Matthew Dempsky [Tue, 22 Oct 2019 00:06:02 +0000 (17:06 -0700)]
sync/atomic: suppress checkptr errors for hammerStoreLoadPointer
This test could be updated to use unsafe.Pointer arithmetic properly
(e.g., see discussion at #34972), but it doesn't seem worthwhile. The
test is just checking that LoadPointer and StorePointer are atomic.
Updates #34972.
Change-Id: I85a8d610c1766cd63136cae686aa8a240a362a18
Reviewed-on: https://go-review.googlesource.com/c/go/+/202597
Run-TryBot: Matthew Dempsky <mdempsky@google.com> Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com> Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Bryan C. Mills [Tue, 22 Oct 2019 16:44:44 +0000 (12:44 -0400)]
net: convert TestTCPServer to use subtests
My fix in CL 202618 inadvertently violated an invariant in the inner
loop of TestTCPServer (namely, that len(trchs) == i). That causes a
panic when one or more of the channels is omitted due to a flake.
Instead of trying to fix up the test, let's just factor out a subtest
and skip the whole thing if the transceiver's Dial flakes out.
Robert Griesemer [Mon, 21 Oct 2019 21:25:45 +0000 (14:25 -0700)]
go/types: don't update package-external types when checking validity
The recently added type-validity check uses a new field of Named
types for marking (to detect cycles). That field was modified even
if the type was not part of the current package or belonged to the
Universe scope (error type). This led to race conditions if the
package's type was imported by multiple, concurrently type-checked
packages.
A test would be nice but it's a bit cumbersome to set one up.
Verified manually that package-external types are left alone.
Fixes #35049.
Change-Id: I51686bef47fcca48b99b91ecb1b2e9d58e135ea6
Reviewed-on: https://go-review.googlesource.com/c/go/+/202483 Reviewed-by: Bryan C. Mills <bcmills@google.com>
David du Colombier [Tue, 22 Oct 2019 11:07:15 +0000 (13:07 +0200)]
cmd/go: fix TestScript/list_ambiguous_path on Plan 9
CL 198459 added TestScript/list_ambiguous_path. This
test is failing on Plan 9, because the expected error
doesn't match the error message returned on Plan 9.
This change fixes the test by matching the correct
error message on Plan 9.
Fixes #35072.
Change-Id: If8cdb641e0e9544ae4ac24f8d0c54859a3b23a69
Reviewed-on: https://go-review.googlesource.com/c/go/+/202447
Run-TryBot: David du Colombier <0intro@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Bryan C. Mills <bcmills@google.com>
Ian Lance Taylor [Mon, 21 Oct 2019 16:55:23 +0000 (09:55 -0700)]
runtime: correctly negate errno value for *BSD ARM
Fixes #35037
Change-Id: I0b9bcd001556cd409994d83dabcdd6e32b001d28
Reviewed-on: https://go-review.googlesource.com/c/go/+/202441
Run-TryBot: Ian Lance Taylor <iant@golang.org> Reviewed-by: Bryan C. Mills <bcmills@google.com>
Jason A. Donenfeld [Mon, 21 Oct 2019 14:12:22 +0000 (16:12 +0200)]
syscall: respect permission bits on file opening on Windows
On Windows, os.Chmod and syscall.Chmod toggle the FILE_ATTRIBUTES_
READONLY flag depending on the permission bits. That's a bit odd but I
guess some compromises were made at some point and this is what was
chosen to map to a Unix concept that Windows doesn't really have in the
same way. That's fine. However, the logic used in Chmod was forgotten
from os.Open and syscall.Open, which then manifested itself in various
places, most recently, go modules' read-only behavior.
This makes syscall.Open consistent with syscall.Chmod and adds a test
for the permission _behavior_ using ioutil. By testing the behavior
instead of explicitly testing for the attribute bits we care about, we
make sure this doesn't regress in unforeseen ways in the future, as well
as ensuring the test works on platforms other than Windows.
In the process, we fix some tests that never worked and relied on broken
behavior, as well as tests that were disabled on Windows due to the
broken behavior and had TODO notes.
Fixes #35033
Change-Id: I6f7cf54517cbe5f6b1678d1c24f2ab337edcc7f7
Reviewed-on: https://go-review.googlesource.com/c/go/+/202439
Run-TryBot: Jason A. Donenfeld <Jason@zx2c4.com> Reviewed-by: Bryan C. Mills <bcmills@google.com> Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Ian Lance Taylor [Tue, 22 Oct 2019 07:38:08 +0000 (00:38 -0700)]
runtime: force testing calls of netpoll to run on system stack
Fixes #35053
Change-Id: I31853d434610880044c169e0c1e9732f97ff1bdb
Reviewed-on: https://go-review.googlesource.com/c/go/+/202444
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David du Colombier <0intro@gmail.com>
Cuong Manh Le [Sat, 19 Oct 2019 08:18:34 +0000 (15:18 +0700)]
runtime: fix unsafe.Pointer alignment on Linux
Caught by go test -a -short -gcflags=all=-d=checkptr runtime
TestMincoreErrorSign intentionally uses uintptr(1) to get -EINVAL,
but it violates unsafe pointer rules 2. So use another misaligned
pointer add(new(int32), 1), but do not violate unsafe pointer rules.
TestEpollctlErrorSign passes an unsafe.Pointer of &struct{}{} to
Epollctl, which is then casted to epollevent, causes mis-alignment.
Fixing it by exporting epollevent on runtime_test package, so it can be
passed to Epollctl.
Updates #34972
Change-Id: I78ebfbeaf706fd1d372272af0bbc4e2cabca4631
Reviewed-on: https://go-review.googlesource.com/c/go/+/202157
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com> Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Ian Lance Taylor [Thu, 11 Apr 2019 00:38:26 +0000 (17:38 -0700)]
runtime, syscall, time: prepare for adding timers to P's
Add new fields to runtime.timer, and adjust the various timer
functions in preparation for adding timers to P's. This continues to
use the old timer code.
Updates #6239
Updates #27707
Change-Id: I9adb3814f657e083ec5e22736c4b5b52b77b6a3f
Reviewed-on: https://go-review.googlesource.com/c/go/+/171829
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Michael Knyszek <mknyszek@google.com>
Matthew Dempsky [Thu, 17 Oct 2019 21:29:16 +0000 (14:29 -0700)]
cmd/compile: recognize (*[Big]T)(ptr)[:n:m] pattern for -d=checkptr
A common idiom for turning an unsafe.Pointer into a slice is to write:
s := (*[Big]T)(ptr)[:n:m]
This technically violates Go's unsafe pointer rules (rule #1 says T2
can't be bigger than T1), but it's fairly common and not too difficult
to recognize, so might as well allow it for now so we can make
progress on #34972.
Dmitri Shuralyov [Mon, 21 Oct 2019 19:30:13 +0000 (15:30 -0400)]
net/http: remove parseURL variable
The parseURL variable was introduced in CL 49930 in order to work
around the fact that the name "url" was shadowed by a parameter of
exported functions, and couldn't be renamed without sacrificing
documentation readability. Documentation readability takes higher
priority than internal implementation details.
Back then, I considered renaming the net/url import but saw that it
would be too disruptive of a change to the large net/http package.
Now I see a better way: it's possible to import net/url both as url
and as urlpkg (the package is still imported just once, but it becomes
available via two names). This way we eliminate the need for wasting
(a little) memory on the parseURL variable, improve code readability
slightly, and delete some lines of code and comments.
Dmitri Shuralyov [Mon, 21 Oct 2019 19:04:54 +0000 (15:04 -0400)]
database/sql: remove forced log import from test
This var _ = log.Printf line was added 8 years ago, in CL 4973055,
which created the database/sql package and its tests. There was no
goimports back then, so this was likely added to make it easier to
use log package during development of tests.
It's no longer needed, so remove it. It can always be conveniently
re-added via goimports whenever needed.
Bryan C. Mills [Mon, 21 Oct 2019 18:12:26 +0000 (14:12 -0400)]
crypto/tls: retry net.Dial flakes on Dragonfly
localPipe currently flakes in various crypto/tls tests. Since that
function doesn't seem to flake anywhere else, I suspect a kernel bug.
To make the test less flaky, retry the Dial if we suspect that it is
affected. (Worst case, we delay the test by a few seconds before
erroring out as usual.)
smasher164 [Mon, 15 Oct 2018 07:14:57 +0000 (03:14 -0400)]
cmd/compile: add fma intrinsic for arm
This change introduces an arm intrinsic that generates the FMULAD
instruction for the fused-multiply-add operation on systems that
support it. System support is detected via cpu.ARM.HasVFPv4. A rewrite
rule translates the generic intrinsic to FMULAD.
Updates #25819.
Change-Id: I8459e5dd1cdbdca35f88a78dbeb7d387f1e20efa
Reviewed-on: https://go-review.googlesource.com/c/go/+/142117
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Ian Lance Taylor [Thu, 11 Apr 2019 00:23:05 +0000 (17:23 -0700)]
runtime, syscall, time: add and use resettimer
As a small step toward speeding up timers, restrict modification
of the timer.when field to the timer code itself. Other code that
wants to change the when field of an existing timer must now call
resettimer rather than changing the when field and calling addtimer.
The new resettimer function also works for a new timer.
This is just a refactoring in preparation for later code.
Updates #27707
Change-Id: Iccd5dcad415ffbeac4c2a3cf015e91f82692acf8
Reviewed-on: https://go-review.googlesource.com/c/go/+/171825
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Michael Knyszek <mknyszek@google.com> Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
To permit ssa-level optimization, this change introduces an amd64 intrinsic
that generates the VFMADD231SD instruction for the fused-multiply-add
operation on systems that support it. System support is detected via
cpu.X86.HasFMA. A rewrite rule can then translate the generic ssa intrinsic
("Fma") to VFMADD231SD.
The benchmark compares the software implementation (old) with the intrinsic
(new).
name old time/op new time/op delta
Fma-4 27.2ns ± 1% 1.0ns ± 9% -96.48% (p=0.008 n=5+5)
Updates #25819.
Change-Id: I966655e5f96817a5d06dff5942418a3915b09584
Reviewed-on: https://go-review.googlesource.com/c/go/+/137156
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Ian Lance Taylor [Fri, 5 Apr 2019 22:53:12 +0000 (15:53 -0700)]
runtime: add netpollBreak
The new netpollBreak function can be used to interrupt a blocking netpoll.
This function is not currently used; it will be used by later CLs.
Updates #27707
Change-Id: I5cb936609ba13c3c127ea1368a49194fc58c9f4d
Reviewed-on: https://go-review.googlesource.com/c/go/+/171824
Run-TryBot: Ian Lance Taylor <iant@golang.org> Reviewed-by: Michael Knyszek <mknyszek@google.com>
smasher164 [Thu, 30 Aug 2018 00:57:33 +0000 (20:57 -0400)]
cmd/compile: introduce generic ssa intrinsic for fused-multiply-add
In order to make math.FMA a compiler intrinsic for ISAs like ARM64,
PPC64[le], and S390X, a generic 3-argument opcode "Fma" is provided and
rewritten as
ARM64: (Fma x y z) -> (FMADDD z x y)
PPC64: (Fma x y z) -> (FMADD x y z)
S390X: (Fma x y z) -> (FMADD z x y)
Akhil Indurti [Thu, 2 Aug 2018 03:22:46 +0000 (23:22 -0400)]
math: add guaranteed-precision FMA implementation
Currently, the precision of the float64 multiply-add operation
(x * y) + z varies across architectures. While generated code for
ppc64, s390x, and arm64 can guarantee that there is no intermediate
rounding on those platforms, other architectures like x86, mips, and
arm will exhibit different behavior depending on available instruction
set. Consequently, applications cannot rely on results being identical
across GOARCH-dependent codepaths.
This CL introduces a software implementation that performs an IEEE 754
double-precision fused-multiply-add operation. The only supported
rounding mode is round-to-nearest ties-to-even. Separate CLs include
hardware implementations when available. Otherwise, this software
fallback is given as the default implementation.
Specifically,
- arm64, ppc64, s390x: Uses the FMA instruction provided by all
of these ISAs.
- mips[64][le]: Falls back to this software implementation. Only
release 6 of the ISA includes a strict FMA instruction with
MADDF.D (not implementation defined). Because the number of R6
processors in the wild is scarce, the assembly implementation
is left as a future optimization.
- x86: Guards the use of VFMADD213SD by checking cpu.X86.HasFMA.
- arm: Guards the use of VFMA by checking cpu.ARM.HasVFPv4.
- software fallback: Uses mostly integer arithmetic except
for input that involves Inf, NaN, or zero.
Updates #25819.
Change-Id: Iadadff2219638bacc9fec78d3ab885393fea4a08
Reviewed-on: https://go-review.googlesource.com/c/go/+/127458
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Bryan C. Mills [Fri, 18 Oct 2019 20:52:12 +0000 (16:52 -0400)]
cmd/go: add a flag to avoid creating unwritable directories in the module cache
This change adds the '-modcacherw' build flag, which leaves
newly-created directories (but not the files!) in the module cache
read-write instead of making them unwritable.
Fixes #31481
Change-Id: I7c21a53dd145676627c3b51096914ce797991d99
Reviewed-on: https://go-review.googlesource.com/c/go/+/202079
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Jay Conrod <jayconrod@google.com>
Ian Lance Taylor [Fri, 5 Apr 2019 18:42:37 +0000 (11:42 -0700)]
runtime: change read and write to return negative errno value
The internal read and write functions used to return -1 on error;
change them to return a negative errno value instead.
This will be used by later CLs in this series.
For most targets this is a simplification, although for ones that call
into libc it is a complication.
Updates #27707
Change-Id: Id02bf9487f03e7e88e4f2b85e899e986738697ad
Reviewed-on: https://go-review.googlesource.com/c/go/+/171823
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Michael Knyszek <mknyszek@google.com>
Tobias Klauser [Mon, 21 Oct 2019 12:02:18 +0000 (14:02 +0200)]
vendor: re-vendor golang.org/x/net to fix Dragonfly build
This picks up CL 202317 which fixes golang.org/x/net for Dragonfly
master (upcoming 5.8 release). Also re-enable the interface tests
disabled in CL 201482.
Vendored using:
$ go get golang.org/x/net@24d2ffbea1e8
$ go mod tidy
$ go mod vendor
Giovanni Bajo [Thu, 19 Sep 2019 22:12:15 +0000 (00:12 +0200)]
cmd/compile: in prove, give a different name to each poset
Instead of using a two-slot array and having to remember which
index is the signed poset, and which is the unsigned one, just
use two different variables.
Change-Id: Ic7f7676436c51bf43a182e999a926f8b7f69434b
Reviewed-on: https://go-review.googlesource.com/c/go/+/196678 Reviewed-by: Keith Randall <khr@golang.org>
Alex Brainman [Sat, 19 Oct 2019 06:47:12 +0000 (17:47 +1100)]
cmd/link: use libmsvcrt.a during internal link
When using recent versions of gcc with cgo, internal link fails with
c:\>go test debug/pe
--- FAIL: TestInternalLinkerDWARF (0.94s)
file_test.go:394: building test executable for linktype 2 failed: exit status 2 # command-line-arguments
runtime/cgo(.text): relocation target __acrt_iob_func not defined for ABI0 (but is defined for ABI0)
runtime/cgo(.text): relocation target __acrt_iob_func not defined for ABI0 (but is defined for ABI0)
runtime/cgo(.text): relocation target __acrt_iob_func not defined for ABI0 (but is defined for ABI0)
FAIL
FAIL debug/pe 4.572s
FAIL
It appears that __acrt_iob_func is defined in libmsvcrt.a. And this
change adds libmsvcrt.a to the list of libraries always used byi
internal linker.
libmsvcrt.a also implements __imp___acrt_iob_func. So this change
also prevents rewriting __imp___acrt_iob_func name into
__acrt_iob_func, otherwise we end up with duplicate __acrt_iob_func
symbol error.
Fixes #23649
Change-Id: Ie9864cd17e907501e9a8a3672bbc33e02ca20e5c
Reviewed-on: https://go-review.googlesource.com/c/go/+/197977
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Caleb Spare [Fri, 18 Oct 2019 22:02:49 +0000 (15:02 -0700)]
cmd/vet: re-vendor x/tools upstream for printf verb fix
This pulls in the x/tools fix from
https://go-review.googlesource.com/c/tools/+/202041
so that cmd/vet won't flag %x/%X verbs incorrectly for floating-point
and complex types.
Cherry Zhang [Sat, 19 Oct 2019 02:18:27 +0000 (22:18 -0400)]
runtime: fix past-the-end write of wasmStack
On Wasm, at program start, we set the SP to
wasmStack+sizeof(wasmStack), and start to write on it. This write
is actually past the end of wasmStack. This may scribble to some
other variable next to it in the data segment. Or if wasmStack
happens to be the last object in the data segment, we'll scribble
to unreserved memory and cause the next sysReserve return
non-zero memory. Either way, this is bad. Fix this by subtracting
16 before writing.
Found while debugging the new page allocator (CL 190622 and the
stack) with Michael. We found that on Wasm, the first sysReserve
may return memory with the first a few words being non-zero.
Matthew Dempsky [Thu, 17 Oct 2019 23:31:19 +0000 (16:31 -0700)]
cmd/compile: only escape unsafe.Pointer conversions when -d=checkptr=2
Escaping all unsafe.Pointer conversions for -d=checkptr seems like it
might be a little too aggressive to enable for -race/-msan mode, since
at least some tests are written to expect unsafe.Pointer conversions
to not affect escape analysis.
So instead only enable that functionality behind -d=checkptr=2.
cmd/internal/obj/arm64: add support of NOOP instruction
This patch uses symbol NOOP to support arm64 instruction NOP. In
arm64, NOP stands for that No Operation does nothing, other than
advance the value of the program counter by 4. This instruction
can be used for instruction alignment purposes. This patch uses
NOOP to support arm64 instruction NOP, because we have a generic
"NOP" instruction, which is a zero-width pseudo-instruction.
In arm64, instruction NOP is an alias of HINT #0. This patch adds
test cases for instruction HINT #0.
Alberto Donizetti [Sat, 28 Sep 2019 17:02:15 +0000 (19:02 +0200)]
math/bits: add Rem, Rem32, Rem64
The Div functions in math/bits (Div, Div32, and Div64) compute both
quotients and remainders, but they panic if the quotients do not not
fit a 32/64 uint.
Since, on the other hand, the remainder will always fit the size of
the divisor, it is useful to have Div variants that only compute the
remainder, and don't panic on a quotient overflow.
This change adds to the math/bits package three new functions:
Rem(hi, lo, y uint) uint
Rem32(hi, lo, y uint32) uint32
Rem64(hi, lo, y uint64) uint64
which can be used to compute (hi,lo)%y even when the quotient
overflows the uint size.
Fixes #28970
Change-Id: I119948429f737670c5e5ceb8756121e6a738dbdc
Reviewed-on: https://go-review.googlesource.com/c/go/+/197838
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Robert Griesemer [Thu, 17 Oct 2019 23:28:38 +0000 (16:28 -0700)]
go/types: don't update the underlying type of an imported type
Updating the underlying type of an imported type (even though
is was set to the same type again) leads to a race condition
if the imported package is imported by separate, concurrently
type-checked packages.
Matthew Dempsky [Thu, 17 Oct 2019 20:16:15 +0000 (13:16 -0700)]
cmd/compile: fix -d=checkptr for named unsafe.Pointer types
We need to explicitly convert pointers to unsafe.Pointer before
passing to the runtime checkptr instrumentation in case the user
declared their own type with underlying type unsafe.Pointer.
Updates #22218.
Fixes #34966.
Change-Id: I3baa2809d77f8257167cd78f57156f819130baa8
Reviewed-on: https://go-review.googlesource.com/c/go/+/201782
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Matthew Dempsky [Thu, 17 Oct 2019 19:31:07 +0000 (12:31 -0700)]
cmd/compile: escape unsafe.Pointer conversions when -d=checkptr
This CL tweaks escape analysis to treat unsafe.Pointer(ptr) as an
escaping operation when -d=checkptr is enabled. This allows better
detection of unsafe pointer arithmetic and conversions, because the
runtime checkptr instrumentation can currently only detect object
boundaries for heap objects, not stack objects.
Updates #22218.
Fixes #34959.
Change-Id: I856812cc23582fe4d0d401592583323e95919f28
Reviewed-on: https://go-review.googlesource.com/c/go/+/201781
Run-TryBot: Matthew Dempsky <mdempsky@google.com> Reviewed-by: Keith Randall <khr@golang.org> Reviewed-by: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Keith Randall [Thu, 17 Oct 2019 19:31:53 +0000 (12:31 -0700)]
regexp: skip long-running benchmarks if -short is specified
This CL helps race.bash finish in a reasonable amount of
time. Otherwise the Match/Hard1/32M benchmark takes over 1200 seconds
to finish on arm64, triggering a timeout. With this change the regexp
benchmarks as a whole take only about a minute.
Bryan C. Mills [Thu, 17 Oct 2019 13:57:42 +0000 (09:57 -0400)]
os/exec: re-enable TestExtraFiles checks skipped on various OSes
The issues associated with these skipped checks are closed.
If they are working around unfixed bugs, the issues should remain open.
If they are working around unfixable properties of the system, the skips
should refer to those properties rather than closed issues.
Matthew Dempsky [Wed, 13 Feb 2019 03:40:42 +0000 (19:40 -0800)]
cmd/compile: add -d=checkptr to validate unsafe.Pointer rules
This CL adds -d=checkptr as a compile-time option for adding
instrumentation to check that Go code is following unsafe.Pointer
safety rules dynamically. In particular, it currently checks two
things:
1. When converting unsafe.Pointer to *T, make sure the resulting
pointer is aligned appropriately for T.
2. When performing pointer arithmetic, if the result points to a Go
heap object, make sure we can find an unsafe.Pointer-typed operand
that pointed into the same object.
These checks are currently disabled for the runtime, and can also be
disabled through a new //go:nocheckptr annotation. The latter is
necessary for functions like strings.noescape, which intentionally
violate safety rules to workaround escape analysis limitations.
Fixes #22218.
Change-Id: If5a51273881d93048f74bcff10a3275c9c91da6a
Reviewed-on: https://go-review.googlesource.com/c/go/+/162237
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Eugene Kalinin [Fri, 28 Jun 2019 20:59:49 +0000 (23:59 +0300)]
encoding/json: correct caller's name in encoding errors
1. Change mapencode.encode to use fmt.Error rather than MarshalerError.
MarshalerError refer to MarshalJSON, but mapencode.encode does not use that.
2. Add sourceFunc field to MarshalerError to record the name of the function
that creates the error, so that the Error method can report it correctly.
Fixes #29753
Change-Id: I186c2fac8470ae2f9e300501de3730face642230
Reviewed-on: https://go-review.googlesource.com/c/go/+/184119
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Michael Munday [Wed, 16 Oct 2019 20:49:09 +0000 (21:49 +0100)]
crypto/ecdsa: remove s390x assembly
This a revert of CL 174437 and follow up fix CL 201317.
The s390x assembly in this package makes use of an instruction
(specifically KDSA) which is not supported by the current build
machine. Remove this assembly for now, we can revisit this
functionality once we have a newer build machine and can ensure
that this assembly is well tested.
Emmanuel T Odeke [Wed, 2 Oct 2019 20:36:27 +0000 (13:36 -0700)]
io/ioutil: support predictable prefix,suffix for TempDir with *
Allow TempDir to create directories with predictable
prefixes and suffixes, separated by the last "*", for example:
"prefix*suffix"
will now expand to
"prefix" + <RANDOM_VALUE> + "suffix"
RELNOTE=yes
Fixes #33805.
Change-Id: I85fa73ae6a684ce820d1810c82a60765eb9c4a42
Reviewed-on: https://go-review.googlesource.com/c/go/+/198488
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Dan Scales [Mon, 24 Jun 2019 19:59:22 +0000 (12:59 -0700)]
cmd/compile, cmd/link, runtime: make defers low-cost through inline code and extra funcdata
Generate inline code at defer time to save the args of defer calls to unique
(autotmp) stack slots, and generate inline code at exit time to check which defer
calls were made and make the associated function/method/interface calls. We
remember that a particular defer statement was reached by storing in the deferBits
variable (always stored on the stack). At exit time, we check the bits of the
deferBits variable to determine which defer function calls to make (in reverse
order). These low-cost defers are only used for functions where no defers
appear in loops. In addition, we don't do these low-cost defers if there are too
many defer statements or too many exits in a function (to limit code increase).
When a function uses open-coded defers, we produce extra
FUNCDATA_OpenCodedDeferInfo information that specifies the number of defers, and
for each defer, the stack slots where the closure and associated args have been
stored. The funcdata also includes the location of the deferBits variable.
Therefore, for panics, we can use this funcdata to determine exactly which defers
are active, and call the appropriate functions/methods/closures with the correct
arguments for each active defer.
In order to unwind the stack correctly after a recover(), we need to add an extra
code segment to functions with open-coded defers that simply calls deferreturn()
and returns. This segment is not reachable by the normal function, but is returned
to by the runtime during recovery. We set the liveness information of this
deferreturn() to be the same as the liveness at the first function call during the
last defer exit code (so all return values and all stack slots needed by the defer
calls will be live).
I needed to increase the stackguard constant from 880 to 896, because of a small
amount of new code in deferreturn().
The -N flag disables open-coded defers. '-d defer' prints out the kind of defer
being used at each defer statement (heap-allocated, stack-allocated, or
open-coded).
Cost of defer statement [ go test -run NONE -bench BenchmarkDefer$ runtime ]
With normal (stack-allocated) defers only: 35.4 ns/op
With open-coded defers: 5.6 ns/op
Cost of function call alone (remove defer keyword): 4.4 ns/op
Text size increase (including funcdata) for go cmd without/with open-coded defers: 0.09%
The average size increase (including funcdata) for only the functions that use
open-coded defers is 1.1%.
The cost of a panic followed by a recover got noticeably slower, since panic
processing now requires a scan of the stack for open-coded defer frames. This scan
is required, even if no frames are using open-coded defers:
Cost of panic and recover [ go test -run NONE -bench BenchmarkPanicRecover runtime ]
Without open-coded defers: 62.0 ns/op
With open-coded defers: 255 ns/op
A CGO Go-to-C-to-Go benchmark got noticeably faster because of open-coded defers:
CGO Go-to-C-to-Go benchmark [cd misc/cgo/test; go test -run NONE -bench BenchmarkCGoCallback ]
Without open-coded defers: 443 ns/op
With open-coded defers: 347 ns/op
Fazlul Shahriar [Wed, 16 Oct 2019 02:58:40 +0000 (22:58 -0400)]
net: fix multicast and IPv6 related issues on Plan 9
Fix issues that make these tests pass:
- TestDialerLocalAddr: return error if local address is not IPv4 for
"tcp4" network.
- TestInterfaceAddrs, TestInterfaceUnicastAddrs: don't assume each
interface has only one address. It may have more than one or none.
- TestConcurrentPreferGoResolversDial: should be skipped on Plan 9.
- TestListenMulticastUDP: remove IP from `announce` command and don't
mix IPv4 address with IPv6 address in `addmulti` command.
Agniva De Sarker [Sat, 12 Oct 2019 15:43:28 +0000 (21:13 +0530)]
cmd/internal/obj/wasm,cmd/link/internal/wasm: add fast path for writeUleb128
While building a simple hello world binary, there are total 858277 calls
to writeUleb during the assembler phase out of which 836625 (97%) are less than 7 bits.
Using a simple micro-benchmark like this:
func BenchmarkUleb(b *testing.B) {
var buf bytes.Buffer
for i := 0; i < b.N; i++ {
writeUleb128(&buf, 42)
buf.Reset()
}
}
We get the following results with the fast path enabled.
name old time/op new time/op delta
Uleb-4 8.45ns ± 2% 7.51ns ± 2% -11.16% (p=0.000 n=10+10)
Applying the time taken to the number of calls, we get roughly 6% improvement
in total time taken for writeUleb128.
We also apply the change to the function in linker to make it consistent.
Cherry Zhang [Tue, 15 Oct 2019 21:51:32 +0000 (17:51 -0400)]
cmd/compile: on Wasm and AIX, let deferred nil function panic at invocation
The Go spec requires
If a deferred function value evaluates to nil, execution
panics when the function is invoked, not when the "defer"
statement is executed.
On Wasm and AIX, currently we actually emit a nil check at the
point of defer statement, which will make it panic too early.
This CL fixes this.
Also, on Wasm, now the nil function will be passed through
deferreturn to jmpdefer, which does an explicit nil check and
calls sigpanic if it is nil. This sigpanic, being called from
assembly, is ABI0. So change the assembler backend to also
handle sigpanic in ABI0.
Robert Griesemer [Tue, 15 Oct 2019 20:44:22 +0000 (13:44 -0700)]
math/big: make Rat accessors safe for concurrent use
Do not modify the underlying Rat denominator when calling
one of the accessors Float32, Float64; verify that we don't
modify the Rat denominator when calling Inv, Sign, IsInt, Num.
Fixes #34919.
Reopens #33792.
Change-Id: Ife6d1252373f493a597398ee51e7b5695b708df5
Reviewed-on: https://go-review.googlesource.com/c/go/+/201205 Reviewed-by: Ian Lance Taylor <iant@golang.org>