We want to issue loads as soon as possible, especially when they
are going to miss in the cache. Using a conditional move (CMOV) here:

i := ...
if cond {
   i++
}
... = a[i]

means that we have to wait for cond to be computed before the load
is issued. Without a CMOV, if the branch is predicted correctly the
load can be issued in parallel with computing cond.
Even if the branch is predicted incorrectly, maybe the speculative
load is close to the real load, and we get a prefetch for free.
In the worst case, when the prediction is wrong and the address is
way off, we only lose by the time difference between the CMOV
latency (~2 cycles) and the mispredict restart latency (~15 cycles).

We only squash CMOVs that affect load addresses. Results of CMOVs
that are used for other things (store addresses, store values) we
use as before.

Fixes #26306

Change-Id: I82ca14b664bf05e1d45e58de8c4d9c775a127ca1
Reviewed-on: https://go-review.googlesource.com/c/145717
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>

This commit fixes a mistake made in CL 144538.
This nilcheck can be removed because OpPPC64LoweredMove will fault if
arg0 is nil, as it's used to store. Further information can be found in
cmd/compile/internal/ssa/nilcheck.go.

Change-Id: Ifec0080c00eb1f94a8c02f8bf60b93308e71b119
Reviewed-on: https://go-review.googlesource.com/c/151298
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

After CL 151139 introduced a plugin test, the macOS linker for iOS
outputs:

ld: warning: -flat_namespace is deprecated on iOS

Omit the -flat_namespace flag on iOS; plugins are not supported on
iOS, and unlikely to ever be.

Change-Id: I2d08f8b984efcfd442d572b4a0f3a2c95c551b9f
Reviewed-on: https://go-review.googlesource.com/c/151300
Run-TryBot: Elias Naur <elias.naur@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

TestImportTableInUnknownSection uses kernel32.dll file, but the error
message mentions atmfd.dll. Adjust error message to match the test.

This change should have been part of CL 151137.

Updates #27904

Change-Id: Ifc31a12134b328472191122f8426ab6ed234fbd4
Reviewed-on: https://go-review.googlesource.com/c/151477
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>

When building windows/386 executable that imports "plugin" package,
cmd/link adds reference to DLL with blank name. Running

objdump -x a.exe

reports

...
The Import Tables (interpreted .idata section contents)
...
DLL Name:
vma:  Hint/Ord Member-Name Bound-To
25308a     0  _GLOBAL_OFFSET_TABLE_
...

So, obviously, executable cannot run, because Windows complains
that it cannot find DLL when trying to run it.

Stop using _GLOBAL_OFFSET_TABLE_ on windows/386.

Fixes #28789

Change-Id: Idd489eafd998f6e329f40c5d90a2a8965ab1d873
Reviewed-on: https://go-review.googlesource.com/c/151139
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

We were using the mode reported by ReadDir to decide whether each
entry is a file, but in the case of symlinks that isn't sufficient: a
symlink could point to either a file or a directory, and if it is a
file we should treat it as such.

Fixes #28107

Change-Id: Icf6e495dce427a7b1124c9cc9f085e40a215c169
Reviewed-on: https://go-review.googlesource.com/c/141097
Run-TryBot: Bryan C. Mills <bcmills@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Russ Cox <rsc@golang.org>

Div panics when y<=hi because either the quotient overflows
the size of the output or division by zero occurs when y==0.
This provides a uniform behavior for all implementations.

Fixes #28316

Change-Id: If23aeb10e0709ee1a60b7d614afc9103d674a980
Reviewed-on: https://go-review.googlesource.com/c/149517Reviewed-by: Robert Griesemer <gri@golang.org>

Note that the intrinsic implementation panics separately for overflow and
divide by zero, which matches the behavior of the pure go implementation.
There is a modest performance improvement after intrinsic implementation.

name     old time/op  new time/op  delta
Div-4    53.0ns ± 1%  47.0ns ± 0%  -11.28%  (p=0.008 n=5+5)
Div32-4  18.4ns ± 0%  18.5ns ± 1%     ~     (p=0.444 n=5+5)
Div64-4  53.3ns ± 0%  47.5ns ± 4%  -10.77%  (p=0.008 n=5+5)

Updates #28273

Change-Id: Ic1688ecc0964acace2e91bf44ef16f5fb6b6bc82
Reviewed-on: https://go-review.googlesource.com/c/144378
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

Explicitly check for divide-by-zero/overflow and panic with the appropriate
runtime error.  The additional checks have basically no effect on performance
since the branch is easily predicted.

name     old time/op  new time/op  delta
Div-4    53.9ns ± 1%  53.0ns ± 1%  -1.59%  (p=0.016 n=4+5)
Div32-4  17.9ns ± 0%  18.4ns ± 0%  +2.56%  (p=0.008 n=5+5)
Div64-4  53.5ns ± 0%  53.3ns ± 0%    ~     (p=0.095 n=5+5)

Updates #28316

Change-Id: I36297ee9946cbbc57fefb44d1730283b049ecf57
Reviewed-on: https://go-review.googlesource.com/c/144377
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

Don't convert values that aren't Go constants, like
uintptr(unsafe.Pointer(nil)), to a literal constant. This avoids
assuming they are constants for things like indexing, array sizes,
case duplication, etc.

Also, nil is an allowed duplicate in switches. CTNILs aren't Go constants.

Fixes #28078
Fixes #28079

Change-Id: I9ab8af47098651ea09ef10481787eae2ae2fb445
Reviewed-on: https://go-review.googlesource.com/c/151320
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

XCOFF files can't have multiples text or data sections. The name
of each type section must be .text, .data and .bss.

This commit also updates cmd/internal/objfile/xcoff.go to retrieve Go
sections using runtime symbols.

Change-Id: Ib6315f19dad2d154a4531fc6508e7cbd8bc94743
Reviewed-on: https://go-review.googlesource.com/c/151037
Run-TryBot: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

Fixes #28955

Change-Id: I738ad0c76f7bf8fc504a48cf55d3becd5ed7a9d6
Reviewed-on: https://go-review.googlesource.com/c/151337Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

On XCOFF, it is forbidden relocation of a DATA pointer to a text
section. It happens when a RODATA symbol needs a DATA symbol's address.
This commit moves every RODATA symbols with a R_ADDR on a data symbol to
.data sections to avoid these relocations.

Change-Id: I7f34d8e0ebdc8352a74e6b40e4c893d8d9419f4d
Reviewed-on: https://go-review.googlesource.com/c/146977Reviewed-by: Ian Lance Taylor <iant@golang.org>

When a slice composite literal is sparse, initialize it dynamically
instead of statically.

s := []int{5:5, 20:20}

To initialize the backing store for s, use 2 constant writes instead
of copying from a static array with 21 entries.

This CL also fixes pathologies in the compiler when the slice is
*very* sparse.

Fixes #23780

Change-Id: Iae95c6e6f6a0e2994675cbc750d7a4dd6436b13b
Reviewed-on: https://go-review.googlesource.com/c/151319
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

Abort evconst if its argument isn't a Go constant. The SSA backend
will do the optimizations in question later. They tend to be weird
cases, like uintptr(unsafe.Pointer(uintptr(1))).

Fix OADDSTR and OCOMPLEX cases in isGoConst.
OADDSTR has its arguments in n.List, not n.Left and n.Right.
OCOMPLEX might have a 2-result function as its arg in List[0]
(in which case it isn't a Go constant).

Fixes #24760

Change-Id: Iab312d994240d99b3f69bfb33a443607e872b01d
Reviewed-on: https://go-review.googlesource.com/c/151338
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

This experiment is less effective and less needed since the
introduction of stack objects.

We can't clobber stack objects because we don't know statically
whether they are live or not.

We don't really need this experiment that much any more, as it was
primarily used to test the complicated ambiguously-live logic in the
liveness analysis, which has been removed in favor of stack objects.

It is also ~infeasible to maintain once we have safepoints everywhere.

Fixes #27326

Change-Id: I3bdde480b93dd508d048703055d4586b496176af
Reviewed-on: https://go-review.googlesource.com/c/151317
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

This patch merges in support for reading indexed type export data,
from the gofrontend CL https://golang.org/cl/143022 (which includes
a change in the export data version number from V2 to V3).

Also fixes the key tests to insure that they run both in gccgo builds
and main Go repo builds if "gccgo" is present (prior to this the tests
were not running in either scenario); this required fixing up some of
the expected results.

Fixes #28961.

Change-Id: I644d171f2a46be9160f89dada06ab3c20468bab7
Reviewed-on: https://go-review.googlesource.com/c/149957
Run-TryBot: Than McIntosh <thanm@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

In assembly free packages (aka "complete" or "pure go"), allow
bodyless functions if they are linkname'd to something else.

Presumably the thing the function is linkname'd to has a definition.
If not, the linker will complain. And linkname is unsafe, so we expect
users to know what they are doing.

Note this handles only one direction, where the linkname directive
is in the local package. If the linkname directive is in the remote
package, this CL won't help. (See os/signal/sig.s for an example.)

Fixes #23311

Change-Id: I824361b4b582ee05976d94812e5b0e8b0f7a18a6
Reviewed-on: https://go-review.googlesource.com/c/151318
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>

The exception handler modifies the stack and continuation context so
it looks like the faulting code calls sigpanic() directly. The call was
not set up correctly on ARM, because it did not handle the link register
correctly. This change handles the link register correctly for ARM.

Updates #28854

Change-Id: I7ccf838adfc05cd968a5edd7d19ebba6a2478360
Reviewed-on: https://go-review.googlesource.com/c/150957
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>

This commit adapts compile tool to create correct nilchecks for AIX.

AIX allows to load a nil pointer. Therefore, the default nilcheck
which issues a load must be replaced by a CMP instruction followed by a
store at 0x0 if the value is nil. The store will trigger a SIGSEGV as on
others OS.

The nilcheck algorithm must be adapted to do not remove nilcheck if it's
only a read. Stores are detected with v.Type.IsMemory().

Tests related to nilptr must be adapted to the previous changements.
nilptr.go cannot be used as it's because the AIX address space starts at
1<<32.

Change-Id: I9f5aaf0b7e185d736a9b119c0ed2fe4e5bd1e7af
Reviewed-on: https://go-review.googlesource.com/c/144538
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

This commit allows the runtime to handle 64bits addresses returned by
mmap syscall on AIX.

Mmap syscall returns addresses on 59bits on AIX. But the Arena
implementation only allows addresses with less than 48 bits.
This commit increases the arena size up to 1<<60 for aix/ppc64.

Update: #25893

Change-Id: Iea72e8a944d10d4f00be915785e33ae82dd6329e
Reviewed-on: https://go-review.googlesource.com/c/138736Reviewed-by: Austin Clements <austin@google.com>

Now that maps are printed in deterministic order, the map example
can have multiple keys without breaking the build.

Change-Id: Iccec0cd76a3d41c75d8d4eb768ec0ac09ad9f2ad
Reviewed-on: https://go-review.googlesource.com/c/151218Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>

The String method is n-squared and overwrites its receiver.
Fix both issues, with only a slight loss of clarity.

Fixes #28773

Change-Id: I588f69d4cbd72931b28b984671512834473bd466
Reviewed-on: https://go-review.googlesource.com/c/151217Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

TestImportTableInUnknownSection was introduced in CL 110555 to
test PE executable with import table located in section other than
".idata". We used atmfd.dll for that purpose, but it seems
atmfd.dll is not present on some systems.

Use kernel32.dll instead. kernel32.dll import table is located in
".rdata" section, so it should do the job. And every Windows
system has kernel32.dll file.

Also make TestImportTableInUnknownSection run on windows-arm,
since windows-arm should also have kernel32.dll file.

Updates #27904

Change-Id: Ie005ee10e46ae0c06e83929d581e89f86c051eea
Reviewed-on: https://go-review.googlesource.com/c/151137
Run-TryBot: Alex Brainman <alex.brainman@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

This makes it easier to track names of function arguments
for debugging purposes.

Change-Id: Ic34856fe0b910005e1c7bc051d769d489a4b158e
Reviewed-on: https://go-review.googlesource.com/c/150098
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

This CL adds a button that collapses values list of a block.
Button is displayed for every block with non-empty values.

Change-Id: I4b65af81e25349f38341df487d42698c9d006a00
Reviewed-on: https://go-review.googlesource.com/c/144557
Run-TryBot: Yury Smolsky <yury@smolsky.by>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

On some platforms, assembly in internal/syscall/unix references
unexported runtime symbols. Catch these references so the compiler can
generate the necessary ABI wrappers.

Fixes #28769.
Updates #27539.

Change-Id: I118eebfb8b3d907b4c3562198e6afb49854f5827
Reviewed-on: https://go-review.googlesource.com/c/149817
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
Reviewed-by: Clément Chigot <clement.chigot@atos.net>
Reviewed-by: Keith Randall <khr@golang.org>

When using softfloat, floating point ops are rewritten to integer
ops. The types of store ops were not rewritten. This may lower
to floating point stores, which are problematic. This CL fixes
this by rewriting the store types as well.

This fixes test/fixedbugs/issue28688.go on Wasm. Softfloat mode
is not used by default on Wasm, and it is not needed as Wasm spec
supports floating points. But it is nice to have the correct
types.

Change-Id: Ib5e19e19fa9491b15c2f60320f8724cace5cefb5
Reviewed-on: https://go-review.googlesource.com/c/149965
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>

UserHomeDir used to return an empty string if the corresponding
environment variable was not set. Changed it to return an error if the
variable is not set, to have the same signature and behaviour as UserCacheDir.

Fixes #28562

Change-Id: I42c497e8011ecfbbadebe7de1751575273be221c
Reviewed-on: https://go-review.googlesource.com/c/150418
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

The performance improvement is not as big as we hoped.
Until the API is feature complete, we postpone the release
and avoid added complexity.

This change was prepared by reverting all the changes affected
src/cmd/trace and src/internal/traceparser packages after
golang.org/cl/137635, and then bringing back MMU computation
APIs (originally in src/internal/traceparser) to the
src/internal/trace package.

Revert "cmd/trace: use new traceparser to parse the raw trace files"
This reverts https://golang.org/cl/145457
  (commit 08816cb8).

Revert "internal/traceparser: provide parser that uses less space and parses segments of runtime trace files"
This reverts https://golang.org/cl/137635
  (commit daaf361f).

Change-Id: Ic2a068a7dbaf4053cd9674ca7bde9c58e74385b4
Reviewed-on: https://go-review.googlesource.com/c/150517
Run-TryBot: Hyang-Ah Hana Kim <hyangah@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Austin Clements <austin@google.com>

CL 150497 enabled TestRemoveAllDot on "noat" systems.

However, this test is failing on Plan 9 because the rmdir
system call allows to remove "." on Plan 9.

This change prevents the "noat" implementation of RemoveAll to
remove ".", so it remains consistent with the "at" implementation.

Fixes #28903.

Change-Id: Ifc8fe36bdd8053a4e416f0590663c844c97ce72a
Reviewed-on: https://go-review.googlesource.com/c/150621
Run-TryBot: David du Colombier <0intro@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>

This case was missed entirely and caused such params to be
unprintable.  This change gives them stack addresses
for the entire function (which is correct).

Change-Id: Ia4f706450219e48bce65b6395d3d9792df142fb5
Reviewed-on: https://go-review.googlesource.com/c/150657
Run-TryBot: David Chase <drchase@google.com>
Reviewed-by: Heschi Kreinick <heschi@google.com>

CL 142517 has used some formats incorrectly. This change fixes it
by using %v for errors and invoking Block.Kind.String().
Format map stays intact.

Updates #28177

Change-Id: If53b6cc54ba3c1ffc17b005225787e3b546de404
Reviewed-on: https://go-review.googlesource.com/c/150798
Run-TryBot: Yury Smolsky <yury@smolsky.by>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>

This adds a debug check to throw immediately if any pointers are added
to the gcWork buffer after the mark completion barrier. The intent is
to catch the source of the cached GC work that occasionally produces
"P has cached GC work at end of mark termination" failures.

The result should be that we get "throwOnGCWork" throws instead of "P
has cached GC work at end of mark termination" throws, but with useful
stack traces.

This should be reverted before the release. I've been unable to
reproduce this issue locally, but this issue appears fairly regularly
on the builders, so the intent is to catch it on the builders.

This probably slows down the GC slightly.

For #27993.

Change-Id: I5035e14058ad313bfbd3d68c41ec05179147a85c
Reviewed-on: https://go-review.googlesource.com/c/149969
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

For #27993.

Change-Id: I20127e8a9844c2c488f38e1ab1f8f5a27a5df03e
Reviewed-on: https://go-review.googlesource.com/c/149968
Run-TryBot: Austin Clements <austin@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

Fixes #28904

Change-Id: I8d416c47479a266735a39c926fd2f0f2bb25d57b
GitHub-Last-Rev: 3a7865a5be27937833cf4f65c242c639e51665c4
GitHub-Pull-Request: golang/go#28907
Reviewed-on: https://go-review.googlesource.com/c/150737Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>

This CL adds CFGs to ssa.html.
It execs dot to generate SVG,
which then gets inlined into the html.
Some standard naming and javascript hacks
enable integration with the rest of ssa.html.
Clicking on blocks highlights the relevant
part of the CFG, and vice versa.

Sample output and screenshots can be seen in #28177.

CFGs can be turned on with the suffix mask:
:*            - dump CFG for every phase
:lower        - just the lower phase
:lower-layout - lower through layout
:w,x-y        - phases w and x through y

Calling dot after every pass is noticeably slow,
instead use the range of phases.

Dead blocks are not displayed on CFG.

User can zoom and pan individual CFG
when the automatic adjustment has failed.

Dot-related errors are reported
without bringing down the process.

Fixes #28177

Change-Id: Id52c42d86c4559ca737288aa10561b67a119c63d
Reviewed-on: https://go-review.googlesource.com/c/142517
Run-TryBot: Yury Smolsky <yury@smolsky.by>
Reviewed-by: David Chase <drchase@google.com>

In race mode, these functions are defined and declared in
different packages, which therefore don't have implicit arg maps.
When they are defer'd, and the stack needs to move, the runtime
fails with missing stack maps. This CL adds arg maps (FUNCDATA)
to them.

Updates #28848

Change-Id: I0271563b7e78e7797ce2990c303dced957efaa86
Reviewed-on: https://go-review.googlesource.com/c/150457
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>

Using fmt.Scanln in a browser environment caused a panic, since there
was no stub for fs.read. This commit adds a stub that returns ENOSYS.

Fixes #27773.

Change-Id: I79b019039e4bc90da51d71a4edddf3bd7809ff45
Reviewed-on: https://go-review.googlesource.com/c/150617
Run-TryBot: Richard Musiol <neelance@gmail.com>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

The playground is not using GOOS=js, so it is not able to use the
package syscall/js. Examples that depend on syscall/js should not
show a "Run" button.

Fixes #28526.

Change-Id: I8b2fcdd0c0ee517a5c3864bf459f813129542389
Reviewed-on: https://go-review.googlesource.com/c/148918Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>