]> Cypherpunks repositories - gostls13.git/log
gostls13.git
3 years agocmd/cgo: add missing tab in exports for a result of void
Leonard Wang [Sat, 24 Jul 2021 14:17:21 +0000 (22:17 +0800)]
cmd/cgo: add missing tab in exports for a result of void

Change-Id: I45575afbad364c13b7179ebe3f3dfc4ed9671d2b
Reviewed-on: https://go-review.googlesource.com/c/go/+/336891
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Meng Zhuo <mzh@golangcn.org>
3 years agocmd/compile: remove unneeded early transforms, with dictionary change
Dan Scales [Sun, 12 Sep 2021 19:21:48 +0000 (12:21 -0700)]
cmd/compile: remove unneeded early transforms, with dictionary change

Now that we are computing the dictionary format on the instantiated
functions, we can remove the early transformation code that was needed
to create the implicit CONVIFACE nodes in the generic function.

Change-Id: I1695484e7d59bccbfb757994f3e40e84288759a5
Reviewed-on: https://go-review.googlesource.com/c/go/+/349614
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Dan Scales <danscales@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
3 years agocmd/compile: switch to computing dict format on instantiated functions
Dan Scales [Mon, 13 Sep 2021 20:53:18 +0000 (13:53 -0700)]
cmd/compile: switch to computing dict format on instantiated functions

Change to computing the dictionary form on each shape-instantiated
function, rather than once on the underlying generic function/method.
The problem with computing the dictionary format on the generic function
is that we had to force early transformations to create all the
needed/implicit CONVIFACE nodes, since many of these nodes cause the
need for a dictionary entry. Also, the dictionary entries needed can
different with different instantiations of the same generic function,
especially depending on whether a type argument is a non-interface or
interface type, or a instantiated type vs. a non-instantiated type.

By computing the dictionary format on the instantiated function, we are
scanning a function where all the transformations have been done to
create implicit CONVFIFACE nodes, and we know the above relevant
information about the type params (which are shapes).

Much of the change is more mechanical changes from typeparams to shapes,
and generic functions/info to instantiated functions/info. Some of the
most important non-mechanical changes are:

 - Separated out the dictionary transformations to nodes into a separate
   dictPass, since we need to analyze instantiated functions after
   stenciling, but before the dictionary transformations.

 - Added type param index to shape types, since we need to be able
   distinguish type params of an instantiation which are different but
   happen to have the same shape.

 - Allow the type substituter to work with shapes again (since for the
   dictionary entries we need to substitute shape params to the concrete
   type args).

 - Support types.IdentityStrict() that does strict type comparison (no
   special case for shapes). This needed for type substitution,
   formatting and creating dictionaries, etc. We can maybe create better
   names for this function.

 - Add new information to instInfo to contain a mapping from the shape
   type params to their instantiated type bound. This is needed when
   doing the dictionary transformations related to type bounds.

Change-Id: I1c3ca312c5384f318c4dd7d0858dba9766396ff6
Reviewed-on: https://go-review.googlesource.com/c/go/+/349613
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
3 years agocmd/internal/obj/ppc64: generate prologue code compatible with new ABI
Lynn Boger [Mon, 16 Aug 2021 14:37:07 +0000 (09:37 -0500)]
cmd/internal/obj/ppc64: generate prologue code compatible with new ABI

This changes the ppc64 prologue to avoid clobbering the registers
that could contain incoming argument values. This means preserving
the values in R3 - R10 and R14 - R19 for ppc64.

Instead of modifying R3, R4, R5 and R6 the registers R22, R23, R24
and R25 are used.

The argument registers that could be clobbered by the call to
morestack are saved and restored around that call.

Change-Id: If354c3dc73f2c8537ef4e513e5a4c05d7bd22866
Reviewed-on: https://go-review.googlesource.com/c/go/+/343872
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agoruntime: fix setting of cpu features for amd64
nimelehin [Wed, 15 Sep 2021 18:37:08 +0000 (18:37 +0000)]
runtime: fix setting of cpu features for amd64

Because of wrong case of letters, the cpu features flags were not
set properly for amd64.

Fixes #48406.

Change-Id: If19782851670e91fd31d119f4701c47373fa7e71
GitHub-Last-Rev: 91c7321ca49343c86917f071babec8a004ca5c77
GitHub-Pull-Request: golang/go#48403
Reviewed-on: https://go-review.googlesource.com/c/go/+/350151
Trust: Keith Randall <khr@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/go/internal/modload: prevent tidy downgrading disambiguating modules
Jay Conrod [Tue, 24 Aug 2021 18:51:07 +0000 (11:51 -0700)]
cmd/go/internal/modload: prevent tidy downgrading disambiguating modules

If an indirectly required module does not provide any packages needed
to build packages in the main module but is needed to disambiguate
imports, 'go mod tidy' may keep an indirect requirement on that module
to prevent it from being downgraded. This can prevent the introduction
of new ambiguities. This also ensures tidy keeps sums needed to load
all packages.

Fixes #47738

Change-Id: Ib8e33422b95394707894cda7cfbb71a4b111e0ed
Reviewed-on: https://go-review.googlesource.com/c/go/+/344572
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocmd/compile: emit DWARF info about dictionary entries
Alessandro Arzilli [Wed, 25 Aug 2021 08:08:07 +0000 (10:08 +0200)]
cmd/compile: emit DWARF info about dictionary entries

When emitting the DIE of the instantiation of a generic function also
emit one DW_TAG_typedef_type entry for each dictionary entry in use,
referencing the shape type and having a custom attribute containing the
index inside the dictionary.

When emitting the DIE of variables that have an instantiated parametric
type, instead of referencing the shape type directly go through the
DW_TAG_typedef_type entry emitted for the dictionary entry describing
the real type of the variable.

Change-Id: Ia45d157ecd4c25e2b60300469e43bbb27a663582
Reviewed-on: https://go-review.googlesource.com/c/go/+/344929
Run-TryBot: Alessandro Arzilli <alessandro.arzilli@gmail.com>
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Trust: Dan Scales <danscales@google.com>
Trust: Jeremy Faller <jeremy@golang.org>

3 years agocmd/compile: mark wrapper functions with DW_AT_trampoline
Alessandro Arzilli [Fri, 3 Sep 2021 13:04:52 +0000 (15:04 +0200)]
cmd/compile: mark wrapper functions with DW_AT_trampoline

Change DWARF generation to tag wrapper functions with the
"DW_AT_trampoline attribute". The intent is that debuggers can pick up
on this attr so as to skip through the wrapper to the eventual target.

DWARF standard allows for a couple of different possible variants of
the trampoline attr; this is the simplest variant (all it tells the
debugger is that the function is a wrapper, doesn't include a
reference to the wrapper routine).

This implementation keys off the WRAPPER LSym attribute, which is set
for method wrappers, ABI wrappers, and a selected set of runtime
assembly routines (ex: "runtime.call32").

Change-Id: Ib53e1bc56c02b86ca3ac5e7da1a541ec262726cf
Reviewed-on: https://go-review.googlesource.com/c/go/+/347352
Reviewed-by: Than McIntosh <thanm@google.com>
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Jeremy Faller <jeremy@golang.org>

3 years agogo/internal/gcimporter,cmd/compile: minor clean-up in iimport.go
Robert Findley [Wed, 15 Sep 2021 00:40:09 +0000 (20:40 -0400)]
go/internal/gcimporter,cmd/compile: minor clean-up in iimport.go

Make two superficial fixes to iimport.go: rename instType to
instanceType (suggested in CL 349949), and fix a stale comment.

Done in both go/internal/gcimporter and cmd/compile/internal/importer.

Change-Id: Idfdda11a59b036a35824bbb1c101cba3652aeff4
Reviewed-on: https://go-review.googlesource.com/c/go/+/350031
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/compile/internal/types2: eliminate Named.instPos
Robert Griesemer [Wed, 15 Sep 2021 00:13:32 +0000 (17:13 -0700)]
cmd/compile/internal/types2: eliminate Named.instPos

This is a clean port of CL 349411 from go/types to types2.

Change-Id: Id5fa04c53f286dad263d7ba7911cb49eebf47b0e
Reviewed-on: https://go-review.googlesource.com/c/go/+/350030
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: merge Named type loading and expansion
Robert Griesemer [Wed, 15 Sep 2021 00:02:30 +0000 (17:02 -0700)]
cmd/compile/internal/types2: merge Named type loading and expansion

Clean port of CL 349410 from go/types to types2 with 2 adjustments:
using syntax.Pos instead of token.Pos, and using TypeHash instead
of typeHash.

Fixes #47887.

Change-Id: Ifd8495e4187b5e30aaf80702768d82aad5e10cf4
Reviewed-on: https://go-review.googlesource.com/c/go/+/349995
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: implement Identical for *Union types
Robert Griesemer [Tue, 14 Sep 2021 20:23:37 +0000 (13:23 -0700)]
cmd/compile/internal/types2: implement Identical for *Union types

This is a clean port of CL 349413 from go/types to types2.

Change-Id: I18bad5e29b1e719b30a73fb2aa32fe252538496e
Reviewed-on: https://go-review.googlesource.com/c/go/+/349992
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: remove some unnecessary loading/expansion of Named types
Robert Griesemer [Tue, 14 Sep 2021 20:18:40 +0000 (13:18 -0700)]
cmd/compile/internal/types2: remove some unnecessary loading/expansion of Named types

This is a clean port of CL 349409 from go/types to types2.

Change-Id: I2deb9ce46e6dcda736fda2169912c02163930d7d
Reviewed-on: https://go-review.googlesource.com/c/go/+/349991
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: export TypeHash, return value without blanks
Robert Griesemer [Tue, 14 Sep 2021 20:09:19 +0000 (13:09 -0700)]
cmd/compile/internal/types2: export TypeHash, return value without blanks

Change the typeWriter to produce blank-free hashes where easily possible
if used as a type hasher, and replace remaining blanks with '#' is needed.

Exported Environment.TypeHash for use by the compiler.

Change-Id: Icbd364c207f9c139a7a1844bb695512a0c56a4e4
Reviewed-on: https://go-review.googlesource.com/c/go/+/349990
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agoruntime: in adjustTimers back up as far as necessary
Ian Lance Taylor [Fri, 20 Aug 2021 23:55:04 +0000 (16:55 -0700)]
runtime: in adjustTimers back up as far as necessary

When the adjustTimers function removed a timer it assumed it was
sufficient to continue the heap traversal at that position.
However, in some cases a timer will be moved to an earlier
position in the heap. If that timer is timerModifiedEarlier,
that can leave timerModifiedEarliest not correctly representing
the earlier such timer.

Fix the problem by restarting the heap traversal at the earliest
changed position.

Fixes #47762

Change-Id: I152bbe62793ee40a680baf49967bcb89b1f94764
Reviewed-on: https://go-review.googlesource.com/c/go/+/343882
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
3 years agocmd/go: remove subcommand prefix from error messages
Jay Conrod [Mon, 13 Sep 2021 17:58:25 +0000 (10:58 -0700)]
cmd/go: remove subcommand prefix from error messages

For example, errors that started before with "go mod download: " now
start with "go: " instead.

Previously, we had a mix of errors with and without subcommand
prefixes, even in packages like modload that ostensibly aren't tied
to any specific command. This change makes usage more consistent,
which makes refactoring much easier.

These prefixes didn't add useful information: the user should know the
subcommand they just ran. But see CL 347152 for an attempt at making
the opposite change: always printing the subcommand prefix.

Note that there are a number of errors that don't start with "go: " or
any subcommand prefix. This CL doesn't affect those.

Change-Id: I16430d8c39ea3f4d0870f55a5205f06fb21943c0
Reviewed-on: https://go-review.googlesource.com/c/go/+/349597
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agogo/types: implement Identical for *Union types
Robert Findley [Sun, 12 Sep 2021 18:33:04 +0000 (14:33 -0400)]
go/types: implement Identical for *Union types

This aligns with the API proposal (#47916).

Change-Id: I732e5b107e729718ed37e053ad3f434993a97ecd
Reviewed-on: https://go-review.googlesource.com/c/go/+/349413
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: minor cleanup of instantiation
Robert Findley [Sun, 12 Sep 2021 03:06:41 +0000 (23:06 -0400)]
go/types: minor cleanup of instantiation

This CL addresses a couple TODOs related to instantiation:
 - factor out resolving the best environment
 - don't eagerly resolve substituted instances

Change-Id: I4a5de7ea7939b6f272991071f591d622dec04b53
Reviewed-on: https://go-review.googlesource.com/c/go/+/349429
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: instantiate methods when instantiating Named types
Robert Findley [Fri, 10 Sep 2021 20:28:01 +0000 (16:28 -0400)]
go/types: instantiate methods when instantiating Named types

In the API proposal we decided that instantiation must also instantiate
methods. This CL does that, and eliminates the special handling for lazy
instantiation in lookupMethod.

It is possible that we expand an instance before all method signatures
have been type-checked, so for simplicity we introduce a new flag on
Func, 'isIncompleteMethod', which controls whether we must fully
substitute methods before using them. We could avoid this flag by using
some convention for the structure of an incomplete method (such as the
receiver has no position), but in practice using a flag was cleaner and
didn't increase the size of the Func struct.

Updates #47916

Change-Id: I352baa6664cd07f61b06924744382897805f9d29
Reviewed-on: https://go-review.googlesource.com/c/go/+/349412
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: eliminate Named.instPos
Robert Findley [Fri, 10 Sep 2021 19:12:57 +0000 (15:12 -0400)]
go/types: eliminate Named.instPos

We no longer need to use the nilness of Named.instPos to signal whether
instance expansion has occurred, so remove it from the Named struct by
instead closing over the instantiation position in the resolver.

This means we cannot print instance markers for unexpanded instances:
instances may escape the type checking pass without being fully
expanded, and we can not check whether they have been expanded in a
concurrency-safe way without introducing a more heavy-weight
syncronization mechanism.

With this change, instantiation should be concurrency safe, modulo bugs
of course as we have little test coverage of concurrency (see #47729).

Fixes #47910

Change-Id: Ifeef6df296f00105579554b333a44d08aae113c9
Reviewed-on: https://go-review.googlesource.com/c/go/+/349411
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: merge Named type loading and expansion
Robert Findley [Fri, 10 Sep 2021 14:23:23 +0000 (10:23 -0400)]
go/types: merge Named type loading and expansion

Named type expansion and loading were conceptually similar: a mechanism
for lazily resolving type information in a concurrency-safe manner.
Unify them into a 'resolve' method, that delegates to a resolver func to
produce type parameters, underlying, and methods.

By leveraging the sync.Once field on Named for instance expansion, we
get closer to making instance expansion concurrency-safe, and remove the
requirement that instPos guard instantiation. This will be cleaned up
in a follow-up CL.

This also fixes #47887 by causing substituted type instances to be
expanded (in the old code, this could be fixed by setting instPos when
substituting).

For #47910
Fixes #47887

Change-Id: Ifc52a420dde76e3a46ce494fea9bd289bc8aca4c
Reviewed-on: https://go-review.googlesource.com/c/go/+/349410
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/compile: set IsShape based on type being in the Shapes pkg
Dan Scales [Tue, 14 Sep 2021 15:39:08 +0000 (08:39 -0700)]
cmd/compile: set IsShape based on type being in the Shapes pkg

Move ShapePkg to types, and change types.NewNamed to automatically set
IsShape/HasShape if a type is in the shapes pkg. This means that
imported shape types will automatically have the correct
IsShape/HasShape flags, even though we are not explicitly
exporting/importing those flags.

Updates #48337

Change-Id: I8b6131a663205f73f395943c9d0c8bdb2a213401
Reviewed-on: https://go-review.googlesource.com/c/go/+/349869
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Dan Scales <danscales@google.com>

3 years agocmd/compile: fix test/typeparam/mdempsky/4.go for -G=3
wdvxdr [Fri, 10 Sep 2021 12:56:29 +0000 (20:56 +0800)]
cmd/compile: fix test/typeparam/mdempsky/4.go for -G=3

Change-Id: I894ee000561a3c6afede8df697b1bce4576ceef0
Reviewed-on: https://go-review.googlesource.com/c/go/+/349012
Reviewed-by: Dan Scales <danscales@google.com>
Trust: Dan Scales <danscales@google.com>
Trust: Keith Randall <khr@golang.org>

3 years agoruntime: avoid loop variable capture in test
Michael Pratt [Wed, 4 Aug 2021 20:07:47 +0000 (16:07 -0400)]
runtime: avoid loop variable capture in test

In TestSegv, the t.Run closure captures the loop variable 'test'. Since
the subtest calls t.Parallel, the parent test is allowed to keep
running, changing the loop variable and thus changing the value of
'test' in the subtest.

Change-Id: I021ddc50304de08a341e6ffe486aa54e573d3b94
Reviewed-on: https://go-review.googlesource.com/c/go/+/339911
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agogo/internal/gcimporter: remove outdated comment
Cuong Manh Le [Tue, 14 Sep 2021 03:36:29 +0000 (10:36 +0700)]
go/internal/gcimporter: remove outdated comment

CL 349010 ported the fix from CL 349009, but forgot to remove the
outdated comment from old solution. This CL removes that one.

Change-Id: Ia401295e9d0984f4a088ddce5db09d306bfd89b6
Reviewed-on: https://go-review.googlesource.com/c/go/+/349729
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: Robert Findley <rfindley@google.com>
3 years agosyscall: remove use of IN_KUBERNETES in test
Tobias Klauser [Tue, 14 Sep 2021 06:56:54 +0000 (08:56 +0200)]
syscall: remove use of IN_KUBERNETES in test

CL 201737 dropped the use of IN_KUBERNETES in tests, but it looks like
it did not catch all occurrences.

For #12815
For #34956

Change-Id: I72b89bfb850ba2890e9e6aa39b87167291ab7e9f
Reviewed-on: https://go-review.googlesource.com/c/go/+/349789
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
3 years agocmd/internal/obj/ppc64: allow VR register arguments to VS registers
Paul E. Murphy [Tue, 9 Mar 2021 22:55:31 +0000 (16:55 -0600)]
cmd/internal/obj/ppc64: allow VR register arguments to VS registers

Likewise, reorder register numbers such that extended mnemonics which
use FPR arguments can be transparently encoded as a VSR argument for
the move to/from VSR class of instructions. Specifically, ensure the
following holds for all FPx and VRx constants: FPRx & 63 == x, and
VRx & 63 == x + 32.

This simplifies encoding machine instructions, and likewise helps
ppc64 assembly writers to avoid hokey workarounds when switching from
vector to vector-scalar register notation. Notably, many VSX
instructions are limited to vector operands due to encoding
restrictions.

Secondly, this explicitly rejects dubious usages of the m[tf]vsr
family of instructions which had previously been accepted.

 * Reject two GPR arguments for non-MTVSRDD opcodes. These
   have no defined behavior today, and may set RFU bits. e.g
   MTVSRD R1, R2, VS1

 * Reject FPR destinations for MTVSRDD, and only accept with two GPR
   arguments. This copies two GPR values into either half of a VSR. e.g
   MTVSRDD R1, R2, F1
   MTVSRDD R1, F1

Change-Id: If13dd88c3791d1892dbd18ef0e34675a5285fff9
Reviewed-on: https://go-review.googlesource.com/c/go/+/342929
Run-TryBot: Paul Murphy <murp@ibm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/compile: prevent typecheck importer reading type parameter twice
Cuong Manh Le [Fri, 10 Sep 2021 01:34:03 +0000 (08:34 +0700)]
cmd/compile: prevent typecheck importer reading type parameter twice

This is a port of CL 349009 to typecheck importer.

Fixes #48306

Change-Id: Iec3f078089346bd85f0ab739896e079940325011
Reviewed-on: https://go-review.googlesource.com/c/go/+/349011
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 agogo/internal/gcimporter: prevent importReader reading type parameter twice
Cuong Manh Le [Fri, 10 Sep 2021 01:20:28 +0000 (08:20 +0700)]
go/internal/gcimporter: prevent importReader reading type parameter twice

This is port of CL 349009 to go/internal/gcimporter.

Updates #48280

Change-Id: I7d40d8b67333538ca58fe012535d54e891d0ed16
Reviewed-on: https://go-review.googlesource.com/c/go/+/349010
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: Robert Findley <rfindley@google.com>
3 years agocmd/compile: prevent importReader reading type parameter twice
Cuong Manh Le [Fri, 10 Sep 2021 00:44:02 +0000 (07:44 +0700)]
cmd/compile: prevent importReader reading type parameter twice

The importReader always reads type parameter before declaring type stub
declaration. Thus, for recursive type, the type parameter is going to be
read twice, cause the bound more than once error.

To fix this, only read the type parameter after declaring stub obj, thus
r.doDecl can see the type was already inserted and terminate the
recursive call earlier.

Fixes #48280

Change-Id: I272e2f214f739fb8ec71a8628ba297477e1b7755
Reviewed-on: https://go-review.googlesource.com/c/go/+/349009
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 agoall: remove some unused code
Alexander Melentyev [Sat, 11 Sep 2021 06:59:17 +0000 (06:59 +0000)]
all: remove some unused code

Change-Id: I519b8021fa79dccc5c0ee79134547491116fc4cc
GitHub-Last-Rev: 48869f5434c1255d33c3a14714747509235c94df
GitHub-Pull-Request: golang/go#48071
Reviewed-on: https://go-review.googlesource.com/c/go/+/346231
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Keith Randall <khr@golang.org>

3 years agoruntime: change time.now to ABIInternal
Ian Lance Taylor [Fri, 30 Apr 2021 02:08:54 +0000 (19:08 -0700)]
runtime: change time.now to ABIInternal

This reduces the number of instructions executed for time.now by nine,
by eliminating the wrapper. Somehow BenchmarkNow is 0.2ns slower.
On the other hand BenchmarkNowUnixNano is 0.8ns faster.

name                                                       old time/op      new time/op      delta
AfterFunc-12                                                   66.7µs ± 4%      67.3µs ± 2%     ~     (p=0.573 n=20+18)
After-12                                                       97.6µs ± 4%      97.4µs ± 4%     ~     (p=0.758 n=20+20)
Stop-12                                                        66.7µs ±12%      64.8µs ±10%     ~     (p=0.072 n=20+20)
SimultaneousAfterFunc-12                                        109µs ± 0%       110µs ± 1%   +1.47%  (p=0.000 n=17+20)
StartStop-12                                                   31.9µs ±15%      32.7µs ±14%     ~     (p=0.799 n=20+20)
Reset-12                                                       3.67µs ± 2%      3.68µs ± 2%     ~     (p=0.132 n=20+20)
Sleep-12                                                        132µs ± 2%       133µs ± 2%   +0.70%  (p=0.035 n=20+19)
Ticker-12                                                      32.4µs ± 1%      32.3µs ± 2%     ~     (p=0.270 n=20+19)
TickerReset-12                                                 3.71µs ± 2%      3.74µs ± 2%   +0.89%  (p=0.012 n=20+20)
TickerResetNaive-12                                            65.7µs ±10%      67.2µs ±10%     ~     (p=0.174 n=20+20)
Now-12                                                         29.6ns ± 1%      29.8ns ± 0%   +0.78%  (p=0.000 n=17+17)
NowUnixNano-12                                                 31.1ns ± 1%      30.3ns ± 0%   -2.69%  (p=0.000 n=19+18)
NowUnixMilli-12                                                30.9ns ± 0%      31.1ns ± 0%   +0.90%  (p=0.000 n=18+20)
NowUnixMicro-12                                                30.9ns ± 0%      31.1ns ± 1%   +0.68%  (p=0.000 n=20+18)
Format-12                                                       304ns ± 1%       301ns ± 2%   -0.81%  (p=0.004 n=18+19)
FormatNow-12                                                    187ns ± 2%       185ns ± 2%   -0.90%  (p=0.036 n=20+18)
MarshalJSON-12                                                  267ns ± 3%       265ns ± 3%   -1.00%  (p=0.004 n=18+18)
MarshalText-12                                                  267ns ± 2%       265ns ± 3%   -0.87%  (p=0.038 n=19+20)
Parse-12                                                        150ns ± 1%       149ns ± 1%   -0.83%  (p=0.000 n=18+20)
ParseDuration-12                                               79.6ns ± 0%      80.1ns ± 1%   +0.61%  (p=0.000 n=20+20)
Hour-12                                                        4.42ns ± 1%      4.45ns ± 0%   +0.83%  (p=0.000 n=20+20)
Second-12                                                      4.42ns ± 0%      4.42ns ± 1%     ~     (p=0.075 n=18+20)
Year-12                                                        11.1ns ± 1%      11.1ns ± 1%     ~     (p=0.489 n=20+19)
Day-12                                                         14.8ns ± 1%      14.8ns ± 0%     ~     (p=0.616 n=20+18)
ISOWeek-12                                                     17.2ns ± 1%      17.2ns ± 0%     ~     (p=0.179 n=20+19)

name                                                       old avg-late-ns  new avg-late-ns  delta
ParallelTimerLatency-12                                          380k ± 4%        379k ± 3%     ~     (p=0.879 n=20+19)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=1-12         137k ± 3%        137k ± 2%     ~     (p=0.261 n=19+18)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=2-12         106k ±16%         95k ± 8%   -9.76%  (p=0.003 n=19+20)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=3-12        88.6k ±22%       74.6k ± 3%  -15.78%  (p=0.000 n=19+20)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=4-12        76.1k ±18%       70.8k ± 5%   -7.04%  (p=0.020 n=20+20)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=5-12        67.3k ±27%       65.6k ±13%     ~     (p=0.211 n=16+18)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=6-12        59.5k ±24%       57.3k ±32%     ~     (p=0.607 n=19+20)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=7-12        41.8k ±34%       46.2k ±33%  +10.54%  (p=0.039 n=17+20)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=8-12        57.5k ±37%       65.6k ±46%     ~     (p=0.283 n=17+20)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=9-12         118k ±60%        136k ±59%     ~     (p=0.169 n=19+18)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=10-12      3.66M ±236%       2.55M ±36%     ~     (p=0.158 n=16+20)
StaggeredTickerLatency/work-dur=2ms/tickers-per-P=1-12          81.7k ± 4%       80.7k ± 5%     ~     (p=0.107 n=20+19)

name                                                       old max-late-ns  new max-late-ns  delta
ParallelTimerLatency-12                                        5.88M ±124%      7.28M ±183%     ~     (p=0.640 n=20+20)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=1-12         384k ±17%        371k ±11%     ~     (p=0.540 n=17+17)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=2-12        503k ±180%        373k ±19%     ~     (p=0.057 n=17+18)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=3-12        519k ±129%        340k ±17%  -34.47%  (p=0.000 n=18+19)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=4-12        491k ±141%        341k ±26%  -30.52%  (p=0.015 n=18+17)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=5-12        457k ±123%        405k ±48%     ~     (p=0.786 n=17+17)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=6-12         491k ±85%        502k ±74%     ~     (p=0.916 n=18+19)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=7-12        572k ±100%        574k ±65%     ~     (p=0.858 n=18+17)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=8-12       1.95M ±205%      1.65M ±155%     ~     (p=0.641 n=18+19)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=9-12       7.77M ±104%      8.72M ±103%     ~     (p=0.512 n=20+20)
StaggeredTickerLatency/work-dur=300µs/tickers-per-P=10-12      29.5M ±187%       18.5M ±43%     ~     (p=0.186 n=18+20)
StaggeredTickerLatency/work-dur=2ms/tickers-per-P=1-12           981k ±14%       1033k ±12%   +5.30%  (p=0.048 n=20+18)

Change-Id: Ie794a932a929b46053a6c3020b67d640b98d2335
Reviewed-on: https://go-review.googlesource.com/c/go/+/315369
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
3 years agoreflect: use Value.Len instead of conversion to slice header
Nevkontakte [Mon, 13 Sep 2021 22:40:30 +0000 (22:40 +0000)]
reflect: use Value.Len instead of conversion to slice header

This change is functionally equivalent, but reduces reliance on unsafe
features. This would allow GopherJS to avoid an additional patch to the
standard library we'd have to maintain in order to remain compatible
with Go 1.17+.

Change-Id: I4f113db0c572ec0b81ebfecf5a137145f6c8c41d
GitHub-Last-Rev: 94ebb393bac93579b6455555822691c0d69e2d42
GitHub-Pull-Request: golang/go#48346
Reviewed-on: https://go-review.googlesource.com/c/go/+/349469
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agospec: fix prose about terminating statements
Robert Griesemer [Sat, 11 Sep 2021 00:06:43 +0000 (17:06 -0700)]
spec: fix prose about terminating statements

CL 85215 added prose to provide some minimal intuition for the
definition of a "terminating statement". While the original definition
was perfectly fine, the added prose was actually incorrect: If the
terminating statement is a goto, it might jump to a labeled statement
following that goto in the same block (it could be the very next
statement), and thus a terminating statement does not in fact
"prevent execution of all statements that lexically appear after
it in the same block".

Rather than explaining the special case for gotos with targets that
are lexically following the goto in the same block, this CL opts for
a simpler approach.

Thanks to @3bodar (Github) for finding this.

Fixes #48323.

Change-Id: I8031346250341d038938a1ce6a75d3e687d32c37
Reviewed-on: https://go-review.googlesource.com/c/go/+/349172
Trust: Robert Griesemer <gri@golang.org>
Trust: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocmd/compile: save the note of fields when translating struct
korzhao [Fri, 10 Sep 2021 17:14:54 +0000 (01:14 +0800)]
cmd/compile: save the note of fields when translating struct

Fixes #48317

Change-Id: I756ae6253022870071004332dd8f49169307f7e6
Reviewed-on: https://go-review.googlesource.com/c/go/+/349013
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Trust: Dan Scales <danscales@google.com>

3 years agocmd/go: add missing parenthesis in a call to "PrintVersion"
Hossein Zolfi [Mon, 13 Sep 2021 19:16:28 +0000 (19:16 +0000)]
cmd/go: add missing parenthesis in a call to "PrintVersion"

For #45713

Change-Id: I16e548e6c10e58da815d08897f4ba5d71eeb17e4
GitHub-Last-Rev: 4a0c5d0cdaba94e5950effdcb0ef6b736c9556d1
GitHub-Pull-Request: golang/go#48360
Reviewed-on: https://go-review.googlesource.com/c/go/+/349599
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Trust: Bryan C. Mills <bcmills@google.com>
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/link/internal/ld: re-enable DWARF tests on solaris/illumos
Tobias Klauser [Mon, 13 Sep 2021 15:26:03 +0000 (17:26 +0200)]
cmd/link/internal/ld: re-enable DWARF tests on solaris/illumos

It looks like these are fixed on current tip after CL 84655
marked them to be skipped.

Fixes #23168

Change-Id: I0020e6da1042f723eb54186ef0fe925df5326230
Reviewed-on: https://go-review.googlesource.com/c/go/+/349250
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoreflect: add a floating point section to DeepEqual tests
Josh Bleecher Snyder [Mon, 13 Sep 2021 19:49:08 +0000 (12:49 -0700)]
reflect: add a floating point section to DeepEqual tests

The floating point tests were all added into the Inequalities section,
instead of separated into Equalities vs Inequalities.

Rather than separate them, add a new floating point section.

Change-Id: I3713a5aff5850dcc0caf68a754633d695a03ded9
Reviewed-on: https://go-review.googlesource.com/c/go/+/349612
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Joe Tsai <joetsai@digital-static.net>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Joe Tsai <joetsai@digital-static.net>
Reviewed-by: Joe Tsai <joetsai@digital-static.net>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agoreflect: add fast paths for common, simple Kinds to DeepEqual
Josh Bleecher Snyder [Sat, 8 May 2021 01:39:36 +0000 (18:39 -0700)]
reflect: add fast paths for common, simple Kinds to DeepEqual

Though the normal equality check suffices, it allocates.
Handle common, simple kinds without allocating.

For some real world types compared with DeepEqual
in the Tailscale code base, the impact of these changes:

name             old time/op    new time/op    delta
HostInfoEqual-8    25.9µs ± 0%    14.4µs ± 0%  -44.32%  (p=0.008 n=5+5)

name             old alloc/op   new alloc/op   delta
HostInfoEqual-8    18.8kB ± 0%    12.6kB ± 0%  -32.58%  (p=0.008 n=5+5)

name             old allocs/op  new allocs/op  delta
HostInfoEqual-8       548 ± 0%        90 ± 0%  -83.58%  (p=0.008 n=5+5)

For the benchmarks added in this commit:

name                      old time/op    new time/op    delta
DeepEqual/int8-8            40.1ns ± 4%    31.3ns ± 4%   -21.79%  (p=0.000 n=19+20)
DeepEqual/[]int8-8           169ns ± 1%     123ns ± 1%   -27.05%  (p=0.000 n=17+18)
DeepEqual/int16-8           39.9ns ± 2%    30.8ns ± 4%   -22.81%  (p=0.000 n=17+20)
DeepEqual/[]int16-8          172ns ± 0%     122ns ± 1%   -28.91%  (p=0.000 n=17+18)
DeepEqual/int32-8           40.4ns ± 3%    31.1ns ± 4%   -23.07%  (p=0.000 n=20+20)
DeepEqual/[]int32-8          180ns ± 1%     124ns ± 1%   -31.06%  (p=0.000 n=20+18)
DeepEqual/int64-8           40.1ns ± 2%    31.4ns ± 4%   -21.82%  (p=0.000 n=20+20)
DeepEqual/[]int64-8          180ns ± 1%     124ns ± 1%   -31.47%  (p=0.000 n=16+19)
DeepEqual/int-8             40.1ns ± 4%    30.9ns ± 3%   -22.88%  (p=0.000 n=20+18)
DeepEqual/[]int-8            180ns ± 0%     123ns ± 2%   -31.59%  (p=0.000 n=19+20)
DeepEqual/uint8-8           39.8ns ± 3%    31.9ns ± 5%   -19.72%  (p=0.000 n=20+20)
DeepEqual/[]uint8-8          168ns ± 1%     114ns ± 1%   -32.48%  (p=0.000 n=18+19)
DeepEqual/uint16-8          40.3ns ± 4%    31.4ns ± 6%   -22.14%  (p=0.000 n=20+20)
DeepEqual/[]uint16-8         173ns ± 1%     124ns ± 1%   -28.20%  (p=0.000 n=20+16)
DeepEqual/uint32-8          40.1ns ± 3%    30.7ns ± 3%   -23.48%  (p=0.000 n=20+20)
DeepEqual/[]uint32-8         180ns ± 1%     123ns ± 1%   -31.56%  (p=0.000 n=20+18)
DeepEqual/uint64-8          40.0ns ± 4%    31.3ns ± 4%   -21.80%  (p=0.000 n=19+19)
DeepEqual/[]uint64-8         180ns ± 1%     124ns ± 0%   -31.45%  (p=0.000 n=18+18)
DeepEqual/uint-8            39.8ns ± 3%    31.1ns ± 4%   -21.95%  (p=0.000 n=19+20)
DeepEqual/[]uint-8           181ns ± 1%     122ns ± 1%   -32.33%  (p=0.000 n=17+20)
DeepEqual/uintptr-8         40.3ns ± 3%    31.2ns ± 3%   -22.66%  (p=0.000 n=20+18)
DeepEqual/[]uintptr-8        181ns ± 1%     124ns ± 1%   -31.46%  (p=0.000 n=19+16)
DeepEqual/float32-8         40.3ns ± 2%    31.2ns ± 3%   -22.52%  (p=0.000 n=16+20)
DeepEqual/[]float32-8        180ns ± 1%     122ns ± 1%   -32.18%  (p=0.000 n=17+17)
DeepEqual/float64-8         40.6ns ± 3%    30.9ns ± 5%   -23.91%  (p=0.000 n=19+20)
DeepEqual/[]float64-8        182ns ± 2%     121ns ± 1%   -33.33%  (p=0.000 n=18+20)
DeepEqual/complex64-8       43.0ns ±11%    32.1ns ± 5%   -25.33%  (p=0.000 n=20+18)
DeepEqual/[]complex64-8      182ns ± 1%     122ns ± 2%   -32.60%  (p=0.000 n=18+19)
DeepEqual/complex128-8      42.4ns ± 4%    34.2ns ± 3%   -19.35%  (p=0.000 n=20+19)
DeepEqual/[]complex128-8     197ns ± 1%     122ns ± 1%   -38.01%  (p=0.000 n=19+19)
DeepEqual/bool-8            40.3ns ± 3%    32.9ns ± 5%   -18.33%  (p=0.000 n=20+20)
DeepEqual/[]bool-8           169ns ± 1%     124ns ± 1%   -26.90%  (p=0.000 n=18+19)
DeepEqual/string-8          41.4ns ± 3%    33.7ns ± 5%   -18.50%  (p=0.000 n=19+20)
DeepEqual/[]string-8         216ns ± 0%     128ns ± 1%   -41.05%  (p=0.000 n=19+17)
DeepEqual/[]uint8#01-8       507ns ± 1%     112ns ± 2%   -77.92%  (p=0.000 n=20+20)
DeepEqual/[][]uint8-8        613ns ± 1%     210ns ± 1%   -65.76%  (p=0.000 n=18+19)
DeepEqual/[6]uint8-8         228ns ± 1%     162ns ± 1%   -29.00%  (p=0.000 n=20+19)
DeepEqual/[][6]uint8-8       546ns ± 2%     269ns ± 1%   -50.72%  (p=0.000 n=20+19)

name                      old alloc/op   new alloc/op   delta
DeepEqual/int8-8             0.00B          0.00B           ~     (all equal)
DeepEqual/[]int8-8           2.00B ± 0%     0.00B       -100.00%  (p=0.000 n=20+20)
DeepEqual/int16-8            0.00B          0.00B           ~     (all equal)
DeepEqual/[]int16-8          4.00B ± 0%     0.00B       -100.00%  (p=0.000 n=20+20)
DeepEqual/int32-8            0.00B          0.00B           ~     (all equal)
DeepEqual/[]int32-8          8.00B ± 0%     0.00B       -100.00%  (p=0.000 n=20+20)
DeepEqual/int64-8            0.00B          0.00B           ~     (all equal)
DeepEqual/[]int64-8          16.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/int-8              0.00B          0.00B           ~     (all equal)
DeepEqual/[]int-8            16.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint8-8            0.00B          0.00B           ~     (all equal)
DeepEqual/[]uint8-8          2.00B ± 0%     0.00B       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint16-8           0.00B          0.00B           ~     (all equal)
DeepEqual/[]uint16-8         4.00B ± 0%     0.00B       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint32-8           0.00B          0.00B           ~     (all equal)
DeepEqual/[]uint32-8         8.00B ± 0%     0.00B       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint64-8           0.00B          0.00B           ~     (all equal)
DeepEqual/[]uint64-8         16.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint-8             0.00B          0.00B           ~     (all equal)
DeepEqual/[]uint-8           16.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/uintptr-8          0.00B          0.00B           ~     (all equal)
DeepEqual/[]uintptr-8        16.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/float32-8          0.00B          0.00B           ~     (all equal)
DeepEqual/[]float32-8        8.00B ± 0%     0.00B       -100.00%  (p=0.000 n=20+20)
DeepEqual/float64-8          0.00B          0.00B           ~     (all equal)
DeepEqual/[]float64-8        16.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/complex64-8        0.00B          0.00B           ~     (all equal)
DeepEqual/[]complex64-8      16.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/complex128-8       0.00B          0.00B           ~     (all equal)
DeepEqual/[]complex128-8     32.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/bool-8             0.00B          0.00B           ~     (all equal)
DeepEqual/[]bool-8           2.00B ± 0%     0.00B       -100.00%  (p=0.000 n=20+20)
DeepEqual/string-8           0.00B          0.00B           ~     (all equal)
DeepEqual/[]string-8         32.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/[]uint8#01-8       12.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/[][]uint8-8        12.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)
DeepEqual/[6]uint8-8         0.00B          0.00B           ~     (all equal)
DeepEqual/[][6]uint8-8       12.0B ± 0%      0.0B       -100.00%  (p=0.000 n=20+20)

name                      old allocs/op  new allocs/op  delta
DeepEqual/int8-8              0.00           0.00           ~     (all equal)
DeepEqual/[]int8-8            2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/int16-8             0.00           0.00           ~     (all equal)
DeepEqual/[]int16-8           2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/int32-8             0.00           0.00           ~     (all equal)
DeepEqual/[]int32-8           2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/int64-8             0.00           0.00           ~     (all equal)
DeepEqual/[]int64-8           2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/int-8               0.00           0.00           ~     (all equal)
DeepEqual/[]int-8             2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint8-8             0.00           0.00           ~     (all equal)
DeepEqual/[]uint8-8           2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint16-8            0.00           0.00           ~     (all equal)
DeepEqual/[]uint16-8          2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint32-8            0.00           0.00           ~     (all equal)
DeepEqual/[]uint32-8          2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint64-8            0.00           0.00           ~     (all equal)
DeepEqual/[]uint64-8          2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/uint-8              0.00           0.00           ~     (all equal)
DeepEqual/[]uint-8            2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/uintptr-8           0.00           0.00           ~     (all equal)
DeepEqual/[]uintptr-8         2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/float32-8           0.00           0.00           ~     (all equal)
DeepEqual/[]float32-8         2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/float64-8           0.00           0.00           ~     (all equal)
DeepEqual/[]float64-8         2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/complex64-8         0.00           0.00           ~     (all equal)
DeepEqual/[]complex64-8       2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/complex128-8        0.00           0.00           ~     (all equal)
DeepEqual/[]complex128-8      2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/bool-8              0.00           0.00           ~     (all equal)
DeepEqual/[]bool-8            2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/string-8            0.00           0.00           ~     (all equal)
DeepEqual/[]string-8          2.00 ± 0%      0.00       -100.00%  (p=0.000 n=20+20)
DeepEqual/[]uint8#01-8        12.0 ± 0%       0.0       -100.00%  (p=0.000 n=20+20)
DeepEqual/[][]uint8-8         12.0 ± 0%       0.0       -100.00%  (p=0.000 n=20+20)
DeepEqual/[6]uint8-8          0.00           0.00           ~     (all equal)
DeepEqual/[][6]uint8-8        12.0 ± 0%       0.0       -100.00%  (p=0.000 n=20+20)

Change-Id: Ic21f0e2305f2cf5e6674c81b9ca609120b3006d9
Reviewed-on: https://go-review.googlesource.com/c/go/+/318169
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Joe Tsai <joetsai@digital-static.net>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Joe Tsai <joetsai@digital-static.net>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Joe Tsai <joetsai@digital-static.net>
3 years agoreflect: fix _faststr optimization
Keith Randall [Mon, 13 Sep 2021 16:17:22 +0000 (09:17 -0700)]
reflect: fix _faststr optimization

CL 345486 introduced an optimization to reflect's map accesses
which is not quite correct. We can't use the optimized code if the
value type is >128 bytes.

See cmd/compile/internal/walk/walk.go:mapfast

Fixes #48357

Change-Id: I8e3c7858693083dd4393a8de48ca5fa47bab66f2
Reviewed-on: https://go-review.googlesource.com/c/go/+/349593
Trust: Keith Randall <khr@golang.org>
Trust: Joe Tsai <joetsai@digital-static.net>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Trust: Martin Möhrmann <martin@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Run-TryBot: Joe Tsai <joetsai@digital-static.net>
Reviewed-by: Joe Tsai <joetsai@digital-static.net>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Martin Möhrmann <martin@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/go: add test to check for a potential workspace loading issue
Michael Matloob [Wed, 8 Sep 2021 21:28:09 +0000 (17:28 -0400)]
cmd/go: add test to check for a potential workspace loading issue

This test checks that we load the same graph regardless of the path to
the requested module in the workspace.

We currently don't. This will be fixed in a future change that redoes
workspace mode's usage of the Requirements structure.

For #45713

Change-Id: Id02cbb60a38619d840dbf1e70173ce853c0c167a
Reviewed-on: https://go-review.googlesource.com/c/go/+/348649
Trust: Michael Matloob <matloob@golang.org>
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agostrings: add Clone function
Martin Möhrmann [Sat, 28 Aug 2021 15:54:10 +0000 (17:54 +0200)]
strings: add Clone function

The new strings.Clone function copies the input string
without the returned cloned string referencing the
input strings memory.

goarch: amd64
cpu: Intel(R) Core(TM) i5-1038NG7 CPU @ 2.00GHz

name     time/op
Clone-8  24.2ns ± 2%

name     alloc/op
Clone-8   48.0B ± 0%

name     allocs/op
Clone-8    1.00 ± 0%

Update #45038
Fixes #40200

Change-Id: Id9116c21c14328ec3931ef9a67a2e4f30ff301f9
Reviewed-on: https://go-review.googlesource.com/c/go/+/345849
Trust: Martin Möhrmann <martin@golang.org>
Run-TryBot: Martin Möhrmann <martin@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Joe Tsai <joetsai@digital-static.net>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocmd/link: minor code cleanup in dwarf gen
Than McIntosh [Thu, 1 Jul 2021 13:27:46 +0000 (09:27 -0400)]
cmd/link: minor code cleanup in dwarf gen

Minor code cleanup to get rid of a few unused parameters and return
values in the linker's dwarf generation code. No functional changes.

Change-Id: I1a68ebe0f08d8d32ca7adfdd2fb9db573a4fd5f5
Reviewed-on: https://go-review.googlesource.com/c/go/+/332070
Trust: Than McIntosh <thanm@google.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/go: document 'go install cmd@version' ignores vendor directories
Jay Conrod [Mon, 13 Sep 2021 16:16:39 +0000 (09:16 -0700)]
cmd/go: document 'go install cmd@version' ignores vendor directories

For #48332

Change-Id: I708eb3e8f3f386f03210b7117d9ab8b0be2125bb
Reviewed-on: https://go-review.googlesource.com/c/go/+/349591
Trust: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agogo/types: remove some unnecessary loading/expansion of Named types
Robert Findley [Thu, 9 Sep 2021 22:41:30 +0000 (18:41 -0400)]
go/types: remove some unnecessary loading/expansion of Named types

For Identical an u.nify, only type arguments and pointer identity is
needed.

Change-Id: Id4018d2a53044fa20fd26d28890f28b37b6d6d70
Reviewed-on: https://go-review.googlesource.com/c/go/+/349409
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 agobufio: add Writer.AvailableBuffer
Joe Tsai [Wed, 4 Aug 2021 08:22:45 +0000 (01:22 -0700)]
bufio: add Writer.AvailableBuffer

This adds a new Writer.AvailableBuffer method that returns
an empty buffer with a possibly non-empty capacity for use
with append-like APIs.

The typical usage pattern is something like:
b := bw.AvailableBuffer()
b = appendValue(b, v)
bw.Write(b)

It allows logic combining append-like APIs with bufio.Writer to avoid
needing to allocate and manage buffers themselves and allows the
append-like APIs to directly write into the buffer for a bufio.Writer.

Fixes #47527

Change-Id: I9cd169f3f8e8c7cd40818caf3daf1944c826fc66
Reviewed-on: https://go-review.googlesource.com/c/go/+/345569
Trust: Joe Tsai <joetsai@digital-static.net>
Run-TryBot: Joe Tsai <joetsai@digital-static.net>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agoreflect: optimize for maps with string keys
Joe Tsai [Fri, 27 Aug 2021 02:13:22 +0000 (19:13 -0700)]
reflect: optimize for maps with string keys

Over 80% of all Go map types use a string as the key.
The Go runtime already has a specialized implementation for such maps
in runtime/map_faststr.go. However, the Go reflection implementation
has not historically made use of that implementation.

This CL plumbs the appropriate logic to be accessible from Go reflection
so that it can benefit as well.

    name                            old time/op    new time/op    delta
    Map/StringKeys/MapIndex-4       4.65us ± 5%    2.95us ± 3%  -36.50%  (p=0.016 n=4+5)
    Map/StringKeys/SetMapIndex-4    7.47us ± 5%    5.27us ± 2%  -29.40%  (p=0.008 n=5+5)
    Map/Uint64Keys/MapIndex-4       3.79us ± 3%    3.75us ± 2%     ~     (p=0.548 n=5+5)
    Map/Uint64Keys/SetMapIndex-4    6.13us ± 3%    6.09us ± 1%     ~     (p=0.746 n=5+5)

Change-Id: I5495d48948d8caf2d004a03ae1820ab5f8729670
Reviewed-on: https://go-review.googlesource.com/c/go/+/345486
Trust: Joe Tsai <joetsai@digital-static.net>
Run-TryBot: Joe Tsai <joetsai@digital-static.net>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
3 years agobufio: make Reader.Reset and Writer.Reset work on the zero value
Joe Tsai [Wed, 4 Aug 2021 07:57:07 +0000 (00:57 -0700)]
bufio: make Reader.Reset and Writer.Reset work on the zero value

For batch allocation reasons, it would be useful to nest a
bufio.Reader or bufio.Writer in a struct as a value,
rather than a pointer. When the Reset method is called,
have it use the default buffer size if the buffer is nil.

Fixes #45374

Change-Id: I80df18a13575431428a42ed150a1579de1282637
Reviewed-on: https://go-review.googlesource.com/c/go/+/345570
Trust: Joe Tsai <joetsai@digital-static.net>
Run-TryBot: Joe Tsai <joetsai@digital-static.net>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agodoc/asm: fix HTML markup
WANG Xuerui [Sun, 1 Aug 2021 16:56:25 +0000 (00:56 +0800)]
doc/asm: fix HTML markup

Change-Id: I33bde4835d3b83fafd55beea483f6236c4c62840
Reviewed-on: https://go-review.googlesource.com/c/go/+/338990
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Than McIntosh <thanm@google.com>

3 years agodebug/elf: retain original error message when getSymbols fails.
Florin Papa [Fri, 7 May 2021 00:17:59 +0000 (17:17 -0700)]
debug/elf: retain original error message when getSymbols fails.

The original error is currently discarded, and that makes it difficult
to know what failed, in case we want to retry only certain errors.

Change-Id: Id7e927ec242464249c4dfa5cda0f264adef3c898
Reviewed-on: https://go-review.googlesource.com/c/go/+/317851
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Than McIntosh <thanm@google.com>

3 years agotime: reference -tags=timetzdata in testing panic
Josh Bleecher Snyder [Wed, 14 Jul 2021 20:21:11 +0000 (13:21 -0700)]
time: reference -tags=timetzdata in testing panic

This will spare anyone who hits it having to search for the workaround.

Change-Id: Iff0d449212f2675ac78e30ae5ffc8efb4d924088
Reviewed-on: https://go-review.googlesource.com/c/go/+/334611
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 agotesting: increase alternation precedence
Ethan Reesor [Fri, 20 Aug 2021 23:57:45 +0000 (18:57 -0500)]
testing: increase alternation precedence

Updates handling of go test flags -run and -bench to give alternation
precendence over the / delimiter. Currently, `A/B|C/D` is effectively
`A/(B|C)/D` - with this change, it changes to effectively `(A/B)|(C/D)`.

Fixes #39904

Change-Id: Iebe5efd8d91c72eed6351bd63b4689b0fcb0ed0f
Reviewed-on: https://go-review.googlesource.com/c/go/+/343883
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Than McIntosh <thanm@google.com>

3 years agocmd/compile/internal/types2: better error message for invalid array decls
Robert Griesemer [Thu, 9 Sep 2021 05:08:14 +0000 (22:08 -0700)]
cmd/compile/internal/types2: better error message for invalid array decls

Fixes #43527.

Change-Id: I988a4d49f2f54b4b1741688fb52a55bf313d39e1
Reviewed-on: https://go-review.googlesource.com/c/go/+/348731
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/syntax: correct follow token for type parameter lists
Robert Griesemer [Thu, 9 Sep 2021 22:43:19 +0000 (15:43 -0700)]
cmd/compile/internal/syntax: correct follow token for type parameter lists

When parsing a type parameter declaration, parts of the code still
expected a ) as closing token. Use the correct follow token ) or ]
depending on parameter list kind.

Also, consistently use tokstring (not tok.String()) for user-facing
(error) messages.

Follow-up on comment in CL 348730.

For #43527.

Change-Id: Ib1d4feb526771a1668a54c3bb7a671f6c8a65940
Reviewed-on: https://go-review.googlesource.com/c/go/+/348742
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal: better AST line highlight in ssa.html
Zheng Xu [Wed, 1 Sep 2021 05:48:48 +0000 (13:48 +0800)]
cmd/compile/internal: better AST line highlight in ssa.html

We tend to set div class with the line number in HTML AST nodes. So that
the AST nodes can be highlighted with corresponding source and ssa ir.

The pure AST text dump is created first. And then it is parsed and
written to the HTML file.

CL 275785 changed the format of the line information in AST node dump,
which makes the HTMLWriter fail to parse the line information.

This CL updates the code in HTMLWriter to align with the format of AST
node dump.

Fix #48133

Change-Id: I2b56fc5e3e9771456d91f22caf23a427c235eb12
Reviewed-on: https://go-review.googlesource.com/c/go/+/347269
Reviewed-by: Cherry Mui <cherryyz@google.com>
Trust: Than McIntosh <thanm@google.com>

3 years agoembed: guarantee the returned file of FS.Open implements io.Seeker
Hajime Hoshi [Mon, 26 Apr 2021 13:32:21 +0000 (22:32 +0900)]
embed: guarantee the returned file of FS.Open implements io.Seeker

Fixes golang/go#45745

Change-Id: Ib49a9605a38074f544a5d28116862e191cea8c0f
Reviewed-on: https://go-review.googlesource.com/c/go/+/313352
Run-TryBot: Hajime Hoshi <hajimehoshi@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Than McIntosh <thanm@google.com>

3 years agocmd/compile: add support for Abs and Copysign intrinsics on riscv64
Michael Munday [Thu, 9 Sep 2021 22:47:14 +0000 (23:47 +0100)]
cmd/compile: add support for Abs and Copysign intrinsics on riscv64

Also, add the FABSS and FABSD pseudo instructions to the assembler.
The compiler could use FSGNJX[SD] directly but there doesn't seem
to be much advantage to doing so and the pseudo instructions are
easier to understand.

Change-Id: Ie8825b8aa8773c69cc4f07a32ef04abf4061d80d
Reviewed-on: https://go-review.googlesource.com/c/go/+/348989
Trust: Michael Munday <mike.munday@lowrisc.org>
Run-TryBot: Michael Munday <mike.munday@lowrisc.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Joel Sing <joel@sing.id.au>
3 years agocmd/compile: simiplify arm64 bitfield optimizations
fanzha02 [Wed, 21 Oct 2020 10:51:42 +0000 (18:51 +0800)]
cmd/compile: simiplify arm64 bitfield optimizations

In some rewrite rules for arm64 bitfield optimizations, the
bitfield lsb value and the bitfield width value are related
to datasize, some of them use datasize directly to check the
bitfield lsb value is valid, to get the bitfiled width value,
but some of them call isARM64BFMask() and arm64BFWidth()
functions. In order to be consistent, this patch changes them
all to use datasize.

Besides, this patch sorts the codegen test cases.

Run the "toolstash-check -all" command and find one inconsistent code
is as the following.

new:    src/math/fma.go:104      BEQ 247
master: src/math/fma.go:104      BEQ 248

The above inconsistence is due to this patch changing the range of the
field lsb value in "UBFIZ" optimization rules from "lc+(32|16|8)<64" to
"lc<64", so that the following code is generated as "UBFIZ". The logical
of changed code is still correct.

The code of src/math/fma.go:160:
  const uvinf = 0x7FF0000000000000
  func FMA(a, b uint32) float64 {
   ps := a+b
   return Float64frombits(uint64(ps)<<63 | uvinf)
  }

The new assembly code:
  TEXT    "".FMA(SB), LEAF|NOFRAME|ABIInternal, $0-16
  MOVWU   "".a(FP), R0
  MOVWU   "".b+4(FP), R1
  ADD     R1, R0, R0
  UBFIZ   $63, R0, $1, R0
  ORR     $9218868437227405312, R0, R0
  MOVD    R0, "".~r2+8(FP)
  RET     (R30)

The master assembly code:
  TEXT    "".FMA(SB), LEAF|NOFRAME|ABIInternal, $0-16
  MOVWU   "".a(FP), R0
  MOVWU   "".b+4(FP), R1
  ADD     R1, R0, R0
  MOVWU   R0, R0
  LSL     $63, R0, R0
  ORR     $9218868437227405312, R0, R0
  MOVD    R0, "".~r2+8(FP)
  RET     (R30)

Change-Id: I9061104adfdfd3384d0525327ae1e5c8b0df5c35
Reviewed-on: https://go-review.googlesource.com/c/go/+/265038
Trust: fannie zhang <Fannie.Zhang@arm.com>
Run-TryBot: fannie zhang <Fannie.Zhang@arm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/compile: fix test case for unified IR (fix build)
Robert Griesemer [Thu, 9 Sep 2021 22:55:05 +0000 (15:55 -0700)]
cmd/compile: fix test case for unified IR (fix build)

For #48301.

Change-Id: Ie5f57dcce86773c06c5140abf13a6cfff79eb323
Reviewed-on: https://go-review.googlesource.com/c/go/+/348743
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile: don't crash while reporting invalid alias cycle
Robert Griesemer [Thu, 9 Sep 2021 21:42:42 +0000 (14:42 -0700)]
cmd/compile: don't crash while reporting invalid alias cycle

Add a missing nil check in the formatting code for expression
nodes. Matches the nil checks in the same code.

Fixes #48301.

Change-Id: Ia9bfd3535254a94996ee190b544d95e15433d252
Reviewed-on: https://go-review.googlesource.com/c/go/+/348740
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
3 years agocmd/cgo, runtime/cgo: avoid GCC/clang conversion warnings
Ian Lance Taylor [Thu, 9 Sep 2021 21:04:43 +0000 (14:04 -0700)]
cmd/cgo, runtime/cgo: avoid GCC/clang conversion warnings

Add explicit conversions to avoid warnings from -Wsign-conversion and
-Wshorten-64-to-32. Also avoid runtime errors from -fsanitize=undefined.

Fixes #48121

Change-Id: I29dc8d976884fc42826392c10f1e1759bb1a3989
Reviewed-on: https://go-review.googlesource.com/c/go/+/348739
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/syntax: better error message for missing type constraint
Robert Griesemer [Thu, 9 Sep 2021 04:30:01 +0000 (21:30 -0700)]
cmd/compile/internal/syntax: better error message for missing type constraint

For #43527.

Change-Id: I8c706e68572286d5675383eb2dfd75b5618b646b
Reviewed-on: https://go-review.googlesource.com/c/go/+/348730
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agotime: propagate "," separator for fractional seconds into Format
korzhao [Sat, 28 Aug 2021 21:21:17 +0000 (05:21 +0800)]
time: propagate "," separator for fractional seconds into Format

In CL 300996 that fixed issue #6189, we made Parse recognize
"," as a separator for fractional seconds.
However, we didn't modify Format to propagate the separator
verbatim from Parse. Without this change, we break prior
functionality that relied on a comma being used in Format.

Fixes #48037

Change-Id: I6565a25e8657ca3747a58b25acba58f27cdcddc0
Reviewed-on: https://go-review.googlesource.com/c/go/+/345438
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
Trust: Cherry Mui <cherryyz@google.com>

3 years agocmd/compile: fix implement for closure in a global assignment
korzhao [Thu, 9 Sep 2021 13:51:43 +0000 (21:51 +0800)]
cmd/compile: fix implement for closure in a global assignment

If closure in a global assignment and has a method receiver.
We should assign receiver as a global variable, not a local variable.

Fixes #48225

Change-Id: I8f65dd6e8baf66a5eff24028d28ad0a594091add
Reviewed-on: https://go-review.googlesource.com/c/go/+/348512
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Dan Scales <danscales@google.com>

3 years agocmd/link: enable internal linker in more cases for ppc64le
Paul E. Murphy [Tue, 23 Mar 2021 21:17:00 +0000 (16:17 -0500)]
cmd/link: enable internal linker in more cases for ppc64le

The internal linker is capable of linking the ppc64le linux
race detector and approved cgo packages.

Likewise, ppc64/linux and ppc64/aix do not support the race
detector. Thus, extra code to enforce external linking when
using the race detector on ppc64/ppc64le can be removed
entirely.

Fixes #21961

Change-Id: I10db14f65ee616ee3291e17409e8333e3af7d4df
Reviewed-on: https://go-review.googlesource.com/c/go/+/304459
Run-TryBot: Paul Murphy <murp@ibm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agotest: add compiler regress tests for #46461
Matthew Dempsky [Thu, 9 Sep 2021 18:17:39 +0000 (11:17 -0700)]
test: add compiler regress tests for #46461

gri@ reports that types2 now correctly handles when type parameters
recursively refer back to the parameterized type, so we might as well
add tests to exercise that. Unified IR also correctly handles
importing and exporting these types, but -G=3 currently does not.

Updates #46461.

Change-Id: I272102aa08c40c980b9aeeca9f834291dfbbcc3e
Reviewed-on: https://go-review.googlesource.com/c/go/+/348738
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 agocmd/compile: fix case where init info of OAS node is dropped
Dan Scales [Wed, 8 Sep 2021 15:41:54 +0000 (08:41 -0700)]
cmd/compile: fix case where init info of OAS node is dropped

When an OAS node is converted to an OSELRECV2 node in tcSelect(), the
possible DCL node in the Init field was being dropped, since a
completely new node was being created and the Init field was not set. I
don't expect n.Init() to be set for the ORECV case, but the code now
deals with that too.

Fixed bug in both tcSelect() and transformSelect().

Fixes #48289

Change-Id: I09918a70f7cbaa4aa9a17546169f908a8787df15
Reviewed-on: https://go-review.googlesource.com/c/go/+/348569
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
3 years agogo/types, types2: rename RParams -> RecvTypeParams
Robert Findley [Thu, 9 Sep 2021 13:26:40 +0000 (09:26 -0400)]
go/types, types2: rename RParams -> RecvTypeParams

To be consistent with CL 348376, spell out 'RecvTypeParams' in go/types
and types2 API.

Updates #47916

Change-Id: If8b3fd4274ccb944bd0ff04d7007e94e5fba61c1
Reviewed-on: https://go-review.googlesource.com/c/go/+/348810
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 agoreflect: add hooks for dealing with narrow width floats
Michael Anthony Knyszek [Fri, 3 Sep 2021 21:35:59 +0000 (21:35 +0000)]
reflect: add hooks for dealing with narrow width floats

Currently on amd64 and arm64, float32 values just live in the bottom 32
bits of the register, so reflect simply places them there in a RegArgs
for reflectcall to load them. This works fine because both of these
platforms don't care what the upper 32 bits are, and have instructions
to operate on float32 values specifically that we use. In sum, the
representation of the float32 in memory is identical to that of the
representation in a register.

On other platforms, however, the representation of FP values differ
depending on whether they're in memory or in a register. For instance,
on ppc64, all floating point values get promoted to a float64 when
loaded to a register (i.e. there's only one set of FP instructions). As
another example, on riscv64, narrow-width floats in registers need to be
NaN-boxed.

What all this means is that for supporting the register ABI on these
platforms, reflect needs to do a little extra work to ensure that the
representation of FP values in a RegArgs matches the representation it
takes on in a register. For this purpose, this change abstracts away the
action of storing values into a RegArgs a little bit and adds a
platform-specific hook which currently does nothing but copy the value.

For #40724.

Change-Id: I65dcc7d86d5602a584f86026ac204564617f4c5a
Reviewed-on: https://go-review.googlesource.com/c/go/+/347566
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agonet: deprecate (net.Error).Temporary
Damien Neil [Fri, 6 Aug 2021 17:33:08 +0000 (10:33 -0700)]
net: deprecate (net.Error).Temporary

Fixes #45729.

Change-Id: I819dd0cc79fc23baac46cafd78bb80f5133c992b
Reviewed-on: https://go-review.googlesource.com/c/go/+/340261
Trust: Damien Neil <dneil@google.com>
Trust: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocmd/compile: stenciled conversions might be NOPs
Keith Randall [Thu, 9 Sep 2021 15:09:57 +0000 (08:09 -0700)]
cmd/compile: stenciled conversions might be NOPs

A generic conversion might be required for when converting T->interface{}.
When stenciled with T=interface{}, then that conversion doesn't need
to do anything.

Fixes #48276

Change-Id: Ife65d01c99fbd0895cb7eec79df9e93e752b1fa5
Reviewed-on: https://go-review.googlesource.com/c/go/+/348736
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agotest: re-enable AsmCheck tests for types2-based frontends
Matthew Dempsky [Wed, 8 Sep 2021 21:03:53 +0000 (14:03 -0700)]
test: re-enable AsmCheck tests for types2-based frontends

run.go has logic for being able to run tests with various -G flags,
but not all test types (in particular, "asmcheck" tests) support
configuring non-default -G levels. The intention was that these tests
would continue running in the default mode (at the time -G=0), but at
some point it seems like we accidentally disabled them all
together (if it ever worked correctly in the first place).

Fixes #48247.

Change-Id: I13917cb0012cbe522d29b23b888de6136872ead4
Reviewed-on: https://go-review.googlesource.com/c/go/+/348671
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Munday <mike.munday@lowrisc.org>
3 years agogo/types: reduce number of delayed functions
Robert Findley [Wed, 8 Sep 2021 23:03:37 +0000 (19:03 -0400)]
go/types: reduce number of delayed functions

This is a port of CL 348018 to go/types. It differs from that CL due to
the way that field lists are represented in go/ast.

Change-Id: Ib5a0243b44d0bf9e95d039f624c668f8c329f8fa
Reviewed-on: https://go-review.googlesource.com/c/go/+/348691
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: handle recursive type parameter constraints
Robert Findley [Wed, 8 Sep 2021 22:27:31 +0000 (18:27 -0400)]
go/types: handle recursive type parameter constraints

This is a port of CL 348090 to go/types. Notably, unlike in types2,
declareTypeParams was previously setting the default constraint to the
empty interface, not nil, because this was missed in CL 335034 (no
changes were made to declareTypeParams). This CL fixes this discrepancy.

Change-Id: I0fa54a660ba14c6cbefa81a27ab7eb193df3be20
Reviewed-on: https://go-review.googlesource.com/c/go/+/348690
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: detect constraint type inference cycles
Robert Findley [Wed, 8 Sep 2021 22:16:36 +0000 (18:16 -0400)]
go/types: detect constraint type inference cycles

This is a port of CL 347300 to go/types. The test was adjusted to match
the differing error positioning in go/types: errors are placed on the
ast.CallExpr.Fun, rather than the Lparen.

Change-Id: I3dee5d7d75bae18817cc1f04ab0d357d3a6a8198
Reviewed-on: https://go-review.googlesource.com/c/go/+/348689
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 agotest/typeparam: fix issue48094b test build
Tobias Klauser [Thu, 9 Sep 2021 08:10:54 +0000 (10:10 +0200)]
test/typeparam: fix issue48094b test build

CL 347529 broke the longtest builders due to missing -G=3 flag when
compiling the added test.

Change-Id: I73007801dd6871a8cf3554e957d247f5f56fd641
Reviewed-on: https://go-review.googlesource.com/c/go/+/348330
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agosyscall: drop fallback to pipe in Pipe on linux/arm
Tobias Klauser [Wed, 8 Sep 2021 08:41:56 +0000 (10:41 +0200)]
syscall: drop fallback to pipe in Pipe on linux/arm

Follow-up for CL 346789

The minimum required Linux kernel version for Go 1.18 will be changed to
2.6.32, see #45964. The pipe2 syscall was added in 2.6.27, so the
fallback to use pipe in Pipe on linux/arm can be removed.

For #45964

Change-Id: I8b18244ca1f849f10e90565b4fef80ce777fef69
Reviewed-on: https://go-review.googlesource.com/c/go/+/347349
Trust: Tobias Klauser <tobias.klauser@gmail.com>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
3 years agocmd/compile: fix unified IR panic when expanding nested inline function
Cuong Manh Le [Fri, 3 Sep 2021 03:49:32 +0000 (10:49 +0700)]
cmd/compile: fix unified IR panic when expanding nested inline function

When reading body of inlining function, which has another inlined
function in the body, the reader still add this inlined function to
todoBodies, which it shouldn't because the inlined function was read
already.

To fix this, introduce new flag to signal that we are done construting
all functions in todoBodies, thus the addBody shouldn't add anything
to todoBodies then.

Updates #48094

Change-Id: I45105dd518f0a7b69c6dcbaf23b957623f271203
Reviewed-on: https://go-review.googlesource.com/c/go/+/347529
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: Matthew Dempsky <mdempsky@google.com>
3 years agointernal/poll: report open fds when TestSplicePipePool fails
Ian Lance Taylor [Wed, 8 Sep 2021 23:21:33 +0000 (16:21 -0700)]
internal/poll: report open fds when TestSplicePipePool fails

For #48066

Change-Id: I1152a1c15756df35b71b27d3e7025d97da9e70b0
Reviewed-on: https://go-review.googlesource.com/c/go/+/348579
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emmanuel@orijtech.com>
3 years agocmd/compile: fix wrong instantiated type for embedded receiver
Cuong Manh Le [Wed, 8 Sep 2021 18:11:26 +0000 (01:11 +0700)]
cmd/compile: fix wrong instantiated type for embedded receiver

In case of embedded field, if the receiver was fully instantiated, we
must use its instantiated type, instead of passing the type params of
the base receiver.

Fixes #47797
Fixes #48253

Change-Id: I97613e7e669a72605137e82406f7bf5fbb629378
Reviewed-on: https://go-review.googlesource.com/c/go/+/348549
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 agocmd/compile: move checkptr alignment to SSA generation
Cuong Manh Le [Fri, 27 Aug 2021 13:07:00 +0000 (20:07 +0700)]
cmd/compile: move checkptr alignment to SSA generation

This is followup of CL 343972, moving the checkptr alignment
instrumentation during SSA generation instead of walk.

Change-Id: I29b2953e4eb8631277fe2e0f44b9d987dd7a69f9
Reviewed-on: https://go-review.googlesource.com/c/go/+/345430
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: Matthew Dempsky <mdempsky@google.com>
3 years agocmd/compile: fold handling OCONV logic to separate function
Cuong Manh Le [Sat, 28 Aug 2021 09:40:34 +0000 (16:40 +0700)]
cmd/compile: fold handling OCONV logic to separate function

So next CL can re-use the logic to perform checkptr instrumentation.

Change-Id: I3241e9c3c84da04db71fd1d4fd83cb76b2e18521
Reviewed-on: https://go-review.googlesource.com/c/go/+/345435
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: Matthew Dempsky <mdempsky@google.com>
3 years agocmd/internal/obj/arm64: add checks for incorrect use of REGTMP register
fanzha02 [Tue, 24 Aug 2021 04:46:54 +0000 (12:46 +0800)]
cmd/internal/obj/arm64: add checks for incorrect use of REGTMP register

The current assembler uses REGTMP as a temporary destination register,
when optimizing one instruction into a multi-instruction sequence. But
in some cases, when the source register is REGTMP, this behavior is
incorrect.

For example:
  ADD $0x1234567, R27, R3

The current assembler encodes it into
  MOVD $17767, R27
  MOVK $(291<<16), R27
  ADD R27, R27, R3

It is illegal to overwrite R27. This CL adds the related checks.

Add test cases.

Change-Id: I0af373d9fd23d8f067c093778dd4cc76748faf38
Reviewed-on: https://go-review.googlesource.com/c/go/+/344689
Reviewed-by: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: fannie zhang <Fannie.Zhang@arm.com>

3 years agocmd/compile: remove 'ext' fields from unified IR reader/writer types
Matthew Dempsky [Wed, 8 Sep 2021 00:24:50 +0000 (17:24 -0700)]
cmd/compile: remove 'ext' fields from unified IR reader/writer types

This is a vestigial artifact of how I initially split apart the public
and private data for objects. But now objects are split into more
parts, and it's proven easier to just keep them as separate variables.
So it's time to cleanup the initial public/private code to follow the
same approach.

Change-Id: I3976b19fb433cbe21d299d3799ec616f9e59561e
Reviewed-on: https://go-review.googlesource.com/c/go/+/348412
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
3 years agocmd/compile: simplify value coding for unified IR
Matthew Dempsky [Tue, 7 Sep 2021 20:23:08 +0000 (13:23 -0700)]
cmd/compile: simplify value coding for unified IR

In indexed export, values are always exported along with their type
and are encoded in a type-sensitive manner, because this matches how
cmd/compile handled constants internally.

However, go/types intentionally differs from this, decoupling type
from value representation. As unified IR strives to be more
go/types-centric, it makes sense to embrace this and make values a
more first-class encoding.

Change-Id: If21d849c4f610358bd776d5665469d180bcd5f6e
Reviewed-on: https://go-review.googlesource.com/c/go/+/348014
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
3 years agocmd/compile: extrapolate $GOROOT in unified IR
Matthew Dempsky [Wed, 8 Sep 2021 22:59:11 +0000 (15:59 -0700)]
cmd/compile: extrapolate $GOROOT in unified IR

This ensures that diagnostics for files within $GOROOT continue to be
reported using their full filepath, rather than the abbreviated
filepath. Notably, this is necessary for test/run.go, which has tests
that expect to see the full filepath.

Updates #48247.

Change-Id: I440e2c6dd6109ca059d81cee49e476bba805d703
Reviewed-on: https://go-review.googlesource.com/c/go/+/348670
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agogo/types, types2: add Environment to Config
Robert Griesemer [Wed, 8 Sep 2021 23:03:57 +0000 (16:03 -0700)]
go/types, types2: add Environment to Config

Port to types2 and adjust compiler accordingly.

Change-Id: I2e72b151ef834977dca64cb2e62cedcac4e46062
Reviewed-on: https://go-review.googlesource.com/c/go/+/348578
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/compile/internal/types2: spell out 'Type' in type parameter APIs
Robert Griesemer [Wed, 8 Sep 2021 22:22:20 +0000 (15:22 -0700)]
cmd/compile/internal/types2: spell out 'Type' in type parameter APIs

This is a port of CL 348376 with the necessary adjustments
in the compiler.

Change-Id: Ib11ee841b194746ff231ee493aa56bf9b3a4a67f
Reviewed-on: https://go-review.googlesource.com/c/go/+/348577
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile: add automated rewrite cycle detection
Josh Bleecher Snyder [Sun, 5 Sep 2021 02:29:08 +0000 (19:29 -0700)]
cmd/compile: add automated rewrite cycle detection

A common bug during development is to introduce rewrite rule cycles.
This is annoying because it takes a while to notice that
make.bash is a bit too slow this time, and to remember why.
And then you have to manually arrange to debug.

Make this all easier by automating it.
Detect cycles, and when we detect one, print the sequence
of rewrite rules that occur within a single cycle before crashing.

Change-Id: I8dadda13990ab925a81940d4833c9e5243368435
Reviewed-on: https://go-review.googlesource.com/c/go/+/347829
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
3 years agocmd/compile/internal/types2: temporarily pin the Checker to Interface during checking
Robert Griesemer [Wed, 8 Sep 2021 20:52:46 +0000 (13:52 -0700)]
cmd/compile/internal/types2: temporarily pin the Checker to Interface during checking

This is a clean port of CL 348371.

Change-Id: I3a61a5a8928279bc783ef16739f7607de3b6ecf3
Reviewed-on: https://go-review.googlesource.com/c/go/+/348575
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: move NewTypeParam off of Checker
Robert Griesemer [Wed, 8 Sep 2021 20:45:16 +0000 (13:45 -0700)]
cmd/compile/internal/types2: move NewTypeParam off of Checker

This is a port of CL 347561.

A comment was corrected both in types2 and go/types, and the
compiler adjusted for the updated NewTypeParameter function.

Change-Id: I4381f0dd8e43228e1d037c5d997d421b7838f905
Reviewed-on: https://go-review.googlesource.com/c/go/+/348574
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: move typeHash to environment.go
Robert Griesemer [Wed, 8 Sep 2021 20:37:34 +0000 (13:37 -0700)]
cmd/compile/internal/types2: move typeHash to environment.go

This is a clean port of CL 347560.

Change-Id: I0d56f5a818df1a66e603415d5198d909b0aef228
Reviewed-on: https://go-review.googlesource.com/c/go/+/348573
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agocmd/compile/internal/types2: implement deduplication of instances using the Environment
Robert Griesemer [Wed, 8 Sep 2021 20:30:36 +0000 (13:30 -0700)]
cmd/compile/internal/types2: implement deduplication of instances using the Environment

This is a port of CL 344390 with adjustments to names to make it
work for types2.

Change-Id: I05c33d9858f973adfbf48d8a1faaf377280f6985
Reviewed-on: https://go-review.googlesource.com/c/go/+/348572
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
3 years agogo/ast: rename MultiIndexExpr to IndexListExpr
Robert Findley [Wed, 8 Sep 2021 19:51:10 +0000 (15:51 -0400)]
go/ast: rename MultiIndexExpr to IndexListExpr

As discussed in #47781, IndexListExpr is one character shorter and has
the advantage of being next to IndexExpr in documentation.

Updates #47781

Change-Id: I709d5c1a79b4f9aebcd6445e4ab0cd6dae45bab7
Reviewed-on: https://go-review.googlesource.com/c/go/+/348609
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agocmd/go: print offending -mod value in workspace mode
Michael Matloob [Fri, 3 Sep 2021 22:11:00 +0000 (18:11 -0400)]
cmd/go: print offending -mod value in workspace mode

Workspace mode only allows -mod to be set to 'readonly'. When returning
the error for other values of -mod, specify the the value in the error
so it's easier to see what's going on when the go command is invoked
through another tool.

For #45713, #48186

Change-Id: I3ba206a71c7a20c18aeccfa48f3c9dc935a180a4
Reviewed-on: https://go-review.googlesource.com/c/go/+/347593
Trust: Michael Matloob <matloob@golang.org>
Trust: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
3 years agocmd/go: improve the creation and editing of go.work files
Michael Matloob [Fri, 3 Sep 2021 17:33:37 +0000 (13:33 -0400)]
cmd/go: improve the creation and editing of go.work files

This change changes go.work files so that directory paths are clearly
distinguished from module paths by either being rooted absolute paths or
starting with '.' or '..' path elements if they are relative paths.
go mod initwork now checks that the go.work file doesn't already exist
before creating it, and gomod initwork and gomod editwork look up the
module path corresponding to a directory and write it to the directory
directive's comment.

For #45713

Change-Id: I6983779059b7de6fc83d359280ceffb263f6b641
Reviewed-on: https://go-review.googlesource.com/c/go/+/347591
Trust: Michael Matloob <matloob@golang.org>
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
3 years agogo/types: temporarily pin the Checker to Interface during checking
Robert Findley [Wed, 8 Sep 2021 13:58:44 +0000 (09:58 -0400)]
go/types: temporarily pin the Checker to Interface during checking

While type checking expressions involving interface types, it is
possible that their type set is used before delayed actions are
processed. As a result, computeInterfaceTypeSet is called with a nil
checker, and errors in the interface type definition result in panics
(see #48234).

To avoid the panics, store a *Checker on Interface for use in between
checking of the interface type expression and processing of delayed
actions.

Fixes #48234

Change-Id: I5509bc1c01b55edac52446b9e075fbe8fcc01874
Reviewed-on: https://go-review.googlesource.com/c/go/+/348371
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: spell out 'Type' in type parameter APIs
Robert Findley [Wed, 8 Sep 2021 15:57:04 +0000 (11:57 -0400)]
go/types: spell out 'Type' in type parameter APIs

As discussed on the go/types proposal (#47916), we should spell out the
word 'Type', rather than using 'T'.

Change-Id: I5f51255eedc07fea61f909b7ecb3093a7fab765e
Reviewed-on: https://go-review.googlesource.com/c/go/+/348376
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agogo/ast: rename TParams fields to TypeParams
Robert Findley [Wed, 8 Sep 2021 15:46:58 +0000 (11:46 -0400)]
go/ast: rename TParams fields to TypeParams

As discussed in the ast proposal (#47781), there's not really a strong
reason to avoid spelling out 'Type'.

Change-Id: I0ba1bf03b112ea60509a78a89a050a302779d9d0
Reviewed-on: https://go-review.googlesource.com/c/go/+/348375
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>

3 years agogo/types: move NewTypeParam off of Checker
Robert Findley [Fri, 3 Sep 2021 15:22:18 +0000 (11:22 -0400)]
go/types: move NewTypeParam off of Checker

This aligns with the API proposal.

Change-Id: I9967a317196392ffa5ddbe5391d7aba5f6e7bad2
Reviewed-on: https://go-review.googlesource.com/c/go/+/347561
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>