Alberto Donizetti [Mon, 24 Sep 2018 21:05:39 +0000 (23:05 +0200)]
test: restore binary.BigEndian use in checkbce
CL 136855 removed the encoding/binary dependency from the checkbce.go
test by defining a local Uint64 to fix the noopt builder; then a more
general mechanism to skip tests on the noopt builder was introduced in
CL 136898, so we can now restore the binary.Uint64 calls in testbce.
Change-Id: I3efbb41be0bfc446a7e638ce6a593371ead2684f
Reviewed-on: https://go-review.googlesource.com/137056
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com> Reviewed-by: Giovanni Bajo <rasky@develer.com> Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Rob Pike [Thu, 20 Sep 2018 23:47:31 +0000 (09:47 +1000)]
fmt: unify the printing examples
Provide an example for each of the printing functions (Print,
Sprintf, Fprintln etc.), and make them all produce the same output
so their usage can be compared.
Also add a package-level example explaining the difference between
how Printf, Println, and Print behave.
There are more examples to come.
Update #27554.
Change-Id: Ide03e5233f3762a9ee2ac0269f534ab927562ce2
Reviewed-on: https://go-review.googlesource.com/136615 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Robert Griesemer [Sat, 22 Sep 2018 04:03:35 +0000 (21:03 -0700)]
spec: be more precise about the moment deferred functions are executed
Fixes #27802.
Change-Id: I7ea9f7279300a55b0cb851893edc591a6f84e324
Reviewed-on: https://go-review.googlesource.com/136758 Reviewed-by: Rob Pike <r@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Rob Pike [Fri, 21 Sep 2018 00:48:05 +0000 (10:48 +1000)]
fmt: add a package-level example illustrating basic formats
There is much left out here—the space of possibilities is very
large—but this example shows all that most programmers will need
to know for most printing problems.
Update #27554.
Change-Id: Ib6ae651d5c3720cf7fe1a05ffd0859a5b56a9157
Reviewed-on: https://go-review.googlesource.com/136616 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Change-Id: Ic687e7e0e171690e8d937c7bb29b0e55316f874a
Reviewed-on: https://go-review.googlesource.com/137015 Reviewed-by: Ian Lance Taylor <iant@golang.org>
compress: reduce copies of new text for compression testing
The previous book was 387 KiB decompressed and 119 KiB compressed, the
new book is 567 KiB decompressed and 132 KiB compressed. Overall, this
change will reduce the release binary size by 196 KiB. The new book will
allow for slightly more extensive compression testing with a larger
text.
Command to run the benchmark tests used with benchstat:
`../bin/go test -run='^$' -count=4 -bench=. compress/bzip2 compress/flate`
When running the benchmarks locally, changed "Newton" to "Twain" and
filtered the tests with the -bench flag to include only those which were
relevant to these changes.
net: don't reject domain names with only numbers and hyphens
From https://github.com/golang/go/issues/17659#issuecomment-423113606 ...
> In kubernetes , isDomainName reject Pods "A Record" "pod-ip-address",
> for example: "172-17-0-16", as RFC 3696 section 2 requires
> "top-level domain names not be all-numeric", but this example has
> three hyphen, so I think it should not be reject.
Updates #17659
Change-Id: Ibd8ffb9473d69c45c91525953c09c6749233ca20
Reviewed-on: https://go-review.googlesource.com/136900
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org> Reviewed-by: Ian Gudger <igudger@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
net: don't use splice for unix{packet,gram} connections
As pointed out in the aftermath of CL 113997, splice is not supported
for SOCK_SEQPACKET or SOCK_DGRAM unix sockets. Don't call poll.Splice
in those cases.
Elias Naur [Sat, 22 Sep 2018 08:30:05 +0000 (10:30 +0200)]
syscall: replace lstat, lchown, stat to please Android O
Implement Lstat with fstatat and Lchown with Fchownat on
linux/amd64, linux/arm and linux/386. Furthermore, implement Stat
with fstatat on linux/arm and linux/386. Linux/arm64 already had
similar replacements.
The fstatat and fchownat system calls were added in kernel 2.6.16,
which is before the Go minimum, 2.6.23.
The three syscalls then match the android bionic implementation
and avoids the Android O seccomp filter.
Alberto Donizetti [Sun, 23 Sep 2018 11:48:46 +0000 (13:48 +0200)]
test: fix bcecheck test on noopt builder
The noopt builder is configured by setting GO_GCFLAGS=-N -l, but the
test/run.go test harness doesn't look at GO_GCFLAGS when processing
"errorcheck" files, it just calls compile:
This is working as intended, since it makes the tests more robust and
independent from the environment; errorcheck files are supposed to set
additional building flags, when needed, like in:
// errorcheck -0 -N -l
The test/bcecheck.go test used to work on the noopt builder (even if
bce is not active on -N -l) because the test was auto-contained and
the file always compiled with optimizations enabled.
In CL 107355, a new bce test dependent on an external package
(encoding.binary) was added. On the noopt builder the external package
is built using -N -l, and this causes a test failure that broke the
noopt builder:
$ go install -a -gcflags="-N -l" std
$ go run run.go -- checkbce.go
This change fixes the noopt builder breakage by removing the bce test
dependency on encoding/binary by defining a local Uint64() function to
be used in the test.
Eugene Kalinin [Tue, 19 Jun 2018 18:19:47 +0000 (21:19 +0300)]
mime: derestrict value backslash unescaping for all encodings
Previously consumeValue performed consumption of "unnecessary backslashes"
strictly for non-ASCII and non-token runes. Thus if it encountered a
backslash before a rune that is out of the ASCII range, it would
erroneously skip that backslash. This change now derestricts
"unnecessary backslash" unescaping for all character encodings,
using "isTSpecial" instead of "!isTokenChar".
This change is a follow-up of CL 32175.
Johan Brandhorst [Fri, 24 Aug 2018 11:10:01 +0000 (12:10 +0100)]
net/http: ensure null body in Fetch response is not read
The Fetch API returns a null body if there is no response body,
on browsers that support streaming the response body. This
change ensures we check for both undefined and null bodies
before attempting to read the body.
Fixes #27196
Change-Id: I0da86b61284fe394418b4b431495e715a037f335
Reviewed-on: https://go-review.googlesource.com/131236 Reviewed-by: Richard Musiol <neelance@gmail.com>
Run-TryBot: Richard Musiol <neelance@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Robert Griesemer [Thu, 20 Sep 2018 22:22:33 +0000 (15:22 -0700)]
cmd/compile/internal/gc: fix Node.copy and introduce (raw|sep)copy
Node.copy used to make a shallow copy of a node. Often, this is not
correct: If a node n's Orig field pointed to itself, the copy's Orig
field has to be adjusted to point to the copy. Otherwise, if n is
modified later, the copy's Orig appears modified as well (because it
points to n).
This was fixed for one specific case with
https://go-review.googlesource.com/c/go/+/136395 (issue #26855).
This change instead addresses copy in general:
In two cases we don't want the Orig adjustment as it causes escape
analysis output to fail (not match the existing error messages).
rawcopy is used in those cases.
In several cases Orig is set to the copy immediately after making
a copy; a new function sepcopy is used there.
Updates #26855.
Fixes #27765.
Change-Id: Idaadeb5c4b9a027daabd46a2361348f7a93f2b00
Reviewed-on: https://go-review.googlesource.com/136540 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Ben Shi [Thu, 20 Sep 2018 01:26:17 +0000 (01:26 +0000)]
cmd/compile: optimize 386's assembly generator
The ADDconstmodify has similar logic with other constmodify like
instructions. This CL optimize them to share code via fallthrough.
And the size of pkg/linux_386/cmd/compile/internal/x86.a decreases
about 0.3KB.
Change-Id: Ibdf06228afde875e8fe8e30851b50ca2be513dd9
Reviewed-on: https://go-review.googlesource.com/136398
Run-TryBot: Ben Shi <powerman1st@163.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Robert Griesemer [Wed, 19 Sep 2018 03:50:04 +0000 (20:50 -0700)]
cmd/compiler/internal/gc: remove flag argument from fconv (cleanup)
The fconv flag arguments were 0, FmtSharp, and FmtSharp|FmtSign.
The 0 value was used for binary representation only, which was
readily available via Mpflt.String. Otherwise, FmtSharp was always
passed. FmtSign was used to print the '+' sign in case of a positive
number and only needed for complex number formatting. Instead
implemented cconv and handled it there.
Change-Id: I1f77282f995be9cfda05efb71a0e027836a9da26
Reviewed-on: https://go-review.googlesource.com/136195 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Joe Tsai [Tue, 18 Sep 2018 19:52:39 +0000 (12:52 -0700)]
fmt: fix usage of sync.Pool
The current usage of sync.Pool is leaky because it stores an arbitrary
sized buffer into the pool. However, sync.Pool assumes that all items in the
pool are interchangeable from a memory cost perspective. Due to the unbounded
size of a buffer that may be added, it is possible for the pool to eventually
pin arbitrarily large amounts of memory in a live-lock situation.
As a simple fix, we just set a maximum size that we permit back into the pool.
We do not need to fix the use of a sync.Pool in scan.go since the free method
has always enforced a maximum capacity since the first commit of the scan logic.
Fixes #27740
Updates #23199
Change-Id: I875278f7dba42625405df36df3e9b028252ce5e3
Reviewed-on: https://go-review.googlesource.com/136116 Reviewed-by: Bryan C. Mills <bcmills@google.com>
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Some variants of Mercurial respond differently to “permission denied” errors
than to “file not found”, and we set HOME to point to an absolute path that may
produce the former instead of the latter.
To discourage Mercurial from trying HOME, give it an explicit (empty)
configuration in the working directory instead.
Change-Id: I82ae99a6892bba7fc3d41b77209ca181d24315e2
Reviewed-on: https://go-review.googlesource.com/136135 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Ben Shi [Mon, 17 Sep 2018 02:05:22 +0000 (02:05 +0000)]
cmd/compile: optimize AMD64's bit wise operation
Currently "arr[idx] |= 0x80" is compiled to MOVLload->BTSL->MOVLstore.
And this CL optimizes it to a single BTSLconstmodify. Other bit wise
operations with a direct memory operand are also implemented.
1. The size of the executable bin/go decreases about 4KB, and the total size
of pkg/linux_amd64 (excluding cmd/compile) decreases about 0.6KB.
Ben Shi [Tue, 18 Sep 2018 01:53:42 +0000 (01:53 +0000)]
cmd/compile: simplify AMD64's assembly generator
AMD64's ADDQconstmodify/ADDLconstmodify have similar logic with
other constmodify like operators, but seperated case statements.
This CL simplify them with a fallthrough.
Change-Id: Ia73ffeaddc5080182f68c06c9d9b48fe32a14e38
Reviewed-on: https://go-review.googlesource.com/135855
Run-TryBot: Ben Shi <powerman1st@163.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
Jordan Rhee [Wed, 8 Aug 2018 21:44:42 +0000 (14:44 -0700)]
runtime: support windows/arm
Updates #26148
Change-Id: I8f68b2c926c7b11dc86c9664ed7ff2d2f78b64b4
Reviewed-on: https://go-review.googlesource.com/128715
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Keith Randall [Mon, 17 Sep 2018 19:25:36 +0000 (12:25 -0700)]
runtime: ignore races between close and len/cap
They aren't really races, or at least they don't have any
observable effect. The spec is silent on whether these are actually
races or not.
Fix this problem by not using the address of len (or of cap)
as the location where channel operations are recorded to occur.
Use a random other field of hchan for that.
I'm not 100% sure we should in fact fix this. Opinions welcome.
On Linux, sysUnused currently uses madvise(MADV_DONTNEED) to signal the
kernel that a range of allocated memory contains unneeded data. After a
successful call, the range (but not the data it contained before the
call to madvise) is still available but the first access to that range
will unconditionally incur a page fault (needed to 0-fill the range).
A faster alternative is MADV_FREE, available since Linux 4.5. The
mechanism is very similar, but the page fault will only be incurred if
the kernel, between the call to madvise and the first access, decides to
reuse that memory for something else.
In sysUnused, test whether MADV_FREE is supported and fall back to
MADV_DONTNEED in case it isn't. This requires making the return value of
the madvise syscall available to the caller, so change runtime.madvise
to return it.
Fixes #23687
Change-Id: I962c3429000dd9f4a00846461ad128b71201bb04
Reviewed-on: https://go-review.googlesource.com/135395
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
cmd/link: fix DWARF refs so that they always point to the typedef entry
For types defined as:
type typename struct { ... }
the linker produces two DIEs: (1) a DW_TAG_structure_type DIE and (2) a
DW_TAG_typedef_type DIE having (1) as its type attribute.
All subsequent references to 'typename' should use the
DW_TAG_typedef_type DIE, not the DW_TAG_structure_type. Mostly this is
true but sometimes one reference will use the DW_TAG_structure_type
directly. In particular, this happens to the 'first' reference to the
type in question (where 'first' means whatever happens first in the way
the linker scans its symbols).
This isn't only true of struct types: pointer types, array types, etc.
can also be affected.
This fix solves the problem by always returning the typedef DIE in
newtype, when one is created.
Ben Shi [Sun, 16 Sep 2018 03:05:35 +0000 (03:05 +0000)]
cmd/compile: fix wrong comment message in AMD64Ops.go
According to AMD64.rules, BTS&BTR&BTC use arg1 as the bit index,
while BT uses arg0. This CL fixes the wrong comment message in
AMD64Ops.go, which indicates all bit indexes are in arg0.
Change-Id: Idb78f4d39f7ef5ea78065ad8bc651324597e2a8a
Reviewed-on: https://go-review.googlesource.com/135419 Reviewed-by: Keith Randall <khr@golang.org>
Robert Griesemer [Mon, 17 Sep 2018 20:43:35 +0000 (13:43 -0700)]
go/types: don't report cycle error if clearer error follows
If a cyclic declaration uses a non-type object where it expects
a type, don't report the cycle error in favor of the clearer and
more informative error about the missing type.
Fixes #25790.
Change-Id: If937078383def878efb4c69686e5b4b2a495fd5d
Reviewed-on: https://go-review.googlesource.com/135700 Reviewed-by: Alan Donovan <adonovan@google.com>
Hana Kim [Mon, 17 Sep 2018 18:46:50 +0000 (14:46 -0400)]
cmd/trace: don't drop sweep slice details
For sweep events, we used to modify the ViewerEvent returned from
ctx.emitSlice later in order to embed more details about the sweep
operation. The trick no longer works after the change
https://golang.org/cl/92375 and caused a regression.
ctx.emit method encodes the ViewerEvent, so any modification to the
ViewerEvent object after ctx.emit returns will not be reflected.
Refactor ctx.emitSlice, so ctx.makeSlice can be used when producing
slices for SWEEP. ctx.emit* methods are meant to truely emit
ViewerEvents.
Michael Munday [Wed, 12 Sep 2018 11:16:50 +0000 (12:16 +0100)]
cmd/compile: avoid more float32 <-> float64 conversions in compiler
Use the new custom truncate/extension code when storing or extracting
float32 values from AuxInts to avoid the value being changed by the
host platform's floating point conversion instructions (e.g. sNaN ->
qNaN).
Updates #27516.
Change-Id: Id39650f1431ef74af088c895cf4738ea5fa87974
Reviewed-on: https://go-review.googlesource.com/134855
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org>
cmd/compile/internal/gc: handle array slice self-assign in esc.go
Instead of skipping all OSLICEARR, skip only ones with non-pointer
array type. For pointers to arrays, it's safe to apply the
self-assignment slicing optimizations.
Refactored the matching code into separate function for readability.
This is an extension to already existing optimization.
On its own, it does not improve any code under std, but
it opens some new optimization opportunities. One
of them is described in the referenced issue.
Ian Davis [Mon, 10 Sep 2018 08:27:47 +0000 (09:27 +0100)]
image/png: minor cleanup of a few tests
Removes a redundant err check and replaces some returns in a testing
loop with continue to prevent skipping unrelated test cases when
a failure is encountered.
Change-Id: Ic1a560751b95bb0ef8dfa957e057e0fa0c2b281d
Reviewed-on: https://go-review.googlesource.com/134236
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Clarify the behavior of splice on older kernels, merge comments so
control flow becomes more obvious, as discussed in CL 133575.
Change-Id: I95855991bd0b1fa1c78a900b39c4382f65d83468
Reviewed-on: https://go-review.googlesource.com/135436 Reviewed-by: Ian Lance Taylor <iant@golang.org>
The cmd/compile/internal/ld/go.go file not exist, actually cmd/link/internal/ld/go.go.
Also, write line number is not good because it changes every commit of the file.
Change-Id: Id2b9f2c9904390adb011dab357716ee8e2fe84fc
Reviewed-on: https://go-review.googlesource.com/135516 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Ben Burkert [Wed, 5 Sep 2018 17:20:26 +0000 (10:20 -0700)]
internal/poll: fall back on unsupported splice from unix socket
Gracefully fallback to a userspace copy when the kernel does not support
splice(2) on a unix domain socket. EINVAL is returned by the splice
syscall if it does not support unix domain sockets. Keeping the handled
return value as false when the first splice call fails with EINVAL will
cause the caller to fall back to a userspace copy.
Dmitri Shuralyov [Wed, 12 Sep 2018 17:58:18 +0000 (13:58 -0400)]
doc/go1.11, cmd/go: elaborate on new GOFLAGS environment variable
In Go 1.11, cmd/go gained support for the GOFLAGS environment variable.
It was added and described in detail in CL 126656.
Mention it in the Go 1.11 release notes, link to the cmd/go documentation,
and add more details there.
Fixes #27282.
Change-Id: Ifc35bfe3e0886a145478d36dde8e80aedd8ec68e
Reviewed-on: https://go-review.googlesource.com/135035 Reviewed-by: Bryan C. Mills <bcmills@google.com> Reviewed-by: Rob Pike <r@golang.org>
Agniva De Sarker [Sat, 30 Jun 2018 20:05:38 +0000 (01:35 +0530)]
time: improve error message for LoadLocation
Currently, when a tz file was being checked inside a zoneInfo dir,
a syscall.ENOENT error was being returned, which caused it to look
in the zoneinfo.zip file and return an error for that case.
We return a syscall.ENOENT error for the zip file case too, so that
it falls through to the end of the loop and returns an uniform error
for both cases.
Fixes #20969
Change-Id: If1de068022ac7693caabb5cffd1c929878460140
Reviewed-on: https://go-review.googlesource.com/121877
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 [Thu, 13 Sep 2018 12:16:46 +0000 (13:16 +0100)]
cmd/compile: skip float32 constant folding test on 387 builder
The 387 unit always quietens float64 and float32 signaling NaNs,
even when just loading and storing them. This makes it difficult
to propagate such values in the compiler. This is a hard problem
to fix and it is also very obscure.
Roberto Selbach [Mon, 27 Aug 2018 15:57:32 +0000 (15:57 +0000)]
cmd/go/internal/modfetch: stop cutting the last character of versions
When a zip archive for a module contains an unexpected file, the error
message removes the last character in the version number, e.g. an invalid
archive for "somemod@v1.2.3" would generate the following error:
"zip for somemod@1.2. has unexpected file..."
Lynn Boger [Wed, 15 Aug 2018 21:34:06 +0000 (17:34 -0400)]
cmd/compile: improve rules for PPC64.rules
This adds some improvements to the rules for PPC64 to eliminate
unnecessary zero or sign extends, and fix some rule for truncates
which were not always using the correct sign instruction.
This reduces of size of many functions by 1 or 2 instructions and
can improve performance in cases where the execution time depends
on small loops where at least 1 instruction was removed and where that
loop contributes a significant amount of the total execution time.
Included is a testcase for codegen to verify the sign/zero extend
instructions are omitted.
Robert Griesemer [Wed, 12 Sep 2018 23:45:10 +0000 (16:45 -0700)]
go/types: fix scope printing (debugging support)
Printing of scopes was horribly wrong: If a scope contained
no object declarations, it would abort printing even if the
scope had children scopes. Also, the line breaks were not
inserted consistently. The actual test case (ExampleScope)
was incorrect as well.
Fixed and simplified printing, and adjusted example which
tests the printing output.
Change-Id: If21c1d4ad71b15a517d4a54da16de5e6228eb4b5
Reviewed-on: https://go-review.googlesource.com/135115
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Alan Donovan <adonovan@google.com>
13:10:32 --- FAIL: TestGcSys (0.06s)
13:10:32 gc_test.go:30: expected "OK\n", but got "using too much memory: 71401472 bytes\n"
13:10:32 FAIL
Updates #27636
Change-Id: Ifd4cfce167d8054dc6f037bd34368d63c7f68ed4
Reviewed-on: https://go-review.googlesource.com/135155
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Ian Lance Taylor [Fri, 29 Jun 2018 17:39:44 +0000 (10:39 -0700)]
path/filepath: rewrite walkSymlinks
Rather than try to work around Clean and Join on intermediate steps,
which can remove ".." components unexpectedly, just do everything in
walkSymlinks. Use a single loop over path components.
Fixes #23444
Change-Id: I4f15e50d0df32349cc4fd55e3d224ec9ab064379
Reviewed-on: https://go-review.googlesource.com/121676
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
Alan Donovan [Thu, 13 Sep 2018 15:15:30 +0000 (11:15 -0400)]
regexp: fix incorrect name in Match doc comment
Change-Id: I628aad9a3abe9cc0c3233f476960e53bd291eca9
Reviewed-on: https://go-review.googlesource.com/135235 Reviewed-by: Ralph Corderoy <ralph@inputplus.co.uk> Reviewed-by: Ian Lance Taylor <iant@golang.org>
This change compares the first two characters instead of the first one,
and if they match, the entire string is compared. Comparing the first two
characters helps to filter out the case where the first character matches
but the subsequent characters do not match, thereby improving the substring
search speed in this case.
Benchmarks with no effect or minimal impact (less than 5%) is not listed,
the following are improved benchmarks:
On arm64:
strings:
IndexPeriodic/IndexPeriodic16-8 172890.00ns +- 2% 124156.20ns +- 0% -28.19% (p=0.008 n=5+5)
IndexPeriodic/IndexPeriodic32-8 78092.80ns +- 0% 65138.60ns +- 0% -16.59% (p=0.008 n=5+5)
IndexPeriodic/IndexPeriodic64-8 42322.20ns +- 0% 34661.60ns +- 0% -18.10% (p=0.008 n=5+5)
bytes:
IndexPeriodic/IndexPeriodic16-8 183468.20ns +- 6% 123759.00ns +- 0% -32.54% (p=0.008 n=5+5)
IndexPeriodic/IndexPeriodic32-8 84776.40ns +- 0% 63907.80ns +- 0% -24.62% (p=0.008 n=5+5)
IndexPeriodic/IndexPeriodic64-8 45835.60ns +- 0% 34194.20ns +- 0% -25.40% (p=0.008 n=5+5)
Change-Id: I2d9e7e138d29e960d20a203eb74dc2ec976a9d71
Reviewed-on: https://go-review.googlesource.com/131177
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Robert Griesemer [Thu, 13 Sep 2018 01:07:51 +0000 (18:07 -0700)]
cmd/vet: avoid internal error for implicitly declared type switch vars
For type switches using a short variable declaration of the form
switch t := x.(type) {
case T1:
...
go/types doesn't declare the symbolic variable (t in this example)
with the switch; thus such variables are not found in types.Info.Defs.
Instead they are implicitly declared with each type switch case,
and can be found in types.Info.Implicits.
Adjust the shadowing code accordingly.
Added a test case to verify that the issue is fixed, and a test
case verifying that the shadowing code now considers implicitly
declared variables introduces in type switch cases.
While at it, also fixed the (internal) error reporting to provide
more accurate information.
Fixe #26725.
Change-Id: If408ed9e692bf47c640f81de8f46bf5eb43415b0
Reviewed-on: https://go-review.googlesource.com/135117
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Daniel Martí <mvdan@mvdan.cc> Reviewed-by: Alan Donovan <adonovan@google.com>
erifan01 [Tue, 22 May 2018 06:58:32 +0000 (06:58 +0000)]
cmd/compile: intrinsify math.RoundToEven and math.Abs on arm64
math.RoundToEven can be done by one arm64 instruction FRINTND, intrinsify it to improve performance.
The current pure Go implementation of the function Abs is translated into five instructions on arm64:
str, ldr, and, str, ldr. The intrinsic implementation requires only one instruction, so in terms of
performance, intrinsify it is worthwhile.
Benchmarks:
name old time/op new time/op delta
Abs-8 3.50ns ± 0% 1.50ns ± 0% -57.14% (p=0.000 n=10+10)
RoundToEven-8 9.26ns ± 0% 1.50ns ± 0% -83.80% (p=0.000 n=10+10)
runtime: regression test for semasleep indefinite hang
A regression test in which: for a program that invokes semasleep,
we send non-terminal signals such as SIGIO.
Since the signal wakes up pthread_cond_timedwait_relative_np,
after CL 133655, we should only re-spin for the amount of
time left, instead of re-spinning with the original duration
which would cause an indefinite spin.
erifan01 [Mon, 7 May 2018 08:08:30 +0000 (08:08 +0000)]
internal/bytealg: optimize Equal on arm64
Currently the 16-byte loop chunk16_loop is implemented with NEON instructions LD1, VMOV and VCMEQ.
Using scalar instructions LDP and CMP to achieve this loop can reduce the number of clock cycles.
For cases where the length of strings are between 4 to 15 bytes, loading the last 8 or 4 bytes at
a time to reduce the number of comparisons.
cmd/go: avoid type names in __debug__modinfo__ variable injected in package main
If we use the name 'string' to refer to the built-in type, that name
can be shadowed by a local declaration. Use a string constant instead,
but keep the init function to populate it so that //go:linkname will
still work properly.
Fixes #27584.
Change-Id: I850cad6663e566f70fd123107d2e4e742c93b450
Reviewed-on: https://go-review.googlesource.com/134915 Reviewed-by: Ian Lance Taylor <iant@golang.org>
The implementations of the functions in mem_darwin.go is identical to
the ones defined in mem_bsd.go for all other BSD-like GOOSes. Also use
them on Darwin.
cmd/dist: make raceDetectorSupported an exact copy of cmd/internal/sys.RaceDetectorSupported
The comment states that cmd/internal/sys.RaceDetectorSupported is a copy,
so make the two identical. No functional difference, since ppce64le is
only supported on linux anyway.
Change-Id: Id3e4d445fb700b9b3bb53bf15ea05b8911b4f95e
Reviewed-on: https://go-review.googlesource.com/134595
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Ian Lance Taylor <iant@golang.org>
runtime: convert initial timediv quotient increments to bitsets
At the very beginning of timediv, inside a for loop,
we reduce the base value by at most (1<<31)-1, while
incrementing the quotient result by 1<<uint(bit).
However, since the quotient value was 0 to begin with,
we are essentially just doing bitsets.
This change is in the hot path of various concurrency and
scheduling operations that require sleeping, waiting
on mutexes and futexes etc. On the following OSes:
* Dragonfly
* FreeBSD
* Linux
* NetBSD
* OpenBSD
* Plan9
* Windows
and paired with architectures that provide the BTS instruction, this
change shaves off a couple of nanoseconds per invocation of timediv.
Lynn Boger [Mon, 27 Aug 2018 20:34:53 +0000 (16:34 -0400)]
internal/bytealg: implement bytes.Count in asm for ppc64x
This adds an asm implementation for the Count function in ppc64x.
The Go code that manipulates a byte at a time is especially
inefficient on ppc64x, so an asm implementation is a significant
improvement.
test: fix the wrong test of math.Copysign(c, -1) for arm64
The CL 132915 added the wrong codegen test for math.Copysign(c, -1),
it should test that AND is not emitted. This CL fixes this error.
Change-Id: Ida1d3d54ebfc7f238abccbc1f70f914e1b5bfd91
Reviewed-on: https://go-review.googlesource.com/134815 Reviewed-by: Giovanni Bajo <rasky@develer.com> Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Giovanni Bajo <rasky@develer.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Daniel Martí [Wed, 12 Sep 2018 07:26:31 +0000 (09:26 +0200)]
encoding/json: use panics for phase errors
Having these panic-like errors used to be ok, since they were used in
the internal decoder state instead of passed around via return
parameters.
Recently, the decoder was rewritten to use explicit error returns
instead. This error is a terrible fit for error returns; a handful of
functions must return an error because of it, and their callers must
check for an error that should never happen.
This is precisely what panics are for, so use them. The test coverage of
the package goes up from 91.3% to 91.6%, and performance is unaffected.
We can also get rid of some unnecessary verbosity in the code.
name old time/op new time/op delta
CodeDecoder-4 27.5ms ± 1% 27.5ms ± 1% ~ (p=0.937 n=6+6)
Change-Id: I01033b3f5b7c0cf0985082fa272754f96bf6353c
Reviewed-on: https://go-review.googlesource.com/134835
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Joe Tsai <thebrokentoaster@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Joe Tsai <thebrokentoaster@gmail.com>
Ian Lance Taylor [Fri, 7 Sep 2018 19:53:52 +0000 (12:53 -0700)]
cmd/link: don't pass all linker args when testing flag
Some linker flags can actually be input files, which can cause
misleading errors when doing the trial link, which can cause the
linker to incorrectly decide that the flag is not supported, which can
cause the link to fail.
Fixes #27510
Updates #27110
Updates #27293
Change-Id: I70c1e913cee3c813e7b267bf779bcff26d4d194a
Reviewed-on: https://go-review.googlesource.com/134057
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com> Reviewed-by: Damien Neil <dneil@google.com>
cmd/link,compress/zip,image/png: use binary.{Big,Little}Endian methods
Use the binary.{Big,Little}Endian integer encoding methods rather than
variations found in local implementations. The functions in
the binary package have been tested to ensure they inline correctly and
don't add unnecessary bounds checking.