// package and imports snipped
func TestCache(t *testing.T) {
tmp := os.TempDir()
for i := 0; i < 1000000; i++ {
os.Stat(filepath.Join(tmp, fmt.Sprintf("%d", i)))
}
}
ijt:~/gopath/src/issue26562$ time ~/github/go/bin/go test -count=1
PASS
ok issue26562 9.444s
real 0m10.021s
user 0m2.344s
sys 0m7.835s
ijt:~/gopath/src/issue26562$ time ~/github/go/bin/go test .
ok issue26562 (cached)
real 0m0.802s
user 0m0.551s
sys 0m0.306s
Updates #26562
Updates #26726
Change-Id: I7914ee57dc75bcbd7f0ea01c70bed97d67c810ea
Reviewed-on: https://go-review.googlesource.com/127715
Run-TryBot: Ian Lance Taylor <iant@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
math: ensure Erfc is not called with out-of-expected-range arguments on s390x
The existing implementation produces correct results with a wide range of inputs,
but invalid results asymptotically. With this change we ensure correct asymptotic results
on s390x
Fixes #26477
Change-Id: I760c1f8177f7cab2d7622ab9a926dfb1f8113b49
Reviewed-on: https://go-review.googlesource.com/127119 Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
// package and imports snipped
func TestCache(t *testing.T) {
tmp := os.TempDir()
for i := 0; i < 1000000; i++ {
os.Stat(filepath.Join(tmp, fmt.Sprintf("%d", i)))
}
}
ijt:~/gopath/src/issue26562$ time ~/github/go/bin/go test -count=1
PASS
ok issue26562 9.444s
real 0m10.021s
user 0m2.344s
sys 0m7.835s
ijt:~/gopath/src/issue26562$ time ~/github/go/bin/go test .
ok issue26562 (cached)
real 0m0.802s
user 0m0.551s
sys 0m0.306s
Change-Id: I3ce7f7b68bb5b9e802069f277e79e1ed3c162622
Reviewed-on: https://go-review.googlesource.com/127635
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Ian Lance Taylor [Thu, 2 Aug 2018 19:35:32 +0000 (12:35 -0700)]
go/scanner: continue adding directory to file name
Before CL 97795, when go/scanner saw a //line comment, it would clean
the path and, if the path was relative, prepend the directory from the
file name. This was not the best API because it meant that the
behavior changed based on whether the code was running on Windows or
not, and it meant that information from the //line directive was lost.
So in CL 97795, among other changes, go/scanner was changed to simply
return the filename given in the //line comment.
Unfortunately existing tools such as unparam and unconvert expected
the old behavior. In order to avoid breaking those tools, revert that
part of the change.
Fixes #26671
Change-Id: Ifa06542bd19cda9d682ac33766ab9080444ba050
Reviewed-on: https://go-review.googlesource.com/127658
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Richard Musiol [Thu, 2 Aug 2018 20:39:00 +0000 (22:39 +0200)]
cmd/link/internal/wasm: add Go version as a custom wasm section
The interface between the wasm binary and wasm_exec.js is experimental
and likely to change in the future. Still, there are some early adopters
who experiment with non-web wasm runtimes. They can't use wasm_exec.js
and have to provide their own equivalent. Adding the Go version as a
custom wasm sections allows for them to support a stable Go version and
the latest devel at the same time.
Brad Fitzpatrick [Thu, 2 Aug 2018 20:42:00 +0000 (20:42 +0000)]
encoding/xml: document HTMLAutoClose and HTMLEntity more
They didn't even have public types, which made them pretty mysterious.
Give them types and reference the Decoder, which uses them.
Also, refer them qualified by their package name in the examples, as
we usually do in example*.go files, which usually use package foo_test
specifically so we can show the package names along with the symbols.
Change-Id: I50ebbbf43778c1627bfa526f8824f52c7953454f
Reviewed-on: https://go-review.googlesource.com/127663 Reviewed-by: Bryan C. Mills <bcmills@google.com>
Brad Fitzpatrick [Thu, 2 Aug 2018 19:34:38 +0000 (19:34 +0000)]
syscall: use deprecation convention in package comment, simplify
Change-Id: I612041d31c01d49135947796fe2a09db3e6894d4
Reviewed-on: https://go-review.googlesource.com/127657 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Xudong Zheng [Wed, 1 Aug 2018 21:33:09 +0000 (21:33 +0000)]
syscall/js: move callback helper code to misc/wasm to avoid using eval()
When using the compiled .wasm with misc/wasm/wasm_exec.js, we get an error message if the site prohibits eval() via the Content-Security-Policy header. This can be resolved by moving the callback helper code from src/syscall/js/callback.go to misc/wasm/wasm_exec.js.
David Wimmer [Wed, 25 Jul 2018 18:44:07 +0000 (18:44 +0000)]
runtime: fix syscall error returns on mips/mips64/ppc64
The linux syscall functions used in runtime are designed around the calling
convention of returning errors as negative numbers. On some other systems
(like mips and ppc) the actual syscalls signal errors in other ways. This
means that the assembly implementations of the syscall functions on these
platforms need to transform the return values in the error cases to match
the expected negative errno values. This was addressed for certain syscalls
in https://golang.org/cl/19455 and https://golang.org/cl/89235. This patch
handles the rest of the syscall functions in sys_linux_*.s that return any
value for mips/mips64/ppc64.
Fixes #23446
Change-Id: I302100261231f76d5850ab2c2ea080170d7dba72
GitHub-Last-Rev: e358e2b08c76897b13f917cfa12b5085e20337f9
GitHub-Pull-Request: golang/go#26606
Reviewed-on: https://go-review.googlesource.com/125895
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Keith Randall [Wed, 1 Aug 2018 04:31:19 +0000 (21:31 -0700)]
cmd/cgo: stop expanding typedefs once we reach __builtin types
Expanding __builtin types (__builtin_va_list, particularly) leads
to problems because they are expanded by the compiler itself - the
expansions are not generated by anything in a .h file. The types
a __builtin type expand to are thus very confusing to cgo.
See CL 126275.
Change-Id: I66eb6a4f27f652f1b934ba702f580f6daa62a566
Reviewed-on: https://go-review.googlesource.com/127156
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Previously, cmd/doc treated GOROOT/src and GOPATH/src
as the roots of the directory trees holding packages, assuming
that the import path would be the path elements after the src directory.
With modules, each module serves as its own root of a file tree,
and the import path prefix starts with the module path before
adding the path elements after the module root.
There are ways we could make this more efficient,
but for now this is a fairly small adjustment to get 'go doc'
working OK for modules for Go 1.11.
cmd/go: move module cache from $GOPATH/src/mod to $GOPATH/pkg/mod
Using $GOPATH/src/mod confuses too many tools.
$GOPATH/pkg/mod seems better for now.
It's also next to dep's cache, $GOPATH/pkg/dep.
If we do eliminate GOPATH/pkg for holding .a files (#4719)
then we could still keep it around for pkg/mod.
(Or we could move the module cache again then.)
Fixes #26401.
Fixes #26635.
Change-Id: I18f7da216ed9f490eded3c00d837fb086ae5b6a4
Reviewed-on: https://go-review.googlesource.com/126755
Run-TryBot: Russ Cox <rsc@golang.org> Reviewed-by: Bryan C. Mills <bcmills@google.com> Reviewed-by: Rob Pike <r@golang.org>
Relative directory paths have always worked.
This CL makes absolute directory paths be handled the same way.
(It was an oversight that they were excluded.)
It also fixes the case of naming the directory holding source code
for a package in a module dependency.
The old -getmode flag had two settings:
-getmode=local meant don't download from the network.
-getmode=vendor meant only use the vendor directory.
The new -mod flag has two settings:
-mod=readonly means refuse to automatically update go.mod (mainly for CI testing).
-mod=vendor means only use the vendor directory.
The old GOPROXY variable had two settings:
a proxy URL or else the empty string (direct connect).
The new GOPROXY variable has three settings:
a proxy URL, the string "off" (no network use allowed),
or else the empty string or the explicit string "direct" (direct connection).
We anticipate allow a comma-separated sequence in a future release,
so commas are disallowed entirely right now.
People sometimes want to turn on a particular go command flag by default.
In Go 1.11 we have at least two different cases where users may need this.
1. Linking can be noticeably slower on underpowered systems
due to DWARF, and users may want to set -ldflags=-w by default.
2. For modules, some users or CI systems will want vendoring always,
so they want -getmode=vendor (soon to be -mod=vendor) by default.
This CL generalizes the problem to “set default flags for the go command.”
$GOFLAGS can be a space-separated list of flag settings, but each
space-separated entry in the list must be a standalone flag.
That is, you must do 'GOFLAGS=-ldflags=-w' not 'GOFLAGS=-ldflags -w'.
The latter would mean to pass -w to go commands that understand it
(if any do; if not, it's an error to mention it).
cmd/go: change list -compiled to populate new CompiledGoFiles list
CL 108156 added -cgo during the Go 1.11 cycle.
To avoid adding a new field to Package, it redefined the
meaning of the CgoFiles list to be the cgo output instead
of the cgo input.
This was awkward in the go command itself, since the meaning
of the list changed midway through the build.
But, worse, it is awkward to users of go list.
When gathering information about a tree of packages,
we may want the names of both the cgo inputs and the cgo outputs
(golang.org/x/tools/go/packages does, it turns out),
or when combined with -deps (CL 107776),
we may only care about one list or the other depending
on whether the package was requested explicitly or is
being returned as a dependency.
Also, it's not general enough. SWIGFiles turn into cgo files
and then end up in the list too. And maybe there will be others
in the future. What clients really want is the list of files that
are presented to the go compiler, so that they can parse
and type-check them as if they were the compiler instead.
Eliminate all this awkwardness by dropping -cgo and adding
a new -compiled that populates a new CompiledGoFiles list.
Change-Id: I5f152da17cfb2692eedde61721d01ec13067c57d
Reviewed-on: https://go-review.googlesource.com/126695
Run-TryBot: Russ Cox <rsc@golang.org> Reviewed-by: Michael Matloob <matloob@golang.org> Reviewed-by: Bryan C. Mills <bcmills@google.com>
The current "go mod" command does too many things.
The design is unclear.
It looks like "everything you might want to do with modules"
which causes people to think all module operations go through
"go mod", which is the opposite of the seamless integration we're
going for. In particular too many people think "go mod -require"
and "go get" are the same.
It does make sense to put the module-specific functionality
under "go mod", but not as flags. Instead, split "go mod" into
multiple subcommands:
go mod edit # old go mod -require ...
go mod fix # old go mod -fix
go mod graph # old go mod -graph
go mod init # old go mod -init
go mod tidy # old go mod -sync
go mod vendor # old go mod -vendor
go mod verify # old go mod -verify
Splitting out the individual commands makes both the docs
and the implementations dramatically easier to read.
It simplifies the command lines
(go mod -init -module m is now 'go mod init m')
and allows command-specific flags.
We've avoided subcommands in the go command to date, and we
should continue to avoid adding them unless it really makes
the experience significantly better. In this case, it does.
Creating subcommands required some changes in the core
command-parsing and help logic to generalize from one
level to multiple levels.
As part of having "go mod init" be a separate command,
this CL changes the failure behavior during module initialization
to be delayed until modules are actually needed.
Initialization can still happen early, but the base.Fatalf
is delayed until something needs to use modules.
This fixes a bunch of commands like 'go env' that were
unhelpfully failing with GO111MODULE=on when not in a
module directory.
We aren't planning to use this or advertise it much yet,
but having support for it now will make it easier to start
using in the future - older go commands will understand
what 'go 1.20' means and that they don't have go 1.20.
Saves ~105KB of heap for callers who don't use html.UnescapeString.
(EscapeString is much more common).
Also saves 70KB of binary size, because now the linker can do dead
code elimination. (because #2559 is still open and global maps always
generate init code)
Kir Kolyshkin [Mon, 30 Jul 2018 14:05:49 +0000 (17:05 +0300)]
cmd/cgo: add column number to line directives
Due to a new specification of //line: directives, missing
column info is now treated as column 0, aka "unknown column"
(see https://github.com/golang/go/issues/24183 for details).
As cgo does not add column number to generated //line: directive,
resulting files parsed do not have column info.
Fix by adding column of 1 to generated line directives.
Fixes #26692
Change-Id: Ie9263c0cf666b92d19c34240e745e8f32ffe7174
Reviewed-on: https://go-review.googlesource.com/126675
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Keith Randall [Thu, 26 Jul 2018 23:33:27 +0000 (16:33 -0700)]
cmd/cgo: make sure we FinishType everything
Ensure that we call FinishType on all the types added to the ptrs map.
We only add a key to ptrKeys once. Once we FinishType for that key,
we'll never look at that key again. But we can add a new type under that
key later, and we'll never finish it.
Make sure we add the key to the ptrKeys list every time we make the list
of types for that key non-empty.
This makes sure we FinishType each pointer type exactly once.
Fixes #26517
Change-Id: Iad86150d516fcfac167591daf5a26c38bec7d143
Reviewed-on: https://go-review.googlesource.com/126275 Reviewed-by: Ian Lance Taylor <iant@golang.org>
cmd/go/internal/modfetch/codehost: quote arguments to commands printed from -x
Some of the arguments — particularly format strings passed to git commands — may
contain spaces, and it's useful to be able to paste commands from 'go get -x
foo' directly into a shell to reproduce their output.
Richard Musiol [Sat, 28 Jul 2018 11:45:02 +0000 (13:45 +0200)]
syscall: remove support for O_NONBLOCK and O_SYNC on js/wasm
This commit removes O_NONBLOCK on js/wasm. O_SYNC can't be
removed, because it is referenced by the os package, so instead
its use returns an error.
On Windows, the options O_NONBLOCK and O_SYNC are not available
when opening a file with Node.js. This caused the initialization
of the syscall package to panic.
The simplest solution is to not support these two options on js/wasm
at all. Code written for js/wasm is supposed to be portable,
so platform-specific options should not be used.
Richard Musiol [Tue, 31 Jul 2018 13:17:31 +0000 (15:17 +0200)]
syscall/js: improve documentation about mappings to JavaScript values
This commit moves the documentation about how Go values are mapped to
JavaScript values to the functions that apply the mapping, instead of
mentioning them in the documentation of the types being mapped. This
should be easier to read.
Roger Peppe [Fri, 20 Jul 2018 10:31:39 +0000 (11:31 +0100)]
go/doc: do not treat methods as test functions
The example code was treating a method starting with Test
as a test function when considering whether to produce
a whole-file example or not. As a method can never be
a test function, this isn't correct.
Change-Id: Idd8ec9eaf0904af076e941d7fe7d967f6b7eef78
Reviewed-on: https://go-review.googlesource.com/125257 Reviewed-by: Daniel Martí <mvdan@mvdan.cc> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org> Reviewed-by: Andrew Bonventre <andybons@golang.org>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Paul Jolly [Mon, 30 Jul 2018 21:31:01 +0000 (22:31 +0100)]
doc: add link to go command modules docs from Go 1.11 release notes
The FAQ already has a link to the release notes and the go command docs.
Add a link from the release notes to the go command docs, to ensure that
people ultimately end up there (the docs that then signpost
people to the relevant other help docs).
cmd/go: forbid use of one module with two different paths
If a single module is imported via two different paths (e.g., as itself and as a
replacement for something else), some users may be surprised if the two paths
do not share the same package-level state. Others may be surprised if the two
paths do share state.
Punt on the question for now by rejecting that condition explicitly.
cmd/go/internal/modload: inline doImport into doPkg
doImport is itself a thin wrapper around Import, and doPkg is its only call
site. I'm having trouble following what doPkg is doing due to the indirection,
so I'm removing it.
Michael Matloob [Fri, 27 Jul 2018 20:23:16 +0000 (16:23 -0400)]
cmd/go: add test code packages in list -test
Previously go list -test <pkg> would return pkg and, if it exists,
pkg.test, the test main package. This change will also list the
two test code packages (if they exist) that contain testing functions,
<pkg> [<pkg>.test] and <pkg>_test [<pkg>.test].
These packages which contain testing code are usually the packages
go list users desire to know about, so they should be surfaced
in go list -test.
John Gibb [Sun, 29 Jul 2018 15:19:17 +0000 (15:19 +0000)]
path: present the correct error message in TestMatch
The message was hardcoded to indicate that the test wanted a nil error, even
though in some cases a specific error was wanted. This patch fixes the
message to print the wanted error.
David du Colombier [Sun, 29 Jul 2018 11:15:25 +0000 (13:15 +0200)]
cmd/go: fix TestScript/mod_gobuild_import on Plan 9
CL 125296 added TestScript/mod_gobuild_import. This
test is failing on Plan 9, because go/build invokes
the go tool which cannot be found in the path.
The "PATH" environment variable has been updated to
contain the path to the go tool on Unix and Windows,
but on Plan 9, the analogous environment variable is
called "path".
This change fixes the script engine by setting
the "path" environment variable on Plan 9.
Fixes #26669.
Change-Id: If1be50e14baceccee591f4f76b7e698f5e12a2d4
Reviewed-on: https://go-review.googlesource.com/126608
Run-TryBot: David du Colombier <0intro@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Constantin Konstantinidis [Sat, 28 Jul 2018 08:16:40 +0000 (10:16 +0200)]
cmd/go: uses SID of group Guests to test ACL
The test TestACL failed when ran on a Windows set
up in another language as the "Guest" account
name is translated. The SID of the group of Guests
always exist and is used instead.
Fixes #26658
Change-Id: Ia885d08a9e50563787e389c2d2dc2547881a2943
Reviewed-on: https://go-review.googlesource.com/126598 Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
cmd/go: ignore unknown directives in dependency go.mod files
This will help with forward compatibility when we add additional
directives that only matter for the main module (or that can be
safely ignored otherwise).
Change-Id: Ida1e186fb2669b128aeb5a9a1187e2535b72b763
Reviewed-on: https://go-review.googlesource.com/125936 Reviewed-by: Bryan C. Mills <bcmills@google.com>
cmd/go: add go list -m -f {{.GoMod}} to show path to go.mod file
"go env GOMOD" gives this for the main module already
but it's useful to be able to query other modules.
Using {{.Dir}} does not work if the go.mod was auto-synthesized.
Change-Id: If4844571e9e429b541de0d40c36ff4c5743b2031
Reviewed-on: https://go-review.googlesource.com/125656 Reviewed-by: Bryan C. Mills <bcmills@google.com>
cmd/cgo: document that #including source files in subdirectories is a bad idea
Suggested in discussion on #26366.
Change-Id: Id9ad2e429a915f88b4c4b30fc415c722eebe0ea4
Reviewed-on: https://go-review.googlesource.com/125297 Reviewed-by: Ian Lance Taylor <iant@golang.org>
go/build: invoke go command to find modules during Import, Context.Import
The introduction of modules has broken (intentionally) the rule
that the source code for a package x/y/z is in GOPATH/src/x/y/z
(or GOROOT/src/x/y/z). This breaks the code in go/build.Import,
which uses that rule to find the directory for a package.
In the long term, the fix is to move programs that load packages
off of go/build and onto golang.org/x/tools/go/packages, which
we hope will eventually become go/packages. That code invokes
the go command to learn what it needs to know about where
packages are.
In the short term, though, there are lots of programs that use go/build
and will not be able to find code in module dependencies.
To help those programs, go/build now runs the go command to
ask where a package's source code can be found, if it sees that
modules are in use. (If modules are not in use, it falls back to the
usual lookup code and does not invoke the go command, so that
existing uses are unaffected and not slowed down.)
Helps #24661.
Fixes #26504.
Change-Id: I0dac68854cf5011005c3b2272810245d81b7cc5a
Reviewed-on: https://go-review.googlesource.com/125296 Reviewed-by: Michael Matloob <matloob@golang.org> Reviewed-by: Bryan C. Mills <bcmills@google.com>
net/http: document Transport.Proxy's https support
The net/http module added support for HTTPS proxies in CL 68550, but the
Transport.Proxy docstring was never updated to reflect this. This (doc-only)
update adds "https" to the list of supported schemes.
Ian Lance Taylor [Fri, 27 Jul 2018 17:48:18 +0000 (10:48 -0700)]
syscall: support Faccessat flags argument
The Linux kernel faccessat system call does not take a flags parameter.
The flag parameter to the C library faccessat function is implemented in C.
The syscall.Faccessat function takes a flags parameter. In older releases
we have passed the flags parameter to the kernel, which ignored it.
In CL 120015 we started returning an error if any flags were set.
That seems clearly better than ignoring them, but it turns out that some
code was using the flags. The code was previously subtly broken.
Now it is obviously broken. That is better, but we can do better still:
we can implement the flags as the C library does. That is what this CL does.
Change-Id: I259bd6f240c3951e939b81c3032dead3d9c567b4
Reviewed-on: https://go-review.googlesource.com/126415
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Change-Id: I03b699391351a30ee60a15d7aa712c6c66444cf9
Reviewed-on: https://go-review.googlesource.com/125875 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Constantin Konstantinidis [Thu, 26 Jul 2018 14:01:34 +0000 (16:01 +0200)]
cmd/link: added test of availability of gcc
The test RuntimeTypeAttr always failed when gcc
was unavailable. The test is duplicated for internal
and external linking. The usual verification
of host linker is added at the beginning of the
external link test.
Ian Lance Taylor [Wed, 25 Jul 2018 22:24:27 +0000 (15:24 -0700)]
cmd/cgo: don't resolve typedefs in -godefs mode
In -godefs mode any typedefs that appear in struct fields and the like
will presumably be defined in the input file. If we resolve to the
base type, those cross-references will not work. So for -godefs mode,
keep the Go 1.10 behavior and don't resolve the typedefs in a loop.
Fixes #26644
Change-Id: I48cf72d9eb5016353c43074e6aff6495af326f35
Reviewed-on: https://go-review.googlesource.com/125995 Reviewed-by: Keith Randall <khr@golang.org>
Martin Möhrmann [Fri, 20 Jul 2018 14:41:26 +0000 (16:41 +0200)]
crypto/x509: skip TestSystemRoots
cgo and non-cgo code paths can disagree
on the number of root certificates:
=== RUN TestSystemRoots
--- FAIL: TestSystemRoots (0.31s)
root_darwin_test.go:31: cgo sys roots: 93.605184ms
root_darwin_test.go:32: non-cgo sys roots: 213.998586ms
root_darwin_test.go:44: got 168 roots
root_darwin_test.go:44: got 427 roots
root_darwin_test.go:73: insufficient overlap between cgo and non-cgo roots; want at least 213, have 168
FAIL
exit status 1
Daniel Martí [Fri, 27 Jul 2018 10:24:00 +0000 (11:24 +0100)]
cmd/go/internal/par: fix TestWorkParallel retries
When the test retried multiple times, it reused the same Work variable,
causing in the builders being flaky due to panics. I was able to
immediately reproduce the failure with stress and -race:
$ go test -race -c && stress -p 32 ./par.test -test.run=TestWorkParallel$
/tmp/go-stress909062277
--- FAIL: TestWorkParallel (0.07s)
panic: par.Work.Do: already called Do [recovered]
panic: par.Work.Do: already called Do
Instead, use a new Work variable at each retry. Now, the line above
seems to never fail. Of course, much higher 'stress -p' values will
still result in "does not seem to be parallel" test failures since the
machine lacks resources. But those are test failures, not panics.
Fixes #26642.
Change-Id: I5e962eca7602cf413d911ff5669f56d4f52da5a7
Reviewed-on: https://go-review.googlesource.com/126355
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Bryan C. Mills <bcmills@google.com>
David Wimmer [Wed, 25 Jul 2018 19:16:07 +0000 (19:16 +0000)]
syscall: implement pipe() on linux/mips
Change the Pipe() function to use the pipe() syscall (which has a unique
calling convention on linux/mips) instead of using pipe2(). This allows
it work on kernels <2.6.27 when pipe2() was introduced.
Change-Id: I65dfbd2a02b64e777a8eb13013d718e356521be6
GitHub-Last-Rev: c483a06168387c2cf151b6419e4e16576fab640a
GitHub-Pull-Request: golang/go#26608
Reviewed-on: https://go-review.googlesource.com/125915
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Vladimir Stefanovic <vladimir.stefanovic@mips.com> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Minaev Mike [Fri, 26 Jan 2018 09:17:46 +0000 (09:17 +0000)]
crypto/tls: fix deadlock when Read and Close called concurrently
The existing implementation of TLS connection has a deadlock. It occurs
when client connects to TLS server and doesn't send data for
handshake, so server calls Close on this connection. This is because
server reads data under locked mutex, while Close method tries to
lock the same mutex.
Ian Lance Taylor [Wed, 25 Jul 2018 17:03:40 +0000 (10:03 -0700)]
cmd/go: for missing binary-only package, say where it should be
Before this CL the user effectively has to guess at the expected
location of a binary-only package. While the location is normally
obvious ($GOPATH/pkg/GOOS_GOARCH/PATH/PKG.a) it is much less so when
building with options that implicitly add an -installsufix option.
Ian Lance Taylor [Wed, 25 Jul 2018 21:09:02 +0000 (14:09 -0700)]
runtime: ignore GNU/Linux sigaction errors for signals 32 and 33
This avoids problems when running under QEMU. It seems that at least
some QEMU versions turn the sigaction implementation into a call to
the C library sigaction function. The C library function will reject
attempts to set the signal handler for signals 32 and 33. Ignore
errors in that case.
Change-Id: Id443a9a32f6fb0ceef5c59a398e7ede30bf71646
Reviewed-on: https://go-review.googlesource.com/125955
Run-TryBot: Ian Lance Taylor <iant@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Ian Gudger [Tue, 24 Jul 2018 23:19:19 +0000 (16:19 -0700)]
net: fix handling of Conns created by Resolver.Dial
The DNS client in net is documented to treat Conns returned by
Resolver.Dial which implement PacketConn as UDP and those which don't as
TCP regardless of what was requested. golang.org/cl/37879 changed the
DNS client to assume that the Conn returned by Resolver.Dial was the
requested type which broke compatibility.
Filippo Valsorda [Fri, 30 Mar 2018 23:09:59 +0000 (19:09 -0400)]
net/http: fix and normalize the [Server.][ListenAnd]Serve[TLS] docs
The only inaccurate part was the HTTP/2 caveat in Server.ServeTLS, which
only applies to the plain Serve variant.
The restriction implemented in shouldConfigureHTTP2ForServe is not on
the setupHTTP2_ServeTLS codepath because ServeTLS owns the tls.Listener,
so we fix it for the user instead of disabling HTTP/2.
Keith Randall [Mon, 23 Jul 2018 20:09:48 +0000 (13:09 -0700)]
cmd/compile: set stricter inlining threshold in large functions
If we're compiling a large function, be more picky about how big
the function we're inlining is. If the function is >5000 nodes,
we lower the inlining threshold from a cost of 80 to 20.
Turns out reflect.Value's cost is exactly 80. That's the function
at issue in #26546.
20 was chosen as a proxy for "inlined body is smaller than the call would be".
Simple functions still get inlined, like this one at cost 7:
There are a lot more smaller (~1000 node) functions in the stdlib.
The function in #26546 has 12477 nodes.
At some point it might be nice to have a better heuristic for "inlined
body is smaller than the call", a non-cliff way to scale down the cost
as the function gets bigger, doing cheaper inlined calls first, etc.
All that can wait for another release. I'd like to do this CL for
1.11.
Fixes #26546
Update #17566
Change-Id: Idda13020e46ec2b28d79a17217f44b189f8139ac
Reviewed-on: https://go-review.googlesource.com/125516
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
Dmitry Vyukov [Tue, 26 Jun 2018 11:28:04 +0000 (13:28 +0200)]
net/rpc: clarify requirements for connections and codecs
1. Connections and codecs need to be partially safe for concurrent use.
Namely, read side is serialized by one mutex,
and writing side is serialized by another.
Current comment says that they need to be fully thread-safe,
which makes the default implementations (gobClientCodec/gobServerCodec)
non-conforming.
2. Say that ServerCodec.Close can be called multiple times
and must be idempotent. Server requires this and gobServerCodec
accounts for this, but the requirement is not documented.
Change-Id: Ie877e37891fed28056e3d9d1722edaed8e154067
Reviewed-on: https://go-review.googlesource.com/120818 Reviewed-by: Rob Pike <r@golang.org>
net/http: document that Client methods always return *url.Error
Updates #9424
Change-Id: If117ba3e7d031f84b30d3a721ef99fe622734de2
Reviewed-on: https://go-review.googlesource.com/125575 Reviewed-by: Ian Lance Taylor <iant@golang.org>
There's a 50ms threshold in net/http.Transport that this test
sometimes hitting on slower devices. That was unrelated to what this
test was trying to test. So instead just t.Skip on RoundTrip errors
unless the failure was quick (under 25ms), in which case the error
must've been about something else. Our fast machines should catch
regressions there.
Fixes #25366
Change-Id: Ibe8e2716a5b68558b57d0b8b5c46f38e46a2cba2
Reviewed-on: https://go-review.googlesource.com/125555
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
net/http: update Serve docs on when HTTP/2 is enabled
Contains portions and modified portions of CL 103815
Fixes #24607
Change-Id: Ic330850a0f098f183315f04ea4780eded46c5b77
Reviewed-on: https://go-review.googlesource.com/125515 Reviewed-by: Ian Lance Taylor <iant@golang.org>
misc/cgo/test: disable some Darwin tests in cgo mode
For unknown reasons, linking against CoreFoundation on macOS 10.10
sometimes causes mmap to ignore the hint address, which makes the Go
allocator incompatible with TSAN. Currently, the effect of this is to
run the allocator out of arena hints on the very first allocation,
causing a "too many address space collisions for -race mode" panic.
This CL skips the cgo tests that link against CoreFoundation in race
mode.
Updates #26475.
Updates #26513.
Change-Id: I52ec638c99acf5d4966e68ff0054f7679680dac6
Reviewed-on: https://go-review.googlesource.com/125304
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Daniel Martí [Thu, 5 Jul 2018 13:17:46 +0000 (14:17 +0100)]
doc: explain minor change to gofmt in go1.11
Add the tools section with a Gofmt sub-section, just like in
go1.10.html. Instead of copying the two last paragraphs from 1.10, which
warn users about the hidden complexity of enforcing gofmt, move that to
go/format and link to it.
While at it, remove a duplicate "Tools" header that was likely added by
accident.
Fixes #26228.
Change-Id: Ic511c44b2b86f82a41f2b78dd7e7482d694b6c62
Reviewed-on: https://go-review.googlesource.com/122295 Reviewed-by: Ian Lance Taylor <iant@golang.org>