]> Cypherpunks repositories - gostls13.git/log
gostls13.git
3 years agocmd/doc, go/doc: add basic support for generic code
Robert Findley [Fri, 29 Oct 2021 21:41:41 +0000 (17:41 -0400)]
cmd/doc, go/doc: add basic support for generic code

Update cmd/doc and go/doc for the generics, by adding handling for type
parameters and the new embedded interface elements.

Specifically:
 - Format type parameters when summarizing type and function nodes.
 - Find the origin type name for instantiation expressions, so that
   methods are associated with generic type declarations.
 - Generalize the handling of embedding 'error' in interfaces to
   arbitrary predeclared types.
 - Keep embedded type literals.
 - Update filtering to descend into embedded type literals.

Also add "any" to the list of predeclared types.

Updates #49210

Change-Id: I6ea82869f19c3cdbc3c842f01581c8fc7e1c2ee7
Reviewed-on: https://go-review.googlesource.com/c/go/+/359778
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agoruntime: allow builtin write function to be redirected with function pointer
Jason A. Donenfeld [Thu, 3 Dec 2020 12:29:58 +0000 (13:29 +0100)]
runtime: allow builtin write function to be redirected with function pointer

The x/sys/windows package currently uses go:linkname for other facilities
inside of runtime that are not suitable to be exposed as a public API
due to their dangers but are still necessary for manipulating any
low-level plumbing that the runtime controls.

Logging, via the built-in println and panic handler, is one such
low-level plumbing feature. In this case, x/sys/windows/svc needs to be
able to redirect panics to the Windows event log. Because the event log
is a complicated interface, this requires a bit more fiddling than the
simple solution used on Android (baking it into runtime itself), and
because Windows services are very diverse, the event log might not even
always be a desirable destination.

This commit accomplishes this by exposing a function pointer called
"overrideWrite" that low-level runtime packages like x/sys/windows/svc
can use to redirect output logs toward the event log or otherwise.

It is not safe or acceptable to use as a generic mechanism, and for that
reason, we wouldn't want to expose this as a real stable API, similar to
the other instances of go:linkname in x/sys/windows. But for packages
that must interoperate with low-level Go runtime fundamentals, this is a
safety hatch for packages that are developed in tandem with the runtime.
x/sys/windows is one such package.

Updates #42888.

Change-Id: I77a32ff7e1494324e8cc38e792e007f86d32672d
Reviewed-on: https://go-review.googlesource.com/c/go/+/278792
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
3 years agonet/http: reduce TestClientTimeout_h{1,2} latency
Bryan C. Mills [Thu, 4 Nov 2021 02:15:17 +0000 (22:15 -0400)]
net/http: reduce TestClientTimeout_h{1,2} latency

The test had been setting an arbitrary 200ms timeout to allow the
server's handler to set up before timing out. That is not only
potentially flaky on slow machines, but also typically much longer
than necessary. Replace the hard-coded timeout with a much shorter
initial timeout, and use exponential backoff to lengthen it if needed.

This allows the test to be run about 20x faster in the typical case,
which may make it easier to reproduce rare failure modes by running
with a higher -count flag.

For #43120

Change-Id: I1e0d0ec99d5a107fff56e3bcc7174d686ec582d1
Reviewed-on: https://go-review.googlesource.com/c/go/+/361275
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
3 years agoruntime: do not generate crash dump on Windows 7
Alex Brainman [Sun, 31 Oct 2021 06:58:34 +0000 (17:58 +1100)]
runtime: do not generate crash dump on Windows 7

It appears Windows 7 ignores WER_FAULT_REPORTING_NO_UI WerSetFlags
API flag.

And now after CL 307372, runtime will display WER GUI dialogue.

We don't want to introduce random GUI dialogues during Go program
execution. So disable dump crash creation on Windows 7 altogether.

Updates #20498

Change-Id: Ie268a7d4609f8a0eba4fe9ecf250856b0a61b331
Reviewed-on: https://go-review.googlesource.com/c/go/+/360617
Trust: Alex Brainman <alex.brainman@gmail.com>
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Patrik Nyblom <pnyb@google.com>
3 years agocmd/compile/internal/types2: implement compiler helper functions without using under
Robert Griesemer [Wed, 3 Nov 2021 19:39:43 +0000 (12:39 -0700)]
cmd/compile/internal/types2: implement compiler helper functions without using under

These functions are exported for the compiler and are used after
type checking is finished. There is no need to call under() in
their implementations; they can rely entirely on the public API.
This opens the door to moving them into the compiler eventually.
They may also be slightly more efficient.

Change-Id: Ib4f83d2dcf82e3c319c3147e01ecaea684553ea5
Reviewed-on: https://go-review.googlesource.com/c/go/+/361214
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: minor cleanups in predicates.go
Robert Griesemer [Wed, 3 Nov 2021 04:31:50 +0000 (21:31 -0700)]
cmd/compile/internal/types2: minor cleanups in predicates.go

- reordered some functions for better organization
- renamed single arguments typ to t for consistency
- updated some comments

No functional changes.

Change-Id: I4362ac48044595cdf5c3d9eb7b2f7b94e776d65b
Reviewed-on: https://go-review.googlesource.com/c/go/+/360956
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: rename is_X predicates back to isX (step 2 of 2)
Robert Griesemer [Wed, 3 Nov 2021 15:39:41 +0000 (08:39 -0700)]
cmd/compile/internal/types2: rename is_X predicates back to isX (step 2 of 2)

This is s/is_/is/ throughout. No other changes.

Change-Id: I1be77a209133edc68a6dec0677a4991a7683f116
Reviewed-on: https://go-review.googlesource.com/c/go/+/361134
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: rename isX predicates to allX, add simple is_X (step...
Robert Griesemer [Wed, 3 Nov 2021 00:31:51 +0000 (17:31 -0700)]
cmd/compile/internal/types2: rename isX predicates to allX, add simple is_X (step 1 of 2)

Rename the isX predicates to allX to clearly identify that these
predicates are looking inside type parameters.

Introduce is_X as predicates that do not look
inside type parameters so we can see all call sites.
The next CL will rename them all back to isX.

Review all call sites and use correct predicate.

Replace the single helper function is with isBasic and allBasic.

Change-Id: I3430ccfc466fdedf4b58a6158f95d47b9020f7a5

Change-Id: I81116b87cf8f2e17526723c7440676d133057aca
Reviewed-on: https://go-review.googlesource.com/c/go/+/360955
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: remove qualification from type parameter strings
Robert Griesemer [Thu, 4 Nov 2021 01:06:42 +0000 (18:06 -0700)]
cmd/compile/internal/types2: remove qualification from type parameter strings

This is a clean port of CL 360938 from go/types to types2.

For #49215.

Change-Id: If20779c862b71641cb6afe4c281fbad5488a8abe
Reviewed-on: https://go-review.googlesource.com/c/go/+/361266
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile: fix missing update source type in storeArgOrLoad
Cuong Manh Le [Mon, 1 Nov 2021 07:32:39 +0000 (14:32 +0700)]
cmd/compile: fix missing update source type in storeArgOrLoad

After removing trivial wrapper types, the source needs to be updated
with new type, otherwise, it leads to mismatch between field offset and
the source type for selecting struct/array.

Fixes #49249

Change-Id: I26f9440bcb2e78bcf0617afc21d9d40cdbe4aca6
Reviewed-on: https://go-review.googlesource.com/c/go/+/360057
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
3 years agosyscall: fix and document uses of RegEnumKeyEx
Michael Anthony Knyszek [Wed, 3 Nov 2021 18:18:18 +0000 (18:18 +0000)]
syscall: fix and document uses of RegEnumKeyEx

RegEnumKeyEx has an undocumented requirement that subsequent calls need
to be made from the same thread. This change documents that requirement
and fixes uses of it in std.

Fixes #49320.

Change-Id: I6d182227e22dd437996a878b3a978943f01b2d61
Reviewed-on: https://go-review.googlesource.com/c/go/+/361154
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocmd/compile: add extra test for the non-mono pass
Dan Scales [Wed, 3 Nov 2021 22:29:43 +0000 (15:29 -0700)]
cmd/compile: add extra test for the non-mono pass

Just add a test for another function that is not monomorphisable, which
comes from the Featherweight Go paper.

Updates #48018

Change-Id: I664e3f48412b02678e32b50204dc4befae90374c
Reviewed-on: https://go-review.googlesource.com/c/go/+/361262
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: remove qualification from type parameter strings
Robert Findley [Tue, 2 Nov 2021 23:49:15 +0000 (19:49 -0400)]
go/types: remove qualification from type parameter strings

Type parameters only exist within the scope of a type or function
declaration, so there is really no reason to package-qualify them. It is
also confusing to do so, as it makes their type string look like a
defined type.

Fixes #49215

Change-Id: I986c527e78d45f3cadd75bc3b244e23dabe707ee
Reviewed-on: https://go-review.googlesource.com/c/go/+/360938
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agocmd/trace: use fmt.Print for newline-ending fixed string
Zvonimir Pavlinovic [Wed, 3 Nov 2021 23:44:44 +0000 (16:44 -0700)]
cmd/trace: use fmt.Print for newline-ending fixed string

This redundancy is now caught by the improved printf vet checker.

Updates #49322

Change-Id: Ic7a931b8d4838be02ebb855b69624b95093bd114
Reviewed-on: https://go-review.googlesource.com/c/go/+/361265
Run-TryBot: Zvonimir Pavlinovic <zpavlinovic@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Zvonimir Pavlinovic <zpavlinovic@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocmd/compile/internal/syntax: better error message when type parameters are not permitted
Robert Griesemer [Wed, 3 Nov 2021 21:58:59 +0000 (14:58 -0700)]
cmd/compile/internal/syntax: better error message when type parameters are not permitted

Fixes #48382.

Change-Id: I215896a4429839c41c9136b6922b1b748ed47734
Reviewed-on: https://go-review.googlesource.com/c/go/+/361259
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/cover: use fmt.Print for newline-ending fixed string
Zvonimir Pavlinovic [Wed, 3 Nov 2021 23:31:11 +0000 (16:31 -0700)]
cmd/cover: use fmt.Print for newline-ending fixed string

This redundancy is now caught by the improved printf vet checker
(golang/go#30436).

Updates #49322

Change-Id: Id450247adc6fa28a9244c019be3c1b52c2d17f49
Reviewed-on: https://go-review.googlesource.com/c/go/+/361263
Run-TryBot: Zvonimir Pavlinovic <zpavlinovic@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Zvonimir Pavlinovic <zpavlinovic@google.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
3 years agosyscall: unify unix Recvmsg epilogues
Josh Bleecher Snyder [Tue, 2 Nov 2021 19:25:23 +0000 (12:25 -0700)]
syscall: unify unix Recvmsg epilogues

These were identical. This is a preliminary step
towards remove allocs per UDP receive.

Change-Id: I83106cd3f1fe4bc5bae2d1b0ebd23eedd820abed
Reviewed-on: https://go-review.googlesource.com/c/go/+/361258
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agosyscall: unify unix Sendmsg prologues
Josh Bleecher Snyder [Tue, 2 Nov 2021 18:59:56 +0000 (11:59 -0700)]
syscall: unify unix Sendmsg prologues

These were identical. This is a preliminary step
towards remove allocs per UDP send.

Change-Id: I21e1264c7d4747baa626ddb93afff4c1cf225d13
Reviewed-on: https://go-review.googlesource.com/c/go/+/361256
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agointernal/poll: remove alloc sending IPv6 UDP on Windows
Josh Bleecher Snyder [Wed, 3 Nov 2021 20:54:08 +0000 (13:54 -0700)]
internal/poll: remove alloc sending IPv6 UDP on Windows

This was an oversight in CL 331511.

Change-Id: Ibc20bf6ea80a8675d43d9691ed551dffab1d9215
Reviewed-on: https://go-review.googlesource.com/c/go/+/361254
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agonet/netip: fix various typos in comments
Jason A. Donenfeld [Wed, 3 Nov 2021 15:18:36 +0000 (16:18 +0100)]
net/netip: fix various typos in comments

Change-Id: I67998f70969ac6e6ff1b52a56e49a61ef167ed50
Reviewed-on: https://go-review.googlesource.com/c/go/+/361055
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agodatabase/sql: fix tx.Prepare documentation
Mostafa Solati [Tue, 3 Aug 2021 17:09:29 +0000 (21:39 +0430)]
database/sql: fix tx.Prepare documentation

Fixes: #45978
Change-Id: I3d5ccd621d670f97bb1a642b4e84b6629347a315
Reviewed-on: https://go-review.googlesource.com/c/go/+/339291
Reviewed-by: Daniel Theophanes <kardianos@gmail.com>
Trust: Cherry Mui <cherryyz@google.com>

3 years agodatabase/sql: use errors.Is when checking ErrBadConn
Daniel Theophanes [Mon, 12 Jul 2021 14:25:04 +0000 (09:25 -0500)]
database/sql: use errors.Is when checking ErrBadConn

When drivers return driver.ErrBadConn, no meaningful
information about what the cause of the problem is
returned. Ideally the driver.ErrBadConn would be
always caught with the retry loop, but this is not
always the case. Drivers today must choose between
returning a useful error and use the rety logic.
This allows supporting both.

Fixes #47142

Change-Id: I454573028f041dfdf874eed6c254fb194ccf6d96
Reviewed-on: https://go-review.googlesource.com/c/go/+/333949
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Trust: Ian Lance Taylor <iant@golang.org>
Trust: Daniel Theophanes <kardianos@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agosyscall: migrate new UDP send/recv API to internal/syscall/unix
Josh Bleecher Snyder [Wed, 3 Nov 2021 20:43:12 +0000 (13:43 -0700)]
syscall: migrate new UDP send/recv API to internal/syscall/unix

CL 331490 and friends added new API to package syscall.
This was a mistake that we need to fix before Go 1.18 is released.

Change-Id: I697c9a4fa649d564822f585dc163df5ab9e5ae08
Reviewed-on: https://go-review.googlesource.com/c/go/+/361216
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agocmd/compile: fix up shape type package
Keith Randall [Wed, 3 Nov 2021 18:26:37 +0000 (11:26 -0700)]
cmd/compile: fix up shape type package

Use go.shape instead of .shape as the package the compiler uses
to store shape types.

Prevent path escaping for compiler-internal types, so we don't
need to see %2e everywhere.

Change-Id: I98e39c3b6472560113bdea7e0ba6eb7b81cb35e0
Reviewed-on: https://go-review.googlesource.com/c/go/+/361174
Trust: Keith Randall <khr@golang.org>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
3 years agodatabase/sql: Fix idle connection reuse
Steven Hartland [Tue, 9 Jun 2020 07:58:08 +0000 (08:58 +0100)]
database/sql: Fix idle connection reuse

Fix idle connection reuse so that ConnMaxIdleTime clears down excessive
idle connections.

This now ensures that db.freeConn is ordered by returnedAt and that
connections that have been idle for the shortest period are reused
first.

In addition connectionCleanerRunLocked updates the next check deadline
based on idle and maximum life time information so that we avoid waiting
up to double MaxIdleTime to close connections.

Corrected the calling timer of connectionCleaner.

Fixes #39471

Change-Id: I6d26b3542179ef35aa13e5265a89bc0f08ba7fa1
Reviewed-on: https://go-review.googlesource.com/c/go/+/237337
Reviewed-by: Tamás Gulácsi <tgulacsi78@gmail.com>
Reviewed-by: Daniel Theophanes <kardianos@gmail.com>
Trust: Ian Lance Taylor <iant@golang.org>

3 years agoruntime: on windows, read nanotime with one instruction or issue barrier
Jason A. Donenfeld [Wed, 3 Nov 2021 17:19:04 +0000 (18:19 +0100)]
runtime: on windows, read nanotime with one instruction or issue barrier

On 64-bit, this is more efficient, and on ARM64, this prevents the time
from moving backwards due to the weaker memory model. On ARM32 due to
the weaker memory model, we issue a memory barrier.

Updates #48072.

Change-Id: If4695716c3039d8af14e14808af217f5c99fc93a
Reviewed-on: https://go-review.googlesource.com/c/go/+/361057
Trust: Jason A. Donenfeld <Jason@zx2c4.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
3 years agocmd/compile: make pointers to arrays their own shape
Keith Randall [Wed, 3 Nov 2021 16:40:32 +0000 (09:40 -0700)]
cmd/compile: make pointers to arrays their own shape

Pointers to arrays can be used to cast from a slice. We need
the shape of such type params to be different so we can compile
those casts correctly.

This is kind of a big hammer to fix #49295. It would be nice to
only do this when we know there's a []T->*[N]T conversion.

Fixes #49295

Change-Id: Ibda33057fab2dd28162537aab0f1244211d68e3f
Reviewed-on: https://go-review.googlesource.com/c/go/+/361135
Trust: Keith Randall <khr@golang.org>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
3 years agointernal/fuzz: improve error for mismatched types
Katie Hockman [Wed, 3 Nov 2021 14:58:59 +0000 (10:58 -0400)]
internal/fuzz: improve error for mismatched types

Fixes #48635

Change-Id: Ia3cde119d5eb31bc771fe3a39acb2372dbd988ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/361114
Trust: Katie Hockman <katie@golang.org>
Trust: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Katie Hockman <katie@golang.org>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agoruntime/pprof: disable GC during calibration test
Rhys Hiltner [Sat, 30 Oct 2021 00:35:30 +0000 (17:35 -0700)]
runtime/pprof: disable GC during calibration test

TestCPUProfileMultithreadMagnitude compares Go's CPU profile against the
OS's accounting of the process's execution time, expecting them to be
near equal. Background work from the runtime (especially in the garbage
collector) can add significant noise to that measurement and flakiness
to the test. Disable automatic GC cycles during the test.

Updates #49065

Change-Id: Ie88895bfea17374278c5187f3a83e9f486bd37fb
Reviewed-on: https://go-review.googlesource.com/c/go/+/359934
Reviewed-by: Michael Pratt <mpratt@google.com>
Trust: Michael Pratt <mpratt@google.com>
Trust: Michael Knyszek <mknyszek@google.com>

3 years agotest/typeparam: add test for indexing on typeparams with various constraints
Dan Scales [Wed, 3 Nov 2021 03:42:21 +0000 (20:42 -0700)]
test/typeparam: add test for indexing on typeparams with various constraints

Add test for indexing on variables whose types are constrained to
various kinds of types.

Change-Id: I991eecfe39dba5d817c0fbe259ba558d4881ea84
Reviewed-on: https://go-review.googlesource.com/c/go/+/360867
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agoall: add support for c-archive and c-shared on linux/riscv64
Joel Sing [Thu, 15 Jul 2021 17:06:18 +0000 (03:06 +1000)]
all: add support for c-archive and c-shared on linux/riscv64

This provides the runtime glue (_rt0_riscv64_linux_lib) for c-archive and c-shared
support, along with enabling both of these buildmodes on linux/riscv64.

Both misc/cgo/testcarchive and misc/cgo/testcshared now pass on this platform.

Fixes #47100

Change-Id: I7ad75b23ae1d592dbac60d15bba557668287711f
Reviewed-on: https://go-review.googlesource.com/c/go/+/334872
Trust: Joel Sing <joel@sing.id.au>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/asm,cmd/compile,runtime: stop using X3 (aka GP) on riscv64
Joel Sing [Sun, 26 Sep 2021 09:21:29 +0000 (19:21 +1000)]
cmd/asm,cmd/compile,runtime: stop using X3 (aka GP) on riscv64

The X3 (aka GP) register will potentially be loaded with the __global_pointer$ symbol
during program start up (usually by the dynamic linker). As such, non-Go code may depend
on the contents of GP and calculate offsets based on it, including code called via cgo
and signal handlers installed by non-Go code. As such, stop using the X3 register so
that there are fewer issues interacting between Go and non-Go code.

While here remove the X4 (TP) name from the assembler such that any references must
use the 'TP' name. This should reduce the likelihood of accidental use (like we do
for the 'g' register). The same applies for X3 (GP) when the -shared flag is given.

Updates #47100

Change-Id: I72e82b5ca3f80c46a781781345ca0432a4111b74
Reviewed-on: https://go-review.googlesource.com/c/go/+/351859
Trust: Joel Sing <joel@sing.id.au>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile/internal/types2: port nomono check from go/types
Matthew Dempsky [Tue, 2 Nov 2021 19:50:05 +0000 (12:50 -0700)]
cmd/compile/internal/types2: port nomono check from go/types

Same logic as CL 357449 (including CL 360815), just ported to types2.

Updates #48098.

Change-Id: I4578f7329bb4ffc42410025bb6cb97e24697ebfd
Reviewed-on: https://go-review.googlesource.com/c/go/+/360857
Trust: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agoruntime: install sigPreempt signal handler for c-archive/c-shared
Ian Lance Taylor [Tue, 2 Nov 2021 21:12:05 +0000 (14:12 -0700)]
runtime: install sigPreempt signal handler for c-archive/c-shared

Fixes #49288

Change-Id: I7bfcbecbefa68871a3e556935a73f241fff44c0e
Reviewed-on: https://go-review.googlesource.com/c/go/+/360861
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/compile/internal/types2: report cause for failing const conversions
Robert Griesemer [Tue, 2 Nov 2021 16:58:12 +0000 (09:58 -0700)]
cmd/compile/internal/types2: report cause for failing const conversions

Follow-up on CL 360396.

Change-Id: Icd802baffb1fef91f8fef0070b6167a438ceda1a
Reviewed-on: https://go-review.googlesource.com/c/go/+/360795
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agogo/types: report cause for failing const conversions
Robert Findley [Tue, 2 Nov 2021 22:42:51 +0000 (18:42 -0400)]
go/types: report cause for failing const conversions

This is a port of CL 360795 to go/types. Error messages were adjusted
accordingly, with a TODO to fix the discrepancy.

Change-Id: Ifd7d8248fa11a31fde391021f3c5f1840877892f
Reviewed-on: https://go-review.googlesource.com/c/go/+/360937
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: fix indexing of generic types
Robert Findley [Tue, 2 Nov 2021 22:36:36 +0000 (18:36 -0400)]
go/types: fix indexing of generic types

This is a clean port of CL 360603 to go/types.

Change-Id: Iadb312f07e509ff83339d5525765b7b7987bf233
Reviewed-on: https://go-review.googlesource.com/c/go/+/360936
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: adjust printing of type parameter in error
Robert Findley [Tue, 2 Nov 2021 22:29:27 +0000 (18:29 -0400)]
go/types: adjust printing of type parameter in error

This is a clean port of CL 360514 to go/types.

Change-Id: Ia13638b3758b3b8017867934d09136ac5f9a62ca
Reviewed-on: https://go-review.googlesource.com/c/go/+/360935
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: fix conversions of constants to type parameter
Robert Findley [Tue, 2 Nov 2021 22:25:21 +0000 (18:25 -0400)]
go/types: fix conversions of constants to type parameter

This is a port of both CL 360396 and CL 360796 to go/types. The latter
is added to avoid introducing an intermediate bug.

An error message was adjusted in issue49296.go2, with a TODO to switch
to the types2 error.

Change-Id: Iede294b69b4e097e53876498f039ee18667568c4
Reviewed-on: https://go-review.googlesource.com/c/go/+/360934
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agoruntime/asan: declare asan reporting functions in C
Ian Lance Taylor [Tue, 2 Nov 2021 20:56:22 +0000 (13:56 -0700)]
runtime/asan: declare asan reporting functions in C

Fixes #49287

Change-Id: I3f970d6db0b1dee886df2f7bb6cbeb56a37c42c2
Reviewed-on: https://go-review.googlesource.com/c/go/+/360860
Trust: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile: remove TODO that is no longer relevant
Dan Scales [Tue, 2 Nov 2021 19:30:37 +0000 (12:30 -0700)]
cmd/compile: remove TODO that is no longer relevant

types2 methods are now sorted in the same way as types1, so this TODO is
no longer needed.  (Comment change only).

Change-Id: Ic975ce001a5d54f15381a9cb7b6969dff795e3b4
Reviewed-on: https://go-review.googlesource.com/c/go/+/360856
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
3 years agodoc/go1.18: remove some redundancy from netip docs
Brad Fitzpatrick [Tue, 2 Nov 2021 14:59:16 +0000 (07:59 -0700)]
doc/go1.18: remove some redundancy from netip docs

Change-Id: I593862331add3366f6b337844588568301746b80
Reviewed-on: https://go-review.googlesource.com/c/go/+/360608
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Brad Fitzpatrick <bradfitz@golang.org>

3 years agogo/types: differently named types are not assignable
Robert Findley [Tue, 2 Nov 2021 15:46:25 +0000 (11:46 -0400)]
go/types: differently named types are not assignable

This is a clean port of CL 360274 to go/types.

Change-Id: Idfa584fab95f7226e10b1a7c5b06d56a0bf9d757
Reviewed-on: https://go-review.googlesource.com/c/go/+/360759
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: disallow lone type parameter on RHS of type declaration
Robert Findley [Tue, 2 Nov 2021 15:42:54 +0000 (11:42 -0400)]
go/types: disallow lone type parameter on RHS of type declaration

This is a clean port of CL 359177 to go/types.

Change-Id: I6c1574e3a1c23ea326eb8d4aacea0e78415703ab
Reviewed-on: https://go-review.googlesource.com/c/go/+/360758
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: implement singleType and structure (type)
Robert Findley [Tue, 2 Nov 2021 15:34:11 +0000 (11:34 -0400)]
go/types: implement singleType and structure (type)

This is a clean port of CL 359015 to go/types.

Change-Id: Iea4e3bfe0a4ae0e5a9052cb6e66c01405bd57c3d
Reviewed-on: https://go-review.googlesource.com/c/go/+/360756
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: rename isNamed predicate to hasName
Robert Findley [Tue, 2 Nov 2021 15:23:19 +0000 (11:23 -0400)]
go/types: rename isNamed predicate to hasName

This is a clean port of CL 358621 to go/types.

Change-Id: I4e858b1b70cff69b6e0e76bb8a58a70ff54990c9
Reviewed-on: https://go-review.googlesource.com/c/go/+/360755
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: clean up asT converters (step 2 of 2)
Robert Findley [Tue, 2 Nov 2021 15:20:10 +0000 (11:20 -0400)]
go/types: clean up asT converters (step 2 of 2)

This is a clean port of CL 358774 to go/types.

Change-Id: Icba54336de2b8de7c2002d2a44cac856907178c9
Reviewed-on: https://go-review.googlesource.com/c/go/+/360754
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agoerrors: improve wording in As doc comment
Josh Bleecher Snyder [Tue, 2 Nov 2021 18:05:03 +0000 (11:05 -0700)]
errors: improve wording in As doc comment

"so" didn't have an antecedent.

Change-Id: I27f7b334decea7bc34bfa3f3f2d3a79874c6fe90
Reviewed-on: https://go-review.googlesource.com/c/go/+/360797
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agoarchive/zip: don't panic on (*Reader).Open
Jason7602 [Tue, 14 Sep 2021 13:38:19 +0000 (21:38 +0800)]
archive/zip: don't panic on (*Reader).Open

Previously, opening a zip with (*Reader).Open could result in a panic if
the zip contained a file whose name was exclusively made up of slash
characters or ".." path elements.

Open could also panic if passed the empty string directly as an argument.

Now, any files in the zip whose name could not be made valid for
fs.FS.Open will be skipped, and no longer added to the fs.FS file list,
although they are still accessible through (*Reader).File.

Note that it was already the case that a file could be accessible from
(*Reader).Open with a name different from the one in (*Reader).File, as
the former is the cleaned name, while the latter is the original one.

Finally, made the actual panic site robust as a defense-in-depth measure.

Fixes CVE-2021-41772
Fixes #48085

Co-authored-by: Filippo Valsorda <filippo@golang.org>
Change-Id: I6271a3f2892e7746f52e213b8eba9a1bba974678
Reviewed-on: https://go-review.googlesource.com/c/go/+/349770
Run-TryBot: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Katie Hockman <katie@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Trust: Katie Hockman <katie@golang.org>
Trust: Julie Qiu <julie@golang.org>

3 years agogo/types: clean up asT converters (step 1 of 2)
Robert Findley [Mon, 1 Nov 2021 19:27:13 +0000 (15:27 -0400)]
go/types: clean up asT converters (step 1 of 2)

This is a port of CL 358597 to go/types. A comment was missing in the
base of applyTypeFunc, which had been there since the initial check-in
of types2; somehow it was not in go/types.

Change-Id: If08efd92d782dd3099b26254ae6e311c6cea8c3b
Reviewed-on: https://go-review.googlesource.com/c/go/+/360477
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agosyscall: fix finalizer fd close bugs in TestFcntlFlock and TestPassFD
Michael Anthony Knyszek [Mon, 1 Nov 2021 22:35:29 +0000 (22:35 +0000)]
syscall: fix finalizer fd close bugs in TestFcntlFlock and TestPassFD

Currently, the syscall test suite takes very little time to run. It
stands to reason that pretty much every time, zero GCs execute.

With CL 309869, this changes because the minimum heap size is lowered,
triggering two bugs in the test suite.

One bug is in TestFcntlFlock, where a raw FD is wrapped in an os.File
whose last reference is passed into a Cmd. That FD is then closed by a
defer syscall.Close, instead of the os.File's Close, so the finalizer
may fire *after* that FD has already been reused by another test.

The second bug is in the child helper process of TestPassFD, where
there's a small window in which a temp file's FD is encoded for an
out-of-band unix domain socket message to the parent, but not yet sent.
The point of encoding is also the last reference that FD's os.File, so a
finalizer may run at any time. While it's safe for the finalizer to run
after the FD is sent, if it runs before, the send will fail, since unix
domain sockets require that any sent FDs are valid.

Change-Id: I2d1bd7e6db6efcc6763273217fd85cb5b9764274
Reviewed-on: https://go-review.googlesource.com/c/go/+/360575
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agogo/types: avoid closure allocations in mono check
Matthew Dempsky [Tue, 2 Nov 2021 18:23:34 +0000 (11:23 -0700)]
go/types: avoid closure allocations in mono check

This CL replaces monoEdge's "report" field with fields "pos" and
"typ", and pushes the logic for formatting them into the report
loop. This avoids needing to allocate a function closure for each
edge.

Also tweak a test case so the two type parameters involved in the
cycle aren't both "T" so they're easier to understand.

Change-Id: I9d392ad1d99a4c5e89da4613084e885149ebad07
Reviewed-on: https://go-review.googlesource.com/c/go/+/360815
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agonet/http: deflake TestServerKeepAlivesEnabled_h{1,2}
Damien Neil [Tue, 2 Nov 2021 18:52:36 +0000 (11:52 -0700)]
net/http: deflake TestServerKeepAlivesEnabled_h{1,2}

This test assumes that two successive TCP connections will use different
source ports. This does not appear to be a universally safe assumption.

Rewrite the test to use httptrace to detect connection reuse instead.

Fixes #46707

Change-Id: Iebfbdfdeb77a1e6663a0c654dc847cc270c5d54d
Reviewed-on: https://go-review.googlesource.com/c/go/+/360854
Trust: Damien Neil <dneil@google.com>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocmd/compile/internal/abi: fix some typos in comments
Than McIntosh [Tue, 2 Nov 2021 17:44:05 +0000 (13:44 -0400)]
cmd/compile/internal/abi: fix some typos in comments

Fix a few types in the header comment for 'ComputePadding'.

Change-Id: If38911e6dcbec571845ae44eef30fd6c33f589cb
Reviewed-on: https://go-review.googlesource.com/c/go/+/360814
Trust: Than McIntosh <thanm@google.com>
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agonet: add ReadFromUDPAddrPort
Josh Bleecher Snyder [Mon, 1 Nov 2021 20:45:21 +0000 (13:45 -0700)]
net: add ReadFromUDPAddrPort

It is now possible to do completely allocation-free UDP.

This is implemented completely separately from ReadFromUDP
because ReadFromUDP exists in a delicate balance to allow
mid-stack inlining. After performance-sensitive callers have
migrated to ReadFromUDPAddrPort, we may be able to simplify
ReadFromUDP to call ReadFromUDPAddrPort.

name                          old time/op    new time/op    delta
WriteToReadFromUDPAddrPort-8    4.71µs ± 2%    4.81µs ± 5%    +2.18%  (p=0.000 n=14+14)

name                          old alloc/op   new alloc/op   delta
WriteToReadFromUDPAddrPort-8     4.00B ± 0%     0.00B       -100.00%  (p=0.000 n=15+15)

name                          old allocs/op  new allocs/op  delta
WriteToReadFromUDPAddrPort-8      1.00 ± 0%      0.00       -100.00%  (p=0.000 n=15+15)

Change-Id: I37f5ad9416a1d4333ed48d83474b2cf933b2a1be
Reviewed-on: https://go-review.googlesource.com/c/go/+/360600
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agonet: optimize WriteToUDPAddrPort
Josh Bleecher Snyder [Mon, 1 Nov 2021 20:23:39 +0000 (13:23 -0700)]
net: optimize WriteToUDPAddrPort

See the recent change optimizing WriteMsgUDPAddrPort
for an explanation of why this change includes copy/paste/modify
instead of implementing WriteToUDP using WriteToUDPAddrPort.

name                          old time/op    new time/op    delta
WriteToReadFromUDPAddrPort-8    5.02µs ± 3%    4.71µs ± 2%   -6.31%  (p=0.000 n=15+14)

name                          old alloc/op   new alloc/op   delta
WriteToReadFromUDPAddrPort-8     68.0B ± 0%      4.0B ± 0%  -94.12%  (p=0.000 n=15+15)

name                          old allocs/op  new allocs/op  delta
WriteToReadFromUDPAddrPort-8      3.00 ± 0%      1.00 ± 0%  -66.67%  (p=0.000 n=15+15)

Change-Id: I301715e774de07eb6ccb4e329ccf2e554609abc1
Reviewed-on: https://go-review.googlesource.com/c/go/+/360599
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agonet: add BenchmarkWriteToReadFromUDPAddrPort
Josh Bleecher Snyder [Mon, 1 Nov 2021 21:07:30 +0000 (14:07 -0700)]
net: add BenchmarkWriteToReadFromUDPAddrPort

The main item of interest is the allocs.

BenchmarkWriteToReadFromUDPAddrPort-8      237735       5152 ns/op       68 B/op        3 allocs/op

Change-Id: I5cb3f2e1ebbb02acc51039c2266e99b90d388a99
Reviewed-on: https://go-review.googlesource.com/c/go/+/360598
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agocmd/compile: mark type descriptors as always dupok
Keith Randall [Sat, 30 Oct 2021 21:39:29 +0000 (14:39 -0700)]
cmd/compile: mark type descriptors as always dupok

The types of the two interfaces should be equal, but they aren't.
We end up with multiple descriptors for a type when we need type
descriptors to be unique.

Fixes #49241

Change-Id: I8a6c70da541c6088a92a01392bc83b61cc130eba
Reviewed-on: https://go-review.googlesource.com/c/go/+/360134
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
3 years agodebug/gosym: bulk allocate Syms
Josh Bleecher Snyder [Wed, 6 Oct 2021 19:40:35 +0000 (12:40 -0700)]
debug/gosym: bulk allocate Syms

name                old time/op    new time/op    delta
115/NewLineTable-8    80.7ns ± 1%    81.7ns ± 3%     ~     (p=0.408 n=10+8)
115/NewTable-8        76.3µs ± 2%    63.3µs ± 3%  -17.03%  (p=0.000 n=9+9)
115/LineToPC-8        56.8µs ± 2%    56.8µs ± 3%     ~     (p=0.905 n=10+9)
115/PCToLine-8         252ns ± 1%     256ns ± 5%     ~     (p=0.183 n=10+10)

name                old alloc/op   new alloc/op   delta
115/NewLineTable-8      384B ± 0%      384B ± 0%     ~     (all equal)
115/NewTable-8         164kB ± 0%     172kB ± 0%   +4.97%  (p=0.000 n=10+10)
115/LineToPC-8         0.00B          0.00B          ~     (all equal)
115/PCToLine-8         0.00B          0.00B          ~     (all equal)

name                old allocs/op  new allocs/op  delta
115/NewLineTable-8      3.00 ± 0%      3.00 ± 0%     ~     (all equal)
115/NewTable-8         1.04k ± 0%     0.01k ± 0%  -98.75%  (p=0.000 n=10+10)
115/LineToPC-8          0.00           0.00          ~     (all equal)
115/PCToLine-8          0.00           0.00          ~     (all equal)

Change-Id: Ibfe6c0cb86737dee8209cb1f58f5b796aa7b30aa
Reviewed-on: https://go-review.googlesource.com/c/go/+/353134
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agocmd/go: disallow profile flags when using -fuzz
Roland Shoemaker [Mon, 1 Nov 2021 17:28:52 +0000 (10:28 -0700)]
cmd/go: disallow profile flags when using -fuzz

Disallow the usage of coverprofile, blockprofile, cpuprofile, memprofile,
mutexprofile, and trace when fuzzing, since these currently do not work
as the user would expect.

Fixes #48178

Change-Id: Ibe6dfaef9b28a91c60c26fcad38c33a2a959d869
Reviewed-on: https://go-review.googlesource.com/c/go/+/360395
Trust: Roland Shoemaker <roland@golang.org>
Trust: Katie Hockman <katie@golang.org>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Katie Hockman <katie@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agogo/types: add check that code is monomorphizable
Matthew Dempsky [Wed, 20 Oct 2021 20:17:07 +0000 (13:17 -0700)]
go/types: add check that code is monomorphizable

This CL adds a check to ensure that generic Go code doesn't involve
any unbounded recursive instantiation, which are incompatible with an
implementation that uses static instantiation (i.e., monomorphization
or compile-time dictionary construction).

Updates #48098.

Change-Id: I9d051f0f9369ab881592a361a5d0e2a716788a6b
Reviewed-on: https://go-review.googlesource.com/c/go/+/357449
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>

3 years agocomd/compile/internal/types2: add missing nil check in const conversion
Robert Griesemer [Tue, 2 Nov 2021 17:14:29 +0000 (10:14 -0700)]
comd/compile/internal/types2: add missing nil check in const conversion

Follow-up on CL 360396.

Fixes #49296.

Change-Id: Ie08f86ae884da4cfd5db557cbf4f721a237dc39f
Reviewed-on: https://go-review.googlesource.com/c/go/+/360796
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agobytes: test for page boundary crosses on sep of Index
Paul E. Murphy [Tue, 19 Oct 2021 16:17:53 +0000 (11:17 -0500)]
bytes: test for page boundary crosses on sep of Index

Improve TestIndexNearPageBoundary to verify needles
ending on a page boundary don't cause a segfault.

Change-Id: I2edb13db63a71dc9955e266f6b97026ee13bf76e
Reviewed-on: https://go-review.googlesource.com/c/go/+/356889
Reviewed-by: Cherry Mui <cherryyz@google.com>
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>

3 years agogo/types: generalize assignability to generic types
Robert Findley [Mon, 1 Nov 2021 19:11:31 +0000 (15:11 -0400)]
go/types: generalize assignability to generic types

This is a port of CL 357917 to go/types. Some error messages in
assignability.go2 had to be adjusted. I left a TODO to investigate
whether we should align error messages.

Change-Id: Ia323ffe18bc08e82de62044f35b8b0f3edd7dc08
Reviewed-on: https://go-review.googlesource.com/c/go/+/360476
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: clarify is/underIs semantics and implementation
Robert Findley [Mon, 1 Nov 2021 19:03:53 +0000 (15:03 -0400)]
go/types: clarify is/underIs semantics and implementation

This is a port of CL 358594 to go/types. Some code in conversions.go had
to be trivially reorganized to align with types2 -- I'm not sure how
go/types diverged from the base.

Change-Id: I40ce247bbb3b9d0e87ce88c50e440c12774c0745
Reviewed-on: https://go-review.googlesource.com/c/go/+/360475
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types: better error messages for empty type sets
Robert Findley [Mon, 1 Nov 2021 18:46:35 +0000 (14:46 -0400)]
go/types: better error messages for empty type sets

This is a clean port of CL 358175 to go/types.

Change-Id: If1b4e51d1579fd168e651d79d031335ff09ca128
Reviewed-on: https://go-review.googlesource.com/c/go/+/360474
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agoruntime: check amd64 microarchitecture level at startup
vpachkov [Sun, 31 Oct 2021 11:06:27 +0000 (11:06 +0000)]
runtime: check amd64 microarchitecture level at startup

Make Go runtime throw if it's been compiled to assume instruction
set extensions that aren't available on the CPU.
Updates #48506

Change-Id: Ic4d6696e1cd6b28d389a86fe64e8175ea3ca135a
GitHub-Last-Rev: ba338377f549344b416fbd519576e95ce5d523be
GitHub-Pull-Request: golang/go#48514
Reviewed-on: https://go-review.googlesource.com/c/go/+/351191
Run-TryBot: Martin Möhrmann <martin@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Martin Möhrmann <martin@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Martin Möhrmann <martin@golang.org>

3 years agocmd/compile: explain special treatment when rewrite slice literal args
Cuong Manh Le [Tue, 2 Nov 2021 03:19:24 +0000 (10:19 +0700)]
cmd/compile: explain special treatment when rewrite slice literal args

Followup discussion in CL 360055.

Change-Id: I36212c2a497b152d01ed86d244d5f57bd34a64a6
Reviewed-on: https://go-review.googlesource.com/c/go/+/360614
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile: fix rewriting slice literal call argument
Cuong Manh Le [Sat, 30 Oct 2021 17:20:13 +0000 (00:20 +0700)]
cmd/compile: fix rewriting slice literal call argument

When seeing Key:Value expression in slice literal, the compiler only
needs to emit tmp var for the Value, not the whole expression.

Fixes #49240

Change-Id: I7bda3c796a93c0fa1974f7c5930f38025dfa665c
Reviewed-on: https://go-review.googlesource.com/c/go/+/360055
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
3 years agoRevert "runtime: fix missing pprof labels"
Bryan C. Mills [Tue, 2 Nov 2021 15:38:47 +0000 (15:38 +0000)]
Revert "runtime: fix missing pprof labels"

This reverts CL 351751.

Reason for revert: new test is failing on many builders.

Change-Id: I066211c9f25607ca9eb5299aedea2ecc5069e34f
Reviewed-on: https://go-review.googlesource.com/c/go/+/360757
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agocmd/compile/internal/types2: fix indexing of generic types
Robert Griesemer [Mon, 1 Nov 2021 23:05:14 +0000 (16:05 -0700)]
cmd/compile/internal/types2: fix indexing of generic types

Correctly track if the index expression is addressable.
Rewrote code slightly.

Fixes #49275.

Change-Id: Ic54edd0213a091173ff5403ab0e3e1f1fca0e361
Reviewed-on: https://go-review.googlesource.com/c/go/+/360603
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: adjust printing of type parameter in error
Robert Griesemer [Mon, 1 Nov 2021 20:41:32 +0000 (13:41 -0700)]
cmd/compile/internal/types2: adjust printing of type parameter in error

For constraint type inference failures where the type parameter doesn't
match the constraint, print the type parameter rather than its type name
object. This provides more flexibility for improving the error message
down the road.

Change-Id: I188871d6f26a16cd96e59770966a1ec65607b128
Reviewed-on: https://go-review.googlesource.com/c/go/+/360514
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: fix conversions of constants to type parameter
Robert Griesemer [Mon, 1 Nov 2021 19:14:25 +0000 (12:14 -0700)]
cmd/compile/internal/types2: fix conversions of constants to type parameter

When converting a constant to a type parameter, the result is never
constant (type parameters are not constant types), but we still need
to verfy that the constant is representable by each specific type in
the type set of the type parameter.

Fixes #49247.

Change-Id: I460983c7070b33baadce25dd23210e10930cfb2c
Reviewed-on: https://go-review.googlesource.com/c/go/+/360396
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agointernal/fuzz: minimize bytes to be human readable
Amelia Downs [Mon, 27 Sep 2021 20:54:39 +0000 (16:54 -0400)]
internal/fuzz: minimize bytes to be human readable

Try to replace every byte with one of the following printable
characters: "012789ABCXYZabcxyz !\"#$%&'()*+,.".

Fixes #48129

Change-Id: Ie58f6bbc3431d50d9f0a3f608ba63e854ac6ce79
Reviewed-on: https://go-review.googlesource.com/c/go/+/352614
Reviewed-by: Katie Hockman <katie@golang.org>
Reviewed-by: Amelia Downs <adowns@vmware.com>
Trust: Katie Hockman <katie@golang.org>
Trust: Julie Qiu <julie@golang.org>
Run-TryBot: Katie Hockman <katie@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoruntime: fix missing pprof labels
Felix Geisendörfer [Mon, 20 Sep 2021 14:09:47 +0000 (16:09 +0200)]
runtime: fix missing pprof labels

Use gp.m.curg instead of the gp when recording cpu profiler stack
traces. This ensures profiler labels are captured when systemstack or similar
is executing on behalf of the current goroutine.

After this there are still rare cases of samples containing the labelHog
function, so more work might be needed. This patch should fix ~99% of the
problem.

Fixes #48577.

Change-Id: I27132110e3d09721ec3b3ef417122bc70d8f3279
Reviewed-on: https://go-review.googlesource.com/c/go/+/351751
Reviewed-by: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Michael Knyszek <mknyszek@google.com>

3 years agosyscall: remove GOMAXPROCS change in TestExecHelper
Michael Anthony Knyszek [Mon, 1 Nov 2021 19:39:18 +0000 (19:39 +0000)]
syscall: remove GOMAXPROCS change in TestExecHelper

TestExec and TestExecHelper check for a workaround of a particular OS
bug on darwin that's triggered more often via asynchronous preemption.

As part of this, the test sets up 100 CPU-bound goroutines, and sets
GOMAXPROCS to 50, sleeping for a little bit before calling Exec. Thus
far, this is fine because the scheduler runs the Execing goroutine in a
timely manner. However, CL 309869 will reduce the minimum heap size,
causing a GC to happen during the test.

On a 16 CPU machine, with GOMAXPROCS at 50, and 100 CPU-bound
goroutines, both the OS scheduler and the Go scheduler are severly
oversaturated. As a result, the test often (not always, but often) runs
out for the full lifetime of those 100 goroutines, which run for about 1
second.

The minimum heap size reduction is not necessary to trigger this; an
additional call to runtime.GC in the helper is also sufficient to
trigger this delay.

The delay on its own isn't great, since it adds a whole second to
all.bash on its own. However, it also seems correlated with other
subprocess tests in the syscall package, namely TestPassFD and
TestFcntlFlock. These tests fail in a fairly superficial way: the file
descriptor for the temporary directories they make gets clobbered, is
closed, or becomes stale.

Change-Id: I213dd5e38967d19a8b317e6d4c5024b57f9e3fed
Reviewed-on: https://go-review.googlesource.com/c/go/+/360574
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agoA+C: add Leonard Wang (individual CLA)
Leonard Wang [Sat, 30 Oct 2021 14:26:56 +0000 (22:26 +0800)]
A+C: add Leonard Wang (individual CLA)

Change-Id: I9fa687804c7f0f1d5a987b95ab1c3110468d1b65
Reviewed-on: https://go-review.googlesource.com/c/go/+/360054
Reviewed-by: Meng Zhuo <mzh@golangcn.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Meng Zhuo <mzh@golangcn.org>
Run-TryBot: Meng Zhuo <mzh@golangcn.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/dist: add asan tests in misc/cgo/testsanitizers package
fanzha02 [Tue, 5 Jan 2021 09:52:43 +0000 (17:52 +0800)]
cmd/dist: add asan tests in misc/cgo/testsanitizers package

Add asan tests to check the use of Go with -asan option.

Currenly, the address sanitizer in Go only checks for error
memory access to heap objects.

TODO: Enable check for error memory access to global objects.

Updates #44853.

Change-Id: I83579f229f117b5684a369fc8f365f4dea140648
Reviewed-on: https://go-review.googlesource.com/c/go/+/298615
Trust: fannie zhang <Fannie.Zhang@arm.com>
Run-TryBot: fannie zhang <Fannie.Zhang@arm.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agodoc/go1.18: add net/netip
Brad Fitzpatrick [Tue, 2 Nov 2021 05:25:55 +0000 (22:25 -0700)]
doc/go1.18: add net/netip

Updates #46518

Change-Id: Iefc2fec93becc7b36ba2b933ae79f7c9424317fc
Reviewed-on: https://go-review.googlesource.com/c/go/+/360634
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Brad Fitzpatrick <bradfitz@golang.org>

3 years agocmd/go, internal/cpu: use internal/godebug in tests
Brad Fitzpatrick [Tue, 2 Nov 2021 03:25:48 +0000 (20:25 -0700)]
cmd/go, internal/cpu: use internal/godebug in tests

Change-Id: Ifdf67e778e88ee70780428aa5479d2e091752a3a
Reviewed-on: https://go-review.googlesource.com/c/go/+/360605
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
Trust: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agoruntime: add padding to Linux kernel structures
Rhys Hiltner [Thu, 30 Sep 2021 00:35:27 +0000 (17:35 -0700)]
runtime: add padding to Linux kernel structures

Go exchanges siginfo and sigevent structures with the kernel. They
contain unions, but Go's use is limited to the first few fields. Pad out
the rest so the size Go sees is the same as what the Linux kernel sees.

This is a follow-up to CL 342052 which added the sigevent struct without
padding, and to CL 353136 which added the padding but with an assertion
that confused several type-checkers. It updates the siginfo struct as
well so there are no bad examples in the defs_linux_*.go files.

Reviewed-on: https://go-review.googlesource.com/c/go/+/353136

Change-Id: I9610632ff0ec43eba91f560536f5441fa907b36f
Reviewed-on: https://go-review.googlesource.com/c/go/+/360094
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agoruntime, syscall: add calls to asan functions
fanzha02 [Tue, 5 Jan 2021 09:52:43 +0000 (17:52 +0800)]
runtime, syscall: add calls to asan functions

Add explicit address sanitizer instrumentation to the runtime and
syscall packages. The compiler does not instrument the runtime
package. It does instrument the syscall package, but we need to add
a couple of cases that it can't see.

Refer to the implementation of the asan malloc runtime library,
this patch also allocates extra memory as the redzone, around the
returned memory region, and marks the redzone as unaddressable to
detect the overflows or underflows.

Updates #44853.

Change-Id: I2753d1cc1296935a66bf521e31ce91e35fcdf798
Reviewed-on: https://go-review.googlesource.com/c/go/+/298614
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: fannie zhang <Fannie.Zhang@arm.com>

3 years agonet: optimize WriteMsgUDPAddrPort
Josh Bleecher Snyder [Mon, 1 Nov 2021 20:34:08 +0000 (13:34 -0700)]
net: optimize WriteMsgUDPAddrPort

This is one step towards optimizing WriteMsgUDPAddrPort.
Further steps remain, namely to avoid the syscall.Sockaddr interface,
as we do for UDPConn.WriteToUDP and UDPConn.ReadFromUDP.

A previous change optimized ReadMsgUDPAddrPort by having
ReadMsgUDP call ReadMsgUDPAddrPort rather than the other way around.

This change does not implement WriteMsgUDP in terms of WriteMsgUDPAddrPort,
because a few layers deep, on posix platforms only
(in ipToSockaddrInet4 and ipToSockaddrInet6),
is special handling of zero-length IP addresses and IPv4zero.
It treats IP(nil) as equivalent to 0.0.0.0 or ::,
and 0.0.0.0 as equivalent to :: in an IPv6 context.

Based on the comments, I suspect that this treatment was intended
for the Listen* API, not the Write* API, but it affects both,
and I am nervous about changing the behavior for Write*.

The netip package doesn't have a way to represent a "zero-length IP address"
as distinct from an invalid IP address (which is a good thing),
so to implement WriteMsgUDP using WriteMsgUDPAddrPort,
we would have to duplicate this special handling at the start of WriteMsgUDP.
But this handling depends on whether the UDPConn is an IPv4 or an IPv6 conn,
which is also platform-specific information.

As a result, every attempt I made to implement WriteMsgUDP using
WriteMsgUDPAddrPort while preserving behavior ended up
being considerably worse than copy/paste/modify.

This does mean that WriteMsgUDP and WriteMsgUDPAddrPort will have
different behavior in these cases.

name                       old time/op    new time/op    delta
ReadWriteMsgUDPAddrPort-8    5.29µs ± 6%    5.02µs ± 7%   -5.14%  (p=0.000 n=13+15)

name                       old alloc/op   new alloc/op   delta
ReadWriteMsgUDPAddrPort-8      128B ± 0%       64B ± 0%  -50.00%  (p=0.000 n=15+15)

name                       old allocs/op  new allocs/op  delta
ReadWriteMsgUDPAddrPort-8      4.00 ± 0%      2.00 ± 0%  -50.00%  (p=0.000 n=15+15)

Change-Id: Ia78eb49734f4301d7772dfdbb5a87e4d303a9f7a
Reviewed-on: https://go-review.googlesource.com/c/go/+/360597
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agonet: optimize ReadMsgUDPAddrPort
Josh Bleecher Snyder [Mon, 1 Nov 2021 19:44:43 +0000 (12:44 -0700)]
net: optimize ReadMsgUDPAddrPort

Instead of implementing ReadMsgUDPAddrPort in terms of ReadMsgUDP,
do it the other way around. This keeps the code minimal while
still avoiding allocs.

We could also rearrange ReadMsgUDP to be mid-stack inlined to avoid
allocating the *UDPAddr, but anyone who's trying to eliminate
allocs should use ReadMsgUDPAddrPort instead anyway,
because ReadMsgUDP will always allocate at least once (the IP slice).

name                       old time/op    new time/op    delta
ReadWriteMsgUDPAddrPort-8    5.26µs ± 3%    5.29µs ± 6%     ~     (p=0.429 n=12+13)

name                       old alloc/op   new alloc/op   delta
ReadWriteMsgUDPAddrPort-8      176B ± 0%      128B ± 0%  -27.27%  (p=0.000 n=15+15)

name                       old allocs/op  new allocs/op  delta
ReadWriteMsgUDPAddrPort-8      5.00 ± 0%      4.00 ± 0%  -20.00%  (p=0.000 n=15+15)

Change-Id: I15228cb4ec4f13f2f390407b6c62c44c228e7201
Reviewed-on: https://go-review.googlesource.com/c/go/+/360596
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agonet: add BenchmarkReadWriteMsgUDPAddrPort
Josh Bleecher Snyder [Mon, 1 Nov 2021 21:05:14 +0000 (14:05 -0700)]
net: add BenchmarkReadWriteMsgUDPAddrPort

The main item of interest here is the allocs.

BenchmarkReadWriteMsgUDPAddrPort-8      195920       6450 ns/op      176 B/op        5 allocs/op

Change-Id: I358bd18888ce254b6c6d5d5c920e6d22ce5a2ef4
Reviewed-on: https://go-review.googlesource.com/c/go/+/360595
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agoencoding: simplify some code
Josh Bleecher Snyder [Tue, 2 Nov 2021 00:27:09 +0000 (17:27 -0700)]
encoding: simplify some code

Change-Id: I63aac663da4ce7bdd198d385f87929c74f6a5cf8
Reviewed-on: https://go-review.googlesource.com/c/go/+/360601
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agosyscall,internal/poll: copy arrays by assignment instead of looping
Josh Bleecher Snyder [Tue, 2 Nov 2021 00:28:19 +0000 (17:28 -0700)]
syscall,internal/poll: copy arrays by assignment instead of looping

golang.org/x/sys contains similar code and also needs updating.

Change-Id: Id00177397639075d4792eb253829d8042941b70c
Reviewed-on: https://go-review.googlesource.com/c/go/+/360602
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agocmd/compile: fix panic when refer to method of imported instantiated type
Cuong Manh Le [Sun, 31 Oct 2021 19:04:13 +0000 (02:04 +0700)]
cmd/compile: fix panic when refer to method of imported instantiated type

In case of reference to method call of an imported fully-instantiated
type, nameNode.Func will be nil causes checkFetchBody panic. To fix
this, make sure checkFetchBody is only called when Func is not nil.

Fixes #49246

Change-Id: I32e9208385a86d4600d8ebf6f5efd8fca571ea16
Reviewed-on: https://go-review.googlesource.com/c/go/+/360056
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
3 years agonet/netip: add new IP address package
Brad Fitzpatrick [Mon, 2 Aug 2021 21:55:51 +0000 (14:55 -0700)]
net/netip: add new IP address package

Co-authored-by: Alex Willmer <alex@moreati.org.uk> (GitHub @moreati)
Co-authored-by: Alexander Yastrebov <yastrebov.alex@gmail.com>
Co-authored-by: David Anderson <dave@natulte.net> (Tailscale CLA)
Co-authored-by: David Crawshaw <crawshaw@tailscale.com> (Tailscale CLA)
Co-authored-by: Dmytro Shynkevych <dmytro@tailscale.com> (Tailscale CLA)
Co-authored-by: Elias Naur <mail@eliasnaur.com>
Co-authored-by: Joe Tsai <joetsai@digital-static.net> (Tailscale CLA)
Co-authored-by: Jonathan Yu <jawnsy@cpan.org> (GitHub @jawnsy)
Co-authored-by: Josh Bleecher Snyder <josharian@gmail.com> (Tailscale CLA)
Co-authored-by: Maisem Ali <maisem@tailscale.com> (Tailscale CLA)
Co-authored-by: Manuel Mendez (Go AUTHORS mmendez534@...)
Co-authored-by: Matt Layher <mdlayher@gmail.com>
Co-authored-by: Noah Treuhaft <noah.treuhaft@gmail.com> (GitHub @nwt)
Co-authored-by: Stefan Majer <stefan.majer@gmail.com>
Co-authored-by: Terin Stock <terinjokes@gmail.com> (Cloudflare CLA)
Co-authored-by: Tobias Klauser <tklauser@distanz.ch>
Fixes #46518

Change-Id: I0041f9e1115d61fa6e95fcf32b01d9faee708712
Reviewed-on: https://go-review.googlesource.com/c/go/+/339309
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
Trust: Brad Fitzpatrick <bradfitz@golang.org>

3 years agonet/url: consider ForceQuery in ResolveReference
Timothy Gu [Fri, 7 May 2021 08:03:59 +0000 (01:03 -0700)]
net/url: consider ForceQuery in ResolveReference

Previously, when resolving references of form
  (https://golang.org/?hello).ResolveReference(?)
we only used URL.RawQuery to determine whether or not a query part is
defined. Go 1.7 introduced URL.ForceQuery as a flag for the situation
where a query part is provided but empty. But we did not use it in
ResolveReference. This leads to the erroneous output
  https://golang.org/?hello
when the correct output should be
  https://golang.org/?
This commit rectifies that error.

Fixes #46033

Change-Id: I05bc0b48bf2bbf13b4ddc0dd10599ea613dc2188
Reviewed-on: https://go-review.googlesource.com/c/go/+/317930
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Damien Neil <dneil@google.com>

3 years agogo/types, types2: remove confusing debugging output in inference error message
Robert Griesemer [Mon, 1 Nov 2021 22:00:16 +0000 (15:00 -0700)]
go/types, types2: remove confusing debugging output in inference error message

When type inference cannot infer type arguments it prints the list of
inferred type arguments (often empty) at the end of the error message.
This was meant as debugging support only. Removed for now.

Eventually we should provide a better error message.

Fixes #49272.

Change-Id: I68d43a6ace91081009cab0f2fbad7bfbddf5e76b
Reviewed-on: https://go-review.googlesource.com/c/go/+/360554
Trust: Robert Griesemer <gri@golang.org>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
3 years agosync: add PoolStarvation benchmark
Ruslan Andreev [Mon, 1 Nov 2021 18:17:49 +0000 (18:17 +0000)]
sync: add PoolStarvation benchmark

This benchmark simulates object starvation in order to force Ps to steal
objects from other Ps. Extracted from CL 314229.

Change-Id: Iee31df355ba04d80fbd91c4414e397a375e6d6d7
Reviewed-on: https://go-review.googlesource.com/c/go/+/360256
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Trust: Cherry Mui <cherryyz@google.com>

3 years agoruntime: disable pacer lock held assertions in tests
Michael Anthony Knyszek [Fri, 29 Oct 2021 22:42:50 +0000 (22:42 +0000)]
runtime: disable pacer lock held assertions in tests

Fixes #49234.

Change-Id: I64c1eab0dce2bbe990343b43a32858a6c9f3dcda
Reviewed-on: https://go-review.googlesource.com/c/go/+/359878
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
3 years agointernal/fuzz: don't add duplicate corpus entries
Roland Shoemaker [Fri, 29 Oct 2021 18:21:45 +0000 (11:21 -0700)]
internal/fuzz: don't add duplicate corpus entries

If a identical input is already present in the corpus, don't re-add it.
This may happen when the same input produces a different coverage map,
causing the coordinator to think it has found a new input.

This fixes a race between reading/writing cached inputs.

Fixes #48721

Change-Id: I4807602f433c2b99396d25ceaa58b827796b3555
Reviewed-on: https://go-review.googlesource.com/c/go/+/359755
Trust: Roland Shoemaker <roland@golang.org>
Trust: Katie Hockman <katie@golang.org>
Run-TryBot: Roland Shoemaker <roland@golang.org>
Run-TryBot: Katie Hockman <katie@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Katie Hockman <katie@golang.org>
3 years agogo/types: fix TypeName.IsAlias for type parameter names
zhouguangyuan [Fri, 29 Oct 2021 07:24:28 +0000 (15:24 +0800)]
go/types: fix TypeName.IsAlias for type parameter names

Fixes #49213

Change-Id: I2bfc151b74b0d14efbd00e5d28584f4180126c5d
Reviewed-on: https://go-review.googlesource.com/c/go/+/359656
Trust: Robert Griesemer <gri@golang.org>
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile/internal/types2: fix TypeName.IsAlias for type parameter names
Robert Griesemer [Mon, 1 Nov 2021 00:27:12 +0000 (17:27 -0700)]
cmd/compile/internal/types2: fix TypeName.IsAlias for type parameter names

This is a port of CL 359656 from go/types to types2.

For #49213.

Change-Id: Ib98f9a344c1397af92e061cafd519ea374fd60bc
Reviewed-on: https://go-review.googlesource.com/c/go/+/360294
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: remove subscripts from type parameter names
Robert Griesemer [Sun, 31 Oct 2021 19:08:32 +0000 (12:08 -0700)]
cmd/compile/internal/types2: remove subscripts from type parameter names

This is a port of CL 357814 from go/types to types2 with minor
adjustments due to small differences in error handling code.

Change-Id: I72ecc4532e8349f569cabb38006f3d8ff517bf30
Reviewed-on: https://go-review.googlesource.com/c/go/+/360276
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: differently named types are not assignable
Robert Griesemer [Sun, 31 Oct 2021 17:37:15 +0000 (10:37 -0700)]
cmd/compile/internal/types2: differently named types are not assignable

When checking assignability, a value of a named type (incl. a type parameter)
can never be assigned to a variable of a differently named type. Specifically,
if the types are two differently named type parameters, then values of one are
never assignable to variables of the other.

This CL clarifies the assignabiliy rules and simplifies the implementation.

Fixes #49242.

Change-Id: Id72a2c9bed5cdb726855e7a707137db1009e7953
Reviewed-on: https://go-review.googlesource.com/c/go/+/360274
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agostrings: align Clone output with string([]byte(""))
Martin Möhrmann [Mon, 1 Nov 2021 05:42:01 +0000 (06:42 +0100)]
strings: align Clone output with string([]byte(""))

Add a comment how strings of length 0 are treated and
that they always will result in the return of a string
equal to the constant string "".

The previous implementation would return a string header that uses
runtime.zerobase as the backing array pointer while the string constant
"" has 0 as pointer value.

Using 0 has the backing array pointer is also the behaviour of
string([]byte(input)) which makes the new behaviour a better drop in
replacement.

Change-Id: Ic5460e9494b6708edbdfa4361e878d50db54ba10
Reviewed-on: https://go-review.googlesource.com/c/go/+/360255
Trust: Martin Möhrmann <martin@golang.org>
Run-TryBot: Martin Möhrmann <martin@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>
3 years agotest/codegen: updated arithmetic tests to verify on ppc64,ppc64le
Archana R [Tue, 26 Oct 2021 04:57:46 +0000 (23:57 -0500)]
test/codegen: updated arithmetic tests to verify on ppc64,ppc64le

Updated multiple tests in test/codegen/arithmetic.go to verify
on ppc64/ppc64le as well

Change-Id: I79ca9f87017ea31147a4ba16f5d42ba0fcae64e1
Reviewed-on: https://go-review.googlesource.com/c/go/+/358546
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>