cmd/compile: in escape analysis, use element type for OIND of slice
The escape analysis models the flow of "content" of X with a
level of "indirection" (OIND node) of X. This content can be
pointer dereference, or slice/string element. For the latter
case, the type of the OIND node should be the element type of
the slice/string. This CL fixes this. In particular, this
matters when the element type is pointerless, where the data
flow should not cause any escape.
Fixes #15730.
Change-Id: Iba9f92898681625e7e3ddef76ae65d7cd61c41e0
Reviewed-on: https://go-review.googlesource.com/107597 Reviewed-by: David Chase <drchase@google.com>
The documentation for %v behavior for compound objects uses an ellipsis
to indicate indefinite lenght of elements. This is done for struct
fields as well as elements of arrays and slices. This adds the missing
ellipsis for maps.
Matthew Dempsky [Tue, 17 Apr 2018 21:56:29 +0000 (14:56 -0700)]
cmd/compile: use empty package name for runtime/{race,msan}
These fake imports are just so we can emit build dependencies for the
linker, so the package name isn't really necessary. Also, the package
import logic assumes that if we have the name for a package, then
we've already read some package data for it.
Using the empty string allows the importers to correctly populate it
the first time these packages are seen in package export data.
Michael Fraenkel [Tue, 17 Apr 2018 22:38:06 +0000 (18:38 -0400)]
encoding/json: simplify dominantField
Fixes #18037
Change-Id: I20e27bcc013b00b726eb348daf5ca86b138ddcc2
Reviewed-on: https://go-review.googlesource.com/107598
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Ilya Tocar [Thu, 22 Mar 2018 21:53:52 +0000 (16:53 -0500)]
compress/flate: optimize huffSym
By using local variables and assigning them back to decompressor
at the end of huffSym, we allow compiler to keep them in registers
and avoid reloading/storing them repeatedly. To archive this,
moreBits was inlined and specialized to work with local variables.
Also move EOF error conversion to helper function, to make inlined
part of moreBits more readable. Together this results in nice speed-up:
cmd/internal/obj/arm, runtime: delete old ARM softfloat code
CL 106735 changed to the new softfloat support on GOARM=5.
ARM assembly code that uses FP instructions not guarded on GOARM,
if any, will break. The easiest way to fix is probably to use Go
implementation on GOARM=5, like
MOVB runtime·goarm(SB), R11
CMP $5, R11
BEQ arm5
... FP instructions ...
RET
arm5:
CALL or JMP to Go implementation
Alberto Donizetti [Mon, 16 Apr 2018 11:35:35 +0000 (13:35 +0200)]
fmt: document that Scan etc. accept 'p' format floats
In the Scan functions documentation, clarify that for float/complex
literals in scientific notation both decimal (e) and binary (p)
exponents are accepted.
Fixes #24453
Change-Id: Ic6dcdb0c36e088ffb65177038aff7a57ab56b805
Reviewed-on: https://go-review.googlesource.com/107416 Reviewed-by: Rob Pike <r@golang.org>
Tim Cooper [Sat, 27 Jan 2018 02:29:55 +0000 (22:29 -0400)]
os/exec: remove "binary" when talking about executables
The use of binary was incorrect as executable files can also be scripts.
The docs for Error are also reworded. The old docs implied that Error was
returned when attempting to start an executable, which is not correct: it
was returned by LookPath when the file was not found or did not have the
attributes of an executable.
erifan01 [Tue, 28 Nov 2017 03:38:03 +0000 (03:38 +0000)]
math: add a testcase for Mod and Remainder respectively
One might try to implement the Mod or Remainder function with the expression
x - TRUNC(x/y + 0.5)*y, but in fact this method is wrong, because the rounding
of (x/y + 0.5) to initialize the argument of TRUNC may lose too much precision.
However, the current test cases can not detect this error. This CL adds two
test cases to prevent people from continuing to do such attempts.
Change-Id: I6690f5cffb21bf8ae06a314b7a45cafff8bcee13
Reviewed-on: https://go-review.googlesource.com/84275 Reviewed-by: Robert Griesemer <gri@golang.org>
Rob Pike [Mon, 16 Apr 2018 21:16:40 +0000 (07:16 +1000)]
doc/contribute.html: adjust wording from previous CL
The previous CL, 107197, overclarified the need for short subject
lines. Tweak the wording to be a guideline (keep it short) rather
than a limit (76 characters), which is more the Go way.
Michael Munday [Thu, 12 Apr 2018 14:24:08 +0000 (15:24 +0100)]
cmd/compile: generate constants for NeqPtr, EqPtr and IsNonNil ops
If both inputs are constant offsets from the same pointer then we
can evaluate NeqPtr and EqPtr at compile time. Triggers a few times
during all.bash. Removes a conditional branch in the following
code:
copy(x[1:], x[:])
This branch was recently added as an optimization in CL 94596. We
now skip the memmove if the pointers are equal. However, in the
above code we know at compile time that they are never equal.
Also, when the offset is variable, check if the offset is zero
rather than if the pointers are equal. For example:
copy(x[a:], x[:])
This would now skip the copy if a == 0, rather than if x + a == x.
Finally I've also added a rule to make IsNonNil true for pointers
to values on the stack. The nil check elimination pass will catch
these anyway, but eliminating them here might eliminate branches
earlier.
Change-Id: If72f436fef0a96ad0f4e296d3a1f8b6c3e712085
Reviewed-on: https://go-review.googlesource.com/106635
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
cmd/dist, os/user: test os/user in osusergo mode as well, fix plan9 & windows
Would've caught two regressions so far, and found two more.
Updates #24841
Updates #24845 (package net remains)
Change-Id: I57ad06eb54e04b8c99b5d2e7f24c77ad865224e8
Reviewed-on: https://go-review.googlesource.com/107300 Reviewed-by: Ian Lance Taylor <iant@golang.org>
David Url [Mon, 2 Apr 2018 10:57:59 +0000 (12:57 +0200)]
net/http: omit forbidden Trailer headers from response
Use the vendored ValidTrailerHeader function from x/net/http/httpguts to
check Trailer headers according to RFC 7230. The previous implementation
only omitted illegal Trailer headers defined in RFC 2616.
This CL adds x/net/http/httpguts from CL 104042 (git rev a35a21de97)
Change-Id: I71d0d7f75108bf4ad606733a45bb71baa66a4e91
Reviewed-on: https://go-review.googlesource.com/107197 Reviewed-by: Alberto Donizetti <alb.donizetti@gmail.com> Reviewed-by: Giovanni Bajo <rasky@develer.com>
Daniel Martí [Sun, 8 Apr 2018 12:39:10 +0000 (13:39 +0100)]
cmd/compile/internal/gc: add some Node methods
Focus on "isfoo" funcs that take a *Node, and conver them to isFoo
methods instead. This makes for more idiomatic Go code, and also more
readable func names.
Found candidates with grep, and applied most changes with sed. The funcs
chosen were isgoconst, isnil, and isblank. All had the same signature,
func(*Node) bool.
While at it, camelCase the isliteral and iszero function names. Don't
move these to methods, as they are only used in the backend part of gc,
which might one day be split into a separate package.
Passes toolstash -cmp on std cmd.
Change-Id: I4df081b12d36c46c253167c8841c5a841f1c5a16
Reviewed-on: https://go-review.googlesource.com/105555
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Giovanni Bajo [Sun, 15 Apr 2018 20:53:58 +0000 (22:53 +0200)]
test: small cleanup of code and comments in run.go
While writing CL 107315, I went back and forth for the syntax used for
constraints of build environments in which the architecture did not
support varitants ("plan9/amd64" vs "plan9/amd64/"). I eventually
settled for the latter because the code required less heuristics
(think parsing "plan9/386" vs "386/sse2") but there were a few
leftovers in code and comments.
Change-Id: I9d9a008f3814f9a1642609650eb571e7f1a675cf
Reviewed-on: https://go-review.googlesource.com/107338
Run-TryBot: Giovanni Bajo <rasky@develer.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
runtime: use internal/cpu.X86.HasAVX2 instead of support_avx2
After CL 104636 cpu.X86.HasAVX is set early enough that it can be used
in runtime·memclrNoHeapPointers. Add an offset to use in assembly and
replace the only occurence of support_avx2.
Giovanni Bajo [Sun, 15 Apr 2018 17:00:27 +0000 (19:00 +0200)]
test: run codegen tests on all supported architecture variants
This CL makes the codegen testsuite automatically test all
architecture variants for architecture specified in tests. For
instance, if a test file specifies a "arm" test, it will be
automatically run on all GOARM variants (5,6,7), to increase
the coverage.
The CL also introduces a syntax to specify only a specific
variant (eg: "arm/7") in case the test makes sense only there.
The same syntax also allows to specify the operating system
in case it matters (eg: "plan9/386/sse2").
Fixes #24658
Change-Id: I2eba8b918f51bb6a77a8431a309f8b71af07ea22
Reviewed-on: https://go-review.googlesource.com/107315
Run-TryBot: Giovanni Bajo <rasky@develer.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
CL 106463 introduced what seems to trim the "go:" prefix in pramas
comments twice, so "//go:go:foo" would be handled, too.
So either the strings.TrimPrefix or the [3:]-slicing is not needed.
I opted to remove the [3:]-slicing.
Change-Id: I1325bbc08a9be9ae100c5a7775b0a23f9ed0a419
Reviewed-on: https://go-review.googlesource.com/107256 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
text/tabwriter: reduce allocations from tracking cells
The tabwriter tracks cells on a line-by-line basis.
This can be memory-hungry when working with large input.
This change adds two optimizations.
First, when there's an existing cell slice for a line,
don't overwrite it by appending.
This helps when re-using a Writer,
or when the output is broken into groups,
e.g. by a blank line.
We now re-use that existing cell slice.
Second, we predict that the number of cells in a line
will probably match those of the previous line,
since tabwriter is most often used to format tables.
This has a noticeable impact on cmd/objdump (#24725).
It reduces allocated space by about 55%.
It also speeds it up some.
Using "benchcmd -n 10 Objdump go tool objdump `which go`":
name old time/op new time/op delta
ObjdumpCompile 9.03s ± 1% 8.51s ± 1% -5.81% (p=0.000 n=10+10)
It might also imaginably speed up gofmt on some
large machine-generated code.
Aman Gupta [Tue, 7 Nov 2017 07:02:21 +0000 (23:02 -0800)]
net: implement (*syscall.RawConn).Read/Write on Windows
RawRead assumes the callback will perform either (a) a blocking read
and always return true, (b) a blocking read with a SO_RCVTIMEO set
returning false on WSAETIMEDOUT, or (c) a non-blocking read
returning false on WSAEWOULDBLOCK. In the latter two cases, it uses
a 0-byte overlapped read for notifications from the IOCP runtime
when the socket becomes readable before trying again.
RawWrite assumes the callback will perform blocking write and will
always return true, and makes no effort to tie into the runtime loop.
Alex Brainman [Sat, 14 Apr 2018 04:14:28 +0000 (14:14 +1000)]
cmd/internal/obj/arm64: do not clear environment in TestLarge and TestNoRet
Windows process cannot run properly, if it only has
GOOS and GOARCH environment variables set. It needs
other environment variables. So adjust TestLarge and
TestNoRet to add GOOS and GOARCH to the existing
variables set instead of clearing environment.
Alex Brainman [Sat, 14 Apr 2018 02:48:55 +0000 (12:48 +1000)]
syscall: remove WSAEMSGSIZE
CL 92475 added WSAEMSGSIZE const to syscall package. But there
is already copy of WSAEMSGSIZE in internal/syscall/windows.
So delete syscall.WSAEMSGSIZE
cmd/compile/internal/ssa: prefer non-indexed stores on amd64
We sometimes generate code like this:
v473 = MOVQconst <uintptr> // constant..
v580 = MOVBstoreidx1 <mem> v1056 v473 v819 v491 // ..only used as an index
Rewrite indexed stores to non-indexed version, where possible.
This allows to eliminate const->register move, reducing codesize and lowering register pressure.
Ilya Tocar [Fri, 23 Mar 2018 20:45:03 +0000 (15:45 -0500)]
runtime: avoid division in growslice
Add a special case for power-of-2 sized elements.
We can replace div/mul with left/right shift and avoid expensive operation.
growslice is hotter for short slices of small elements, such as int16, so
add an int16 version for GrowSlice benchmark.
Eric Daniels [Wed, 4 Apr 2018 01:35:46 +0000 (21:35 -0400)]
runtime/traceback: support tracking goroutine ancestor tracebacks with GODEBUG="tracebackancestors=N"
Currently, collecting a stack trace via runtime.Stack captures the stack for the
immediately running goroutines. This change extends those tracebacks to include
the tracebacks of their ancestors. This is done with a low memory cost and only
utilized when debug option tracebackancestors is set to a value greater than 0.
Cherry Zhang [Wed, 14 Feb 2018 21:26:39 +0000 (16:26 -0500)]
all: use new softfloat on GOARM=5
Use the new softfloat support in the compiler, originally added
for softfloat on MIPS. This support is portable, so we can just
use it for softfloat on ARM.
In the old softfloat support on ARM, the compiler generates
floating point instructions, then the assembler inserts calls
to _sfloat before FP instructions. _sfloat decodes the following
FP instructions and simulates them.
In the new scheme, the compiler generates runtime calls to do FP
operations at a higher level. It doesn't generate FP instructions,
and therefore the assembler won't insert _sfloat calls, i.e. the
old mechanism is automatically suppressed.
The old method may be still be triggered with assembly code
using FP instructions. In the standard library, the only
occurance is math/sqrt_arm.s, which is rewritten to call to the
Go implementation instead.
Some significant speedups for code using floating points:
Hana Kim [Thu, 12 Apr 2018 21:06:33 +0000 (17:06 -0400)]
cmd/trace: change span id computation for trace view use
golang.org/cl/102697 attempted to fix the span presentation by utilizing
the position of the span in the span slices of a task. But it is
not complete either.
First, id=0 is omitted in json encoding and the trace viewer silently
drops entries with the missing id field, so we must avoid zero-value id.
Second, it is possible that a goroutine handles multiple tasks. Then,
id collisions will happen.
This takes a simpler approach - have a counter that increments for every
emitSpan call, and use the value as the id value.
Change-Id: Idaa9505634acf6d327c6f00af32d8260955b85e1
Reviewed-on: https://go-review.googlesource.com/106755 Reviewed-by: Heschi Kreinick <heschi@google.com>
Run-TryBot: Hyang-Ah Hana Kim <hyangah@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
fanzha02 [Wed, 21 Mar 2018 10:49:24 +0000 (10:49 +0000)]
cmd/internal/obj/arm64: fix the bug of incorrect handling negative offset of LDP/STP/LDPW/STPW
The current assembler will report error when the negative offset is in
the range of [-256, 0) and is not the multiples of 4/8.
The fix introduces C_NSAUTO_8, C_NSAUTO_4 and C_NAUTO4K. C_NPAUTO
includes C_NSAUTO_8 instead of C_NSAUTO, C_NAUTO4K includes C_NSAUTO_8,
C_NSAUTO_4 and C_NSAUTO. So that assembler will encode the negative offset
that is greater than -4095 and is not the multiples of 4/8 as two instructions.
cmd/compile: in escape analysis, propagate loop depth to field
The escape analysis models "loop depth". If the address of an
expression is assigned to something defined at a lower (outer)
loop depth, the escape analysis decides it escapes. However, it
uses the loop depth of the address operator instead of where
the RHS is defined. This causes an unnecessary escape if there is
an assignment inside a loop but the RHS is defined outside the
loop. This CL propagates the loop depth.
Change-Id: I7536aca1e90717283bd6a3bb4b1bab059b0cf720
Reviewed-on: https://go-review.googlesource.com/104677
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Tom Limoncelli [Sun, 8 Apr 2018 20:05:04 +0000 (16:05 -0400)]
io/ioutil: change TempFile prefix to a pattern
Users of TempFile need to be able to supply the suffix, especially
when using operating systems that give semantic meaning to the
filename extension such as Windows. Renaming the file to include
an extension after the fact is insufficient as it could lead to
race conditions.
If the string given to TempFile includes a "*", the random string
replaces the "*". For example "myname.*.bat" will result in a random
filename such as "myname.123456.bat". If no "*' is included the
old behavior is retained, and the random digits are appended to the
end.
If multiple "*" are included, the final one is replaced, thus
permitting a pathological programmer to create filenames such as
"foo*.123456.bat" but not "foo.123456.*.bat"
Ben Shi [Mon, 9 Apr 2018 11:12:15 +0000 (11:12 +0000)]
cmd/internal/obj/arm64: optimize constant pool
"MOVD $0xaaaaaaaa, R2"
"MOVD $-0x55555555, R3"
For the above instructions, 64-bit constants 0x00000000 aaaaaaaa
and 0xffffffff aaaaaaab are stored in the constant pool.
This CL optimizes them to
"MOVWU $0xaaaaaaaa, R2"
"MOVW $-0x05555555, R3"
and 32-bit constants 0xaaaaaaaa and 0xaaaaaaab are stored in the
constant pool.
There is a little size reduction (about total 5KB) in both the go
executable and the library files.
Elias Naur [Thu, 12 Apr 2018 18:33:47 +0000 (20:33 +0200)]
misc/ios: speed up the iOS exec wrapper
First, take the exclusive lock that ensures only one running binary
later: after assembling the gotest.app directory and signing it.
Second, don't pass -r to ios-deploy. The -r flag uninstalls the
app before installing it. It seems unnecessary, takes extra time
and if there was only the one developer app on the phone, it
will drop the developer permission on uninstall.
Change-Id: Ia222d3e5c2e1e2285f53074eb952941fd45fadd9
Reviewed-on: https://go-review.googlesource.com/106676 Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
This commit includes efficiency improvements in two places in the
database/sql package where an "if err != nil" was redundant and
the error can be returned as-is (most of the code in the standard
library and even in the file I changed does it my suggested way).
Change-Id: Ib9dac69ed01ee846e570a776164cb87c2caee6ca
Reviewed-on: https://go-review.googlesource.com/106555 Reviewed-by: Daniel Theophanes <kardianos@gmail.com>
Run-TryBot: Daniel Theophanes <kardianos@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The receiver was renamed 6 years ago in https://golang.org/cl/5674065
but the docs weren't updated to match.
Change-Id: I5e72cedc0e0f067382545d272f48a9c7dfb5a9b7
Reviewed-on: https://go-review.googlesource.com/104116 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Michael Hudson-Doyle [Wed, 4 Apr 2018 22:07:41 +0000 (10:07 +1200)]
cmd/link: do not pass -no-pie to host linker when -linkshared is passed
As the comment above the code I'm changing says, when building with
-buildmode=exe, the default compiler flags produce code incompatible with PIE.
But when -linkshared is passed, the default compiler flags are not used so this
does not apply. And now I've found a system (linux/arm64 with glibc 2.27) where
this combination of flags causes a problem, albeit for reasons I don't really
understand, so stop passing -no-pie when -linkshared is passed.
Change-Id: I412ec7941dc0cb89e6d1b171fc29288aadcb9f20
Reviewed-on: https://go-review.googlesource.com/104815
Run-TryBot: Michael Hudson-Doyle <michael.hudson@canonical.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Matthew Dempsky [Wed, 11 Apr 2018 22:37:16 +0000 (15:37 -0700)]
cmd/compile, cmd/link: encode cgo directives using JSON
The standard library has plenty of polished encoder/decoder
implementations. No need for another ad-hoc one.
I considered using encoding/gob instead, but these strings go into the
package data part of the object file, so it's important they don't
contain "\n$$\n". Package json escapes newlines in strings, so it's
safe to use here.
os: document that Chown with -1 means to leave values unchanged, like POSIX
And fix the nacl implementation.
Fixes #24710
Change-Id: I31ffeea03a72dac5021ffb183fde31e9ffd060ad
Reviewed-on: https://go-review.googlesource.com/106464 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Elias Naur [Tue, 10 Apr 2018 23:24:09 +0000 (01:24 +0200)]
misc/ios: don't wait for response to lldb run in the exec wrapper
CL 106096 changed the iOS exec wrapper to directly run the binary
without waiting for a SIGINT signal, but did so in a way that
expects a "(lldb)" response from lldb in 2 seconds. Lldb might
not out output anything until the program finishes, so change the
exec wrapper to just fire and forget the the run command and go
straight to waiting for exit, successfully or otherwise.
Robert Griesemer [Thu, 5 Apr 2018 23:04:56 +0000 (16:04 -0700)]
go/printer, gofmt: handle raw string literals containing newlines better
A raw string containing newlines breaks whatever columns structure
has been established so far. Recognize the situation and force a
new section of alignment with the first line break seen after the
the raw string.
Applied gofmt to src and misc.
Fixes #9064.
Change-Id: I961e94b529b1fd421908311f366b113e2ec9b7f0
Reviewed-on: https://go-review.googlesource.com/105040 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
runtime: make Playground timestamps change when the stream fd changes
The process reading the output of the binary may read stderr and
stdout separately, and may interleave reads from the two streams
arbitrarily. Because we explicitly serialize writes on the writer
side, we can reuse a timestamp within a single stream without losing
information; however, if we use the same timestamp for write on both
streams, the reader can't tell how to interleave them.
This change ensures that every time we change between the two fds, we
also bump the timestamp. That way, writes within a stream continue to
show the same timestamp, but a sorted merge of the contents of the two
streams always interleaves them in the correct order.
This still requires a corresponding change to the Playground parser to
actually reconstruct the correct interleaving. It currently merges the
two streams without reordering them; it should instead buffer them
separately and perform a sorted merge. (See
https://golang.org/cl/105496.)
Updates golang/go#24615.
Updates golang/go#24659.
Change-Id: Id789dfcc02eb4247906c9ddad38dac50cf829979
Reviewed-on: https://go-review.googlesource.com/105235
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Yury Smolsky <yury@smolsky.by> Reviewed-by: Ian Lance Taylor <iant@golang.org>
cmd/link: set S_ATTR_PURE_INSTRUCTIONS bit for text section on darwin
Mac otool and llvm-objdump distinguishes a Mach-O section is
text or data by looking at S_ATTR_PURE_INSTRUCTIONS bit. Without
this bit it thinks our function symbols are data, not functions.
Set this bit for text section to make otool/objdump happy.
Fixes #24706.
Change-Id: I5236482cb9a72474c23fbea0f35d5b5cc8491ea4
Reviewed-on: https://go-review.googlesource.com/105256
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Nick Patavalis [Sun, 11 Mar 2018 17:11:33 +0000 (19:11 +0200)]
os: use poller when NewFile is called with a blocking descriptor.
If NewFile is called with a file descriptor that is already set to
non-blocking mode, it tries to return a pollable file (one for which
SetDeadline methods work) by adding the filedes to the poll/netpoll
mechanism. If called with a filedes in blocking mode, it returns a
non-pollable file, as it always did.
Fixes #22939
Updates #24331
Change-Id: Id54c8be1b83e6d35e14e76d7df0e57a9fd64e176
Reviewed-on: https://go-review.googlesource.com/100077
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
David Chase [Thu, 5 Apr 2018 20:14:42 +0000 (16:14 -0400)]
cmd/compile: use Block.Likely to put likely branch first
When a neither of a conditional block's successors follows,
the block must end with a conditional branch followed by a
an unconditional branch. If the (conditional) branch is
"unlikely", invert it and swap successors to make it
likely instead.
This doesn't matter to most benchmarks on amd64, but in one
instance on amd64 it caused a 30% improvement, and it is
otherwise harmless. The problematic loop is
for i := 0; i < w; i++ {
if pw[i] != 0 {
return true
}
}
compiled under GOEXPERIMENT=preemptibleloops
This the very worst-case benchmark for that experiment.
Also in this CL is a commoning up of heavily-repeated
boilerplate, which made it much easier to see that the
changes were applied correctly. In the future this should
allow un-exporting of SSAGenState.Branches once the
boilerplate-replacement is done everywhere.