cca828c9ba234
Reviewed-on: https://go-review.googlesource.com/c/go/+/194561
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agoencoding/json: fix and optimize marshal for quoted string
Lucas Bremgartner [Tue, 10 Sep 2019 18:52:02 +0000 (18:52 +0000)]
encoding/json: fix and optimize marshal for quoted string

Since Go 1.2 every string can be marshaled to JSON without error even if it
contains invalid UTF-8 byte sequences. Therefore there is no need to use
Marshal again for the only reason of enclosing the string in double quotes.
Not using Marshal here also removes the error check as there has not been a
way for Marshal to fail anyway.

name          old time/op    new time/op    delta
Issue34127-4     360ns ± 3%     200ns ± 3%  -44.56%  (p=0.008 n=5+5)

name          old alloc/op   new alloc/op   delta
Issue34127-4     56.0B ± 0%     40.0B ± 0%  -28.57%  (p=0.008 n=5+5)

name          old allocs/op  new allocs/op  delta
Issue34127-4      3.00 ± 0%      2.00 ± 0%  -33.33%  (p=0.008 n=5+5)

Fixes #34154

Change-Id: Ib60dc11980f9b20d8bef2982de7168943d632263
GitHub-Last-Rev: 9b0ac1d4c5318b6bf9ed7930320f2bd755f9939c
GitHub-Pull-Request: golang/go#34127
Reviewed-on: https://go-review.googlesource.com/c/go/+/193604
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/compile/internal/s390x: replace 4-byte NOP with a 2-byte NOP on s390x
Ruixin(Peter) Bao [Thu, 22 Aug 2019 13:53:54 +0000 (09:53 -0400)]
cmd/compile/internal/s390x: replace 4-byte NOP with a 2-byte NOP on s390x

Added a new instruction, NOPH, with the encoding [0x0700](i.e: bcr 0, 0) and
replace the current 4-byte nop that was encoded using the WORD instruction.

This reduces the size of .text section in go binary by around 17KB and make
generated code easier to read.

Change-Id: I6a756df39e93c4415ea6d038ba4af001b8ccb286
Reviewed-on: https://go-review.googlesource.com/c/go/+/194344
Reviewed-by: Michael Munday <mike.munday@ibm.com>
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agosyscall: implement rawVforkSyscall for linux/arm64
Howard Zhang [Tue, 6 Aug 2019 09:42:37 +0000 (02:42 -0700)]
syscall: implement rawVforkSyscall for linux/arm64

This allows the use of CLONE_VFORK and CLONE_VM for fork/exec, preventing
"fork/exec ...: cannot allocate memory" failures from occuring when attempting
to execute commands from a Go process that has a large memory footprint.
Additionally, this should reduce the latency of fork/exec on linux/arm64.

With CLONE_VM the child process shares the same memory with the parent
process. On its own this would lead to conflicting use of the same
memory, so CLONE_VFORK is used to suspend the parent process until the
child releases the memory when switching to the new program binary
via the exec syscall. When the parent process continues to run, one
has to consider the changes to memory that the child process did,
namely the return address of the syscall function needs to be restored
from a register.

exec.Command() callers can start in a faster manner, as child process who
do exec commands job can be cloned faster via vfork than via fork on arm64.

The same problem was addressed on linux/amd64 via issue #5838.

Updates #31936
Contributed by Howard Zhang <howard.zhang@arm.com> and Bin Lu <bin.lu@arm.com>

Change-Id: Ia99d81d877f564ec60d19f17e596276836576eaf
Reviewed-on: https://go-review.googlesource.com/c/go/+/189418
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoruntime: fix crash during VDSO calls on arm
Yuichi Nishiwaki [Wed, 11 Sep 2019 02:26:02 +0000 (02:26 +0000)]
runtime: fix crash during VDSO calls on arm

As discussed in #32912, a crash occurs when go runtime calls a VDSO function (say
__vdso_clock_gettime) and a signal arrives to that thread.
Since VDSO functions temporarily destroy the G register (R10),
Go functions asynchronously executed in that thread (i.e. Go's signal
handler) can try to load data from the destroyed G, which causes
segmentation fault.

To fix the issue a guard is inserted in front of sigtrampgo, so that the control escapes from
signal handlers without touching G in case the signal occurred in the VDSO context.
The test case included in the patch is take from discussion in a relevant thread on github:
https://github.com/golang/go/issues/32912#issuecomment-517874531.
This patch not only fixes the issue on AArch64 but also that on 32bit ARM.

Fixes #32912

Change-Id: I657472e54b7aa3c617fabc5019ce63aa4105624a
GitHub-Last-Rev: 28ce42c4a02a060f08c1b0dd1c9a392123fd2ee9
GitHub-Pull-Request: golang/go#34030
Reviewed-on: https://go-review.googlesource.com/c/go/+/192937
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoencoding/base32: increase performance and code reuse
Sven Taute [Mon, 9 Sep 2019 21:56:46 +0000 (21:56 +0000)]
encoding/base32: increase performance and code reuse

Add benchmarks for the Encode/Decode functions operating on []byte and increase decoding performance by removing the calls to strings.Map/bytes.Map and reusing the newline filtering code that is used by NewDecoder.
Cut allocations in half for DecodeString.

Comparison using the new benchmarks:
name            old time/op    new time/op     delta
Encode            16.7µs ± 1%     17.0µs ± 2%    +2.25%  (p=0.000 n=9+9)
EncodeToString    21.1µs ± 1%     20.9µs ± 1%    -0.96%  (p=0.000 n=10+10)
Decode             141µs ± 1%       54µs ± 1%   -61.51%  (p=0.000 n=10+10)
DecodeString      81.4µs ± 0%     54.7µs ± 1%   -32.79%  (p=0.000 n=9+10)

name            old speed      new speed       delta
Encode           492MB/s ± 1%    481MB/s ± 2%    -2.19%  (p=0.000 n=9+9)
EncodeToString   389MB/s ± 1%    392MB/s ± 1%    +0.97%  (p=0.000 n=10+10)
Decode          93.0MB/s ± 1%  241.6MB/s ± 1%  +159.82%  (p=0.000 n=10+10)
DecodeString     161MB/s ± 0%    240MB/s ± 1%   +48.78%  (p=0.000 n=9+10)

Change-Id: Id53633514a9e14ecd0389d52114b2b8ca64370cb
GitHub-Last-Rev: f4be3cf55caf5b89d76d14b7f32422faff39e3c3
GitHub-Pull-Request: golang/go#30376
Reviewed-on: https://go-review.googlesource.com/c/go/+/163598
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoreflect: fix panic in DeepEqual when checking a cycle
Huan Du [Thu, 29 Aug 2019 09:18:53 +0000 (17:18 +0800)]
reflect: fix panic in DeepEqual when checking a cycle

Before this change, when DeepEqual checks values with cycle, it may
panic due to stack overflow.

Here is a sample to reproduce the issue.

    makeCycleMap := func() interface{} {
        cycleMap := map[string]interface{}{}
        cycleMap["foo"] = cycleMap
        return cycleMap
    }

    m1 := makeCycleMap()
    m2 := makeCycleMap()
    reflect.DeepEqual(m1, m2) // stack overflow

The root cause is that DeepEqual fails to cache interface values
in visited map, which is used to detect cycle. DeepEqual calls
CanAddr to check whether a value should be cached or not. However,
all values referenced by interface don't have flagAddr thus all these
values are not cached.

THe fix is to remove CanAddr calls and use underlying ptr in value
directly. As ptr is only read-only in DeepEqual for caching, it's
safe to do so. We don't use UnsafeAddr this time, because this method
panics when CanAddr returns false.

Fixes #33907

Change-Id: I2aa88cc060a2c2192b1d34c129c0aad4bd5597e7
Reviewed-on: https://go-review.googlesource.com/c/go/+/191940
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoreflect: enhance docs for IsZero and IsValid
Marko Kungla [Sat, 7 Sep 2019 09:07:08 +0000 (12:07 +0300)]
reflect: enhance docs for IsZero and IsValid

Make it clear that IsValid checks that we have valid
reflect.Value and not the value of `v`

fixes #34152

Change-Id: Ib3d359eeb3a82bf733b9ed17c777fc4c143bc29c
Reviewed-on: https://go-review.googlesource.com/c/go/+/193841
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agocmd/compile: better integrate parameter tagging with escape.go
Matthew Dempsky [Wed, 4 Sep 2019 22:16:25 +0000 (15:16 -0700)]
cmd/compile: better integrate parameter tagging with escape.go

This CL moves parameter tagging to before escape analysis is complete,
so we still have access to EscLocation. This will be useful once
EscLocation starts tracking higher-fidelity escape details.

Notably, this CL stops using n.Esc to record parameter escape analysis
details. Now escape analysis only ever sets n.Esc to EscNone or
EscHeap. (It still defaults to EscUnknown, and is set to EscNever in
some places though.)

Passes toolstash-check.

Updates #33981.

Change-Id: I50a91ea1e38c442092de6cd14e20b211f8f818c9
Reviewed-on: https://go-review.googlesource.com/c/go/+/193178
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoregexp: add example for NumSubexp
Pantelis Sampaziotis [Tue, 10 Sep 2019 19:00:16 +0000 (19:00 +0000)]
regexp: add example for NumSubexp

Updates #21450

Change-Id: Idf276e97f816933cc0f752cdcd5e713b5c975833
GitHub-Last-Rev: 198e585f92db6e7ac126b49cd751b333e9a44b93
GitHub-Pull-Request: golang/go#33490
Reviewed-on: https://go-review.googlesource.com/c/go/+/189138
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agogo/internal/gccgoimporter: remove guard on some assertions
Than McIntosh [Tue, 10 Sep 2019 18:42:50 +0000 (14:42 -0400)]
go/internal/gccgoimporter: remove guard on some assertions

Remove unnecessary conditional guard for a couple of assertions in the
type parser's update() method (inspired by comment from Robert). No
change in functionality.

Change-Id: I706a54569e75c6960768247889b7dec3f267dde9
Reviewed-on: https://go-review.googlesource.com/c/go/+/194565
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agocmd/go/internal/modfetch: report the module path for errors in (*codeRepo).Versions
Bryan C. Mills [Thu, 5 Sep 2019 18:26:29 +0000 (14:26 -0400)]
cmd/go/internal/modfetch: report the module path for errors in (*codeRepo).Versions

Updates #34094

Change-Id: Ifd10b51c2b4ebe77c4f8f68726e411f54c13b9c9
Reviewed-on: https://go-review.googlesource.com/c/go/+/194560
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go/internal/module: document Version usage for filesystem replacements
Bryan C. Mills [Thu, 5 Sep 2019 16:47:07 +0000 (12:47 -0400)]
cmd/go/internal/module: document Version usage for filesystem replacements

Updates #34085

Change-Id: I57250d0c51a27f0bd3e223588dde9d6d955e0fcf
Reviewed-on: https://go-review.googlesource.com/c/go/+/193618
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agogo/internal/gccgoimporter: support embedded field in pointer loop
Ian Lance Taylor [Tue, 10 Sep 2019 02:22:46 +0000 (19:22 -0700)]
go/internal/gccgoimporter: support embedded field in pointer loop

If an embedded field refers to a type via a pointer, the parser needs
to know the name of the embedded field. It is possible that the
pointer type is not yet resolved. This CL fixes the parser to handle
that case by setting the pointer element type to the unresolved named
type while the pointer is being resolved.

Fixes #34182

Change-Id: I48435e0404362a85effd7463685c502290fa3c57
Reviewed-on: https://go-review.googlesource.com/c/go/+/194440
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years agocmd/compile: refactor escape analysis parameter tagging
Matthew Dempsky [Wed, 4 Sep 2019 22:12:08 +0000 (15:12 -0700)]
cmd/compile: refactor escape analysis parameter tagging

No behavior change; just inverting the loop ordering so the
per-parameter behavior is a bit clearer.

Passes toolstash-check.

Updates #33981.

Change-Id: I9bfcd7d0a4aff65a27ced157767ca2ba8038319a
Reviewed-on: https://go-review.googlesource.com/c/go/+/193177
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoencoding/json: don't reset before returning buffer to pool
andig [Mon, 9 Sep 2019 18:47:43 +0000 (18:47 +0000)]
encoding/json: don't reset before returning buffer to pool

Reset is already performed when retrieving from pool

Change-Id: Ia810dd18d3e55a1565a5ad435a00d1e46724576c
GitHub-Last-Rev: d9df74a4aeb86e5d292c9fc33568a3c9a64a967d
GitHub-Pull-Request: golang/go#34195
Reviewed-on: https://go-review.googlesource.com/c/go/+/194338
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
5 years agocmd/go: for gccgo, look for tool build ID before hashing entire file
Ian Lance Taylor [Tue, 10 Sep 2019 02:54:41 +0000 (19:54 -0700)]
cmd/go: for gccgo, look for tool build ID before hashing entire file

Also fix the key used to store the ID.

This is a significant speedup in cmd/go run time when using an
unreleased toolchain. For example, the TestGoBuildTestOnly cmd/go test
goes from 15 seconds to 1 second.

Change-Id: Ibfd697d55084db059c6b563f70f71f635e935391
Reviewed-on: https://go-review.googlesource.com/c/go/+/194441
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/compile: improve s390x sign/zero extension removal
Michael Munday [Thu, 25 Apr 2019 13:41:46 +0000 (09:41 -0400)]
cmd/compile: improve s390x sign/zero extension removal

This CL gets rid of the MOVDreg and MOVDnop SSA operations on
s390x. They were originally inserted to help avoid situations
where a sign/zero extension was elided but a spill invalidated
the optimization. It's not really clear we need to do this though
(amd64 doesn't have these ops for example) so long as we are
careful when removing sign/zero extensions. Also, the MOVDreg
technique doesn't work if the register is spilled before the
MOVDreg op (I haven't seen that in practice).

Removing these ops reduces the complexity of the rules and also
allows us to unblock optimizations. For example, the compiler can
now merge the loads in binary.{Big,Little}Endian.PutUint16 which
it wasn't able to do before. This CL reduces the size of the .text
section in the go tool by about 4.7KB (0.09%).

Change-Id: Icaddae7f2e4f9b2debb6fabae845adb3f73b41db
Reviewed-on: https://go-review.googlesource.com/c/go/+/173897
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/compile: optimize const rotates for wasm architecture
Agniva De Sarker [Sun, 8 Sep 2019 17:25:01 +0000 (22:55 +0530)]
cmd/compile: optimize const rotates for wasm architecture

This removes the unnecessary code to check whether the shift
is within limits or not when the shift amount is a constant.

The rules hit 23034 times when building std cmd.

grep -E "Wasm.rules:(106|107|121|122|139|140)" rulelog | wc -l
23034

Reduces the size of pkg/js_wasm by 132 bytes.

Change-Id: I64a2b8faca08c3b5039d6a027d4676130d2db18d
Reviewed-on: https://go-review.googlesource.com/c/go/+/194239
Run-TryBot: Agniva De Sarker <agniva.quicksilver@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Richard Musiol <neelance@gmail.com>
5 years agoencoding/json: don't indirect pointers when decoding null
Romain Baugue [Wed, 1 May 2019 12:52:57 +0000 (14:52 +0200)]
encoding/json: don't indirect pointers when decoding null

The indirect method checked the type of the child when indirecting a
pointer. If the current value is a pointer and we are decoding null, we
can skip this entirely and return early, avoiding the whole descent.

Fixes #31776

Change-Id: Ib8b2a2357572c41f56fceac59b5a858980f3f65e
Reviewed-on: https://go-review.googlesource.com/c/go/+/174699
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
5 years agocmd/compile/internal/scanner: report at most one lexical error per number literal
Robert Griesemer [Fri, 30 Aug 2019 00:34:17 +0000 (17:34 -0700)]
cmd/compile/internal/scanner: report at most one lexical error per number literal

Leave reporting of multiple errors for strings alone for now;
we probably want to see all incorrect escape sequences in
runes/strings independent of other errors.

Fixes #33961.

Change-Id: Id722e95f802687963eec647d1d1841bd6ed17d35
Reviewed-on: https://go-review.googlesource.com/c/go/+/192499
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
5 years agocmd/compile: remove vestigial TDDDFIELD
Matthew Dempsky [Mon, 9 Sep 2019 22:58:46 +0000 (15:58 -0700)]
cmd/compile: remove vestigial TDDDFIELD

Change-Id: I4a582f8efcf413665a7513c163334fa8d978a7e9
Reviewed-on: https://go-review.googlesource.com/c/go/+/194437
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agoencoding/asn1: handle ASN1's string type BMPString
sergeilem [Sat, 23 Mar 2019 06:06:08 +0000 (06:06 +0000)]
encoding/asn1: handle ASN1's string type BMPString

This code enables handling of ASN1's string type BMPString, used in some digital signatures.
Parsing code taken from golang.org/x/crypto/pkcs12.

Change-Id: Ibeae9cf4d8ae7c18f8b5420ad9244a16e117ff6b
GitHub-Last-Rev: 694525351411f2ec3982a6bf4ac33be892ce1b12
GitHub-Pull-Request: golang/go#26690
Reviewed-on: https://go-review.googlesource.com/c/go/+/126624
Run-TryBot: Andrew Bonventre <andybons@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Andrew Bonventre <andybons@golang.org>
5 years agopath: improve documentation to call out cases where Clean is called
Eric Rutherford [Mon, 9 Sep 2019 18:56:53 +0000 (13:56 -0500)]
path: improve documentation to call out cases where Clean is called

Rewording the comments for Join to do a better job of calling out
when Clean is called. Also clarifing other portions of the comment.

Fixes #29875

Change-Id: Ied43983bb10a97922898d28af133de0930224496
Reviewed-on: https://go-review.googlesource.com/c/go/+/194339
Reviewed-by: Rob Pike <r@golang.org>
5 years agocmd/compile: report more precise errors about untyped constants
Matthew Dempsky [Fri, 6 Sep 2019 23:05:36 +0000 (16:05 -0700)]
cmd/compile: report more precise errors about untyped constants

Previously, we used a single "untyped number" type for all untyped
numeric constants. This led to vague error messages like "string(1.0)"
reporting that "1 (type untyped number)" can't be converted to string,
even though "string(1)" is valid.

This CL makes cmd/compile more like go/types by utilizing
types.Ideal{int,rune,float,complex} instead of types.Types[TIDEAL],
and keeping n.Type in sync with n.Val().Ctype() during constant
folding.

Thanks to K Heller for looking into this issue, and for the included
test case.

Fixes #21979.

Change-Id: Ibfea88c05704bc3c0a502a455d018a375589754d
Reviewed-on: https://go-review.googlesource.com/c/go/+/194019
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agosrc: gofmt -s
Ainar Garipov [Mon, 9 Sep 2019 06:00:53 +0000 (09:00 +0300)]
src: gofmt -s

Change-Id: I56d7eeaf777ac30886ee77428ca1ac72b77fbf7d
Reviewed-on: https://go-review.googlesource.com/c/go/+/193849
Run-TryBot: Dave Cheney <dave@cheney.net>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agoencoding/json: clarify Unmarshal behavior for map keys
Eli Bendersky [Fri, 2 Aug 2019 21:13:23 +0000 (14:13 -0700)]
encoding/json: clarify Unmarshal behavior for map keys

This is a documentation-only change

Fixes #33298

Change-Id: I816058a872b57dc868dff11887214d9de92d9342
Reviewed-on: https://go-review.googlesource.com/c/go/+/188821
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoA+C: change email address for Akhil Indurti
smasher164 [Sun, 8 Sep 2019 07:34:30 +0000 (03:34 -0400)]
A+C: change email address for Akhil Indurti

Change-Id: I27ca0d71cdc7b13388556aa7c1987320b6b41849
Reviewed-on: https://go-review.googlesource.com/c/go/+/194178
Reviewed-by: Matt Layher <mdlayher@gmail.com>
5 years agoRevert "compile: prefer an AND instead of SHR+SHL instructions"
Martin Möhrmann [Mon, 9 Sep 2019 07:22:14 +0000 (07:22 +0000)]
Revert "compile: prefer an AND instead of SHR+SHL instructions"

This reverts commit 9ec7074a946b7c2812a1a044e84276a36f46d14d.

Reason for revert: broke s390x (copysign, abs) and arm64 (bitfield) tests.

Change-Id: I16c1b389c062e8c4aa5de079f1d46c9b25b0db52
Reviewed-on: https://go-review.googlesource.com/c/go/+/193850
Run-TryBot: Martin Möhrmann <moehrmann@google.com>
Reviewed-by: Agniva De Sarker <agniva.quicksilver@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocompile: prefer an AND instead of SHR+SHL instructions
Martin Möhrmann [Sat, 24 Aug 2019 08:38:23 +0000 (10:38 +0200)]
compile: prefer an AND instead of SHR+SHL instructions

On modern 64bit CPUs a SHR, SHL or AND instruction take 1 cycle to execute.
A pair of shifts that operate on the same register will take 2 cycles
and needs to wait for the input register value to be available.

Large constants used to mask the high bits of a register with an AND
instruction can not be encoded as an immediate in the AND instruction
on amd64 and therefore need to be loaded into a register with a MOV
instruction.

However that MOV instruction is not dependent on the output register and
on many CPUs does not compete with the AND or shift instructions for
execution ports.

Using a pair of shifts to mask high bits instead of an AND to mask high
bits of a register has a shorter encoding and uses one less general
purpose register but is slower due to taking one clock cycle longer
if there is no register pressure that would make the AND variant need to
generate a spill.

For example the instructions emitted for (x & 1 << 63) before this CL are:
48c1ea3f                SHRQ $0x3f, DX
48c1e23f                SHLQ $0x3f, DX

after this CL the instructions are the same as GCC and LLVM use:
48b80000000000000080    MOVQ $0x8000000000000000, AX
4821d0                  ANDQ DX, AX

Some platforms such as arm64 already have SSA optimization rules to fuse
two shift instructions back into an AND.

Removing the general rule to rewrite AND to SHR+SHL speeds up this benchmark:

var GlobalU uint

func BenchmarkAndHighBits(b *testing.B) {
x := uint(0)
for i := 0; i < b.N; i++ {
x &= 1 << 63
}
GlobalU = x
}

amd64/darwin on Intel(R) Core(TM) i7-3520M CPU @ 2.90GHz:
name           old time/op  new time/op  delta
AndHighBits-4  0.61ns ± 6%  0.42ns ± 6%  -31.42%  (p=0.000 n=25+25):

Updates #33826
Updates #32781

Change-Id: I862d3587446410c447b9a7265196b57f85358633
Reviewed-on: https://go-review.googlesource.com/c/go/+/191780
Run-TryBot: Martin Möhrmann <moehrmann@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agosyscall: minor cleanup of duplicated code
Keisuke Kishimoto [Sun, 8 Sep 2019 16:39:46 +0000 (16:39 +0000)]
syscall: minor cleanup of duplicated code

Call the Nano methods of Timespec and Timeval in TimespecToNsec and
TimevalToNsec respectively, instead of duplicating the implementation.

Change-Id: I17551ea54c59c1e45ce472e029c625093a67251a
GitHub-Last-Rev: fecf43d163f4ebe72e8bb1d3854d4ad962c08b03
GitHub-Pull-Request: golang/go#33390
Reviewed-on: https://go-review.googlesource.com/c/go/+/188397
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoall: fix typos
Ainar Garipov [Sun, 8 Sep 2019 16:36:13 +0000 (19:36 +0300)]
all: fix typos

Use the following (suboptimal) script to obtain a list of possible
typos:

  #!/usr/bin/env sh

  set -x

  git ls-files |\
    grep -e '\.\(c\|cc\|go\)$' |\
    xargs -n 1\
    awk\
    '/\/\// { gsub(/.*\/\//, ""); print; } /\/\*/, /\*\// { gsub(/.*\/\*/, ""); gsub(/\*\/.*/, ""); }' |\
    hunspell -d en_US -l |\
    grep '^[[:upper:]]\{0,1\}[[:lower:]]\{1,\}$' |\
    grep -v -e '^.\{1,4\}$' -e '^.\{16,\}$' |\
    sort -f |\
    uniq -c |\
    awk '$1 == 1 { print $2; }'

Then, go through the results manually and fix the most obvious typos in
the non-vendored code.

Change-Id: I3cb5830a176850e1a0584b8a40b47bde7b260eae
Reviewed-on: https://go-review.googlesource.com/c/go/+/193848
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agosyscall: re-generate zsyscall_darwin_arm*.s
Elias Naur [Sat, 7 Sep 2019 23:13:14 +0000 (01:13 +0200)]
syscall: re-generate zsyscall_darwin_arm*.s

I missed that in CL 193843.

Updates #34133

Change-Id: I70b420f022cc7f8289f07375bfc2ade20cf3ffe7
Reviewed-on: https://go-review.googlesource.com/c/go/+/193846
Run-TryBot: Elias Naur <mail@eliasnaur.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agonet,os: disable more sysctl tests on iOS
Elias Naur [Sat, 7 Sep 2019 22:21:57 +0000 (00:21 +0200)]
net,os: disable more sysctl tests on iOS

Updates #34133

Change-Id: I27c75993176cf876f2d80f70982528258c509b68
Reviewed-on: https://go-review.googlesource.com/c/go/+/193845
Run-TryBot: Elias Naur <mail@eliasnaur.com>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agonet/http: make copyValues append instead of calling Add
smasher164 [Wed, 21 Aug 2019 03:11:22 +0000 (23:11 -0400)]
net/http: make copyValues append instead of calling Add

This results in a performance boost:

name          old time/op    new time/op    delta
CopyValues-4    3.46µs ± 3%    1.53µs ± 3%  -55.85%  (p=0.000 n=18+19)

name          old alloc/op   new alloc/op   delta
CopyValues-4    1.52kB ± 0%    0.74kB ± 0%  -51.58%  (p=0.000 n=20+20)

name          old allocs/op  new allocs/op  delta
CopyValues-4      24.0 ± 0%      11.0 ± 0%  -54.17%  (p=0.000 n=20+20)

Fixes #33744.

Change-Id: Ibc653fb076a9a6aaa775fcc9ca720fb90e68cf96
Reviewed-on: https://go-review.googlesource.com/c/go/+/191057
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agolog/syslog: skip unsupported tests on iOS
Elias Naur [Sat, 7 Sep 2019 19:15:55 +0000 (21:15 +0200)]
log/syslog: skip unsupported tests on iOS

CL 193843 disabled sysctl on iOS. This change disables two tests that
rely on sysctl.

Updates #34133

Change-Id: I7c569a1992a50ad6027a294c1fd535cccddcfc4e
Reviewed-on: https://go-review.googlesource.com/c/go/+/193844
Run-TryBot: Elias Naur <mail@eliasnaur.com>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
5 years agosyscall: disable sysctl on iOS
Elias Naur [Sat, 7 Sep 2019 15:53:54 +0000 (17:53 +0200)]
syscall: disable sysctl on iOS

Sysctl is blocked by the App Store submission checks.

Fixes #34133

Change-Id: I9e83cf87e942d6249e9bb67a95dba230e44badd9
Reviewed-on: https://go-review.googlesource.com/c/go/+/193843
Run-TryBot: Elias Naur <mail@eliasnaur.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
5 years agocmd/internal/obj: instructions and registers for RISC-V
Joel Sing [Thu, 5 Sep 2019 17:13:11 +0000 (03:13 +1000)]
cmd/internal/obj: instructions and registers for RISC-V

Start implementing an assembler for RISC-V - this provides register
definitions and instruction mnemonics as defined in the RISC-V
Instruction Set Manual, along with instruction encoding.

The instruction encoding is generated by the parse_opcodes script with
the "opcodes" and "opcodes-pseudo" files from (`make inst.go`):

  https://github.com/riscv/riscv-opcodes

This is based on the riscv-go port:

  https://github.com/riscv/riscv-go

Contributors to the riscv-go port are:

  Amol Bhave <ammubhave@gmail.com>
  Benjamin Barenblat <bbaren@google.com>
  Josh Bleecher Snyder <josharian@gmail.com>
  Michael Pratt <michael@pratt.im>
  Michael Yenik <myenik@google.com>
  Ronald G. Minnich <rminnich@gmail.com>
  Stefan O'Rear <sorear2@gmail.com>

This port has been updated to Go 1.13:

  https://github.com/4a6f656c/riscv-go

Updates #27532

Change-Id: I257b6de87e9864df61a2b0ce9be15968c1227b49
Reviewed-on: https://go-review.googlesource.com/c/go/+/193677
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/asm/internal/arch: consolidate LinkArch handling
Joel Sing [Fri, 6 Sep 2019 18:40:38 +0000 (04:40 +1000)]
cmd/asm/internal/arch: consolidate LinkArch handling

Rather than manually setting the LinkArch for each case, pass the correct
*obj.LinkArch to the arch* function, as is already done for archX86().

Change-Id: I4cf950780aa30a1385e785fb1d26edacb99bda79
Reviewed-on: https://go-review.googlesource.com/c/go/+/193818
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agonet: handle >=2GiB files with sendfile on Windows
Emmanuel T Odeke [Fri, 30 Aug 2019 06:07:43 +0000 (23:07 -0700)]
net: handle >=2GiB files with sendfile on Windows

CL 187037 applied a fix to handle the case where
files larger than 2GiB were not being sendfile-d,
in one shot, rejecting any files whose size was
larger than the 2GiB.

This CL allows files that are larger than limit
by SendFile-ing in chunks of upto 2GiB per chunk.

The test has been excluded as testing with 3GB
requires creating a local file, flushing it
and then doing sendfile which takes a while
and could cause flakes on computers without capacity,
but the test can be retroactively accessed at:
https://go-review.googlesource.com/c/go/+/192518/8/src/net/sendfile_windows_test.go

Fixes #33193.

Change-Id: If57c25bc289aec82b748890ac1ac4f55798d6a5e
Reviewed-on: https://go-review.googlesource.com/c/go/+/192518
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
5 years agocmd/compile: rewrite untyped constant conversion logic
Matthew Dempsky [Thu, 5 Sep 2019 20:33:06 +0000 (13:33 -0700)]
cmd/compile: rewrite untyped constant conversion logic

This CL detangles the hairy mess that was convlit+defaultlit. In
particular, it makes the following changes:

1. convlit1 now follows the standard typecheck behavior of setting
"n.Type = nil" if there's an error. Notably, this means for a lot of
test cases, we now avoid reporting useless follow-on error messages.
For example, after reporting that "1 << s + 1.0" has an invalid shift,
we no longer also report that it can't be assigned to string.

2. Previously, assignconvfn had some extra logic for trying to
suppress errors from convlit/defaultlit so that it could provide its
own errors with better context information. Instead, this extra
context information is now passed down into convlit1 directly.

3. Relatedly, this CL also removes redundant calls to defaultlit prior
to assignconv. As a consequence, when an expression doesn't make sense
for a particular assignment (e.g., assigning an untyped string to an
integer), the error messages now say "untyped string" instead of just
"string". This is more consistent with go/types behavior.

4. defaultlit2 is now smarter about only trying to convert pairs of
untyped constants when it's likely to succeed. This allows us to
report better error messages for things like 3+"x"; instead of "cannot
convert 3 to string" we now report "mismatched types untyped number
and untyped string".

Passes toolstash-check.

Change-Id: I26822a02dc35855bd0ac774907b1cf5737e91882
Reviewed-on: https://go-review.googlesource.com/c/go/+/187657
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agocmd/compile: use CTNIL for pointer-typed OLITERALs
Matthew Dempsky [Thu, 5 Sep 2019 00:45:50 +0000 (17:45 -0700)]
cmd/compile: use CTNIL for pointer-typed OLITERALs

We used to be more aggressive about constant folding in the frontend,
handling expressions that the Go spec does not consider constant;
e.g., "(*int)(unsafe.Pointer(uintptr(200)))". However, that led to a
lot of subtle Go spec conformance issues, so we've since abandoned
that effort (CL 151320), leaving SSA to handle these cases instead.

As such, the only time we now end up with pointer-typed OLITERALs is
when "nil" is implicitly converted to a pointer-typed variable.
Instead of representing these OLITERALs with an CTINT of 0, we can
just use CTNIL.

Saves a few bytes of memory and lines of code.

Change-Id: Ibc5c756b992fdc89c3bdaf4fda3aa352e8e2b101
Reviewed-on: https://go-review.googlesource.com/c/go/+/193437
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agoRevert "cmd/compile: improve errors for invalid conversions of consts"
Matthew Dempsky [Fri, 6 Sep 2019 20:54:36 +0000 (20:54 +0000)]
Revert "cmd/compile: improve errors for invalid conversions of consts"

This reverts commit 2da9c3e0f99b72cf8f3bdca01b57e7a68a546c5b.

Reason for revert: while the new error messages are more informative,
they're not strictly correct. This CL also conflicts with CL 187657.

Change-Id: I1c36cf7e86c2f35ee83a4f98918ee38aa1f59965
Reviewed-on: https://go-review.googlesource.com/c/go/+/193977
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agotext/template: refer to sorted map keys as "ordered" not "comparable" in docs
Mihai Borobocea [Fri, 6 Sep 2019 17:30:51 +0000 (20:30 +0300)]
text/template: refer to sorted map keys as "ordered" not "comparable" in docs

Consistent with the spec's definition of "ordered" and "comparable".

Fixes #34147

Change-Id: Id13186df5343588d80eaebfeb23092596a846d51
Reviewed-on: https://go-review.googlesource.com/c/go/+/193840
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Rob Pike <r@golang.org>
5 years agocmd/compile: rename Etype to ctxType
Matthew Dempsky [Fri, 6 Sep 2019 19:26:57 +0000 (12:26 -0700)]
cmd/compile: rename Etype to ctxType

golang.org/cl/150140 renamed the other Efoo constants to ctxFoo, but
forgot about Etype.

gorename -from '"cmd/compile/internal/gc".Etype -to ctxType

Change-Id: I142dd42ca84a398f8d2316d75ead3331c023b820
Reviewed-on: https://go-review.googlesource.com/c/go/+/193958
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
5 years agocmd/compile: improve errors for invalid conversions of consts
K. "pestophagous" Heller [Wed, 21 Aug 2019 06:13:25 +0000 (23:13 -0700)]
cmd/compile: improve errors for invalid conversions of consts

Follow-up to Change-Id: If6e52c59eab438599d641ecf6f110ebafca740a9

This addresses the remaining tech debt on issue 21979.

The aforementioned previous CL silenced one of two mostly redundant
compiler errors. However, the silenced error was the more expressive
error. This CL now imbues the surviving error with the same level
of expressiveness as the old semi-redundant error.

Fixes #21979

Change-Id: I3273d48c88bbab073fabe53421d801df621ce321
Reviewed-on: https://go-review.googlesource.com/c/go/+/191079
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agocmd/internal/sys: provide architecture definition for riscv64
Joel Sing [Thu, 5 Sep 2019 17:27:12 +0000 (03:27 +1000)]
cmd/internal/sys: provide architecture definition for riscv64

Updates #27532

Change-Id: I7ecf5239d6bc49408a2f155d0f5398ee716fd443
Reviewed-on: https://go-review.googlesource.com/c/go/+/193678
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/link: use gold when dynamic linking on arm,arm64 only on Linux and Android
Yuval Pavel Zholkover [Thu, 5 Sep 2019 05:50:11 +0000 (08:50 +0300)]
cmd/link: use gold when dynamic linking on arm,arm64 only on Linux and Android

Fixes freebsd/arm build.

Change-Id: Id9b1905a5335f86d317dab7514e0ce7cb74aba1d
Reviewed-on: https://go-review.googlesource.com/c/go/+/193537
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agomisc/cgo/test: use __atomic intrinsics instead of __sync
Ian Lance Taylor [Thu, 5 Sep 2019 21:35:28 +0000 (14:35 -0700)]
misc/cgo/test: use __atomic intrinsics instead of __sync

GCC has supported the __atomic intrinsics since 4.7, and clang
supports them as well. They are better than the __sync intrinsics in
that they specify a memory model and, more importantly for our purposes,
they are reliably implemented either in the compiler or in libatomic.

Change-Id: I5e0036ea3300f65c28b1c3d1f3b93fb61c1cd646
Reviewed-on: https://go-review.googlesource.com/c/go/+/193603
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years agocmd/go: add a test for -sectcreate in LDFLAGS
Sjoerd Siebinga [Thu, 22 Aug 2019 14:36:20 +0000 (16:36 +0200)]
cmd/go: add a test for -sectcreate in LDFLAGS

It was already covered by a regex pattern, but it didn't have a test.

To fix the issue for good, added regression test.

Fixes #28832.

Change-Id: I861e3bed92d3b9484fd8671270dbd2e264b10d2d
Reviewed-on: https://go-review.googlesource.com/c/go/+/191311
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/compile: remove tempname usages
Cuong Manh Le [Fri, 6 Sep 2019 10:55:35 +0000 (17:55 +0700)]
cmd/compile: remove tempname usages

CL 59610 merged tempname to tempAt, but some of comments and error
message still refer to tempname. So changing to tempAt instead.

Change-Id: I032f3bedc135d17124b0daaf22c97d0d5ada0a6f
Reviewed-on: https://go-review.googlesource.com/c/go/+/193817
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
5 years agocmd/link: memoize/cache whether plugin.Open symbol available
Than McIntosh [Wed, 4 Sep 2019 15:25:06 +0000 (11:25 -0400)]
cmd/link: memoize/cache whether plugin.Open symbol available

Perform a single lookup of "plugin.Open" at the point where we set the
loaded flag for the context, then cache whether the result is nil, so
that we can consult this cached value later on (instead of having to
look up the symbol each time). This helps speed up the DynLinkingGo()
context method, which is called from within some very hot loops in the
linker (when linking 'hyperkube' from kubernetes, reduces total calls
to "sym.(*Symbols).ROLookup" from 6.5M to 4.3M)

Change-Id: I92a2ea2b21d24f67aec0a7afeef4acc77c095adf
Reviewed-on: https://go-review.googlesource.com/c/go/+/193260
Reviewed-by: Jeremy Faller <jeremy@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agotest: tweak test to avoid unpreemptible loop with gccgo
Than McIntosh [Thu, 5 Sep 2019 19:43:32 +0000 (15:43 -0400)]
test: tweak test to avoid unpreemptible loop with gccgo

This test contains a very tight loop with locking/unlocking that can
wind up as an unpreemptible when compiled with gccgo, depending on
inlining. Tweak the test slightly to avoid this problem.

Change-Id: I155fd2b4bfea961244eb6c6594c24ab03d32d41c
Reviewed-on: https://go-review.googlesource.com/c/go/+/193619
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agotest: add test that failed with gccgo
Than McIntosh [Wed, 4 Sep 2019 16:28:03 +0000 (12:28 -0400)]
test: add test that failed with gccgo

Test with some code that triggered a compilation error bug in gccgo.

Updates #33866.

Change-Id: Ib2f226bbbebbfae33b41037438fe34dc5f2ad034
Reviewed-on: https://go-review.googlesource.com/c/go/+/193261
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/compile: restore lineno before returning in defaultlitreuse
Matthew Dempsky [Thu, 5 Sep 2019 22:14:02 +0000 (15:14 -0700)]
cmd/compile: restore lineno before returning in defaultlitreuse

This affects the position information that's associated with the
implicit OCONV nodes created in assignconvfn.

Not super important and the followup rewrite CL fixes this too, but
fixing it separately is easy and makes toolstash-check happier.

Change-Id: Ifd65dc524b367812d14a4d996647a5d40665fb38
Reviewed-on: https://go-review.googlesource.com/c/go/+/193606
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agocmd/dist: default to clang on OpenBSD
Ian Lance Taylor [Thu, 5 Sep 2019 23:04:06 +0000 (16:04 -0700)]
cmd/dist: default to clang on OpenBSD

OpenBSD ships with GCC 4.2, the last version of GCC that used GPLv2.
As that is quite old (current GCC version is GCC 9, GCC 4.2 was
released in 2007), default to clang.

Change-Id: Ib93e7b4f4f3ffb9e047e60ffca3696d26ab08aac
Reviewed-on: https://go-review.googlesource.com/c/go/+/193621
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/compile: preserve loop depth when evaluating block
Cuong Manh Le [Thu, 5 Sep 2019 04:15:59 +0000 (11:15 +0700)]
cmd/compile: preserve loop depth when evaluating block

Add block method to preserve loop depth when evaluating statements in a
block, so escape analysis can handle looping label more precisely.

Updates #22438

Change-Id: I39b306544a6c0ee3fcbebbe0d0ee735cb71773e6
Reviewed-on: https://go-review.googlesource.com/c/go/+/193517
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
5 years agoerrors: clarify docs for when As returns false
Ian Lance Taylor [Thu, 5 Sep 2019 20:24:26 +0000 (13:24 -0700)]
errors: clarify docs for when As returns false

Change-Id: Ic8d8399f726c1f9376499fdae92bea41632586ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/193602
Reviewed-by: Jonathan Amsterdam <jba@google.com>
5 years agoregexp: add example for ReplaceAll
psampaz [Tue, 6 Aug 2019 16:56:33 +0000 (16:56 +0000)]
regexp: add example for ReplaceAll

Updates #21450

Change-Id: Ia31c20b52bae5daeb33d918234c2f0944a8aeb07
GitHub-Last-Rev: cc8554477024277c3c1b4122344e9d14427680b3
GitHub-Pull-Request: golang/go#33489
Reviewed-on: https://go-review.googlesource.com/c/go/+/189137
Run-TryBot: Sylvain Zimmer <sylvinus@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agocmd/internal/obj/wasm: refactor handling of wasm variables
Richard Musiol [Wed, 21 Aug 2019 19:57:59 +0000 (21:57 +0200)]
cmd/internal/obj/wasm: refactor handling of wasm variables

This commit improves how registers get mapped to wasm variables. This
is a preparation for future improvements (e.g. adding 32 bit float
registers).

Change-Id: I374c80b2d6c9bcce6b0e373fe921b5ad4dee40ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/191777
Run-TryBot: Richard Musiol <neelance@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoruntime: use hard heap goal if we've done more scan work than expected
Michael Knyszek [Thu, 27 Jun 2019 18:46:29 +0000 (14:46 -0400)]
runtime: use hard heap goal if we've done more scan work than expected

This change makes it so that if we're already finding ourselves in a
situation where we've done more scan work than expected in the
steady-state (that is, 50% of heap_scan for GOGC=100), then we fall back
on the hard heap goal instead of continuing to assume the expected case.

In some cases its possible that we're already doing more scan work than
expected, and if GC assists come in just at that window where we notice
it, they might accumulate way too much assist credit, causing undue heap
growths if GOMAXPROCS=1 (since the fractional background worker isn't
guaranteed to fire). This case seems awfully specific, and that's
because it's exactly the case for TestGcSys, which has been flaky for
some time as a result.

Fixes #28574, #27636, and #27156.

Change-Id: I771f42bed34739dbb1b84ad82cfe247f70836031
Reviewed-on: https://go-review.googlesource.com/c/go/+/184097
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/go/internal/test: prepend -test.timeout rather than appending
Bryan C. Mills [Wed, 4 Sep 2019 16:59:41 +0000 (12:59 -0400)]
cmd/go/internal/test: prepend -test.timeout rather than appending

Tests may accept positional arguments, in which case the -test.timeout
flag must be passed before those arguments.

Fixes #34072

Change-Id: I5b92d7c0edc4f9e1efb63b0733937b76236c0eff
Reviewed-on: https://go-review.googlesource.com/c/go/+/193297
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoruntime: platform-independent faketime support
Austin Clements [Sun, 1 Sep 2019 14:49:27 +0000 (10:49 -0400)]
runtime: platform-independent faketime support

This adds a platform-independent implementation of nacl's faketime
support. It can be enabled by setting the faketime build tag.

Updates #30439.

Change-Id: Iee097004d56d796e6d2bfdd303a092c067ade87e
Reviewed-on: https://go-review.googlesource.com/c/go/+/192740
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agosyscall: redirect writes to runtime.write in faketime mode
Austin Clements [Sun, 1 Sep 2019 18:19:19 +0000 (14:19 -0400)]
syscall: redirect writes to runtime.write in faketime mode

If the faketime build tag is set, this causes syscall.Write for FDs 1
and 2 to redirect to runtime.write, since that's where we'll apply the
faketime framing. This is equivalent to what nacl currently does in
naclFile.write.

We do this on all of the platforms except nacl, which has its own
faketime support and we're about to remove, and Windows, which would
require other changes to support faketime so we're leaving alone for
now.

Updates #30439.

Change-Id: I138a5ca63577d92d15b5437d037bd3159fa84ee7
Reviewed-on: https://go-review.googlesource.com/c/go/+/192739
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoruntime: wrap nanotime, walltime, and write
Austin Clements [Sun, 1 Sep 2019 14:37:44 +0000 (10:37 -0400)]
runtime: wrap nanotime, walltime, and write

In preparation for general faketime support, this renames the existing
nanotime, walltime, and write functions to nanotime1, walltime1, and
write1 and wraps them with trivial Go functions. This will let us
inject different implementations on all platforms when faketime is
enabled.

Updates #30439.

Change-Id: Ice5ccc513a32a6d89ea051638676d3ee05b00418
Reviewed-on: https://go-review.googlesource.com/c/go/+/192738
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/go/internal/get: avoid panic in metaImportsForPrefix if web.Get fails
Bryan C. Mills [Wed, 4 Sep 2019 15:51:06 +0000 (11:51 -0400)]
cmd/go/internal/get: avoid panic in metaImportsForPrefix if web.Get fails

Fixes #34049

Change-Id: I817b83ee2d0ca6d01ec64998f14bc4f32e365d66
Reviewed-on: https://go-review.googlesource.com/c/go/+/193259
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
5 years agoerrors: fix wrong code in package doc
Jonathan Amsterdam [Wed, 4 Sep 2019 17:01:16 +0000 (13:01 -0400)]
errors: fix wrong code in package doc

You can't call Unwrap on the return value of fmt.Errorf, but
you can pass the result to errors.Unwrap.

Also, move the description of the Unwrap function up so the
example makes sense.

Fixes #34061.

Change-Id: Ica07c44665c5e65deea4aa6a146fc543a5a0a99d
Reviewed-on: https://go-review.googlesource.com/c/go/+/193298
Run-TryBot: Jonathan Amsterdam <jba@google.com>
Reviewed-by: Katie Hockman <katie@golang.org>
5 years agospec: clarify that shift count must be non-negative
Robert Griesemer [Wed, 4 Sep 2019 15:42:12 +0000 (08:42 -0700)]
spec: clarify that shift count must be non-negative

Fixes #34056.

Change-Id: I2c9b7a20d19f458df5dcc376e29bee6be1f09f7a
Reviewed-on: https://go-review.googlesource.com/c/go/+/193277
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agoruntime: don't hold worldsema across mark phase
Michael Anthony Knyszek [Mon, 17 Jun 2019 19:03:09 +0000 (19:03 +0000)]
runtime: don't hold worldsema across mark phase

This change makes it so that worldsema isn't held across the mark phase.
This means that various operations like ReadMemStats may now stop the
world during the mark phase, reducing latency on such operations.

Only three such operations are still no longer allowed to occur during
marking: GOMAXPROCS, StartTrace, and StopTrace.

For the former it's because any change to GOMAXPROCS impacts GC mark
background worker scheduling and the details there are tricky.

For the latter two it's because tracing needs to observe consistent GC
start and GC end events, and if StartTrace or StopTrace may stop the
world during marking, then it's possible for it to see a GC end event
without a start or GC start event without an end, respectively.

To ensure that GOMAXPROCS and StartTrace/StopTrace cannot proceed until
marking is complete, the runtime now holds a new semaphore, gcsema,
across the mark phase just like it used to with worldsema.

Fixes #19812.

Change-Id: I15d43ed184f711b3d104e8f267fb86e335f86bf9
Reviewed-on: https://go-review.googlesource.com/c/go/+/182657
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoall: add a space before +build in build tag comments
Tobias Klauser [Wed, 4 Sep 2019 09:40:03 +0000 (11:40 +0200)]
all: add a space before +build in build tag comments

Add a space before build tag comments so it corresponds to the format
documented at https://golang.org/pkg/go/build/.

Change-Id: I8349d0343597e304b97fb5479847231ed8945b1a
Reviewed-on: https://go-review.googlesource.com/c/go/+/193237
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocrypto/x509: remove IsCA exception for broken Entrust root
Filippo Valsorda [Tue, 3 Sep 2019 18:01:57 +0000 (14:01 -0400)]
crypto/x509: remove IsCA exception for broken Entrust root

The exception allowed a specific intermediate [1] to chain up to a
broken root that lacked the CA:TRUE X509v3 Basic Constraint.

The broken root [2] is expiring at the end of 2019, so we can remove the
exception in Go 1.14.

Moreover, there is a reissued version of that root [3] (same Subject and
SPKI, valid CA) which expires in 2029, so root stores should have
migrated to it already, making the exception unnecessary.

[1]: https://crt.sh/?caid=57
[2]: https://crt.sh/?id=1616049
[3]: https://crt.sh/?id=55

Change-Id: I43f51100982791b0e8bac90d143b60851cd46dfc
Reviewed-on: https://go-review.googlesource.com/c/go/+/193038
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/compile,runtime: generate hash functions only for types which are map keys
Keith Randall [Tue, 6 Aug 2019 22:22:51 +0000 (15:22 -0700)]
cmd/compile,runtime: generate hash functions only for types which are map keys

Right now we generate hash functions for all types, just in case they
are used as map keys. That's a lot of wasted effort and binary size
for types which will never be used as a map key. Instead, generate
hash functions only for types that we know are map keys.

Just doing that is a bit too simple, since maps with an interface type
as a key might have to hash any concrete key type that implements that
interface. So for that case, implement hashing of such types at
runtime (instead of with generated code). It will be slower, but only
for maps with interface types as keys, and maybe only a bit slower as
the aeshash time probably dominates the dispatch time.

Reorg where we keep the equals and hash functions. Move the hash function
from the key type to the map type, saving a field in every non-map type.
That leaves only one function in the alg structure, so get rid of that and
just keep the equal function in the type descriptor itself.

cmd/go now has 10 generated hash functions, instead of 504. Makes
cmd/go 1.0% smaller. Update #6853.

Speed on non-interface keys is unchanged. Speed on interface keys
is ~20% slower:

name                  old time/op  new time/op  delta
MapInterfaceString-8  23.0ns ±21%  27.6ns ±14%  +20.01%  (p=0.002 n=10+10)
MapInterfacePtr-8     19.4ns ±16%  23.7ns ± 7%  +22.48%   (p=0.000 n=10+8)

Change-Id: I7c2e42292a46b5d4e288aaec4029bdbb01089263
Reviewed-on: https://go-review.googlesource.com/c/go/+/191198
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Martin Möhrmann <moehrmann@google.com>
5 years agodoc: add Go 1.13 to release history
Andrew [Tue, 3 Sep 2019 20:00:13 +0000 (16:00 -0400)]
doc: add Go 1.13 to release history

Change-Id: I3340561c0b17bf28d8d480e00f5bc8afb2a897ef
Reviewed-on: https://go-review.googlesource.com/c/go/+/193041
Run-TryBot: Andrew Bonventre <andybons@golang.org>
Reviewed-by: Katie Hockman <katie@golang.org>
5 years agocmd/compile/internal/gc: use GoToolPath in TestDeps
Ian Lance Taylor [Tue, 3 Sep 2019 17:30:24 +0000 (10:30 -0700)]
cmd/compile/internal/gc: use GoToolPath in TestDeps

Updates #31563
Fixes #34041

Change-Id: Ib9fdcd2f83d867fd31b42eab3a813f5cef88860e
Reviewed-on: https://go-review.googlesource.com/c/go/+/193077
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
5 years agobytes/hash: disable seed test on 32-bit platforms
Keith Randall [Tue, 3 Sep 2019 17:27:48 +0000 (10:27 -0700)]
bytes/hash: disable seed test on 32-bit platforms

The distribution of hash outputs when varying the seed is
not good enough on 32-bit platforms.

This isn't a likely attack vector (the adversary controlling the seed),
so it isn't a huge deal. Would be nice to fix, though. For now, just
skip this test.

Fixes #33988

Change-Id: Icf51ab687fc215422a5492ae78e6f414b33e04cb
Reviewed-on: https://go-review.googlesource.com/c/go/+/193078
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/compile: extend ssa.go to handle 1-element array and 1-field struct
Cuong Manh Le [Tue, 3 Sep 2019 16:24:35 +0000 (23:24 +0700)]
cmd/compile: extend ssa.go to handle 1-element array and 1-field struct

Assinging to 1-element array/1-field struct variable is considered clobbering
the whole variable. By emitting OpVarDef in this case, liveness analysis
can now know the variable is redefined.

Also, the isfat is not necessary anymore, and will be removed in follow up CL.

Fixes #33916

Change-Id: Iece0d90b05273f333d59d6ee5b12ee7dc71908c2
Reviewed-on: https://go-review.googlesource.com/c/go/+/192979
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
5 years agodebug/dwarf: better handling for DW_FORM_indirect
Than McIntosh [Wed, 14 Aug 2019 19:04:05 +0000 (15:04 -0400)]
debug/dwarf: better handling for DW_FORM_indirect

Fix a buglet in abbrev processing related to DW_FORM_indirect. When
reading an abbrev entry if we encounter an attribute with form
DW_FORM_indirect, leave the class as ClassUnknown, then when the
abbrev is walked during the reading of the DIE fill in the class based
on the value read at that point (code for handling DW_FORM_indirect
seems to be already partially in place in the DIE reader).

Updates #33488.

Change-Id: I9dc89abf5cc8d7ea96824c0011bef979de0540bf
Reviewed-on: https://go-review.googlesource.com/c/go/+/190158
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
5 years agocmd/compile: silence esc diagnostics about directiface OCONVIFACEs
Matthew Dempsky [Fri, 30 Aug 2019 17:56:30 +0000 (10:56 -0700)]
cmd/compile: silence esc diagnostics about directiface OCONVIFACEs

In general, a conversion to interface type may require values to be
boxed, which in turn necessitates escape analysis to determine whether
the boxed representation can be stack allocated.

However, esc.go used to unconditionally print escape analysis
decisions about OCONVIFACE, even for conversions that don't require
boxing (e.g., pointers, channels, maps, functions).

For test compatibility with esc.go, escape.go similarly printed these
useless diagnostics. This CL removes the diagnostics, and updates test
expectations accordingly.

Change-Id: I97c57a4a08e44d265bba516c78426ff4f2bf1e12
Reviewed-on: https://go-review.googlesource.com/c/go/+/192697
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/compile: replace copytype to setUnderlying
Matthew Dempsky [Fri, 30 Aug 2019 22:59:16 +0000 (15:59 -0700)]
cmd/compile: replace copytype to setUnderlying

While here, change the params to be easier to understand: "t" is now
always the type being updated, and "underlying" is now used to
represent the underlying type.

Updates #33658.

Change-Id: Iabb64414ca3abaa8c780e4c9093e0c77b76fabf9
Reviewed-on: https://go-review.googlesource.com/c/go/+/192724
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agocmd/compile: simplify {defer,resume}checkwidth logic
Matthew Dempsky [Fri, 30 Aug 2019 21:54:21 +0000 (14:54 -0700)]
cmd/compile: simplify {defer,resume}checkwidth logic

This CL extends {defer,resume}checkwidth to support nesting, which
simplifies usage.

Updates #33658.

Change-Id: Ib3ffb8a7cabfae2cbeba74e21748c228436f4726
Reviewed-on: https://go-review.googlesource.com/c/go/+/192721
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agodebug/dwarf, debug/elf: support DWARF 5
Ian Lance Taylor [Fri, 3 May 2019 05:15:05 +0000 (22:15 -0700)]
debug/dwarf, debug/elf: support DWARF 5

Change-Id: I6e9d47865c198299d497911c58235cd40f775e34
Reviewed-on: https://go-review.googlesource.com/c/go/+/175138
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years agodebug/dwarf: add new constants defined in DWARF 5
Ian Lance Taylor [Thu, 2 May 2019 00:48:46 +0000 (17:48 -0700)]
debug/dwarf: add new constants defined in DWARF 5

Also add a few other missing constants.

Also rerun current version of stringer.

Change-Id: I004ef27f6b40fe2cab64c52d169255dfea43fa01
Reviewed-on: https://go-review.googlesource.com/c/go/+/175137
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
5 years agoruntime: one lock per order
Changkun Ou [Thu, 1 Aug 2019 16:22:28 +0000 (16:22 +0000)]
runtime: one lock per order

This CL implements one lock per order of stackpool. It improves performance when mutator stack growth deeply, see benchmark below:

```
name                old time/op  new time/op  delta
StackGrowth-8       3.60ns ± 5%  3.59ns ± 1%    ~     (p=0.794 n=10+9)
StackGrowthDeep-8    370ns ± 1%   335ns ± 1%  -9.47%  (p=0.000 n=9+9)
StackCopyPtr-8      72.6ms ± 0%  71.6ms ± 1%  -1.31%  (p=0.000 n=9+9)
StackCopy-8         53.5ms ± 0%  53.2ms ± 1%  -0.54%  (p=0.006 n=8+9)
StackCopyNoCache-8   100ms ± 0%    99ms ± 0%  -0.70%  (p=0.000 n=8+8)
```

Change-Id: I1170d3fd9e6ff8516e25f669d0aaf1861311420f
GitHub-Last-Rev: 13b820cddd8008079c98d01ac22dd123a46a6603
GitHub-Pull-Request: golang/go#33399
Reviewed-on: https://go-review.googlesource.com/c/go/+/188478
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agoplugin: add freebsd/amd64 plugin support
Ayan George [Fri, 23 Aug 2019 15:38:12 +0000 (11:38 -0400)]
plugin: add freebsd/amd64 plugin support

Change-Id: I4e7b0e99fd0f6f7d6a8ef60e9d3cb5baeb80a2b6
Reviewed-on: https://go-review.googlesource.com/c/go/+/191617
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agodoc/go1.13: remove announcements of arm64 support on NetBSD and OpenBSD
Bryan C. Mills [Tue, 3 Sep 2019 13:28:17 +0000 (09:28 -0400)]
doc/go1.13: remove announcements of arm64 support on NetBSD and OpenBSD

Those configurations currently lack builders, and it is not clear to
me what state their tests are in.

The Go porting policy¹ requires builders for all active ports, so
let's not claim support until that requirement is met.

¹https://golang.org/wiki/PortingPolicy#requirements-for-a-new-port

Updates #30824
Updates #31656
Updates #34035
Updates #34036

Change-Id: I6496de9d92fb8546048abf139cf10546a47e314b
Reviewed-on: https://go-review.googlesource.com/c/go/+/192997
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
5 years agoRevert "runtime: remove slow time compatibility hacks for wine"
Daniel Martí [Mon, 2 Sep 2019 14:34:20 +0000 (14:34 +0000)]
Revert "runtime: remove slow time compatibility hacks for wine"

This reverts CL 191759.

Reason for revert: broke most Go programs using the time package on Wine,
including on 4.15, the latest stable version. Only wine-staging (with
experimental patches) contains an upstream fix we could rely on.

Change-Id: Ic8ba126022e54f412174042fbb9abed82d5eb318
Reviewed-on: https://go-review.googlesource.com/c/go/+/192622
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
5 years agodoc: document Go 1.13
Andrew [Mon, 2 Sep 2019 20:33:07 +0000 (16:33 -0400)]
doc: document Go 1.13

Change-Id: Icf7e1dab82aa48cc693eb4da8a564dff23312741
Reviewed-on: https://go-review.googlesource.com/c/go/+/192745
Run-TryBot: Andrew Bonventre <andybons@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
5 years agodoc/1.13: remove draft note and make various fixes
Andrew Bonventre [Wed, 28 Aug 2019 20:07:39 +0000 (16:07 -0400)]
doc/1.13: remove draft note and make various fixes

Fixes #33954

Change-Id: Idfe71bf825adcd7cbf70cd139b3e779963394ff6
Reviewed-on: https://go-review.googlesource.com/c/go/+/192105
Run-TryBot: Andrew Bonventre <andybons@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Katie Hockman <katie@golang.org>
5 years agocmd/doc: make -src mode deterministic
Ivan Trubach [Mon, 2 Sep 2019 13:57:09 +0000 (13:57 +0000)]
cmd/doc: make -src mode deterministic

These changes make cmd/doc -src deterministic, or, more precisely,
go/ast.MergePackageFiles, which is used by cmd/doc. So far the order of
comments depended on the package file map iteration order.

cmd/doc with -src flag has been inserting and omitting random comments
ever since the addition of -src flag. After investigating the code path
with the debugger, I’ve noticed that ast.File.Comments slice order changes
between invocations of the command. The bug was introduced in 3e24f2d,
which ironically claimed to “fix formatting of -src output”. The commit
implemented the collection of comments by iterating over the map and
c7cdce1 “godoc: make ?m=src mode deterministic” did’t actually make
go/ast.MergePackageFiles deterministic.

I’ve found this issue after running “go doc -src sync.WaitGroup.Wait”.
There are likely other packages and functions affected, but the bug
should be somewhat reproducible across all Go versions.

Change-Id: Iae223e99550c0a3b54005c5cde36f909e655b66b
GitHub-Last-Rev: c49532f79fcd7529d7ad1a369c467abc59d4df0a
GitHub-Pull-Request: golang/go#33553
Reviewed-on: https://go-review.googlesource.com/c/go/+/189477
Run-TryBot: Agniva De Sarker <agniva.quicksilver@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agonet: document concurrency safety and example for Dialer
tomocy [Mon, 2 Sep 2019 14:34:16 +0000 (14:34 +0000)]
net: document concurrency safety and example for Dialer

Fixes #33743.

Change-Id: I80621321d56b6cf312a86e272800f1ad03c5544c
GitHub-Last-Rev: d91cb3697516ccfc3f956e97837404cbfee5922f
GitHub-Pull-Request: golang/go#33856
Reviewed-on: https://go-review.googlesource.com/c/go/+/191879
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>