This change is the first of a series of changes which replace the
current page allocator (which is based on the contents of mgclarge.go
and some of mheap.go) with one based on free/used bitmaps.
It adds in the key constants for the page allocator as well as a comment
describing the implementation.
Updates #35112.
Change-Id: I839d3a07f46842ad379701d27aa691885afdba63
Reviewed-on: https://go-review.googlesource.com/c/go/+/190619
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
zip.Create is now used to filter and translate zip files from VCS tools.
zip.Unzip is now used instead of Unzip.
Fixes#35290
Change-Id: I4aa41b2e96bf147c09db43d1d189b8393cafb06f
Reviewed-on: https://go-review.googlesource.com/c/go/+/204917
Run-TryBot: Jay Conrod <jayconrod@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
This changes makes it so that sysReserve, which creates a PROT_NONE
mapping, maps that memory as NORESERVE. Before this change, relatively
large PROT_NONE mappings could cause fork to fail with ENOMEM, reported
as "not enough space". Presumably this refers to swap space, since
adding this flag causes the failures to go away.
This helps unblock page allocator work, since it allows us to make large
PROT_NONE mappings on solaris safely.
Updates #35112.
Change-Id: Ic3cba310c626e93d5db0f27269e2569bb7bc393e
Reviewed-on: https://go-review.googlesource.com/c/go/+/205759
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
This CL adds a new test package which downloads specific versions of
~1000 modules in direct mode and verifies that modules have the same
sums and the zip files have the same SHA-256 hashes.
This test takes a long time to run and depends heavily on external
data that may disappear. It must be enabled manually with -zipsum.
Fixes#35290
Change-Id: Ic6959e685096e8b09cea291f19d5bd0255432284
Reviewed-on: https://go-review.googlesource.com/c/go/+/204838
Reviewed-by: Bryan C. Mills <bcmills@google.com>
This makes it a little less likely the portable FMA will be
broken without realizing it.
Change-Id: I7f7f4509b35160a9709f8b8a0e494c09ea6e410a
Reviewed-on: https://go-review.googlesource.com/c/go/+/205337
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
This API was added for #25819, where it was discussed as math.FMA.
The commit adding it used math.Fma, presumably for consistency
with the rest of the unusual names in package math
(Sincos, Acosh, Erfcinv, Float32bits, etc).
I believe that using an idiomatic Go name is more important here
than consistency with these other names, most of which are historical
baggage from C's standard library.
Early additions like Float32frombits happened before "uppercase for export"
(so they were originally like "float32frombits") and they were not properly
reconsidered when we uppercased the symbols to export them.
That's a mistake we live with.
The names of functions we have added since then, and even a few
that were legacy, are more properly Go-cased, such as IsNaN, IsInf,
and RoundToEven, rather than Isnan, Isinf, and Roundtoeven.
And also constants like MaxFloat32.
For new API, we should keep using proper Go-cased symbols
instead of minimally-upper-cased-C symbols.
So math.FMA, not math.Fma.
This API has not yet been released, so this change does not break
the compatibility promise.
This CL also modifies cmd/compile, since the compiler knows
the name of the function. I could have stopped at changing the
string constants, but it seemed to make more sense to use a
consistent casing everywhere.
Change-Id: I0f6f3407f41e99bfa8239467345c33945088896e
Reviewed-on: https://go-review.googlesource.com/c/go/+/205317
Run-TryBot: Russ Cox <rsc@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
This CL was verified by running:
go test -gcflags=all=-d=checkptr=2 internal/syscall/windows
internal/syscall/windows.TestRunAtLowIntegrity uses code in question.
Updates #34972
Change-Id: I434530058e2d41f132e9bf154e8c64c03894e9c4
Reviewed-on: https://go-review.googlesource.com/c/go/+/204117
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
No need to check as pieces further down do so anyway:
% go doc '&&.%$^'
doc: symbol && is not a type in package fmt installed in "fmt"
exit status 1
%
Removing this check allows 'go doc sort.interface' or 'go doc
types.type' to discover sort.Interface and go/types.Type.
Easily
Fixes#34656.
Change-Id: I84352e83dd7f91a232f45a44d1a52f019a1a9a06
Reviewed-on: https://go-review.googlesource.com/c/go/+/205778
Reviewed-by: Caleb Spare <cespare@gmail.com>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
Allow the inputs a and b to be zero or negative to GCD
with the following definitions.
If x or y are not nil, GCD sets their value such that z = a*x + b*y.
Regardless of the signs of a and b, z is always >= 0.
If a == b == 0, GCD sets z = x = y = 0.
If a == 0 and b != 0, GCD sets z = |b|, x = 0, y = sign(b) * 1.
If a != 0 and b == 0, GCD sets z = |a|, x = sign(a) * 1, y = 0.
Fixes#28878
Change-Id: Ia83fce66912a96545c95cd8df0549bfd852652f3
Reviewed-on: https://go-review.googlesource.com/c/go/+/164972
Run-TryBot: Brian Kessler <brian.m.kessler@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
When we have already assigned the semaphore ticket to a specific
waiter, we want to get the waiter running as fast as possible since
no other G waiting on the semaphore can acquire it optimistically.
The net effect is that, when a sync.Mutex is contented, the code in
the critical section guarded by the Mutex gets a priority boost.
Fixes#33747
Change-Id: I9967f0f763c25504010651bdd7f944ee0189cd45
Reviewed-on: https://go-review.googlesource.com/c/go/+/200577
Reviewed-by: Rhys Hiltner <rhys@justin.tv>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
On iOS, the address space is not 48 bits as one might believe, since
it's arm64 hardware. In fact, all pointers are truncated to 33 bits, and
the OS only gives applications access to the range [1<<32, 2<<32).
While today this has no effect on the Go runtime, future changes which
care about address space size need this to be correct.
Updates #35112.
Change-Id: Id518a2298080f7e3d31cf7d909506a37748cc49a
Reviewed-on: https://go-review.googlesource.com/c/go/+/205758
Run-TryBot: Michael Knyszek <mknyszek@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
This change removes a hack which was added to deal with Darwin 10.10's
weird ignorance of mapping hints which would cause race mode to fail
since it requires the heap to live within a certain address range.
We no longer support 10.10, and this is potentially causing problems
related to the page allocator, so drop this code.
Updates #26475.
Updates #35112.
Change-Id: I0e1c6f8c924afe715a2aceb659a969d7c7b6f749
Reviewed-on: https://go-review.googlesource.com/c/go/+/205757
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
The test deliberately constructs an invalid pointer, so don't check it.
Fixes#35379
Change-Id: Ifeff3484740786b0470de3a4d2d4103d91e06f5d
Reviewed-on: https://go-review.googlesource.com/c/go/+/205717
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
This change is based on the previous discussion in CL 202442.
Fixes#34634
Change-Id: I1319aa26d5cfcd034bc576555787b3ca79968c38
Reviewed-on: https://go-review.googlesource.com/c/go/+/205637
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
This change employs the same strategy as in CL 203017
to detect when vendoring is in use, and if so treats
the vendor directory as a (non-module, prefixless) root.
The integration test also verifies that the 'std' and 'cmd'
modules are included and their vendored dependencies are
visible (as they are with 'go list') even when outside of
those modules.
Fixes#35224
Change-Id: I18cd01218e9eb97c1fc6e2401c1907536b0b95f7
Reviewed-on: https://go-review.googlesource.com/c/go/+/205577
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Factor out the direct CALL identification code from objabi.IsDirectJump and
use this in two places that have separately maintained lists of reloc types.
Provide an objabi.IsDirectCallOrJump function that implements the original
behaviour of objabi.IsDirectJump.
Change-Id: I48131bae92b2938fd7822110d53df0b4ffb35766
Reviewed-on: https://go-review.googlesource.com/c/go/+/196577
Run-TryBot: Joel Sing <joel@sing.id.au>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Don't print to stdout in non-verbose (-v) test mode.
Exception: Timing output (2 lines) of TestStdLib. If
we want to disable that as well we should use another
flag to differenciate between -verbose output and
measurement results. Leaving alone for now.
Fixes#35223.
Change-Id: Ie8160760e8db1138f9031888d654eaeab202128c
Reviewed-on: https://go-review.googlesource.com/c/go/+/204039
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
This CL adds basic encode test for mips64x and
most of the instructions are cross checked with 'gas'
Update #35008
Change-Id: I18bb524897aa745bfe23db43fcbb44c3b009463c
Reviewed-on: https://go-review.googlesource.com/c/go/+/204297
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Otherwise, we can get into a deadlock: sysmon takes the scheduler lock
and calls timeSleepUntil which takes each P's timer lock. Simultaneously,
some P calls runtimer (holding the P's own timer lock) which wakes up
the scavenger, calling goready, calling wakep, calling startm, getting
the scheduler lock. Now the sysmon thread is holding the scheduler lock
and trying to get a P's timer lock, while some other thread running on
that P is holding the P's timer lock and trying to get the scheduler lock.
So change sysmon to call timeSleepUntil without holding the scheduler
lock, and change timeSleepUntil to use allpLock, which is only held for
limited periods of time and should never compete with timer locks.
This hopefully
Fixes#35375
At least it should fix the linux-arm64-packet builder problems,
which occurred more reliably as that system has GOMAXPROCS == 96,
giving a lot more scope for this deadlock.
Change-Id: I7a7917daf7a4882e0b27ca416e4f6300cfaaa774
Reviewed-on: https://go-review.googlesource.com/c/go/+/205558
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
The -Wl,-headerpad, -Wl,-no_pie, -Wl,-pagezero_size flags are
incompatible with the bitcode-related flags used for iOS.
We already omitted the flags on darwin/arm and darwin/arm64; this change
omits the flags on all platforms != macOS so that building for the iOS
simulator works.
Updates #32963
Change-Id: Ic9af0daf01608f5ae0f70858e3045e399de7e95b
Reviewed-on: https://go-review.googlesource.com/c/go/+/205340
Run-TryBot: Elias Naur <mail@eliasnaur.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
WaitForSigusr1 registers a callback to be called on SIGUSR1 directly
from the runtime signal handler. Currently, this callback has a write
barrier in it, which can crash with a nil P if the GC is active and
the signal arrives on an M that doesn't have a P.
Fix this by recording the ID of the M that receives the signal instead
of the M itself, since that's all we needed anyway. To make sure there
are no other problems, this also lifts the callback into a package
function and marks it "go:nowritebarrierrec".
Fixes#35248.
Updates #35276, since in principle a write barrier at exactly the
wrong time while entering the scheduler could cause issues, though I
suspect that bug is unrelated.
Change-Id: I47b4bc73782efbb613785a93e381d8aaf6850826
Reviewed-on: https://go-review.googlesource.com/c/go/+/204620
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
token.IsExported expects to be passed a token, and does not check for
non-token arguments such as "C:\workdir\go\src\text".
While we're at it, clean up a few other parts of the code that
are assuming a package path where a directory may be passed instead.
There are probably others lurking around here, but I believe this
change is sufficient to get past the test failures on the
windows-amd64-longtest builder.
Fixes#35236
Change-Id: Ic79fa035531ca0777f64b1446c2f9237397b1bdf
Reviewed-on: https://go-review.googlesource.com/c/go/+/204442
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Rob Pike <r@golang.org>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
With buildmode=c-archive, "runtime.types" type isn't STYPE but
STYPERELRO.
On AIX, this symbol is present in the symbol table and not under
typerel.* outersymbol. Therefore, the size of typerel.* must be adapted.
Fixes#35342
Change-Id: Ib982c6557d9b41bc3d8775e4825650897f9e0ee6
Reviewed-on: https://go-review.googlesource.com/c/go/+/205338
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
We seem to lack any tests for some corner cases of itab.init
(multiple methods with the same name, breaking itab.init doesn't
seem to fail any tests). We also lack tests that fix text of panics.
Add more tests for itab.init.
Change-Id: Id6b536179ba6b0d45c3cb9dc1c66b9311d0ab85e
Reviewed-on: https://go-review.googlesource.com/c/go/+/202451
Run-TryBot: Dmitry Vyukov <dvyukov@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
The check is not relevant anymore.
The comment claims that go run does not rebuild packages,
but this is not true. And we use go build anyway.
We may have added the check because without caching
rebuilding everything starting from runtime for each test
takes a while. But now we have caching.
So from every side this check just adds code and pain.
Change-Id: Ifbbb643724100622e5f9db884339b67cde4ba729
Reviewed-on: https://go-review.googlesource.com/c/go/+/202450
Run-TryBot: Dmitry Vyukov <dvyukov@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
The hash is used in type switches. However, compiler statically generates itab's
for all interface/type pairs used in switches (which are added to itabTable
in itabsinit). The dynamically-generated itab's never participate in type switches,
and thus the hash is irrelevant.
Change-Id: I4f6e37be31b8f5605cca7a1806cb04708e948cea
Reviewed-on: https://go-review.googlesource.com/c/go/+/202448
Run-TryBot: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Previously we would always “upgrade” to the semantically-highest
version, even if a newer compatible version exists.
That made certain classes of mistakes irreversible: in general we
expect users to address bad releases by releasing a new (higher)
version, but if the bad release was an unintended +incompatible
version, then no release that includes a go.mod file can ever have a
higher version, and the bad release will be treated as “latest”
forever.
Instead, when considering a +incompatible version we now consult the
latest compatible (v0 or v1) release first. If the compatible release
contains a go.mod file, we ignore the +incompatible releases unless
they are expicitly requested (by version, commit ID, or branch name).
Fixes#34165
Updates #34189
Change-Id: I7301eb963bbb91b21d3b96a577644221ed988ab7
Reviewed-on: https://go-review.googlesource.com/c/go/+/204440
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Jay Conrod <jayconrod@google.com>
Also revert an incidental 'gofmt' of a vendored file from CL 205240.
Updates #34822
Change-Id: I82a015d865db4d865b4776a8013312f25dbb9181
Reviewed-on: https://go-review.googlesource.com/c/go/+/205539
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
codeRepo.Versions previously checked every possible +incompatible
version for a 'go.mod' file. That is wasteful and counterproductive.
It is wasteful because typically, a project will adopt modules at some
major version, after which they will (be required to) use semantic
import paths for future major versions.
It is counterproductive because it causes an accidental
'+incompatible' tag to exist, and no compatible tag can have higher
semantic precedence.
This change prunes out some of the +incompatible versions in
codeRepo.Versions, eliminating the “wasteful” part but not all of the
“counterproductive” part: the extraneous versions can still be fetched
explicitly, and proxies may include them in the @v/list endpoint.
Updates #34165
Updates #34189
Updates #34533
Change-Id: Ifc52c725aa396f7fde2afc727d0d5950acd06946
Reviewed-on: https://go-review.googlesource.com/c/go/+/204439
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Jay Conrod <jayconrod@google.com>
We had some issues with reports being marked as spam, so I added a
filter to never mark as spam something that mentions the word
"vulnerability". We get too much spam at that address to disable the
filter entirely, so instead meantion the bypass in the docs.
Change-Id: Idb4dabcf51a9dd8234a2d571cd020c970b0a582c
Reviewed-on: https://go-review.googlesource.com/c/go/+/205538
Reviewed-by: Katie Hockman <katie@golang.org>
dsa.Verify might currently use a nil s inverse in a
multiplication if the public key contains a non-prime Q,
causing a panic. Change this to check that the mod
inverse exists before using it.
Fixes CVE-2019-17596
Fixes#34960
Change-Id: I94d5f3cc38f1b5d52d38dcb1d253c71b7fd1cae7
Reviewed-on: https://team-review.git.corp.google.com/c/golang/go-private/+/572809
Reviewed-by: Filippo Valsorda <valsorda@google.com>
Reviewed-on: https://go-review.googlesource.com/c/go/+/205441
Run-TryBot: Katie Hockman <katie@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Filippo Valsorda <filippo@golang.org>
In the dev.link branch we implemented the new object file format
and (part of) the linker improvements described in
https://golang.org/s/better-linker
The new object file is index-based and provides random access.
The linker maps the object files into read-only memory, and
access symbols on-demand using indices, as opposed to reading
all object files sequentially into the heap with the old format.
The linker carries symbol informations using indices (as opposed
to Symbol data structure). Symbols are created after the
reachability analysis, and only created for reachable symbols.
This reduces the linker's memory usage.
Linking cmd/compile, it creates ~25% fewer Symbols, and reduces
memory usage (inuse_space) by ~15%. (More results from Than.)
Currently, both the old and new object file formats are supported.
The old format is used by default. The new format can be turned
on by using the compiler/assembler/linker's -newobj flag. Note
that the flag needs to be specified consistently to all
compilations, i.e.
go build -gcflags=all=-newobj -asmflags=all=-newobj -ldflags=-newobj
Change-Id: Ia0e35306b5b9b5b19fdc7fa7c602d4ce36fa6abd
Flip back to the old object files for Go 1.14.
Change-Id: I4ad499460fb7156b63fc63e9c6ea4f7099e20af2
Reviewed-on: https://go-review.googlesource.com/c/go/+/204098
Run-TryBot: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
When using cgo, we save G to TLS, and when a signal happens, we
load G from TLS in sigtramp. This should give us a valid G. Don't
try to fetch from the signal stack. In particular, C code may
change the signal stack or call our signal handler directly (e.g.
TSAN), so we are not necessarily running on the original gsignal
stack where we saved G.
Also skip saving G on the signal stack when using cgo.
Updates #35249.
Change-Id: I40749ce6682709bd4ebfdfd9f23bd0f317fc197d
Reviewed-on: https://go-review.googlesource.com/c/go/+/204519
Reviewed-by: Ian Lance Taylor <iant@golang.org>
In the normal case, sigFetchG just returns the G register. But in
the case that sigFetchG fetches the G from somewhere else, the G
register still holding an invalid value. Setg here to make sure
they match.
This is particularly useful because setGsignalStack, called by
adjustSignalStack from sigtrampgo before setg to gsignal,
accesses the G register.
Should fix#35249.
Change-Id: I64c85143cb05cdb2ecca7f9936dbd8bfec186c2d
Reviewed-on: https://go-review.googlesource.com/c/go/+/204441
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Before this CL adjustTimers left timers being moved in an inconsistent
state: status timerWaiting but not on a P. Simplify the code by
leaving the timers in timerMoving status until they are actually moved.
Other functions (deltimer, modtimer) will wait until the move is complete
before changing anything on the timer. This does leave timers in timerMoving
state for longer, but still not all that long.
Fixes#35367
Change-Id: I31851002fb4053bd6914139125b4c82a68bf6fb2
Reviewed-on: https://go-review.googlesource.com/c/go/+/205418
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Michael Knyszek <mknyszek@google.com>
TestFormats adds ~3s of running time to the test, which may be
slightly annoying in an edit/compile/test cycle but is negligible in a
TryBot run.
The test keeps regressing in the longtest builders, requiring a manual
fix. Instead, run it even in short mode on the builders, so that
TryBot runs will detect regressions ahead of time.
Updates #34907
Updates #33915
Updates #28621
Change-Id: I6f9bf0f2ca929a743438310b86d85d8673c720bf
Reviewed-on: https://go-review.googlesource.com/c/go/+/205440
Run-TryBot: Bryan C. Mills <bcmills@google.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Once defined, a stack slot holding an open-coded defer arg should always be marked
live, since it may be used at any time if there is a panic. These stack slots are
typically kept live naturally by the open-defer code inlined at each return/exit point.
However, we need to do extra work to make sure that they are kept live if a
function has an infinite loop or a panic exit.
For this fix, only in the case of a function that is using open-coded defers, we
compute the set of blocks (most often empty) that cannot reach a return or a
BlockExit (panic) because of an infinite loop. Then, for each block b which
cannot reach a return or BlockExit or is a BlockExit block, we mark each defer arg
slot as live, as long as the definition of the defer arg slot dominates block b.
For this change, had to export (*Func).sdom (-> Sdom) and SparseTree.isAncestorEq
(-> IsAncestorEq)
Updates #35277
Change-Id: I7b53c9bd38ba384a3794386dd0eb94e4cbde4eb1
Reviewed-on: https://go-review.googlesource.com/c/go/+/204802
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
