Matthew Dempsky [Tue, 27 Mar 2018 22:35:51 +0000 (15:35 -0700)]
cmd/compile: disable instrumentation for no-race packages earlier
Rather than checking for each function whether the package supports
instrumentation, check once up front.
Relatedly, tweak the logic for preventing inlining calls to runtime
functions from instrumented packages. Previously, we simply disallowed
inlining runtime functions altogether when instrumenting. With this
CL, it's only disallowed from packages that are actually being
instrumented. That is, now intra-runtime calls can be inlined.
Giovanni Bajo [Mon, 2 Apr 2018 21:13:43 +0000 (23:13 +0200)]
go/types: fix column reporting of invalid selector names
Fixes #24645
Change-Id: I914674451b6667c3ebaf012893503d9de58991ee
Reviewed-on: https://go-review.googlesource.com/104155
Run-TryBot: Giovanni Bajo <rasky@develer.com> Reviewed-by: Robert Griesemer <gri@golang.org>
compress/gzip: do not count header bytes written in Write
Before, if an underlying writer errored within 10 bytes (plus any gzip
header metadata), a gzip.Write would erroneously report up to 10 bytes
written that were not actually written of the input slice. This is
especially problematic when the input slice is less than 10 bytes.
The error came from counting the 10 header byte write. If writing the
header is completely successful, the 10 bytes written is overridden by
the flate write with the input slice.
This removes counting the 10 required header bytes, and also changes the
return to use zero until the slice is used.
The old Write could return one byte written when it actually was not.
This is difficult to verify because the smallest input slice is one
byte; a test checking that the input slice was the byte written would be
quite involved. Thankfully, gzip's minimum header write is 10 bytes. If
we test that two bytes are not falsely written, we indirectly cover the
one byte case.
Fixes #24625
Change-Id: I1c1f8cd791e0c4cffc22aa8acd95186582c832ba
Reviewed-on: https://go-review.googlesource.com/103861 Reviewed-by: Joe Tsai <joetsai@google.com>
Run-TryBot: Joe Tsai <joetsai@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
When test/run script was removed, these two tests
were changed to be executed by test/run.go.
Because errchk does not exit with non-zero status on
errors, they were silently failing for a while.
This change makes 2 things:
1. Compile tested packages in GOROOT/test to match older runner script
behavior (strictly required only in bug345, optional in bug248)
2. Check command output with "(?m)^BUG" regexp.
It approximates older `grep -q '^BUG' that was used before.
Matthew Dempsky [Sat, 31 Mar 2018 23:49:07 +0000 (16:49 -0700)]
cmd/compile: refactor constant rewriting
Extract all rewrite-to-OLITERAL expressions to use a single setconst
helper function.
Does not pass toolstash-check for two reasons:
1) We now consistently clear Left/Right/etc when rewriting Nodes into
OLITERALs, which results in their inlining complexity being correctly
computed. So more functions can now be inlined.
2) We preserve Pos, so PC line tables change somewhat.
Change-Id: I2b5c293bee7c69c2ccd704677f5aba4ec40e3155
Reviewed-on: https://go-review.googlesource.com/103860
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
When the flag package first begin printing nonzero defaults, the test
was against a fixed set of string representations of zero values.
This worked until the string representation of a time.Duration
changed from "0" to "0s", causing the zero Duration to register as
nonzero. The flag package then added reflect-based code that fell back
to the old test. This failed to work when a nonzero default for a flag
happened to be the string representation of one the original fixed set
of zero values in the original test. This change removes the original
test, allowing the reflect-based code to be the only deciding factor.
Mikio Hara [Tue, 20 Feb 2018 01:21:36 +0000 (10:21 +0900)]
net: improve test coverage for syscall.RawConn
An application using syscall.RawConn in a particular way must take
account of the operating system or platform-dependent behavior.
This change consolidates duplicate code and improves the test coverage
for applications that use socket options.
Bryan Chan [Fri, 23 Mar 2018 08:14:25 +0000 (04:14 -0400)]
cmd/compile/internal/ssa: handle symbol address comparisons consistently
CL 38338 introduced SSA rules to optimize two types of pointer equality
tests: a pointer compared with itself, and comparison of addresses taken
of two symbols which may have the same base. This patch adds rules to
apply the same optimization to pointer inequality tests, which also ensures
that two pointers to zero-width types cannot be both equal and unequal
at the same time.
Daniel Martí [Thu, 8 Mar 2018 21:47:17 +0000 (21:47 +0000)]
cmd/compile/internal/ssa: add initial README
This is the first version of an introductory document that should help
developers who want to get started with this package.
I recently started poking around this part of the compiler, and was
confused by a few basic ideas such as memory arguments. I also hadn't
heard about GOSSAFUNC until another developer pointed it out. Both of
those are essential if one wants to do any non-trivial work here.
This document can of course be expanded with more pointers and tips to
better understand this package's code and behavior. Its intent is not to
cover all of its features; but it should be enough for most developers
to start playing with it without extensive compiler experience.
Samuel Tan [Fri, 30 Mar 2018 00:50:46 +0000 (17:50 -0700)]
html/template: fix one more lint naming error
Change-Id: I629d89d5065271f3b92dde8b12f0e743c9bde8f0
Reviewed-on: https://go-review.googlesource.com/103595 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
CL 40693 added concurrent backend compilation support,
and used it for user-provided functions.
Autogenerated functions were still compiled serially.
This CL brings them into the fold.
As of this CL, when requested,
no functions are compiled serially.
There generally aren't many autogenerated functions.
When there are, this CL can help a lot,
because autogenerated functions are usually short.
Many short functions is the best case scenario
for concurrent compilation; see CL 41192.
One example of such a package comes from Dave Cheney's benchjuju:
github.com/juju/govmomi/vim25/types.
It has thousands of autogenerated functions.
This CL improves performance on the entire benchmark
by around a second on my machine at c=8, or about ~5%.
Richard Musiol [Thu, 8 Mar 2018 23:06:33 +0000 (00:06 +0100)]
cmd/compile: add SSA config options useAvg and useHmul
This commit allows architectures to disable optimizations that need the
Avg* and Hmul* operations.
WebAssembly has no such operations, so using them as an optimization
but then having to emulate them with multiple instructions makes no
sense, especially since the WebAssembly compiler may do the same
optimizations internally.
This is the first commit of a series that will add WebAssembly
as an architecture target. The design document can be found at
https://docs.google.com/document/d/131vjr4DH6JFnb-blm_uRdaC0_Nv3OUwjEY5qVCxCup4.
The GOARCH name "wasm" is the official abbreviation of WebAssembly.
The GOOS name "js" got chosen because initially the host environment
that executes WebAssembly bytecode will be web browsers and Node.js,
which both use JavaScript to embed WebAssembly. Other GOOS values
may be possible later, see:
https://github.com/WebAssembly/design/blob/master/NonWeb.md
Richard Musiol [Sun, 4 Mar 2018 12:53:50 +0000 (13:53 +0100)]
misc/wasm: add scripts for running WebAssembly binaries
This commit adds scripts for running the WebAssembly binaries that the
Go compiler will produce.
The script go_js_wasm_exec uses Node.js to run the binaries. Adding it
to PATH will enable "go run" and "go test" to work for js/wasm
without having to manually provide the -exec flag.
See https://golang.org/cmd/go/#hdr-Compile_and_run_Go_program
for more information.
The web page wasm_exec.html is an example on how to run the same
binaries in a web browser.
Both scripts use wasm_exec.js as a shared library.
Ben Burkert [Thu, 29 Mar 2018 22:04:58 +0000 (15:04 -0700)]
net: map context errors from aborted dial to internal net package errors
Map the error returned when a dial is aborted from the context package
error to the internal net package error. For example, context.Canceled
errors map to errCanceled, and context.DeadlineExceeded errors map to
poll.ErrTimeout.
Fixes #23648
Change-Id: Idf9d3d08052d540740c0b054503aaed931dc5b1e
Reviewed-on: https://go-review.googlesource.com/103518
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Joe Tsai [Thu, 30 Nov 2017 01:11:45 +0000 (17:11 -0800)]
os: add ModeIrregular flag
There is currently no way for os.FileMode.IsRegular to report false
without being one of the following types:
ModeDir | ModeSymlink | ModeNamedPipe | ModeSocket | ModeDevice
This makes it difficult for custom implementations of os.FileInfo to return
a Mode that is explicitly not regular without resorting to setting one
of the types listed above. However, every one of the aforementioned types
are ill-suited as a general-purpose "not regular" file type.
Thus, add a ModeIrregular to serve exactly for that purpose.
The ModeIrregular type carries no information other than the fact that the
file is not regular.
Wèi Cōngruì [Thu, 11 Jan 2018 05:11:12 +0000 (13:11 +0800)]
math: fix Ldexp when result is below ldexp(2, -1075)
Before this change, the smallest result Ldexp can handle was
ldexp(2, -1075), which is SmallestNonzeroFloat64.
There are some numbers below it should also be rounded to
SmallestNonzeroFloat64. The change fixes this.
Fixes #23407
Change-Id: I76f4cb005a6e9ccdd95b5e5c734079fd5d29e4aa
Reviewed-on: https://go-review.googlesource.com/87338
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
Ian Lance Taylor [Thu, 29 Mar 2018 22:00:47 +0000 (15:00 -0700)]
regexp/syntax: update perl script to preserve \s behavior
Incorporate https://code-review.googlesource.com/#/c/re2/+/3050/ from
the re2 repository. Description of that change:
Preserve the original behaviour of \s.
Prior to Perl 5.18, \s did not match vertical tab. Bake that into
make_perl_groups.pl as an override so that perl_groups.cc retains
its current definitions when rebuilt with newer versions of Perl.
This fixes make_perl_groups.pl to generate an unchanged perl_groups.go
with perl versions 5.18 and later.
Fixes #22057
Change-Id: I9a56e9660092ed6c1ff1045b4a3847de355441a7
Reviewed-on: https://go-review.googlesource.com/103517
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Alberto Donizetti [Thu, 29 Mar 2018 16:52:12 +0000 (18:52 +0200)]
test: avoid touching GOOS/GOARCH in codegen driver
This change modifies the codegen test harness driver so that it no
longer modifies the environment GOOS/GOARCH, since that seems to cause
flakiness in other concurrently-running tests.
The change also enables the codegen tests in run.go.
cmd/compile: prefer to evict a rematerializable register
This resolves a long-standing regalloc TODO:
If you must evict a register, choose to evict a register
containing a rematerializable value, since that value
won't need to be spilled.
Provides very minor performance and size improvements.
Paul Jolly [Thu, 29 Mar 2018 14:29:23 +0000 (15:29 +0100)]
cmd/go: make generate pass correct GOPACKAGE to XTest files
The existing behaviour of go generate is to pass GOPACKAGE=p
to all package files, including XTest files. This however is
incorrect as the package name for the XTest files is p_test.
Fixes #24594
Change-Id: I96b6e5777ec511cdcf1a6267a43f4d8c544c4af3
Reviewed-on: https://go-review.googlesource.com/103415
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Alberto Donizetti [Thu, 29 Mar 2018 12:35:57 +0000 (14:35 +0200)]
test: update list of escape reasons
The escape_because.go test file (which tests the "because" escape
explainations printed by `-m -m`) cointains a machine-generated list
of all the escape reasons seen in the escape tests.
The list appears to be outdated; moreove a new escape reason was added
in CL 102895. This change re-generates the list.
Change-Id: Idc721c6bbfe9516895b5cf1e6d09b77deda5a3dd
Reviewed-on: https://go-review.googlesource.com/103375 Reviewed-by: David Chase <drchase@google.com>
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Andrew Bonventre [Wed, 28 Mar 2018 21:42:01 +0000 (17:42 -0400)]
doc: document Go 1.9.5
Change-Id: I8f12b188d2e13555e1d7634b9cc0319fc737cb1d
Reviewed-on: https://go-review.googlesource.com/103216 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Andrew Bonventre [Wed, 28 Mar 2018 21:51:05 +0000 (17:51 -0400)]
doc: document Go 1.10.1
Change-Id: Ibba5d4fd8124db9b1e4e50b8f37366f8dd23f47e
Reviewed-on: https://go-review.googlesource.com/103217 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Andrew Bonventre [Fri, 23 Mar 2018 20:40:15 +0000 (16:40 -0400)]
net/http/pprof: harden handler responses
A very small number of old browsers consider content as HTML
even when it is explicitly stated in the Content-Type header
that it is not. If content served is based on user-supplied
input, then an XSS is possible. Introduce three mitigations:
+ Don't reflect user input in error strings
+ Set a Content-Disposition header when requesting a resource
that should never be displayed in a browser window
+ Set X-Content-Type-Options: nosniff on all responses
Hana Kim [Wed, 28 Mar 2018 18:19:09 +0000 (14:19 -0400)]
internal/trace: fix double counting in span analysis
When snapshotting the execution time stats of a goroutine
we take into account the intermediate values kepts in the GDesc's
gdesc field. At the end of goroutine analysis, we go through all
goroutines and replace the GExecutionStat with the new snapshot.
Here the mistake was that we replaced the GExecutionStat with
the value that reflects the intermediate values, but did clear
the intermediate values. So, when the following finalizeActiveSpans
runs, it trieds to add the intermediate values again (double-counting)
when taking a snapshot.
Fix the issue by using the finalized GExecutionStat instead of
recomputing the snapshot. While we are here, refactor the finalization
logic so it can be used when processing GoEnd, GoStop events.
Change-Id: Ibdb342214c29b65d4ea37e52b1f8b83f1fb20143
Reviewed-on: https://go-review.googlesource.com/103157 Reviewed-by: Peter Weinberger <pjw@google.com>
Agniva De Sarker [Wed, 28 Mar 2018 08:15:26 +0000 (13:45 +0530)]
os: keep the $ if a variable is not detected
If the character after $ cannot be detected as a valid
variable declaration, do not gobble the $.
Fixes #24345
Change-Id: Iec47be1f2e4f8147b8ceb64c30778eae8045b58f
Reviewed-on: https://go-review.googlesource.com/103055
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
cmd/compile/internal/gc: speed up arith const tests
This change reduces the time to run this test on my machine from
real 0m2.491s
user 0m3.020s
sys 0m0.331s
to
real 0m0.237s
user 0m0.180s
sys 0m0.173s
This will make it reasonable to add more constants to the test.
I am also hopeful that it might help a bit with intermittent
cmd/compile/internal/gc test timeouts on the build dashboard
on the slower builders.
The time savings are entirely in compilation time,
by avoiding generating one giant func main.
Instead, generate tables of tests to be run,
which are translated into static data,
and then loop over those tests.
While we're here, do some minor cleanup:
* Remove the _ssa suffix on function names,
as that was only needed during ssa bootstrapping.
* Clean up error handling during test generation.
* Make functions single-line, to reduce future diff sizes.
Diffing giant files is slow.
Alberto Donizetti [Wed, 28 Mar 2018 10:19:46 +0000 (12:19 +0200)]
cmd/compile: print accurate escape reason for non-const-length slices
This change makes `-m -m` print a better explanation for the case
where a slice is marked as escaping and heap-allocated because it
has a non-constant len/cap.
Jakob Weisblat [Fri, 18 Aug 2017 16:24:54 +0000 (18:24 +0200)]
crypto/x509: improve error when parsing bad ECDSA cert
When parsing an ECDSA certificate, improve the error message upon
failing to parse the curve as a named curve, rather than returning
the original ASN1 error.
Ian Lance Taylor [Sat, 17 Feb 2018 03:08:59 +0000 (19:08 -0800)]
cmd/go: if -race, don't run coverage on runtime packages
Don't compile the runtime packages with coverage when using the race
detector. The user can, perhaps accidentally, request coverage for the
runtime by using -coverpkg=all. If using the race detector, the
runtime package coverage will call into the race detector before it
has been initialized. This will cause the program to crash
mysteriously on startup.
Fixes #23882
Change-Id: I9a63867a9138797d8b8afb0856ae21079accdb27
Reviewed-on: https://go-review.googlesource.com/94898
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Andrew Bonventre <andybons@golang.org>
Filippo Valsorda [Sun, 28 Jan 2018 22:30:48 +0000 (23:30 +0100)]
crypto/tls: parse certificate first in X509KeyPair to get better errors
parsePrivateKey can't return useful error messages because it does trial
decoding of multiple formats. Try ParseCertificate first in case it
offers a useful error message.
erifan01 [Tue, 19 Dec 2017 07:49:10 +0000 (07:49 +0000)]
math: optimize Exp and Exp2 on arm64
This CL implements Exp and Exp2 with arm64 assembly. By inlining Ldexp and
using fused instructions(fmadd, fmsub, fnmsub), this CL helps to improve
the performance of functions Exp, Exp2, Sinh, Cosh and Tanh.
Ben Shi [Sat, 10 Mar 2018 15:39:02 +0000 (15:39 +0000)]
cmd/internal/obj/arm64: add LDPW/LDPSW/STPW to arm64 assembler
1. STPW stores the lower 32-bit words of a pair of registers to memory.
2. LDPW loads two 32-bit words from memory, zero extends them to 64-bit,
and then copies to a pair of registers.
3. LDPSW does the same as LDPW, except a sign extension.
This CL implements those 3 instructions and adds test cases.
Matthew Dempsky [Tue, 27 Mar 2018 08:05:48 +0000 (01:05 -0700)]
cmd/compile: fix method set computation for shadowed methods
In expandmeth, we call expand1/expand0 to build a list of all
candidate methods to promote, and then we use dotpath to prune down
which names actually resolve to a promoted method and how.
However, previously we still computed "followsptr" based on the
expand1/expand0 traversal (which is depth-first), rather than
dotpath (which is breadth-first). The result is that we could
sometimes end up miscomputing whether a particular promoted method
involves a pointer traversal, which could result in bad code
generation for method trampolines.
Fixes #24547.
Change-Id: I57dc014466d81c165b05d78b98610dc3765b7a90
Reviewed-on: https://go-review.googlesource.com/102618 Reviewed-by: Robert Griesemer <gri@golang.org>
HaraldNordgren [Sat, 6 Jan 2018 20:30:42 +0000 (21:30 +0100)]
cmd/go: print each import error only once
This change prevents import errors from being printed multiple times.
Creating a bare-bones package 'p' with only one file importing itself
and running 'go build p', the current implementation gives this error
message:
can't load package: import cycle not allowed
package p
imports p
import cycle not allowed
package p
imports p
With this change we will show the message only once.
Updates #23295
Change-Id: I653b34c1c06c279f3df514f12ec0b89745a7e64a
Reviewed-on: https://go-review.googlesource.com/86535 Reviewed-by: Harald Nordgren <haraldnordgren@gmail.com> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Meng Zhuo [Mon, 12 Mar 2018 07:32:28 +0000 (07:32 +0000)]
runtime: use vDSO for clock_gettime on linux/arm64
Use the __vdso_clock_gettime fast path via the vDSO on linux/arm64 to
speed up nanotime and walltime. This results in the following
performance improvement for time.Now on Cavium ThunderX:
name old time/op new time/op delta
TimeNow 442ns ± 0% 163ns ± 0% -63.16% (p=0.000 n=10+10)
Since the initial workList slice allocation is always on the stack
(as the variable does not escape), we can aggressively pre-allocate a
big backing array at (almost) no cost. This will save several
allocations in liveValues calls that end up having a large workList,
with no performance penalties for calls that have a small workList.
David du Colombier [Mon, 26 Mar 2018 08:41:11 +0000 (10:41 +0200)]
os: fix TestDevNullFile on Plan 9
CL 102457 added TestDevNullFile. However, this
test is failing on Plan 9, because it checks
that /dev/null is a character device while there
are no special files on Plan 9.
We fix this issue by changing Stat to consider
all files served by the console device (#c)
as character devices.
Fixes #24534.
Change-Id: I1c60cdf25770358b908790b3fb71910fa914dec0
Reviewed-on: https://go-review.googlesource.com/102424
Run-TryBot: David du Colombier <0intro@gmail.com> Reviewed-by: Alex Brainman <alex.brainman@gmail.com> Reviewed-by: Rob Pike <r@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Tim Wright [Tue, 27 Mar 2018 02:00:15 +0000 (19:00 -0700)]
cmd/go, cmd/link, runtime: enable PIE build mode, cgo race tests on FreeBSD
Fixes #24546
Change-Id: I99ebd5bc18e5c5e42eee4689644a7a8b02405f31
Reviewed-on: https://go-review.googlesource.com/102616
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Brad Fitzpatrick [Mon, 26 Mar 2018 06:56:39 +0000 (06:56 +0000)]
all: use strings.Builder instead of bytes.Buffer where appropriate
I grepped for "bytes.Buffer" and "buf.String" and mostly ignored test
files. I skipped a few on purpose and probably missed a few others,
but otherwise I think this should be most of them.
Updates #18990
Change-Id: I5a6ae4296b87b416d8da02d7bfaf981d8cc14774
Reviewed-on: https://go-review.googlesource.com/102479
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Adam Langley [Wed, 28 Feb 2018 19:35:54 +0000 (11:35 -0800)]
crypto/x509: matching any requested EKU should be sufficient.
The documentation was unclear here and I misremembered the behaviour and
changed it in 1.10: it used to be that matching any EKU was enough but
1.10 requires that all EKUs match.
Restore 1.9 behaviour and clarify the documentation to make it official.
Austin Clements [Sun, 25 Mar 2018 16:20:57 +0000 (12:20 -0400)]
cmd/compile: clarify unsigned interpretation of AuxInt
The way Value.AuxInt represents unsigned numbers is currently
documented in genericOps.go, which is not the most obvious place for
it. Move that documentation to Value.AuxInt. Furthermore, to make it
harder to use incorrectly, introduce a Value.AuxUnsigned accessor that
returns the zero-extended value of Value.AuxInt.
David Chase [Mon, 26 Mar 2018 17:37:00 +0000 (13:37 -0400)]
cmd/compile: invoke gdb more carefully in ssa/debug_test.go
Gdb can be sensitive to contents of .gdbinit, and to run
this test properly needs to have runtime/runtime-gdb.py
on the auto load safe path. Therefore, turn off .gdbinit
loading and explicitly add $GOROOT/runtime to the safe
load path.
This should make ssa/debug_test.go run more consistently.
Updates #24464.
Change-Id: I63ed17c032cb3773048713ce51fca3a3f86e79b6
Reviewed-on: https://go-review.googlesource.com/102598
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
A newish check for branch-likely on single-successor blocks
caught a case where the preemption-check inserter was
setting "likely" on an unconditional branch.
Fixed by checking for that case before setting likely.
Also removed an overconservative restriction on parallel
compilation for GOEXPERIMENT=preemptibleloops; it works
fine, it is just another control-flow transformation.
Michael Munday [Mon, 26 Mar 2018 16:00:29 +0000 (17:00 +0100)]
cmd/compile: use 32-bit comparisons where possible on s390x
We use 32-bit operations for 8- and 16-bit arithmetic, so use them
for comparisons too. This won't change performance but it is more
consistent and makes testing 8- and 16-bit comparison codegen
slightly more straightforward (for follow up CL).
Also fix a typo and add some additional double sign and zero
extension rules to remove the operations inserted by the comparison
rules.
Hana Kim [Thu, 15 Mar 2018 02:22:31 +0000 (22:22 -0400)]
internal/trace: compute span stats as computing goroutine stats
Move part of UserSpan event processing from cmd/trace.analyzeAnnotations
to internal/trace.GoroutineStats that returns analyzed per-goroutine
execution information. Now the execution information includes list of
spans and their execution information.
cmd/trace.analyzeAnnotations utilizes the span execution information
from internal/trace.GoroutineStats and connects them with task
information.
Change-Id: Ib7f79a3ba652a4ae55cd81ea17565bcc7e241c5c
Reviewed-on: https://go-review.googlesource.com/101917
Run-TryBot: Hyang-Ah Hana Kim <hyangah@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Heschi Kreinick <heschi@google.com> Reviewed-by: Peter Weinberger <pjw@google.com>