]> Cypherpunks repositories - gostls13.git/log
gostls13.git
10 years agocmd/gc: add 2-, 3-, 4-word write barrier specializations
Russ Cox [Tue, 14 Oct 2014 20:31:09 +0000 (16:31 -0400)]
cmd/gc: add 2-, 3-, 4-word write barrier specializations

Assignments of 2-, 3-, and 4-word values were handled
by individual MOV instructions (and for scalars still are).
But if there are pointers involved, those assignments now
go through the write barrier routine. Before this CL, they
went to writebarrierfat, which calls memmove.
Memmove is too much overhead for these small
amounts of data.

Instead, call writebarrierfat{2,3,4}, which are specialized
for the specific amount of data being copied.
Today the write barrier does not care which words are
pointers, so size alone is enough to distinguish the cases.
If we keep these distinctions in Go 1.5 we will need to
expand them for all the pointer-vs-scalar possibilities,
so the current 3 functions will become 3+7+15 = 25,
still not a large burden (we deleted more morestack
functions than that when we dropped segmented stacks).

BenchmarkBinaryTree17           3250972583  3123910344  -3.91%
BenchmarkFannkuch11             3067605223  2964737839  -3.35%
BenchmarkFmtFprintfEmpty        101         96.0        -4.95%
BenchmarkFmtFprintfString       267         235         -11.99%
BenchmarkFmtFprintfInt          261         253         -3.07%
BenchmarkFmtFprintfIntInt       444         402         -9.46%
BenchmarkFmtFprintfPrefixedInt  374         346         -7.49%
BenchmarkFmtFprintfFloat        472         449         -4.87%
BenchmarkFmtManyArgs            1537        1476        -3.97%
BenchmarkGobDecode              13986528    12432985    -11.11%
BenchmarkGobEncode              13120323    12537420    -4.44%
BenchmarkGzip                   451925758   437500578   -3.19%
BenchmarkGunzip                 113267612   110053644   -2.84%
BenchmarkHTTPClientServer       103151      77100       -25.26%
BenchmarkJSONEncode             25002733    23435278    -6.27%
BenchmarkJSONDecode             94213717    82568789    -12.36%
BenchmarkMandelbrot200          4804246     4713070     -1.90%
BenchmarkGoParse                4646114     4379456     -5.74%
BenchmarkRegexpMatchEasy0_32    163         158         -3.07%
BenchmarkRegexpMatchEasy0_1K    433         391         -9.70%
BenchmarkRegexpMatchEasy1_32    154         138         -10.39%
BenchmarkRegexpMatchEasy1_1K    1481        1132        -23.57%
BenchmarkRegexpMatchMedium_32   282         270         -4.26%
BenchmarkRegexpMatchMedium_1K   92421       86149       -6.79%
BenchmarkRegexpMatchHard_32     5209        4718        -9.43%
BenchmarkRegexpMatchHard_1K     158141      147921      -6.46%
BenchmarkRevcomp                699818791   642222464   -8.23%
BenchmarkTemplate               132402383   108269713   -18.23%
BenchmarkTimeParse              509         478         -6.09%
BenchmarkTimeFormat             462         456         -1.30%

LGTM=r
R=r
CC=golang-codereviews
https://golang.org/cl/156200043

10 years agocmd/gc: fix 'make' in cmd/gc directory
Russ Cox [Tue, 14 Oct 2014 18:58:25 +0000 (14:58 -0400)]
cmd/gc: fix 'make' in cmd/gc directory

Right now, go tool 6g -A fails complaining about 'any' type.

TBR=r
CC=golang-codereviews
https://golang.org/cl/156200044

10 years agoruntime: a few optimizations of scanblock.
Keith Randall [Tue, 14 Oct 2014 16:22:47 +0000 (09:22 -0700)]
runtime: a few optimizations of scanblock.

Lowers gc pause time by 5-10% on test/bench/garbage

LGTM=rsc, dvyukov
R=rsc, dvyukov
CC=golang-codereviews
https://golang.org/cl/157810043

10 years agocrypto/x509: continue to recognise MaxPathLen of zero as "no value".
Adam Langley [Tue, 14 Oct 2014 01:35:53 +0000 (18:35 -0700)]
crypto/x509: continue to recognise MaxPathLen of zero as "no value".

In [1] the behaviour of encoding/asn1 with respect to marshaling
optional integers was changed. Previously, a zero valued integer would
be omitted when marshaling. After the change, if a default value was
set then the integer would only be omitted if it was the default value.

This changed the behaviour of crypto/x509 because
Certificate.MaxPathLen has a default value of -1 and thus zero valued
MaxPathLens would no longer be omitted when marshaling. This is
arguably a bug-fix -- a value of zero for MaxPathLen is valid and
meaningful and now could be expressed. However it broke users
(including Docker) who were not setting MaxPathLen at all.

This change again causes a zero-valued MaxPathLen to be omitted and
introduces a ZeroMathPathLen member that indicates that, yes, one
really does want a zero. This is ugly, but we value not breaking users.

[1] https://code.google.com/p/go/source/detail?r=4218b3544610e8d9771b89126553177e32687adf

LGTM=rsc
R=rsc
CC=golang-codereviews, golang-dev
https://golang.org/cl/153420045

10 years agomath/big: Fixes issue 8920
Casey Marshall [Mon, 13 Oct 2014 19:41:14 +0000 (12:41 -0700)]
math/big: Fixes issue 8920

(*Rat).SetString checks for denominator.

LGTM=gri
R=golang-codereviews, gri
CC=golang-codereviews
https://golang.org/cl/159760043

10 years agotest: add test for gccgo-specific issue 8079
Chris Manghane [Mon, 13 Oct 2014 19:33:45 +0000 (12:33 -0700)]
test: add test for gccgo-specific issue 8079

LGTM=gri
R=gri
CC=golang-codereviews, iant
https://golang.org/cl/159850044

10 years agoos: handle 'no parent' error as IsNotExist on Plan 9
David du Colombier [Mon, 13 Oct 2014 18:39:46 +0000 (20:39 +0200)]
os: handle 'no parent' error as IsNotExist on Plan 9

This error is returned by lib9p when removing a file
without parent. It should fix TestRemoveAllRace
when running on ramfs.

LGTM=bradfitz, aram
R=rsc, bradfitz, aram
CC=golang-codereviews, mischief
https://golang.org/cl/153410044

10 years agonet/rpc: fix mutex comment
Rob Pike [Mon, 13 Oct 2014 17:27:51 +0000 (10:27 -0700)]
net/rpc: fix mutex comment
Fixes #8086.

LGTM=bradfitz
R=golang-codereviews, bradfitz
CC=golang-codereviews
https://golang.org/cl/153420044

10 years agoreflect: generated unrolled GC bitmask directly
Ian Lance Taylor [Mon, 13 Oct 2014 17:01:34 +0000 (10:01 -0700)]
reflect: generated unrolled GC bitmask directly

The code for a generated type is already generating an
unrolled GC bitmask.  Rather than unrolling the the source
type bitmasks and copying them, just generate the required
bitmask directly.  Don't mark it as an unrolled GC program,
since there is no need to do so.

Fixes #8917.

LGTM=rsc
R=dvyukov, rsc
CC=golang-codereviews
https://golang.org/cl/156930044

10 years agogo1.4: add note about change to bufio.SplitFunc at EOF
Ian Lance Taylor [Sun, 12 Oct 2014 18:41:55 +0000 (11:41 -0700)]
go1.4: add note about change to bufio.SplitFunc at EOF

CC=golang-codereviews
https://golang.org/cl/159800044

10 years agocmd/ld: do not assume that only pe section names start with '.'
Alex Brainman [Sat, 11 Oct 2014 11:01:04 +0000 (22:01 +1100)]
cmd/ld: do not assume that only pe section names start with '.'

Our current pe object reader assumes that every symbol starting with
'.' is section. It appeared to be true, until now gcc 4.9.1 generates
some symbols with '.' at the front. Change that logic to check other
symbol fields in addition to checking for '.'. I am not an expert
here, but it seems reasonable to me.

Added test, but it is only good, if tested with gcc 4.9.1. Otherwise
the test PASSes regardless.

Fixes #8811.
Fixes #8856.

LGTM=jfrederich, iant, stephen.gutekanst
R=golang-codereviews, jfrederich, stephen.gutekanst, iant
CC=alex.brainman, golang-codereviews
https://golang.org/cl/152410043

10 years agocmd/ld: correct pe section names if longer then 8 chars
Alex Brainman [Sat, 11 Oct 2014 10:34:10 +0000 (21:34 +1100)]
cmd/ld: correct pe section names if longer then 8 chars

gcc 4.9.1 generates pe sections with names longer then 8 charters.

From IMAGE_SECTION_HEADER definition:

Name
An 8-byte, null-padded UTF-8 string. There is no terminating null character
if the string is exactly eight characters long. For longer names, this
member contains a forward slash (/) followed by an ASCII representation
of a decimal number that is an offset into the string table.

Our current pe object file reader does not read string table when section
names starts with /. Do that, so (issue 8811 example)

c:\go\path\src\isssue8811>go build
# isssue8811
isssue8811/glfw(.text): isssue8811/glfw(/76): not defined
isssue8811/glfw(.text): undefined: isssue8811/glfw(/76)

becomes

c:\go\path\src\isssue8811>go build
# isssue8811
isssue8811/glfw(.text): isssue8811/glfw(.rdata$.refptr._glfwInitialized): not defined
isssue8811/glfw(.text): undefined: isssue8811/glfw(.rdata$.refptr._glfwInitialized)

Small progress to

Update #8811

LGTM=iant, jfrederich
R=golang-codereviews, iant, jfrederich
CC=golang-codereviews
https://golang.org/cl/154210044

10 years agocmd/ld: fix off-by-one error when emitting symbol names
Shenghou Ma [Sat, 11 Oct 2014 00:30:24 +0000 (20:30 -0400)]
cmd/ld: fix off-by-one error when emitting symbol names

I diffed the output of `nm -n gofmt' before and after this change,
and verified that all changes are correct and all corrupted symbol
names are fixed.

Fixes #8906.

LGTM=iant, cookieo9
R=golang-codereviews, iant, cookieo9
CC=golang-codereviews
https://golang.org/cl/159750043

10 years agonet: link skipped TestReadFromUDP to the issue
Alex Brainman [Fri, 10 Oct 2014 02:12:32 +0000 (13:12 +1100)]
net: link skipped TestReadFromUDP to the issue

LGTM=minux
R=bradfitz, minux
CC=golang-codereviews
https://golang.org/cl/154220043

10 years agoencoding/asn1: fix explicitly tagged Times.
Adam Langley [Fri, 10 Oct 2014 00:37:40 +0000 (17:37 -0700)]
encoding/asn1: fix explicitly tagged Times.

https://golang.org/cl/153770043/ tried to fix the case where a
implicitly tagged Time, that happened to have the same tag as
GENERALIZEDTIME, shouldn't be parsed as a GENERALIZEDTIME.

It did so, mistakenly, by testing whether params.tag != nil. But
explicitly tagged values also have a non-nil tag and there the inner
tag actually does encode the type of the value.

This change instead tests whether the tag class is UNIVERSAL before
assuming that the tag contains type information.

LGTM=iant
R=iant
CC=golang-codereviews
https://golang.org/cl/152380044

10 years agonet: skip new TestReadFromUDP on nacl and plan9 (fixes build)
Alex Brainman [Thu, 9 Oct 2014 22:46:41 +0000 (09:46 +1100)]
net: skip new TestReadFromUDP on nacl and plan9 (fixes build)

TBR=0intro
R=golang-codereviews
CC=golang-codereviews
https://golang.org/cl/157820043

10 years agonet: disable SIO_UDP_CONNRESET behavior on windows.
Ron Hashimoto [Thu, 9 Oct 2014 22:21:32 +0000 (09:21 +1100)]
net: disable SIO_UDP_CONNRESET behavior on windows.

Fixes #5834.

LGTM=alex.brainman
R=golang-codereviews, bradfitz, alex.brainman, mikioh.mikioh, in60jp, iant
CC=golang-codereviews
https://golang.org/cl/149510043

10 years agodebug/elf: add comments explaining applyRelocations for amd64/arm64
Ian Lance Taylor [Thu, 9 Oct 2014 21:38:45 +0000 (14:38 -0700)]
debug/elf: add comments explaining applyRelocations for amd64/arm64

LGTM=bradfitz
R=golang-codereviews, bradfitz
CC=golang-codereviews
https://golang.org/cl/155190043

10 years agoruntime: add comment to mgc0.h
Dmitriy Vyukov [Thu, 9 Oct 2014 13:05:38 +0000 (17:05 +0400)]
runtime: add comment to mgc0.h
Missed that comment in CL 153990043.

LGTM=khr
R=khr
CC=golang-codereviews
https://golang.org/cl/156010043

10 years agoC: add Hana Kim (Google CLA)
Brad Fitzpatrick [Thu, 9 Oct 2014 13:00:23 +0000 (15:00 +0200)]
C: add Hana Kim (Google CLA)

LGTM=crawshaw
R=golang-codereviews, crawshaw
CC=golang-codereviews, iant
https://golang.org/cl/150690043

10 years agonet/rpc: fix build
David du Colombier [Thu, 9 Oct 2014 09:21:21 +0000 (11:21 +0200)]
net/rpc: fix build

LGTM=adg
R=golang-codereviews, adg
CC=golang-codereviews
https://golang.org/cl/151620043

10 years agonet/rpc: skip TestGobError on Plan 9
David du Colombier [Thu, 9 Oct 2014 09:12:03 +0000 (11:12 +0200)]
net/rpc: skip TestGobError on Plan 9

LGTM=bradfitz
R=rsc, bradfitz
CC=aram, golang-codereviews
https://golang.org/cl/154140043

10 years agoundo CL 145150043 / 8b3d26697b8d
Alex Brainman [Thu, 9 Oct 2014 06:24:34 +0000 (17:24 +1100)]
undo CL 145150043 / 8b3d26697b8d

That was complete failure - builders are broken,
but original cl worked fine on my system.
I will need access to builders
to test this change properly.

««« original CL description
runtime: handle all windows exception

Fixes #8006.

LGTM=rsc
R=golang-codereviews, rsc
CC=golang-codereviews
https://golang.org/cl/145150043
»»»

TBR=rsc
R=golang-codereviews
CC=golang-codereviews
https://golang.org/cl/154180043

10 years agoruntime: handle all windows exception
Alex Brainman [Thu, 9 Oct 2014 05:52:28 +0000 (16:52 +1100)]
runtime: handle all windows exception

Fixes #8006.

LGTM=rsc
R=golang-codereviews, rsc
CC=golang-codereviews
https://golang.org/cl/145150043

10 years agomisc/cgo/test: add Test8517
Alex Brainman [Thu, 9 Oct 2014 05:10:51 +0000 (16:10 +1100)]
misc/cgo/test: add Test8517

CL 145890044 did not provide test. Rectify that.

Update #8517

LGTM=rsc
R=golang-codereviews, rsc
CC=golang-codereviews
https://golang.org/cl/148790043

10 years agoruntime: zero a few more dead pointers.
Keith Randall [Thu, 9 Oct 2014 00:22:34 +0000 (17:22 -0700)]
runtime: zero a few more dead pointers.

In channels, zeroing of gp.waiting is missed on a closed channel panic.
m.morebuf.g is not zeroed.

I don't expect the latter causes any problems, but just in case.

LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/151610043

10 years agocmd/ld: don't add line number info for the final address of an FDE
Ian Lance Taylor [Wed, 8 Oct 2014 23:17:34 +0000 (16:17 -0700)]
cmd/ld: don't add line number info for the final address of an FDE

This makes dwardump --verify happy.

Update #8846

LGTM=r
R=golang-codereviews, r
CC=golang-codereviews
https://golang.org/cl/150370043

10 years agoreflect: add direct call tests to TestMakeFuncVariadic
Michael Hudson-Doyle [Wed, 8 Oct 2014 22:58:56 +0000 (15:58 -0700)]
reflect: add direct call tests to TestMakeFuncVariadic

TestMakeFuncVariadic only called the variadic function via Call and
CallSlice, not via a direct function call.

I thought these tests would fail under gccgo tip, but they don't.  Still seems worth having though.

LGTM=iant
R=golang-codereviews, gobot, iant
CC=golang-codereviews
https://golang.org/cl/152060043

10 years agoruntime: delay freeing of shrunk stacks until gc is done.
Keith Randall [Wed, 8 Oct 2014 22:57:20 +0000 (15:57 -0700)]
runtime: delay freeing of shrunk stacks until gc is done.

This change prevents confusion in the garbage collector.
The collector wants to make sure that every pointer it finds
isn't junk.  Its criteria for junk is (among others) points
to a "free" span.

Because the stack shrinker modifies pointers in the heap,
there is a race condition between the GC scanner and the
shrinker.  The GC scanner can see old pointers (pointers to
freed stacks).  In particular this happens with SudoG.elem
pointers.

Normally this is not a problem, as pointers into stack spans
are ok.  But if the freed stack is the last one in its span,
the span is marked as "free" instead of "contains stacks".

This change makes sure that even if the GC scanner sees
an old pointer, the span into which it points is still
marked as "contains stacks", and thus the GC doesn't
complain about it.

This change will make the GC pause a tiny bit slower, as
the stack freeing now happens in serial with the mark pause.
We could delay the freeing until the mutators start back up,
but this is the simplest change for now.

TBR=dvyukov
CC=golang-codereviews
https://golang.org/cl/158750043

10 years agoreflect: add tests for variadic method calls
Ian Lance Taylor [Wed, 8 Oct 2014 22:48:46 +0000 (15:48 -0700)]
reflect: add tests for variadic method calls

These tests fail when using gccgo.  In gccgo using Interface
on the value of a method function is implemented using a
variant of MakeFunc.  That approach did not correctly handle
variadic functions.

LGTM=r
R=golang-codereviews, r
CC=golang-codereviews
https://golang.org/cl/151280043

10 years agoA+C: Ron Hashimoto (individual CLA)
Ian Lance Taylor [Wed, 8 Oct 2014 14:10:04 +0000 (07:10 -0700)]
A+C: Ron Hashimoto (individual CLA)

Generated by a+c.

R=gobot
CC=golang-codereviews
https://golang.org/cl/153240043

10 years agoruntime: faster GC scan
Dmitriy Vyukov [Wed, 8 Oct 2014 09:51:12 +0000 (13:51 +0400)]
runtime: faster GC scan
The change contains 3 spot optimizations to scan loop:
1. Don't use byte vars, use uintptr's instead.
This seems to alleviate some codegen issue,
and alone accounts to a half of speedup.
2. Remove bitmap cache. Currently we cache only 1 byte,
so caching is not particularly effective anyway.
Removal of the cache simplifies code and positively affects regalloc.
3. Replace BitsMultiword switch with if and
do debug checks only in Debug mode.
I've benchmarked changes separately and ensured that
each of them provides speedup on top of the previous one.
This change as a whole fixes the unintentional regressions
of scan loop that were introduced during development cycle.
Fixes #8625.
Fixes #8565.

On go.benchmarks/garbage benchmark:
GOMAXPROCS=1
time: -3.13%
cputime: -3.22%
gc-pause-one: -15.71%
gc-pause-total: -15.71%

GOMAXPROCS=32
time: -1.96%
cputime: -4.43%
gc-pause-one: -6.22%
gc-pause-total: -6.22%

LGTM=khr, rsc
R=golang-codereviews, khr
CC=golang-codereviews, rlh, rsc
https://golang.org/cl/153990043

10 years agoruntime: clear Defer.fn before removing from the G.defer list
Russ Cox [Wed, 8 Oct 2014 04:03:50 +0000 (00:03 -0400)]
runtime: clear Defer.fn before removing from the G.defer list

Should fix the remaining 'invalid heap pointer' build failures.

TBR=khr
CC=golang-codereviews
https://golang.org/cl/152360043

10 years agoruntime: fix windows/amd64 build
Russ Cox [Wed, 8 Oct 2014 03:39:00 +0000 (23:39 -0400)]
runtime: fix windows/amd64 build

Out of stack space due to new 2-word call in freedefer.
Go back to smaller function calls.

TBR=brainman
CC=golang-codereviews
https://golang.org/cl/152340043

10 years agoruntime: change Windows M.thread from void* to uintptr
Russ Cox [Wed, 8 Oct 2014 03:27:25 +0000 (23:27 -0400)]
runtime: change Windows M.thread from void* to uintptr

It appears to be an opaque bit pattern more than a pointer.
The Go garbage collector has discovered that for m0
it is set to 0x4c.

Should fix Windows build.

TBR=brainman
CC=golang-codereviews
https://golang.org/cl/149640043

10 years agoruntime: clear Defer.panic before removing from G.defer list
Russ Cox [Wed, 8 Oct 2014 03:17:31 +0000 (23:17 -0400)]
runtime: clear Defer.panic before removing from G.defer list

Another dangling stack pointer in a cached structure.
Same as SudoG.elem and SudoG.selectdone.

Definitely a fix, and the new test in freedefer makes the
crash reproducible, but probably not a complete fix.
I have seen one dangling pointer in a Defer.panic even
after this fix; I cannot see where it could be coming from.

I think this will fix the solaris build.
I do not think this will fix the occasional failure on the darwin build.

TBR=khr
R=khr
CC=golang-codereviews
https://golang.org/cl/155080043

10 years agonet/rpc: listen on localhost, let kernel pick port
Russ Cox [Wed, 8 Oct 2014 03:08:07 +0000 (23:08 -0400)]
net/rpc: listen on localhost, let kernel pick port

This avoids a pop-up box on OS X and it avoids
a test failure if something is using 5555.
I apologize for not noticing this during the review.

TBR=r
CC=golang-codereviews
https://golang.org/cl/152320044

10 years agodoc: use "keyed" instead of "tagged" in Go 1 compatibility doc
Andrew Gerrand [Wed, 8 Oct 2014 02:23:05 +0000 (13:23 +1100)]
doc: use "keyed" instead of "tagged" in Go 1 compatibility doc

LGTM=bradfitz, r
R=r, bradfitz
CC=golang-codereviews
https://golang.org/cl/156730043

10 years agocmd/go: add ImportComment to Package struct
Andrew Gerrand [Wed, 8 Oct 2014 02:22:31 +0000 (13:22 +1100)]
cmd/go: add ImportComment to Package struct

It seems reasonable that people might want to look up the
ImportComment with "go list".

LGTM=r
R=golang-codereviews, r
CC=golang-codereviews
https://golang.org/cl/143600043

10 years agoruntime: zero pointer-looking scalararg values
Keith Randall [Tue, 7 Oct 2014 22:21:00 +0000 (15:21 -0700)]
runtime: zero pointer-looking scalararg values

I have a CL which at every gc looks through data and bss
sections for nonpointer data (according to gc maps) that
looks like a pointer.  These are potential missing roots.

The only thing it finds are begnign, storing stack pointers
into m0.scalararg[1] and never cleaning them up.  Let's
clean them up now so the test CL passes all.bash cleanly.

The test CL can't be checked in because we might store
pointer-looking things in nonpointer data by accident.

LGTM=iant
R=golang-codereviews, iant, khr
CC=golang-codereviews
https://golang.org/cl/153210043

10 years agoruntime: update heap dump format for 1.4
Keith Randall [Tue, 7 Oct 2014 20:36:16 +0000 (13:36 -0700)]
runtime: update heap dump format for 1.4

We no longer have full type information in the heap, so
we can't dump that any more.  Instead we dump ptr/noptr
maps so at least we can compute graph connectivity.

In addition, we still dump Iface/Eface types so together
with dwarf type info we might be able to reconstruct
types of most things in the heap.

LGTM=dvyukov
R=golang-codereviews, dvyukov, rsc, khr
CC=golang-codereviews
https://golang.org/cl/155940043

10 years agoruntime: fix _cgo_allocate(0)
Russ Cox [Tue, 7 Oct 2014 20:27:40 +0000 (16:27 -0400)]
runtime: fix _cgo_allocate(0)

Fixes a SWIG bug reported off-list.

LGTM=iant
R=iant
CC=golang-codereviews
https://golang.org/cl/155990043

10 years agomath/big: fix doc comments
Robert Griesemer [Tue, 7 Oct 2014 17:56:58 +0000 (10:56 -0700)]
math/big: fix doc comments

Fixes #8904.

TBR=iant
R=iant
CC=golang-codereviews
https://golang.org/cl/148650043

10 years agonet/rpc: add test for issue 7689 (gob error should cause EOF)
Rob Pike [Tue, 7 Oct 2014 17:52:16 +0000 (10:52 -0700)]
net/rpc: add test for issue 7689 (gob error should cause EOF)
Helpfully supplied by tommi.virtanen in issue 8173.

LGTM=rsc
R=golang-codereviews, rsc
CC=golang-codereviews
https://golang.org/cl/151370043

10 years agocmd/5c, cmd/6c, cmd/8c: make failure to optimize fatal
Russ Cox [Tue, 7 Oct 2014 16:07:40 +0000 (12:07 -0400)]
cmd/5c, cmd/6c, cmd/8c: make failure to optimize fatal

LGTM=bradfitz, dave, r
R=r, bradfitz, dave
CC=golang-codereviews
https://golang.org/cl/152250044

10 years agocmd/gc: fix print format
Russ Cox [Tue, 7 Oct 2014 16:03:48 +0000 (12:03 -0400)]
cmd/gc: fix print format

Fixes 386 build.

TBR=r
CC=golang-codereviews
https://golang.org/cl/149620043

10 years agoruntime: crash if we see an invalid pointer into GC arena
Russ Cox [Tue, 7 Oct 2014 15:07:18 +0000 (11:07 -0400)]
runtime: crash if we see an invalid pointer into GC arena

This will help find bugs during the release freeze.
It's not clear it should be kept for the release itself.
That's issue 8861.

The most likely thing that would trigger this is stale
pointers that previously were ignored or caused memory
leaks. These were allowed due to the use of conservative
collection. Now that everything is precise, we should not
see them anymore.

The small number check reinforces what the stack copier
is already doing, catching the storage of integers in pointers.
It caught issue 8864.

The check is disabled if _cgo_allocate is linked into the binary,
which is to say if the binary is using SWIG to allocate untyped
Go memory. In that case, there are invalid pointers and there's
nothing we can do about it.

LGTM=rlh
R=golang-codereviews, dvyukov, rlh
CC=golang-codereviews, iant, khr, r
https://golang.org/cl/148470043

10 years agoencoding/json: fix handling of null with ,string fields
Russ Cox [Tue, 7 Oct 2014 15:07:04 +0000 (11:07 -0400)]
encoding/json: fix handling of null with ,string fields

Fixes #8587.

LGTM=bradfitz
R=golang-codereviews, bradfitz
CC=golang-codereviews, iant, r
https://golang.org/cl/152270044

10 years agoruntime: remove type-punning for Type.gc[0], gc[1]
Russ Cox [Tue, 7 Oct 2014 15:06:51 +0000 (11:06 -0400)]
runtime: remove type-punning for Type.gc[0], gc[1]

Depending on flags&KindGCProg,
gc[0] and gc[1] are either pointers or inlined bitmap bits.
That's not compatible with a precise garbage collector:
it needs to be always pointers or never pointers.

Change the inlined bitmap case to store a pointer to an
out-of-line bitmap in gc[0]. The out-of-line bitmaps are
dedup'ed, so that for example all pointer types share the
same out-of-line bitmap.

Fixes #8864.

LGTM=r
R=golang-codereviews, dvyukov, r
CC=golang-codereviews, iant, khr, rlh
https://golang.org/cl/155820043

10 years agonet/http: fix authentication info leakage in Referer header (potential security risk)
Jens Frederich [Tue, 7 Oct 2014 14:13:42 +0000 (07:13 -0700)]
net/http: fix authentication info leakage in Referer header (potential security risk)

http.Client calls URL.String() to fill in the Referer header, which may
contain authentication info. This patch removes authentication info from
the Referer header without introducing any API changes.

A new test for net/http is also provided.

This is the polished version of Alberto García Hierro's
https://golang.org/cl/9766046/

It should handle https Referer right.

Fixes #8417

LGTM=bradfitz
R=golang-codereviews, gobot, bradfitz, mikioh.mikioh
CC=golang-codereviews
https://golang.org/cl/151430043

10 years agodoc/go1.4.txt: useless change to tickle builders
Brad Fitzpatrick [Tue, 7 Oct 2014 05:22:47 +0000 (22:22 -0700)]
doc/go1.4.txt: useless change to tickle builders

Debugging the builders and need a commit to make them run.

R=adg
CC=golang-codereviews
https://golang.org/cl/149600043

10 years agocmd/ld: ignore .Linfo_stringNN variables in clang .o files
Russ Cox [Tue, 7 Oct 2014 00:51:05 +0000 (20:51 -0400)]
cmd/ld: ignore .Linfo_stringNN variables in clang .o files

http://build.golang.org/log/c7a91b6eac8f8daa2bd17801be273e58403a15f2

# cmd/pprof
/linux-386-clang-9115aad1dc4a/go/pkg/linux_386/net.a(_all.o): sym#16: ignoring .Linfo_string0 in section 16 (type 0)
/linux-386-clang-9115aad1dc4a/go/pkg/linux_386/net.a(_all.o): sym#17: ignoring .Linfo_string1 in section 16 (type 0)
/linux-386-clang-9115aad1dc4a/go/pkg/linux_386/net.a(_all.o): sym#18: ignoring .Linfo_string2 in section 16 (type 0)
/linux-386-clang-9115aad1dc4a/go/pkg/linux_386/net.a(_all.o): sym#20: ignoring .Linfo_string0 in section 16 (type 0)
/linux-386-clang-9115aad1dc4a/go/pkg/linux_386/net.a(_all.o): sym#21: ignoring .Linfo_string1 in section 16 (type 0)
...

I don't know what these are. Let's ignore them and see if we get any further.

TBR=iant
CC=golang-codereviews
https://golang.org/cl/155030043

10 years agoos, syscall: test Chtimes on directories, fix on Windows
Russ Cox [Mon, 6 Oct 2014 23:22:48 +0000 (19:22 -0400)]
os, syscall: test Chtimes on directories, fix on Windows

Fixes #8090.

LGTM=alex.brainman
R=alex.brainman
CC=golang-codereviews
https://golang.org/cl/154020043

10 years agostrings: use fast path for IndexRune
Brad Fitzpatrick [Mon, 6 Oct 2014 22:10:51 +0000 (15:10 -0700)]
strings: use fast path for IndexRune

Noticed while reviewing https://golang.org/cl/147690043/

I'd never seen anybody use IndexRune before, and
unsurprisingly it doesn't use the other fast paths in the
strings/bytes packages. IndexByte uses assembly.

Also, less code this way.

LGTM=r
R=r
CC=golang-codereviews
https://golang.org/cl/147700043

10 years agogo/build: update docs for GOOS.go change
Brad Fitzpatrick [Mon, 6 Oct 2014 22:08:31 +0000 (15:08 -0700)]
go/build: update docs for GOOS.go change

Forgotten in https://golang.org/cl/147690043/

Update #8838

LGTM=r
R=r
CC=golang-codereviews, rsc
https://golang.org/cl/152220045

10 years agodoc/go1.4.txt: GOOS suffixes require underscores to act as build tags
Rob Pike [Mon, 6 Oct 2014 21:58:59 +0000 (14:58 -0700)]
doc/go1.4.txt: GOOS suffixes require underscores to act as build tags

CC=golang-codereviews
https://golang.org/cl/151490043

10 years agogo/build: do not consider "android.go" to be android-specific
Rob Pike [Mon, 6 Oct 2014 21:50:58 +0000 (14:50 -0700)]
go/build: do not consider "android.go" to be android-specific
A file name must have a non-empty underscore-separated
prefix before its suffix matches GOOS. This is what the
documentation already said but is not what the code did.

Fixes #8838.

This needs to be called out in the release notes.
The he single affected file
        code.google.com/p/go.text/collate/tools/colcmp/darwin.go
could use a renaming but works because it has a build tag inside.

LGTM=adg, rsc
R=golang-codereviews, adg, rsc
CC=golang-codereviews
https://golang.org/cl/147690043

10 years agocmd/gc: prohibit short variable declarations containing duplicate symbols
Evan Kroske [Mon, 6 Oct 2014 21:16:39 +0000 (17:16 -0400)]
cmd/gc: prohibit short variable declarations containing duplicate symbols

Fixes #6764.
Fixes #8435.

LGTM=rsc
R=golang-codereviews, r, gobot, rsc
CC=golang-codereviews
https://golang.org/cl/116440046

10 years agoos: recomment MkdirAll
Russ Cox [Mon, 6 Oct 2014 19:49:33 +0000 (15:49 -0400)]
os: recomment MkdirAll

The internal comments are not completely precise about
what is going on, and they are causing confusion.

Fixes #8283.

LGTM=r
R=r
CC=golang-codereviews
https://golang.org/cl/151460043

10 years agoos: make Process.Signal 'process finished' error consistent on Unix
Russ Cox [Mon, 6 Oct 2014 19:49:19 +0000 (15:49 -0400)]
os: make Process.Signal 'process finished' error consistent on Unix

While we're here, fix the implementation of Release on both
Unix and Windows: Release is supposed to make Signal an error.

While we're here, make sure we never Signal pid 0.
(Don't try this at home.)

Fixes #7658.

LGTM=r
R=golang-codereviews, r
CC=golang-codereviews, iant
https://golang.org/cl/152240043

10 years agonet/url: document result of String
Russ Cox [Mon, 6 Oct 2014 19:49:07 +0000 (15:49 -0400)]
net/url: document result of String

Fixes #8742.

LGTM=bradfitz
R=golang-codereviews
CC=adg, bradfitz, golang-codereviews, iant
https://golang.org/cl/155910043

10 years agoencoding/json: document that embedded interfaces look like non-embedded ones
Russ Cox [Mon, 6 Oct 2014 19:48:17 +0000 (15:48 -0400)]
encoding/json: document that embedded interfaces look like non-embedded ones

Fixes #8386.

LGTM=r
R=golang-codereviews, r
CC=golang-codereviews, iant
https://golang.org/cl/149570043

10 years agoregexp/syntax: regenerate doc.go from re2 syntax
Russ Cox [Mon, 6 Oct 2014 19:32:11 +0000 (15:32 -0400)]
regexp/syntax: regenerate doc.go from re2 syntax

Generated using re2/doc/mksyntaxgo.

Fixes #8505.

LGTM=iant
R=r, iant
CC=golang-codereviews
https://golang.org/cl/155890043

10 years agocmd/go: fix 'go vet' of package with external tests
Russ Cox [Mon, 6 Oct 2014 18:49:22 +0000 (14:49 -0400)]
cmd/go: fix 'go vet' of package with external tests

For example, fixes 'go vet syscall', which has source
files in package syscall_test.

Fixes #8511.

LGTM=r
R=golang-codereviews, r
CC=golang-codereviews, iant
https://golang.org/cl/152220044

10 years agoruntime: update docs, code for SetFinalizer
Russ Cox [Mon, 6 Oct 2014 18:18:09 +0000 (14:18 -0400)]
runtime: update docs, code for SetFinalizer

At last minute before 1.3 we relaxed SetFinalizer to avoid
crashes when you pass the result of a global alloc to it.
This avoids the crash but makes SetFinalizer a bit too relaxed.

Document that the finalizer of a global allocation may not run.

Tighten the SetFinalizer check to ignore a global allocation but
not ignore everything else.

Fixes #7656.

LGTM=r, iant
R=golang-codereviews, iant, r
CC=dvyukov, golang-codereviews, khr, rlh
https://golang.org/cl/145930043

10 years agocmd/8l: accept R_386_GOT32 in push instruction
Russ Cox [Mon, 6 Oct 2014 18:17:48 +0000 (14:17 -0400)]
cmd/8l: accept R_386_GOT32 in push instruction

Fixes #8382.

LGTM=iant
R=iant
CC=golang-codereviews
https://golang.org/cl/149540045

10 years agoA+C: Jens Frederich (individual CLA)
Brad Fitzpatrick [Mon, 6 Oct 2014 17:41:24 +0000 (10:41 -0700)]
A+C: Jens Frederich (individual CLA)

Generated by a+c.

R=gobot
CC=golang-codereviews
https://golang.org/cl/151450043

10 years agonet: disable TestDualStackUDPListener in short mode
Russ Cox [Mon, 6 Oct 2014 17:09:14 +0000 (13:09 -0400)]
net: disable TestDualStackUDPListener in short mode

This test is flaky, just like TestDualStackTCPListener.
That one was disabled. Disable this one too.

Update #5001

LGTM=bradfitz
R=rlh, bradfitz
CC=golang-codereviews
https://golang.org/cl/154950043

10 years agosyscall: another attempt to keep windows syscall pointers live
Alex Brainman [Sun, 5 Oct 2014 02:15:13 +0000 (13:15 +1100)]
syscall: another attempt to keep windows syscall pointers live

This approach was suggested in
https://golang.org/cl/138250043/#msg15.
Unlike current version of mksyscall_windows.go,
new code could be used in go.sys and other external
repos without help from asm.

LGTM=iant
R=golang-codereviews, iant, r
CC=golang-codereviews
https://golang.org/cl/143160046

10 years agodoc/go1.4: document memory consumption reduction
Dmitriy Vyukov [Sat, 4 Oct 2014 08:02:04 +0000 (12:02 +0400)]
doc/go1.4: document memory consumption reduction

R=golang-codereviews, bradfitz
CC=golang-codereviews
https://golang.org/cl/149370043

10 years agodoc/go1.4.txt: fmt: document printing of &map rather than just pointer
Rob Pike [Sat, 4 Oct 2014 03:30:06 +0000 (20:30 -0700)]
doc/go1.4.txt: fmt: document printing of &map rather than just pointer

CC=golang-codereviews
https://golang.org/cl/152160043

10 years agofmt: print &map like &slice and &struct
Rob Pike [Sat, 4 Oct 2014 03:27:08 +0000 (20:27 -0700)]
fmt: print &map like &slice and &struct
It was inconsistent.
Also test these better.
Also document the default format for types.
This wasn't written down.

Fixes #8470.

LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/154870043

10 years agodoc/go1.4: mention that we are going to need a new version of SWIG
Ian Lance Taylor [Sat, 4 Oct 2014 01:12:05 +0000 (18:12 -0700)]
doc/go1.4: mention that we are going to need a new version of SWIG

Update #8750

CC=golang-codereviews
https://golang.org/cl/153950044

10 years agocrypto/x509: add OpenELEC system root location.
Dave Cheney [Fri, 3 Oct 2014 22:09:39 +0000 (08:09 +1000)]
crypto/x509: add OpenELEC system root location.

Fixes #8349.

LGTM=rsc
R=rsc
CC=golang-codereviews
https://golang.org/cl/147320043

10 years agomisc/cgo/test: fail issue 7978 with clearer error when GOTRACEBACK != 2
Russ Cox [Fri, 3 Oct 2014 20:42:18 +0000 (16:42 -0400)]
misc/cgo/test: fail issue 7978 with clearer error when GOTRACEBACK != 2

The test doesn't work with GOTRACEBACK != 2.
Diagnose that failure mode.

LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews, r
https://golang.org/cl/152970043

10 years agofmt: part 2 of the great flag rebuild: make %+v work in formatters
Rob Pike [Fri, 3 Oct 2014 20:23:35 +0000 (13:23 -0700)]
fmt: part 2 of the great flag rebuild: make %+v work in formatters
Apply a similar transformation to %+v that we did to %#v, making it
a top-level setting separate from the + flag itself. This fixes the
appearance of flags in Formatters and cleans up the code too,
probably making it a little faster.

Fixes #8835.

LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/154820043

10 years agoruntime: clear sg.selectdone before saving in SudoG cache
Russ Cox [Fri, 3 Oct 2014 19:33:29 +0000 (15:33 -0400)]
runtime: clear sg.selectdone before saving in SudoG cache

Removes another dangling pointer that might
cause a memory leak in 1.4 or crash the GC in 1.5.

LGTM=rlh
R=golang-codereviews
CC=golang-codereviews, iant, khr, r, rlh
https://golang.org/cl/150520043

10 years agoruntime: clear stale values from G.param and SudoG.elem
Russ Cox [Fri, 3 Oct 2014 17:36:48 +0000 (13:36 -0400)]
runtime: clear stale values from G.param and SudoG.elem

This change was necessary on the dev.garbage branch
to keep the garbage collector from seeing pointers into
invalid heap areas.

On this default (Go 1.4) branch, the change removes
some possibility for memory leaks.

LGTM=khr
R=golang-codereviews, khr
CC=golang-codereviews, iant, r, rlh
https://golang.org/cl/155760043

10 years agocmd/cc, runtime: disallow structs without tags
Russ Cox [Fri, 3 Oct 2014 16:44:20 +0000 (12:44 -0400)]
cmd/cc, runtime: disallow structs without tags

Structs without tags have no unique name to use in the
Go definitions generated from the C types.
This caused issue 8812, fixed by CL 149260043.
Avoid future problems by requiring struct tags.

Update runtime as needed.
(There is no other C code in the tree.)

LGTM=bradfitz, iant
R=golang-codereviews, bradfitz, dave, iant
CC=golang-codereviews, khr, r
https://golang.org/cl/150360043

10 years agoundo CL 138250043 / 4eda5e4001fd
Alex Brainman [Fri, 3 Oct 2014 00:36:54 +0000 (10:36 +1000)]
undo CL 138250043 / 4eda5e4001fd

I will use different approach to solve this problem.
See CL 143160046 for details.

««« original CL description
syscall: keep Windows syscall pointers live too

Like https://golang.org/cl/139360044

LGTM=rsc, alex.brainman
R=alex.brainman, rsc
CC=golang-codereviews
https://golang.org/cl/138250043
»»»

LGTM=r
R=golang-codereviews, bradfitz, r
CC=golang-codereviews
https://golang.org/cl/147440043

10 years agofmt: make the %#v verb a special flag
Rob Pike [Thu, 2 Oct 2014 21:16:58 +0000 (14:16 -0700)]
fmt: make the %#v verb a special flag
The %#v verb is special: it says all values below need to print as %#v.
However, for some situations the # flag has other meanings and this
causes some issues, particularly in how Formatters work. Since %#v
dominates all formatting, translate it into actual state of the formatter
and decouple it from the # flag itself within the calculations (although
it must be restored when methods are doing the work.)
The result is cleaner code and correct handling of # for Formatters.
TODO: Apply the same thinking to the + flag in a followup CL.

Also, the wasString return value in handleMethods is always false,
so eliminate it.

Update #8835

LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/146650043

10 years agomath/big: math.Exp should return result >= 0 for |m| > 0
Robert Griesemer [Thu, 2 Oct 2014 20:02:25 +0000 (13:02 -0700)]
math/big: math.Exp should return result >= 0 for |m| > 0

The documentation states that Exp(x, y, m)
computes x**y mod |m| for m != nil && m > 0.
In math.big, Mod is the Euclidean modulus,
which is always >= 0.

Fixes #8822.

LGTM=agl, r, rsc
R=agl, r, rsc
CC=golang-codereviews
https://golang.org/cl/145650043

10 years agoencoding/binary: fix error message
Robert Griesemer [Thu, 2 Oct 2014 19:53:51 +0000 (12:53 -0700)]
encoding/binary: fix error message

In the process, simplified internal sizeOf and
dataSize functions. Minor positive impact on
performance. Added test case.

benchmark                         old ns/op     new ns/op     delta
BenchmarkReadSlice1000Int32s      14006         14122         +0.83%
BenchmarkReadStruct               2508          2447          -2.43%
BenchmarkReadInts                 921           928           +0.76%
BenchmarkWriteInts                2086          2081          -0.24%
BenchmarkWriteSlice1000Int32s     13440         13497         +0.42%
BenchmarkPutUvarint32             28.5          26.3          -7.72%
BenchmarkPutUvarint64             81.3          76.7          -5.66%

benchmark                         old MB/s     new MB/s     speedup
BenchmarkReadSlice1000Int32s      285.58       283.24       0.99x
BenchmarkReadStruct               27.90        28.60        1.03x
BenchmarkReadInts                 32.57        32.31        0.99x
BenchmarkWriteInts                14.38        14.41        1.00x
BenchmarkWriteSlice1000Int32s     297.60       296.36       1.00x
BenchmarkPutUvarint32             140.55       151.92       1.08x
BenchmarkPutUvarint64             98.36        104.33       1.06x

Fixes #6818.

LGTM=r
R=r
CC=golang-codereviews
https://golang.org/cl/149290045

10 years agosyscall: support UID/GID map files for Linux user namespaces
Mrunal Patel [Thu, 2 Oct 2014 18:37:06 +0000 (11:37 -0700)]
syscall: support UID/GID map files for Linux user namespaces

Fixes #8447.

LGTM=iant
R=golang-codereviews, bradfitz, iant
CC=golang-codereviews
https://golang.org/cl/126190043

10 years agoA+C: add Mrunal Patel (Red Hat corporate CLA)
Ian Lance Taylor [Thu, 2 Oct 2014 18:36:39 +0000 (11:36 -0700)]
A+C: add Mrunal Patel (Red Hat corporate CLA)

LGTM=bradfitz
R=golang-codereviews, bradfitz
CC=golang-codereviews
https://golang.org/cl/152010043

10 years agosyscall: fix Setenv for plan 9
Nicolas Owens [Thu, 2 Oct 2014 08:25:56 +0000 (10:25 +0200)]
syscall: fix Setenv for plan 9

envi needs to be updated during Setenv so the key can be correctly deleted later with Unsetenv.

Update #8849.

LGTM=0intro
R=bradfitz, 0intro
CC=golang-codereviews
https://golang.org/cl/149300046

10 years agorun.bat: comment text properly
Alex Brainman [Thu, 2 Oct 2014 06:15:36 +0000 (16:15 +1000)]
run.bat: comment text properly

LGTM=bradfitz
R=golang-codereviews, bradfitz
CC=golang-codereviews
https://golang.org/cl/153830044

10 years agotime: A Timer must be properly created before use (documentation).
Robert Griesemer [Wed, 1 Oct 2014 23:44:52 +0000 (16:44 -0700)]
time: A Timer must be properly created before use (documentation).

Fixes #8776.

LGTM=bradfitz
R=rsc, bradfitz
CC=golang-codereviews
https://golang.org/cl/145710043

10 years agoencoding/json: don't panic on incorrect map argument
Robert Griesemer [Wed, 1 Oct 2014 23:24:17 +0000 (16:24 -0700)]
encoding/json: don't panic on incorrect map argument

Fixes #8305.

LGTM=rsc
R=rsc
CC=golang-codereviews
https://golang.org/cl/145680044

10 years agoencoding/binary: slightly better documentation
Robert Griesemer [Wed, 1 Oct 2014 23:09:55 +0000 (16:09 -0700)]
encoding/binary: slightly better documentation

Fixes #7306.

LGTM=r
R=r
CC=golang-codereviews
https://golang.org/cl/153820044

10 years agodoc/go_faq.html: explain the policy about unused imports a little better
Rob Pike [Wed, 1 Oct 2014 22:25:56 +0000 (15:25 -0700)]
doc/go_faq.html: explain the policy about unused imports a little better
This new text won't stop the whining but it might focus the whining a little more.

LGTM=adg
R=golang-codereviews, adg
CC=golang-codereviews
https://golang.org/cl/146680044

10 years agofmt: fix internal unknownType function
Rob Pike [Wed, 1 Oct 2014 21:35:12 +0000 (21:35 +0000)]
fmt: fix internal unknownType function
This thing should never be called, but before
151960044 it was being called, incorrectly.
This is now just a precaution but let's pretend it
Fixes #8843
even though that was fixed by 151960044.
The test case was already there and ran, another mystery.

LGTM=rsc
R=rsc
CC=golang-codereviews
https://golang.org/cl/151970043

10 years agoreflect: fix IsValid vs Kind mismatch after Elem of nil interface
Russ Cox [Wed, 1 Oct 2014 20:51:32 +0000 (16:51 -0400)]
reflect: fix IsValid vs Kind mismatch after Elem of nil interface

LGTM=r
R=r
CC=golang-codereviews
https://golang.org/cl/151960044

10 years agotesting: clearer comment
Robert Griesemer [Wed, 1 Oct 2014 20:19:40 +0000 (13:19 -0700)]
testing: clearer comment

Fixes #8797.

LGTM=r
R=r
CC=golang-codereviews
https://golang.org/cl/146680043

10 years agonet/rpc: shut down connection if gob has error
Rob Pike [Wed, 1 Oct 2014 20:18:44 +0000 (13:18 -0700)]
net/rpc: shut down connection if gob has error
The nicest solution would be to buffer the message and only write
it if it encodes correctly, but that adds considerable memory and
CPU overhead for a very rare condition. Instead, we just shut
down the connection if this happens.
Fixes #7689.

LGTM=rsc
R=rsc
CC=golang-codereviews
https://golang.org/cl/146670043

10 years agoos, syscall: add Unsetenv
Brad Fitzpatrick [Wed, 1 Oct 2014 18:17:15 +0000 (11:17 -0700)]
os, syscall: add Unsetenv

Also address a TODO, making Clearenv pass through to cgo.

Based largely on Minux's earlier https://golang.org/cl/82040044

Fixes #6423

LGTM=iant, alex.brainman, r, rsc
R=rsc, iant, r, alex.brainman
CC=golang-codereviews
https://golang.org/cl/148370043

10 years agotag go1.3.3
Andrew Gerrand [Wed, 1 Oct 2014 03:38:00 +0000 (13:38 +1000)]
tag go1.3.3

LGTM=crawshaw
R=golang-codereviews, crawshaw
CC=golang-codereviews
https://golang.org/cl/146610043

10 years agomisc/nacl/README: fix typo
Mikio Hara [Wed, 1 Oct 2014 00:16:55 +0000 (09:16 +0900)]
misc/nacl/README: fix typo

LGTM=dave
R=golang-codereviews, dave
CC=golang-codereviews
https://golang.org/cl/151870043

10 years agodoc: document go1.3.3
Andrew Gerrand [Tue, 30 Sep 2014 22:30:51 +0000 (08:30 +1000)]
doc: document go1.3.3

LGTM=r
R=r
CC=golang-codereviews
https://golang.org/cl/149280044

10 years agox509: Fixed ASN.1 encoding in CRL Distribution Points extension
Paul van Brouwershaven [Tue, 30 Sep 2014 20:38:48 +0000 (13:38 -0700)]
x509: Fixed ASN.1 encoding in CRL Distribution Points extension

The ASN.1 encoding of the CRL Distribution Points extension showed an invalid false 'IsCompound' which caused a display problem in the Windows certificate viewer.

LGTM=agl
R=agl
CC=golang-codereviews
https://golang.org/cl/143320043