]> Cypherpunks repositories - gostls13.git/log
gostls13.git
6 years agoruntime: fix stale comments about mheap and mspan
Michael Anthony Knyszek [Wed, 17 Oct 2018 20:16:45 +0000 (20:16 +0000)]
runtime: fix stale comments about mheap and mspan

As of 07e738e all spans are allocated out of a treap, and not just
large spans or spans for large objects. Also, now we have a separate
treap for spans that have been scavenged.

Change-Id: I9c2cb7b6798fc536bbd34835da2e888224fd7ed4
Reviewed-on: https://go-review.googlesource.com/c/142958
Reviewed-by: Austin Clements <austin@google.com>
6 years agocmd/trace: notes on MMU view improvements
Austin Clements [Tue, 25 Jul 2017 19:03:44 +0000 (15:03 -0400)]
cmd/trace: notes on MMU view improvements

Change-Id: Ib9dcdc76095f6718f1cdc83349503f52567c76d4
Reviewed-on: https://go-review.googlesource.com/c/60801
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agocmd/trace: display p99.9, p99 and p95 MUT
Austin Clements [Thu, 17 Aug 2017 22:11:01 +0000 (18:11 -0400)]
cmd/trace: display p99.9, p99 and p95 MUT

This uses the mutator utilization distribution to compute the p99.9,
p99, and p95 mutator utilization topograph lines and display them
along with the MMU.

Change-Id: I8c7e0ec326aa4bc00619ec7562854253f01cc802
Reviewed-on: https://go-review.googlesource.com/c/60800
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agointernal/trace: support for mutator utilization distributions
Austin Clements [Thu, 17 Aug 2017 15:31:03 +0000 (11:31 -0400)]
internal/trace: support for mutator utilization distributions

This adds support for computing the quantiles of a mutator utilization
distribution.

Change-Id: Ia8b3ed14bf415c234e2f567360fd1b361d28bd40
Reviewed-on: https://go-review.googlesource.com/c/60799
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agocmd/trace: expose MMU analysis flags in web UI
Austin Clements [Fri, 28 Jul 2017 20:30:05 +0000 (16:30 -0400)]
cmd/trace: expose MMU analysis flags in web UI

Change-Id: I672240487172380c9eef61837b41698021aaf834
Reviewed-on: https://go-review.googlesource.com/c/60798
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agointernal/trace: add "per-P" MMU analysis
Austin Clements [Fri, 28 Jul 2017 20:26:51 +0000 (16:26 -0400)]
internal/trace: add "per-P" MMU analysis

The current MMU analysis considers all Ps together, so if, for
example, one of four Ps is blocked, mutator utilization is 75%.
However, this is less useful for understanding the impact on
individual goroutines because that one blocked goroutine could be
blocked for a very long time, but we still appear to have good
utilization.

Hence, this introduces a new flag that does a "per-P" analysis where
the utilization of each P is considered independently. The MMU is then
the combination of the MMU for each P's utilization function.

Change-Id: Id67b980d4d82b511d28300cdf92ccbb5ae8f0c78
Reviewed-on: https://go-review.googlesource.com/c/60797
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agointernal/trace: flags for what to include in GC utilization
Austin Clements [Fri, 28 Jul 2017 17:51:58 +0000 (13:51 -0400)]
internal/trace: flags for what to include in GC utilization

Change-Id: I4ba963b003cb25b39d7575d423f17930d84f3f69
Reviewed-on: https://go-review.googlesource.com/c/60796
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agocmd/trace: list and link to worst mutator utilization windows
Austin Clements [Fri, 28 Jul 2017 14:24:39 +0000 (10:24 -0400)]
cmd/trace: list and link to worst mutator utilization windows

This adds the ability to click a point on the MMU graph to show a list
of the worst 10 mutator utilization windows of the selected size. This
list in turn links to the trace viewer to drill down on specifically
what happened in each specific window.

Change-Id: Ic1b72d8b37fbf2212211c513cf36b34788b30133
Reviewed-on: https://go-review.googlesource.com/c/60795
Run-TryBot: Austin Clements <austin@google.com>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agointernal/trace: track worst N mutator utilization windows
Austin Clements [Wed, 26 Jul 2017 16:21:15 +0000 (12:21 -0400)]
internal/trace: track worst N mutator utilization windows

This will let the trace viewer show specifically when poor utilization
happened and link to specific instances in the trace.

Change-Id: I1f03a0f9d9a7570009bb15762e7b8b6f215e9423
Reviewed-on: https://go-review.googlesource.com/c/60793
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agointernal/trace: use banding to optimize MMU computation
Austin Clements [Tue, 8 Aug 2017 22:15:32 +0000 (18:15 -0400)]
internal/trace: use banding to optimize MMU computation

This further optimizes MMU construction by first computing a
low-resolution summary of the utilization curve. This "band" summary
lets us compute the worst-possible window starting in each of these
low-resolution bands (even without knowing where in the band the
window falls). This in turn lets us compute precise minimum mutator
utilization only in the worst low-resolution bands until we can show
that any remaining bands can't possibly contain a worse window.

This slows down MMU construction for small traces, but these are
reasonably fast to compute either way. For large traces (e.g.,
150,000+ utilization changes) it's significantly faster.

Change-Id: Ie66454e71f3fb06be3f6173b6d91ad75c61bda48
Reviewed-on: https://go-review.googlesource.com/c/60792
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agointernal/trace: use MU slope to optimize MMU
Austin Clements [Mon, 28 Aug 2017 16:29:07 +0000 (12:29 -0400)]
internal/trace: use MU slope to optimize MMU

This commit speeds up MMU construction by ~10X (and reduces the number
of windows considered by ~20X) by using an observation about the
maximum slope of the windowed mutator utilization function to advance
the window time in jumps if the window's current mean mutator
utilization is much larger than the current minimum.

Change-Id: If3cba5da0c4adc37b568740f940793e491e96a51
Reviewed-on: https://go-review.googlesource.com/c/60791
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agocmd/trace: add minimum mutator utilization (MMU) plot
Austin Clements [Mon, 24 Jul 2017 15:37:59 +0000 (11:37 -0400)]
cmd/trace: add minimum mutator utilization (MMU) plot

This adds an endpoint to the trace tool that plots the minimum mutator
utilization curve using information on mark assists and GC pauses from
the trace.

This commit implements a fairly straightforward O(nm) algorithm for
computing the MMU (and tests against an even more direct but slower
algorithm). Future commits will extend and optimize this algorithm.

This should be useful for debugging and understanding mutator
utilization issues like #14951, #14812, #18155. #18534, #21107,
particularly once follow-up CLs add trace cross-referencing.

Change-Id: Ic2866869e7da1e6c56ba3e809abbcb2eb9c4923a
Reviewed-on: https://go-review.googlesource.com/c/60790
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agocmd/addr2line: defer closing objfile
Muhammad Falak R Wani [Mon, 5 Nov 2018 16:34:46 +0000 (22:04 +0530)]
cmd/addr2line: defer closing objfile

Change-Id: I19ff9d231c4cc779b0737802c3c40ee2e00934dd
Reviewed-on: https://go-review.googlesource.com/c/147477
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agoruntime/internal/sys: regenerate zgoos_*.go files
Tobias Klauser [Mon, 5 Nov 2018 11:39:11 +0000 (12:39 +0100)]
runtime/internal/sys: regenerate zgoos_*.go files

zgoos_aix.go is missing GoosJs, the order of GoosAndroid and GoosAix is
mixed up in all files and GoosHurd was added after CL 146023 introduced
GOOS=hurd.

Change-Id: I7e2f5a15645272e9020cfca86e44c364fc072a2b
Reviewed-on: https://go-review.googlesource.com/c/147397
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoruntime: avoid arm64 8.1 atomics on Android
Elias Naur [Sun, 4 Nov 2018 08:36:25 +0000 (09:36 +0100)]
runtime: avoid arm64 8.1 atomics on Android

The kernel on some Samsung S9+ models reports support for arm64 8.1
atomics, but in reality only some of the cores support them. Go
programs scheduled to cores without support will crash with SIGILL.

This change unconditionally disables the optimization on Android.
A better fix is to precisely detect the offending chipset.

Fixes #28431

Change-Id: I35a1273e5660603824d30ebef2ce7e429241bf1f
Reviewed-on: https://go-review.googlesource.com/c/147377
Run-TryBot: Elias Naur <elias.naur@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
6 years agostrings: lower running time of TestCompareStrings
Keith Randall [Sun, 4 Nov 2018 05:57:52 +0000 (22:57 -0700)]
strings: lower running time of TestCompareStrings

At each comparison, we're making a copy of the whole string.
Instead, use unsafe to share the string backing store with a []byte.

It reduces the test time from ~4sec to ~1sec on my machine
(darwin/amd64).  Some builders were having much more trouble with this
test (>3min), it may help more there.

Fixes #26174
Fixes #28573
Fixes #26155
Update #26473

Change-Id: Id5856fd26faf6ff46e763a088f039230556a4116
Reviewed-on: https://go-review.googlesource.com/c/147358
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/compile/internal/gc: remove isforw predicate table (cleanup)
Robert Griesemer [Sat, 3 Nov 2018 05:21:23 +0000 (22:21 -0700)]
cmd/compile/internal/gc: remove isforw predicate table (cleanup)

Was only ever filled with one Etype (TFORW) and only used
in one place. Easier to just check for TFORW.

Change-Id: Icc96da3a22b0af1d7e60bc5841c744916c53341e
Reviewed-on: https://go-review.googlesource.com/c/147285
Reviewed-by: Martin Möhrmann <moehrmann@google.com>
6 years agocmd/compile/internal/gc: don't print "internal error" twice
Robert Griesemer [Sat, 3 Nov 2018 16:33:31 +0000 (09:33 -0700)]
cmd/compile/internal/gc: don't print "internal error" twice

The Fatalf mechanism already prints "compiler internal error:"
when reporting an error. There's no need to have "internal error"
in the error message passed to Fatalf calls. Removed them.

Fixes #28575.

Change-Id: I12b1bea37bc839780f257c27ef9e2005bf334925
Reviewed-on: https://go-review.googlesource.com/c/147287
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Martin Möhrmann <moehrmann@google.com>
6 years agocmd/compile: fix fmt_test (fix long test build)
Robert Griesemer [Sat, 3 Nov 2018 04:51:01 +0000 (21:51 -0700)]
cmd/compile: fix fmt_test (fix long test build)

Follow-up on https://golang.org/cl/146782.

TBR=bradfitz

Change-Id: Idaf5488fedfc05d6ff71706fa0bcd70bf98ab25a
Reviewed-on: https://go-review.googlesource.com/c/147283
Reviewed-by: Robert Griesemer <gri@golang.org>
6 years agocmd/internal/obj: don't dedup symbols in WriteObjFile
Austin Clements [Mon, 29 Oct 2018 22:27:51 +0000 (18:27 -0400)]
cmd/internal/obj: don't dedup symbols in WriteObjFile

Currently, WriteObjFile deduplicates symbols by name. This is a
strange and unexpected place to do this. But, worse, there's no
checking that it's reasonable to deduplicate two symbols, so this
makes it incredibly easy to mask errors involving duplicate symbols.
Dealing with duplicate symbols is better left to the linker. We're
also about to introduce multiple symbols with the same name but
different ABIs/versions, which would make this deduplication more
complicated. We just removed the only part of the compiler that
actually depended on this behavior.

This CL removes symbol deduplication from WriteObjFile, since it is no
longer needed.

For #27539.

Change-Id: I650c550e46e83f95c67cb6c6646f9b2f7f10df30
Reviewed-on: https://go-review.googlesource.com/c/146558
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
6 years agocmd/compile: avoid duplicate GC bitmap symbols
Austin Clements [Mon, 29 Oct 2018 22:21:00 +0000 (18:21 -0400)]
cmd/compile: avoid duplicate GC bitmap symbols

Currently, liveness produces a distinct obj.LSym for each GC bitmap
for each function. These are then named by content hash and only
ultimately deduplicated by WriteObjFile.

For various reasons (see next commit), we want to remove this
deduplication behavior from WriteObjFile. Furthermore, it's
inefficient to produce these duplicate symbols in the first place.

GC bitmaps are the only source of duplicate symbols in the compiler.
This commit eliminates these duplicate symbols by declaring them in
the Ctxt symbol hash just like every other obj.LSym. As a result, all
GC bitmaps with the same content now refer to the same obj.LSym.

The next commit will remove deduplication from WriteObjFile.

For #27539.

Change-Id: I4f15e3d99530122cdf473b7a838c69ef5f79db59
Reviewed-on: https://go-review.googlesource.com/c/146557
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
6 years agocmd/internal/obj/arm64: fix encoding of 32-bit negated logical instructions
Cherry Zhang [Fri, 2 Nov 2018 16:03:31 +0000 (12:03 -0400)]
cmd/internal/obj/arm64: fix encoding of 32-bit negated logical instructions

32-bit negated logical instructions (BICW, ORNW, EONW) with
constants were mis-encoded, because they were missing in the
cases where we handle 32-bit logical instructions. This CL
adds the missing cases.

Fixes #28548

Change-Id: I3d6acde7d3b72bb7d3d5d00a9df698a72c806ad5
Reviewed-on: https://go-review.googlesource.com/c/147077
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Ben Shi <powerman1st@163.com>
Reviewed-by: Ben Shi <powerman1st@163.com>
6 years agonet/http: speed up ServeMux matching
Carl Mastrangelo [Fri, 26 Oct 2018 05:03:43 +0000 (22:03 -0700)]
net/http: speed up ServeMux matching

Scanning through all path patterns is not necessary, since the
paths do not change frequently.  Instead, maintain a sorted list
of path prefixes and return the first match.

name            old time/op  new time/op  delta
ServerMatch-12   134ns ± 3%    17ns ± 4%  -86.95%  (p=0.000 n=19+20)

Change-Id: I15b4483dc30db413321435ee6815fc9bf2bcc546
Reviewed-on: https://go-review.googlesource.com/c/144937
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agoall: use "reports whether" consistently in the few places that didn't
Brad Fitzpatrick [Fri, 2 Nov 2018 15:18:43 +0000 (15:18 +0000)]
all: use "reports whether" consistently in the few places that didn't

Go documentation style for boolean funcs is to say:

    // Foo reports whether ...
    func Foo() bool

(rather than "returns true if")

This CL also replaces 4 uses of "iff" with the same "reports whether"
wording, which doesn't lose any meaning, and will prevent people from
sending typo fixes when they don't realize it's "if and only if". In
the past I think we've had the typo CLs updated to just say "reports
whether". So do them all at once.

(Inspired by the addition of another "returns true if" in CL 146938
in fd_plan9.go)

Created with:

$ perl -i -npe 's/returns true if/reports whether/' $(git grep -l "returns true iff" | grep -v vendor)
$ perl -i -npe 's/returns true if/reports whether/' $(git grep -l "returns true if" | grep -v vendor)

Change-Id: Ided502237f5ab0d25cb625dbab12529c361a8b9f
Reviewed-on: https://go-review.googlesource.com/c/147037
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/compile/internal/gc: add tracing support to debug type checking
Robert Griesemer [Thu, 1 Nov 2018 20:46:50 +0000 (13:46 -0700)]
cmd/compile/internal/gc: add tracing support to debug type checking

The compiler must first be built with the constant enableTrace set
to true (typecheck.go). After that, the -t flag becomes available
which enables tracing output of type-checking functions.

With enableTrace == false, the tracing code becomes dead code
and won't affect the compiler.

Typical output might look like this:

path/y.go:4:6: typecheck 0xc00033e180 DCLTYPE <node DCLTYPE> tc=0
path/y.go:4:6: . typecheck1 0xc00033e180 DCLTYPE <node DCLTYPE> tc=2
path/y.go:4:6: . . typecheck 0xc000331a40 TYPE T tc=1
path/y.go:4:6: . . . typecheck1 0xc000331a40 TYPE T tc=2
path/y.go:4:6: . . . . typecheckdef 0xc000331a40 TYPE T tc=2
path/y.go:4:6: . . . . => 0xc000331a40 TYPE T tc=2 type=*T
path/y.go:4:6: . . . => 0xc000331a40 TYPE T tc=2 type=*T
path/y.go:4:6: . . => 0xc000331a40 TYPE T tc=1 type=*T
path/y.go:4:6: . => 0xc00033e180 DCLTYPE <node DCLTYPE> tc=2 type=<T>
path/y.go:4:6: => 0xc00033e180 DCLTYPE <node DCLTYPE> tc=1 type=<T>

Disabled by default.

Change-Id: Ifd8385290d1cf0d3fc5e8468b2f4ab84e8eff338
Reviewed-on: https://go-review.googlesource.com/c/146782
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
6 years agocrypto/tls: implement TLS 1.3 server handshake (base)
Filippo Valsorda [Fri, 2 Nov 2018 04:57:30 +0000 (00:57 -0400)]
crypto/tls: implement TLS 1.3 server handshake (base)

Implement a basic TLS 1.3 server handshake, only enabled if explicitly
requested with MaxVersion.

This CL intentionally leaves for future CLs:
  - PSK modes and resumption
  - client authentication
  - compatibility mode ChangeCipherSpecs
  - early data skipping
  - post-handshake messages
  - downgrade protection
  - KeyLogWriter support
  - TLS_FALLBACK_SCSV processing

It also leaves a few areas up for a wider refactor (maybe in Go 1.13):
  - the certificate selection logic can be significantly improved,
    including supporting and surfacing signature_algorithms_cert, but
    this isn't new in TLS 1.3 (see comment in processClientHello)
  - handshake_server_tls13.go can be dried up and broken into more
    meaningful, smaller functions, but it felt premature to do before
    PSK and client auth support
  - the monstrous ClientHello equality check in doHelloRetryRequest can
    get both cleaner and more complete with collaboration from the
    parsing layer, which can come at the same time as extension
    duplicates detection

Updates #9671

Change-Id: Id9db2b6ecc2eea21bf9b59b6d1d9c84a7435151c
Reviewed-on: https://go-review.googlesource.com/c/147017
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Adam Langley <agl@golang.org>
6 years agocrypto/tls: implement TLS 1.3 client handshake (base)
Filippo Valsorda [Thu, 1 Nov 2018 05:01:09 +0000 (01:01 -0400)]
crypto/tls: implement TLS 1.3 client handshake (base)

Implement a basic TLS 1.3 client handshake, only enabled if explicitly
requested with MaxVersion.

This CL intentionally leaves for future CLs:
  - PSK modes and resumption
  - client authentication
  - post-handshake messages
  - downgrade protection
  - KeyLogWriter support

Updates #9671

Change-Id: Ieb6130fb6f25aea4f0d39e3a2448dfc942e1de7a
Reviewed-on: https://go-review.googlesource.com/c/146559
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Adam Langley <agl@golang.org>
6 years agocrypto/tls: advertise and accept rsa_pss_rsae signature algorithms
Filippo Valsorda [Wed, 31 Oct 2018 16:14:51 +0000 (12:14 -0400)]
crypto/tls: advertise and accept rsa_pss_rsae signature algorithms

crypto/x509 already supports PSS signatures (with rsaEncryption OID),
and crypto/tls support was added in CL 79736. Advertise support for the
algorithms and accept them as a peer.

Note that this is about PSS signatures from regular RSA public keys.
RSA-PSS only public keys (with RSASSA-PSS OID) are supported in neither
crypto/tls nor crypto/x509. See RFC 8446, Section 4.2.3.

testdata/Server-TLSv12-ClientAuthRequested* got modified because the
CertificateRequest carries the supported signature algorithms.

The net/smtp tests changed because 512 bits keys are too small for PSS.

Based on Peter Wu's CL 79738, who did all the actual work in CL 79736.

Updates #9671

Change-Id: I4a31e9c6e152ff4c50a5c8a274edd610d5fff231
Reviewed-on: https://go-review.googlesource.com/c/146258
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Adam Langley <agl@golang.org>
6 years agocrypto/tls: implement TLS 1.3 version negotiation
Filippo Valsorda [Wed, 31 Oct 2018 13:34:10 +0000 (09:34 -0400)]
crypto/tls: implement TLS 1.3 version negotiation

RFC 8446 recommends using the supported_versions extension to negotiate
lower versions as well, so begin by implementing it to negotiate the
currently supported versions.

Note that pickTLSVersion was incorrectly negotiating the ServerHello
version down on the client. If the server had illegally sent a version
higher than the ClientHello version, the client would have just
downgraded it, hopefully failing later in the handshake.

In TestGetConfigForClient, we were hitting the record version check
because the server would select TLS 1.1, the handshake would fail on the
client which required TLS 1.2, which would then send a TLS 1.0 record
header on its fatal alert (not having negotiated a version), while the
server would expect a TLS 1.1 header at that point. Now, the client gets
to communicate the minimum version through the extension and the
handshake fails on the server.

Updates #9671

Change-Id: Ie33c7124c0c769f62e10baad51cbed745c424e5b
Reviewed-on: https://go-review.googlesource.com/c/146217
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Adam Langley <agl@golang.org>
6 years agocrypto/tls: implement TLS 1.3 version-specific messages
Filippo Valsorda [Sun, 28 Oct 2018 22:04:54 +0000 (18:04 -0400)]
crypto/tls: implement TLS 1.3 version-specific messages

Note that there is significant code duplication due to extensions with
the same format appearing in different messages in TLS 1.3. This will be
cleaned up in a future refactor once CL 145317 is merged.

Enforcing the presence/absence of each extension in each message is left
to the upper layer, based on both protocol version and extensions
advertised in CH and CR. Duplicated extensions and unknown extensions in
SH, EE, HRR, and CT will be tightened up in a future CL.

The TLS 1.2 CertificateStatus message was restricted to accepting only
type OCSP as any other type (none of which are specified so far) would
have to be negotiated.

Updates #9671

Change-Id: I7c42394c5cc0af01faa84b9b9f25fdc6e7cfbb9e
Reviewed-on: https://go-review.googlesource.com/c/145477
Reviewed-by: Adam Langley <agl@golang.org>
6 years agocrypto/tls: implement TLS 1.3 cryptographic computations
Filippo Valsorda [Sun, 28 Oct 2018 15:04:24 +0000 (11:04 -0400)]
crypto/tls: implement TLS 1.3 cryptographic computations

Vendors golang.org/x/crypto/hkdf at e84da0312774c21d64ee2317962ef669b27ffb41

Updates #9671

Change-Id: I2610c4a66756e2a6f21f9823dcbe39edd9c9ea21
Reviewed-on: https://go-review.googlesource.com/c/145298
Reviewed-by: Adam Langley <agl@golang.org>
6 years agocrypto/tls: implement TLS 1.3 record layer and cipher suites
Filippo Valsorda [Sat, 27 Oct 2018 16:50:25 +0000 (12:50 -0400)]
crypto/tls: implement TLS 1.3 record layer and cipher suites

Updates #9671

Change-Id: I1ea7b724975c0841d01f4536eebb23956b30d5ea
Reviewed-on: https://go-review.googlesource.com/c/145297
Reviewed-by: Adam Langley <agl@golang.org>
6 years agocrypto/tls: implement TLS 1.3 extensions for ClientHello and ServerHello
Filippo Valsorda [Thu, 25 Oct 2018 19:11:13 +0000 (15:11 -0400)]
crypto/tls: implement TLS 1.3 extensions for ClientHello and ServerHello

Updates #9671

Change-Id: Ia68224aca866dc3c98af1fccbe56bfb3f22da9f6
Reviewed-on: https://go-review.googlesource.com/c/144578
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Adam Langley <agl@golang.org>
6 years agogo/build: add go1.12 release tag
Alan Donovan [Fri, 2 Nov 2018 20:56:09 +0000 (16:56 -0400)]
go/build: add go1.12 release tag

Change-Id: I82e3f9140e5d14f02beef64c474a3ae88fe256e1
Reviewed-on: https://go-review.googlesource.com/c/147219
Run-TryBot: Alan Donovan <adonovan@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/compile: disallow converting string to notinheap slice
Matthew Dempsky [Tue, 16 Oct 2018 22:31:07 +0000 (15:31 -0700)]
cmd/compile: disallow converting string to notinheap slice

Unlikely to happen in practice, but easy enough to prevent and might
as well do so for completeness.

Fixes #28243.

Change-Id: I848c3af49cb923f088e9490c6a79373e182fad08
Reviewed-on: https://go-review.googlesource.com/c/142719
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Austin Clements <austin@google.com>
6 years agodoc: document Go 1.10.5
Andrew Bonventre [Fri, 2 Nov 2018 19:31:37 +0000 (15:31 -0400)]
doc: document Go 1.10.5

Change-Id: I11adca150ab795607b832fb354a3e065655e1020
Reviewed-on: https://go-review.googlesource.com/c/147179
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agodoc: document Go 1.11.2
Andrew Bonventre [Fri, 2 Nov 2018 19:28:32 +0000 (15:28 -0400)]
doc: document Go 1.11.2

Change-Id: Iaff03911f1807d462f1966590626bd486807f53d
Reviewed-on: https://go-review.googlesource.com/c/147178
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/cgo: accept expressions as untyped constants
Ian Lance Taylor [Fri, 2 Nov 2018 13:55:48 +0000 (06:55 -0700)]
cmd/cgo: accept expressions as untyped constants

Fixes #28545

Change-Id: I31c57ce11aca651cacc72235c7753e0c0fd170ef
Reviewed-on: https://go-review.googlesource.com/c/146900
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
6 years agocmd/go: don't let "go test -c -o /dev/null" overwrite /dev/null
Ian Lance Taylor [Fri, 2 Nov 2018 14:34:52 +0000 (07:34 -0700)]
cmd/go: don't let "go test -c -o /dev/null" overwrite /dev/null

Fixes #28549

Change-Id: Iba71bb2edd0759004e0c7df92b2b8f1197bd62d3
Reviewed-on: https://go-review.googlesource.com/c/146901
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agosyscall: add Syscall18 on Windows
Hajime Hoshi [Fri, 2 Nov 2018 17:19:47 +0000 (02:19 +0900)]
syscall: add Syscall18 on Windows

There are some OpenGL functions that take more than 15 arguments.
This CL adds Syscall18 to enable to call such functions on Windows
via syscall functions.

Fixes #28434

Change-Id: Ic7e37dda9cadf4516183e98166bfc52844ad2bbe
Reviewed-on: https://go-review.googlesource.com/c/147117
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>

6 years agodatabase/sql: add description to String method of IsolationLevel struct.
Vladimir Kovpak [Fri, 2 Nov 2018 16:56:42 +0000 (16:56 +0000)]
database/sql: add description to String method of IsolationLevel struct.

Add simple description to String method of IsolationLevel struct.

Change-Id: I8bdf829c81d4895b8542a3f21437bed61c6e925d
GitHub-Last-Rev: 7e16d6a93a9560cf3a09413dfc47d0cb04bb2fde
GitHub-Pull-Request: golang/go#28560
Reviewed-on: https://go-review.googlesource.com/c/147040
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agoruntime: look up runtime env variables case insensitively on Windows
Brad Fitzpatrick [Fri, 2 Nov 2018 15:47:40 +0000 (15:47 +0000)]
runtime: look up runtime env variables case insensitively on Windows

Fixes #28557

Change-Id: Ifca958b78e8c62fbc66515e693f528d799e8e84b
Reviewed-on: https://go-review.googlesource.com/c/147039
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoall: skip unsupported tests on AIX
Clément Chigot [Fri, 2 Nov 2018 09:09:30 +0000 (10:09 +0100)]
all: skip unsupported tests on AIX

This commit skips tests which aren't yet supported on AIX.

nosplit.go is disabled because stackGuardMultiplier is increased for
syscalls.

Change-Id: Ib5ff9a4539c7646bcb6caee159f105ff8a160ad7
Reviewed-on: https://go-review.googlesource.com/c/146939
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>

6 years agointernal/syscall/windows: add LockFileEx and UnlockFileEx for use in cmd/go
Bryan C. Mills [Tue, 16 Oct 2018 20:05:51 +0000 (16:05 -0400)]
internal/syscall/windows: add LockFileEx and UnlockFileEx for use in cmd/go

Updates #26794

Change-Id: Ic1d3078176721f3d2e5d8188c234383037babbaf
Reviewed-on: https://go-review.googlesource.com/c/145177
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
6 years agoall: fix tests for older versions of AIX 7.2
Clément Chigot [Fri, 2 Nov 2018 09:55:12 +0000 (10:55 +0100)]
all: fix tests for older versions of AIX 7.2

This commit fixes tests which fail on some versions of AIX 7.2 due
to internal bugs.

getsockname isn't working properly with unix networks.

Timezone files aren't returning a correct output.

Change-Id: I4ff15683912be62ab86dfbeeb63b73513404d086
Reviewed-on: https://go-review.googlesource.com/c/146940
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agointernal/poll, os/exec, runtime: replace PollDescriptor by IsPollDescriptor
Clément Chigot [Fri, 2 Nov 2018 09:02:38 +0000 (10:02 +0100)]
internal/poll, os/exec, runtime: replace PollDescriptor by IsPollDescriptor

This commit changes poll.PollDescriptor by poll.IsPollDescriptor. This
is needed for OS like AIX which have more than one FD using inside their
netpoll implementation.

Change-Id: I49e12a8d74045c501e19fdd8527cf166a3c64850
Reviewed-on: https://go-review.googlesource.com/c/146938
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoruntime: use StorepNoWB instead of atomicstorep in netpoll
Dmitry Vyukov [Wed, 31 Oct 2018 18:36:21 +0000 (19:36 +0100)]
runtime: use StorepNoWB instead of atomicstorep in netpoll

We only need the memory barrier from these stores,
and we only store nil over nil or over a static function value.
The write barrier is unnecessary.

name                  old time/op  new time/op  delta
TCP4OneShotTimeout-6  17.0µs ± 0%  17.0µs ± 0%  -0.43%  (p=0.032 n=5+5)
SetReadDeadline-6      205ns ± 1%   205ns ± 1%    ~     (p=0.683 n=5+5)

Update #25729

Change-Id: I66c097a1db7188697ddfc381f31acec053dfed2c
Reviewed-on: https://go-review.googlesource.com/c/146345
Run-TryBot: Dmitry Vyukov <dvyukov@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoos: add support for long path names on aix RemoveAll
Tobias Klauser [Fri, 2 Nov 2018 08:12:08 +0000 (09:12 +0100)]
os: add support for long path names on aix RemoveAll

Follow CL 146020 and enable RemoveAll based on Unlinkat and Openat on
aix.

Updates #27029

Change-Id: I78b34ed671166ee6fa651d5f2025b88548ee6c68
Reviewed-on: https://go-review.googlesource.com/c/146937
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Clément Chigot <clement.chigot@atos.net>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoruntime: avoid runtimeNano call on a common netpoll path
Dmitry Vyukov [Wed, 31 Oct 2018 14:07:57 +0000 (15:07 +0100)]
runtime: avoid runtimeNano call on a common netpoll path

runtimeNano is slower than nanotime, so pass the duration
to runtime_pollSetDeadline as is. netpoll can add nanotime itself.
Arguably a bit simpler because, say, a negative duration
clearly represents already expired timer, no need to compare to
nanotime again.
This may also fix an obscure corner case when a deadline in past
which happens to be nanotime 0 is confused with no deadline at all,
which are radically different things.
Also don't compute any durations and times if Time is zero
(currently we first compute everything and then reset d back to 0,
which is wasteful).

name                  old time/op  new time/op  delta
TCP4OneShotTimeout-6  17.1µs ± 0%  17.0µs ± 0%     ~     (p=0.421 n=5+5)
SetReadDeadline-6      230ns ± 0%   205ns ± 1%  -10.63%  (p=0.008 n=5+5)

Change-Id: I2aad699270289a5b9ead68f5e44ec4ec6d96baa0
Reviewed-on: https://go-review.googlesource.com/c/146344
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Dmitry Vyukov <dvyukov@google.com>

6 years agoruntime: execute memory barrier conditionally when changing netpoll timers
Dmitry Vyukov [Wed, 31 Oct 2018 16:56:14 +0000 (17:56 +0100)]
runtime: execute memory barrier conditionally when changing netpoll timers

We only need the memory barrier in poll_runtime_pollSetDeadline only
when one of the timers has fired, which is not the expected case.
Memory barrier can be somewhat expensive on some archs,
so execute it only if one of the timers has in fact fired.

name                  old time/op  new time/op  delta
TCP4OneShotTimeout-6  17.0µs ± 0%  17.1µs ± 0%  +0.35%  (p=0.032 n=5+5)
SetReadDeadline-6      232ns ± 0%   230ns ± 0%  -1.03%  (p=0.000 n=4+5)

Update #25729

Change-Id: Ifce6f505b9e7ba3717bad8f454077a2e94ea6e75
Reviewed-on: https://go-review.googlesource.com/c/146343
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoruntime: move nanotime wrappers to time and poll packages
Dmitry Vyukov [Wed, 31 Oct 2018 16:42:15 +0000 (17:42 +0100)]
runtime: move nanotime wrappers to time and poll packages

The nanotime wrappers in runtime introduce a bunch
of unnecessary code onto hot paths, e.g.:

0000000000449d70 <time.runtimeNano>:
  449d70:       64 48 8b 0c 25 f8 ff    mov    %fs:0xfffffffffffffff8,%rcx
  449d77:       ff ff
  449d79:       48 3b 61 10             cmp    0x10(%rcx),%rsp
  449d7d:       76 26                   jbe    449da5 <time.runtimeNano+0x35>
  449d7f:       48 83 ec 10             sub    $0x10,%rsp
  449d83:       48 89 6c 24 08          mov    %rbp,0x8(%rsp)
  449d88:       48 8d 6c 24 08          lea    0x8(%rsp),%rbp
  449d8d:       e8 ae 18 01 00          callq  45b640 <runtime.nanotime>
  449d92:       48 8b 04 24             mov    (%rsp),%rax
  449d96:       48 89 44 24 18          mov    %rax,0x18(%rsp)
  449d9b:       48 8b 6c 24 08          mov    0x8(%rsp),%rbp
  449da0:       48 83 c4 10             add    $0x10,%rsp
  449da4:       c3                      retq
  449da5:       e8 56 e0 00 00          callq  457e00 <runtime.morestack_noctxt>
  449daa:       eb c4                   jmp    449d70 <time.runtimeNano>

Move them to the corresponding packages which eliminates all of this.

name                  old time/op  new time/op  delta
TCP4OneShotTimeout-6  17.1µs ± 1%  17.0µs ± 0%  -0.66%  (p=0.032 n=5+5)
SetReadDeadline-6      234ns ± 1%   232ns ± 0%  -0.77%  (p=0.016 n=5+4)

Update #25729

Change-Id: Iee05027adcdc289ba895c5f5a37f154e451bc862
Reviewed-on: https://go-review.googlesource.com/c/146342
Run-TryBot: Dmitry Vyukov <dvyukov@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agotime: speed up Since and Until
Dmitry Vyukov [Wed, 31 Oct 2018 16:36:51 +0000 (17:36 +0100)]
time: speed up Since and Until

time.now is somewhat expensive (much more expensive than nanotime),
in the common case when Time has monotonic time we don't actually
need to call time.now in Since/Until as we can do calculation
based purely on monotonic times.

name                  old time/op  new time/op  delta
TCP4OneShotTimeout-6  17.0µs ± 0%  17.1µs ± 1%     ~     (p=0.151 n=5+5)
SetReadDeadline-6      261ns ± 0%   234ns ± 1%  -10.35%  (p=0.008 n=5+5)

Benchmark that only calls Until:

benchmark            old ns/op     new ns/op     delta
BenchmarkUntil       54.0          29.5          -45.37%

Update #25729

Change-Id: I5ac5af3eb1fe9f583cf79299f10b84501b1a0d7d
Reviewed-on: https://go-review.googlesource.com/c/146341
Run-TryBot: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agoruntime, time: refactor startNano handling
Dmitry Vyukov [Wed, 31 Oct 2018 16:27:16 +0000 (17:27 +0100)]
runtime, time: refactor startNano handling

Move startNano from runtime to time package.
In preparation for a subsequent change that speeds up Since and Until.
This also makes code simpler as we have less assembly as the result,
monotonic time handling is better localized in time package.
This changes values returned from nanotime on windows
(it does not account for startNano anymore), current comments state
that it's important, but it's unclear how it can be important
since no other OS does this.

Update #25729

Change-Id: I2275d57b7b5ed8fd0d53eb0f19d55a86136cc555
Reviewed-on: https://go-review.googlesource.com/c/146340
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoruntime: add and use modtimer in netpoll
Dmitry Vyukov [Wed, 31 Oct 2018 16:03:35 +0000 (17:03 +0100)]
runtime: add and use modtimer in netpoll

Currently when netpoll deadline is incrementally prolonged,
we delete and re-add timer each time.
Add modtimer function that does both and use it when we need
to modify an existing netpoll timer to avoid unnecessary lock/unlock.

TCP4OneShotTimeout-6  17.2µs ± 0%  17.0µs ± 0%  -0.82%  (p=0.008 n=5+5)
SetReadDeadline-6      274ns ± 2%   261ns ± 0%  -4.89%  (p=0.008 n=5+5)

Update #25729

Change-Id: I08b89dbbc1785dd180e967a37b0aa23b0c4613a8
Reviewed-on: https://go-review.googlesource.com/c/146339
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoruntime: don't recreate netpoll timers if they don't change
Dmitry Vyukov [Wed, 31 Oct 2018 15:18:36 +0000 (16:18 +0100)]
runtime: don't recreate netpoll timers if they don't change

Currently we always delete both read and write timers and then
add them again. However, if user setups read and write deadline
separately, then we don't need to touch the other one.

name                  old time/op  new time/op  delta
TCP4OneShotTimeout-6  17.2µs ± 0%  17.2µs ± 0%     ~     (p=0.310 n=5+5)
SetReadDeadline-6      319ns ± 1%   274ns ± 2%  -13.94%  (p=0.008 n=5+5)

Update #25729

Change-Id: I4c869c3083521de6d0cd6ca99a7609d4dd84b4e4
Reviewed-on: https://go-review.googlesource.com/c/146338
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoruntime: don't wake timeproc needlessly
Dmitry Vyukov [Wed, 31 Oct 2018 10:31:31 +0000 (11:31 +0100)]
runtime: don't wake timeproc needlessly

It's not always necessary to wake timerproc even if we add
a new timer to the top of the heap. Since we don't wake and
reset timerproc when we remove timers, it still can be sleeping
with shorter timeout. It such case it's more profitable to let it
sleep and then update timeout when it wakes on its own rather than
proactively wake it, let it update timeout and go to sleep again.

name                  old time/op  new time/op  delta
TCP4OneShotTimeout-6  18.6µs ± 1%  17.2µs ± 0%   -7.66%  (p=0.008 n=5+5)
SetReadDeadline-6      562ns ± 5%   319ns ± 1%  -43.27%  (p=0.008 n=5+5)

Update #25729

Change-Id: Iec8eacb8563dbc574a82358b3bac7ac479c16826
Reviewed-on: https://go-review.googlesource.com/c/146337
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocrypto/md5: simplify generic implementation
Michael Munday [Tue, 29 May 2018 06:26:17 +0000 (07:26 +0100)]
crypto/md5: simplify generic implementation

This change uses library functions such as bits.RotateLeft32 to
reduce the amount of code needed in the generic implementation.
Since the code is now shorter I've also removed the option to
generate a non-unrolled version of the code.

I've also tried to remove bounds checks where possible to make
the new version performant, however that is not the primary goal
of this change since most architectures have assembly
implementations already.

Assembly performance:

name                 old speed      new speed      delta
Hash8Bytes           50.3MB/s ± 1%  59.1MB/s ± 0%  +17.63%  (p=0.000 n=9+8)
Hash1K                590MB/s ± 0%   597MB/s ± 0%   +1.25%  (p=0.000 n=9+9)
Hash8K                636MB/s ± 1%   638MB/s ± 1%     ~     (p=0.072 n=10+10)
Hash8BytesUnaligned  50.5MB/s ± 0%  59.1MB/s ± 1%  +17.09%  (p=0.000 n=10+10)
Hash1KUnaligned       589MB/s ± 1%   596MB/s ± 1%   +1.23%  (p=0.000 n=9+10)
Hash8KUnaligned       638MB/s ± 1%   640MB/s ± 0%   +0.35%  (p=0.002 n=10+10)

Pure Go performance:

name                 old speed      new speed      delta
Hash8Bytes           30.3MB/s ± 1%  42.8MB/s ± 0%  +41.20%  (p=0.000 n=9+9)
Hash1K                364MB/s ± 4%   394MB/s ± 1%   +8.27%  (p=0.000 n=10+10)
Hash8K                404MB/s ± 1%   420MB/s ± 0%   +4.17%  (p=0.000 n=10+9)
Hash8BytesUnaligned  30.3MB/s ± 1%  42.8MB/s ± 1%  +40.92%  (p=0.000 n=9+10)
Hash1KUnaligned       368MB/s ± 0%   394MB/s ± 0%   +7.07%  (p=0.000 n=9+9)
Hash8KUnaligned       404MB/s ± 1%   411MB/s ± 3%   +1.91%  (p=0.026 n=9+10)

Change-Id: I9a91fb52ea8d62964d5351bdf121e9fbc9282852
Reviewed-on: https://go-review.googlesource.com/c/137355
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agoos: use CreateFile for Stat of symlinks
Alex Brainman [Sat, 20 Oct 2018 05:30:57 +0000 (16:30 +1100)]
os: use CreateFile for Stat of symlinks

Stat uses Windows FindFirstFile + CreateFile to gather symlink
information - FindFirstFile determines if file is a symlink,
and then CreateFile follows symlink to capture target details.

Lstat only uses FindFirstFile.

This CL replaces current approach with just a call to CreateFile.
Lstat uses FILE_FLAG_OPEN_REPARSE_POINT flag, that instructs
CreateFile not to follow symlink. Other than that both Stat and
Lstat look the same now. New code is simpler.

CreateFile + GetFileInformationByHandle (unlike FindFirstFile)
does not report reparse tag of a file. I tried to ignore reparse
tag altogether. And it works for symlinks and mount points.
Unfortunately (see https://github.com/moby/moby/issues/37026),
files on deduped disk volumes are reported with
FILE_ATTRIBUTE_REPARSE_POINT attribute set and reparse tag set
to IO_REPARSE_TAG_DEDUP. So, if we ignore reparse tag, Lstat
interprets deduped volume files as symlinks. That is incorrect.
So I had to add GetFileInformationByHandleEx call to gather
reparse tag after calling CreateFile and GetFileInformationByHandle.

Fixes #27225
Fixes #27515

Change-Id: If60233bcf18836c147597cc17450d82f3f88c623
Reviewed-on: https://go-review.googlesource.com/c/143578
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Kirill Kolyshkin <kolyshkin@gmail.com>
6 years agopath/filepath: change IsAbs("NUL") to return true
Alex Brainman [Sun, 21 Oct 2018 03:57:58 +0000 (14:57 +1100)]
path/filepath: change IsAbs("NUL") to return true

This CL changes IsAbs to return true for "NUL" and other Windows
reserved filenames (search
https://docs.microsoft.com/en-us/windows/desktop/fileio/naming-a-file
for NUL for details). os.Open("NUL") and os.Stat("NUL") work
regardless of what current directory is, and it is mistake to join
"NUL" with current directory when building full path. Changing
IsAbs("NUL") to return true fixes that mistake.

Fixes #28035

Change-Id: Ife8f8aee48400702613ede8fc6834fd43e6e0f03
Reviewed-on: https://go-review.googlesource.com/c/145220
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/cgo: don't update each call in place
Ian Lance Taylor [Fri, 2 Nov 2018 05:06:51 +0000 (22:06 -0700)]
cmd/cgo: don't update each call in place

Updating each call in place broke when there were multiple cgo calls
used as arguments to another cgo call where some required rewriting.
Instead, rewrite calls to strings via the existing mangling mechanism,
and only substitute the top level call in place.

Fixes #28540

Change-Id: Ifd66f04c205adc4ad6dd5ee8e79e57dce17e86bb
Reviewed-on: https://go-review.googlesource.com/c/146860
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Dmitri Shuralyov <dmitshur@golang.org>
6 years ago.github: don't render author-facing text in ISSUE_TEMPLATE
Dmitri Shuralyov [Thu, 27 Sep 2018 23:29:39 +0000 (19:29 -0400)]
.github: don't render author-facing text in ISSUE_TEMPLATE

Fixes #27914

Change-Id: Ic359a099661e959eb25d0f763ea16a6f48a3d4c7
Reviewed-on: https://go-review.googlesource.com/c/138295
Reviewed-by: Andrew Bonventre <andybons@golang.org>
6 years agoruntime: only check the existence of variables in gdb info locals test
Martin Möhrmann [Thu, 1 Nov 2018 10:20:11 +0000 (11:20 +0100)]
runtime: only check the existence of variables in gdb info locals test

As discussed in golang.org/cl/28499:
Only test that all expected variables are listed in 'info locals' since
different versions of gdb print variables in different order and with
differing amount of information and formats.

Fixes #28499

Change-Id: I76627351170b5fdf2bf8cbf143e54f628b45dc4e
Reviewed-on: https://go-review.googlesource.com/c/146598
Reviewed-by: Heschi Kreinick <heschi@google.com>
6 years agocmd/link: don't link sections not named .o
Alan Donovan [Wed, 31 Oct 2018 17:18:17 +0000 (13:18 -0400)]
cmd/link: don't link sections not named .o

For many build systems, modular static analysis is most conveniently
implemented by saving analysis facts (which are analogous to export
data) in an additional section in the archive file, similar to
__PKGDEF. See golang.org/x/tools/go/analysis for an overview.

Because such sections are not object files, the linker must not
attempt to link them. This change causes the linker to skip special
sections whose name does not end with .o (and is short enough not to
be truncated).

Fixes #28429

Change-Id: I830852decf868cb017263308b114f72838032993
Reviewed-on: https://go-review.googlesource.com/c/146297
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/cgo: rewrite pointer checking to use more function literals
Ian Lance Taylor [Wed, 17 Oct 2018 15:04:01 +0000 (08:04 -0700)]
cmd/cgo: rewrite pointer checking to use more function literals

Fixes #14210
Fixes #25941

Change-Id: Idde2d032290da3edb742b5b4f6ffeb625f05b494
Reviewed-on: https://go-review.googlesource.com/c/142884
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agodoc/go1.12: initial add of release notes for Go 1.12
Andrew Bonventre [Wed, 31 Oct 2018 16:03:15 +0000 (12:03 -0400)]
doc/go1.12: initial add of release notes for Go 1.12

Change-Id: I9cb3c80ea397d964fe745b74d595df3fd8982a47
Reviewed-on: https://go-review.googlesource.com/c/146257
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agoos: add support for long path names on freebsd RemoveAll
Tobias Klauser [Thu, 1 Nov 2018 08:02:46 +0000 (08:02 +0000)]
os: add support for long path names on freebsd RemoveAll

Follow CL 146020 and enable RemoveAll based on Unlinkat and Openat on
freebsd.

Since the layout of syscall.Stat_t changes in FreeBSD 12, Fstatat needs
a compatibility wrapper akin to Fstatat in x/sys/unix. See CL 138595 and
CL 136816 for details.

Updates #27029

Change-Id: I8851a5b7fa658eaa6e69a1693150b16d9a68f36a
Reviewed-on: https://go-review.googlesource.com/c/146597
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Yuval Pavel Zholkover <paulzhol@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agonet: enable RFC 6555 Fast Fallback by default
Brad Fitzpatrick [Thu, 1 Nov 2018 16:29:42 +0000 (16:29 +0000)]
net: enable RFC 6555 Fast Fallback by default

The Dialer.DualStack field is now meaningless and documented as
deprecated.

To disable fallback, set FallbackDelay to a negative value.

Fixes #22225

Change-Id: Icc212fe07bb69d7651ab81e539b8b3e3d3372fa9
Reviewed-on: https://go-review.googlesource.com/c/146659
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocrypto/cipher: add VSX implementation of xorBytes for ppc64x
Carlos Eduardo Seo [Wed, 19 Sep 2018 21:37:08 +0000 (18:37 -0300)]
crypto/cipher: add VSX implementation of xorBytes for ppc64x

This change adds asm implementations of xorBytes for ppc64x that
takes advantage of VSX registers and instructions.

name                   old time/op    new time/op     delta
XORBytes/8Bytes-8        16.4ns ± 0%     11.1ns ± 0%   -32.32%  (p=0.000 n=5+4)
XORBytes/128Bytes-8      45.6ns ± 0%     16.2ns ± 0%   -64.50%  (p=0.008 n=5+5)
XORBytes/2048Bytes-8      433ns ±13%      129ns ± 1%   -70.29%  (p=0.000 n=5+4)
XORBytes/32768Bytes-8    7.16µs ± 0%     1.83µs ± 0%   -74.39%  (p=0.008 n=5+5)

name                   old speed      new speed       delta
XORBytes/8Bytes-8       488MB/s ± 0%    721MB/s ± 0%   +47.75%  (p=0.016 n=5+4)
XORBytes/128Bytes-8    2.80GB/s ± 0%   7.89GB/s ± 0%  +181.33%  (p=0.008 n=5+5)
XORBytes/2048Bytes-8   4.77GB/s ±13%  15.87GB/s ± 0%  +232.68%  (p=0.016 n=5+4)
XORBytes/32768Bytes-8  4.58GB/s ± 0%  17.88GB/s ± 0%  +290.47%  (p=0.008 n=5+5)

Change-Id: Ic27d9b858f8ec2d597fdabc68a288d6844eba701
Reviewed-on: https://go-review.googlesource.com/c/145997
Run-TryBot: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
6 years agonet/textproto: do not buffer a line if we know the next line is empty
Travis Bischel [Sun, 9 Sep 2018 22:00:38 +0000 (15:00 -0700)]
net/textproto: do not buffer a line if we know the next line is empty

readContinuedLineSlice intends to buffer a continued line of text, where
a continued line can continue through newlines so long as the next line
begins with a space or tab.

The current optimization is to not try to buffer and build a line if we
immediately see that the next line begins with an ASCII character.

This adds avoiding copying the line if we see that the next line is \n
or \r\n as well.

Notably, headers always end in \r\n\r\n. In the general, well formatted
header case, we can now avoid ever allocating textproto.Reader's
internal reusable buf.

This can mildly be seen in net/http's BenchmarkClientServer:

name            old time/op    new time/op    delta
ClientServer-4    66.4µs ± 0%    66.2µs ± 0%  -0.35%  (p=0.004 n=10+10)

name            old alloc/op   new alloc/op   delta
ClientServer-4    4.87kB ± 0%    4.82kB ± 0%  -1.01%  (p=0.000 n=6+10)

name            old allocs/op  new allocs/op  delta
ClientServer-4      64.0 ± 0%      63.0 ± 0%  -1.56%  (p=0.000 n=10+10)

Change-Id: Id8c2ab69086ac481b90abda289396dcb7bfe8851
Reviewed-on: https://go-review.googlesource.com/c/134227
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agodoc: improve language in contribute
tkivisik [Wed, 17 Oct 2018 12:42:54 +0000 (12:42 +0000)]
doc: improve language in contribute

Previous wording was incorrect.

Change-Id: I91406c775161d9724ec60824c156e8e9a925bcd7
GitHub-Last-Rev: 2b6b4e136a78a431cc847a2717cbeae4341ed7bc
GitHub-Pull-Request: golang/go#28248
Reviewed-on: https://go-review.googlesource.com/c/142879
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
6 years agocmd/compile: document Sym flags
Austin Clements [Wed, 24 Oct 2018 21:59:25 +0000 (17:59 -0400)]
cmd/compile: document Sym flags

Change-Id: Id14b417095628c7a1dc7a8e47bc28cfa392b5262
Reviewed-on: https://go-review.googlesource.com/c/146498
Reviewed-by: Robert Griesemer <gri@golang.org>
6 years agoos: don't create files in local directory
Ian Lance Taylor [Wed, 31 Oct 2018 22:06:59 +0000 (15:06 -0700)]
os: don't create files in local directory

Also, use a random temporary directory rather than os.TempDir.  Defer
removal of existing random temporary directories.

Change-Id: Id7549031cdf78a2bab28c07b6eeff621bdf6e49c
Reviewed-on: https://go-review.googlesource.com/c/146457
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
6 years agocmd/compile: gofmt
Austin Clements [Thu, 1 Nov 2018 00:46:57 +0000 (20:46 -0400)]
cmd/compile: gofmt

I don't know how this file wasn't gofmted.

Change-Id: I9b3765ae63970b7bc4dc87107f546e64a78e2830
Reviewed-on: https://go-review.googlesource.com/c/146497
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoruntime: exit early when scanning map buckets
Keith Randall [Mon, 15 Oct 2018 22:14:48 +0000 (15:14 -0700)]
runtime: exit early when scanning map buckets

Divide the "empty" slot state into two, "emptyOne" and "emptyRest".
emptyOne means just that slot is empty. emptyRest means all subsequent
slots in that bucket are empty and the overflow pointer is nil.

When scanning a bucket, we can often stop at emptyRest, reducing
the total work we have to do. (This is similar to how tombstones
work in open addressing.)

Ideally on delete we have to figure out whether to zero the slot
with an emptyOne or emptyRest marker. For now, we choose the safe
but non-optimal choice. (Fix in subsequent CL?)

This is a simpler CL than some others we've tried, including my
CL sequence 11835[5-8] and Ilya's CL 115616.

Update #19495

name                           old time/op    new time/op    delta
MegMap                           8.96ns ± 2%    8.74ns ± 6%   -2.44%  (p=0.020 n=10+10)
MegOneMap                        8.91ns ± 2%    5.53ns ± 2%  -37.99%  (p=0.000 n=10+10)
MegEqMap                         46.0µs ± 1%    45.8µs ± 3%     ~     (p=0.315 n=9+10)
MegEmptyMap                      2.50ns ± 0%    2.50ns ± 2%     ~     (p=0.957 n=8+10)
SmallStrMap                      8.54ns ± 1%    8.71ns ± 2%   +2.01%  (p=0.000 n=10+10)
MapStringKeysEight_16            8.61ns ± 3%    8.71ns ± 3%   +1.20%  (p=0.026 n=9+9)
MapStringKeysEight_32            8.54ns ± 2%    8.97ns ± 1%   +5.05%  (p=0.000 n=10+9)
MapStringKeysEight_64            8.66ns ± 2%    8.99ns ± 2%   +3.87%  (p=0.000 n=10+10)
MapStringKeysEight_1M            8.57ns ± 2%    8.95ns ± 2%   +4.51%  (p=0.000 n=10+9)
IntMap                           6.69ns ± 1%    7.46ns ± 1%  +11.60%  (p=0.000 n=9+9)
MapFirst/1                       3.69ns ± 1%    3.63ns ± 3%   -1.52%  (p=0.040 n=10+10)
MapFirst/2                       3.70ns ± 2%    3.63ns ± 2%   -1.95%  (p=0.001 n=9+9)
MapFirst/3                       3.74ns ± 2%    3.66ns ± 2%   -2.12%  (p=0.000 n=8+10)
MapFirst/4                       3.71ns ± 2%    3.66ns ± 4%     ~     (p=0.073 n=9+10)
MapFirst/5                       3.69ns ± 1%    3.62ns ± 2%   -1.88%  (p=0.000 n=9+10)
MapFirst/6                       3.68ns ± 2%    3.62ns ± 1%   -1.83%  (p=0.001 n=10+9)
MapFirst/7                       3.67ns ± 1%    3.60ns ± 1%   -1.98%  (p=0.000 n=10+8)
MapFirst/8                       3.68ns ± 2%    3.61ns ± 2%   -1.87%  (p=0.000 n=10+10)
MapFirst/9                       8.03ns ± 4%    7.89ns ± 2%   -1.76%  (p=0.007 n=10+10)
MapFirst/10                      7.99ns ± 2%    7.86ns ± 3%   -1.64%  (p=0.009 n=9+10)
MapFirst/11                      7.96ns ± 1%    7.80ns ± 2%   -2.01%  (p=0.000 n=10+10)
MapFirst/12                      7.96ns ± 1%    7.82ns ± 1%   -1.67%  (p=0.000 n=10+10)
MapFirst/13                      8.06ns ± 3%    7.92ns ± 3%     ~     (p=0.055 n=10+10)
MapFirst/14                      7.95ns ± 1%    7.80ns ± 1%   -1.88%  (p=0.000 n=10+9)
MapFirst/15                      8.01ns ± 2%    7.80ns ± 2%   -2.57%  (p=0.000 n=10+10)
MapFirst/16                      8.05ns ± 2%    7.90ns ± 2%   -1.84%  (p=0.005 n=9+10)
MapMid/1                         4.00ns ± 1%    3.94ns ± 2%   -1.30%  (p=0.021 n=8+9)
MapMid/2                         4.39ns ± 2%    4.32ns ± 4%     ~     (p=0.128 n=10+10)
MapMid/3                         4.40ns ± 2%    4.27ns ± 2%   -2.93%  (p=0.000 n=10+9)
MapMid/4                         4.76ns ± 2%    4.65ns ± 1%   -2.26%  (p=0.000 n=10+9)
MapMid/5                         4.76ns ± 1%    4.65ns ± 1%   -2.27%  (p=0.000 n=10+10)
MapMid/6                         5.11ns ± 2%    4.98ns ± 2%   -2.55%  (p=0.000 n=10+10)
MapMid/7                         5.12ns ± 1%    5.01ns ± 3%   -2.02%  (p=0.003 n=9+9)
MapMid/8                         5.71ns ± 3%    5.97ns ± 1%   +4.51%  (p=0.000 n=10+9)
MapMid/9                         8.72ns ±10%    8.89ns ±10%     ~     (p=0.458 n=9+10)
MapMid/10                        10.1ns ±15%     9.6ns ± 7%     ~     (p=0.080 n=9+10)
MapMid/11                        9.88ns ±10%    9.44ns ±11%     ~     (p=0.065 n=10+10)
MapMid/12                        9.90ns ±13%   10.04ns ± 9%     ~     (p=1.000 n=10+8)
MapMid/13                        9.67ns ±14%   10.23ns ±10%     ~     (p=0.209 n=10+9)
MapMid/14                        9.12ns ±14%    9.14ns ±13%     ~     (p=0.927 n=10+10)
MapMid/15                        9.16ns ±12%    9.15ns ±16%     ~     (p=0.955 n=10+10)
MapMid/16                        9.37ns ±11%    9.60ns ±23%     ~     (p=0.825 n=9+10)
MapLast/1                        4.08ns ± 1%    3.92ns ± 0%   -3.91%  (p=0.000 n=10+9)
MapLast/2                        4.37ns ± 1%    4.28ns ± 1%   -1.95%  (p=0.000 n=10+10)
MapLast/3                        4.94ns ± 2%    4.65ns ± 1%   -5.79%  (p=0.000 n=9+8)
MapLast/4                        5.40ns ± 3%    5.02ns ± 2%   -7.13%  (p=0.000 n=9+9)
MapLast/5                        5.88ns ± 2%    5.67ns ± 2%   -3.57%  (p=0.000 n=10+10)
MapLast/6                        6.48ns ± 3%    5.90ns ± 2%   -8.89%  (p=0.000 n=10+10)
MapLast/7                        7.01ns ± 2%    6.27ns ± 5%  -10.56%  (p=0.000 n=10+10)
MapLast/8                        7.60ns ± 2%    6.62ns ± 2%  -12.93%  (p=0.000 n=9+10)
MapLast/9                        10.6ns ± 9%    10.9ns ±15%     ~     (p=0.344 n=9+10)
MapLast/10                       11.0ns ±12%    10.9ns ±14%     ~     (p=0.985 n=10+10)
MapLast/11                       11.4ns ±12%    11.8ns ±22%     ~     (p=0.671 n=10+10)
MapLast/12                       11.6ns ±10%    12.1ns ±19%     ~     (p=0.617 n=10+10)
MapLast/13                       12.5ns ±23%    11.8ns ±13%     ~     (p=0.827 n=10+9)
MapLast/14                       10.5ns ±22%    10.4ns ± 5%     ~     (p=0.797 n=10+9)
MapLast/15                       10.0ns ±15%    10.3ns ±16%     ~     (p=0.565 n=10+10)
MapLast/16                       10.4ns ±12%    10.5ns ±13%     ~     (p=0.889 n=10+9)
MapCycle                         22.3ns ± 1%    22.0ns ± 2%   -1.43%  (p=0.002 n=9+10)
RepeatedLookupStrMapKey32        16.4ns ± 1%    16.6ns ± 1%   +1.24%  (p=0.000 n=10+9)
RepeatedLookupStrMapKey1M        35.6µs ± 0%    35.4µs ± 1%   -0.62%  (p=0.002 n=10+10)
NewEmptyMap                      5.36ns ± 1%    9.05ns ± 1%  +69.02%  (p=0.000 n=10+8)
NewSmallMap                      51.2ns ± 2%    33.7ns ± 1%  -34.22%  (p=0.000 n=10+9)
MapIter                          83.8ns ± 1%    88.4ns ± 1%   +5.55%  (p=0.000 n=10+10)
MapIterEmpty                     4.32ns ± 3%    5.54ns ± 3%  +28.12%  (p=0.000 n=10+10)
SameLengthMap                    4.31ns ± 1%    4.59ns ± 2%   +6.41%  (p=0.000 n=9+10)
BigKeyMap                        24.2ns ± 2%    24.3ns ± 1%     ~     (p=0.432 n=10+10)
BigValMap                        24.3ns ± 1%    24.4ns ± 2%     ~     (p=0.200 n=10+9)
SmallKeyMap                      17.5ns ± 1%    18.5ns ± 2%   +5.81%  (p=0.000 n=9+10)
MapPopulate/1                    29.0ns ± 4%    18.8ns ± 1%  -35.27%  (p=0.000 n=10+9)
MapPopulate/10                    736ns ± 5%     693ns ± 4%   -5.92%  (p=0.000 n=10+10)
MapPopulate/100                  11.3µs ± 2%    10.8µs ± 3%   -4.38%  (p=0.000 n=10+10)
MapPopulate/1000                  139µs ± 8%     132µs ± 4%   -5.10%  (p=0.002 n=10+10)
MapPopulate/10000                1.21ms ± 5%    1.16ms ± 5%   -4.56%  (p=0.002 n=10+10)
MapPopulate/100000               12.2ms ± 3%    11.8ms ± 5%     ~     (p=0.052 n=10+10)
ComplexAlgMap                    73.9ns ± 1%    74.4ns ± 2%     ~     (p=0.161 n=9+10)
GoMapClear/Reflexive/1           36.0ns ± 1%    26.9ns ± 2%  -25.31%  (p=0.000 n=10+10)
GoMapClear/Reflexive/10          35.2ns ± 1%    24.4ns ± 1%  -30.62%  (p=0.000 n=10+10)
GoMapClear/Reflexive/100         69.6ns ± 2%    59.2ns ± 1%  -14.92%  (p=0.000 n=10+10)
GoMapClear/Reflexive/1000        1.06µs ± 2%    1.05µs ± 1%   -1.16%  (p=0.013 n=10+9)
GoMapClear/Reflexive/10000       11.7µs ± 1%    11.7µs ± 1%     ~     (p=0.542 n=10+10)
GoMapClear/NonReflexive/1        96.3ns ± 1%    90.0ns ± 1%   -6.52%  (p=0.000 n=10+10)
GoMapClear/NonReflexive/10        110ns ± 2%     101ns ± 0%   -8.10%  (p=0.000 n=10+7)
GoMapClear/NonReflexive/100       270ns ± 2%     235ns ± 2%  -12.94%  (p=0.000 n=10+10)
GoMapClear/NonReflexive/1000     3.02µs ± 2%    2.48µs ± 1%  -17.92%  (p=0.000 n=10+10)
GoMapClear/NonReflexive/10000    23.7µs ± 1%    19.6µs ± 1%  -17.30%  (p=0.000 n=10+9)
MapPop100                        9.65µs ± 6%    9.18µs ± 8%   -4.82%  (p=0.008 n=9+10)
MapPop1000                        162µs ± 6%     148µs ± 4%   -8.67%  (p=0.000 n=9+9)
MapPop10000                      3.05ms ± 8%    2.82ms ±15%   -7.66%  (p=0.023 n=10+10)
MapAssign/Int32/256              15.7ns ± 4%    14.6ns ± 2%   -7.08%  (p=0.000 n=10+10)
MapAssign/Int32/65536            29.8ns ± 1%    30.4ns ± 0%   +2.04%  (p=0.000 n=10+8)
MapAssign/Int64/256              14.9ns ± 5%    14.8ns ± 4%     ~     (p=0.611 n=10+10)
MapAssign/Int64/65536            30.3ns ± 2%    30.4ns ± 1%   +0.54%  (p=0.046 n=10+9)
MapAssign/Str/256                17.8ns ± 3%    19.8ns ± 4%  +11.08%  (p=0.000 n=10+10)
MapAssign/Str/65536              35.7ns ± 1%    36.4ns ± 1%   +1.82%  (p=0.000 n=10+10)
MapOperatorAssign/Int32/256      18.8ns ± 5%    14.6ns ± 3%  -22.57%  (p=0.000 n=10+10)
MapOperatorAssign/Int32/65536    29.8ns ± 1%    30.5ns ± 1%   +2.39%  (p=0.000 n=10+10)
MapOperatorAssign/Int64/256      16.6ns ± 4%    15.0ns ± 6%   -9.34%  (p=0.000 n=10+10)
MapOperatorAssign/Int64/65536    30.1ns ± 1%    31.7ns ± 2%   +5.21%  (p=0.000 n=10+10)
MapOperatorAssign/Str/256        1.70µs ± 1%    1.61µs ± 2%   -5.55%  (p=0.000 n=10+8)
MapOperatorAssign/Str/65536       289ns ± 7%     294ns ± 4%     ~     (p=0.425 n=10+10)
MapAppendAssign/Int32/256        34.3ns ± 2%    31.0ns ± 3%   -9.59%  (p=0.000 n=9+9)
MapAppendAssign/Int32/65536      51.8ns ± 3%    47.1ns ±13%   -9.17%  (p=0.002 n=9+10)
MapAppendAssign/Int64/256        32.5ns ± 8%    31.2ns ± 6%     ~     (p=0.065 n=10+10)
MapAppendAssign/Int64/65536      51.4ns ± 4%    47.2ns ±10%   -8.07%  (p=0.005 n=9+10)
MapAppendAssign/Str/256           105ns ±12%     109ns ± 4%     ~     (p=0.138 n=10+8)
MapAppendAssign/Str/65536         101ns ±14%      81ns ± 8%  -19.82%  (p=0.000 n=10+9)
MapDelete/Int32/100              32.0ns ± 1%    35.0ns ± 2%   +9.59%  (p=0.000 n=9+10)
MapDelete/Int32/1000             27.0ns ± 3%    30.3ns ± 1%  +12.10%  (p=0.000 n=10+9)
MapDelete/Int32/10000            29.2ns ± 1%    32.9ns ± 2%  +12.80%  (p=0.000 n=10+10)
MapDelete/Int64/100              31.5ns ± 1%    35.7ns ± 2%  +13.16%  (p=0.000 n=10+10)
MapDelete/Int64/1000             27.0ns ± 2%    30.6ns ± 1%  +13.21%  (p=0.000 n=10+10)
MapDelete/Int64/10000            30.3ns ± 1%    34.4ns ± 3%  +13.47%  (p=0.000 n=10+10)
MapDelete/Str/100                23.4ns ± 8%    26.7ns ± 6%  +14.10%  (p=0.000 n=10+9)
MapDelete/Str/1000               31.0ns ± 2%    35.1ns ± 3%  +13.19%  (p=0.000 n=10+9)
MapDelete/Str/10000              38.8ns ± 1%    43.4ns ± 2%  +12.02%  (p=0.000 n=9+10)

Change-Id: I564ce0f40936589f0f9b837f7f2bbcca4c4a1070
Reviewed-on: https://go-review.googlesource.com/c/142437
Reviewed-by: Giovanni Bajo <rasky@develer.com>
Reviewed-by: Martin Möhrmann <martisch@uos.de>
Run-TryBot: Martin Möhrmann <martisch@uos.de>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/go: don't pass empty string to ar when using -compiler=gccgo
Ian Lance Taylor [Wed, 31 Oct 2018 16:32:51 +0000 (09:32 -0700)]
cmd/go: don't pass empty string to ar when using -compiler=gccgo

Correct error introduced in CL 145417 on non-AIX systems.

Fixes #28511

Change-Id: I6624939061425af19faccedd271f465d1fe6b975
Reviewed-on: https://go-review.googlesource.com/c/146277
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agodoc: replace command line usages of godoc with go doc
Agniva De Sarker [Mon, 29 Oct 2018 06:36:58 +0000 (12:06 +0530)]
doc: replace command line usages of godoc with go doc

Effective Go and the FAQ still had some instances which showed the command line
usage of godoc. Changed them to use go doc.

Updates #25443

Change-Id: If550963322034e6848bc466f79e968e7220e4a88
Reviewed-on: https://go-review.googlesource.com/c/145222
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/compile/internal/gc: more direct noder.nod implementation (cleanup)
Robert Griesemer [Wed, 31 Oct 2018 18:38:37 +0000 (11:38 -0700)]
cmd/compile/internal/gc: more direct noder.nod implementation (cleanup)

Also, renamed

- noder.lineno -> noder.setlineno (because that's what it does)
- noder.setlineno -> noder.pos    (and return the src.XPos)

Change-Id: I5d3442cf2af97028afcab028290152ce8d062927
Reviewed-on: https://go-review.googlesource.com/c/146317
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
6 years agoos: add support for long path names on solaris RemoveAll
Tobias Klauser [Tue, 30 Oct 2018 00:40:24 +0000 (00:40 +0000)]
os: add support for long path names on solaris RemoveAll

Follow CL 146020 and enable RemoveAll based on Unlinkat and Openat on
solaris.

Updates #27029

Change-Id: I0b0e92f4422fa960a13dcd3e9adb57cd23f09ed4
Reviewed-on: https://go-review.googlesource.com/c/145839
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agotime: display results in examples rather soon
Yury Smolsky [Wed, 31 Oct 2018 15:15:32 +0000 (17:15 +0200)]
time: display results in examples rather soon

We have fixed the playground to display results of
the program when it was timed out.

This CL fixes how soon results will be displayed to the user.

Change-Id: Ifb75828e0de12c726c8ca6e2d04947e01913dc73
Reviewed-on: https://go-review.googlesource.com/c/146237
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agointernal/cpu, runtime: add CPU feature detection support for AIX
Clément Chigot [Thu, 25 Oct 2018 09:23:53 +0000 (11:23 +0200)]
internal/cpu, runtime: add CPU feature detection support for AIX

AIX doesn't have HWCAP/HWCAP2 variables like Linux. Therefore, it relies on
getsystemcfg syscall which can provide some information about the CPU.

Change-Id: Ic0dc927e80890d4bf8f0bdfb43fad1e2b890d7a0
Reviewed-on: https://go-review.googlesource.com/c/144959
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Martin Möhrmann <moehrmann@google.com>
6 years agogo/build, cmd/go: add "hurd" as a GOOS value
Ian Lance Taylor [Wed, 31 Oct 2018 14:36:42 +0000 (07:36 -0700)]
go/build, cmd/go: add "hurd" as a GOOS value

RELNOTES=yes

Change-Id: Ie7090b5a6edd548a49bdc4295e01cb686dfe6522
Reviewed-on: https://go-review.googlesource.com/c/146023
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocrypto/tls: cache Leaf certificate during BuildNameToCertificate
Darien Raymond [Wed, 31 Oct 2018 14:37:26 +0000 (14:37 +0000)]
crypto/tls: cache Leaf certificate during BuildNameToCertificate

I am working on a TLS server program, which issues new TLS certificates
on demand. The new certificates will be added into tls.Config.Certificates.
BuildNameToCertificate will be called to refresh the name table afterwards.
This change will reduce some workload on existing certificates.

Note that you can’t modify the Certificates field (or call BuildNameToCertificate)
on a Config in use by a Server. You can however modify an unused Config that gets
cloned in GetConfigForClient with appropriate locking.

Change-Id: I7bdb7d23fc5d68df83c73f3bfa3ba9181d38fbde
GitHub-Last-Rev: c3788f4116be47f2fdb777935c421e7dd694f5c8
GitHub-Pull-Request: golang/go#24920
Reviewed-on: https://go-review.googlesource.com/c/107627
Reviewed-by: Filippo Valsorda <filippo@golang.org>
6 years agoos: add support for long path names on unix RemoveAll
Oliver Stenbom [Wed, 31 Oct 2018 10:55:24 +0000 (10:55 +0000)]
os: add support for long path names on unix RemoveAll

On unix systems, long enough path names will fail when performing syscalls
like `Lstat`. The current RemoveAll uses several of these syscalls, and so
will fail for long paths. This can be risky, as it can let users "hide"
files from the system or otherwise make long enough paths for programs
to fail. By using `Unlinkat` and `Openat` syscalls instead, RemoveAll is
safer on unix systems. Initially implemented for linux, darwin, dragonfly,
netbsd and openbsd. Not yet implemented on freebsd due to fstatat 64-bit
inode compatibility issues.

Fixes #27029

Co-authored-by: Giuseppe Capizzi <gcapizzi@pivotal.io>
Co-authored-by: Julia Nedialkova <yulia.nedyalkova@sap.com>
Change-Id: I978a6a4986878fe076d3c7af86e7927675624a96
GitHub-Last-Rev: 9235489c81b90c228210144b7c25b28a46bb80b7
GitHub-Pull-Request: golang/go#28494
Reviewed-on: https://go-review.googlesource.com/c/146020
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd: add XCOFF objfile and adapt cmd/nm tests
Clément Chigot [Mon, 22 Oct 2018 15:00:37 +0000 (17:00 +0200)]
cmd: add XCOFF objfile and adapt cmd/nm tests

This commit adds a new file format in cmd/internal/objfile for XCOFF.
It also adapts tests inside cmd/nm for AIX.

Updates: #25893

Change-Id: I1e55ea0b7f7d08a871343bee27d11e2d3baad254
Reviewed-on: https://go-review.googlesource.com/c/145397
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agoos: use Stat instead of Lstat in Symlink
Alex Brainman [Fri, 26 Oct 2018 07:44:17 +0000 (18:44 +1100)]
os: use Stat instead of Lstat in Symlink

Windows implementation of Symlink uses CreateSymbolicLink Windows
API. The API requires to identify the target type: file or
directory. Current Symlink implementation  uses Lstat to determine
symlink type, but Lstat will not be able to determine correct
result if destination is symlink. Replace Lstat call with Stat.

Fixes #28432

Change-Id: Ibee6d8ac21e2246bf8d0a019c4c66d38b09887d4
Reviewed-on: https://go-review.googlesource.com/c/145217
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/compile: better documentation around checkwidth
Robert Griesemer [Mon, 29 Oct 2018 20:44:44 +0000 (13:44 -0700)]
cmd/compile: better documentation around checkwidth

Change-Id: I5c7ec9676b5573c883c196459acea85aa9ff8130
Reviewed-on: https://go-review.googlesource.com/c/146021
Run-TryBot: Robert Griesemer <gri@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/doc: allow -all to apply to individual items
Rob Pike [Wed, 24 Oct 2018 02:33:14 +0000 (13:33 +1100)]
cmd/doc: allow -all to apply to individual items

It really only matters for types, and the code already worked but was
blocked by a usage check.

Fixes #25595

Change-Id: I823f313b682b37616ea555aee079e2fe39f914c2
Reviewed-on: https://go-review.googlesource.com/c/144357
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd: allow build with gccgo on AIX
Clément Chigot [Mon, 29 Oct 2018 12:39:53 +0000 (13:39 +0100)]
cmd: allow build with gccgo on AIX

This commit adapts cmd/internal/buildid and cmd/go to allow the use of
gccgo on AIX.
Buildid is supported only for AIX archives.

Change-Id: I14c790a8994ae8d2ee629d8751e04189c30ffd94
Reviewed-on: https://go-review.googlesource.com/c/145417
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
6 years agocmd/compile: implement some moves using non-overlapping reads&writes
Keith Randall [Tue, 30 Oct 2018 19:58:50 +0000 (12:58 -0700)]
cmd/compile: implement some moves using non-overlapping reads&writes

For moves >8,<16 bytes, do a move using non-overlapping loads/stores
if it would require no more instructions.

This helps a bit with the case when the move is from a static
constant, because then the code to materialize the value being moved
is smaller.

Change-Id: Ie47a5a7c654afeb4973142b0a9922faea13c9b54
Reviewed-on: https://go-review.googlesource.com/c/146019
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
6 years agostrings: declare IndexByte as noescape
Keith Randall [Tue, 30 Oct 2018 17:56:02 +0000 (10:56 -0700)]
strings: declare IndexByte as noescape

This lets []byte->string conversions which are used as arguments to
strings.IndexByte and friends have their backing store allocated on
the stack.

It only prevents allocation when the string is small enough (32
bytes), so it isn't perfect. But reusing the []byte backing store
directly requires a bunch more compiler analysis (see #2205 and
related issues).

Fixes #25864.

Change-Id: Ie52430422196e3c91e5529d6e56a8435ced1fc4c
Reviewed-on: https://go-review.googlesource.com/c/146018
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
6 years agocmd/compile: when comparing 0-size types, make sure expr side-effects survive
Keith Randall [Tue, 30 Oct 2018 00:02:42 +0000 (17:02 -0700)]
cmd/compile: when comparing 0-size types, make sure expr side-effects survive

Fixes #23837

Change-Id: I53f524d87946a0065f28a4ddbe47b40f2b43c459
Reviewed-on: https://go-review.googlesource.com/c/145757
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
6 years agocrypto/cipher: use SIMD for xor on amd64
templexxx [Fri, 20 Jul 2018 20:18:08 +0000 (04:18 +0800)]
crypto/cipher: use SIMD for xor on amd64

cpu: Intel(R) Core(TM) i7-7700HQ CPU @ 2.80GHz

Benchmark: xor

name                   old time/op    new time/op     delta
XORBytes/8Bytes-8        8.21ns ± 1%     6.35ns ± 3%   -22.66%  (p=0.008 n=5+5)
XORBytes/128Bytes-8      17.9ns ± 1%     10.4ns ± 1%   -41.68%  (p=0.008 n=5+5)
XORBytes/2048Bytes-8      187ns ± 1%       78ns ± 0%   -58.44%  (p=0.008 n=5+5)
XORBytes/32768Bytes-8    2.87µs ± 1%     1.38µs ± 0%   -52.05%  (p=0.008 n=5+5)

name                   old speed      new speed       delta
XORBytes/8Bytes-8       974MB/s ± 1%   1260MB/s ± 2%   +29.33%  (p=0.008 n=5+5)
XORBytes/128Bytes-8    7.15GB/s ± 0%  12.25GB/s ± 1%   +71.17%  (p=0.008 n=5+5)
XORBytes/2048Bytes-8   10.9GB/s ± 1%   26.4GB/s ± 0%  +140.99%  (p=0.008 n=5+5)
XORBytes/32768Bytes-8  11.4GB/s ± 1%   23.8GB/s ± 0%  +108.52%  (p=0.008 n=5+5)

Benchmark: cipher

name               old time/op    new time/op    delta
AESGCMSeal1K-8        269ns ± 6%     261ns ± 2%     ~     (p=0.246 n=5+5)
AESGCMOpen1K-8        242ns ± 1%     240ns ± 2%     ~     (p=0.190 n=5+5)
AESGCMSign8K-8        869ns ± 0%     870ns ± 1%     ~     (p=0.683 n=5+5)
AESGCMSeal8K-8       1.64µs ± 6%    1.59µs ± 7%     ~     (p=0.151 n=5+5)
AESGCMOpen8K-8       1.48µs ± 2%    1.46µs ± 0%   -1.39%  (p=0.008 n=5+5)
AESCFBEncrypt1K-8    1.88µs ± 5%    1.62µs ± 1%  -13.52%  (p=0.008 n=5+5)
AESCFBDecrypt1K-8    1.76µs ± 1%    1.58µs ± 1%  -10.24%  (p=0.016 n=4+5)
AESOFB1K-8           1.10µs ± 4%    1.03µs ± 2%   -6.36%  (p=0.008 n=5+5)
AESCTR1K-8           1.24µs ± 1%    1.17µs ± 0%   -5.96%  (p=0.008 n=5+5)
AESCBCEncrypt1K-8    1.74µs ± 0%    1.14µs ± 1%  -34.36%  (p=0.008 n=5+5)
AESCBCDecrypt1K-8    1.28µs ± 1%    1.10µs ± 1%  -14.04%  (p=0.008 n=5+5)

name               old speed      new speed      delta
AESGCMSeal1K-8     3.81GB/s ± 6%  3.91GB/s ± 2%     ~     (p=0.310 n=5+5)
AESGCMOpen1K-8     4.23GB/s ± 1%  4.27GB/s ± 2%     ~     (p=0.222 n=5+5)
AESGCMSign8K-8     9.43GB/s ± 0%  9.41GB/s ± 1%     ~     (p=0.841 n=5+5)
AESGCMSeal8K-8     5.01GB/s ± 6%  5.16GB/s ± 6%     ~     (p=0.151 n=5+5)
AESGCMOpen8K-8     5.54GB/s ± 2%  5.62GB/s ± 0%   +1.41%  (p=0.008 n=5+5)
AESCFBEncrypt1K-8   543MB/s ± 5%   627MB/s ± 1%  +15.55%  (p=0.008 n=5+5)
AESCFBDecrypt1K-8   580MB/s ± 1%   646MB/s ± 1%  +11.40%  (p=0.016 n=4+5)
AESOFB1K-8          925MB/s ± 4%   988MB/s ± 2%   +6.73%  (p=0.008 n=5+5)
AESCTR1K-8          821MB/s ± 1%   873MB/s ± 1%   +6.34%  (p=0.008 n=5+5)
AESCBCEncrypt1K-8   588MB/s ± 1%   897MB/s ± 1%  +52.36%  (p=0.008 n=5+5)
AESCBCDecrypt1K-8   799MB/s ± 1%   929MB/s ± 1%  +16.32%  (p=0.008 n=5+5)

Change-Id: I42e6ba66c23dad853d33c924fca7b0ed805cefdd
Reviewed-on: https://go-review.googlesource.com/c/125316
Reviewed-by: Ilya Tocar <ilya.tocar@intel.com>
Run-TryBot: Ilya Tocar <ilya.tocar@intel.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agoRevert "os: add support for long path names on unix RemoveAll"
Katie Hockman [Tue, 30 Oct 2018 16:42:55 +0000 (16:42 +0000)]
Revert "os: add support for long path names on unix RemoveAll"

This reverts commit 85143d355493c6bba994d49ed154b4df8b78874b.

Reason for revert: Breaking all Darwin and FreeBSD builds. Trybots did not pass for this.

Change-Id: I5494e14ad5ab9cf6e1e225a25b2e8b38f3359d13
Reviewed-on: https://go-review.googlesource.com/c/145897
Reviewed-by: Katie Hockman <katie@golang.org>
Run-TryBot: Katie Hockman <katie@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

6 years agocmd/trace: use new traceparser to parse the raw trace files
Peter Weinberger [Mon, 29 Oct 2018 14:18:05 +0000 (10:18 -0400)]
cmd/trace: use new traceparser to parse the raw trace files

Change-Id: I8b224ae48a2f8acd5a64c9ff283e97821479a9a8
Reviewed-on: https://go-review.googlesource.com/c/145457
Run-TryBot: Peter Weinberger <pjw@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Hyang-Ah Hana Kim <hyangah@gmail.com>
6 years agocrypto/cipher: make stream examples runnable in the playground
Yury Smolsky [Tue, 30 Oct 2018 15:22:05 +0000 (17:22 +0200)]
crypto/cipher: make stream examples runnable in the playground

Updates #9679

Change-Id: I53412cf0142364de5f76e8affc15d607bfa2ad23
Reviewed-on: https://go-review.googlesource.com/c/145838
Run-TryBot: Yury Smolsky <yury@smolsky.by>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
6 years agoruntime: add physical memory scavenging test
Michael Anthony Knyszek [Mon, 1 Oct 2018 19:58:01 +0000 (19:58 +0000)]
runtime: add physical memory scavenging test

This change introduces a test to malloc_test which checks for overuse
of physical memory in the large object treap. Due to fragmentation,
there may be many pages of physical memory that are sitting unused in
large-object space.

For #14045.

Change-Id: I3722468f45063b11246dde6301c7ad02ae34be55
Reviewed-on: https://go-review.googlesource.com/c/138918
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agoruntime: scavenge large spans before heap growth
Michael Anthony Knyszek [Thu, 4 Oct 2018 15:33:08 +0000 (15:33 +0000)]
runtime: scavenge large spans before heap growth

This change scavenges the largest spans before growing the heap for
physical pages to "make up" for the newly-mapped space which,
presumably, will be touched.

In theory, this approach to scavenging helps reduce the RSS of an
application by marking fragments in memory as reclaimable to the OS
more eagerly than before. In practice this may not necessarily be
true, depending on how sysUnused is implemented for each platform.

Fixes #14045.

Change-Id: Iab60790be05935865fc71f793cb9323ab00a18bd
Reviewed-on: https://go-review.googlesource.com/c/139719
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agoruntime: sysUsed spans after trimming
Michael Anthony Knyszek [Fri, 5 Oct 2018 18:11:02 +0000 (18:11 +0000)]
runtime: sysUsed spans after trimming

Currently, we mark a whole span as sysUsed before trimming, but this
unnecessarily tells the OS that the trimmed section from the span is
used when it may have been scavenged, if s was scavenged. Overall,
this just makes invocations of sysUsed a little more fine-grained.

It does come with the caveat that now heap_released needs to be managed
a little more carefully in allocSpanLocked. In this case, we choose to
(like before this change) negate any effect the span has on
heap_released before trimming, then add it back if the trimmed part is
scavengable.

For #14045.

Change-Id: Ifa384d989611398bfad3ca39d3bb595a5962a3ea
Reviewed-on: https://go-review.googlesource.com/c/140198
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agoruntime: extend ReadMemStatsSlow to re-compute HeapReleased
Michael Anthony Knyszek [Mon, 15 Oct 2018 23:00:58 +0000 (23:00 +0000)]
runtime: extend ReadMemStatsSlow to re-compute HeapReleased

This change extends the test function ReadMemStatsSlow to re-compute
the HeapReleased statistic such that it is checked in testing to be
consistent with the bookkeeping done in the runtime.

Change-Id: I49f5c2620f5731edea8e9f768744cf997dcd7c22
Reviewed-on: https://go-review.googlesource.com/c/142397
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
6 years agoruntime: remove npreleased in favor of boolean
Michael Anthony Knyszek [Thu, 4 Oct 2018 15:59:47 +0000 (15:59 +0000)]
runtime: remove npreleased in favor of boolean

This change removes npreleased from mspan since spans may now either be
scavenged or not scavenged; how many of its pages were actually scavenged
doesn't matter. It saves some space in mpsan overhead too, as the boolean
fits into what would otherwise be struct padding.

For #14045.

Change-Id: I63f25a4d98658f5fe21c6a466fc38c59bfc5d0f5
Reviewed-on: https://go-review.googlesource.com/c/139737
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>