Austin Clements [Tue, 24 May 2016 02:14:53 +0000 (22:14 -0400)]
runtime: pass gcWork to scanstack
Currently scanstack obtains its own gcWork from the P for the duration
of the stack scan and then, if called during mark termination,
disposes the gcWork.
However, this means that the number of workbufs allocated will be at
least the number of stacks scanned during mark termination, which may
be very high (especially during a STW GC). This happens because, in
steady state, each scanstack will obtain a fresh workbuf (either from
the empty list or by allocating it), fill it with the scan results,
and then dispose it to the full list. Nothing is consuming from the
full list during this (and hence nothing is recycling them to the
empty list), so the length of the full list by the time mark
termination starts draining it is at least the number of stacks
scanned.
Fix this by pushing the gcWork acquisition up the stack to either the
gcDrain that calls markroot that calls scanstack (which batches across
many stack scans and is the path taken during STW GC) or to newstack
(which is still a single scanstack call, but this is roughly bounded
by the number of Ps).
This fix reduces the workbuf allocation for the test program from
issue #15319 from 213 MB (roughly 2KB * 1e5 goroutines) to 10 MB.
Fixes #15319.
Note that there's potentially a similar issue in write barriers during
mark 2. Fixing that will be more difficult since there's no broader
non-preemptible context, but it should also be less of a problem since
the full list is being drained during mark 2.
Some overall improvements in the go1 benchmarks, plus the usual noise.
No significant change in the garbage benchmark (time/op or GC memory).
Change-Id: I3ce8d4016854d41860c5a9f05a54cda3de49f337
Reviewed-on: https://go-review.googlesource.com/23430 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Marcel van Lohuizen [Wed, 25 May 2016 09:19:17 +0000 (11:19 +0200)]
compress/flate: simplify using subtests and sub-benchmarks
This causes the large files to be loaded only once per benchmark.
This CL also serves as an example use case of sub(tests|-benchmarks).
This CL ensures that names are identical to the original
except for an added slashes. Things could be
simplified further if this restriction were dropped.
David Crawshaw [Tue, 24 May 2016 23:04:51 +0000 (19:04 -0400)]
reflect: remove type info for unexported methods
Also remove some of the now unnecessary corner case handling and
tests I've been adding recently for unexported method data.
For #15673
Change-Id: Ie0c7b03f2370bbe8508cdc5be765028f08000bd7
Reviewed-on: https://go-review.googlesource.com/23410 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Elias Naur [Wed, 25 May 2016 11:24:36 +0000 (13:24 +0200)]
cmd/link: fix ARM gold linker check
CL 23400 introduced a check to make sure the gold linker is used
on ARM host links. The check itself works, but the error checking
logic was reversed; fix it.
I manually verified that the check now correctly rejects host links
on my RPi2 running an ancient rasbian without the gold linker
installed.
Updates #15696
Change-Id: I927832620f0a60e91a71fdedf8cbd2550247b666
Reviewed-on: https://go-review.googlesource.com/23421
Run-TryBot: Elias Naur <elias.naur@gmail.com> Reviewed-by: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Elias Naur [Mon, 16 May 2016 13:51:07 +0000 (15:51 +0200)]
runtime,runtime/cgo: save callee-saved FP register on arm
Other GOARCHs already handle their callee-saved FP registers, but
arm was missing. Without this change, code using Cgo and floating
point code might fail in mysterious and hard to debug ways.
There are no floating point registers when GOARM=5, so skip the
registers when runtime.goarm < 6.
darwin/arm doesn't support GOARM=5, so the check is left out of
rt0_darwin_arm.s.
Fixes #14876
Change-Id: I6bcb90a76df3664d8ba1f33123a74b1eb2c9f8b2
Reviewed-on: https://go-review.googlesource.com/23140
Run-TryBot: Elias Naur <elias.naur@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Minux Ma <minux@golang.org>
Ian Lance Taylor [Tue, 24 May 2016 18:39:48 +0000 (11:39 -0700)]
encoding/csv: clarify that this package supports RFC 4180
The intent of this comment is to reduce the number of issues opened
against the package to add support for new kinds of CSV formats, such as
issues #3150, #8458, #12372, #12755.
Change-Id: I452c0b748e4ca9ebde3e6cea188bf7774372148e
Reviewed-on: https://go-review.googlesource.com/23401 Reviewed-by: Andrew Gerrand <adg@golang.org>
Austin Clements [Fri, 20 May 2016 18:57:55 +0000 (14:57 -0400)]
runtime: update SP when jumping stacks in traceback
When gentraceback starts on a system stack in sigprof, it is
configured to jump to the user stack when it reaches the end of the
system stack. Currently this updates the current frame's FP, but not
its SP. This is okay on non-LR machines (x86) because frame.sp is only
used to find defers, which the bottom-most frame of the user stack
will never have.
However, on LR machines, we use frame.sp to find the saved LR. We then
use to resolve the function of the next frame, which is used to
resolved the size of the next frame. Since we're not updating frame.sp
on a stack jump, we read the saved LR from the system stack instead of
the user stack and wind up resolving the wrong function and hence the
wrong frame size for the next frame.
This has had remarkably few ill effects (though the resulting profiles
must be wrong). We noticed it because of a bad interaction with stack
barriers. Specifically, once we get the next frame size wrong, we also
get the location of its LR wrong. If we happen to get a stack slot
that contains a stale stack barrier LR (for a stack barrier we already
hit) and hasn't been overwritten with something else as we re-grew the
stack, gentraceback will fail with a "found next stack barrier at ..."
error, pointing at the slot that it thinks is an LR, but isn't.
Jeff R. Allen [Tue, 24 May 2016 18:12:22 +0000 (00:12 +0600)]
doc: add notes on good commit messages
Explain Brad's algorithm for generating commit headlines.
Fixes #15700
Change-Id: Ic602f17629b3dd7675e2bb1ed119062c03353ee9
Reviewed-on: https://go-review.googlesource.com/23355 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Jeff R. Allen [Tue, 24 May 2016 17:00:06 +0000 (23:00 +0600)]
math/rand: Doc fix for how many bits Seed uses
Document the fact that the default Source uses only
the bottom 31 bits of the given seed.
Fixes #15788
Change-Id: If20d1ec44a55c793a4a0a388f84b9392c2102bd1
Reviewed-on: https://go-review.googlesource.com/23352 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Jeff R. Allen [Tue, 24 May 2016 17:46:11 +0000 (23:46 +0600)]
cmd/go: document testdata directory in "go help test"
Document the correct use of the testdata directory
where test writers might be expecting to find it.
It seems that alldocs.go was out of date, so it
has picked up some other changes with this commit.
Fixes #14715.
Change-Id: I0a22676bb7a64b2a61b56495f7ea38db889d8b37
Reviewed-on: https://go-review.googlesource.com/23353 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
David Crawshaw [Tue, 24 May 2016 17:40:02 +0000 (13:40 -0400)]
cmd/link: ensure -fuse-ld=gold uses gold
Fixes #15696
Change-Id: I134e918dc56f79a72a04aa54f415371884113d2a
Reviewed-on: https://go-review.googlesource.com/23400 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Russ Cox [Tue, 24 May 2016 13:42:33 +0000 (09:42 -0400)]
io: remove SizedReaderAt
It's not clear we want to enshrine an io interface in which Size cannot
return an error. Because this requires more thought before committing
to the API, remove from Go 1.7.
Russ Cox [Tue, 24 May 2016 06:52:31 +0000 (02:52 -0400)]
cmd/compile, go/types: omit needless word in error message
CL 21462 and CL 21463 made this message say explicitly that the problem
was a struct field in a map, but the word "directly" is unnecessary,
sounds wrong, and makes the error long.
Russ Cox [Mon, 23 May 2016 16:28:56 +0000 (12:28 -0400)]
encoding/json: rename Indent method to SetIndent
CL 21057 added this method during the Go 1.7 cycle
(so it is not yet released and still possible to revise).
This makes it clearer that the method is not doing something
(like func Indent does), but just changing a setting about doing
something later.
Also document that this is in some sense irreversible.
I think that's probably a mistake but the original CL discussion
claimed it as a feature, so I'll leave it alone.
For #6492.
Change-Id: If4415c869a9196501056c143811a308822d5a420
Reviewed-on: https://go-review.googlesource.com/23295 Reviewed-by: Ian Lance Taylor <iant@golang.org> Reviewed-by: Andrew Gerrand <adg@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
Russ Cox [Mon, 23 May 2016 15:41:00 +0000 (11:41 -0400)]
encoding/json: change DisableHTMLEscaping to SetEscapeHTML
DisableHTMLEscaping is now SetEscapeHTML, allowing the escaping
to be toggled, not just disabled. This API is new for Go 1.7,
so there are no compatibility concerns (quite the opposite,
the point is to fix the API before we commit to it in Go 1.7).
Change-Id: I96b9f8f169a9c44995b8a157a626eb62d0b6dea7
Reviewed-on: https://go-review.googlesource.com/23293 Reviewed-by: Ian Lance Taylor <iant@golang.org> Reviewed-by: Andrew Gerrand <adg@golang.org>
Russ Cox [Tue, 24 May 2016 00:40:52 +0000 (20:40 -0400)]
net: revise IP.String result for malformed IP address to add ? back
In earlier versions of Go the result was simply "?".
A change in this cycle made the result echo back the hex bytes
of the address, which is certainly useful, but now the result is
not clearly indicating an error. Put the "?" back, at the beginning
of the hex string, to make the invalidity of the string clearer.
Russ Cox [Mon, 23 May 2016 16:21:57 +0000 (12:21 -0400)]
encoding/json: additional tests and fixes for []typedByte encoding/decoding
CL 19725 changed the encoding of []typedByte to look for
typedByte.MarshalJSON and typedByte.MarshalText.
Previously it was handled like []byte, producing a base64 encoding of the underlying byte data.
CL 19725 forgot to look for (*typedByte).MarshalJSON and (*typedByte).MarshalText,
as the marshaling of other slices would. Add test and fix for those.
This CL also adds tests that the decoder can handle both the old and new encodings.
(This was true even in Go 1.6, which is the only reason we can consider this
not an incompatible change.)
For #13783.
Change-Id: I7cab8b6c0154a7f2d09335b7fa23173bcf856c37
Reviewed-on: https://go-review.googlesource.com/23294 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Andrew Gerrand <adg@golang.org>
Robert Griesemer [Sat, 21 May 2016 00:26:24 +0000 (17:26 -0700)]
go/types: don't drop type in n:1 var decl if one is given
In n:1 variable declarations (multiple lhs variables with single
multi-valued initialization expression) where also a variable
type is provided, make sure that that type is assigned to all
variables on the lhs before the init expression assignment is
checked. Otherwise, (some) variables are assumed to take the type
of the corresponding value of the multi-valued init expression.
Fixes #15755.
Change-Id: I969cb5a95c85e28dbb38abd7fa7df16ff5554c03
Reviewed-on: https://go-review.googlesource.com/23313 Reviewed-by: Alan Donovan <adonovan@google.com>
Kenny Grant [Sat, 21 May 2016 15:57:37 +0000 (16:57 +0100)]
time: run genzabbrs.go with new source data
The source xml data has changed, so running genzabbrs.go
regenerates a new time zone file in zoneinfo_abbrs_windows.go
which adds some zones and adjusts others.
Now set export ZONEINFO=$GOROOT/lib/time/zoneinfo.zip to use zoneinfo.zip in go tip.
Change-Id: I19f72359cc808094e5dcb420e480a00c6b2205d7
Reviewed-on: https://go-review.googlesource.com/23321 Reviewed-by: Alex Brainman <alex.brainman@gmail.com>
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
djherbis [Sat, 21 May 2016 20:25:47 +0000 (13:25 -0700)]
AUTHORS: correcting my last name Herbis -> Herbison
Change-Id: I91608b15e00c8eaf732db3a99a890d4ceeb41955
Reviewed-on: https://go-review.googlesource.com/23317 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Brad Fitzpatrick [Fri, 20 May 2016 23:28:56 +0000 (23:28 +0000)]
A+C: automated update (subrepos)
Add Abe Haskins (individual CLA)
Add Ahmy Yulrizka (individual CLA)
Add Akihiro Suda (individual CLA)
Add Alex Vaghin (corporate CLA for Google Inc.)
Add Arlo Breault (individual CLA)
Add Audrey Lim (individual CLA)
Add Benjamin Wester (corporate CLA for Square, Inc.)
Add Bryan Chan (corporate CLA for IBM)
Add Christy Perez (corporate CLA for IBM)
Add Colin Edwards (individual CLA)
Add David Brophy (individual CLA)
Add David Sansome (individual CLA)
Add Diwaker Gupta (individual CLA)
Add Doug Anderson (corporate CLA for Google Inc.)
Add Dustin Carlino (corporate CLA for Google Inc.)
Add Ernest Chiang (individual CLA)
Add Ethan Burns (corporate CLA for Google Inc.)
Add Gary Elliott (corporate CLA for Google Inc.)
Add Hallgrimur Gunnarsson (corporate CLA for Google Inc.)
Add Hironao OTSUBO (individual CLA)
Add Holden Huang (individual CLA)
Add Idora Shinatose (individual CLA)
Add Irieda Noboru (individual CLA)
Add Jeff Craig (corporate CLA for Google Inc.)
Add Joe Henke (individual CLA)
Add John Schnake (individual CLA)
Add Jonathan Amsterdam (corporate CLA for Google Inc.)
Add Kenji Kaneda (individual CLA)
Add Kenneth Shaw (individual CLA)
Add Mark Severson (individual CLA)
Add Martin Garton (individual CLA)
Add Mathias Leppich (individual CLA)
Add Maxwell Krohn (individual CLA)
Add Niall Sheridan (individual CLA)
Add Nick Patavalis (individual CLA)
Add Nick Petroni (individual CLA)
Add Omar Jarjur (corporate CLA for Google Inc.)
Add Özgür Kesim (individual CLA)
Add Peter Gonda (corporate CLA for Google Inc.)
Add Pierre Durand (individual CLA)
Add Quentin Smith (corporate CLA for Google Inc.)
Add Ricardo Padilha (individual CLA)
Add Riku Voipio (corporate CLA for Linaro Limited)
Add Roland Shoemaker (individual CLA)
Add Sam Hug (individual CLA)
Add Sam Whited (individual CLA)
Add Sami Commerot (corporate CLA for Google Inc.)
Add Scott Mansfield (corporate CLA for Netflix, Inc.)
Add Sean Harger (corporate CLA for Google Inc.)
Add Simon Jefford (individual CLA)
Add Sridhar Venkatakrishnan (individual CLA)
Add Tim Swast (corporate CLA for Google Inc.)
Add Timothy Studd (individual CLA)
Add Tipp Moseley (corporate CLA for Google Inc.)
Add Toby Burress (corporate CLA for Google Inc.)
Add Tzu-Jung Lee (corporate CLA for Currant)
Add Vadim Grek (individual CLA)
Add Xudong Zhang (individual CLA)
Updates #12042
Change-Id: I4119a8829119a2b8a9abbea9f52ceebb04878764
Reviewed-on: https://go-review.googlesource.com/23306 Reviewed-by: Ian Lance Taylor <iant@golang.org> Reviewed-by: Andrew Gerrand <adg@golang.org>
Brad Fitzpatrick [Fri, 20 May 2016 21:52:59 +0000 (21:52 +0000)]
A+C: automated updates
Add Aiden Scandella (individual CLA)
Add Alessandro Arzilli (individual CLA)
Add Augusto Roman (individual CLA)
Add Brady Catherman (individual CLA)
Add Brady Sullivan (individual CLA)
Add Caio Marcelo de Oliveira Filho (corporate CLA for Intel Corporation)
Add Catalin Nicutar (corporate CLA for Google Inc.)
Add Cherry Zhang (corporate CLA for Google Inc.)
Add Chris Zou (corporate CLA for IBM)
Add Christopher Nelson (individual CLA)
Add Conrad Irwin (individual CLA)
Add Cuihtlauac ALVARADO (corporate CLA for Orange)
Add Daniel Speichert (individual CLA)
Add Datong Sun (individual CLA)
Add Denys Honsiorovskyi (individual CLA)
Add Derek Shockey (individual CLA)
Add Dmitriy Dudkin (individual CLA)
Add Dustin Herbis (individual CLA)
Add Frits van Bommel (individual CLA)
Add Harshavardhana (individual CLA)
Add Hitoshi Mitake (individual CLA)
Add James Bardin (individual CLA)
Add James Chacon (corporate CLA for Google Inc.)
Add Jamil Djadala (individual CLA)
Add Jess Frazelle (individual CLA)
Add Joe Sylve (individual CLA)
Add Johan Sageryd (individual CLA)
Add John Jeffery (individual CLA)
Add Julia Hansbrough (corporate CLA for Google Inc.)
Add Jure Ham (corporate CLA for Zemanta d.o.o.)
Add Kamal Aboul-Hosn (corporate CLA for Google Inc.)
Add Kevin Burke (individual CLA)
Add Kevin Kirsche (individual CLA)
Add Kevin Vu (individual CLA)
Add Lee Hinman (individual CLA)
Add Luan Santos (individual CLA)
Add Marc-Antoine Ruel (corporate CLA for Google Inc.)
Add Matt Robenolt (individual CLA)
Add Michael McConville (individual CLA)
Add Michael Munday (corporate CLA for IBM)
Add Michael Pratt (corporate CLA for Google Inc.)
Add Michel Lespinasse (corporate CLA for Google Inc.)
Add Mike Danese (corporate CLA for Google Inc.)
Add Mikhail Gusarov (individual CLA)
Add Monty Taylor (individual CLA)
Add Morten Siebuhr (individual CLA)
Add Muhammed Uluyol (individual CLA)
Add Niels Widger (individual CLA)
Add Niko Dziemba (individual CLA)
Add Olivier Poitrey (individual CLA)
Add Paul Wankadia (corporate CLA for Google Inc.)
Add Philip Hofer (individual CLA)
Add Prashant Varanasi (individual CLA)
Add Rhys Hiltner (corporate CLA for Amazon.com, Inc)
Add Richard Miller (individual CLA)
Add Scott Bell (individual CLA)
Add Shahar Kohanim (individual CLA)
Add Shinji Tanaka (individual CLA)
Add Suharsh Sivakumar (corporate CLA for Google Inc.)
Add Tal Shprecher (individual CLA)
Add Tilman Dilo (individual CLA)
Add Tim Ebringer (individual CLA)
Add Tom Bergan (corporate CLA for Google Inc.)
Add Vishvananda Ishaya (individual CLA)
Add Wedson Almeida Filho (corporate CLA for Google Inc.)
Add Zhongwei Yao (corporate CLA for ARM Ltd.)
Updates #12042
Change-Id: Ia118adc2eb38e5ffc8448de2d9dd3ca792ee7227
Reviewed-on: https://go-review.googlesource.com/23303 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Brad Fitzpatrick [Fri, 20 May 2016 22:43:14 +0000 (22:43 +0000)]
time: document that After uses memory until duration times out
Fixes #15698
Change-Id: I616fc06dcf04092bafdaf56fb1afba2a998a6d83
Reviewed-on: https://go-review.googlesource.com/23304 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Brad Fitzpatrick [Thu, 19 May 2016 17:35:23 +0000 (17:35 +0000)]
net/http: update bundled http2
Updates x/net/http2 to git rev 202ff482 for https://golang.org/cl/23235 (Expect:
100-continue support for HTTP/2)
Fixes a flaky test too, and changes the automatic HTTP/2 behavior to
no longer special-case the DefaultTransport, because
ExpectContinueTimeout is no longer unsupported by the HTTP/2
transport.
Fixes #13851
Fixes #15744
Change-Id: I3522aace14179a1ca070fd7063368a831167a0f7
Reviewed-on: https://go-review.googlesource.com/23254 Reviewed-by: Andrew Gerrand <adg@golang.org>
Jess Frazelle [Thu, 19 May 2016 01:47:24 +0000 (18:47 -0700)]
syscall: add Unshare flags to SysProcAttr on Linux
This patch adds Unshare flags to SysProcAttr for Linux systems.
Fixes #1954
Change-Id: Id819c3f92b1474e5a06dd8d55f89d74a43eb770c
Reviewed-on: https://go-review.googlesource.com/23233
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, 19 May 2016 00:43:15 +0000 (17:43 -0700)]
cmd/compile: don't exit early because of hidden error messages
Non-syntax errors are always counted to determine if to exit
early, but then deduplication eliminates them. This can lead
to situations which report "too many errors" and only one
error is shown.
De-duplicate non-syntax errors early, at least the ones that
appear consecutively, and only count the ones actually being
shown. This doesn't work perfectly as they may not appear in
sequence, but it's cheap and good enough.
Fixes #14136.
Change-Id: I7b11ebb2e1e082f0d604b88e544fe5ba967af1d7
Reviewed-on: https://go-review.googlesource.com/23259 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
In Go versions 1 up to and including Go 1.6,
ResponseRecorder.HeaderMap was both the map that handlers got access
to, and was the map tests checked their results against. That did not
mimic the behavior of the real HTTP server (Issue #8857), so HeaderMap
was changed to be a snapshot at the first write in
https://golang.org/cl/20047. But that broke cases where the Handler
never did a write (#15560), so revert the behavior.
Instead, introduce the ResponseWriter.Result method, returning an
*http.Response. It subsumes ResponseWriter.Trailers which was added
for Go 1.7 in CL 20047. Result().Header now contains the correct
answer, and HeaderMap is unchanged in behavior from previous Go
releases, so we don't break people's tests. People wanting the correct
behavior can use ResponseWriter.Result.
Austin Clements [Tue, 17 May 2016 22:46:03 +0000 (18:46 -0400)]
runtime: fix goroutine priority elevation
Currently it's possible for user code to exploit the high scheduler
priority of the GC worker in conjunction with the runnext optimization
to elevate a user goroutine to high priority so it will always run
even if there are other runnable goroutines.
For example, if a goroutine is in a tight allocation loop, the
following can happen:
1. Goroutine 1 allocates, triggering a GC.
2. G 1 attempts an assist, but fails and blocks.
3. The scheduler runs the GC worker, since it is high priority.
Note that this also starts a new scheduler quantum.
4. The GC worker does enough work to satisfy the assist.
5. The GC worker readies G 1, putting it in runnext.
6. GC finishes and the scheduler runs G 1 from runnext, giving it
the rest of the GC worker's quantum.
7. Go to 1.
Even if there are other goroutines on the run queue, they never get a
chance to run in the above sequence. This requires a confluence of
circumstances that make it unlikely, though not impossible, that it
would happen in "real" code. In the test added by this commit, we
force this confluence by setting GOMAXPROCS to 1 and GOGC to 1 so it's
easy for the test to repeated trigger GC and wake from a blocked
assist.
We fix this by making GC always put user goroutines at the end of the
run queue, instead of in runnext. This makes it so user code can't
piggy-back on the GC's high priority to make a user goroutine act like
it has high priority. The only other situation where GC wakes user
goroutines is waking all blocked assists at the end, but this uses the
global run queue and hence doesn't have this problem.
Fixes #15706.
Change-Id: I1589dee4b7b7d0c9c8575ed3472226084dfce8bc
Reviewed-on: https://go-review.googlesource.com/23172 Reviewed-by: Rick Hudson <rlh@golang.org>
Austin Clements [Tue, 17 May 2016 22:21:54 +0000 (18:21 -0400)]
runtime: add 'next' flag to ready
Currently ready always puts the readied goroutine in runnext. We're
going to have to change this for some uses, so add a flag for whether
or not to use runnext.
For now we always pass true so this is a no-op change.
Ian Lance Taylor [Thu, 19 May 2016 17:07:41 +0000 (10:07 -0700)]
cmd/cgo: mark stub functions as no_sanitize_thread
When the generated stub functions write back the results to the stack,
they can in some cases be writing to the same memory on the g0 stack.
There is no race here (assuming there is no race in the Go code), but
the thread sanitizer does not know that. Turn off the thread sanitizer
for the stub functions to prevent false positive warnings.
Current clang suggests the no_sanitize("thread") attribute, but that
does not work with clang 3.6 or GCC. clang 3.6, GCC, and current clang
all support the no_sanitize_thread attribute, so use that
unconditionally.
The test case and first version of the patch are from Dmitriy Vyukov.
Change-Id: I80ce92824c6c8cf88ea0fe44f21cf50cf62474c9
Reviewed-on: https://go-review.googlesource.com/23252
Run-TryBot: Ian Lance Taylor <iant@golang.org> Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Joel Sing [Wed, 18 May 2016 13:39:23 +0000 (23:39 +1000)]
runtime/cgo: make cgo work with openbsd ABI changes
OpenBSD 6.0 (due out November 2016) will support PT_TLS, which will
allow for the OpenBSD cgo pthread_create() workaround to be removed.
However, in order for Go to continue working on supported OpenBSD
releases (the current release and the previous release - 5.9 and 6.0,
once 6.0 is released), we cannot enable PT_TLS immediately. Instead,
adjust the existing code so that it works with the previous TCB
allocation and the new TIB allocation. This allows the same Go
runtime to work on 5.8, 5.9 and later 6.0.
Once OpenBSD 5.9 is no longer supported (May 2017, when 6.1 is
released), PT_TLS can be enabled and the additional cgo runtime
code removed.
Change-Id: I3eed5ec593d80eea78c6656cb12557004b2c0c9a
Reviewed-on: https://go-review.googlesource.com/23197 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Joel Sing <joel@sing.id.au>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Ian Lance Taylor [Mon, 16 May 2016 22:39:43 +0000 (15:39 -0700)]
runtime: don't do a plain throw when throwsplit == true
The test case in #15639 somehow causes an invalid syscall frame. The
failure is obscured because the throw occurs when throwsplit == true,
which causes a "stack split at bad time" error when trying to print the
throw message.
This CL fixes the "stack split at bad time" by using systemstack. No
test because there shouldn't be any way to trigger this error anyhow.
Update #15639.
Change-Id: I4240f3fd01bdc3c112f3ffd1316b68504222d9e1
Reviewed-on: https://go-review.googlesource.com/23153
Run-TryBot: Ian Lance Taylor <iant@golang.org> Reviewed-by: Austin Clements <austin@google.com>
Brad Fitzpatrick [Wed, 18 May 2016 21:54:12 +0000 (21:54 +0000)]
net: don't return io.EOF from zero byte reads
Updates #15735
Change-Id: I42ab2345443bbaeaf935d683460fc2c941b7679c
Reviewed-on: https://go-review.googlesource.com/23227 Reviewed-by: Ian Lance Taylor <iant@golang.org>
David Benjamin [Mon, 15 Feb 2016 16:41:40 +0000 (11:41 -0500)]
crypto/tls: Never resume sessions across different versions.
Instead, decline the session and do a full handshake. The semantics of
cross-version resume are unclear, and all major client implementations
treat this as a fatal error. (This doesn't come up very much, mostly if
the client does the browser version fallback without sharding the
session cache.)
Brad Fitzpatrick [Wed, 18 May 2016 20:07:16 +0000 (20:07 +0000)]
net/http: further restrict when Transport's automatic HTTP/2 happens
Make the temporary, conservative restrictions from rev 79d9f48c in Go
1.6 permanent, and also don't do automatic TLS if the user configured
a Dial or DialTLS hook. (Go 1.7 has Transport.Dialer instead, for
tweaking dialing parameters)
Ian Lance Taylor [Wed, 18 May 2016 20:43:50 +0000 (13:43 -0700)]
doc/go1.7: add runtime.KeepAlive
Update #13347.
Change-Id: I04bf317ed409478a859355f833d4a5e30db2b9c9
Reviewed-on: https://go-review.googlesource.com/23226 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Keith Randall [Fri, 22 Apr 2016 02:28:28 +0000 (19:28 -0700)]
cmd/compile: keep pointer input arguments live throughout function
Introduce a KeepAlive op which makes sure that its argument is kept
live until the KeepAlive. Use KeepAlive to mark pointer input
arguments as live after each function call and at each return.
We do this change only for pointer arguments. Those are the
critical ones to handle because they might have finalizers.
Doing compound arguments (slices, structs, ...) is more complicated
because we would need to track field liveness individually (we do
that for auto variables now, but inputs requires extra trickery).
Turn off the automatic marking of args as live. That way, when args
are explicitly nulled, plive will know that the original argument is
dead.
The KeepAlive op will be the eventual implementation of
runtime.KeepAlive.
Brad Fitzpatrick [Sun, 1 May 2016 02:57:28 +0000 (19:57 -0700)]
net/http: fix spurious logging in Transport when server closes idle conn
In https://golang.org/3210, Transport errors occurring before
receiving response headers were wrapped in another error type to
indicate to the retry logic elsewhere that the request might be
re-tryable. But a check for err == io.EOF was missed, which then became
false once io.EOF was wrapped in the beforeRespHeaderError type.
The beforeRespHeaderError was too fragile. Remove it. I tried to fix
it in an earlier version of this CL and just broke different things
instead.
Also remove the "markBroken" method. It's redundant and confusing.
Also, rename the checkTransportResend method to shouldRetryRequest and
make it return a bool instead of an error. This also helps readability.
Now the code recognizes the two main reasons we'd want to retry a
request: because we never wrote the request in the first place (so:
count the number of bytes we've written), or because the server hung
up on us before we received response headers for an idempotent request.
As an added bonus, this could make POST requests safely re-tryable
since we know we haven't written anything yet. But it's too late in Go
1.7 to enable that, so we'll do that later (filed #15723).
This also adds a new internal (package http) test, since testing this
blackbox at higher levels in transport_test wasn't possible.
Lee Hinman [Tue, 8 Mar 2016 04:31:31 +0000 (22:31 -0600)]
crypto/x509: add Admin & User Keychains to FetchPEMRoots on Darwin
in root_cgo_darwin.go only certificates from the System Domain
were being used in FetchPEMRoots. This patch adds support for
getting certificates from all three domains (System, Admin,
User). Also it will only read trusted certificates from those
Keychains. Because it is possible to trust a non Root certificate,
this patch also adds a checks to see if the Subject and Issuer
name are the same.
Adam Langley [Thu, 14 Apr 2016 20:52:56 +0000 (13:52 -0700)]
crypto/ecdsa: reject negative inputs.
The fact that crypto/ecdsa.Verify didn't reject negative inputs was a
mistake on my part: I had unsigned numbers on the brain. However, it
doesn't generally cause problems. (ModInverse results in zero, which
results in x being zero, which is rejected.)
The amd64 P-256 code will crash when given a large, negative input.
This fixes both crypto/ecdsa to reject these values and also the P-256
code to ignore the sign of inputs.
Monty Taylor [Tue, 17 May 2016 13:24:18 +0000 (08:24 -0500)]
vcs: Add support for git.openstack.org
Go is being proposed as an officially supported language for elements of
OpenStack:
https://review.openstack.org/#/c/312267/
As such, repos that exist in OpenStack's git infrastructure
are likely to become places from which people might want to go get
things. Allow optional .git suffixes to allow writing code that depends
on git.openstack.org repos that will work with older go versions while
we wait for this support to roll out.
David Chase [Mon, 9 May 2016 18:59:25 +0000 (14:59 -0400)]
cmd/compile: reduce element size of arrays in sparse{map,set}
sparseSet and sparseMap only need 32 bit integers in their
arrays, since a sparseEntry key is also limited to 32 bits.
This appears to reduce the space allocated for at least
one pathological compilation by 1%, perhaps more.
Not necessarily for 1.7, but it saves a little and is very
low-risk.
Change-Id: Icf1185859e9f5fe1261a206b441e02c34f7d02fd
Reviewed-on: https://go-review.googlesource.com/22972
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Keith Randall <khr@golang.org> Reviewed-by: Russ Cox <rsc@golang.org>
Cuihtlauac ALVARADO [Tue, 17 May 2016 07:27:00 +0000 (09:27 +0200)]
runtime: in tests, make sure gdb does not start with a shell
On some systems, gdb is set to: "startup-with-shell on". This
breaks runtime_test. This just make sure gdb does not start by
spawning a shell.
Fixes #15354
Change-Id: Ia040931c61dea22f4fdd79665ab9f84835ecaa70
Reviewed-on: https://go-review.googlesource.com/23142 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
which can leave g's temporary value on the stack as a uintptr, rather
than an unsafe.Pointer. Instead, recognize uintptr-to-unsafe.Pointer
conversions when reordering function calls to instead produce:
Mikio Hara [Wed, 18 May 2016 01:54:20 +0000 (10:54 +0900)]
net: deflake TestLookupPort on Android
Looks like some version of Android still fails with "servname not
supported for ai_socktype". It probably doesn't support
ai_socktype=SOCK_STREAM.
Updates #14576.
Change-Id: I77ecff147d5b759e3281b3798c60f150a4aab811
Reviewed-on: https://go-review.googlesource.com/23194 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Brad Fitzpatrick [Wed, 18 May 2016 01:00:32 +0000 (01:00 +0000)]
regexp/syntax: clarify that \Z means Perl's \Z
Fixes #14793
Change-Id: I408056d096cd6a999fa5e349704b5ea8e26d2e4e
Reviewed-on: https://go-review.googlesource.com/23201 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Ian Lance Taylor [Wed, 18 May 2016 01:16:47 +0000 (18:16 -0700)]
runtime: deflake TestSignalExitStatus
The signal might get delivered to a different thread, and that thread
might not run again before the currently running thread returns and
exits. Sleep to give the other thread time to pick up the signal and
crash.
Not tested for all cases, but, optimistically:
Fixes #14063.
Change-Id: Iff58669ac6185ad91cce85e0e86f17497a3659fd
Reviewed-on: https://go-review.googlesource.com/23203
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Mikio Hara <mikioh.mikioh@gmail.com>
Brad Fitzpatrick [Wed, 18 May 2016 00:35:43 +0000 (00:35 +0000)]
net/http: add test confirming a connection reuse case
Verify that for a server doing chunked encoding, with the final data
and EOF arriving together, the client will reuse the connection even
if it closes the body without seeing an EOF. The server sends at least
one non-zero chunk and one zero chunk. This verifies that the client's
bufio reading reads ahead and notes the EOF, so even if the JSON
decoder doesn't read the EOF itself, as long as somebody sees it, a
close won't forcible tear down the connection. This was true at least
of https://golang.org/cl/21291
No code change. Test already passed (even with lots of runs, including
in race mode with randomized goroutine scheduling).