Currently the linker figures out where runtime-gdb.py should be by
looking for the path to runtime/debug.go. However, debug.go contains
only a few symbols and can easily get dead-code eliminated entirely,
especially from simple binaries. When this happens, the resulting
binary lacks a reference to runtime-gdb.py, so the GDB helpers don't
get loaded.

Fix this by instead sniffing for runtime/proc.go. This contains
runtime.main and the scheduler, so it's not going anywhere.

Change-Id: Ie3380c77c840d28614fada68b8c5861625f2aff5
Reviewed-on: https://go-review.googlesource.com/68019
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Alex Brainman <alex.brainman@gmail.com>

All of the callers want a *obj.LSym instead of a *types.Sym, and the
runtime type descriptors don't need Go source symbols anyway.

Passes toolstash-check.

Change-Id: I8ae4b64380fbb547857f49b42465118f41884eed
Reviewed-on: https://go-review.googlesource.com/68251
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Marvin Stenger <marvin.stenger94@gmail.com>
Reviewed-by: David Crawshaw <crawshaw@golang.org>

Change-Id: Ib042e472e62f48a6afaba1762beaf102a9b99cf5
Reviewed-on: https://go-review.googlesource.com/68290Reviewed-by: Russ Cox <rsc@golang.org>

Before CL 62530 fastrand always returned non-zero value, and one
condition in sema.go depends on this behavior.

fastrand is used to generate random weight for treap of sudog, and
it is checked against zero to verify sudog were inserted into treap or
wait queue.

Since its precision is not very important for correctness, lets just
always set its lowest bit in this place.

Updates #22047
Updates #21806

Change-Id: Iba0b56d81054e6ef9c49ffd293fc5d92a6a31e9b
Reviewed-on: https://go-review.googlesource.com/68050Reviewed-by: Austin Clements <austin@google.com>
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

Change-Id: Ib42fabc6829b6033373c0378713733f88e73e73d
Reviewed-on: https://go-review.googlesource.com/68230Reviewed-by: Russ Cox <rsc@golang.org>

PlainAuth originally refused to send passwords to non-TLS servers
and was documented as such.

In 2013, issue #5184 was filed objecting to the TLS requirement,
despite the fact that it is spelled out clearly in RFC 4954.
The only possibly legitimate use case raised was using PLAIN auth
for connections to localhost, and the suggested fix was to let the
server decide: if it advertises that PLAIN auth is OK, believe it.
That approach was adopted in CL 8279043 and released in Go 1.1.

Unfortunately, this is exactly wrong. The whole point of the TLS
requirement is to make sure not to send the password to the wrong
server or to a man-in-the-middle. Instead of implementing this rule,
CL 8279043 blindly trusts the server, so that if a man-in-the-middle
says "it's OK, you can send me your password," PlainAuth does.
And the documentation was not updated to reflect any of this.

This CL restores the original TLS check, as required by RFC 4954
and as promised in the documentation for PlainAuth.
It then carves out a documented exception for connections made
to localhost (defined as "localhost", "127.0.0.1", or "::1").

Change-Id: I1d3729bbd33aa2f11a03f4c000e6bb473164957b
Reviewed-on: https://go-review.googlesource.com/68170
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: David Crawshaw <crawshaw@golang.org>

These are redundant with checking x.Sym.Pkg == localpkg.

Passes toolstash-check -all.

Change-Id: Iebe25f7932cd15a036141b468ad75c239abcdcf7
Reviewed-on: https://go-review.googlesource.com/66670
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>

Avoids generating useless equality and hash functions.

Shrinks cmd/go executable size on linux/amd64 by ~17kB.

Change-Id: Ifde5315cc5cbceb3a7260195c8803cace952359f
Reviewed-on: https://go-review.googlesource.com/66650
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>

This can only lead to confusion.

Change-Id: Iae84c6404ab5eeb6950faa2364f97a017c67c506
Reviewed-on: https://go-review.googlesource.com/68110
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Crawshaw <crawshaw@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Simon Rawet <Simon@rawet.se>

This removes some of the []byte/string conversions currently
existing in the (un)marshaling methods of Int and Rat.

For Int we introduce a new function (*Int).setFromScanner() essentially
implementing the SetString method being given an io.ByteScanner instead
of a string. So we can handle the string case in (*Int).SetString with
a *strings.Reader and the []byte case in (*Int).UnmarshalText() with a
*bytes.Reader now avoiding the []byte/string conversion here.

For Rat we introduce a new function (*Rat).marshal() essentially
implementing the String method outputting []byte instead of string.
Using this new function and the same formatting rules as in
(*Rat).RatString we can implement (*Rat).MarshalText() without
the []byte/string conversion it used to have.

Change-Id: Ic5ef246c1582c428a40f214b95a16671ef0a06d9
Reviewed-on: https://go-review.googlesource.com/65950Reviewed-by: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

It was only waiting 0.1 seconds for the two GCs it wanted.
Let it wait 1 second.

Change-Id: Ib3cdc8127cbf95694a9f173643c02529a85063af
Reviewed-on: https://go-review.googlesource.com/68151
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

Before this change the ct == 0 check could never be true. Moreover the
values were not properly indirected.

Change-Id: Ice47e36e3492babc4b47d2f9099e8772be231c96
Reviewed-on: https://go-review.googlesource.com/68130Reviewed-by: Daniel Theophanes <kardianos@gmail.com>
Run-TryBot: Daniel Theophanes <kardianos@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

Preparation for moving the object file reader to its own package.

For #22095

Change-Id: I31fe4a10a2c465f8ea4bf548f40918807e4ec6b5
Reviewed-on: https://go-review.googlesource.com/67314
Run-TryBot: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Fix incorrect reference to string instead of byte slice.

Change-Id: I95553da32acfbcf5dde9613b07ea38408cb31ae8
Reviewed-on: https://go-review.googlesource.com/68090Reviewed-by: Ian Lance Taylor <iant@golang.org>

Also reduce the passed context from *Link to *sys.Arch, so fewer
data dependencies need to be wired through all the code dealing
with symbols.

For #22095

Change-Id: I50969405d6562c5152bd1a3c443b72413e9b70bc
Reviewed-on: https://go-review.googlesource.com/67313
Run-TryBot: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Update golang/go#22120

Change-Id: Ie7dbd0e7b01b116c960243a8cd3fa9fd121f317d
Reviewed-on: https://go-review.googlesource.com/68021Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Dave Cheney <dave@cheney.net>

All current darwin architectures seem to take at least 100ms to dial a closed port,
and that was making the all.bash script fail.

Fixes #22062

Change-Id: Ib79c4b7a5db2373c95ce5d993cdcbee55cc0667f
Reviewed-on: https://go-review.googlesource.com/67350Reviewed-by: Ian Lance Taylor <iant@golang.org>

Fixes #19418

Change-Id: I98205f40c1915cd68a5d20438469ba06f1efb160
Reviewed-on: https://go-review.googlesource.com/67432
Run-TryBot: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Begin passing coutbuf by as a parameter. To make the initial plumbing
pass easier, it is also a field in the standard ctxt parameter.

Consolidate the byte writing functions into the OutBuf object.
The result is less architecture-dependent initialization.

To avoid plumbing out everywhere we want to report an error, move
handling of out file deletion to an AtExit function.

For #22095

Change-Id: I0863695241562e0662ae3669666c7922b8c846f9
Reviewed-on: https://go-review.googlesource.com/67318
Run-TryBot: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

For #22095

Change-Id: I9d1f0d93f8fd701a24af826dc903eea2bc235de2
Reviewed-on: https://go-review.googlesource.com/67317
Run-TryBot: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

No need to type this global as an unsafe.Pointer, we know
what type the referent is.

Change-Id: I7b1374065b53ccf1373754a21d54adbedf1fd587
Reviewed-on: https://go-review.googlesource.com/67990Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Reviewed-by: David Crawshaw <crawshaw@golang.org>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

Latest macOS High Sierra changed how the wall time information
is exported into the commpage. Backward compatibility was partly
preserved, that is previous Go versions are basically forced to
go through a syscall which is much slower and is not able to
get nanosecond precision.

Implement the new commpage layout and wall time computation,
using a version check to fallback to the previous code on
older operating systems.

Fixes #22037

Change-Id: I8c2176eaca83a5d7be23443946a6b4c653ec7f68
Reviewed-on: https://go-review.googlesource.com/67332
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

cmd/cgo is special among the build tools because it is (re)built on demand
when needed for a package using cgo, to avoid additional bootstrap logic
elsewhere. (This is in contrast to cmd/compile, cmd/link, and so on, which
must be specially built before even invoking the go command.)

When the go command starts using content-based decisions about staleness,
it is important that the build of cmd/cgo never use -linkmode=external,
because that depends on runtime/cgo, which in turn depends on cmd/cgo.

Change-Id: I72a2be748606d1ed4b93a54f2a5c7084e87d5fbc
Reviewed-on: https://go-review.googlesource.com/67310Reviewed-by: Ian Lance Taylor <iant@golang.org>

Change-Id: Iab8f0d201780bd571541a6806f071e883a553d35
Reviewed-on: https://go-review.googlesource.com/56286Reviewed-by: David Crawshaw <crawshaw@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Dave Cheney <dave@cheney.net>

This is an intermediate step toward not being able to predict
the final generated file name for a package build, so that
parent builds can refer directly to cache files.

Change-Id: I4dea5e8d8b80e6b995b3d9dc1d8c6f0ac9b88d72
Reviewed-on: https://go-review.googlesource.com/56285Reviewed-by: David Crawshaw <crawshaw@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

These are rarely used and can be computed on demand,
to make clear that they are never out of sync with the
lists in the non-internal Package fields.

Change-Id: I8c621dceaff1aeb39a3ed83f18e848adf14d7106
Reviewed-on: https://go-review.googlesource.com/56284Reviewed-by: David Crawshaw <crawshaw@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Actually execute topological sort to get those special dependencies right.

Mistake introduced in CL 67650.

Change-Id: I22fd6efb4f033deaf7f191431c0401b59b8a97d0
Reviewed-on: https://go-review.googlesource.com/67870
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

When variables need to be spilled to the stack, they usually get their
own stack slot. Local variables have a slot allocated if they need one,
and arguments start out on the stack. Before this CL, the debug
information made the assumption that this was always the case, and so
didn't bother storing an actual stack offset during SSA analysis.

There's at least one case where this isn't true: variables that alias
arguments. Since the argument is the source of the variable, the
variable will begin its life on the stack in the argument's stack slot,
not its own. Therefore the debug info needs to track the actual stack
slot for each location entry.

No detectable performance change, despite the O(N) loop in getHomeSlot.

Change-Id: I2701adb7eddee17d4524336cb7aa6786e8f32b46
Reviewed-on: https://go-review.googlesource.com/67231Reviewed-by: Alessandro Arzilli <alessandro.arzilli@gmail.com>
Reviewed-by: David Chase <drchase@google.com>

Some tests in misc/cgo/test are run with various options including
'-linkmode=external "-extldflags=-pie"'. On ppc64x passing -pie to
the external linker with code that was not compiled as position
independent is incorrect. This works by luck in many cases but is
not guaranteed to work. I suspect it is an issue on other targets
as well.

This will now run the tests using -buildmode=pie for the platforms
that support that buildmode option.

Fixes #21954

Change-Id: I25fc7573f2d3cb5b0d1c691a0ac91aef7715404f
Reviewed-on: https://go-review.googlesource.com/66870
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Previously, only the empty interface could be formatted to print on a
single line. This behaviour made short one-method interfaces in function
definitions and type assertions more verbose than they had to be.

For example, the following type assertion:

    if c, ok := v.(interface {
        Close() error
    }); ok {
    }

Can now be formatted as:

    if c, ok := v.(interface{ Close() error }); ok {
    }

Fixes #21952

Change-Id: I896f796c5a30b9f4da2be3fe67cb6fea5871b835
Reviewed-on: https://go-review.googlesource.com/66130
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>

s/TypeSymLink/TypeSymlink/g

Change-Id: I2550843248eb27d90684d0036fe2add0b247ae5a
Reviewed-on: https://go-review.googlesource.com/67810Reviewed-by: Ian Lance Taylor <iant@golang.org>

Apply gofmt to src, misc.

Fixes #22111.

Change-Id: Ib1bda0caaf2c1787a8137b7a61bbef7a341cc68c
Reviewed-on: https://go-review.googlesource.com/67633
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

High Sierra has a new commpage layout (this is issue #3188), so
we need to adjust the code to handle multiple versions of the
layout.

In preparation for this change, we rename the existing offset
macros with a prefix that identifies the commpage version they
refer to.

Updates #22037

Change-Id: Idca4b7a855a2ff6dbc434cd12453fc3194707aa8
Reviewed-on: https://go-review.googlesource.com/67331
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Loading location data from tzdata files was only supported
from default paths on android. This change enables support on
all OS via the ZONEINFO environment variable and reduces the
amount of android specific code significantly.
Furthermore, unsuccessful calls to LoadLocation now return the
first error encountered, including errors from attempting to
load a location from the source specified by ZONEINFO.
Errors indicating that the source or location was not found are
ignored until all possible sources have been traversed.

Change-Id: I45bc23b92253c9447f12f95f3ca29a7e613ed995
Reviewed-on: https://go-review.googlesource.com/67170Reviewed-by: Ian Lance Taylor <iant@golang.org>

Fixes #22077

Change-Id: I8d1b18e78719514fa2a410e85c9d3934c47cd2fe
Reviewed-on: https://go-review.googlesource.com/66930Reviewed-by: Joe Tsai <thebrokentoaster@gmail.com>

Fixes #21159.

Change-Id: I2c5ad505d673e213a548e5d632a5b3ad706e0dde
Reviewed-on: https://go-review.googlesource.com/67635
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

The lldb child process is killed if a test runs too long. Also
kill it when the setup times out (and is retried).

Might help with builder flakes where all 5 attempts to start up
lldb fail even though the tests before and after the timeouts
succeed. For example:

...
ok  	vendor/golang_org/x/net/route	37.797s
lldb setup error: command timeout (lldb start for 17s)
start timeout, trying again
lldb setup error: command timeout (lldb start for 17s)
start timeout, trying again
lldb setup error: command timeout (lldb start for 17s)
start timeout, trying again
lldb setup error: command timeout (lldb start for 17s)
start timeout, trying again
lldb setup error: command timeout (lldb start for 17s)
go_darwin_arm_exec: failed to start test harness (retry attempted)
FAIL	vendor/golang_org/x/text/transform	115.185s
ok  	vendor/golang_org/x/text/unicode/norm	122.773s
...

Change-Id: I6638860522896491dccfa12f1e520c0f23df6d66
Reviewed-on: https://go-review.googlesource.com/67791
Run-TryBot: Elias Naur <elias.naur@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

The point of this code is to print a warning about repeated go test
invocations rebuilding the same packages over and over.
The new cache will eliminate this failure mode and with it
the need for the warning and this field.

Change-Id: Ied79b3ca67d51a61f44629de6ae4974e6c8dd5a9
Reviewed-on: https://go-review.googlesource.com/56282Reviewed-by: David Crawshaw <crawshaw@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Mistake introduced just before submitting CL 67650 that somehow
did not break a final pre-submit all.bash on my laptop.

Not sure why all.bash passes locally when mkdeps.go doesn't build.
I guess the test only runs on builders?

Change-Id: I18fb91ada47605035345ba4b2f5e360a5c4b7f6e
Reviewed-on: https://go-review.googlesource.com/67850
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

In addition to the obvious formatting change, this also drops
from deps.go any indirect dependencies, so that when you add
a new import to one package, the resulting diff only affects that
one package, not every package that imports that package
directly or indirectly. That makes the file a bit easier to understand,
if you need to debug it or deal with a possible merge conflict.

The code to trim the import lists (but not too much) was more
than I wanted to do in shell, so I rewrote mkdeps in Go.

The shell script is left behind for backwards-compatibility with
people who have learned to run ./mkdeps.bash (or documentation
suggesting the same).

Change-Id: I0bf27b5b27d0440e11ea830b00735c73f58eae03
Reviewed-on: https://go-review.googlesource.com/67650Reviewed-by: David Crawshaw <crawshaw@golang.org>