Moves the network poller from net package into runtime.

benchmark                           old ns/op    new ns/op    delta
BenchmarkTCP4OneShot                   316386       287061   -9.27%
BenchmarkTCP4OneShot-2                 339822       313424   -7.77%
BenchmarkTCP4OneShot-3                 330057       306589   -7.11%
BenchmarkTCP4OneShotTimeout            341775       287061  -16.01%
BenchmarkTCP4OneShotTimeout-2          380835       295849  -22.32%
BenchmarkTCP4OneShotTimeout-3          398412       328070  -17.66%
BenchmarkTCP4Persistent                 40622        33392  -17.80%
BenchmarkTCP4Persistent-2               44528        35736  -19.74%
BenchmarkTCP4Persistent-3               44919        36907  -17.84%
BenchmarkTCP4PersistentTimeout          45309        33588  -25.87%
BenchmarkTCP4PersistentTimeout-2        50289        38079  -24.28%
BenchmarkTCP4PersistentTimeout-3        51559        37103  -28.04%
BenchmarkTCP6OneShot                   361305       345645   -4.33%
BenchmarkTCP6OneShot-2                 361305       331976   -8.12%
BenchmarkTCP6OneShot-3                 376929       347598   -7.78%
BenchmarkTCP6OneShotTimeout            361305       322212  -10.82%
BenchmarkTCP6OneShotTimeout-2          378882       333928  -11.86%
BenchmarkTCP6OneShotTimeout-3          388647       335881  -13.58%
BenchmarkTCP6Persistent                 47653        35345  -25.83%
BenchmarkTCP6Persistent-2               49215        35736  -27.39%
BenchmarkTCP6Persistent-3               38474        37493   -2.55%
BenchmarkTCP6PersistentTimeout          56637        34369  -39.32%
BenchmarkTCP6PersistentTimeout-2        42575        38079  -10.56%
BenchmarkTCP6PersistentTimeout-3        44137        37689  -14.61%

R=dvyukov
CC=golang-dev
https://golang.org/cl/8670044

Trivial, but annoying while debugging this code.

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/11656043

Fixes #5898.

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/11556043

Calls into math/rand are very slow, especially under race
detector because of heap accesses.

go test -bench . -run none -benchtime .1s
Before: 23.0s
After:  17.4s

Fixes #5837.

R=golang-dev, dave, r
CC=golang-dev
https://golang.org/cl/11564044

R=golang-dev, dave, bradfitz, rsc
CC=golang-dev
https://golang.org/cl/10030043

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/11623043

R=golang-dev, dave
CC=golang-dev
https://golang.org/cl/11620044

This does not change the default compiler on OS X to clang.
It appears that for now we can keep using gcc as long as we
enable a few more warning settings that are on-by-default
elsewhere.

R=golang-dev, bradfitz, dave
CC=golang-dev
https://golang.org/cl/11610044

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/11604043

Fixes #5924.

R=golang-dev, khr, rsc
CC=golang-dev
https://golang.org/cl/11606043

This CL introduces a FUNCDATA number for runtime-specific
garbage collection metadata, changes the C and Go compilers
to emit that metadata, and changes the runtime to expect it.

The old pseudo-instructions that carried this information
are gone, as is the linker code to process them.

R=golang-dev, dvyukov, cshapiro
CC=golang-dev
https://golang.org/cl/11406044

So far no checked-in assembly needs these,
but it matches having them for M and G.
I needed these for some manual testing.

R=golang-dev, dvyukov
CC=golang-dev
https://golang.org/cl/11595043

where it belongs.

R=rsc
CC=golang-dev
https://golang.org/cl/11596043

whose argument size is unknown (C vararg functions, and
assembly code without an explicit specification).

We used to use 0 to mean "unknown" and 1 to mean "zero".
Now we use ArgsSizeUnknown (0x80000000) to mean "unknown".

R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/11590043

This holds the last finalized object and arguments to its finalizer.
Fixes #5348.

R=golang-dev, iant
CC=golang-dev
https://golang.org/cl/11454044

R=golang-dev, iant
CC=golang-dev
https://golang.org/cl/11432046

It contains pointers to first blocks of lots of spans.

R=golang-dev, iant
CC=golang-dev
https://golang.org/cl/11416046

If the network is not polled for 10ms, sysmon starts polling network
on every iteration (every 20us) until another thread blocks in netpoll.
Fixes #5922.

R=golang-dev, iant
CC=golang-dev
https://golang.org/cl/11569043

It is an expensive test to run, and even more so with -race,
and causes timeouts on builders. It is doubtful that it would
find a race that other tests in this package wouldn't, so there
is little loss in excluding it.

Update #5837.

R=golang-dev, dave
CC=golang-dev
https://golang.org/cl/11568043

It assumes that the m will not change, and the m may
change if the goroutine is preempted.

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/11560043

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/11542044

If a goroutine runs for more than 10ms, preempt it.
Update #543.

R=rsc
CC=golang-dev
https://golang.org/cl/10796043

If we start a garbage collection on g0 during a
stack split or unsplit, we'll see morestack or lessstack
at the top of the stack. Record an argument frame size
for those, and record that they terminate the stack.

R=golang-dev, dvyukov
CC=golang-dev
https://golang.org/cl/11533043

R=ken2
CC=golang-dev
https://golang.org/cl/11397043

Syntax description was moved to the top level comment of this package.

R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/11536043

Deferreturn is synthesizing a new call frame.
It must not be interrupted between copying the args there
and fixing up the program counter, or else the stack will
be in an inconsistent state, one that will confuse the
garbage collector.

R=golang-dev, dvyukov
CC=golang-dev
https://golang.org/cl/11522043

With preemption, _sfloat2 can show up in stack traces.
Write the function prototype in a way that accurately
shows the frame size and the fact that it might contain
pointers.

R=golang-dev, dvyukov
CC=golang-dev
https://golang.org/cl/11523043

Windows was the only one seeing this bug reliably in the builder,
but it was easy to reproduce using 'GOGC=1 go test strconv'.
concatstring looked like it took only one string, but in fact it
takes a long list of strings. Add an explicit ... so that the traceback
will not use the "fixed" frame size and instead look at the
frame size metadata recorded by the caller.

R=golang-dev
TBR=golang-dev
CC=golang-dev
https://golang.org/cl/11531043

This should fix the Windows build, or at least
what's breaking it at the moment.

Fixes #5904.

TBR=golang-dev
CC=golang-dev
https://golang.org/cl/11519044

Would fix build if build hadn't already been broken with something else.

TBR=golang-dev
CC=golang-dev
https://golang.org/cl/11525043

R=golang-dev, r, dave
CC=golang-dev
https://golang.org/cl/11494043

The change to addr2line makes it easy to test by hand.

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/11485044

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/11495043

Be consistent with os.File, strings.Reader, bytes.Reader, etc,
which all allow seeks past the end.

R=golang-dev, r
CC=golang-dev
https://golang.org/cl/11403043

R=r, nigeltao
CC=golang-dev
https://golang.org/cl/11446043

Update #5905

Don't emit #pragma cgo_ldflag and cgo_import_static
declarations that gccgo doesn't understand and can't use.

R=golang-dev, rsc, bradfitz
CC=golang-dev
https://golang.org/cl/11490043

R=golang-dev, r, rsc
CC=golang-dev
https://golang.org/cl/11436043

R=golang-dev, dsymonds, rogpeppe
CC=golang-dev
https://golang.org/cl/11432044

Otherwise the tests in pkg/runtime fail:

runtime: unknown argument frame size for runtime.deferreturn called from 0x48657b [runtime_test.func·022]
fatal error: invalid stack
...

R=golang-dev, dave
CC=golang-dev
https://golang.org/cl/11483043

This will behave like similar "*-other-window" functions in Emacs.

Default key bind is C-x 4 C-c C-j – while awkward, it follows both
the convention for other-window functions and the convention for
not using user- or emacs-reserved keys.

R=golang-dev, adonovan
CC=golang-dev
https://golang.org/cl/10707045