1) The arrayindexof lookup function is O(n). Replace with O(1) lookups.

2) The checkptxt function is O(n²) and is purely for debugging.
Only run when the debugging flags are turned on.

3) Iterating over sparse bitmaps can be done faster word by word.
Introduce and use bvnext for that.

Run times before and after, on my 2.5 GHz Core i5 MacBook Pro.

x.go       9.48  0.84  issue 8259

x100.go    0.01  0.01  issue 8354
x1000.go   0.10  0.10
x2000.go   0.62  0.19
x3000.go   1.33  0.34
x4000.go   2.29  0.49
x5000.go   3.89  0.67
x6000.go   5.00  0.90
x7000.go   6.70  1.13
x8000.go   9.44  1.38
x9000.go  11.23  1.87
x10000.go 13.78  2.09

Fixes #8259.
Fixes #8354.

LGTM=iant, r
R=golang-codereviews, iant, r
CC=golang-codereviews
https://golang.org/cl/125720043

LGTM=minux
R=minux
CC=golang-codereviews
https://golang.org/cl/118690043

LGTM=minux
R=minux
CC=golang-codereviews
https://golang.org/cl/124800043

The linker currently produces the DWARF 3 DW_TAG_unspecified_type tag, however the Reader in debug/dwarf will panic whenever that tag is encountered.

Fixes #8437.

LGTM=rsc
R=golang-codereviews, bradfitz, iant, rsc
CC=golang-codereviews
https://golang.org/cl/117280043

Generated by a+c.

R=gobot
CC=golang-codereviews
https://golang.org/cl/125750043

This is an experiment. See mail on golang-dev
(subject: "an experiment: development branches").

LGTM=minux
R=minux
CC=golang-codereviews
https://golang.org/cl/117660043

It can happen legitimately if a profiling signal arrives at just the wrong moment.
It's harmless.

Fixes #8153.

LGTM=minux
R=golang-codereviews, minux
CC=golang-codereviews, iant, r
https://golang.org/cl/118670043

Revised version of https://golang.org/cl/81260045/

LGTM=agl
R=golang-codereviews, gobot, agl, ox
CC=golang-codereviews
https://golang.org/cl/107400043

Fixes #8478.

LGTM=iant
R=iant
CC=golang-codereviews
https://golang.org/cl/122150043

Left over from cl/119490044.

LGTM=bradfitz
R=rsc, bradfitz
CC=golang-codereviews
https://golang.org/cl/125730043

Full spans can't be passed to UncacheSpan since we get rid of free.

LGTM=rsc
R=golang-codereviews
CC=golang-codereviews, khr, rsc
https://golang.org/cl/119490044

Since CL 115060044, mkanames declares an empty
array in anames8.c and anames6.c, which is not
valid for the Plan 9 compiler.

char* cnames8[] = {
};

This change makes mkanames not declaring the
cnames array when no C_ constants are found.

LGTM=iant
R=minux, iant
CC=golang-codereviews
https://golang.org/cl/117680043

LGTM=dvyukov
R=golang-codereviews, dvyukov
CC=golang-codereviews
https://golang.org/cl/117670043

Broke freebsd/amd64 due to exposure of a latent bug.

««« original CL description
cmd/ld: fix operator precedence

LGTM=rsc
R=rsc, iant
CC=golang-codereviews
https://golang.org/cl/114420043
»»»

TBR=dfc
R=dave
CC=golang-codereviews
https://golang.org/cl/120630043

LGTM=rsc
R=golang-codereviews, gobot, rsc
CC=golang-codereviews
https://golang.org/cl/118480046

The helps certain diagnostics and also removed duplicated enums as a side effect.

LGTM=dave, rsc
R=rsc, dave
CC=golang-codereviews
https://golang.org/cl/115060044

LGTM=rsc
R=rsc, iant
CC=golang-codereviews
https://golang.org/cl/118610043

LGTM=rsc
R=rsc, iant
CC=golang-codereviews
https://golang.org/cl/120220043

LGTM=rsc
R=rsc, iant
CC=golang-codereviews
https://golang.org/cl/115300044

LGTM=rsc
R=rsc, iant
CC=golang-codereviews
https://golang.org/cl/114420043

The manpages says SA_RESTORER is obsolete, and indeed, not every architecture
support it. However, sadly it's required on x86_64, see http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/arch/x86/kernel/signal.c?id=26bcd8b72563b4c54892c4c2a409f6656fb8ae8b#n430, so only use it on x86.

LGTM=rsc
R=rsc, iant
CC=golang-codereviews
https://golang.org/cl/115450043

Rather than switch on thechar.

LGTM=rsc, dave
R=rsc, iant, dave
CC=golang-codereviews
https://golang.org/cl/119330043

LGTM=bradfitz, dave, ruiu
R=rsc, iant, bradfitz, dave, ruiu
CC=golang-codereviews
https://golang.org/cl/116610043

Fixes build for nacl/386.

LGTM=dave
R=khr, bradfitz, dave, dan.kortschak, rsc
CC=golang-codereviews
https://golang.org/cl/121080043

Encode MOV $0, %ax as XOR %eax, %eax instead of
XOR %rax, %rax. If an operand register does not
need REX.w bit (i.e. not one of R8-R15), it is
encoded in 2 bytes instead of 3 bytes.

LGTM=rsc
R=golang-codereviews, gobot, rsc
CC=golang-codereviews
https://golang.org/cl/115580044

Instead of immediately completing pointer type mappings, add them to
a queue to allow them to be completed later.  This fixes issues	caused
by Type() returning arbitrary in-progress type mappings.

Fixes #8368.
Fixes #8441.

LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/122850043

Instead of including <sys/types.h> to get size_t, instead include
the ISO C standard <stddef.h> header, which defines fewer additional
types at risk of colliding with the user code.  In particular, this
prevents collisions between <sys/types.h>'s userspace definitions with
the kernel definitions needed by defs_linux.go.

Also, -cdefs mode uses #pragma pack, so we can keep misaligned fields.

Fixes #8477.

LGTM=iant
R=golang-codereviews, iant
CC=golang-codereviews
https://golang.org/cl/120610043

Before fixing issue 6579 this CL separates DNS transport from
DNS message interaction to make it easier to add builtin DNS
resolver control logic.

Update #6579

LGTM=alex, kevlar
R=golang-codereviews, alex, gobot, iant, minux, kevlar
CC=golang-codereviews
https://golang.org/cl/101220044

Update #6677

When a struct contains an anonymous union, use the type and
name of the first field in the union.

This should make the glibc <sys/resource.h> file work; in that
file struct rusage has fields like

__extension__ union
{
        long int ru_maxrss;
        __syscall_slong_t __ru_maxrss_word;
};

in which the field that matters is ru_maxrss and
__ru_maxrss_word just exists to advance to the next field on
systems where the kernel uses long long fields but userspace
expects long fields.

LGTM=mikioh.mikioh
R=golang-codereviews, mikioh.mikioh
CC=golang-codereviews
https://golang.org/cl/106260044

We call scanblock for lots of small root pieces
e.g. for every stack frame args and locals area.
Every scanblock invocation calls getempty/putempty,
which accesses lock-free stack shared among all worker threads.
One-element local cache allows most scanblock calls
to proceed without accessing the shared stack.

LGTM=rsc
R=golang-codereviews, rlh
CC=golang-codereviews, khr, rsc
https://golang.org/cl/121250043

If the process exits before the spawned goroutine
completes, it'll miss the data race.

LGTM=bradfitz
R=bradfitz
CC=dvyukov, golang-codereviews
https://golang.org/cl/122120043

CC=golang-codereviews
https://golang.org/cl/116710043

Camlistore uses this pattern to do streaming writes, as do
others I imagine, and it was broken by the lazy boundary
change.

LGTM=dvyukov, ruiu
R=ruiu, dvyukov
CC=golang-codereviews, mathieu.lonjaret
https://golang.org/cl/116690043

Breaks Camlistore by introducing a datarace. See comments on
https://golang.org/cl/95760043/ for details.

I'll add a new test to lock-in the current behavior in a
subsequent CL.

I don't think Camlistore is particularly unique here: it's doing
the obvious thing to stream a multipart body to a server
using a goroutine feeding the multipart writer.

««« original CL description
mime/multipart: delay reading random source

If a user sets his/her own boundary string with SetBoundary,
we don't need to call randomBoundary at all.

LGTM=bradfitz
R=golang-codereviews, bradfitz
CC=golang-codereviews
https://golang.org/cl/95760043
»»»

LGTM=ruiu
R=ruiu
CC=golang-codereviews, mathieu.lonjaret
https://golang.org/cl/117600043

Fixes #6736.

LGTM=mikioh.mikioh
R=bradfitz, mikioh.mikioh
CC=golang-codereviews
https://golang.org/cl/108710046

We have an autogenerated version in zruntime_defs.
I am not sure what are the consequences as gdb never printed any values for me.
But it looks unnecessary to manually duplicate it.

LGTM=rsc
R=golang-codereviews, rsc
CC=golang-codereviews, iant, khr
https://golang.org/cl/115660043

Technically a language change, this cleanup is a completely
backward compatible change that brings the boolean results
of comma-ok expressions in line with the boolean results of
comparisons: they are now all untyped booleans.

The implementation effort should be minimal (less than a
handfull lines of code, depending how well factored the
implementation of comma-ok expressions is).

Fixes #8189.

LGTM=iant, r, rsc
R=r, rsc, iant, ken
CC=golang-codereviews
https://golang.org/cl/112320045

Generated by a+c.

R=gobot
CC=golang-codereviews
https://golang.org/cl/121240043

With the recent GC changes large objects are handled somewhat differently.

LGTM=khr
R=khr
CC=golang-codereviews
https://golang.org/cl/114600043

LGTM=bradfitz, khr
R=khr, bradfitz
CC=golang-codereviews
https://golang.org/cl/120400043