]> Cypherpunks repositories - gostls13.git/log
gostls13.git
5 years ago[dev.link] cmd/link: stop overwriting symbol types in loadlibfull
Cherry Zhang [Sat, 25 Apr 2020 21:27:34 +0000 (17:27 -0400)]
[dev.link] cmd/link: stop overwriting symbol types in loadlibfull

symGroupType was needed for dodata. Now that we have converted
dodata to using the loader, stop overwriting it.

Change-Id: Ie94109c0b35dd2f71a19ebb38f8cf20b6a37c624
Reviewed-on: https://go-review.googlesource.com/c/go/+/229994
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years ago[dev.link] cmd/link: delete old dodata
Cherry Zhang [Mon, 27 Apr 2020 20:46:23 +0000 (16:46 -0400)]
[dev.link] cmd/link: delete old dodata

Change-Id: I569bbee235630baad3c35ca0c6598b8bd059307a
Reviewed-on: https://go-review.googlesource.com/c/go/+/230311
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years ago[dev.link] cmd/link: accept more types in Xcoffadddynrel
Cherry Zhang [Mon, 27 Apr 2020 20:01:24 +0000 (16:01 -0400)]
[dev.link] cmd/link: accept more types in Xcoffadddynrel

In dodata we overwrite symbol types to SDATA. Now we'll stop
doing that, so accept more symbol types here. This is basically
a list of all writeable types handled in dodata that could appear
in XCOFF.

Change-Id: Iee35369162f5acd59806a3f0e6c8d3682620067e
Reviewed-on: https://go-review.googlesource.com/c/go/+/230310
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years ago[dev.link] cmd/link: support new dodata for PPC64
Than McIntosh [Fri, 24 Apr 2020 19:32:46 +0000 (15:32 -0400)]
[dev.link] cmd/link: support new dodata for PPC64

Add linux/{ppc64,ppc64le} and aix/ppc64 arch support for the new
dodata() phase.

This completes the picture in terms of architecture support for the
new dodata(), but to be safe this patch leaves the command line flag
in place there are problems on the builders (especially given that we
have a dead aix-ppc64 builder).

Change-Id: I78da615c3b540d8925ed7b3226e199280eb7451d
Reviewed-on: https://go-review.googlesource.com/c/go/+/229983
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years ago[dev.link] cmd/link: remove ctxt.Syms.Allsym
Cherry Zhang [Sat, 25 Apr 2020 02:45:05 +0000 (22:45 -0400)]
[dev.link] cmd/link: remove ctxt.Syms.Allsym

Replace remaining uses with loader.Syms. Reduces some memory
usage.

Change-Id: I6f295b42b8cd734c6c18f08c61a5473506675075
Reviewed-on: https://go-review.googlesource.com/c/go/+/229992
Reviewed-by: Jeremy Faller <jeremy@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years ago[dev.link] cmd/link: stop setting ReadOnly attribute in late stage
Cherry Zhang [Sat, 25 Apr 2020 00:22:39 +0000 (20:22 -0400)]
[dev.link] cmd/link: stop setting ReadOnly attribute in late stage

The ReadOnly attribute was used to do copy on write when applying
relocations to symbols with read-only backing stores. Now that we
always apply relocations in the output buffer (mmap or heap), it
is always writeable. No need to tamper with the ReadOnly
attribute anymore.

Wasm is an exception, where we don't copy symbol contents to the
output buffer first. Do copy-on-write there.

This is in preparation of converting reloc to using the loader.

Change-Id: I15e53b7c162b9124e6689dfd8eb45cbe2ffd7153
Reviewed-on: https://go-review.googlesource.com/c/go/+/229991
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link: always run Asmb before reloc
Cherry Zhang [Sat, 25 Apr 2020 00:49:35 +0000 (20:49 -0400)]
[dev.link] cmd/link: always run Asmb before reloc

Currently, we run Asmb before reloc, except on Wasm, where the
order is reversed. However, Asmb is no-op on Wasm. So we can
always run Asmb first.

Change-Id: Ifb8989d8150ebdd5777deb05cbccec16f8e36d82
Reviewed-on: https://go-review.googlesource.com/c/go/+/229990
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link: enable new dodata on darwin/arm64
Cherry Zhang [Mon, 27 Apr 2020 15:10:14 +0000 (11:10 -0400)]
[dev.link] cmd/link: enable new dodata on darwin/arm64

Change-Id: I6234e7288212e399f766d19fbca675f45c38e12d
Reviewed-on: https://go-review.googlesource.com/c/go/+/230298
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years ago[dev.link] cmd/link: use new dodata on MIPS(64) and RISCV64
Cherry Zhang [Sat, 25 Apr 2020 01:55:24 +0000 (21:55 -0400)]
[dev.link] cmd/link: use new dodata on MIPS(64) and RISCV64

They also don't need to do anything for Adddynrel. So we can just
enable it.

Change-Id: If85fceca63a7b3cb5a09e5db224c3018060e86de
Reviewed-on: https://go-review.googlesource.com/c/go/+/229993
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years ago[dev.link] cmd/link: support Loader in s390x dodata
Jeremy Faller [Fri, 24 Apr 2020 21:56:24 +0000 (17:56 -0400)]
[dev.link] cmd/link: support Loader in s390x dodata

Recreation of CL 229863 that was removed from the repo because it
included the linker binary.

Change-Id: I5e96afa079b1217df6e7cba63a107546bd96ef76
Reviewed-on: https://go-review.googlesource.com/c/go/+/230028
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years ago[dev.link] cmd/link: remove symbol movement workaround in dodata
Cherry Zhang [Sat, 25 Apr 2020 18:25:18 +0000 (14:25 -0400)]
[dev.link] cmd/link: remove symbol movement workaround in dodata

It is supposed to work around symbol movement in machosymorder.
But machosymorder doesn't actually move symbols around.

Change-Id: Ibdc2ad41aaa8cd49e865088aa1ddb7ab399736cd
Reviewed-on: https://go-review.googlesource.com/c/go/+/230279
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link: set symbol alignments after dynreloc2
Cherry Zhang [Sun, 26 Apr 2020 18:56:19 +0000 (14:56 -0400)]
[dev.link] cmd/link: set symbol alignments after dynreloc2

The symbol alignment is set based on its size. In dynreloc2
symbol size may change (e.g. elfdynhash2). So the alignment must
be set after dynreloc2.

Noticed this while debugging nondeterministic build on Solaris.

Idx Name          Size      VMA               LMA               File off  Algn
  8 .hash         000000c8  000000000048add2  000000000048add2  0008add2  2**3
                  CONTENTS, ALLOC, LOAD, READONLY, DATA

This doesn't look right, as the section address is not a multiple
of its alignment.

Change-Id: I23534cbc59695b7bc241838173fcc71dde95b195
Reviewed-on: https://go-review.googlesource.com/c/go/+/230278
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link: sort DynidSyms
Cherry Zhang [Sun, 26 Apr 2020 17:19:32 +0000 (13:19 -0400)]
[dev.link] cmd/link: sort DynidSyms

Sort DynidSyms to ensure a deterministic build.

Fix Solaris build.

Change-Id: I6c01cb3dec5e46b3d881e720e3c2079643b5c7c7
Reviewed-on: https://go-review.googlesource.com/c/go/+/230277
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link: support new dodata for elf/{arm,arm64}
Than McIntosh [Fri, 24 Apr 2020 15:01:10 +0000 (11:01 -0400)]
[dev.link] cmd/link: support new dodata for elf/{arm,arm64}

Add elf/ARM arch support for the new dodata() phase.

Change-Id: Iadd772b01036c6c5be95bcc6017f6c05d45a24c0
Reviewed-on: https://go-review.googlesource.com/c/go/+/229868
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years ago[dev.link] cmd/link: use new dodata on Plan 9 and Wasm
Cherry Zhang [Fri, 24 Apr 2020 21:15:40 +0000 (17:15 -0400)]
[dev.link] cmd/link: use new dodata on Plan 9 and Wasm

They don't have fancy Adddynrel stuff, so we can just enable it.

Change-Id: I84082c3187d8a9ffa3a9c5458959794df0e3c2b6
Reviewed-on: https://go-review.googlesource.com/c/go/+/230030
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years ago[dev.link] cmd/link: use new dodata on darwin/amd64
Cherry Zhang [Fri, 24 Apr 2020 18:57:20 +0000 (14:57 -0400)]
[dev.link] cmd/link: use new dodata on darwin/amd64

This probably breaks darwin/arm64. Will fix.

Change-Id: I8be168985124f971e9d8ab5bc95c303336dd705b
Reviewed-on: https://go-review.googlesource.com/c/go/+/230019
Reviewed-by: Than McIntosh <thanm@google.com>
5 years ago[dev.link] cmd/link: fix buglet in new GCProg generation code
Cherry Zhang [Fri, 24 Apr 2020 19:19:55 +0000 (15:19 -0400)]
[dev.link] cmd/link: fix buglet in new GCProg generation code

It should check the name of the symbol being added, not the
GC data symbol we're generating.

Change-Id: I123679778ee542b8d1f5c15bf090fa3578025c19
Reviewed-on: https://go-review.googlesource.com/c/go/+/230018
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years ago[dev.link] cmd/link: fix minor error on error reporting
Cherry Zhang [Fri, 24 Apr 2020 19:18:54 +0000 (15:18 -0400)]
[dev.link] cmd/link: fix minor error on error reporting

Correctly propagate ... arguments. (Maybe vet should warn on it?)

Reviewed-on: https://go-review.googlesource.com/c/go/+/230017
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Change-Id: Ife56dc2321847cdaf0caea3142c2c7dad8b5924d
Reviewed-on: https://go-review.googlesource.com/c/go/+/230027
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years ago[dev.link] cmd/link: create symbol updated lazily in amd64 adddynrel
Than McIntosh [Fri, 24 Apr 2020 15:07:17 +0000 (11:07 -0400)]
[dev.link] cmd/link: create symbol updated lazily in amd64 adddynrel

Tweak the code in the amd64 version of adddynrel to avoid creating a
symbol updated for the symbol being processed until it's clear we need
to alter its relocations. This should help performance for the
PIE+internal linking scenario.

Reviewed-on: https://go-review.googlesource.com/c/go/+/229866
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Change-Id: Id25adfd81a5bbd2dde0f80a83b976397ba6abfb5
Reviewed-on: https://go-review.googlesource.com/c/go/+/230026
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years ago[dev.link] cmd/link: check fingerprint for index consistency
Cherry Zhang [Thu, 23 Apr 2020 02:20:44 +0000 (22:20 -0400)]
[dev.link] cmd/link: check fingerprint for index consistency

Previous CL introduced index fingerprint in the object files.
This CL implements the second part: checking fingerprint
consistency in the linker when packages are loaded.

Change-Id: I05dd4c4045a65adfd95e77b625d6c75a7a70e4f1
Reviewed-on: https://go-review.googlesource.com/c/go/+/229618
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/internal/goobj2: add index fingerprint to object file
Cherry Zhang [Wed, 22 Apr 2020 23:21:30 +0000 (19:21 -0400)]
[dev.link] cmd/internal/goobj2: add index fingerprint to object file

The new object files use indices for symbol references, instead
of names. Fundamental to the design, it requires that the
importing and imported packages have consistent view of symbol
indices. The Go command should already ensure this, when using
"go build". But in case it goes wrong, it could lead to obscure
errors like run-time crashes. It would be better to check the
index consistency at build time.

To do that, we add a fingerprint to each object file, which is
a hash of symbol indices. In the object file it records the
fingerprints of all imported packages, as well as its own
fingerprint. At link time, the linker checks that a package's
fingerprint matches the fingerprint recorded in the importing
packages, and issue an error if they don't match.

This CL does the first part: introducing the fingerprint in the
object file, and propagating fingerprints through
importing/exporting by the compiler. It is not yet used by the
linker. Next CL will do.

Change-Id: I0aa372da652e4afb11f2867cb71689a3e3f9966e
Reviewed-on: https://go-review.googlesource.com/c/go/+/229617
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link: panic if HeadType is not set
Cherry Zhang [Fri, 24 Apr 2020 17:02:37 +0000 (13:02 -0400)]
[dev.link] cmd/link: panic if HeadType is not set

In the code there are conditions like !ctxt.IsDarwin(). This will
accidentally be true if HeadType is not yet set. Panic when
HeadType is not set, to catch errors.

Change-Id: Ic891123f27f0276fff5a4b5d29e5b1f7ebbb94ee
Reviewed-on: https://go-review.googlesource.com/c/go/+/229869
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years ago[dev.link] cmd/link: set HeadType early
Cherry Zhang [Fri, 24 Apr 2020 16:13:11 +0000 (12:13 -0400)]
[dev.link] cmd/link: set HeadType early

So we can use it to set per-OS flags.

Also set flagnewDoData after archinit, where IsELF is set.

This should correct the logic of setting flagnewDoData.

Change-Id: I18c7252f141aa35119005c252becc9d7cb74f2f7
Reviewed-on: https://go-review.googlesource.com/c/go/+/229867
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years ago[dev.link] cmd/link: support new dodata for elf/386
Than McIntosh [Fri, 24 Apr 2020 00:54:04 +0000 (20:54 -0400)]
[dev.link] cmd/link: support new dodata for elf/386

Add elf/386 arch support for the new dodata() phase.

Change-Id: I78341dfe70a90719d95c0044183980f348a3369f
Reviewed-on: https://go-review.googlesource.com/c/go/+/229797
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years ago[dev.link] cmd/link: remove -a flag
Cherry Zhang [Fri, 24 Apr 2020 03:30:15 +0000 (23:30 -0400)]
[dev.link] cmd/link: remove -a flag

It doesn't do what it says. It has been like that since Go 1.4.
The current ouput is pretty useless. Remove it.

Change-Id: Id9b4ba04139aaf7ea59acbd51428b1c992115389
Reviewed-on: https://go-review.googlesource.com/c/go/+/229859
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years ago[dev.link] cmd/link: begin converting dodata() to loader APIs
Than McIntosh [Tue, 21 Apr 2020 22:37:43 +0000 (18:37 -0400)]
[dev.link] cmd/link: begin converting dodata() to loader APIs

This patch begins the work of converting the linker's dodata phase to
work with loader APIs. Passes all.bash on linux/amd64, but hasn't been
tested on anything else (more arch-specific code needs to be written).
Use of the new dodata() phase is currently gated by a temporary
command line flag ("-newdodata"), and there is code in the linker's
main routine to insure that we only use the new version for the right
GOOS/GOARCH (currently restricted to ELF + AMD64).

Change-Id: Ied3966677d2a450bc3e0990e0f519b3fceaab806
Reviewed-on: https://go-review.googlesource.com/c/go/+/229706
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years ago[dev.link] cmd/link: move more error handling into loader
Than McIntosh [Wed, 22 Apr 2020 19:16:06 +0000 (15:16 -0400)]
[dev.link] cmd/link: move more error handling into loader

Move the guts of ctxt.Errorf into loader.Loader, so that we can make
calls to it from functions that have a "*loader.Loader" available but
not a "ctxt *Link". This is needed to start converting hooks like
"adddynrel" in the arch-specific portions of the linker to use loader
APIs.

Change-Id: Ieedd4583b66504be0e77d7f3fbadafe0d2307a69
Reviewed-on: https://go-review.googlesource.com/c/go/+/229497
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link/internal/loader: add SetRelocType symbolbuilder method
Than McIntosh [Thu, 23 Apr 2020 13:18:44 +0000 (09:18 -0400)]
[dev.link] cmd/link/internal/loader: add SetRelocType symbolbuilder method

Clients may need to invoke the loader.Reloc2.SetType method to reset
the type of a relocation from external flavor to internal flavor,
meaning that the external type add-in needs to be zeroed (this is
needed when adding dynsym entries).

Add a new SymbolBuider method to support mutating the type of a reloc
for an external symbol, so that the external type can be changed as
well (Reloc2 doesn't have access to that). Also add similar methods
for updating target symbol and addend, so as to have a consistent
interface for ext reloc mutation.

Change-Id: I8e26cdae0a0f353019acba5f9c8a0506e3970266
Reviewed-on: https://go-review.googlesource.com/c/go/+/229604
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link/internal/loader: fix buglet in section handling
Than McIntosh [Thu, 23 Apr 2020 12:43:29 +0000 (08:43 -0400)]
[dev.link] cmd/link/internal/loader: fix buglet in section handling

Allow for the possibility that a client could call newExtSym(), then
ask for the section of the new sym before SetSectSym is called on it
(check in SymSect for this case).

Change-Id: I7bd78e7b3b7618943705b616f62ea78c4a1b68d0
Reviewed-on: https://go-review.googlesource.com/c/go/+/229603
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] all: merge branch 'master' into dev.link
Cherry Zhang [Thu, 23 Apr 2020 15:47:20 +0000 (11:47 -0400)]
[dev.link] all: merge branch 'master' into dev.link

Change-Id: Ia6426e8494244fde7e81a7c932c8e3c865676d3d

5 years ago[dev.link] cmd/link: clean up some tests
Cherry Zhang [Thu, 23 Apr 2020 03:25:41 +0000 (23:25 -0400)]
[dev.link] cmd/link: clean up some tests

Use a separate directory for TestBuildFortvOS test files.

Remove a bad comment in TestTrampoline.

Change-Id: I2dc07ae575ec3f73fb7cea26743094b11a41b464
Reviewed-on: https://go-review.googlesource.com/c/go/+/229619
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years agocmd/go: allow generate to process invalid packages
Johan Jansson [Mon, 20 Apr 2020 18:35:57 +0000 (21:35 +0300)]
cmd/go: allow generate to process invalid packages

Allow go generate to process packages that contain invalid code. Ignore
errors when loading the package, but process only files which have a
valid package clause. Set $GOPACKAGE individually for each file, based
on the package clause.

Add test script for go generate and invalid packages.

Fixes #36422

Change-Id: I91ea088346a1548ccd6678b4595a527b948331ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/229097
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/compile: add ssa.Block.truncateValues
Josh Bleecher Snyder [Thu, 23 Apr 2020 04:35:31 +0000 (21:35 -0700)]
cmd/compile: add ssa.Block.truncateValues

It is a common operation.

Passes toolstash-check.

Change-Id: Icc34600b0f79d0ecb19f257e3c7f23b6f01a26ab
Reviewed-on: https://go-review.googlesource.com/c/go/+/229599
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/compile: simplify zcse
Josh Bleecher Snyder [Thu, 23 Apr 2020 04:34:48 +0000 (21:34 -0700)]
cmd/compile: simplify zcse

Minor refactoring.

Passes toolstash-check.

Change-Id: I91e981bf369d4b719163107644fa58f583356c25
Reviewed-on: https://go-review.googlesource.com/c/go/+/229598
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/go/internal/modget: improve GOINSECURE docs
witchard [Wed, 22 Apr 2020 19:38:30 +0000 (19:38 +0000)]
cmd/go/internal/modget: improve GOINSECURE docs

Recommend use of GOINSECURE over -insecure flang and clarify that GOINSECURE
environment variable does not also imply GONOSUMDB.

Updates #37519 by adding documentation as discussed.

Change-Id: Ia8ab6b3ed1aa559343b72e4ca76c372ee6bf1941
GitHub-Last-Rev: 8d86991f0c1a759f45568c1cc6ef6dc41225f2ba
GitHub-Pull-Request: golang/go#38572
Reviewed-on: https://go-review.googlesource.com/c/go/+/229223
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/compile: pre-alloc enough room for Escape.walkAll
Cuong Manh Le [Wed, 22 Apr 2020 19:21:59 +0000 (02:21 +0700)]
cmd/compile: pre-alloc enough room for Escape.walkAll

Slightly reduce allocs, passes toolstash-check.

name          old time/op       new time/op       delta
Template            181ms ± 4%        174ms ± 0%  -3.59%  (p=0.008 n=5+5)

name          old user-time/op  new user-time/op  delta
Template            249ms ± 3%        240ms ± 2%  -3.59%  (p=0.016 n=5+5)

name          old alloc/op      new alloc/op      delta
Template           35.0MB ± 0%       34.9MB ± 0%  -0.09%  (p=0.008 n=5+5)
Unicode            28.6MB ± 0%       28.6MB ± 0%    ~     (p=0.421 n=5+5)
GoTypes             114MB ± 0%        114MB ± 0%  -0.10%  (p=0.008 n=5+5)
Compiler            542MB ± 0%        541MB ± 0%  -0.13%  (p=0.008 n=5+5)
SSA                1.21GB ± 0%       1.21GB ± 0%  -0.10%  (p=0.008 n=5+5)
Flate              22.0MB ± 0%       22.0MB ± 0%  -0.05%  (p=0.016 n=5+5)
GoParser           27.1MB ± 0%       27.0MB ± 0%  -0.05%  (p=0.008 n=5+5)
Reflect            74.8MB ± 0%       74.8MB ± 0%  -0.11%  (p=0.008 n=5+5)
Tar                33.0MB ± 0%       32.9MB ± 0%  -0.07%  (p=0.008 n=5+5)
XML                42.1MB ± 0%       42.1MB ± 0%  -0.07%  (p=0.008 n=5+5)
LinkCompiler        222MB ± 0%        222MB ± 0%    ~     (p=0.690 n=5+5)
[Geo mean]         81.3MB            81.2MB       -0.07%

name          old allocs/op     new allocs/op     delta
Template             347k ± 0%         347k ± 0%  -0.16%  (p=0.008 n=5+5)
Unicode              334k ± 0%         334k ± 0%  -0.03%  (p=0.016 n=5+5)
GoTypes             1.20M ± 0%        1.20M ± 0%  -0.12%  (p=0.008 n=5+5)
Compiler            5.13M ± 0%        5.12M ± 0%  -0.11%  (p=0.008 n=5+5)
SSA                 11.7M ± 0%        11.7M ± 0%  -0.13%  (p=0.008 n=5+5)
Flate                221k ± 0%         221k ± 0%  -0.18%  (p=0.008 n=5+5)
GoParser             280k ± 0%         280k ± 0%  -0.06%  (p=0.008 n=5+5)
Reflect              902k ± 0%         900k ± 0%  -0.28%  (p=0.008 n=5+5)
Tar                  323k ± 0%         322k ± 0%  -0.18%  (p=0.008 n=5+5)
XML                  401k ± 0%         401k ± 0%  -0.10%  (p=0.008 n=5+5)
LinkCompiler         735k ± 0%         735k ± 0%    ~     (p=0.841 n=5+5)
[Geo mean]           753k              752k       -0.12%

Change-Id: I647bd7752f28b74e6f400fa16cb69632f5c952b3
Reviewed-on: https://go-review.googlesource.com/c/go/+/229517
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
5 years agocmd/compile: refactor variadac call desugaring
Matthew Dempsky [Tue, 21 Apr 2020 22:37:29 +0000 (15:37 -0700)]
cmd/compile: refactor variadac call desugaring

In mid-Walk, we rewrite calls to variadic functions to use explicit
slice literals; e.g., rewriting f(a,b,c) into f([]T{a,b,c}...).
However, it would be useful to do that rewrite much earlier in the
compiler, so that other compiler passes can be simplified.

This CL refactors the rewrite logic into a new fixVariadicCall
function, which subsequent CLs can more easily move into earlier
compiler passes.

Passes toolstash-check -race.

Change-Id: I408e655f2d3aa00446a2e6accf8765abc3b16a8a
Reviewed-on: https://go-review.googlesource.com/c/go/+/229486
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
5 years agotesting: make TempDir work for subtests
Andrei Tudor Călin [Thu, 23 Apr 2020 01:04:33 +0000 (04:04 +0300)]
testing: make TempDir work for subtests

ioutil.TempDir doesn't like path separators in its pattern. Modify
(*common).TempDir to replace path separators with underscores before
using the test name as a pattern for ioutil.TempDir.

Fixes #38465.

Change-Id: I9e8ae48b99648b2bf9f561762e845165aff01972
Reviewed-on: https://go-review.googlesource.com/c/go/+/229399
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/compile: be stricter about recognizing safety rule #4
Matthew Dempsky [Wed, 22 Apr 2020 21:37:02 +0000 (14:37 -0700)]
cmd/compile: be stricter about recognizing safety rule #4

unsafe.Pointer safety rule #4 says "The compiler handles a Pointer
converted to a uintptr in the argument list of a call". Within escape
analysis, we've always required this be a single conversion
unsafe.Pointer->uintptr conversion, but the corresponding logic in
order is somewhat laxer, allowing arbitrary chains of OCONVNOPs from
unsafe.Pointer to uintptr.

This CL changes order to be stricter to match escape analysis.

Passes toolstash-check.

Change-Id: Iadd210d2123accb2020f5728ea2a47814f703352
Reviewed-on: https://go-review.googlesource.com/c/go/+/229578
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years ago[dev.link] cmd/link: only allow heap area to grow to 10MB
Jeremy Faller [Wed, 22 Apr 2020 16:13:33 +0000 (12:13 -0400)]
[dev.link] cmd/link: only allow heap area to grow to 10MB

With CL 228782, we've removed file I/O, but we're growing the memory too
much. This change will periodically flush the heap area to the mmapped
area (if possible).

Change-Id: I1622c738ee5a1a6d02bff5abb0a5751caf8095c7
Reviewed-on: https://go-review.googlesource.com/c/go/+/229439
Run-TryBot: Jeremy Faller <jeremy@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years agodatabase/sql: de-flake TestTxCannotCommitAfterRollback
Daniel Theophanes [Wed, 22 Apr 2020 19:45:44 +0000 (12:45 -0700)]
database/sql: de-flake TestTxCannotCommitAfterRollback

Do not cancel rows during test. Only cancel the Tx.
Correct the referenced issue number on the test.

Fixes #38597

Change-Id: I0e8ba1bf2a8ba638d121c9c6938501fec1d5e961
Reviewed-on: https://go-review.googlesource.com/c/go/+/229478
Run-TryBot: Daniel Theophanes <kardianos@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoencoding/base64: improve performance up to 20% total
Kirill Korotaev [Wed, 22 Apr 2020 20:19:21 +0000 (20:19 +0000)]
encoding/base64: improve performance up to 20% total

Improve base64 encoding/decoding performance by
suppressing compiler boundary checks on decode.

name                 old speed      new speed      delta
EncodeToString-8      570MB/s ± 1%   573MB/s ± 1%     ~     (p=0.421 n=5+5)
DecodeString/2-8     88.6MB/s ± 3%  91.6MB/s ± 2%   +3.37%  (p=0.016 n=5+5)
DecodeString/4-8      162MB/s ± 1%   168MB/s ± 0%   +4.12%  (p=0.008 n=5+5)
DecodeString/8-8      203MB/s ± 0%   214MB/s ± 0%   +5.18%  (p=0.008 n=5+5)
DecodeString/64-8     471MB/s ± 1%   520MB/s ± 1%  +10.50%  (p=0.008 n=5+5)
DecodeString/8192-8   757MB/s ± 0%   895MB/s ± 1%  +18.29%  (p=0.008 n=5+5)

Change-Id: I135243c11aa4c974a4a4e95c5c2abb0635d52c8c
GitHub-Last-Rev: 2c87abcb287f4459888941ae2633edcd8eb5b79d
GitHub-Pull-Request: golang/go#36910
Reviewed-on: https://go-review.googlesource.com/c/go/+/217117
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agonet/http/pprof: allow "seconds" parameters to most profiles
Hana (Hyang-Ah) Kim [Wed, 22 Apr 2020 20:24:30 +0000 (16:24 -0400)]
net/http/pprof: allow "seconds" parameters to most profiles

golang.org/cl/147598 added the support for delta computation for mutex
and block profiles. In fact, this delta computation makes sense for
other types of profiles.

For example, /debug/pprof/allocs?seconds=x will provide how much allocation
was made during the specified period. /debug/pprof/goroutine?seconds=x will
provide the changes in the list of goroutines. This also makes sense for
custom profiles.

Update #23401
Update google/pprof#526

Change-Id: I45e9073eb001ea5b3f3d16e5a57f635193610656
Reviewed-on: https://go-review.googlesource.com/c/go/+/229537
Run-TryBot: Hyang-Ah Hana Kim <hyangah@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agonet/http/pprof: make TestDeltaProfile less flaky by retrying
Hana (Hyang-Ah) Kim [Wed, 22 Apr 2020 19:35:45 +0000 (15:35 -0400)]
net/http/pprof: make TestDeltaProfile less flaky by retrying

In some slow environment, the goroutine for mutexHog2 may not run
within 1secs. So, try with increasing seconds parameters,
and declare failure if it still fails with the longest duration
parameter (32sec).

Also, relax the test condition - previously we expected the
profile's duration is within 0.5~2sec. But obviously, in some
slow environment, that's not even guaranteed. Just check we get
non-zero duration in the result.

Update #38544

Change-Id: Ia9b0d51429a2093e6c9eb92cf463ff6952ef3e10
Reviewed-on: https://go-review.googlesource.com/c/go/+/229498
Run-TryBot: Hyang-Ah Hana Kim <hyangah@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agocmd/compile: update PPC64.rules to use typed aux values
Lynn Boger [Tue, 21 Apr 2020 21:21:10 +0000 (17:21 -0400)]
cmd/compile: update PPC64.rules to use typed aux values

Passes toolstash-check.

Change-Id: I874f8b834ef2f94daa971ecef2dbe4e14daf4213
Reviewed-on: https://go-review.googlesource.com/c/go/+/229305
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/compile: clean up codegen for branch-on-carry on s390x
Michael Munday [Mon, 17 Feb 2020 11:43:33 +0000 (03:43 -0800)]
cmd/compile: clean up codegen for branch-on-carry on s390x

This CL optimizes code that uses a carry from a function such as
bits.Add64 as the condition in an if statement. For example:

    x, c := bits.Add64(a, b, 0)
    if c != 0 {
        panic("overflow")
    }

Rather than converting the carry into a 0 or a 1 value and using
that as an input to a comparison instruction the carry flag is now
used as the input to a conditional branch directly. This typically
removes an ADD LOGICAL WITH CARRY instruction when user code is
doing overflow detection and is closer to the code that a user
would expect to generate.

Change-Id: I950431270955ab72f1b5c6db873b6abe769be0da
Reviewed-on: https://go-review.googlesource.com/c/go/+/219757
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years ago[dev.link] cmd/link/internal/loader: fix AttrSubSymbol
Than McIntosh [Wed, 22 Apr 2020 17:40:22 +0000 (13:40 -0400)]
[dev.link] cmd/link/internal/loader: fix AttrSubSymbol

The code that runs as a part of loadlibfull converts the linker's
outer/sub state and sets the sym.Symbol AttrSubSymbol if a symbol has
both A) an outer sym, and B) is listed as a sub-symbol by some other
symbol.

Make sure that we have the same logic in the original loader method,
since we need to use it as part of dodata() prior to loadlibfull.

Change-Id: I200adab741d778a6ba821419e8ea131ad19375bc
Reviewed-on: https://go-review.googlesource.com/c/go/+/229440
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years ago[dev.link] cmd/link/internal/loader: preprocess numeric constants earlier
Than McIntosh [Wed, 22 Apr 2020 15:35:03 +0000 (11:35 -0400)]
[dev.link] cmd/link/internal/loader: preprocess numeric constants earlier

Change the timing for preprocessing of integer/floating point constant
symbols so that we populate them with content at an earlier stage.
This is needed to allow them can be picked up by the loader-API
version of dodata().

Change-Id: Icf09f4f4b318b4f77e11d4a0f0a9cbecd76a1d6b
Reviewed-on: https://go-review.googlesource.com/c/go/+/229438
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years agocmd/compile: switch to typed aux for mips lowering rules
Alberto Donizetti [Wed, 22 Apr 2020 08:08:41 +0000 (10:08 +0200)]
cmd/compile: switch to typed aux for mips lowering rules

This covers most of the lowering rules.

Passes

  GOARCH=mips gotip build -toolexec 'toolstash -cmp' -a std
  GOARCH=mipsle gotip build -toolexec 'toolstash -cmp' -a std

Change-Id: I9d00aaebecb36622e3bdaf556e5a9377670bf86b
Reviewed-on: https://go-review.googlesource.com/c/go/+/229102
Reviewed-by: Keith Randall <khr@golang.org>
5 years ago[dev.link] cmd/link: create loader-specific version of GCProg
Than McIntosh [Mon, 20 Apr 2020 13:51:21 +0000 (09:51 -0400)]
[dev.link] cmd/link: create loader-specific version of GCProg

Create a new version of the GCProg type + methods that use loader APIs
instead of sym.Symbol.

This code isn't actually used just yet, but will be needed once the
wavefront reaches dodata() and we need to convert that phase.

Change-Id: I087521832015818204fe5c2ac99c7bd3f61b2bf0
Reviewed-on: https://go-review.googlesource.com/c/go/+/229037
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocrypto/x509: disallow setting MaxPathLen without IsCA
Katie Hockman [Fri, 17 Apr 2020 14:23:07 +0000 (10:23 -0400)]
crypto/x509: disallow setting MaxPathLen without IsCA

Fixes #38216

Change-Id: I3222abe2153abb4cbfa65a4825c153ce128f56a0
Reviewed-on: https://go-review.googlesource.com/c/go/+/228777
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
5 years agodatabase/sql: count connections expired in foreground with MaxLifetimeClosed
Colin [Wed, 22 Apr 2020 15:53:52 +0000 (15:53 +0000)]
database/sql: count connections expired in foreground with MaxLifetimeClosed

Previously the connection pool would only count connections
expired in the background connectionCleaner goroutine towards the
MaxLifetimeClosed stat.

This change increments the stat correctly when checking for
expiry in when acquiring and releasing a connection.

Fixes #38058

Change-Id: Id707ddd40a42a4c38658d5f2931da131647d6c29
GitHub-Last-Rev: 0f205ede439efa355a628b6112450f2ca9ced005
GitHub-Pull-Request: golang/go#38263
Reviewed-on: https://go-review.googlesource.com/c/go/+/227278
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Daniel Theophanes <kardianos@gmail.com>
5 years agoruntime: fix bad link to issue tracker in test
Alberto Donizetti [Wed, 22 Apr 2020 15:42:41 +0000 (17:42 +0200)]
runtime: fix bad link to issue tracker in test

Change-Id: Ie88ff3f0493f4119be25476a20038877e879c485
Reviewed-on: https://go-review.googlesource.com/c/go/+/229397
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agoruntime/pprof: plumb labels for goroutine profiles
David Finkel [Sun, 4 Aug 2019 19:14:48 +0000 (15:14 -0400)]
runtime/pprof: plumb labels for goroutine profiles

Goroutines are directly associated with labels. It's relatively easy to
plumb those through without creating goroutine-locals in the wild.

This is accomplished by splitting out most of the code from the public
`runtime.GoroutineProfile` into a new unexported
`runtime.goroutineProfileWithLabels`, which then has a thin wrapper
linked into the `runtime/pprof` package as
`runtime_goroutineProfileWithLabels`. (mirroring the way labels get
associated with the `g` for a goroutine in the first place)

Per-#6104, OS-thread creation profiles are a bit useless, as `M`s tend
to be created be created by a background goroutine. As such, I decided
not to add support for capturing the labels at `M`-creation-time, since
the stack-traces seem to always come out `nil` for my simple test
binaries.

This change currently provides labels for debug=0 and debug=1, as
debug=2 is currently entirely generated by the runtime package and I
don't see a clean way of getting the `labelMap` type handled properly
within the `runtime` package.

Update the comment added in cl/131275 to mention goroutine support for
labels.

Updates #23458

Change-Id: Ia4b558893d7d10156b77121cd9b70c4ccd9e1889
Reviewed-on: https://go-review.googlesource.com/c/go/+/189318
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
5 years agomath/big: clean up whitespace in arith_s390x.s file
Ruixin(Peter) Bao [Tue, 21 Apr 2020 19:14:11 +0000 (15:14 -0400)]
math/big: clean up whitespace in arith_s390x.s file

This CL looks big but it only does formatting changes to arith_s390x.s.
The file was formatted using asmfmt(https://github.com/klauspost/asmfmt)
, so there should not be any functional impact. I verified that the
generated assembly of big.test file is identical.

Change-Id: I8b4035ef082a4d0357881869327e25253f2d8be1
Reviewed-on: https://go-review.googlesource.com/c/go/+/229302
Reviewed-by: Michael Munday <mike.munday@ibm.com>
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years ago[dev.link] cmd/link, cmd/oldlink: remove more darwin/386 and darwin/arm code
Cherry Zhang [Wed, 22 Apr 2020 04:00:03 +0000 (00:00 -0400)]
[dev.link] cmd/link, cmd/oldlink: remove more darwin/386 and darwin/arm code

Updates #37610, #37611.

Change-Id: I0a497af03e24ddea40ed3e342f3a9362bf21ac0c
Reviewed-on: https://go-review.googlesource.com/c/go/+/229323
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Austin Clements <austin@google.com>
5 years ago[dev.link] cmd/internal/goobj: add index to symbol name for indexed symbols
Cherry Zhang [Tue, 21 Apr 2020 17:07:22 +0000 (13:07 -0400)]
[dev.link] cmd/internal/goobj: add index to symbol name for indexed symbols

With old object files, when objdump an object file which, for
example, contains a call of fmt.Fprintf, it shows a symbol
reference like

R_CALL:fmt.Fprintf

With new object files, as the symbol reference is indexed, the
reference becomes

R_CALL:fmt.#33

The object file does not contain information of what symbol #33
in the fmt package is.

To make this more useful, print the index when dumping the symbol
definitions. This way, when dumping the fmt package, e.g.
"go tool nm fmt.a", it will print

   6c705 T fmt.Fprintf#33

So we can find out what symbol #33 actually is.

Change-Id: I320776597d28615ce18dd0617c352d2b8180db49
Reviewed-on: https://go-review.googlesource.com/c/go/+/229246
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link: delete ctxt.Reachparent
Cherry Zhang [Tue, 21 Apr 2020 23:54:37 +0000 (19:54 -0400)]
[dev.link] cmd/link: delete ctxt.Reachparent

It is no longer needed as we have converted the fieldtrack pass
to using the loader.

Also free loader.Reachparent after we are done with it.

Change-Id: Ibc4b29f282e1e4aea363a1b549755e31f84b0295
Reviewed-on: https://go-review.googlesource.com/c/go/+/229322
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link: reduce memory usage for storing symbol section information
Cherry Zhang [Tue, 21 Apr 2020 22:50:49 +0000 (18:50 -0400)]
[dev.link] cmd/link: reduce memory usage for storing symbol section information

Currently, we use a dense array to store symbol's sections. The
array element is a *sym.Section, which takes 8 bytes per symbol
on a 64-bit machine. And the array is created upfront.

To reduce memory usage, use a 16-bit index for sections, so we
store 2 bytes per symbol. The array is pointerless, reducing GC
work. Also create the array lazily.

This reduces some memory usage: linking cmd/compile,

name           old alloc/op   new alloc/op   delta
Loadlib_GC       42.1MB ± 0%    36.2MB ± 0%      -14.01%  (p=0.008 n=5+5)

name           old live-B     new live-B     delta
Loadlib_GC        16.8M ± 0%     15.4M ± 0%       -8.36%  (p=0.008 n=5+5)
Archive_GC        98.2M ± 0%     97.2M ± 0%       -1.02%  (p=0.008 n=5+5) # at the end

Change-Id: If8c41eded8859660bca648c5e6fdf5830810fbf6
Reviewed-on: https://go-review.googlesource.com/c/go/+/229306
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years agounicode/utf8: optimize Valid and ValidString for ASCII checks
Martin Möhrmann [Sat, 18 Apr 2020 21:57:28 +0000 (23:57 +0200)]
unicode/utf8: optimize Valid and ValidString for ASCII checks

Add a fastpath that uses 32bit loads and compares to check
8 ASCII characters per loop iteration.

This avoids the overhead of comparing and branching
for every byte individually.

Combining two 32bit loads into an uint32 allows the same
code to be used for 32bit and 64bit platforms.

amd64 (Intel i7-3520M):
name                         old time/op  new time/op  delta
ValidTenASCIIChars           15.6ns ± 4%   8.5ns ±14%  -45.27%  (p=0.000 n=10+10)
ValidTenJapaneseChars        50.0ns ± 2%  52.7ns ±15%     ~     (p=0.469 n=10+10)
ValidStringTenASCIIChars     13.5ns ± 1%   7.9ns ± 5%  -41.56%  (p=0.000 n=10+10)
ValidStringTenJapaneseChars  46.3ns ± 2%  45.8ns ± 2%     ~     (p=0.085 n=10+10)

arm (Raspberry Pi 3):
name                         old time/op  new time/op  delta
ValidTenASCIIChars           87.5ns ± 0%  58.5ns ± 0%  -33.11%  (p=0.000 n=9+10)
ValidTenJapaneseChars         359ns ± 0%   384ns ± 0%   +6.96%  (p=0.000 n=10+9)
ValidStringTenASCIIChars     87.5ns ± 0%  57.5ns ± 0%  -34.31%  (p=0.000 n=10+10)
ValidStringTenJapaneseChars   356ns ± 0%   377ns ± 0%   +5.90%  (p=0.000 n=10+10)

Change-Id: I9da942bddb250ee1f0ef7aabb4a8cb48edd9053e
Reviewed-on: https://go-review.googlesource.com/c/go/+/228823
Run-TryBot: Martin Möhrmann <moehrmann@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/internal: allow ADDE to work with memory location on s390x
Ruixin(Peter) Bao [Tue, 21 Apr 2020 20:56:22 +0000 (16:56 -0400)]
cmd/internal: allow ADDE to work with memory location on s390x

Originally on s390x, ADDE does not work when adding numbers from a memory location.
For example: ADDE (R3), R4 will result in a failure.

Since ADDC, ADD and ADDW already supports adding from memory location,
let's support that for ADDE as well.

Change-Id: I7cbe112ea154733a621b948c6a21bbee63fb0c62
Reviewed-on: https://go-review.googlesource.com/c/go/+/229304
Reviewed-by: Michael Munday <mike.munday@ibm.com>
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/compile: remove ntz function
Cuong Manh Le [Tue, 21 Apr 2020 17:52:19 +0000 (00:52 +0700)]
cmd/compile: remove ntz function

Use ntzX variants instead.

Passes toolstash-check -a.

Change-Id: I7a627f46f75c3d339034bd3e81c190cea5409c88
Reviewed-on: https://go-review.googlesource.com/c/go/+/229140
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agotime: use extended time format past end of zone transitions
Ian Lance Taylor [Tue, 21 Jan 2020 01:00:48 +0000 (17:00 -0800)]
time: use extended time format past end of zone transitions

This gives us better expected information for daylight savings time
transitions in year 2038 and beyond.

Fixes #36654

Change-Id: I5a39aed3c40b184e1d7bb7d6ce3aff5307c4c146
Reviewed-on: https://go-review.googlesource.com/c/go/+/215539
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agoreflect: fix typo on resolveReflectName function documentation
Koichi Shiraishi [Tue, 21 Apr 2020 18:24:58 +0000 (03:24 +0900)]
reflect: fix typo on resolveReflectName function documentation

Change-Id: I250de9db4e8aca6e1069d05c73051571f1712091
Reviewed-on: https://go-review.googlesource.com/c/go/+/229141
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agoruntime: crash on SI_USER SigPanic signal
Ian Lance Taylor [Sun, 19 Apr 2020 03:11:46 +0000 (20:11 -0700)]
runtime: crash on SI_USER SigPanic signal

Clean up the code a little bit to make it clearer:

Don't check throwsplit for a SI_USER signal.

If throwsplit is set for a SigPanic signal, always throw;
discard any other flags.

Fixes #36420

Change-Id: Ic9dcd1108603d241f71c040504dfdc6e528f9767
Reviewed-on: https://go-review.googlesource.com/c/go/+/228900
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
5 years agonet/http: fix Server.Shutdown race where it could miss an active connection
Brad Fitzpatrick [Mon, 6 Jan 2020 19:16:34 +0000 (11:16 -0800)]
net/http: fix Server.Shutdown race where it could miss an active connection

Wait for Listeners to drop to zero too, not just conns.

Fixes #33313

Change-Id: I09350ae38087990d368dcf9302fbde3e95c02fcd
Reviewed-on: https://go-review.googlesource.com/c/go/+/213442
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hasit Bhatt <hasit.p.bhatt@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoruntime: add bitmap-based markrootSpans implementation
Michael Anthony Knyszek [Thu, 20 Feb 2020 20:03:39 +0000 (20:03 +0000)]
runtime: add bitmap-based markrootSpans implementation

Currently markrootSpans, the scanning routine which scans span specials
(particularly finalizers) as roots, uses sweepSpans to shard work and
find spans to mark.

However, as part of a future CL to change span ownership and how
mcentral works, we want to avoid having markrootSpans use the sweep bufs
to find specials, so in this change we introduce a new mechanism.

Much like for the page reclaimer, we set up a per-page bitmap where the
first page for a span is marked if the span contains any specials, and
unmarked if it has no specials. This bitmap is updated by addspecial,
removespecial, and during sweeping.

markrootSpans then shards this bitmap into mark work and markers iterate
over the bitmap looking for spans with specials to mark. Unlike the page
reclaimer, we don't need to use the pageInUse bits because having a
special implies that a span is in-use.

While in terms of computational complexity this design is technically
worse, because it needs to iterate over the mapped heap, in practice
this iteration is very fast (we can skip over large swathes of the heap
very quickly) and we only look at spans that have any specials at all,
rather than having to touch each span.

This new implementation of markrootSpans is behind a feature flag called
go115NewMarkrootSpans.

Updates #37487.

Change-Id: I8ea07b6c11059f6d412fe419e0ab512d989377b8
Reviewed-on: https://go-review.googlesource.com/c/go/+/221178
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
5 years agocmd/compile: more precise analysis of method values
Matthew Dempsky [Mon, 20 Apr 2020 18:14:36 +0000 (11:14 -0700)]
cmd/compile: more precise analysis of method values

Previously for a method value "x.M", we always flowed x directly to
the heap, which led to the receiver argument generally needing to be
heap allocated.

This CL changes it to flow x to the closure and M's receiver
parameter. This allows receiver arguments to be stack allocated as
long as (1) the closure never escapes, *and* (2) method doesn't leak
its receiver parameter.

Within the standard library, this allows a handful of objects to be
stack allocated instead. Listed here are diagnostics that were
previously emitted by "go build -gcflags=-m std cmd" that are no
longer emitted:

archive/tar/writer.go:118:6: moved to heap: f
archive/tar/writer.go:208:6: moved to heap: f
archive/tar/writer.go:248:6: moved to heap: f
cmd/compile/internal/gc/initorder.go:252:2: moved to heap: d
cmd/compile/internal/gc/initorder.go:75:2: moved to heap: s
cmd/go/internal/generate/generate.go:206:7: &Generator literal escapes to heap
cmd/internal/obj/arm64/asm7.go:910:2: moved to heap: c
cmd/internal/obj/mips/asm0.go:415:2: moved to heap: c
cmd/internal/obj/pcln.go:294:22: new(pcinlineState) escapes to heap
cmd/internal/obj/s390x/asmz.go:459:2: moved to heap: c
crypto/tls/handshake_server.go:56:2: moved to heap: hs

Thanks to Cuong Manh Le for help coming up with this solution.

Fixes #27557.

Change-Id: I8c85d671d07fb9b53e11d2dd05949a34dbbd7e17
Reviewed-on: https://go-review.googlesource.com/c/go/+/228263
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/compile: refactor Escape.tagHole
Matthew Dempsky [Mon, 20 Apr 2020 18:14:15 +0000 (11:14 -0700)]
cmd/compile: refactor Escape.tagHole

This CL refactors tagHole to handle all three call situations (unknown
function; known function in same analysis batch; known function in
previous analysis batch). This will make it somewhat easier to reuse
in a followup CL.

Passes toolstash-check.

Change-Id: I764d047a333dfc593d721a881361683e94b485df
Reviewed-on: https://go-review.googlesource.com/c/go/+/229059
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoruntime: allow proflock and mheap.speciallock above globalAlloc.mutex
Michael Pratt [Fri, 17 Apr 2020 19:36:13 +0000 (15:36 -0400)]
runtime: allow proflock and mheap.speciallock above globalAlloc.mutex

During schedinit, these may occur in:

mProf_Malloc
  stkbucket
    newBucket
      persistentalloc
        persistentalloc1

mProf_Malloc
  setprofilebucket
    fixalloc.alloc
      persistentalloc
        persistentalloc1

These seem to be legitimate lock orderings.

Additionally, mheap.speciallock had a defined rank, but it was never
actually used. That is fixed now.

Updates #38474

Change-Id: I0f6e981852eac66dafb72159f426476509620a65
Reviewed-on: https://go-review.googlesource.com/c/go/+/228786
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
5 years agocmd/compile: optimize comparisons with immediates on s390x
Michael Munday [Fri, 17 Apr 2020 12:05:07 +0000 (05:05 -0700)]
cmd/compile: optimize comparisons with immediates on s390x

When generating code for unsigned equals (==) and not equals (!=)
comparisons we currently, on s390x, always use signed comparisons.

This mostly works well, however signed comparisons on s390x sign
extend their immediates and unsigned comparisons zero extend them.
For compare-and-branch instructions which can only have 8-bit
immediates this significantly changes the range of immediate values
we can represent: [-128, 127] for signed comparisons and [0, 255]
for unsigned comparisons.

When generating equals and not equals checks we don't neet to worry
about whether the comparison is signed or unsigned. This CL
therefore adds rules to allow us to switch signedness for such
comparisons if it means that it brings a constant into range for an
8-bit immediate.

For example, a signed equals with an integer in the range [128, 255]
will now be implemented using an unsigned compare-and-branch
instruction rather than separate compare and branch instructions.

As part of this change I've also added support for adding a name
to block control values using the same `x:(...)` syntax we use for
value rules.

Triggers 792 times when compiling cmd and std.

Change-Id: I77fa80a128f0a8ce51a2888d1e384bd5e9b61a77
Reviewed-on: https://go-review.googlesource.com/c/go/+/228642
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agoRevert "runtime/pprof: speed up CPU profiling shutdown"
Josh Bleecher Snyder [Tue, 21 Apr 2020 19:03:53 +0000 (19:03 +0000)]
Revert "runtime/pprof: speed up CPU profiling shutdown"

This reverts commit 1f0738c1577a55a6b7229b821ddfe762b84771d0.

Reason for revert: This May have caused issue 38567.

Change-Id: I2afa6a9d42cb29cfad09e706fb465c57e3774abd
Reviewed-on: https://go-review.googlesource.com/c/go/+/229301
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
5 years agocmd/compile: remove unused nlo function
Cuong Manh Le [Tue, 21 Apr 2020 06:57:41 +0000 (13:57 +0700)]
cmd/compile: remove unused nlo function

Change-Id: I858d666d491f649f78581a43437408ffab33863b
Reviewed-on: https://go-review.googlesource.com/c/go/+/229139
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Michael Munday <mike.munday@ibm.com>
5 years agocmd/compile: remove nlz function
Cuong Manh Le [Tue, 21 Apr 2020 06:04:59 +0000 (13:04 +0700)]
cmd/compile: remove nlz function

Use nlzX variants instead. While at it, also remove tests involve
nlz/nlo/nto/log2, since when we are calling directly "math/bits"
functions.

Passes toolstash-check.

Change-Id: I83899741a29e05bc2c19d73652961ac795001781
Reviewed-on: https://go-review.googlesource.com/c/go/+/229138
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/compile: convert last 386 rules to typed aux
Alberto Donizetti [Tue, 21 Apr 2020 11:48:57 +0000 (13:48 +0200)]
cmd/compile: convert last 386 rules to typed aux

Passes

  GOARCH=386 gotip build -toolexec 'toolstash -cmp' -a std

Change-Id: I4d1ca83d37ab9f628fc3f1261fe40b81e59137ff
Reviewed-on: https://go-review.googlesource.com/c/go/+/229100
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/go/internal/load: load imports for all package data errors
Jay Conrod [Tue, 21 Apr 2020 15:29:51 +0000 (11:29 -0400)]
cmd/go/internal/load: load imports for all package data errors

go/build.Import can return errors for many different reasons like
inconsistent package clauses or errors parsing build constraints.
It will still return a *build.Package with imports from files it was
able to process. Package.load should load these imports, even after an
unknown error.

There is already a special case for scanner.ErrorList (parse
error). This CL expands that behavior for all errors.

Fixes #38568

Change-Id: I871827299c556f1a9a5b12e7755b221e9d8c6e0e
Reviewed-on: https://go-review.googlesource.com/c/go/+/229243
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocmd/compile: detect and diagnose invalid //go: directive placement
Russ Cox [Wed, 1 Apr 2020 19:51:08 +0000 (15:51 -0400)]
cmd/compile: detect and diagnose invalid //go: directive placement

Thie CL changes cmd/compile/internal/syntax to give the gc half of
the compiler more control over pragma handling, so that it can prepare
better errors, diagnose misuse, and so on. Before, the API between
the two was hard-coded as a uint16. Now it is an interface{}.
This should set us up better for future directives.

In addition to the split, this CL emits a "misplaced compiler directive"
error for any directive that is in a place where it has no effect.
I've certainly been confused in the past by adding comments
that were doing nothing and not realizing it. This should help
avoid that kind of confusion.

The rule, now applied consistently, is that a //go: directive
must appear on a line by itself immediately before the declaration
specifier it means to apply to. See cmd/compile/doc.go for
precise text and test/directive.go for examples.

This may cause some code to stop compiling, but that code
was broken. For example, this code formerly applied the
//go:noinline to f (not c) but now will fail to compile:

//go:noinline
const c = 1

func f() {}

Change-Id: Ieba9b8d90a27cfab25de79d2790a895cefe5296f
Reviewed-on: https://go-review.googlesource.com/c/go/+/228578
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years ago[dev.link] cmd/link: separate out DWARF processing from dodata's allocateSections
Than McIntosh [Mon, 20 Apr 2020 11:47:43 +0000 (07:47 -0400)]
[dev.link] cmd/link: separate out DWARF processing from dodata's allocateSections

Split out DWARF symbol-to-section assignment into its own separate
helper routine, to improve readability. No change in functionality.

Change-Id: Ic2e4f4d99afbff65161cbb8bd63e866ea555f322
Reviewed-on: https://go-review.googlesource.com/c/go/+/228957
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years ago[dev.link] cmd/link: refactor section creation in dodata
Than McIntosh [Fri, 17 Apr 2020 21:23:02 +0000 (17:23 -0400)]
[dev.link] cmd/link: refactor section creation in dodata

Additional cleanups and refactorings in the allocateSections portion
of dodata. Introduce some new helper routines to be used for common
cases in creating sections and assigning symbols, with a goal of
reducing duplicated code blocks and having more readable code.

No change in functionality.

Change-Id: I1b020b3ee993674329b2bebfd7c35995e3a2c043
Reviewed-on: https://go-review.googlesource.com/c/go/+/228883
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years agocmd/go/internal/modfetch: add Unlock before return in checkModSum
BurtonQin [Tue, 21 Apr 2020 11:37:41 +0000 (11:37 +0000)]
cmd/go/internal/modfetch: add Unlock before return in checkModSum

In cmd/go/internal/modfetch/fetch.go,
`checkModSum()` forgets Unlock before return, which may lead to deadlock.
https://github.com/golang/go/blob/876c1feb7d5e10a6ff831de9db19b9ff0ea92fa8/src/cmd/go/internal/modfetch/fetch.go#L514-L520
The fix is to add `goSum.mu.Unlock()` before return.

Change-Id: I855b1c1bc00aeada2c1e84aabb5328f02823007d
GitHub-Last-Rev: afeb3763dd1d08c4c15c4c4183fd5705da60dc4c
GitHub-Pull-Request: golang/go#38563
Reviewed-on: https://go-review.googlesource.com/c/go/+/229219
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/trace: fix the broken link in region pages and improve UX
Rohith Ravi [Sun, 19 Apr 2020 02:21:15 +0000 (02:21 +0000)]
cmd/trace: fix the broken link in region pages and improve UX

The trace tool had a broken link due to a parameter encoding error,
which has been corrected.

In addition:

- the user regions page has been enhanced to include links to
pprof style profiles for region specific io, block, syscall and
schedwait profiles.

- sortable table headers have a pointer cursor to indicate they're
clickable.

Fixes #38518

Change-Id: I26cd5157bd9753750f5f53ea03aac5d2d41b021c
Reviewed-on: https://go-review.googlesource.com/c/go/+/228899
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
5 years ago[dev.link] cmd/link: convert symtab pass to new style
Cherry Zhang [Mon, 20 Apr 2020 22:42:35 +0000 (18:42 -0400)]
[dev.link] cmd/link: convert symtab pass to new style

This is more or less a direct translation, to get things going.
There are more things we can do to make it better, especially on
the handling of container symbols.

Change-Id: I11a0087e402be8d42b9d06869385ead531755272
Reviewed-on: https://go-review.googlesource.com/c/go/+/229125
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Jeremy Faller <jeremy@golang.org>
5 years agoos: correct bad PathError message from FileOpen with O_CREATE on Plan 9
Richard Miller [Tue, 21 Apr 2020 10:00:53 +0000 (11:00 +0100)]
os: correct bad PathError message from FileOpen with O_CREATE on Plan 9

On Plan 9, FileOpen with flag O_CREATE & ~O_TRUNC is done in two
steps.  First, syscall.Open is attempted, to avoid truncation when opening
an existing file.  If that fails because the file doesn't exist,
syscall.Create is used to create a new file.  If the Create fails,
for example because we are racing with another process to create a
ModeExclusive file, the PathError returned from FileOpen should reflect
the result of the Create, not the "does not exist" error from the initial
Open attempt.

Fixes #38540

Change-Id: I90c95a301de417ecdf79cd52748591edb1dbf528
Reviewed-on: https://go-review.googlesource.com/c/go/+/229099
Run-TryBot: David du Colombier <0intro@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David du Colombier <0intro@gmail.com>
5 years agotest/codegen, runtime/pprof, runtime: apply fmt
alex-semenyuk [Mon, 20 Apr 2020 20:33:19 +0000 (20:33 +0000)]
test/codegen, runtime/pprof, runtime: apply fmt

Change-Id: Ife4e065246729319c39e57a4fbd8e6f7b37724e1
GitHub-Last-Rev: e71803eaeb366c00f6c156de0b0b2c50927a0e82
GitHub-Pull-Request: golang/go#38527
Reviewed-on: https://go-review.googlesource.com/c/go/+/228901
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
5 years agocmd/compile: switch to typed aux for 386 optimization rules
Alberto Donizetti [Sun, 19 Apr 2020 08:45:04 +0000 (10:45 +0200)]
cmd/compile: switch to typed aux for 386 optimization rules

Convert first section of 386 optimization rules to the typed aux form.

Adds addOffset{32,64} functions that returns ValAndOffs and a
ValAndOff.canAdd32 function that takes an int32.

Passes

  GOARCH=386 gotip build -toolexec 'toolstash -cmp' -a std

Change-Id: I69d2a8ace6936d5e8ba6ba047183002bf07dd5be
Reviewed-on: https://go-review.googlesource.com/c/go/+/228825
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/compile: use cheaper implementation of oneBit
Josh Bleecher Snyder [Mon, 20 Apr 2020 22:40:38 +0000 (15:40 -0700)]
cmd/compile: use cheaper implementation of oneBit

This is the second attempt. The first attempt was CL 229127,
which got rolled back by CL 229177, because it caused
an infinite loop during compilation on some platforms.
I didn't notice that the trybots hadn't completed when I submitted; mea culpa.

The bug was that we were checking x&(x-1)==0, which is also true of 0,
which does not have exactly one bit set.
This caused an infinite rewrite rule loop.

Updates #38547

file    before    after     Δ       %
compile 19678112  19669808  -8304   -0.042%
total   113143160 113134856 -8304   -0.007%

Change-Id: I417a4f806e1ba61277e31bab2e57dd3f1ac7e835
Reviewed-on: https://go-review.googlesource.com/c/go/+/229197
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Michael Munday <mike.munday@ibm.com>
5 years agoreflect: disallow invoking methods on unexported embedded fields
Matthew Dempsky [Sun, 19 Apr 2020 20:59:16 +0000 (13:59 -0700)]
reflect: disallow invoking methods on unexported embedded fields

Given:

    type u struct{}
    func (u) M() {}

    type t struct { u; u2 u }

    var v = reflect.ValueOf(t{})

Package reflect allows:

    v.Method(0)          // v.M
    v.Field(0).Method(0) // v.u.M

but panics from:

    v.Field(1).Method(0) // v.u2.M

because u2 is not an exported field. However, u is not an exported
field either, so this is inconsistent.

It seems like this behavior originates from #12367, where it was
decided to allow traversing unexported embedded fields to be able to
access their exported fields, since package reflect doesn't provide an
alternative way to access promoted fields directly.

But extending that logic to promoted *methods* was inappropriate,
because package reflect's normal method handling logic already handles
promoted methods correctly. This CL corrects that mistake.

Fixes #38521.

Change-Id: If65008965f35927b4e7927cddf8614695288eb19
Reviewed-on: https://go-review.googlesource.com/c/go/+/228902
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoRevert "cmd/compile: use cheaper implementation of oneBit"
Josh Bleecher Snyder [Tue, 21 Apr 2020 04:28:23 +0000 (04:28 +0000)]
Revert "cmd/compile: use cheaper implementation of oneBit"

This reverts commit 066c47ca5fac1c49f754029f1e61323f74f6d93d.

Reason for revert: This appears to have broken a bunch of builders.

Change-Id: I68b4decf3c1892766e195d8eb018844cdff69443
Reviewed-on: https://go-review.googlesource.com/c/go/+/229177
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
5 years agoreflect: return user-visible method name in panic string
Ian Lance Taylor [Mon, 20 Apr 2020 23:30:43 +0000 (16:30 -0700)]
reflect: return user-visible method name in panic string

This was accidentally broken in CL 166462, which introduce another
function in the panicking path without adjusting the argument to
runtime.Caller.

Change-Id: Ib6f9ed8673fefd458c7a4e3a918c45c5b31ca552
Reviewed-on: https://go-review.googlesource.com/c/go/+/229082
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
5 years agocmd/compile: use proper magnitude for (x>>c) & uppermask = 0
Cuong Manh Le [Mon, 20 Apr 2020 20:47:02 +0000 (03:47 +0700)]
cmd/compile: use proper magnitude for (x>>c) & uppermask = 0

This is followup of CL 228860, which rewrite shift rules to use typed
aux. That CL introduced nlz* functions, to refactor left shift rules.
While at it, we realize there's a bug in old rules with both right/left
shift rules, but only fix for left shift rules only.

This CL fixes the bug for right shift rules.

Passes toolstash-check.

Change-Id: Id8f2158b1b66c9e87f3fdeaa7ae3e35dc0666f8b
Reviewed-on: https://go-review.googlesource.com/c/go/+/229137
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agocmd/compile: rewrite shift rules to use typed aux fields
Cuong Manh Le [Sun, 19 Apr 2020 20:26:53 +0000 (03:26 +0700)]
cmd/compile: rewrite shift rules to use typed aux fields

Passes toolstash-check.

Change-Id: I02e78591fe46e19a43dc36913baef0338a014a3d
Reviewed-on: https://go-review.googlesource.com/c/go/+/228860
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
5 years agoA: add Tailscale Inc. (Corporate CLA)
Brad Fitzpatrick [Wed, 15 Apr 2020 21:44:27 +0000 (14:44 -0700)]
A: add Tailscale Inc. (Corporate CLA)

Change-Id: Ic95f6f78fa56169998a6890beb873693852c5798
Reviewed-on: https://go-review.googlesource.com/c/go/+/228419
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agocmd/compile: use cheaper implementation of oneBit
Josh Bleecher Snyder [Mon, 20 Apr 2020 22:40:38 +0000 (15:40 -0700)]
cmd/compile: use cheaper implementation of oneBit

Updates #38547

file    before    after     Δ       %
compile 19678112  19669808  -8304   -0.042%
total   113143160 113134856 -8304   -0.007%

Change-Id: I5f8afe17401dbdb7c7b3d66d95fe40821c499a92
Reviewed-on: https://go-review.googlesource.com/c/go/+/229127
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/compile: use oneBit instead of isPowerOfTwo in bit optimization
Josh Bleecher Snyder [Mon, 20 Apr 2020 21:43:30 +0000 (14:43 -0700)]
cmd/compile: use oneBit instead of isPowerOfTwo in bit optimization

This optimization works on any integer with exactly one bit set.
This is identical to being a power of two, except in the
most negative number. Use oneBit instead.

The rule now triggers in a few more places in std+cmd,
in packages encoding/asn1, crypto/elliptic, and
vendor/golang.org/x/crypto/cryptobyte.

This change obviates the need for CL 222479
by doing this optimization consistently in the compiler.

Change-Id: I983c6235290fdc634fda5e11b10f1f8ce041272f
Reviewed-on: https://go-review.googlesource.com/c/go/+/229124
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/compile: convert two generic rules to be typed
Josh Bleecher Snyder [Mon, 20 Apr 2020 21:36:40 +0000 (14:36 -0700)]
cmd/compile: convert two generic rules to be typed

Prelude to changing the rules.

Passes toolstash-check.

Change-Id: I22fead7f74d2cf97bb3fbeb22741125b42914c43
Reviewed-on: https://go-review.googlesource.com/c/go/+/229123
Run-TryBot: Josh Bleecher Snyder <josharian@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agogo/types: remove duplicate assert call (minor cleanup)
Robert Griesemer [Mon, 20 Apr 2020 00:56:27 +0000 (17:56 -0700)]
go/types: remove duplicate assert call (minor cleanup)

Change-Id: I6051b3305f8ee02bec4ff3dc7ec2217daed38d72
Reviewed-on: https://go-review.googlesource.com/c/go/+/228903
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoRevert "Revert "cmd/compile: adjust RISCV64 rewrite rules to use typed aux fields""
David Finkel [Mon, 20 Apr 2020 22:15:50 +0000 (18:15 -0400)]
Revert "Revert "cmd/compile: adjust RISCV64 rewrite rules to use typed aux fields""

This reverts commit 98c32670fd454939794504225dca1d4ec55045d5.

Rolling-forward with trivial format-string fix

cmd/compile: adjust RISCV64 rewrite rules to use typed aux fields

Also add a typed version of mergeSym to rewrite.go to assist with a few
rules that used mergeSym in the untyped-form.

Remove a few extra int32 overflow checks that no longer make sense, as
adding two int8s or int16s should never overflow an int32.

Passes toolstash-check -all.

Original review: https://go-review.googlesource.com/c/go/+/228882

Change-Id: Ib63db4ee1687446f0f3d9f11575a40dd85cbce55
Reviewed-on: https://go-review.googlesource.com/c/go/+/229126
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years agocmd/cover: add <title> tag to <head> for coverage report HTML template
David Carter [Mon, 20 Apr 2020 22:17:17 +0000 (22:17 +0000)]
cmd/cover: add <title> tag to <head> for coverage report HTML template

Adds a missing <title> tag to the HTML template to make it
more compliant as <title> tags are generally required for valid
HTML documents.

Change-Id: I1ab2a6ee221c8a79d3cc13d9ac6110f6f4963914
GitHub-Last-Rev: 6d519dc9dda01d142e7f367e43e13c37896cc0cf
GitHub-Pull-Request: golang/go#38313
Reviewed-on: https://go-review.googlesource.com/c/go/+/227547
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoruntime: skip TestBigGOMAXPROCS if it runs out of memory
Ian Lance Taylor [Mon, 20 Apr 2020 18:06:21 +0000 (11:06 -0700)]
runtime: skip TestBigGOMAXPROCS if it runs out of memory

Fixes #38541

Change-Id: I0e9ea5865628d953c32f3a5d4b3ccf1c1d0b081e
Reviewed-on: https://go-review.googlesource.com/c/go/+/229077
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoRevert "cmd/compile: adjust RISCV64 rewrite rules to use typed aux fields"
Than McIntosh [Mon, 20 Apr 2020 21:45:01 +0000 (21:45 +0000)]
Revert "cmd/compile: adjust RISCV64 rewrite rules to use typed aux fields"

This reverts commit 7004be998ba708f027c95615d7885efeb9f73e75.

Reason for revert: causing failures on many builders

Change-Id: I9216bd5409bb6814bac18a6a13ef5115db01b5fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/229120
Run-TryBot: Than McIntosh <thanm@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>