]> Cypherpunks repositories - gostls13.git/log
gostls13.git
5 years agoruntime: negate errno value for mips pipe/pipe2
Ian Lance Taylor [Sun, 22 Mar 2020 19:57:36 +0000 (12:57 -0700)]
runtime: negate errno value for mips pipe/pipe2

The callers expect negative errno values, so negate them when necessary.

No test because there is no reasonable way to make pipe/pipe2 fail.
This was reported on a system on which pipe2 returned ENOSYS.

Fixes #37997

Change-Id: I3ad6cbbc2521cf495f8df6ec991a3f781122b508
Reviewed-on: https://go-review.googlesource.com/c/go/+/224592
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
5 years agosyscall/js: make wasm_exec.js compatible with Webpack
nao20010128nao [Sat, 21 Mar 2020 06:52:58 +0000 (06:52 +0000)]
syscall/js: make wasm_exec.js compatible with Webpack

In Webpack, require("fs") will always be empty. This behavior throws an error: "fs.writeSync is not function". It happens when you did "fmt.Println".
This PR avoids such problem and use polyfill in wasm_exec.js on Webpack.

Change-Id: I55f2c75ce86b7f84d2d92e8e217b5decfbe3c8a1
GitHub-Last-Rev: aecc847e3f9d5617ea4b00196ef2810c2458f085
GitHub-Pull-Request: golang/go#35805
Reviewed-on: https://go-review.googlesource.com/c/go/+/208600
Reviewed-by: Richard Musiol <neelance@gmail.com>
5 years agosyscall/js: allow copyBytesTo(Go|JS) to use Uint8ClampedArray
Aurélio A. Heckert [Sun, 22 Mar 2020 23:38:42 +0000 (20:38 -0300)]
syscall/js: allow copyBytesTo(Go|JS) to use Uint8ClampedArray

closes #38011

Change-Id: Ic50f2f27456dccdc3fca1bda076871af1eb81705
Reviewed-on: https://go-review.googlesource.com/c/go/+/224638
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Richard Musiol <neelance@gmail.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/go: list test packages even when the main package has an error
Michael Matloob [Mon, 23 Mar 2020 17:52:37 +0000 (13:52 -0400)]
cmd/go: list test packages even when the main package has an error

List test packages (when list is run with -e) even when the main package
has an error. This is useful to get complete data for go/packages.

Fixes #37971

Change-Id: If6ba0270a319ea5e003d1ed8b1ad39e479e95509
Reviewed-on: https://go-review.googlesource.com/c/go/+/224944
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agoRevert "crypto/rsa,crypto/ecdsa,crypto/ed25519: implement PublicKey.Equal"
Bryan C. Mills [Mon, 23 Mar 2020 20:38:00 +0000 (20:38 +0000)]
Revert "crypto/rsa,crypto/ecdsa,crypto/ed25519: implement PublicKey.Equal"

This reverts CL 223754.

Reason for revert: new tests are failing on all longtest builders.

Change-Id: I2257d106c132f3a02c0af6b20061d4f9a8093c4f
Reviewed-on: https://go-review.googlesource.com/c/go/+/225077
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
Reviewed-by: Katie Hockman <katie@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/compile: disable addressingmodes pass for 386
Keith Randall [Mon, 23 Mar 2020 20:07:06 +0000 (13:07 -0700)]
cmd/compile: disable addressingmodes pass for 386

Update #37881

Change-Id: I1f9a3f57f6215a19c31765c257ee78715eab36b7
Reviewed-on: https://go-review.googlesource.com/c/go/+/225057
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/go: define a build tag for any GOEXPERIMENT which is enabled.
Dan Scales [Wed, 11 Mar 2020 19:08:03 +0000 (12:08 -0700)]
cmd/go: define a build tag for any GOEXPERIMENT which is enabled.

For each experiment that has been enabled in the toolchain, define a build tag
with the same name (but prefixed by "goexperiment.") which can be used for
compiling alternative files for the experiment. This allows changes for the
experiment, like extra struct fields in the runtime, without affecting the base
non-experiment code at all.

I use this capability in my CL for static lock ranking
(https://go-review.googlesource.com/c/go/+/207619), so that static lock ranking
can be fully enabled as a GOEXPERIMENT, but there is no overhead in the runtime
when the experiment is not enabled.

I added a test in cmd/go/testdata/scripts to make sure the build tags are being
defined properly. In order to implement the test, I needed to provide environment
variable GOEXPSTRING to the test scripts (with its value set from
objabi.Expstring(), so that it can determine the experiments baked into the
toolchain.

I filed https://github.com/golang/go/issues/37937 to make a builder with
GOEXPERIMENT set to 'staticlockranking'. This builder will ensure another variant
of GOEXPERIMENT is being tested regularly for this change, as well as checking
static lock ranking in the runtime.

Change-Id: Ieb4b86107238febd105558c1e639d30cfe57ab5c
Reviewed-on: https://go-review.googlesource.com/c/go/+/222925
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agostrings: narrow the search range of IndexByte in Index
Andy Pan [Mon, 23 Mar 2020 04:53:40 +0000 (04:53 +0000)]
strings: narrow the search range of IndexByte in Index

Same as https://golang.org/cl/224589.

Change-Id: I6a9e2ea60bf6e1888a95bad0331c92079a7eff99
GitHub-Last-Rev: 81c13c0f5b2ee6ae7842a7b73799b7821f78be59
GitHub-Pull-Request: golang/go#38016
Reviewed-on: https://go-review.googlesource.com/c/go/+/224593
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/compile: disable mem+op operations on 386
Keith Randall [Sun, 22 Mar 2020 20:27:02 +0000 (13:27 -0700)]
cmd/compile: disable mem+op operations on 386

Rolling back portions of CL 222782 to see if that helps
issue #37881 any.

Update #37881

Change-Id: I9cc3ff8c469fa5e4b22daec715d04148033f46f7
Reviewed-on: https://go-review.googlesource.com/c/go/+/224837
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocrypto/rsa,crypto/ecdsa,crypto/ed25519: implement PublicKey.Equal
Filippo Valsorda [Wed, 18 Mar 2020 00:34:51 +0000 (20:34 -0400)]
crypto/rsa,crypto/ecdsa,crypto/ed25519: implement PublicKey.Equal

This makes all modern public keys in the standard library implement a
common interface (below) that can be used by applications for better
type safety and allows for checking that public (and private keys via
Public()) are equivalent.

interface {
    Equal(crypto.PublicKey) bool
}

Equality for ECDSA keys is complicated, we take a strict interpretation
that works for all secure applications (the ones not using the
unfortunate non-constant time CurveParams implementation) and fails
closed otherwise.

Tests in separate files to make them x_tests and avoid an import loop
with crypto/x509.

Fixes #21704

Change-Id: Id5379c96384a11c5afde0614955360e7470bb1c4
Reviewed-on: https://go-review.googlesource.com/c/go/+/223754
Reviewed-by: Katie Hockman <katie@golang.org>
5 years agocmd/compile: fold LEAQ with constant scale into LEA
Giovanni Bajo [Mon, 23 Mar 2020 08:21:44 +0000 (09:21 +0100)]
cmd/compile: fold LEAQ with constant scale into LEA

Discovered this after rebasing CL196679 (use poset bounds in prove).
Some tests fail with that CL applied:

codegen/smallintiface.go:11: linux/amd64/: opcode not found: "^LEAQ\\truntime.staticuint64s\\+8\\(SB\\)"
codegen/smallintiface.go:16: linux/amd64/: opcode not found: "^LEAQ\\truntime.staticuint64s\\+2024\\(SB\\)"
codegen/smallintiface.go:21: linux/amd64/: opcode not found: "^LEAQ\\truntime.staticuint64s\\+24\\(SB\\)"

The only difference in prove SSA dumps is that a single Lsh64x64
op with constant shift (<< 3) is marked as bounded. This triggers
a different rule matching sequence in lower, which manages to generate
worse code for the above testcases.

This CL fixes the above test after CL196679 is applied. Right now,
these rules never trigger (this CL passes toolstash -cmp), so I can't
write a test.

Change-Id: I353f1c79c1875cac1da82cd8afa1e05e42684f1c
Reviewed-on: https://go-review.googlesource.com/c/go/+/224877
Run-TryBot: Giovanni Bajo <rasky@develer.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocrypto/x509: add RevocationList and CreateRevocationList
Roland Shoemaker [Fri, 20 Mar 2020 20:28:43 +0000 (20:28 +0000)]
crypto/x509: add RevocationList and CreateRevocationList

The existing Certificate.CreateCRL method generates non-conformant CRLs and
as such cannot be used for implementations that require standards
compliance. This change implements a new top level method, CreateCRL, which
generates compliant CRLs, and offers an extensible API if any
extensions/fields need to be supported in the future.

Here is an example Issuer/CRL generated using this change:
-----BEGIN CERTIFICATE-----
MIIBNjCB3aADAgECAgEWMAoGCCqGSM49BAMCMBIxEDAOBgNVBAMTB3Rlc3Rpbmcw
IhgPMDAwMTAxMDEwMDAwMDBaGA8wMDAxMDEwMTAwMDAwMFowEjEQMA4GA1UEAxMH
dGVzdGluZzBZMBMGByqGSM49AgEGCCqGSM49AwEHA0IABLHrudbSM36sn1VBrmm/
OfQTyEsI4tIUV1VmneOKHL9ENBGCiec4GhQm2SGnDT/sZy2bB3c3yozh/roS6cZJ
UZqjIDAeMA4GA1UdDwEB/wQEAwIBAjAMBgNVHQ4EBQQDAQIDMAoGCCqGSM49BAMC
A0gAMEUCIQCoAYN6CGZPgd5Sw5a1rd5VexciT5MCxTfXj+ZfJNfoiAIgQVCTB8AE
Nm2xset7+HOgtQYlKNw/rGd8cFcv5Y9aUzo=
-----END CERTIFICATE-----
-----BEGIN X509 CRL-----
MIHWMH0CAQEwCgYIKoZIzj0EAwIwEjEQMA4GA1UEAxMHdGVzdGluZxgPMDAwMTAx
MDIwMDAwMDBaGA8wMDAxMDEwMzAwMDAwMFowFjAUAgECGA8wMDAxMDEwMTAxMDAw
MFqgHjAcMA4GA1UdIwQHMAWAAwECAzAKBgNVHRQEAwIBBTAKBggqhkjOPQQDAgNJ
ADBGAiEAjqfj/IG4ys5WkjrbTNpDbr+saHGO/NujLJotlLL9KzgCIQDm8VZPzj0f
NYEQgAW4nsiUzlvEUCoHMw0141VCZXv67A==
-----END X509 CRL-----

Fixes #35428

Change-Id: Id96b6f47698d0bed39d586b46bd12374ee6ff88f
GitHub-Last-Rev: c83a6017164e71df3989fe57322b3b4869a09f37
GitHub-Pull-Request: golang/go#36945
Reviewed-on: https://go-review.googlesource.com/c/go/+/217298
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoruntime: use OnesCount64 to count allocated objects in a span
Michael Anthony Knyszek [Mon, 27 Jan 2020 20:41:38 +0000 (20:41 +0000)]
runtime: use OnesCount64 to count allocated objects in a span

This change modifies the implementation of (*mspan).countAlloc by
using OnesCount64 (which on many systems is intrinsified). It does so by
using an unsafe pointer cast, but in this case we don't care about
endianness because we're just counting bits set.

This change means we no longer need the popcnt table which was redundant
in the runtime anyway. We can also simplify the logic here significantly
by observing that mark bits allocations are always 8-byte aligned, so we
don't need to handle any edge-cases due to the fact that OnesCount64
operates on 64 bits at a time: all irrelevant bits will be zero.

Overall, this implementation is significantly faster than the old one on
amd64, and should be similarly faster (or better!) on other systems
which support the intrinsic. On systems which do not, it should be
roughly the same performance because OnesCount64 is implemented using a
table in the general case.

Results on linux/amd64:

name                         old time/op  new time/op  delta
MSpanCountAlloc/bits=64-4    16.8ns ± 0%  12.7ns ± 0%  -24.40%  (p=0.000 n=5+4)
MSpanCountAlloc/bits=128-4   23.5ns ± 0%  12.8ns ± 0%  -45.70%  (p=0.000 n=4+5)
MSpanCountAlloc/bits=256-4   43.5ns ± 0%  12.8ns ± 0%  -70.67%  (p=0.000 n=4+5)
MSpanCountAlloc/bits=512-4   59.5ns ± 0%  15.4ns ± 0%  -74.12%  (p=0.008 n=5+5)
MSpanCountAlloc/bits=1024-4   116ns ± 1%    23ns ± 0%  -79.84%  (p=0.000 n=5+4)

Change-Id: Id4c994be22224653af5333683a69b0937130ed04
Reviewed-on: https://go-review.googlesource.com/c/go/+/216558
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agoruntime: add countAlloc benchmark
Michael Anthony Knyszek [Wed, 18 Mar 2020 18:46:04 +0000 (18:46 +0000)]
runtime: add countAlloc benchmark

This change adds a small microbenchmark for (*mspan).countAlloc, which
we're about to replace. Admittedly this isn't a critical piece of code,
but the benchmark was useful in understanding the performance change.

Change-Id: Iea93c00f571ee95534a42f2ef2ab026b382242b3
Reviewed-on: https://go-review.googlesource.com/c/go/+/224438
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/compile: panic if trying to alias an intrinsic with no definitions
Michael Anthony Knyszek [Fri, 20 Mar 2020 19:08:13 +0000 (19:08 +0000)]
cmd/compile: panic if trying to alias an intrinsic with no definitions

Currently if we try to alias an intrinsic which hasn't been defined for
any architecture (such as by accidentally creating the alias before the
intrinsic is created with addF), then we'll just silently not apply any
intrinsics to those aliases.

Catch this particular case by panicking in alias if we try to apply the
alias and it did nothing.

Change-Id: I98e75fc3f7206b08fc9267cedb8db3e109ec4f5d
Reviewed-on: https://go-review.googlesource.com/c/go/+/224637
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 agocmd/compile: declare runtime bit func aliases after math/bits intrinsics
Michael Anthony Knyszek [Fri, 20 Mar 2020 15:26:46 +0000 (15:26 +0000)]
cmd/compile: declare runtime bit func aliases after math/bits intrinsics

Currently runtime/internal/sys bit-manipulation functions are aliased to
math/bits functions, which are intrinsified. Unfortunately these aliases
are declared before the intrinsified versions are generated, resulting
in the generic version of the code being copied over.

This change moves the aliases for bit operations in runtime/internal/sys
after the addF calls to generate those intrinsics in SSA, so that the
intrinsified SSA representation of those functions actually get copied
over.

This should improve the overall performance of the runtime (especially
the page allocator) since these bit operations will actually be
intrinsified now.

Change-Id: I4377da13f9a7bb6aee608e50df0297148bf8f806
Reviewed-on: https://go-review.googlesource.com/c/go/+/224437
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/compile: skip generating args_stackmap for "pulled" funcs
Matthew Dempsky [Thu, 12 Mar 2020 19:14:02 +0000 (12:14 -0700)]
cmd/compile: skip generating args_stackmap for "pulled" funcs

In golang.org/cl/171464, we cleaned up generation of .stkobj linker
symbols, but we couldn't figure out why a similar cleanup to
.args_stackmap linker symbols caused problems.

The issue is that we only need/want to generate .args_stackmap for
functions that are implemented in assembly in the same package. When
"pulling" a function from another package via //go:linkname, we can
safely skip emitting .args_stackmap, because compiling that package
will have generated it, if necessary.

Fixes #31615.

Change-Id: If8680aa7dd5b4e8f268b6b032d746f1b8536c867
Reviewed-on: https://go-review.googlesource.com/c/go/+/223238
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: Keith Randall <khr@golang.org>
5 years agoruntime: correct the system-call name of kevent in comment
Andy Pan [Sun, 22 Mar 2020 06:22:41 +0000 (06:22 +0000)]
runtime: correct the system-call name of kevent  in comment

Change-Id: Ib1f4a6f7e36d28eff39f597df5c4703bf62654a4
GitHub-Last-Rev: 15ea1b9fa846737bd5d30b7b98d8933f9992e5c7
GitHub-Pull-Request: golang/go#37994
Reviewed-on: https://go-review.googlesource.com/c/go/+/224590
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agobytes: narrow the search of IndexByte in Index
Andy Pan [Mon, 23 Mar 2020 02:51:26 +0000 (02:51 +0000)]
bytes: narrow the search of IndexByte in Index

Change-Id: I5a47b18b64e7f781dcc77440b06de36966e3d01d
GitHub-Last-Rev: 8576f1931d6e87a0823632ee35fcbc0d4eaaeaaf
GitHub-Pull-Request: golang/go#37993
Reviewed-on: https://go-review.googlesource.com/c/go/+/224589
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agogo/ast: fix inflection in comments to match plurality
Allen Li [Mon, 23 Mar 2020 03:23:27 +0000 (03:23 +0000)]
go/ast: fix inflection in comments to match plurality

Change-Id: I3a725c5691a1090952acdc2ae0bed96aaa8d7067
GitHub-Last-Rev: 339a0dda36aeb1c0b38e531b4292048ebfae05ad
GitHub-Pull-Request: golang/go#37256
Reviewed-on: https://go-review.googlesource.com/c/go/+/219737
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agoruntime: speed up receive on empty closed channel
Ben Schwartz [Mon, 10 Jun 2019 16:29:23 +0000 (12:29 -0400)]
runtime: speed up receive on empty closed channel

Currently, nonblocking receive on an open channel is about
700 times faster than nonblocking receive on a closed channel.
This change makes closed channels equally fast.

Fixes #32529.  Includes a correction based on #36714.

relevant benchstat output:
name                       old time/op    new time/op    delta
MakeChan/Byte-40            140ns ± 4%     137ns ± 7%   -2.38%  (p=0.023 n=17+19)
MakeChan/Int-40             174ns ± 5%     173ns ± 6%     ~     (p=0.437 n=18+19)
MakeChan/Ptr-40             315ns ±15%     301ns ±15%     ~     (p=0.051 n=20+20)
MakeChan/Struct/0-40        123ns ± 8%      99ns ±11%  -19.18%  (p=0.000 n=20+17)
MakeChan/Struct/32-40       297ns ± 8%     241ns ±18%  -19.13%  (p=0.000 n=20+20)
MakeChan/Struct/40-40       344ns ± 5%     273ns ±23%  -20.49%  (p=0.000 n=20+20)
ChanNonblocking-40         0.32ns ± 2%    0.32ns ± 2%   -1.25%  (p=0.000 n=19+18)
SelectUncontended-40       5.72ns ± 1%    5.71ns ± 2%     ~     (p=0.326 n=19+19)
SelectSyncContended-40     10.9µs ±10%    10.6µs ± 3%   -2.77%  (p=0.009 n=20+16)
SelectAsyncContended-40    1.00µs ± 0%    1.10µs ± 0%  +10.75%  (p=0.000 n=18+19)
SelectNonblock-40          1.22ns ± 2%    1.21ns ± 4%     ~     (p=0.141 n=18+19)
ChanUncontended-40          240ns ± 4%     233ns ± 4%   -2.82%  (p=0.000 n=20+20)
ChanContended-40           86.7µs ± 0%    82.7µs ± 0%   -4.64%  (p=0.000 n=20+19)
ChanSync-40                 294ns ± 7%     284ns ± 9%   -3.44%  (p=0.006 n=20+20)
ChanSyncWork-40            38.4µs ±19%    34.0µs ± 4%  -11.33%  (p=0.000 n=20+18)
ChanProdCons0-40           1.50µs ± 1%    1.63µs ± 0%   +8.53%  (p=0.000 n=19+19)
ChanProdCons10-40          1.17µs ± 0%    1.18µs ± 1%   +0.44%  (p=0.000 n=19+20)
ChanProdCons100-40          985ns ± 0%     959ns ± 1%   -2.64%  (p=0.000 n=20+20)
ChanProdConsWork0-40       1.50µs ± 0%    1.60µs ± 2%   +6.54%  (p=0.000 n=18+20)
ChanProdConsWork10-40      1.26µs ± 0%    1.26µs ± 2%   +0.40%  (p=0.015 n=20+19)
ChanProdConsWork100-40     1.27µs ± 0%    1.22µs ± 0%   -4.15%  (p=0.000 n=20+19)
SelectProdCons-40          1.50µs ± 1%    1.53µs ± 1%   +1.95%  (p=0.000 n=20+20)
ChanCreation-40            82.1ns ± 5%    81.6ns ± 7%     ~     (p=0.483 n=19+19)
ChanSem-40                  877ns ± 0%     719ns ± 0%  -17.98%  (p=0.000 n=18+19)
ChanPopular-40             1.75ms ± 2%    1.78ms ± 3%   +1.76%  (p=0.002 n=20+19)
ChanClosed-40               215ns ± 1%       0ns ± 6%  -99.82%  (p=0.000 n=20+18)

Previously committed in CL 181543 and reverted in CL 216158.

Change-Id: Ib767b08d724cfad03598d77271dbc1087485feb8
Reviewed-on: https://go-review.googlesource.com/c/go/+/216818
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/compile: remove unused isIntOrdering method
kadern0 [Sun, 22 Mar 2020 06:33:52 +0000 (17:33 +1100)]
cmd/compile: remove unused isIntOrdering method

Fixes #37989

Change-Id: Iabf86529fde3be9a98222b7e8d09ff8301cf8830
Reviewed-on: https://go-review.googlesource.com/c/go/+/224777
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 agobuild: add default GOROOT_BOOTSTRAP in Windows
Giovanni Bajo [Sun, 25 Feb 2018 10:32:59 +0000 (11:32 +0100)]
build: add default GOROOT_BOOTSTRAP in Windows

CL 57753 added support to make.bash and make.rc to
default GOROOT_BOOTSTRAP to 'go env GOROOT'. This
patch does the same in make.bat for Windows.

Updates #18545
Fixes #28641

Change-Id: I9152cc5080ed219b4de5bad0bd12d7725422ee1a
Reviewed-on: https://go-review.googlesource.com/c/go/+/96455
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
5 years agobuild: force all Windows batch files to CRLF
Giovanni Bajo [Sun, 22 Mar 2020 08:28:21 +0000 (09:28 +0100)]
build: force all Windows batch files to CRLF

Batch files should use CRLF endings. LF endings mostly
work but in some situations they cause random errors like
goto commands failing for mysterious reasons. See
golang.org/issue/37791 for more information.

Next CL triggered one of such bug (a label was not being
recognized), so prepare for it by converting to CRLF.

This CL also touches all existing batch files to force git
to update the line endings (unfortunately, changing
.gitattributes only has effect next time the file is checked
out or modified).

Fixes #37791
Updates #9281

Change-Id: I6f9a114351cb7ac9881914400aa210c930eb8cc1
Reviewed-on: https://go-review.googlesource.com/c/go/+/96495
Run-TryBot: Giovanni Bajo <rasky@develer.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
5 years agocmd/go: fix function name in comment
Ian Lance Taylor [Sat, 21 Mar 2020 23:57:26 +0000 (16:57 -0700)]
cmd/go: fix function name in comment

Fixes #37991

Change-Id: Ica58223f8564ec5d501d5b90b4258ffb78c42af1
Reviewed-on: https://go-review.googlesource.com/c/go/+/224587
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agonet/http: remove period at end of error message
d-tsuji [Sat, 21 Mar 2020 11:40:01 +0000 (20:40 +0900)]
net/http: remove period at end of error message

Change-Id: I4ff5411543c200344babb754fc089e10e29e0fe4
Reviewed-on: https://go-review.googlesource.com/c/go/+/224697
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoruntime: ignore error returned by PowerRegisterSuspendResumeNotification
Alex Brainman [Sun, 16 Feb 2020 01:01:02 +0000 (12:01 +1100)]
runtime: ignore error returned by PowerRegisterSuspendResumeNotification

It appears that PowerRegisterSuspendResumeNotification is not supported
when running inside Docker - see issues #35447, #36557 and #37149.

Our current code relies on error number to determine Docker environment.
But we already saw PowerRegisterSuspendResumeNotification return
ERROR_FILE_NOT_FOUND, ERROR_INVALID_PARAMETERS and ERROR_ACCESS_DENIED
(see issues above). So this approach is not sustainable.

Just ignore PowerRegisterSuspendResumeNotification returned error.

Fixes #37149

Change-Id: I2beba9d45cdb8c1efac5e974e747827a6261915a
Reviewed-on: https://go-review.googlesource.com/c/go/+/219657
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Jason A. Donenfeld <Jason@zx2c4.com>
5 years agocmd/compile: indexed loads/stores can't be faultOnNilArg0
Keith Randall [Fri, 20 Mar 2020 16:59:53 +0000 (09:59 -0700)]
cmd/compile: indexed loads/stores can't be faultOnNilArg0

Because of the index, these ops can't guarantee faulting if arg0 is nil.

Clean up the PPC64 index ops - they can't take a sym or an offset.

Noticed while debugging #37881. I don't think it is the cause, but I guess
there is a chance.
Update #37881

Change-Id: Ic22925250bf7b1ba64e3cea1a65638bc4bab390c
Reviewed-on: https://go-review.googlesource.com/c/go/+/224457
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoruntime: handle empty stack in expandFinalInlineFrame
Keith Randall [Fri, 20 Mar 2020 19:25:57 +0000 (12:25 -0700)]
runtime: handle empty stack in expandFinalInlineFrame

Fixes #37967

Change-Id: I6fc22bdd65f0263d5672731b73d09249201ab0aa
Reviewed-on: https://go-review.googlesource.com/c/go/+/224458
Reviewed-by: Michael Pratt <mpratt@google.com>
5 years agomath/big: document that Sqrt doesn't set Accuracy
Alberto Donizetti [Fri, 20 Mar 2020 18:13:07 +0000 (19:13 +0100)]
math/big: document that Sqrt doesn't set Accuracy

Document that the Float.Sqrt method does not set the receiver's
Accuracy field.

Updates #37915

Change-Id: Ief1dcac07eacc0ef02f86bfac9044501477bca1c
Reviewed-on: https://go-review.googlesource.com/c/go/+/224497
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agoruntime: use divRoundUp
Austin Clements [Fri, 20 Mar 2020 15:37:54 +0000 (11:37 -0400)]
runtime: use divRoundUp

There are a handful of places where the runtime wants to round up the
result of a division. We just introduced a helper to do this. This CL
replaces all of the hand-coded round-ups (that I could find) with this
helper.

Change-Id: I465d152157ff0f3cad40c0aa57491e4f2de510ad
Reviewed-on: https://go-review.googlesource.com/c/go/+/224385
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoruntime: fix rounding in materializeGCProg
Austin Clements [Wed, 26 Feb 2020 20:12:33 +0000 (15:12 -0500)]
runtime: fix rounding in materializeGCProg

materializeGCProg allocates a temporary buffer for unrolling a GC
program. Unfortunately, when computing the size of the buffer, it
rounds *down* the number of bytes needed to store bitmap before
rounding up the number of pages needed to store those bytes. The fact
that it rounds up to pages usually mitigates the rounding down, but
the type from #37470 exists right on the boundary where this doesn't
work:

type Sequencer struct {
htable [1 << 17]uint32
buf    []byte
}

On 64-bit, this GC bitmap is exactly 8 KiB of zeros, followed by three
one bits. Hence, this needs 8193 bytes of storage, but the current
math in materializeGCProg rounds *down* the three one bits to 8192
bytes. Since this is exactly pageSize, the next step of rounding up to
the page size doesn't mitigate this error, and materializeGCProg
allocates a buffer that is one byte too small. runGCProg then writes
one byte past the end of this buffer, causing either a segfault (if
you're lucky!) or memory corruption.

Fixes #37470.

Change-Id: Iad24c463c501cd9b1dc1924bc2ad007991a094a0
Reviewed-on: https://go-review.googlesource.com/c/go/+/221197
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agotest: fix -test.v trace output for cgo/testshared
Than McIntosh [Fri, 20 Mar 2020 13:03:39 +0000 (09:03 -0400)]
test: fix -test.v trace output for cgo/testshared

Trace output showing how dummy GOROOT was being set up was incorrect
(sense of the "cp -r" trace messages was inverted). This patch fixes
the problem.

Change-Id: Ib0ee649e305bfa1bc0c49e0d5ba2ea31e0a4f67e
Reviewed-on: https://go-review.googlesource.com/c/go/+/224377
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agodoc/articles/wiki: use correct variable name in closures guide
HowJMay [Fri, 20 Mar 2020 08:40:38 +0000 (08:40 +0000)]
doc/articles/wiki: use correct variable name in closures guide

Fixes non-existent variable TitleValidator to be validPath in
the closures, functions literal section.

Fixes #36779

Change-Id: I59762c358c3e00d1cc03d9d1e2aace03f145321d
GitHub-Last-Rev: a5e9b17a3707fc48c6d87bab3a4968f2ef63ab65
GitHub-Pull-Request: golang/go#36783
Reviewed-on: https://go-review.googlesource.com/c/go/+/216479
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agoruntime/race: test that close synchronizes with read
Ian Lance Taylor [Thu, 23 Jan 2020 23:02:43 +0000 (15:02 -0800)]
runtime/race: test that close synchronizes with read

Add a test to ensure that the race detector sees that closing a
channel synchronizes with a read from that channel.
This test case failed when CL 181543 was in the tree.
CL 181543 was reverted in CL 216158; this adds a test to make
sure that we don't re-introduce the problem at a later date.

For #32529
For #36714

Change-Id: I5a40f744c67c3f8191d6ad822710c180880a7375
Reviewed-on: https://go-review.googlesource.com/c/go/+/216099
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agointernal/poll: update Windows Server Performance URL in comment
Ian Lance Taylor [Fri, 20 Mar 2020 01:17:21 +0000 (18:17 -0700)]
internal/poll: update Windows Server Performance URL in comment

Change-Id: Ida8048720611ecf50d7e7626a42715921bbc753b
Reviewed-on: https://go-review.googlesource.com/c/go/+/224237
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agonet: merge common Unix/Windows methods
Ian Lance Taylor [Thu, 19 Mar 2020 21:29:23 +0000 (14:29 -0700)]
net: merge common Unix/Windows methods

When we added the internal/poll package, the Unix and Windows implementations
of several netFD methods became exactly the same, except for using a
different name for the string passed to wrapSyscallError.

One case is not an exact duplicate: we slightly tweak the implementation
of (*netFD).shutdown on Windows to wrap the error.

Change-Id: I3d87a317d5468ff8f1958d86f6189ea1ba697e9a
Reviewed-on: https://go-review.googlesource.com/c/go/+/224140
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agonet/http: update bundled x/net/http2
Emmanuel T Odeke [Thu, 19 Mar 2020 23:51:13 +0000 (16:51 -0700)]
net/http: update bundled x/net/http2

Updates bundled http2 to x/net git rev 63522dbf7

    http2: reduce allocations of (*clientConnReadLoop).handleReponse
    https://golang.org/cl/223783 (#37853)

    http2: remove unused errors
    https://golang.org/cl/220458

    http2: remove unused stream struct fields
    https://golang.org/cl/219857

    http2: fix typo in comment
    https://golang.org/cl/214602

    http2: workaround TCPConn CloseWrite not being supported on Plan 9
    https://golang.org/cl/209417 (#17906, #35904)

Change-Id: I0e48f32247938c3858170bf419624367d4faef4b
Reviewed-on: https://go-review.googlesource.com/c/go/+/224217
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agointernal/poll: merge a couple of common definitions
Ian Lance Taylor [Thu, 19 Mar 2020 20:52:14 +0000 (13:52 -0700)]
internal/poll: merge a couple of common definitions

(*FD).Shutdown and (*FD).RawControl were exactly identical in fd_unix.go
and fd_windows.go, so merge them into fd_posix.go.

Change-Id: Ie751edb06d293e08f1493682d6bc5f22e8f3e628
Reviewed-on: https://go-review.googlesource.com/c/go/+/224137
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agoos: merge common Unix/Windows methods
Ian Lance Taylor [Thu, 19 Mar 2020 20:58:48 +0000 (13:58 -0700)]
os: merge common Unix/Windows methods

Several method implementations were identical in file_unix.go and
file_windows.go. Merge them into file_posix.go.

Change-Id: I8bcfad468829530f81f52fe426b3a8c042e7bbd6
Reviewed-on: https://go-review.googlesource.com/c/go/+/224138
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agocrypto/x509: bump minimum macOS version to 10.11
Tobias Klauser [Thu, 19 Mar 2020 21:33:40 +0000 (22:33 +0100)]
crypto/x509: bump minimum macOS version to 10.11

The minimum macOS supported version is 10.11 as of Go 1.14, see #23011.
Thus, bump macosx-version-min to 10.11

While at it, drop __MAC_OS_X_VERSION_MAX_ALLOWED as suggested by
Filippo:

  In general, I can see why we'd want to tell the libraries which
  minimum version we target so they drop compatibility with older
  versions. No idea why we'd specify a max version, unless it's to make
  sure we don't use APIs added after that version, but then it would
  have to be 1011 not 1015.

  Let's try dropping that define and see if anything blows up? ¯\_(ツ)_/¯

Change-Id: I6b76623a9404724ccda40311ff95b3475ae8a60c
Reviewed-on: https://go-review.googlesource.com/c/go/+/214059
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
5 years agocmd/go: update 'go help modules' for automatic vendoring
Jay Conrod [Wed, 18 Mar 2020 18:59:44 +0000 (14:59 -0400)]
cmd/go: update 'go help modules' for automatic vendoring

* Mention vendor/modules.txt verification with -mod=vendor.
* Update "Modules and vendoring" section.

Fixes #37930

Change-Id: Ia3ee72457daabaa2fc15b7ea7427eb324c088b8f
Reviewed-on: https://go-review.googlesource.com/c/go/+/223926
Run-TryBot: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocrypto/tls: update the MITM reference to "machine-in-the-middle"
Chris Le Roy [Thu, 19 Mar 2020 21:31:23 +0000 (21:31 +0000)]
crypto/tls: update the MITM reference to "machine-in-the-middle"

Changing "man-in-the-middle" references to "machine-in-the-middle",
it's a more inclusive term and still aligns with the MITM acronym.

Change-Id: I81f954cff3d252433443f159ff9edaf59a28ab9d
GitHub-Last-Rev: 3e8f91424a207233b537984747ae90cbc1f03755
GitHub-Pull-Request: golang/go#37918
Reviewed-on: https://go-review.googlesource.com/c/go/+/223897
Reviewed-by: Filippo Valsorda <filippo@golang.org>
5 years agotesting: fix data race between parallel subtests
Changkun Ou [Fri, 28 Feb 2020 20:53:38 +0000 (21:53 +0100)]
testing: fix data race between parallel subtests

This CL fixes a race condition if there are two subtests, and
one finishing but the other is panicking.

Fixes #37551

Change-Id: Ic33963eb338aec228964b95f7c34a0d207b91e00
Reviewed-on: https://go-review.googlesource.com/c/go/+/221322
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 agosync/atomic: remove panic64
Daniel Martí [Thu, 19 Mar 2020 13:55:42 +0000 (13:55 +0000)]
sync/atomic: remove panic64

The func has been unused since https://golang.org/cl/93637 in 2018.

Change-Id: I1cab6f265aa5058ac080fd7c7cbf0fe85370f073
Reviewed-on: https://go-review.googlesource.com/c/go/+/224077
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matt Layher <mdlayher@gmail.com>
Reviewed-by: Austin Clements <austin@google.com>
5 years agoruntime: improve memmove for short moves on ppc64
Lynn Boger [Thu, 12 Mar 2020 14:26:01 +0000 (10:26 -0400)]
runtime: improve memmove for short moves on ppc64

This improves the performance of memmove for almost all moves <= 16 bytes
for the ppc64 assembler, improving linux/ppc64le, linux/ppc64, aix/ppc64.

Only the forward moves were changed, the backward moves were left as is.
Additional macro defines were added to improve the readability of the asm.

Results from power8:
name                      old time/op    new time/op    delta
Memmove/0                   5.70ns ± 0%    5.69ns ± 0%    -0.18%  (p=0.029 n=4+4)
Memmove/1                   5.54ns ± 0%    5.39ns ± 0%    -2.71%  (p=0.029 n=4+4)
Memmove/2                   6.31ns ± 0%    5.55ns ± 0%   -12.08%  (p=0.029 n=4+4)
Memmove/3                   7.41ns ± 0%    5.54ns ± 0%   -25.24%  (p=0.029 n=4+4)
Memmove/4                   8.41ns ± 0%    5.56ns ± 0%   -33.87%  (p=0.029 n=4+4)
Memmove/5                   10.1ns ± 5%     5.5ns ± 0%   -45.30%  (p=0.029 n=4+4)
Memmove/6                   10.3ns ± 0%     5.6ns ± 0%   -45.92%  (p=0.029 n=4+4)
Memmove/7                   11.4ns ± 0%     5.7ns ± 0%   -50.33%  (p=0.029 n=4+4)
Memmove/8                   5.66ns ± 0%    5.54ns ± 0%    -2.12%  (p=0.029 n=4+4)
Memmove/9                   5.66ns ± 0%    6.47ns ± 0%   +14.31%  (p=0.029 n=4+4)
Memmove/10                  6.67ns ± 0%    6.22ns ± 0%    -6.82%  (p=0.029 n=4+4)
Memmove/11                  7.83ns ± 0%    6.45ns ± 0%   -17.60%  (p=0.029 n=4+4)
Memmove/12                  8.91ns ± 0%    6.25ns ± 0%   -29.85%  (p=0.029 n=4+4)
Memmove/13                  9.81ns ± 0%    6.48ns ± 0%   -33.94%  (p=0.029 n=4+4)
Memmove/14                  10.7ns ± 1%     6.4ns ± 0%   -40.00%  (p=0.029 n=4+4)
Memmove/15                  11.8ns ± 0%     6.7ns ± 0%   -42.84%  (p=0.029 n=4+4)
Memmove/16                  5.63ns ± 0%    5.56ns ± 0%    -1.20%  (p=0.029 n=4+4)

Change-Id: I2de434f543c5a017395e0850fb9b9f7219583bbb
Reviewed-on: https://go-review.googlesource.com/c/go/+/223317
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com>
5 years agodatabase/sql: add method Err on sql.Row
Tim Möhlmann [Fri, 10 Jan 2020 10:06:53 +0000 (12:06 +0200)]
database/sql: add method Err on sql.Row

The Row.Err method is intended to assist wrapping sql.DB.
Because sql.Row is a struct with private fields,
a wrapper in an existing code base cannot easily provide users
with a different implementation without large rewrites.
Adding this method allows query level errors to be handled
centrally.

Fixes #35804

Change-Id: I94e6329de89a7ee1284ce9ef76af4363d2d081f9
Reviewed-on: https://go-review.googlesource.com/c/go/+/214317
Reviewed-by: Daniel Theophanes <kardianos@gmail.com>
Run-TryBot: Daniel Theophanes <kardianos@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoruntime: minor updates to netpoll comments
Ian Lance Taylor [Wed, 18 Mar 2020 23:32:31 +0000 (16:32 -0700)]
runtime: minor updates to netpoll comments

In Go 1.4 we renamed READY to pdReady and WAIT to pdWait as part of
rewriting netpoll from C to Go. Finish updating the comments to use
the new names.

Change-Id: I6cefc698b46c58211fd6be1489bdd70419454962
Reviewed-on: https://go-review.googlesource.com/c/go/+/223998
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agodatabase/sql: fix variable name in example
Harmen [Wed, 18 Mar 2020 17:52:57 +0000 (17:52 +0000)]
database/sql: fix variable name in example

It's a very minor error, but it's a bad copy/paste example.

Change-Id: Ia6a723c31f2205c933857ce2cf715bddf773ebb6
GitHub-Last-Rev: 7f14b1a5c1ea3775aa2009184b0b1790e03225eb
GitHub-Pull-Request: golang/go#37932
Reviewed-on: https://go-review.googlesource.com/c/go/+/223960
Reviewed-by: Daniel Theophanes <kardianos@gmail.com>
5 years agocontext: prevent creation of invalid contexts
Kyle Nusbaum [Wed, 18 Mar 2020 19:17:50 +0000 (19:17 +0000)]
context: prevent creation of invalid contexts

This commit makes it impossible to create derived contexts with nil parents.
Previously it was possible to create derived contexts with nil parents, and
invalid contexts could propogate through the program. Eventually this can
cause a panic downstream, which is difficult to trace back to the source
of the error.

Although `WithCancel` and `WithDeadline` already panic if `parent` is `nil`, this adds explicit checks to give a useful message in the panic.

Fixes #37908

Change-Id: I70fd01f6539c1b0da0e775fc5457e32e7075e52c
GitHub-Last-Rev: 1b7dadd7db9ba42952644ad5e9a49591d6a5191f
GitHub-Pull-Request: golang/go#37898
Reviewed-on: https://go-review.googlesource.com/c/go/+/223777
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: skip the cmd/cgo staleness check on darwin builders
Bryan C. Mills [Wed, 18 Mar 2020 18:17:43 +0000 (14:17 -0400)]
cmd/go: skip the cmd/cgo staleness check on darwin builders

Some of the darwin-amd64 builders are providing a stale environment.
Let's un-break them while we investigate.

Updates #37573
Updates #33598

Change-Id: I8b79778fe4d5aa916557c1ba89fa9c776d130b01
Reviewed-on: https://go-review.googlesource.com/c/go/+/223925
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Andrew Bonventre <andybons@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
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.

This change is being landed once more after being reverted in the Go
1.14 release cycle, since CL 215157 allows it to have a positive
effect on system performance.

For the benchmark BenchmarkReadMemStatsLatency in the runtime, which
measures ReadMemStats latencies while the GC is exercised, the tail of
these latencies reduced dramatically on an 8-core machine:

name                   old 50%tile-ns  new 50%tile-ns  delta
ReadMemStatsLatency-8      4.40M ±74%      0.12M ± 2%  -97.35%  (p=0.008 n=5+5)

name                   old 90%tile-ns  new 90%tile-ns  delta
ReadMemStatsLatency-8       102M ± 6%         0M ±14%  -99.79%  (p=0.008 n=5+5)

name                   old 99%tile-ns  new 99%tile-ns  delta
ReadMemStatsLatency-8       147M ±18%         4M ±57%  -97.43%  (p=0.008 n=5+5)

Fixes #19812.

Change-Id: If66c3c97d171524ae29f0e7af4bd33509d9fd0bb
Reviewed-on: https://go-review.googlesource.com/c/go/+/216557
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agodoc/go1.14: mention Windows change for Open permissions
Ian Lance Taylor [Wed, 18 Mar 2020 15:34:36 +0000 (08:34 -0700)]
doc/go1.14: mention Windows change for Open permissions

For #35033

Change-Id: Ie15353322d5cfe7320199103ad9543fb89a842ed
Reviewed-on: https://go-review.googlesource.com/c/go/+/223957
Reviewed-by: Brendan Jackman <jackmanb@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocmd/internal/moddeps: skip GOROOT/pkg when locating modules
Bryan C. Mills [Wed, 18 Mar 2020 17:50:14 +0000 (13:50 -0400)]
cmd/internal/moddeps: skip GOROOT/pkg when locating modules

Fixes #37929

Change-Id: I1435411de43aed98f40e0d98e450310b0fdf804c
Reviewed-on: https://go-review.googlesource.com/c/go/+/223924
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoruntime: add ReadMemStats latency benchmark
Michael Anthony Knyszek [Wed, 19 Feb 2020 20:54:25 +0000 (20:54 +0000)]
runtime: add ReadMemStats latency benchmark

This change adds a benchmark to the runtime which measures ReadMemStats
latencies. It generates allocations with lots of pointers to keep the GC
busy while hitting ReadMemStats and measuring the time it takes to
complete.

Updates #19812.

Change-Id: I7a76aaf497ba5324d3c7a7b3df32461b3e6c3ac8
Reviewed-on: https://go-review.googlesource.com/c/go/+/220177
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agoruntime: preempt dedicated background mark workers for STW
Michael Anthony Knyszek [Thu, 16 Jan 2020 21:39:42 +0000 (21:39 +0000)]
runtime: preempt dedicated background mark workers for STW

Currently, dedicated background mark workers are essentially always
non-preemptible.

This change makes it so that dedicated background mark workers park if
their preemption flag is set and someone is trying to STW, allowing them
to do so.

This change prepares us for allowing a STW to happen (and happen
promptly) during GC marking in a follow-up change.

Updates #19812.

Change-Id: I67fb6085bf0f0aebd18ca500172767818a1f15e3
Reviewed-on: https://go-review.googlesource.com/c/go/+/215157
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Austin Clements <austin@google.com>
5 years agocmd/go/internal/modfetch/zip_sum_test: remove dead versions
Jay Conrod [Mon, 10 Feb 2020 19:42:02 +0000 (14:42 -0500)]
cmd/go/internal/modfetch/zip_sum_test: remove dead versions

* gitlab.com/yumeko/MumbleEmu: the go-import tag now points to
  gitlab.com/yumeko/mumbleemu, but the module path hasn't changed
  in go.mod.
* github.com/openshift/api: tag v3.9.0 was deleted.
* github.com/AlexStocks/log4go: tag v1.0.5 was deleted.
* github.com/belogik/goes: repository is no longer available.
* llvm.org/llvm: server times out and disconnects after 30-40 mins.

Also, fix a typo in an error message.

With these versions removed, zip_sum_test passes.

Updates #35290

Change-Id: Id3bdb8675a5582f88a6ff4c12dd7d1abe31aa56f
Reviewed-on: https://go-review.googlesource.com/c/go/+/218917
Run-TryBot: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocmd/go: add a "don't care about success" operator to script_test
Bryan C. Mills [Tue, 17 Mar 2020 17:08:22 +0000 (13:08 -0400)]
cmd/go: add a "don't care about success" operator to script_test

Use that operator to make test_race_install_cgo agnostic to whether GOROOT/pkg is writable.

Updates #37573
Updates #30316

Change-Id: I018c63b3c369209345069f917bbb3a52179e2b58
Reviewed-on: https://go-review.googlesource.com/c/go/+/223746
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agoruntime: don't send preemption signal if there is a signal pending
Cherry Zhang [Tue, 17 Mar 2020 00:08:00 +0000 (20:08 -0400)]
runtime: don't send preemption signal if there is a signal pending

If multiple threads call preemptone to preempt the same M, it may
send many signals to the same M such that it hardly make
progress, causing live-lock problem. Only send a signal if there
isn't already one pending.

Fixes #37741.

Change-Id: Id94adb0b95acbd18b23abe637a8dcd81ab41b452
Reviewed-on: https://go-review.googlesource.com/c/go/+/223737
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/go: add a missing curly-brace in the 'stale' command format string
Bryan C. Mills [Tue, 17 Mar 2020 19:17:22 +0000 (15:17 -0400)]
cmd/go: add a missing curly-brace in the 'stale' command format string

The missing brace made the 'stale' command a no-op in the non-error case.

Fix the 'short' skip in install_cross_gobin (it was backward) and
update it to no longer check staleness of a not-necessarily-stale
target and to no longer expect to be able to install into GOROOT/pkg.
(This was missed in #30316 because that part of the test was
erroneously skipped in non-short mode.)

Change-Id: I6a276fec5fa5e5da3fe0daf0c2b5086116ed7c1a
Reviewed-on: https://go-review.googlesource.com/c/go/+/223747
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agodatabase/sql: process all Session Resets synchronously
Daniel Theophanes [Fri, 26 Apr 2019 21:45:46 +0000 (14:45 -0700)]
database/sql: process all Session Resets synchronously

Adds a new interface, driver.ConnectionValidator, to allow
drivers to signal they should not be used again,
separatly from the session resetter interface.
This is done now that the session reset is done
after the connection is put into the connection pool.

Previous behavior attempted to run Session Resets
in a background worker. This implementation had two
problems: untested performance gains for additional
complexity, and failures when the pool size
exceeded the connection reset channel buffer size.

Fixes #31480

Change-Id: I7d483b883c24a362c292471e87a88db5b204d1d0
Reviewed-on: https://go-review.googlesource.com/c/go/+/174122
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agotesting: use "exit code" in documentation consistently
Rodolfo Carvalho [Tue, 17 Mar 2020 10:41:45 +0000 (10:41 +0000)]
testing: use "exit code" in documentation consistently

The documentation for m.Run says it returns an "exit code" to pass to
os.Exit. The argument to os.Exit is named "code".

While "exit code", "exit status" and "exit status code" are all valid ways
to refer to the same concept, prefer to stick to one form for consistency
and to avoid confusing users.

Change-Id: If76ee3fab5cc99c79e05ac1a4e413790a9c93d60
GitHub-Last-Rev: 85a081d2f03c2cf9e8e519916986c59c86aebf57
GitHub-Pull-Request: golang/go#37899
Reviewed-on: https://go-review.googlesource.com/c/go/+/223778
Reviewed-by: Gabriel Aszalos <gabriel.aszalos@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Gabriel Aszalos <gabriel.aszalos@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoruntime: don't report a pointer alignment error for pointer-free base type
Keith Randall [Tue, 17 Mar 2020 20:27:11 +0000 (13:27 -0700)]
runtime: don't report a pointer alignment error for pointer-free base type

Fixes #37298

Change-Id: I8ba9c8b106e16cea7dd25473c7390b0f2ba9a1a5
Reviewed-on: https://go-review.googlesource.com/c/go/+/223781
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/go: replace TestCgoDependsOnSyscall with a simpler script test
Bryan C. Mills [Tue, 17 Mar 2020 14:50:40 +0000 (10:50 -0400)]
cmd/go: replace TestCgoDependsOnSyscall with a simpler script test

The existing test attempted to remove '_race' binaries from
GOROOT/pkg, which could not only fail if GOROOT is read-only, but also
interfere with other tests run in parallel.

Updates #30316
Updates #37573
Updates #17751

Change-Id: Id7e2286ab67f8333baf4d52244b7f4476aa93a46
Reviewed-on: https://go-review.googlesource.com/c/go/+/223745
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoruntime/pprof: export max rss when saving memory profiles.
Jeremy Faller [Tue, 25 Jun 2019 19:43:33 +0000 (15:43 -0400)]
runtime/pprof: export max rss when saving memory profiles.

NB: Adds syscall to deps on runtime/pprof.
Change-Id: I5dd14c2b25eb9c3c446832f5818de45fafd48a27
Reviewed-on: https://go-review.googlesource.com/c/go/+/183844
Run-TryBot: Jeremy Faller <jeremy@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years agocmd/compile: provide Load8/Store8 atomic intrinsics on riscv64
Joel Sing [Sun, 15 Mar 2020 15:38:43 +0000 (02:38 +1100)]
cmd/compile: provide Load8/Store8 atomic intrinsics on riscv64

Updates #36765

Change-Id: Ieeb6bbc54e4841a1348ad50e80342ec4bc675e07
Reviewed-on: https://go-review.googlesource.com/c/go/+/223557
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/internal/obj/riscv: add comments for Go registers
Joel Sing [Sun, 15 Mar 2020 16:31:32 +0000 (03:31 +1100)]
cmd/internal/obj/riscv: add comments for Go registers

Change-Id: Id9aa6ba268eee67f2dc74096d4ec3bc0a80aefe2
Reviewed-on: https://go-review.googlesource.com/c/go/+/223563
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/compile: use NOT pseudo-instruction for riscv64 Slicemask
Joel Sing [Sun, 15 Mar 2020 16:29:19 +0000 (03:29 +1100)]
cmd/compile: use NOT pseudo-instruction for riscv64 Slicemask

Change-Id: Idefb6669d54929065f57e3bd767c91451dc3a375
Reviewed-on: https://go-review.googlesource.com/c/go/+/223562
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agotesting: do not require os.Exit in TestMain
Changkun Ou [Sun, 16 Feb 2020 13:42:29 +0000 (14:42 +0100)]
testing: do not require os.Exit in TestMain

If TestMain reports a wrong exit code to os.Exit, the test will be
exited with exist code inconsist with test results.

This CL eliminates the requirement of calling os.Exit in TestMain.
Now, m.Run records the execution status of its test, the outer
main func will call os.Exit with that exit code if TestMain does
not call os.Exit.

If TestMain does not call m.Run, the outer main func remain calls
os.Exit(0) as before.

Fixes #34129

Change-Id: I9598023e03b0a6260f0217f34df41c231c7d6489
Reviewed-on: https://go-review.googlesource.com/c/go/+/219639
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoruntime: add goroutines returned by poller to local run queue
Ian Lance Taylor [Fri, 24 Jan 2020 04:38:20 +0000 (20:38 -0800)]
runtime: add goroutines returned by poller to local run queue

In Go 1.13, when the network poller found a list of ready goroutines,
they were added to the global run queue. The timer goroutine would
typically sleep in a futex with a timeout, and when the timeout
expired the timer goroutine would either be handed off to an idle P
or added to the global run queue. The effect was that on a busy system
with no idle P's goroutines waiting for timeouts and goroutines waiting
for the network would start at the same priority.

That changed on tip with the new timer code. Now timer functions are
invoked directly from a P, and it happens that the functions used
by time.Sleep and time.After and time.Ticker add the newly ready
goroutines to the local run queue. When a P looks for work it will
prefer goroutines on the local run queue; in fact it will only
occasionally look at the global run queue, and even when it does it
will just pull one goroutine off. So on a busy system with both active
timers and active network connections the system can noticeably prefer
to run goroutines waiting for timers rather than goroutines waiting
for the network.

This CL undoes that change by, when possible, adding goroutines
waiting for the network to the local run queue of the P that checked.
This doesn't affect network poller checks done by sysmon, but it
does affect network poller checks done as each P enters the scheduler.

This CL also makes injecting a list into either the local or global run
queue more efficient, using bulk operations rather than individual ones.

Change-Id: I85a66ad74e4fc3b458256fb7ab395d06f0d2ffac
Reviewed-on: https://go-review.googlesource.com/c/go/+/216198
Reviewed-by: Michael Knyszek <mknyszek@google.com>
5 years agotime: fix time.Before to reuse t.sec(), u.sec()
Paschalis Tsilias [Mon, 3 Feb 2020 08:47:41 +0000 (10:47 +0200)]
time: fix time.Before to reuse t.sec(), u.sec()

Fixes #36987

Change-Id: I91ea1a42f75302de5256a22d382ab7f1b307a498
Reviewed-on: https://go-review.googlesource.com/c/go/+/217360
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/compile: use NEG/NEGW pseudo-instructions on riscv64
Joel Sing [Mon, 2 Mar 2020 16:45:22 +0000 (03:45 +1100)]
cmd/compile: use NEG/NEGW pseudo-instructions on riscv64

Also rewrite subtraction of zero to NEG/NEGW.

Change-Id: I216e286d1860055f2a07fe2f772cd50f366ea097
Reviewed-on: https://go-review.googlesource.com/c/go/+/221691
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/compile: use NOT pseudo-instruction on riscv64
Joel Sing [Mon, 2 Mar 2020 16:43:02 +0000 (03:43 +1100)]
cmd/compile: use NOT pseudo-instruction on riscv64

Change-Id: I24a72c3fb8d72a47cfded4b523c5d7aa2d40419d
Reviewed-on: https://go-review.googlesource.com/c/go/+/221690
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/internal/obj/riscv: add NEG/NEGW pseudo-instructions
Joel Sing [Mon, 2 Mar 2020 16:41:43 +0000 (03:41 +1100)]
cmd/internal/obj/riscv: add NEG/NEGW pseudo-instructions

Provide NEG/NEGW pseudo-instructions, which translate to SUB/SUBW with the
zero register as a source.

Change-Id: I2c1ec1e75611c234c5ee8e39390dd188f8e42bae
Reviewed-on: https://go-review.googlesource.com/c/go/+/221689
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/internal/obj/riscv: add NOT pseudo-instruction
Joel Sing [Mon, 2 Mar 2020 16:40:37 +0000 (03:40 +1100)]
cmd/internal/obj/riscv: add NOT pseudo-instruction

Add a NOT pseudo-instruction that translates to XORI $-1.

Change-Id: I2be4cfe2939e988cd7f8d30260b704701d78475f
Reviewed-on: https://go-review.googlesource.com/c/go/+/221688
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/internal/obj/riscv: add FCLASSS/FCLASSD instructions
Joel Sing [Thu, 20 Feb 2020 16:03:57 +0000 (03:03 +1100)]
cmd/internal/obj/riscv: add FCLASSS/FCLASSD instructions

Add support for floating-point classification instructions.

Change-Id: I64463d477b3db0cca16ff7bced64f154011ef4cb
Reviewed-on: https://go-review.googlesource.com/c/go/+/220542
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/asm,cmd/internal/obj/riscv: add atomic memory operation instructions
Joel Sing [Thu, 20 Feb 2020 15:50:57 +0000 (02:50 +1100)]
cmd/asm,cmd/internal/obj/riscv: add atomic memory operation instructions

Use instructions in place of currently used defines.

Updates #36765

Change-Id: I00bb59e77b1aace549d7857cc9721ba2cb4ac6ca
Reviewed-on: https://go-review.googlesource.com/c/go/+/220541
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/asm,cmd/internal/obj/riscv: add LR/SC instructions
Joel Sing [Thu, 20 Feb 2020 15:30:09 +0000 (02:30 +1100)]
cmd/asm,cmd/internal/obj/riscv: add LR/SC instructions

Add support for Load-Reserved (LR) and Store-Conditional (SC) instructions.

Use instructions in place of currently used defines.

Updates #36765

Change-Id: I77e660639802293ece40cfde4865ac237e3308d6
Reviewed-on: https://go-review.googlesource.com/c/go/+/220540
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/internal/obj/riscv: add FENCE instruction
Joel Sing [Thu, 20 Feb 2020 15:28:37 +0000 (02:28 +1100)]
cmd/internal/obj/riscv: add FENCE instruction

Also remove #define's that were previously in use.

Updates #36765

Change-Id: I90b6a8629c78f549012f3f6c5f3b325336182712
Reviewed-on: https://go-review.googlesource.com/c/go/+/220539
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/internal/obj/riscv: rework instruction generation
Joel Sing [Wed, 18 Dec 2019 15:09:45 +0000 (02:09 +1100)]
cmd/internal/obj/riscv: rework instruction generation

Rework instruction generation so that multiple instructions are generated
from a single obj.Prog, rather than the current approach where obj.Progs
are rewritten. This allows the original obj.Prog to remain intact, before
being converted into an architecture specific instruction form.

This simplifies the code and removes a level of indirection that results
from trying to manipulate obj.Prog.To/obj.Prog.From into forms that match
the instruction encoding. Furthermore, the errors reported make more sense
since it matches up with the actual assembly that was parsed.

Note that the CALL/JMP/JALR type sequences have not yet been migrated to
this framework and will likely be converted at a later time.

Updates #27532

Change-Id: I9fd12562ed1db0a08cfdc32793897d2a1920ebaa
Reviewed-on: https://go-review.googlesource.com/c/go/+/211917
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/internal/obj/riscv: store relocation details in obj.Prog.RestArgs
Joel Sing [Wed, 5 Feb 2020 17:47:52 +0000 (04:47 +1100)]
cmd/internal/obj/riscv: store relocation details in obj.Prog.RestArgs

Store the relocation offset and symbol in obj.Prog.RestArgs, rather than
overloading obj.Prog.From and having to deal with invalid offsets
potentially existing when the instruction is encoded.

Change-Id: Iff0d678361677e78b41b887f6eba08cee94fccb3
Reviewed-on: https://go-review.googlesource.com/c/go/+/218197
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agonet/http: use atomicBool for inShutdown
Robert Kuska [Fri, 13 Mar 2020 18:21:19 +0000 (19:21 +0100)]
net/http: use atomicBool for inShutdown

This removes the TODO leftover by replacing the original int32 for
atomicBool that mimicks atomic operations for boolean.

Change-Id: I1b2cac0c9573c890c7315e9906ce6bfccee3d770
Reviewed-on: https://go-review.googlesource.com/c/go/+/223357
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 agomath/cmplx: disable TanHuge test on s390x
Brian Kessler [Sat, 14 Mar 2020 06:37:20 +0000 (00:37 -0600)]
math/cmplx: disable TanHuge test on s390x

s390x has inaccurate range reduction for the assembly routines
in math so these tests are diabled until these are corrected.

Updates #37854

Change-Id: I1e26acd6d09ae3e592a3dd90aec73a6844f5c6fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/223457
Run-TryBot: Brian Kessler <brian.m.kessler@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>
5 years agomath/cmplx: implement Payne-Hanek range reduction
Brian Kessler [Fri, 12 Apr 2019 04:54:18 +0000 (22:54 -0600)]
math/cmplx: implement Payne-Hanek range reduction

Tan has poles along the real axis. In order to accurately calculate
the value near these poles, a range reduction by Pi is performed and
the result calculated via a Taylor series.  The prior implementation
of range reduction used Cody-Waite range reduction in three parts.
This fails when x is too large to accurately calculate the partial
products in the summation accurately.  Above this threshold, Payne-Hanek
range reduction using a multiple precision value of 1/Pi is required.

Additionally, the threshold used in math/trig_reduce.go for Payne-Hanek
range reduction was not set conservatively enough. The prior threshold
ensured that catastrophic failure did not occur where the argument x
would not actually be reduced below Pi/4. However, errors in reduction
begin to occur at values much lower when z = ((x - y*PI4A) - y*PI4B) - y*PI4C
is not exact because y*PI4A cannot be exactly represented as a float64.
reduceThreshold is lowered to the proper value.

Fixes #31566

Change-Id: I0f39a4171a5be44f64305f18dc57f6c29f19dba7
Reviewed-on: https://go-review.googlesource.com/c/go/+/172838
Reviewed-by: Rob Pike <r@golang.org>
5 years agocontext: deflake time-sensitive tests
Bryan C. Mills [Wed, 11 Mar 2020 19:14:38 +0000 (15:14 -0400)]
context: deflake time-sensitive tests

Many of tests in this package assumed reasonable scheduling latency.
Unfortunately, scheduling latency on builders and CI systems is not
always reasonable.

Rather than expecting that a timeout is detected within a fixed short
interval, we can use (*testing.T).Deadline to portably scale the time
we're willing to wait to something appropriate to the builder.

Some of the tests also included arbitrary-duration sleeps, which are
no longer needed after CL 196521; we can remove those instead of
extending them.

Promptness of timeouts is also an important property, but testing that
property is better suited to benchmarks than to tests proper: unlike
tests, we generally expect benchmarks to be run in a quiet,
low-contention environment.

Fixes #13956

Change-Id: I0797e2267fb778c8ad94add56d797de9e2c885e5
Reviewed-on: https://go-review.googlesource.com/c/go/+/223019
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoruntime: don't crash on mlock failure
Ian Lance Taylor [Thu, 12 Mar 2020 04:51:09 +0000 (21:51 -0700)]
runtime: don't crash on mlock failure

Instead, note that mlock has failed, start trying the mitigation of
touching the signal stack before sending a preemption signal, and,
if the program crashes, mention the possible problem and a wiki page
describing the issue (https://golang.org/wiki/LinuxKernelSignalVectorBug).

Tested on a kernel in the buggy version range, but with the patch,
by using `ulimit -l 0`.

Fixes #37436

Change-Id: I072aadb2101496dffd655e442fa5c367dad46ce8
Reviewed-on: https://go-review.googlesource.com/c/go/+/223121
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agoRevert "misc/spectre: add spectre index test"
Bryan C. Mills [Fri, 13 Mar 2020 20:38:58 +0000 (20:38 +0000)]
Revert "misc/spectre: add spectre index test"

This reverts CL 222978.

Reason for revert: Test is failing frequently in TryBots and builders (https://build.golang.org/log/06a29585eaa9955cf33b50b5792b3bdc42e8fbe2)

Change-Id: I53bcf789726a9cfe78b1d3c55af78873c5c61696
Reviewed-on: https://go-review.googlesource.com/c/go/+/223378
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
5 years agocmd/go/internal/modload: suppress the 'go mod init' hint for GOROOT more aggressively
Bryan C. Mills [Fri, 13 Mar 2020 16:51:09 +0000 (12:51 -0400)]
cmd/go/internal/modload: suppress the 'go mod init' hint for GOROOT more aggressively

Previously, we suppressed a `to create a module there, run: … go mod
init' warning only if the config file itself (such as .git/config) was
found in GOROOT. However, our release tarballs don't include the
.git/config, so that case was not encountered, and the warning could
occur based on a config file found in some parent directory (outside
of GOROOT entirely).

Instead, skip the directory walk completely if the working directory
is anywhere in GOROOT.

Fixes #34191

Change-Id: I9f774901bfbb53b700407c4882f37d6339d023fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/223340
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/compile: more minor cleanup in shortcircuitBlock
Josh Bleecher Snyder [Sun, 8 Mar 2020 18:39:39 +0000 (11:39 -0700)]
cmd/compile: more minor cleanup in shortcircuitBlock

Continue to simplify, rename for clarity,
improve docs, and reduce variable scope.

This is in preparation for this function becoming
more complicated.

Passes toolstash-check.

Updates #37608

Change-Id: I630a4e07c92297c46d18aea69ec29852d6371ff0
Reviewed-on: https://go-review.googlesource.com/c/go/+/222919
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/compile: remove loop in shortcircuit
Josh Bleecher Snyder [Sun, 8 Mar 2020 18:07:52 +0000 (11:07 -0700)]
cmd/compile: remove loop in shortcircuit

shortcircuitBlock contained a loop to handle blocks like

b: <- p q
  v = Phi true false
If v -> t u

in a single execution.
This change makes shortcircuitBlock do it in two instead,
one for each constant phi arg.

Motivation: Upcoming changes will expand the range of
blocks that the shortcircuit pass can handle.
Those changes need to understand what the CFG
will look like after the rewrite in shortcircuitBlock.
Making shortcircuitBlock do only a single CFG
modification at a time significantly simplifies that code.

In theory, this is less efficient, but not measurably so.
There is minor, unimportant churn in the generated code.

Updates #37608

Change-Id: Ia6dce7011e3e19b546ed1e176bd407575a0ab837
Reviewed-on: https://go-review.googlesource.com/c/go/+/222918
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/compile: rename a local variable in shortcircuitBlock
Josh Bleecher Snyder [Sat, 7 Mar 2020 21:53:15 +0000 (13:53 -0800)]
cmd/compile: rename a local variable in shortcircuitBlock

v is pretty generic. Subsequent changes will make this function
more complicated, so rename it now, independently, for easier review.

v is the control value for the block (or its underlying phi);
call it ctl.

Passes toolstash-check.

Updates #37608

Change-Id: I3fbae3344f1c95aff0a69c1e4f61ef637a54774e
Reviewed-on: https://go-review.googlesource.com/c/go/+/222917
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agomisc/spectre: add spectre index test
Russ Cox [Wed, 11 Mar 2020 15:54:56 +0000 (11:54 -0400)]
misc/spectre: add spectre index test

Test for CL 222660.

Change-Id: I1dae41a9746dfc4144a0d29c02201de8ecd216fd
Reviewed-on: https://go-review.googlesource.com/c/go/+/222978
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/asm, cmd/compile, runtime: add -spectre=ret mode
Russ Cox [Fri, 17 Jan 2020 18:54:30 +0000 (13:54 -0500)]
cmd/asm, cmd/compile, runtime: add -spectre=ret mode

This commit extends the -spectre flag to cmd/asm and adds
a new Spectre mitigation mode "ret", which enables the use
of retpolines.

Retpolines prevent speculation about the target of an indirect
jump or call and are described in more detail here:
https://support.google.com/faqs/answer/7625886

Change-Id: I4f2cb982fa94e44d91e49bd98974fd125619c93a
Reviewed-on: https://go-review.googlesource.com/c/go/+/222661
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/compile: add spectre mitigation mode enabled by -spectre
Russ Cox [Mon, 6 Jan 2020 15:31:39 +0000 (10:31 -0500)]
cmd/compile: add spectre mitigation mode enabled by -spectre

This commit adds a new cmd/compile flag -spectre,
which accepts a comma-separated list of possible
Spectre mitigations to apply, or the empty string (none),
or "all". The only known mitigation right now is "index",
which uses conditional moves to ensure that x86-64 CPUs
do not speculate past index bounds checks.

Speculating past index bounds checks may be problematic
on systems running privileged servers that accept requests
from untrusted users who can execute their own programs
on the same machine. (And some more constraints that
make it even more unlikely in practice.)

The cases this protects against are analogous to the ones
Microsoft explains in the "Array out of bounds load/store feeding ..."
sections here:
https://docs.microsoft.com/en-us/cpp/security/developer-guidance-speculative-execution?view=vs-2019#array-out-of-bounds-load-feeding-an-indirect-branch

Change-Id: Ib7532d7e12466b17e04c4e2075c2a456dc98f610
Reviewed-on: https://go-review.googlesource.com/c/go/+/222660
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/compile: convert 386 port to use addressing modes pass
Keith Randall [Tue, 10 Mar 2020 21:08:40 +0000 (14:08 -0700)]
cmd/compile: convert 386 port to use addressing modes pass

Update #36468

Change-Id: Idfdb845d097994689be450d6e8a57fa9adb57166
Reviewed-on: https://go-review.googlesource.com/c/go/+/222782
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
5 years agocmd/cover: skip function declarations with blank names
zikaeroh [Thu, 12 Mar 2020 00:02:50 +0000 (17:02 -0700)]
cmd/cover: skip function declarations with blank names

Function declarations with blank ("_") names do not introduce a binding,
and therefore cannot be referenced or executed (in fact, they do not
make it into the final compiled binary at all). As such, counters
defined while annotating their bodies will always be zero.

These types of functions are commonly used to create compile-time
checks (e.g., stringer) which are not expected to be executed.

Skip over these functions when annotating a file, preventing the unused
counters from being generated and appearing as uncovered lines in
coverage reports.

Fixes #36264

Change-Id: I6b516cf43c430a6248d68d5f483a3902253fbdab
Reviewed-on: https://go-review.googlesource.com/c/go/+/223117
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/go: check for 'svn' binary in PATH in TestAccidentalGitCheckout
zikaeroh [Thu, 12 Mar 2020 00:17:57 +0000 (17:17 -0700)]
cmd/go: check for 'svn' binary in PATH in TestAccidentalGitCheckout

This test requires subversion to run, but does not check to see if it's
available before running as it does for git.

Call testenv.MustHaveExecPath to check beforehand to allow the test to
be skipped if the svn binary does not exist.

Change-Id: I16ae104621b221fc6e96f6c7dcd71bf406caa0c5
Reviewed-on: https://go-review.googlesource.com/c/go/+/223082
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/internal/obj/ppc64: clean up some opcodes
Lynn Boger [Tue, 10 Mar 2020 14:08:02 +0000 (10:08 -0400)]
cmd/internal/obj/ppc64: clean up some opcodes

This does some clean up of the ppc64 opcodes to remove names
from the opcode list that don't actually assemble. At one time
names were added to this list to represent opcode "classes" to
organize other opcodes that have the same set of operand
combinations. Since this is not documented, it is confusing as
to which opcodes can be used in an asm file and which can't, and
which opcodes should be supported in the disassembler. It is
clearer for the user if the list of Go opcodes are all opcodes
that can be assembled with names that match the ppc64 opcode
where possible.

I found this when trying to use Go opcode XXLAND in an asm file
which seems like it should map to ppc64 xxland but when used it
gets this error:

go tool asm test_xxland.s
asm: bad r/r, r/r/r or r/r/r/r opcode XXLAND
asm: assembly failed

This change removes the opcodes that are only used for opcode
"classes" and fixes the case statement where they are referenced.
This also fixes XXLAND and XXPERM which are opcodes that should
assemble to their corresponding ppc64 opcode but do not.

Change-Id: I52300db6b22f7f8b3dd3491c3f35a384b943352c
Reviewed-on: https://go-review.googlesource.com/c/go/+/223138
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agotime: stop referring to timerproc in comment
Ian Lance Taylor [Wed, 11 Mar 2020 05:11:00 +0000 (22:11 -0700)]
time: stop referring to timerproc in comment

The timerproc function has been removed.

Fixes #37774

Change-Id: Ice5e1d8fec91cd6ee7f032e0d21e8315a26bc6a3
Reviewed-on: https://go-review.googlesource.com/c/go/+/222783
Reviewed-by: Alberto Donizetti <alb.donizetti@gmail.com>
5 years agoruntime: emit more specific errors from checkptr
Johan Jansson [Wed, 11 Mar 2020 19:05:21 +0000 (21:05 +0200)]
runtime: emit more specific errors from checkptr

Update error messages for pointer alignment checks and pointer
arithmetic checks so that each type of error has a unique error
message.

Fixes #37488

Change-Id: Ida2c2fa3f041a3307d665879a463f9e8f2c1fd03
Reviewed-on: https://go-review.googlesource.com/c/go/+/223037
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>