Also use clone(2) syscall instead of fork().

Fixes #6214.

R=golang-dev, bradfitz, dave
CC=golang-dev
https://golang.org/cl/13159044

the use of the flag, especially for objects which actually do have
pointers but we don't want the GC to scan them.

R=golang-dev, cshapiro
CC=golang-dev
https://golang.org/cl/13181045

slice type to an array type, the haspointer-ness may change.
Before this change, we'd sometimes get types like [1]int marked
as having pointers.

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/13189044

Update #5780

R=golang-dev, cshapiro, dave, bradfitz
CC=golang-dev
https://golang.org/cl/12869049

Update #6214

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/13160043

Enable remaining packages (crypto, mime, path, time) to build on dragonfly.

R=bradfitz
CC=golang-dev
https://golang.org/cl/13190043

Make the net package build and work on dragonfly.

R=bradfitz
CC=golang-dev
https://golang.org/cl/13173044

Make the os package build and work on dragonfly.

R=bradfitz
CC=golang-dev
https://golang.org/cl/13183044

Add generated z-files for dragonfly/amd64.

R=bradfitz
CC=golang-dev
https://golang.org/cl/13056045

Add syscall support for dragonfly/amd64.

Also add support for generating syscall z* files for dragonfly.

R=bradfitz
CC=golang-dev
https://golang.org/cl/13188043

Go runtime support for dragonfly/amd64, largely based of the existing
FreeBSD runtime (with some clues from the varialus/godfly work).

R=bradfitz
CC=golang-dev
https://golang.org/cl/13088044

Fixes #6227

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/13183043

Generated by addca.

R=gobot
CC=golang-dev
https://golang.org/cl/13189043

Add dragonflydynld to 5l and 8l so that they compile again.

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/12739048

Add dragonfly/amd64 support to the Go compiler, bootstrap and GOOS list.

R=devon.odell, bradfitz
CC=golang-dev
https://golang.org/cl/12796050

Also avoids platform-dependent datagram truncation in raw IP tests.
At least it's different between Windows and others.

Fixes #6122.

R=alex.brainman
CC=golang-dev
https://golang.org/cl/12843043

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/13141045

This is an attempt to make our slow
windows-386 builder more reliable.

R=golang-dev, dave
CC=golang-dev
https://golang.org/cl/12798045

Also makes variable names a bit cleaner.

R=golang-dev, dave, r
CC=golang-dev
https://golang.org/cl/12808047

Update #5344

R=alex.brainman
CC=golang-dev
https://golang.org/cl/12966046

R=golang-dev, adg
CC=golang-dev, mikioh.mikioh
https://golang.org/cl/12848047

This CL adds minimal information for supporting platforms that don't
have a complete list of internet protocol numbers.

Fixes #5344.

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/12898045

The net package consists of thin three layers like the follwoing;

- Exposed API, that contains net.Dial, net.DialUDP, net.DialUnix
- Socket and network file descriptor, that contains net.netFD and
  its methods, helper functions such as dialUDP, dialUnix
- Network pollster, that contains net.pollDesc and its methods

This CL removes redundant argument check which is already done by
API layer.

R=golang-dev, dave, bradfitz
CC=golang-dev
https://golang.org/cl/13092043

Fixes #5630.

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/12884046

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/13105047

GC acquires worldsema, which is a goroutine-level semaphore
which parks goroutines. g0 can not be parked.
Fixes #6193.

R=khr, khr
CC=golang-dev
https://golang.org/cl/12880045

Update the original change but do not read interface types in
the arguments area.  Once the arguments area is zeroed as the
locals area is we can safely read interface type values there
too.

««« original CL description
undo CL 12785045 / 71ce80dc4195

This has broken the 32-bit builds.

««« original CL description
cmd/gc, runtime: use type information to scan interface values

R=golang-dev, rsc, dvyukov
CC=golang-dev
https://golang.org/cl/12785045
»»»

R=khr, golang-dev, khr
CC=golang-dev
https://golang.org/cl/13010045
»»»

R=khr, khr
CC=golang-dev
https://golang.org/cl/13073045

This slightly improves performance when a lot of timers are present

$ misc/benchcmp ../old_timers_m.txt ../new_timers_m.txt
benchmark                           old ns/op    new ns/op    delta
BenchmarkAfterFunc                       6884         6605   -4.05%
BenchmarkAfterFunc-2                     4473         4144   -7.36%
BenchmarkAfterFunc-3                     8601         6185  -28.09%
BenchmarkAfterFunc-4                     9378         8773   -6.45%
BenchmarkAfter                           7237         7278   +0.57%
BenchmarkAfter-2                         4638         3923  -15.42%
BenchmarkAfter-3                         8751         6239  -28.71%
BenchmarkAfter-4                         9223         8737   -5.27%
BenchmarkStop                             603          496  -17.74%
BenchmarkStop-2                           795          577  -27.42%
BenchmarkStop-3                           982          680  -30.75%
BenchmarkStop-4                          1164          739  -36.51%
BenchmarkSimultaneousAfterFunc            657          593   -9.74%
BenchmarkSimultaneousAfterFunc-2          816          757   -7.23%
BenchmarkSimultaneousAfterFunc-3          844          830   -1.66%
BenchmarkSimultaneousAfterFunc-4          785          771   -1.78%
BenchmarkStartStop                        238          239   +0.42%
BenchmarkStartStop-2                      249          234   -6.02%
BenchmarkStartStop-3                      271          268   -1.11%
BenchmarkStartStop-4                      293          295   +0.68%

R=golang-dev, dvyukov, bradfitz, r
CC=golang-dev
https://golang.org/cl/13094043

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/12780045

Replace linked list walk with memset.
This reduces CPU time taken by 'go install -a std' by ~10%.
Before:
real		user		sys
0m23.561s	0m16.625s	0m5.848s
0m23.766s	0m16.624s	0m5.846s
0m23.742s	0m16.621s	0m5.868s
after:
0m22.714s	0m14.858s	0m6.138s
0m22.644s	0m14.875s	0m6.120s
0m22.604s	0m14.854s	0m6.081s

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/13084043

R=r
CC=golang-dev
https://golang.org/cl/13019044

pointer.  An example that triggers the bad behavior on a 64bit
machine http://play.golang.org/p/GrNFakAYLN
        rv1 := reflect.ValueOf(complex128(0))
        rt := rv1.Type()
        rv2 := rv1.Convert(rt)
        rv3 := reflect.New(rt).Elem()
        rv3.Set(rv2)

Running the code fails with the following:
        panic: reflect: internal error: storeIword of 16-byte value

I've tested on a 64bit machine and verified this fixes the panic.  I
haven't tested on a 32bit machine so I haven't verified the other
cases, but they follow logically.

R=golang-dev, r, iant
CC=golang-dev
https://golang.org/cl/12805045

R=r, adg
CC=golang-dev, toddwang
https://golang.org/cl/13149043

Update #5000
Should reduce the flakiness a little. Malloc counting is important
to general testing but not to the build dashboard, which uses -short.

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/12866047

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/12801045

Add eq, lt, etc. to allow one to do simple comparisons.
It's basic types only (booleans, integers, unsigned integers,
floats, complex, string) because that's easy, easy to define,
and covers the great majority of useful cases, while leaving
open the possibility of a more sweeping definition later.

{{if eq .X .Y}}X and Y are equal{{else}}X and Y are unequal{{end}}

R=golang-dev, adg
CC=golang-dev
https://golang.org/cl/13091045

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/12975044

R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/13140043

Not a user-visible API change.  Just a minor garbage
win in contrived cases.

Also, the package is io/ioutil.

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/12987044

This is potentially an API-breaking change, but it is an important bug fix.

The CL https://golang.org/cl/7305072/ added stuff to make
the tar file look more like a file system internally, including providing an
implementation of os.FileInfo for the file headers within the archive.
But the code is incorrect because FileInfo.Name is supposed to return
the base name only; this implementation returns the full path. A round
trip test added in the same shows this in action, as the slashes are
preserved as we create a header using the local implementation of
FileInfo.

The CL here changes the behavior of the tar (and zip) FileInfo to honor
the Go spec for that interface. It also clarifies that the FileInfoHeader
function, which takes a FileInfo as an argument, will therefore create
a header with only the base name of the file recorded, and that
subsequent adjustment may be necessary.

There may be code out there that depends on the broken behavior.
We can call out the risk in the release notes.

Fixes #6180.

R=golang-dev, dsymonds, adg, bradfitz
CC=golang-dev
https://golang.org/cl/13118043