]> Cypherpunks repositories - gostls13.git/log
gostls13.git
5 years agosyscall: fix riscv64 syscall assembly
Joel Sing [Sun, 5 Jan 2020 13:18:56 +0000 (00:18 +1100)]
syscall: fix riscv64 syscall assembly

RISCV has no instruction to store an immediate value to memory, or to
subtract an immediate value. An immediate needs to be moved to a register
first, or in the case of $0, the zero value register (ZERO or R0) can be
used instead. Restore the original riscv-go code so that it compiles.

Updates #27532

Change-Id: I3c60e6f385c59679252a710dbba153357a367516
Reviewed-on: https://go-review.googlesource.com/c/go/+/212761
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 agoall: fix typo in RuneSelf, runeSelf comments
Tim Cooper [Mon, 6 Jan 2020 02:16:26 +0000 (20:16 -0600)]
all: fix typo in RuneSelf, runeSelf comments

Fixes #36396

Change-Id: I52190f450fa9ac52fbf4ecdc814e954dc29029cd
Reviewed-on: https://go-review.googlesource.com/c/go/+/213377
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/link: skip symbol references when looking for missing symbols
Joel Sing [Fri, 30 Aug 2019 15:02:04 +0000 (01:02 +1000)]
cmd/link: skip symbol references when looking for missing symbols

ErrorUnresolved attempts to find the missing symbol in another ABI,
in order to provide more friendly error messages. However, in doing so
it checks the same ABI and can find the symbol reference for the symbol
that it is currently reporting the unresolved error for. Avoid this by
ignoring SXREF symbols, which is the same behaviour used when linking
is performed.

Fixes #33979

Change-Id: I3cb2477b2ad4baa7c2007323b983eb29404b0aac
Reviewed-on: https://go-review.googlesource.com/c/go/+/192597
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
Run-TryBot: Joel Sing <joel@sing.id.au>

5 years agocmd/internal/obj/riscv: correct CALL reloc symbol handling
Joel Sing [Tue, 31 Dec 2019 14:44:54 +0000 (01:44 +1100)]
cmd/internal/obj/riscv: correct CALL reloc symbol handling

Move the CALL reloc symbol back to p.From.Sym, rather than p.To.Sym - while
p.To.Sym is the correct final location, the call to lowerJALR flips the p.To
and p.From addresses (and proves that this form of obj.Prog rewriting is not
a good idea).

Updates #27532

Change-Id: Iaf9fa24b05bda32b963f8d243401c6cd82c1355c
Reviewed-on: https://go-review.googlesource.com/c/go/+/212759
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/internal/obj/riscv: handle MOV rewrites in separate pass
Joel Sing [Tue, 31 Dec 2019 14:28:22 +0000 (01:28 +1100)]
cmd/internal/obj/riscv: handle MOV rewrites in separate pass

Handle the rewriting of MOV pseudo-instructions in a separate pass. This allows
AMOV to be more readily used by other code, including AGETCALLERPC and ARET
implementations.

Updates #27532

Change-Id: Iea794121210048ed23ed0fb8f3c5e3cd5354a311
Reviewed-on: https://go-review.googlesource.com/c/go/+/212758
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoruntime: don't use R11 in nanotime1/walltime1 on ARM
Cherry Zhang [Mon, 30 Dec 2019 16:26:17 +0000 (11:26 -0500)]
runtime: don't use R11 in nanotime1/walltime1 on ARM

R11 a.k.a. REGTMP is the temp register used by the assembler. It
may be clobbered if the assembler needs to synthesize
instructions. In particular, in nanotime1/walltime1, the load of
global variable runtime.iscgo clobbers it. So, avoid using R11
to hold a long-lived value.

Fixes #36309.

Change-Id: Iec2ab9d664532cad8fbf58da17f580e64a744f62
Reviewed-on: https://go-review.googlesource.com/c/go/+/212641
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Andrew G. Morgan <agm@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agodoc: document atexit_sleep_ms flag in race_detector
Daniel Martí [Sun, 13 Oct 2019 12:18:03 +0000 (13:18 +0100)]
doc: document atexit_sleep_ms flag in race_detector

It was pointed out to me that the thread sanitizer library has many more
flags than the ones documented in this doc page. In particular, I was
able to make use of GORACE=atexit_sleep_ms=10 to reduce the amount that
'go test -race' will sleep before finishing.

I'm sure that this flag will be useful to others in the future, so add
it here. This is still not a complete list, but we can simply add useful
flags that work when we think they deserve a mention.

Change-Id: If199d36fd80a0945af90055d110519e5dba27361
Reviewed-on: https://go-review.googlesource.com/c/go/+/200863
Reviewed-by: Heschi Kreinick <heschi@google.com>
5 years agocmd/go: clarify stderr behavior for 'go help mod download'
Jay Conrod [Fri, 3 Jan 2020 18:13:57 +0000 (13:13 -0500)]
cmd/go: clarify stderr behavior for 'go help mod download'

Fixes #36297

Change-Id: I2623e9ee386644eddfa9a348ba368c5f1a81f893
Reviewed-on: https://go-review.googlesource.com/c/go/+/213134
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Michael Matloob <matloob@golang.org>
5 years agocmd/go: convert TestGoTestImportErrorStack to the script framework
Michael Matloob [Tue, 31 Dec 2019 00:45:29 +0000 (19:45 -0500)]
cmd/go: convert TestGoTestImportErrorStack to the script framework

The literal string match had to be turned into a regexp, otherwise
pretty straightforward.

Part of converting all tests to script framework to improve
test parallelism.

Updates #36320
Updates #17751

Change-Id: I46f62312f02dc9adf83ed91c6f807420d29fbb12
Reviewed-on: https://go-review.googlesource.com/c/go/+/212939
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: convert TestGoBuildNotMain to script framework
Michael Matloob [Mon, 30 Dec 2019 21:36:39 +0000 (16:36 -0500)]
cmd/go: convert TestGoBuildNotMain to script framework

Part of converting all tests to script framework to improve
test parallelism.

Updates #36320
Updates #17751

Change-Id: Icd62dc8db55bec52ad326bc370ee7e435aae2559
Reviewed-on: https://go-review.googlesource.com/c/go/+/212812
Run-TryBot: Michael Matloob <matloob@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: convert TestRunInternal to the script test framework
Michael Matloob [Mon, 30 Dec 2019 21:56:00 +0000 (16:56 -0500)]
cmd/go: convert TestRunInternal to the script test framework

This conversion is a bit weird, because the original test runs in the cmd/go
directory, while the script test runs in the GOPATH directory. So even though
it's not necessary for the new test, it changes dircectory to $WORK, so that
its error message regexp can have four components like the original, just
changing the old gopath directory 'testdata' the new one 'gopath'.

Part of converting all tests to script framework to improve test parallelism.

Updates #36320
Updates #17751

Change-Id: Ie5b029c43dc22167278d3104b37c0b57c61326be
Reviewed-on: https://go-review.googlesource.com/c/go/+/212814
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: convert TestCoveragePattern to the script framework
Michael Matloob [Mon, 30 Dec 2019 22:10:23 +0000 (17:10 -0500)]
cmd/go: convert TestCoveragePattern to the script framework

This test already runs in parallel, but still convert it to the
script framework so we can delete the testdata/src directory and
remove any ambiguity about which tests can run in parallel.

Updates #36320

Change-Id: I6470979bd8bad0631dc6ead0d4eb9c83878356e8
Reviewed-on: https://go-review.googlesource.com/c/go/+/212815
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: remove TestGoTestDetectsTestOnlyImportCycles
Michael Matloob [Mon, 30 Dec 2019 23:18:07 +0000 (18:18 -0500)]
cmd/go: remove TestGoTestDetectsTestOnlyImportCycles

The error that's tested in this test is also tested in list_test_err.txt
which uses go list -test -deps. Because both commands are just loading
packages, the difference is not meaningful.

Updates #36320
Updates #17751

Change-Id: Ie712a77d64e8985dd908a1afb515ed3ecc0a9985
Reviewed-on: https://go-review.googlesource.com/c/go/+/212937
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: convert TestPluginNonMain to script framework
Michael Matloob [Mon, 30 Dec 2019 23:49:27 +0000 (18:49 -0500)]
cmd/go: convert TestPluginNonMain to script framework

TestPluginNonMain was broken before this change! (It provided the
wrong directory for testdep/p2: testdata/testdep/p2 instead of
testdata/src/testdep/p2).

Change-Id: Ib815f119bae1d758b500cd8ad82c016cb630d71e
Reviewed-on: https://go-review.googlesource.com/c/go/+/212938
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: convert tests using testdata/src/syntaxerror to scripts
Michael Matloob [Mon, 30 Dec 2019 22:51:16 +0000 (17:51 -0500)]
cmd/go: convert tests using testdata/src/syntaxerror to scripts

This includes TestMatchesNoTestsDoesNotOverrideBuildFailure
and TestErrorMessageForSyntaxErrorInTestGoFileSaysFAIL.
Convert the tests that use the testdata/src/syntaxerror directory
to the script framework.

Part of converting all tests to script framework to improve
test parallelism.

Updates #36320
Updates #17751

Change-Id: I2b2b616e8c124996ae8c8e5b737f15bb493ec588
Reviewed-on: https://go-review.googlesource.com/c/go/+/212816
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agodoc: remove root.html and conduct.html (moved to x/website)
Dmitri Shuralyov [Thu, 2 Jan 2020 22:23:24 +0000 (17:23 -0500)]
doc: remove root.html and conduct.html (moved to x/website)

root.html has been moved to the x/website repo in CL 180959
(commit golang/website@d83058ced3fadc3e8189107d2fb86c333c356b75).

conduct.html has been moved to the x/website repo in CL 207437
(commit golang/website@99763cba2e5b53679a96f69725f917feb319b799).

There should be only one copy, otherwise it may lead to confusion,
or changes made in the wrong place. This CL removes the old copies
from this repo since they're no longer used.

Updates #29206

Change-Id: I41adfb2c34ed3d870fb7a671f48ccc8f90863feb
Reviewed-on: https://go-review.googlesource.com/c/go/+/213157
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agoall: fix invalid invocations of Fatalf in goroutines
Emmanuel T Odeke [Thu, 2 Jan 2020 10:00:55 +0000 (02:00 -0800)]
all: fix invalid invocations of Fatalf in goroutines

Found by running the go vet pass 'testinggoroutine' that
I started in CL 212920.

Change-Id: Ic9462fac85dbafc437fe4a323b886755a67a1efa
Reviewed-on: https://go-review.googlesource.com/c/go/+/213097
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agodoc: add section for checksum database to module reference doc
Katie Hockman [Fri, 13 Dec 2019 00:57:35 +0000 (09:57 +0900)]
doc: add section for checksum database to module reference doc

Updates #33637

Change-Id: Ia782b3fdc5a8873606b96120a34c9bf194a1a346
Reviewed-on: https://go-review.googlesource.com/c/go/+/211197
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agodoc/go1.14: mention sync.Mutex changes
Austin Clements [Thu, 2 Jan 2020 20:23:26 +0000 (15:23 -0500)]
doc/go1.14: mention sync.Mutex changes

Change-Id: Icd92d115e5d7f00b2100598baf2522ebebcdb223
Reviewed-on: https://go-review.googlesource.com/c/go/+/213125
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoruntime: ensure memmove write pointer atomically on ARM64
Cherry Zhang [Fri, 27 Dec 2019 17:02:00 +0000 (12:02 -0500)]
runtime: ensure memmove write pointer atomically on ARM64

If a pointer write is not atomic, if the GC is running
concurrently, it may observe a partially updated pointer, which
may point to unallocated or already dead memory. Most pointer
writes, like the store instructions generated by the compiler,
are already atomic. But we still need to be careful in places
like memmove. In memmove, we don't know which bits are pointers
(or too expensive to query), so we ensure that all aligned
pointer-sized units are written atomically.

Fixes #36101.

Change-Id: I1b3ca24c6b1ac8a8aaf9ee470115e9a89ec1b00b
Reviewed-on: https://go-review.googlesource.com/c/go/+/212626
Reviewed-by: Austin Clements <austin@google.com>
5 years agoruntime: emit trace event in direct semaphore handoff
Rhys Hiltner [Wed, 18 Dec 2019 00:40:46 +0000 (16:40 -0800)]
runtime: emit trace event in direct semaphore handoff

When a goroutine yields the remainder of its time to another goroutine
during direct semaphore handoff (as in an Unlock of a sync.Mutex in
starvation mode), it needs to signal that change to the execution
tracer. The discussion in CL 200577 didn't reach consensus on how best
to describe that, but pointed out that "traceEvGoSched / goroutine calls
Gosched" could be confusing.

Emit a "traceEvGoPreempt / goroutine is preempted" event in this case,
to allow the execution tracer to find a consistent event ordering
without being both specific and inaccurate about why the active
goroutine has changed.

Fixes #36186

Change-Id: Ic4ade19325126db2599aff6aba7cba028bb0bee9
Reviewed-on: https://go-review.googlesource.com/c/go/+/211797
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
5 years agodoc: 2020 is the Year of the Gopher
Dmitri Shuralyov [Wed, 1 Jan 2020 05:00:00 +0000 (00:00 -0500)]
doc: 2020 is the Year of the Gopher

Starting with 2014 (golang.org/cl/46660043), we have enjoyed 6
consecutive years of the gopher. Now, the slice¹ of gophers is
ready to make its way into the next decade, as 2020 is the new
Year of the Gopher.

¹ https://en.wikipedia.org/w/index.php?title=List_of_English_terms_of_venery,_by_animal&oldid=932675028#G

Change-Id: I5f9598dbedb373bd13021964193fa9e44c67693e
Reviewed-on: https://go-review.googlesource.com/c/go/+/213017
Run-TryBot: Dmitri Shuralyov <dmitshur@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
Reviewed-by: David Symonds <dsymonds@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agointernal/poll: use correct fcntl implementations
Tobias Klauser [Fri, 20 Dec 2019 22:34:23 +0000 (23:34 +0100)]
internal/poll: use correct fcntl implementations

Use the libc fcntl (via syscall.fcntl) on aix and solaris like it is
already done for darwin.

For the syscall-based fcntl implementation use FcntlSyscall from
internal/syscall/unix in order to get fcntl64 on 32-bit Linux
systems.

On aix, fcntl with F_DUPFD_CLOEXEC is not supported. Thus, defined
F_DUPFD_CLOEXEC = 0 in the syscall package and check its value before
calling fcntl(fd, syscall.F_DUPFD_CLOEXEC, 0).

On js/wasm, fcntl is not supported thus let its implementation return
ENOSYS directly.

Updates #36211

Change-Id: I96a2ea79e5c4eed2fefd94d0aefd72c940825682
Reviewed-on: https://go-review.googlesource.com/c/go/+/212278
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agointernal/syscall/unix: use libc based fcntl for IsNonblock on aix and solaris
Tobias Klauser [Mon, 30 Dec 2019 14:50:50 +0000 (15:50 +0100)]
internal/syscall/unix: use libc based fcntl for IsNonblock on aix and solaris

On aix and solaris (like on darwin) use libc fcntl to implement
IsNonblock instead of Syscall(SYS_FCNTL, ...) which isn't supported.

Change-Id: I989b02aa0c90b7e2dae025572867dda277fef8be
Reviewed-on: https://go-review.googlesource.com/c/go/+/212600
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agocmd/compile: fix typo in comment
Josh Bleecher Snyder [Mon, 30 Dec 2019 16:35:40 +0000 (08:35 -0800)]
cmd/compile: fix typo in comment

Change-Id: I25fbd63f10ea9892589ad44cc45761926aff0648
Reviewed-on: https://go-review.googlesource.com/c/go/+/212841
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agosyscall: use fcntl64 on 32-bit GNU/Linux systems
Tobias Klauser [Sat, 28 Dec 2019 13:45:35 +0000 (14:45 +0100)]
syscall: use fcntl64 on 32-bit GNU/Linux systems

Use fcntl64Syscall in forkAndExecInChild1 to get fcntl64 on 32-bit Linux
systems.

Updates #36211

Change-Id: Id0e34359256beace970e72102fdace7a987ff2b0
Reviewed-on: https://go-review.googlesource.com/c/go/+/212598
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoruntime: check whether scavAddr is in inUse on scavengeOne fast path
Michael Anthony Knyszek [Thu, 26 Dec 2019 19:49:39 +0000 (19:49 +0000)]
runtime: check whether scavAddr is in inUse on scavengeOne fast path

This change makes it so that we check whether scavAddr is actually
mapped before trying to look at the summary for the fast path, since we
may segfault if that that part of the summary is not mapped in.
Previously this wasn't a problem because we would conservatively map
all memory for the summaries between the lowest mapped heap address and
the highest one.

This change also adds a test for this case.

Change-Id: I2b1d89b5e044dce81745964dfaba829f4becdc57
Reviewed-on: https://go-review.googlesource.com/c/go/+/212637
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoio: show ErrUnexpectedEOF in ExampleReadAtLeast
Ian Lance Taylor [Mon, 23 Dec 2019 23:27:46 +0000 (15:27 -0800)]
io: show ErrUnexpectedEOF in ExampleReadAtLeast

Fixes #36245

Change-Id: I10ce50b0cc28b15f4e7be85b8f12cf9d0e4fac96
Reviewed-on: https://go-review.googlesource.com/c/go/+/212404
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
5 years agodoc/go1.14: mention increased number of EINTR errors
Ian Lance Taylor [Thu, 26 Dec 2019 23:57:13 +0000 (15:57 -0800)]
doc/go1.14: mention increased number of EINTR errors

Updates #36281

Change-Id: I3c4487caaf47566212dc62322b2e884e695ea7f1
Reviewed-on: https://go-review.googlesource.com/c/go/+/212657
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agoruntime: disable pageAlloc tests on OpenBSD in short mode
Michael Anthony Knyszek [Thu, 19 Dec 2019 17:44:58 +0000 (17:44 +0000)]
runtime: disable pageAlloc tests on OpenBSD in short mode

This change disables pageAlloc tests on OpenBSD in short mode because
pageAlloc holds relatively large virtual memory reservations and we make
two during the pageAlloc tests. The runtime may also be carrying one
such reservation making the virtual memory requirement for testing the
Go runtime three times as much as just running a Go binary.

This causes problems for folks who just want to build and test Go
(all.bash) on OpenBSD but either don't have machines with at least 4ish
GiB of RAM (per-process virtual memory limits are capped at some
constant factor times the amount of physical memory) or their
per-process virtual memory limits are low for other reasons.

Fixes #36210.

Change-Id: I8d89cfde448d4cd2fefff4ad6ffed90de63dd527
Reviewed-on: https://go-review.googlesource.com/c/go/+/212177
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agointernal/poll: use fcntl64 on 32-bit GNU/Linux systems
Tobias Klauser [Tue, 24 Dec 2019 00:24:43 +0000 (01:24 +0100)]
internal/poll: use fcntl64 on 32-bit GNU/Linux systems

Use FcntlSyscall from internal/syscall/unix to get fcntl64 on 32-bit
Linux systems.

Updates #36211

Change-Id: If48a6e09606ca9f7f6e22f3e8dc9a25fb3ccaf65
Reviewed-on: https://go-review.googlesource.com/c/go/+/212537
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agodoc/go1.14: change hash/maphash package link
Ariel Mashraki [Tue, 24 Dec 2019 14:28:16 +0000 (16:28 +0200)]
doc/go1.14: change hash/maphash package link

Package name and documentation link were wrong.

Change-Id: I274906afc3cf7a3d88e3da76549cd6ab008fd0c7
Reviewed-on: https://go-review.googlesource.com/c/go/+/212538
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agonet: fix typo in dns client unix test
Isfan Azhabil [Tue, 24 Dec 2019 10:03:33 +0000 (10:03 +0000)]
net: fix typo in dns client unix test

Fix a typo DNS name, it should be [onion.vegetables.com](http://onion.vegetables.com/)

Change-Id: I4122a2305b8f5e774a834651e70c75b54e9386bd
GitHub-Last-Rev: 9907897944275c5f8c54c79667cbb1ef474539de
GitHub-Pull-Request: golang/go#36270
Reviewed-on: https://go-review.googlesource.com/c/go/+/212410
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agocmd/internal/obj/riscv: handle AGETCALLERPC
Joel Sing [Sun, 3 Nov 2019 17:31:20 +0000 (04:31 +1100)]
cmd/internal/obj/riscv: handle AGETCALLERPC

Updates #27532

Change-Id: Ib3e44ec524ed20bbfbbaf40df22cec335460f3ef
Reviewed-on: https://go-review.googlesource.com/c/go/+/204629
Run-TryBot: Joel Sing <joel@sing.id.au>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/internal/obj/riscv: handle FEQ/FNEG/SEQZ/SNEZ
Joel Sing [Sun, 3 Nov 2019 17:08:26 +0000 (04:08 +1100)]
cmd/internal/obj/riscv: handle FEQ/FNEG/SEQZ/SNEZ

Based on riscv-go port.

Updates #27532

Change-Id: I5e7f45955e1dfdb9d09cc6a4e6f3ce81216d411d
Reviewed-on: https://go-review.googlesource.com/c/go/+/204628
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agocmd/internal/obj/riscv: handle call, jmp and branch
Joel Sing [Sun, 3 Nov 2019 15:31:37 +0000 (02:31 +1100)]
cmd/internal/obj/riscv: handle call, jmp and branch

Also provide REG_LR to more clearly define the link register.

Based on the riscv-go port.

Updates #27532

Change-Id: I0805f373682f93b3918a01c21d4ef34eb3817c75
Reviewed-on: https://go-review.googlesource.com/c/go/+/204627
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agointernal/syscall/unix: use fcntl64 on 32-bit GNU/Linux systems
Ian Lance Taylor [Fri, 20 Dec 2019 21:05:24 +0000 (13:05 -0800)]
internal/syscall/unix: use fcntl64 on 32-bit GNU/Linux systems

Patch up runtime testing to use the libc fcntl function on Darwin,
which is what we should be doing anyhow. This is similar to how
we handle fcntl on AIX and Solaris.

Fixes #36211

Change-Id: I47ad87e11df043ce21496a0d59523dad28960f76
Reviewed-on: https://go-review.googlesource.com/c/go/+/212299
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
5 years agotesting: document that benchmark logs are dumped to stdout
Alberto Donizetti [Mon, 23 Dec 2019 12:47:20 +0000 (13:47 +0100)]
testing: document that benchmark logs are dumped to stdout

The doc for testing.T and testing.B said that both test and benchmark
logs are printed to stderr, but in reality that wasn't the case.

CL 24311 fixed the doc for T, this change fixes it for B.

Fixes #36257

Change-Id: I0ff77ff44608f60320a1565b371c81e96039e71c
Reviewed-on: https://go-review.googlesource.com/c/go/+/212457
Reviewed-by: Gabriel Aszalos <gabriel.aszalos@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Gabriel Aszalos <gabriel.aszalos@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agodoc/go1.14: more runtime/compiler release notes
Austin Clements [Wed, 18 Dec 2019 20:24:42 +0000 (15:24 -0500)]
doc/go1.14: more runtime/compiler release notes

This is based on reading through every commit message to runtime and
cmd/{compile,link,internal,asm} since Go 1.13.

Change-Id: I253b1a70ed265f15180fa20c191ceeafa6612ac4
Reviewed-on: https://go-review.googlesource.com/c/go/+/211977
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
5 years agomisc/cgo/test: tweak to pass with GCC 10
Ian Lance Taylor [Fri, 13 Dec 2019 01:00:08 +0000 (17:00 -0800)]
misc/cgo/test: tweak to pass with GCC 10

The test for issue 8945 was marked to only run on gccgo, but there was
no reason for that. It broke for gccgo using GCC 10, because GCC 10
defaults to -fno-common. Make the test run on gc, and split it into
test.go and testx.go to make it work with GCC 10.

The test for issue 9026 used two identical structs which GCC 10 turns
into the same type. The point of the test is not that the structs are
identical, but that they are handled in a particular order. So make
them different.

Updates #8945
Updates #9026

Change-Id: I000fb02f88f346cfbbe5dbefedd944a2c64e8d8e
Reviewed-on: https://go-review.googlesource.com/c/go/+/211217
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
5 years agoruntime: make sure BP is saved in nanotime1/walltime1, else frame pointer may not...
Dan Scales [Thu, 19 Dec 2019 01:27:27 +0000 (17:27 -0800)]
runtime: make sure BP is saved in nanotime1/walltime1, else frame pointer may not be preserved

nanotime1 and walltime1 do not preserve BP on linux amd64. Previously, this
did not cause a problem, because nanotime/walltime do preserve the BP. But now
with mid-stack inlining, nanotime/walltime are usually inlined, so BP is not
preserved. So, the BP is now wrong in any function after a call to
nanotime()/walltime() on amd64. That means the frame pointer on the stack can
be wrong for any further function call made after the nanotime() call (notably
runtime.main and various GC functions). [386 doesn't use framepointer.]

Fix is to set a frame size of 8 for nanotime1 and walltime1, which means the
standard prolog/epilog that saves/restore BP in the stack frame is added.

I noticed this while investigating issue 16638 (use frame pointers for
runtime.Callers). This change would needed for progress on that issue (which
doesn't have a high priority). Verified that this fix works/is useful for issue
16638.

Change-Id: I19e19ef2c1a517d737a34928baae034f2eb0b2c2
Reviewed-on: https://go-review.googlesource.com/c/go/+/212079
Run-TryBot: Dan Scales <danscales@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/go/internal/modfetch/codehost: replace a dubious call to semver.Max
Bryan C. Mills [Fri, 20 Dec 2019 04:17:55 +0000 (23:17 -0500)]
cmd/go/internal/modfetch/codehost: replace a dubious call to semver.Max

The documentation for RecentTag indicates that it returns an actual
tag, not a canonicalized prefix+version blob equivalent to a tag,
so the canonicalization due to semver.Max seems like a bug here.

Fortunately, RecentTag is not currently ever actually used as a tag,
so the removal of metadata does not result in a user-facing bug.
Nonetheless, it may be a subtle source of confusion for maintainers
in the future.

Updates #32700

Change-Id: I525423c1c0c7ec7c36c09e53b180034474f74e5a
Reviewed-on: https://go-review.googlesource.com/c/go/+/212202
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/go: diagnose missing replacement directories
Bryan C. Mills [Fri, 20 Dec 2019 14:34:30 +0000 (09:34 -0500)]
cmd/go: diagnose missing replacement directories

I noticed the missing diagnostic when writing a regression test for #33795.

Change-Id: Ic3249436a6109d71f9ff720b7096f9b872f6a94b
Reviewed-on: https://go-review.googlesource.com/c/go/+/212201
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: avoid erroneous canonicalization when trying to resolve imports using replace...
Bryan C. Mills [Fri, 20 Dec 2019 03:36:13 +0000 (22:36 -0500)]
cmd/go: avoid erroneous canonicalization when trying to resolve imports using replacements

Updates #32700
Fixes #33795

Change-Id: I16897a0a2f3aa2f0b0bf8cf8252f3f39eef2e7ba
Reviewed-on: https://go-review.googlesource.com/c/go/+/212200
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/go: relax validation for replacements for gopkg.in paths
Bryan C. Mills [Thu, 19 Dec 2019 20:46:06 +0000 (15:46 -0500)]
cmd/go: relax validation for replacements for gopkg.in paths

The 'go' command normally requires the 'go.mod' files for replacement
modules to have a major version compatible with the module they are
replacing.

However, prior to CL 206761, the 'go' command erroneously allowed
unversioned paths (which imply major version 0 or 1) to replace
'gopkg.in' paths with any major-version suffix.

An analysis of proxy.golang.org suggests that these replacements,
while uncommon, are not unheard-of. Rather than breaking the modules
that rely on them, we will continue to allow the erroneous replacement
paths for this particular pairing.

Updates #34254

Change-Id: Icb4e745981803edaa96060f17a8720a058219ab1
Reviewed-on: https://go-review.googlesource.com/c/go/+/212105
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: in 'go list -m', print effective go.mod file
Jay Conrod [Thu, 19 Dec 2019 18:00:28 +0000 (13:00 -0500)]
cmd/go: in 'go list -m', print effective go.mod file

When the -modfile flag is in use (either explicitly or from GOFLAGS),
'go list -m' will now print the effective go.mod file for the main
module in the GoMod field in -f or -json output.

Fixes #36220
Updates #34506

Change-Id: I89c2ee40f20e07854bb37c6e4e13eeea0cce7b0d
Reviewed-on: https://go-review.googlesource.com/c/go/+/212100
Run-TryBot: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocmd/go/internal/clean: make 'go clean -testcache' a no-op if the cache directory...
Bryan C. Mills [Thu, 19 Dec 2019 14:33:16 +0000 (09:33 -0500)]
cmd/go/internal/clean: make 'go clean -testcache' a no-op if the cache directory is gone

Fixes #29100

Change-Id: I195191aad825266ab55d38addef9d662cfc72dff
Reviewed-on: https://go-review.googlesource.com/c/go/+/212099
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agotext/template: indirect interfaces before slicing
Daniel Martí [Wed, 18 Dec 2019 10:05:59 +0000 (10:05 +0000)]
text/template: indirect interfaces before slicing

The recently added slice function used indirectInterface, but then
forgot to actually call reflect.Value.Slice on its result. Calling the
Slice method on the original Value without indirectInterface would
result in a panic, if our slice was indeed behind an interface.

Fix that, and add test cases for all three built-in functions that work
with slices.

Fixes #36199.

Change-Id: I9a18f4f604a3b29967eefeb573f8960000936b88
Reviewed-on: https://go-review.googlesource.com/c/go/+/211877
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>
5 years agoreflect: record PkgPath of StructOf arguments
Ian Lance Taylor [Thu, 19 Dec 2019 03:59:41 +0000 (19:59 -0800)]
reflect: record PkgPath of StructOf arguments

Fixes #36190
Fixes #36191

Change-Id: I1213ef37b6595af63dbe202a8ade65741caf1356
Reviewed-on: https://go-review.googlesource.com/c/go/+/212001
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agoruntime: use innermost frame's func name for async preemption check
Cherry Zhang [Wed, 18 Dec 2019 20:19:05 +0000 (15:19 -0500)]
runtime: use innermost frame's func name for async preemption check

We don't asynchronously preempt if we are in the runtime. We do
this by checking the function name. However, it failed to take
inlining into account. If a runtime function gets inlined into
a non-runtime function, it can be preempted, and bad things can
happen. One instance of this is dounlockOSThread inlined into
UnlockOSThread which is in turn inlined into a non-runtime
function.

Fix this by using the innermost frame's function name.

Change-Id: Ifa036ce1320700aaaefd829b4bee0d04d05c395d
Reviewed-on: https://go-review.googlesource.com/c/go/+/211978
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
5 years agodoc/go1.14: release notes for runtime and compiler
Austin Clements [Tue, 17 Dec 2019 22:39:50 +0000 (17:39 -0500)]
doc/go1.14: release notes for runtime and compiler

Change-Id: I9519659983de23f43ff0e05cffd336d8bc351400
Reviewed-on: https://go-review.googlesource.com/c/go/+/211758
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Alexander Rakoczy <alex@golang.org>
5 years agocmd/doc: flag usage() inconsistency
Gert Cuykens [Fri, 18 Oct 2019 02:13:50 +0000 (04:13 +0200)]
cmd/doc: flag usage() inconsistency

Inconsistency between cmd/doc/main.go and cmd/go/internal/doc/doc.go

Fixes #34976

Change-Id: I429200f9305d473edb4505216bb4840ba92af818
Reviewed-on: https://go-review.googlesource.com/c/go/+/201857
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agodoc/go1.14: document changes to -mod=readonly and go.mod file maintenance
Bryan C. Mills [Tue, 17 Dec 2019 19:41:45 +0000 (14:41 -0500)]
doc/go1.14: document changes to -mod=readonly and go.mod file maintenance

Fixes #36169

Change-Id: Ib9a53fdb0112635b53be38d6818834dd1775e70c
Reviewed-on: https://go-review.googlesource.com/c/go/+/211698
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agosrc/go/doc: in Examples, check for len of params.List, not nilness
Michael Matloob [Mon, 16 Dec 2019 22:20:02 +0000 (17:20 -0500)]
src/go/doc: in Examples, check for len of params.List, not nilness

This makes the check the same as the one in the tests vet check.
It's safer to check the number of arguments rather than for a nil
slice.

Change-Id: I8e04e9c612573f334770c1c4245238649656c6e2
Reviewed-on: https://go-review.googlesource.com/c/go/+/211598
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agogo/build: rename WorkingDir to Dir go1.14beta1
Bryan C. Mills [Tue, 17 Dec 2019 14:00:57 +0000 (09:00 -0500)]
go/build: rename WorkingDir to Dir

Fixes #36168

Change-Id: If2b7368671e83657a3a74dd030e66e7c68bf2361
Reviewed-on: https://go-review.googlesource.com/c/go/+/211657
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agotext/template: remove an incorrect comment and clarify docs for AddParseTree
Oliver Powell [Sat, 5 Oct 2019 11:05:56 +0000 (13:05 +0200)]
text/template: remove an incorrect comment and clarify docs for AddParseTree

The comment "If the name is the name of this template, overwrite this template."
is incorrect and should be "is not" instead. This comment is no longer
required once the docs are updated to mention this behaviour instead.

Fixes #34695

Change-Id: I773495b2194d7bb7619b13c1a28cbc76e8f69aac
Reviewed-on: https://go-review.googlesource.com/c/go/+/199139
Reviewed-by: Rob Pike <r@golang.org>
5 years agotime: further clarifications to the (*Timer).Stop docs
Ian Lance Taylor [Wed, 20 Nov 2019 00:14:50 +0000 (16:14 -0800)]
time: further clarifications to the (*Timer).Stop docs

Fixes #35627

Change-Id: I0c5fed46a69a9663e46a9414468ec610063ea05a
Reviewed-on: https://go-review.googlesource.com/c/go/+/207849
Reviewed-by: Rob Pike <r@golang.org>
5 years agodoc: remove Go Security Policy page (moved to x/website)
Dmitri Shuralyov [Fri, 13 Dec 2019 16:23:09 +0000 (11:23 -0500)]
doc: remove Go Security Policy page (moved to x/website)

This page has moved to the x/website repo in CL 211300 (commit
golang/website@3c8b7f99cadaa000e642595d0fabcd9ac496f335).
Remove the old copy in this repo since it's no longer used.

Updates #29206

Change-Id: I8b3396d9e42d1e7262a8cde9577962d33b215836
Reviewed-on: https://go-review.googlesource.com/c/go/+/211301
Reviewed-by: Filippo Valsorda <filippo@golang.org>
5 years agoapi: promote next to go1.14
Alexander Rakoczy [Mon, 16 Dec 2019 18:58:27 +0000 (13:58 -0500)]
api: promote next to go1.14

Change-Id: I14b1a21a8639b3241326e74ab6152673d5d71243
Reviewed-on: https://go-review.googlesource.com/c/go/+/211517
Run-TryBot: Alexander Rakoczy <alex@golang.org>
Run-TryBot: Carlos Amedee <carlos@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Carlos Amedee <carlos@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
5 years agoruntime: do not use PowerRegisterSuspendResumeNotification on systems with "program...
Jason A. Donenfeld [Thu, 21 Nov 2019 15:16:56 +0000 (16:16 +0100)]
runtime: do not use PowerRegisterSuspendResumeNotification on systems with "program time" timer

Systems where PowerRegisterSuspendResumeNotification returns ERROR_
FILE_NOT_FOUND are also systems where nanotime() is on "program time"
rather than "real time".  The chain for this is:

powrprof.dll!PowerRegisterSuspendResumeNotification ->
  umpdc.dll!PdcPortOpen ->
    ntdll.dll!ZwAlpcConnectPort("\\PdcPort") ->
      syscall -> ntoskrnl.exe!AlpcpConnectPort

Opening \\.\PdcPort fails with STATUS_OBJECT_NAME_NOT_FOUND when pdc.sys
hasn't been initialized. Pdc.sys also provides the various hooks for
sleep resumption events, which means if it's not loaded, then our "real
time" timer is actually on "program time". Finally STATUS_OBJECT_NAME_
NOT_FOUND is passed through RtlNtStatusToDosError, which returns ERROR_
FILE_NOT_FOUND. Therefore, in the case where the function returns ERROR_
FILE_NOT_FOUND, we don't mind, since the timer we're using will
correspond fine with the lack of sleep resumption notifications. This
applies, for example, to Docker users.

Fixes #35447
Fixes #35482

Change-Id: I9e1ce5bbc54b9da55ff7a3918b5da28112647eee
Reviewed-on: https://go-review.googlesource.com/c/go/+/208317
Reviewed-by: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Austin Clements <austin@google.com>
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agogo/doc: ignore example functions with arguments
Michael Matloob [Fri, 13 Dec 2019 21:53:28 +0000 (16:53 -0500)]
go/doc: ignore example functions with arguments

An Example function with arguments is not a valid example to be
run with go test. Don't return those functions from Examples. This
means that some functions that were previously showing up in
Examples will no longer show up. But those functions were not being
tested properly so the fact that they were showing up is misleading.

This fixes an issue where a confusing compiler error was showing
up when running go test on a file with an invalid example. While
that issue could have been fixed by returning an error, this is
more consistent with the behavior of go/doc.Examples, and the tests
checker in vet will catch this issue.

Fixes #35284

Change-Id: I2101a7d19f38522ef9c2e50967f9cfb30d28c730
Reviewed-on: https://go-review.googlesource.com/c/go/+/211357
Run-TryBot: Michael Matloob <matloob@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agoos/exec: ignore hungup error while copying stdin on Plan 9
Fazlul Shahriar [Thu, 21 Nov 2019 09:09:17 +0000 (04:09 -0500)]
os/exec: ignore hungup error while copying stdin on Plan 9

Fixes #35753

Change-Id: I38674c59c601785eb25b778dc25efdb92231dd9b
Reviewed-on: https://go-review.googlesource.com/c/go/+/208223
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agospec: reword claim about the grammar being regular
Alberto Donizetti [Fri, 13 Dec 2019 11:07:06 +0000 (12:07 +0100)]
spec: reword claim about the grammar being regular

Since the word "regular" has a precise meaning in the context of
formal languages, the Introduction sentence claiming that Go's grammar
is "compact and regular" may mislead readers.

Reword it using Rob's suggestion.

Fixes #36037

Change-Id: I00c1a5714bdab8878d9a77b36d67dae67d63da0f
Reviewed-on: https://go-review.googlesource.com/c/go/+/211277
Reviewed-by: Robert Griesemer <gri@golang.org>
Reviewed-by: Rob Pike <r@golang.org>
5 years agodoc: fix missing closing tag in module documentation
Jay Conrod [Fri, 13 Dec 2019 16:11:11 +0000 (11:11 -0500)]
doc: fix missing closing tag in module documentation

Updates #33637

Change-Id: If262d1501cf73b404361f832a2e3e17aaa0db78b
Reviewed-on: https://go-review.googlesource.com/c/go/+/211299
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
5 years agoruntime: fix off-by-1 error on address ranges
Cherry Zhang [Thu, 12 Dec 2019 21:41:24 +0000 (16:41 -0500)]
runtime: fix off-by-1 error on address ranges

When growing the address ranges, the new length is the old length + 1.

Fixes #36113.

Change-Id: I1b425f78e473cfa3cbdfe6113e166663f41fc9f3
Reviewed-on: https://go-review.googlesource.com/c/go/+/211157
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
5 years agodoc: remove expiration from PGP key in security.html
Filippo Valsorda [Thu, 12 Dec 2019 22:04:04 +0000 (17:04 -0500)]
doc: remove expiration from PGP key in security.html

The key had expired earlier this year. Simply resigned it with no
expiration, so it maintains the same fingerprint.

Removed the encouragement to use PGP above the fold. We trust the
security of our mail system, so it's really only there for people that
want it.

Also removed the individual keys, as they were never used, and both Adam
and I have access to the security@golang.org key anyway.

Change-Id: Icc5ad6dfb4f0b52128a59a080b7f270b20d3c520
Reviewed-on: https://go-review.googlesource.com/c/go/+/211177
Reviewed-by: Katie Hockman <katie@golang.org>
5 years agoruntime: force segv for nil defer function to be in deferreturn()
Dan Scales [Wed, 11 Dec 2019 01:27:26 +0000 (17:27 -0800)]
runtime: force segv for nil defer function to be in deferreturn()

If the defer function pointer is nil, force the seg fault to happen in deferreturn
rather than in jmpdefer. jmpdefer is used fairly infrequently now because most
functions have open-coded defers.

The open-coded defer implementation calls gentraceback() with a callback when
looking for the first open-coded defer frame. gentraceback() throws an error if it
is called with a callback on an LR architecture and jmpdefer is on the stack,
because the stack trace can be incorrect in that case - see issue #8153. So, we
want to make sure that we don't have a seg fault in jmpdefer.

Fixes #36050

Change-Id: Ie25e6f015d8eb170b40248dedeb26a37b7f9b38d
Reviewed-on: https://go-review.googlesource.com/c/go/+/210978
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agonet/http: fix some awkward wording in Request.WithContext docs
Brad Fitzpatrick [Thu, 12 Dec 2019 17:31:11 +0000 (17:31 +0000)]
net/http: fix some awkward wording in Request.WithContext docs

Change-Id: Ib37db42c7f1fd6aa55f70fd2d65d56bb2ae6d26a
Reviewed-on: https://go-review.googlesource.com/c/go/+/211098
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocmd/go: restore default vet analyzers for targets in GOROOT
Bryan C. Mills [Wed, 11 Dec 2019 17:14:31 +0000 (12:14 -0500)]
cmd/go: restore default vet analyzers for targets in GOROOT

This fixes a regression introduced in CL 209498,
found while investigating #32471.

Also fix $WORK replacement in cmd/go/internal/work.(*Builder).Showcmd
when b.WorkDir includes a backslash and appears in a quoted string.
That fix is needed in order to write a precise test that passes under Windows,
since Windows directories nearly always include backslashes.

Updates #35837

Change-Id: I5fddc5435d5d283a3e598989209d873b59b0a39c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210937
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: allow arguments to 'go test' and 'go vet' to duplicate or override flags...
Bryan C. Mills [Fri, 6 Dec 2019 18:43:23 +0000 (13:43 -0500)]
cmd/go: allow arguments to 'go test' and 'go vet' to duplicate or override flags from GOFLAGS

This is a minimal fix for Go 1.14, but this parsing logic is much too
complex and seems like it will cause more trouble going forward.

I intend to mail a followup change to refactor this logic for 1.15.

Updates #32471

Change-Id: I00ed07dcf3a23c9cd4ffa8cf764921fb5c18bcd6
Reviewed-on: https://go-review.googlesource.com/c/go/+/210940
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agoruntime: use inUse ranges to map in summary memory only as needed
Michael Anthony Knyszek [Mon, 18 Nov 2019 19:23:39 +0000 (19:23 +0000)]
runtime: use inUse ranges to map in summary memory only as needed

Prior to this change, if the heap was very discontiguous (such as in
TestArenaCollision) it's possible we could map a large amount of memory
as R/W and commit it. We would use only the start and end to track what
should be mapped, and we would extend that mapping as needed to
accomodate a potentially fragmented address space.

After this change, we only map exactly the part of the summary arrays
that we need by using the inUse ranges from the previous change. This
reduces the GCSys footprint of TestArenaCollision from 300 MiB to 18
MiB.

Because summaries are no longer mapped contiguously, this means the
scavenger can no longer iterate directly. This change also updates the
scavenger to borrow ranges out of inUse and iterate over only the
parts of the heap which are actually currently in use. This is both an
optimization and necessary for correctness.

Fixes #35514.

Change-Id: I96bf0c73ed0d2d89a00202ece7b9d089a53bac90
Reviewed-on: https://go-review.googlesource.com/c/go/+/207758
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
5 years agotest: add a test for gccgo compiler bug of missing type descriptor
Cherry Zhang [Wed, 11 Dec 2019 16:44:08 +0000 (11:44 -0500)]
test: add a test for gccgo compiler bug of missing type descriptor

The gccgo compiler did not generate type descriptor for a pointer
to a type alias defined in another package, causing linking error.
The fix is CL 210787. This CL adds a test.

Updates #36085.

Change-Id: I3237c7fedb4d92fb2dc610ee2b88087f96dc2a1a
Reviewed-on: https://go-review.googlesource.com/c/go/+/210858
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoruntime: track ranges of address space which are owned by the heap
Michael Anthony Knyszek [Fri, 15 Nov 2019 23:30:30 +0000 (23:30 +0000)]
runtime: track ranges of address space which are owned by the heap

This change adds a new inUse field to the allocator which tracks ranges
of addresses that are owned by the heap. It is updated on each heap
growth.

These ranges are tracked in an array which is kept sorted. In practice
this array shouldn't exceed its initial allocation except in rare cases
and thus should be small (ideally exactly 1 element in size).

In a hypothetical worst-case scenario wherein we have a 1 TiB heap and 4
MiB arenas (note that the address ranges will never be at a smaller
granularity than an arena, since arenas are always allocated
contiguously), inUse would use at most 4 MiB of memory if the heap
mappings were completely discontiguous (highly unlikely) with an
additional 2 MiB leaked from previous allocations. Furthermore, the
copies that are done to keep the inUse array sorted will copy at most 4
MiB of memory in such a scenario, which, assuming a conservative copying
rate of 5 GiB/s, amounts to about 800µs.

However, note that in practice:
1) Most 64-bit platforms have 64 MiB arenas.
2) The copies should incur little-to-no page faults, meaning a copy rate
   closer to 25-50 GiB/s is expected.
3) Go heaps are almost always mostly contiguous.

Updates #35514.

Change-Id: I3ad07f1c2b5b9340acf59ecc3b9ae09e884814fe
Reviewed-on: https://go-review.googlesource.com/c/go/+/207757
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Austin Clements <austin@google.com>
5 years agonet/http: use cancellation instead of a timeout in TestTransportProxyHTTPSConnectTimeout
Bryan C. Mills [Wed, 11 Dec 2019 14:35:31 +0000 (09:35 -0500)]
net/http: use cancellation instead of a timeout in TestTransportProxyHTTPSConnectTimeout

The use of a timeout in this test caused it to be flaky: if the
timeout occurred before the connection was attempted, then the Accept
call on the Listener could hang indefinitely, and its goroutine would
not exit until that Listener was closed. That caused the test to fail.

A longer timeout would make the test less flaky, but it would become
even slower and would still be sensitive to timing.

Instead, replace the timeout with an explicit Context cancellation
after the CONNECT request has been read. That not only ensures that
the cancellation occurs at the appropriate point, but also makes the
test much faster: a test run with -count=1000 now executes in less
than 2s on my machine, whereas before it took upwards of 50s.

Fixes #36082
Updates #28012

Change-Id: I00c20d87365fd3d257774422f39d2acc8791febd
Reviewed-on: https://go-review.googlesource.com/c/go/+/210857
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agoruntime: syscall_forkx on Solaris can return error on success
Andrew Stormont [Wed, 11 Dec 2019 00:31:44 +0000 (00:31 +0000)]
runtime: syscall_forkx on Solaris can return error on success

The syscall_forkx function returns the value of errno even on success.  This can be a problem when using cgo where an atfork handler might be registered; if the atfork handler does something which causes errno to be set the caller of syscall_forkx can be misled into thinking the fork has failed.  This causes the various exec functions in the runtime package to hang.

Change-Id: Ia1842179226078a0cbbea33d541aa1187dc47f68
GitHub-Last-Rev: 4dc4db75c82a826da9a50c323b7e3ddfe46ed6c0
GitHub-Pull-Request: golang/go#36076
Reviewed-on: https://go-review.googlesource.com/c/go/+/210742
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agodoc: remove Release History pages (moved to x/website)
Dmitri Shuralyov [Tue, 10 Dec 2019 20:31:47 +0000 (15:31 -0500)]
doc: remove Release History pages (moved to x/website)

These pages were moved to the x/website repo in CL 210797 (commit
golang/website@9aef1eefbbe663d448b04b7cc1b2b995f4cf4c0b).
Remove the old copies in this repo since they're no longer used.

Updates #36075
Updates #29206

Change-Id: I6e3ffaebd92fa753cb5f3b21e4238edfb7f5f0e8
Reviewed-on: https://go-review.googlesource.com/c/go/+/210798
Reviewed-by: Alexander Rakoczy <alex@golang.org>
5 years agosyscall: use SOCK_CLOEXEC when creating sockets
Lorenz Bauer [Tue, 10 Dec 2019 15:58:24 +0000 (15:58 +0000)]
syscall: use SOCK_CLOEXEC when creating sockets

LsfSocket, SetLsfPromisc and NetlinkRIB currently don't force the CLOEXEC
flag on the sockets they create. While the former two functions are
deprecated, NetlinkRIB is called by various functions related to
net.Interface.

Add a helper to create CLOEXEC sockets, and use it from SetLsfPromisc and
NetlinkRIB. LsfSocket is unchanged since we don't want to break callers.

Fixes #36053

Change-Id: I72fe2b167996797698d8a44b0d28165045c42d3c
Reviewed-on: https://go-review.googlesource.com/c/go/+/210517
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agocmd/compile/internal/logopt: gofmt
Tobias Klauser [Tue, 10 Dec 2019 15:21:56 +0000 (16:21 +0100)]
cmd/compile/internal/logopt: gofmt

Change-Id: Ie9d29645e7702104202ee1f338babdd9e33e1e58
Reviewed-on: https://go-review.googlesource.com/c/go/+/210679
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agotesting: use conventional comments for exported internal funcs & structs
vovapi [Tue, 10 Dec 2019 10:34:34 +0000 (10:34 +0000)]
testing: use conventional comments for exported internal funcs & structs

Change-Id: I2dd5ddc22bfff143b81d5945992d8c5fccf387f4
GitHub-Last-Rev: aa637756e772f5ee9094b802df3be9945c8466c4
GitHub-Pull-Request: golang/go#36054
Reviewed-on: https://go-review.googlesource.com/c/go/+/210497
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agoruntime: use MRS instruction to read system registers on freebsd/arm64
Tobias Klauser [Tue, 10 Dec 2019 10:57:35 +0000 (11:57 +0100)]
runtime: use MRS instruction to read system registers on freebsd/arm64

Support for these was added in CL 189577

Change-Id: Iaf2a774b141995cbbdfb3888aea67ae9c7f928b1
Reviewed-on: https://go-review.googlesource.com/c/go/+/210677
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agonet/http: await state traces earlier in TestServerConnState
Bryan C. Mills [Tue, 10 Dec 2019 15:02:27 +0000 (10:02 -0500)]
net/http: await state traces earlier in TestServerConnState

This approach attempts to ensure that the log for each connection is
complete before the next sequence of states begins.

Updates #32329

Change-Id: I25150d3ceab6568af56a40d2b14b5f544dc87f61
Reviewed-on: https://go-review.googlesource.com/c/go/+/210717
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agonet/http: fix timeout race in Transport proxy CONNECT
Brad Fitzpatrick [Tue, 10 Dec 2019 17:40:22 +0000 (17:40 +0000)]
net/http: fix timeout race in Transport proxy CONNECT

Fixes #36070

Change-Id: I99742aa153202436d802634c9e019a14b9ef9185
Reviewed-on: https://go-review.googlesource.com/c/go/+/210738
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agoall: fix a number of misuses of the word "an"
Daniel Martí [Tue, 10 Dec 2019 14:27:53 +0000 (14:27 +0000)]
all: fix a number of misuses of the word "an"

After golang.org/cl/210124, I wondered if the same error had gone
unnoticed elsewhere. I quickly spotted another dozen mistakes after
reading through the output of:

git grep '\<[Aa]n [bcdfgjklmnpqrtvwyz][a-z]'

Many results are false positives for acronyms like "an mtime", since
it's pronounced "an em-time". However, the total amount of output isn't
that large given how simple the grep pattern is.

Change-Id: Iaa2ca69e42f4587a9e3137d6c5ed758887906ca6
Reviewed-on: https://go-review.googlesource.com/c/go/+/210678
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Zach Jones <zachj1@gmail.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agocmd/go: document 'go env GOMOD' outside modules
Jay Conrod [Mon, 9 Dec 2019 19:29:29 +0000 (14:29 -0500)]
cmd/go: document 'go env GOMOD' outside modules

In documentation for 'go env GOMOD', note that the path will be
os.DevNull in module-aware mode when no go.mod file is present.

Fixes #36052

Change-Id: I30ced1df02ccefe1970bd856190e79d6f0384375
Reviewed-on: https://go-review.googlesource.com/c/go/+/210577
Run-TryBot: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocmd/link: fix imports problem in elf_test.go
Than McIntosh [Mon, 9 Dec 2019 20:13:24 +0000 (15:13 -0500)]
cmd/link: fix imports problem in elf_test.go

Fix an imports problem in this test (doesn't compile).

Updates #35779

Change-Id: Icaeec0384bf2e75696e43d9410df7219f0245940
Reviewed-on: https://go-review.googlesource.com/c/go/+/210578
Run-TryBot: Than McIntosh <thanm@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/link: fix loadelf failed on MIPS family
Meng Zhuo [Fri, 6 Dec 2019 08:13:44 +0000 (16:13 +0800)]
cmd/link: fix loadelf failed on MIPS family

The relocation of MIPS64 family ELF is different with other architecure according
to the document from Linux-MIPS

https://www.linux-mips.org/pub/linux/mips/doc/ABI/elf64-2.4.pdf

In "2.9 Relocation" it shows relocation section contains five parts:

1. r_sym Elf64_Word Symbol index
2. r_ssym Elf64_Byte Special symbol
3. r_type3 Elf64_Byte Relocation type
4. r_type2 Elf64_Byte Relocation type
5. r_type Elf64_Byte Relocation type

This CL makes loadelf aware the difference.

Update #35779

Change-Id: Ib221665641972b1c2bfea5a496e3118e5dc0bc45
Reviewed-on: https://go-review.googlesource.com/c/go/+/209317
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5 years agoruntime: use mprotect in sysMap for aix/ppc64
Clément Chigot [Thu, 14 Nov 2019 14:43:55 +0000 (15:43 +0100)]
runtime: use mprotect in sysMap for aix/ppc64

AIX doesn't allow to mmap an already mmap address. The previous way to
deal with this behavior was to munmap before calling mmap again.
However, mprotect syscall is able to change protections on a memory
range. Thus, memory mapped by sysReserve can be remap using it. Note
that sysMap is always called with a non-nil pointer so mprotect is
always possible.

Updates: #35451

Change-Id: I1fd1e1363d9ed9eb5a8aa7c8242549bd6dad8cd0
Reviewed-on: https://go-review.googlesource.com/c/go/+/207237
Reviewed-by: Austin Clements <austin@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
5 years agonet/http: don't wait indefinitely in Transport for proxy CONNECT response
Brad Fitzpatrick [Fri, 6 Dec 2019 20:47:29 +0000 (20:47 +0000)]
net/http: don't wait indefinitely in Transport for proxy CONNECT response

Fixes #28012

Change-Id: I711ebaabf63194e3d2c608d829da49c51a294d74
Reviewed-on: https://go-review.googlesource.com/c/go/+/210286
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
5 years agoruntime/race: correct typo s/is/in
Muhammad Falak R Wani [Sun, 8 Dec 2019 16:44:52 +0000 (22:14 +0530)]
runtime/race: correct typo s/is/in

Change-Id: Ic79d97c2aa107c0e5c4a8906ad757b0390228bef
Reviewed-on: https://go-review.googlesource.com/c/go/+/210417
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agoruntime: suggest more kernel options for mlock failure
Mark Pulford [Fri, 6 Dec 2019 22:50:17 +0000 (09:50 +1100)]
runtime: suggest more kernel options for mlock failure

Some Linux distributions will continue to provide 5.3.x kernels for a
while rather than 5.4.x.

Updates #35777

Change-Id: I493ef8338d94475f4fb1402ffb9040152832b0fd
Reviewed-on: https://go-review.googlesource.com/c/go/+/210299
Reviewed-by: Austin Clements <austin@google.com>
5 years agocmd/go: clarify -dropreplace help text in 'go help mod edit'
Jay Conrod [Fri, 6 Dec 2019 20:02:41 +0000 (15:02 -0500)]
cmd/go: clarify -dropreplace help text in 'go help mod edit'

Fixes #34930

Change-Id: Ie0ddd8de6e40f5a192c4d2267399126e449fe5f5
Reviewed-on: https://go-review.googlesource.com/c/go/+/210340
Run-TryBot: Jay Conrod <jayconrod@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agoruntime: mlock top of signal stack on both amd64 and 386
Austin Clements [Mon, 9 Dec 2019 03:24:10 +0000 (22:24 -0500)]
runtime: mlock top of signal stack on both amd64 and 386

CL 209899 worked around an issue that corrupts vector registers in
recent versions of the Linux kernel by mlocking the top page of every
signal stack on amd64. However, the underlying issue also affects the
XMM registers on 386. This CL applies the mlock fix to both amd64 and
386.

Fixes #35777 (again).

Change-Id: I9886f2dc4c23625421296bd5518d5fd3288bfe48
Reviewed-on: https://go-review.googlesource.com/c/go/+/210345
Run-TryBot: Austin Clements <austin@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

5 years agogo/ast: fix a few incorrect "an" articles
Daniel Martí [Sun, 8 Dec 2019 20:54:10 +0000 (20:54 +0000)]
go/ast: fix a few incorrect "an" articles

slice, type, and select should all go with "a", not "an", since they all
start with consonant sounds.

Change-Id: I57e96dcbdc571dc21e24096779f2f756ec81103e
Reviewed-on: https://go-review.googlesource.com/c/go/+/210124
Reviewed-by: Robert Griesemer <gri@golang.org>
5 years agodoc: add missing p in install from source page
Alberto Donizetti [Sat, 7 Dec 2019 15:52:10 +0000 (16:52 +0100)]
doc: add missing p in install from source page

The last paragraph in golang.org/doc/install/source#fetch is missing a
p tag, so it doesn't get formatted with the 'max-width: 50rem' like
all the other text in the page.

Add it.

Change-Id: I1a981dd2afde561b4ab21bd90ad99b3a146111f6
Reviewed-on: https://go-review.googlesource.com/c/go/+/210122
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agonet/http: unflake TestTimeoutHandlerSuperfluousLogs
Emmanuel T Odeke [Fri, 22 Nov 2019 19:18:43 +0000 (14:18 -0500)]
net/http: unflake TestTimeoutHandlerSuperfluousLogs

Uses 2 channels to synchronize that test, because
relying on sleeps creates flaky behavior, thus:

a) 1 buffered channel to send back the last spurious line
without having to reason about "happens before" behavior
a) 1 buffered channel at the end of the handler; it'll
be controlled by whether we expect to timeout or not,
but will always be closed when the test ends

Fixes #35051

Change-Id: Iff735aa8d1ed9de8d92b792374ec161cc0a72798
Reviewed-on: https://go-review.googlesource.com/c/go/+/208477
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
5 years agocrypto/tls: fix a typo in TLS handshake comment
Alex Harford [Fri, 6 Dec 2019 23:27:29 +0000 (23:27 +0000)]
crypto/tls: fix a typo in TLS handshake comment

Fix a minor typo in the TLS handshake comment.

Change-Id: I0fd243e5440f6c77c97e844e6669a7974a2c8798
GitHub-Last-Rev: 21e91a1a48e8a9200e6fca89a988e1383ded7bb2
GitHub-Pull-Request: golang/go#36030
Reviewed-on: https://go-review.googlesource.com/c/go/+/210289
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
5 years agocmd/go/internal/modcmd: loosen path validation in "go mod edit"
Bryan C. Mills [Fri, 6 Dec 2019 20:51:34 +0000 (15:51 -0500)]
cmd/go/internal/modcmd: loosen path validation in "go mod edit"

Replaced modules require only valid import paths, not full
module paths that can be fetched with 'go get'.

The 'go' command does not in general reject manually-edited go.mod
files with these paths, so 'go mod edit' should not reject them
either.

Fixes #30513

Change-Id: I4f1a5c65937f91d41478f8d218c8018e0c70f320
Reviewed-on: https://go-review.googlesource.com/c/go/+/210343
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go/internal/modload: revert duplicate path suppression in modload.ImportMissingError
Bryan C. Mills [Fri, 6 Dec 2019 20:40:04 +0000 (15:40 -0500)]
cmd/go/internal/modload: revert duplicate path suppression in modload.ImportMissingError

CL 210338 suppressed duplication for import paths mentioned in an
ImportMissingError.

Unfortunately, that broke one of the cases in
cmd/go/internal/modload.TestImport, and the new error message is still
kind of awkward anyway.

Let's revert that part of the change — we can try again with more
coverage for that case.

Updates #35986

Change-Id: Ib0858aec4f89a7231e32c35ec876da80d80f2098
Reviewed-on: https://go-review.googlesource.com/c/go/+/210342
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/go: include cfg.BuildModReason in 'import lookup disabled' errors
Bryan C. Mills [Fri, 6 Dec 2019 20:02:28 +0000 (15:02 -0500)]
cmd/go: include cfg.BuildModReason in 'import lookup disabled' errors

This location was missed in CL 204521.

Updates #33326
Updates #33848

Change-Id: I0ece6d9b37548d8abb54f79c69be5548a0428c76
Reviewed-on: https://go-review.googlesource.com/c/go/+/210341
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
5 years agocmd/link: skip gaps between PT_LOAD segments in TestPIESize
Cherry Zhang [Fri, 6 Dec 2019 21:52:53 +0000 (16:52 -0500)]
cmd/link: skip gaps between PT_LOAD segments in TestPIESize

There may be gaps between non-writeable and writeable PT_LOAD
segments, and the gaps may be large as the segments may have
large alignment. Don't count those gaps in file size comparison.

Fixes #36023.

Change-Id: I68582bdd0f385ac5c6f87d485d476d06bc96db19
Reviewed-on: https://go-review.googlesource.com/c/go/+/210180
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>