This CL implements the math/bits.OnesCount{8,16,32,64} functions
as intrinsics on s390x using the 'population count' (popcnt)
instruction. This instruction was released as the 'population-count'
facility which uses the same facility bit (45) as the
'distinct-operands' facility which is a pre-requisite for Go on
s390x. We can therefore use it without a feature check.
The s390x popcnt instruction treats a 64 bit register as a vector
of 8 bytes, summing the number of ones in each byte individually.
It then writes the results to the corresponding bytes in the
output register. Therefore to implement OnesCount{16,32,64} we
need to sum the individual byte counts using some extra
instructions. To do this efficiently I've added some additional
pseudo operations to the s390x SSA backend.
Unlike other architectures the new instruction sequence is faster
for OnesCount8, so that is implemented using the intrinsic.
name old time/op new time/op delta
OnesCount 3.21ns ± 1% 1.35ns ± 0% -58.00% (p=0.000 n=20+20)
OnesCount8 0.91ns ± 1% 0.81ns ± 0% -11.43% (p=0.000 n=20+20)
OnesCount16 1.51ns ± 3% 1.21ns ± 0% -19.71% (p=0.000 n=20+17)
OnesCount32 1.91ns ± 0% 1.12ns ± 1% -41.60% (p=0.000 n=19+20)
OnesCount64 3.18ns ± 4% 1.35ns ± 0% -57.52% (p=0.000 n=20+20)
Change-Id: Id54f0bd28b6db9a887ad12c0d72fcc168ef9c4e0
Reviewed-on: https://go-review.googlesource.com/114675
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Teach escape analysis to recognize these assignment patterns
as not causing the src to leak:
val.x = val.y
val.x[i] = val.y[j]
val.x1.x2 = val.x1.y2
... etc
Helps to avoid "leaking param" with assignments showed above.
The implementation is based on somewhat similiar xs=xs[a:b]
special case that is ignored by the escape analysis.
We may figure out more generalized version of this,
but this one looks like a safe step into that direction.
Updates #14858
Change-Id: I6fe5bfedec9c03bdc1d7624883324a523bd11fde
Reviewed-on: https://go-review.googlesource.com/126395
Run-TryBot: Iskander Sharipov <iskander.sharipov@intel.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
This is still not fixed, the testcase reflects that there are still
a few boundchecks. Let's fix the good alternative with an explicit
test though.
Updates #24876
Change-Id: I4da35eb353e19052bd7b69ea6190a69ced8b9b3d
Reviewed-on: https://go-review.googlesource.com/107355
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Giovanni Bajo <rasky@develer.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The codegen testsuite uses regexp to parse the syntax, but it doesn't
have a way to tell line comments containing checks from line comments
containing English sentences. This means that any syntax error (that
is, non-matching regexp) is currently ignored and not reported.
There were some tests in memcombine.go that had an extraneous space
and were thus effectively disabled. It would be great if we could
report it as a syntax error, but for now we just punt and swallow the
spaces as a workaround, to avoid the same mistake again.
Fixes#25452
Change-Id: Ic7747a2278bc00adffd0c199ce40937acbbc9cf0
Reviewed-on: https://go-review.googlesource.com/113835
Run-TryBot: Giovanni Bajo <rasky@develer.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
The words 'the returned' were changed to 'a returned' in
8201b92aae when referring to the value
returned by SystemCertPool. Brad Fitz pointed out after that commit was
merged that it makes the wording of this function doc inconsistent with
rest of the stdlib since 'a returned' is not used anywhere, but 'the
returned' is frequently used.
Fixes#27385
Change-Id: I289b533a5a0b5c63eaf0abb6dec0085388ecf76b
GitHub-Last-Rev: 6c83b80257
GitHub-Pull-Request: golang/go#27438
Reviewed-on: https://go-review.googlesource.com/132776
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
In the CPU profile tests for gccgo, check to make sure that the
runtime's sigprof handler itself doesn't appear in the profile. Add a
"skip if gccgo" guard to one testpoint.
Updates #26595
Change-Id: I92a44161d61f17b9305ce09532134edd229745a7
Reviewed-on: https://go-review.googlesource.com/126316
Run-TryBot: Than McIntosh <thanm@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
This change adds support for VDSO on ppc64x, making it possible to
avoid a syscall in walltime and nanotime.
BenchmarkClockVDSOAndFallbackPaths/vDSO-192 20000000 66.0 ns/op
BenchmarkClockVDSOAndFallbackPaths/Fallback-192 1000000 1456 ns/op
Change-Id: I3373bd804b6f122961de3ae9d034e6ccf35748e6
Reviewed-on: https://go-review.googlesource.com/131135
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
Some versions of Windows (Windows 10 1803) do not set file
position after TransmitFile completes. So just use Seek
to set file position before returning from sendfile.
Fixes#25722
Change-Id: I7a49be10304b5db19dda707b13ac93d338aeb190
Reviewed-on: https://go-review.googlesource.com/131976
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Yasuhiro MATSUMOTO <mattn.jp@gmail.com>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Fence-post implications of the form "x-1 >= w && x > min ⇒ x > w"
were not correctly handling unsigned domain, by always checking signed
limits.
This bug was uncovered once we taught prove that len(x) is always
>= 0 in the signed domain.
In the code being miscompiled (s[len(s)-1]), prove checks
whether len(s)-1 >= len(s) in the unsigned domain; if it proves
that this is always false, it can remove the bound check.
Notice that len(s)-1 >= len(s) can be true for len(s) = 0 because
of the wrap-around, so this is something prove should not be
able to deduce.
But because of the bug, the gate condition for the fence-post
implication was len(s) > MinInt64 instead of len(s) > 0; that
condition would be good in the signed domain but not in the
unsigned domain. And since in CL105635 we taught prove that
len(s) >= 0, the condition incorrectly triggered
(len(s) >= 0 > MinInt64) and things were going downfall.
Fixes#27251Fixes#27289
Change-Id: I3dbcb1955ac5a66a0dcbee500f41e8d219409be5
Reviewed-on: https://go-review.googlesource.com/132495
Reviewed-by: Keith Randall <khr@golang.org>
The benchstat tool computes statistics about benchmarks, including
whether any differences are statistically significant. Recommend its use
in commit messages of performance-related changes rather than the
simpler benchcmp tool.
Change-Id: I4b35c2d892b48e60c3064489b035774792c19c30
Reviewed-on: https://go-review.googlesource.com/132515
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Change-Id: Ifd509c0c6a6ea41094b6ae1f4931414325b152fd
Reviewed-on: https://go-review.googlesource.com/132475
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Internal helper functions for type-checking type expressions were
renamed to make it clearer when they should be used:
typExpr (w/o def) -> typ
typExpr (w/ def) -> definedType
typ -> indirectType
typExprInternal -> typInternal
The rename emphasizes that in most cases Checker.typ should be used
to compute the types.Type from an ast.Type. If the type is defined,
definedType should be used. For composite type elements which are
not "inlined" in memory, indirectType should be used.
In the process, implicitly changed several uses of indirectType
(old: typ) to typ (old: typExpr) by not changing the respective
function call source. These implicit changes are ok in those
places because either call is fine where we are not concerned
about composite type elements. But using typ (old: typExpr) is
more efficient than using indirectType (old: typ).
Change-Id: I4ad14d5357c5f94b6f1c33173de575c4cd05c703
Reviewed-on: https://go-review.googlesource.com/130595
Reviewed-by: Alan Donovan <adonovan@google.com>
Now that most of the type-checker is using the object-coloring mechanism
to detect cycles, remove the explicit path parameter from the functions
that don't rely on it anymore.
Some of the syntactic-based resolver code (for aliases, interfaces)
still use an explicit path; leaving those unchanged for now.
The function cycle was moved from typexpr.go (where it is not used
anymore) to resolver.go (where it's still used). It has not changed.
Fixes#25773.
Change-Id: I2100adc8d66d5da9de9277dee94a1f08e5a88487
Reviewed-on: https://go-review.googlesource.com/130476
Reviewed-by: Alan Donovan <adonovan@google.com>
For Go 1.11, cycle tracking of global (package-level) objects was changed
to use a Checker-level object path rather than relying on the explicit
path parameter that is passed around to some (but not all) type-checker
functions.
This change now uses the same mechanism for the detection of local
type cycles (local non-type objects cannot create cycles by definition
of the spec).
As a result, local alias cycles are now correctly detected as well
(issue #27106).
The path parameter that is explicitly passed around to some type-checker
methods is still present and will be removed in a follow-up CL.
Also:
- removed useCycleMarking flag and respective dead code
- added a couple more tests
- improved documentation
Fixes#27106.
Updates #25773.
Change-Id: I7cbf304bceb43a8d52e6483dcd0fa9ef7e1ea71c
Reviewed-on: https://go-review.googlesource.com/130455
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Alan Donovan <adonovan@google.com>
Previously, pattern matching was good enough to achieve good performance
for the RotateLeft* functions, but the inlining cost for them was much
too high. Make RotateLeft* intrinsic on amd64 as a stop-gap for now to
reduce inlining costs.
This should be done (or at least looked at) for other architectures
as well.
Updates golang/go#17566
Change-Id: I6a106ff00b6c4e3f490650af3e083ed2be00c819
Reviewed-on: https://go-review.googlesource.com/132435
Run-TryBot: Andrew Bonventre <andybons@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>
The sentence in the docs for SystemCertPool that states that mutations
to a returned pool do not affect any other pool is ambiguous as to who
the any other pools are, because pools can be created in multiple ways
that have nothing to do with the system certificate pool. Also the use
of the word 'the' instead of 'a' early in the sentence implies there is
only one shared pool ever returned.
Fixes#27385
Change-Id: I43adbfca26fdd66c4adbf06eb85361139a1dea93
GitHub-Last-Rev: 2f1ba09fa4
GitHub-Pull-Request: golang/go#27388
Reviewed-on: https://go-review.googlesource.com/132378
Reviewed-by: Filippo Valsorda <filippo@golang.org>
"someting" is misspelled and the error handling both clobbers the
error that occurs and distracts from the point of the example, which
is to demonstrate how Printf works. It's better to just panic with the
error.
Change-Id: I5fb0a4a1a8b4772cbe0302582fa878d95e3a4060
Reviewed-on: https://go-review.googlesource.com/132376
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Instead of calling run synchronously, we pass it through bgrun
and immediately wait for it to finish. This pushes all jobs
to execute through the bgwork channel and therefore causes
them to exit cleanly in case of a compiler error.
Fixes#25981
Change-Id: I789a85d23fabf32d144ab85a3c9f53546cb7765a
Reviewed-on: https://go-review.googlesource.com/127776
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The type-checker currently crashes when checking code such as:
_ = map[string][...]int{"": {1, 2, 3}}
In this case, the type checker reports an error for map[string][...]int,
then proceeds to type-check the values of the map literal using a hint
type of [...]int. When type-checking the inner composite (array) literal,
the length of the open array type is computed from the elements,
then the array type is recorded, but the literal has no explicit type
syntax against which to record the type, so this code causes the
type-checker to panic. Add a nil check before calling
check.recordTypeAndValue to avoid that.
Updates #22467
Change-Id: Ic4453ba485b7b88ede2a89f209365eda9e032abc
Reviewed-on: https://go-review.googlesource.com/132355
Reviewed-by: Alan Donovan <adonovan@google.com>
Refactor TestSplice/readerAtEOF to handle cases where we disable
splice on older kernels better.
If splice is disabled, net.splice and poll.Splice do not get to
observe EOF on the reader, because poll.Splice returns immediately
with EINVAL. The test fails unexpectedly, because the splice operation
is reported as not handled.
This change refactors the test to handle the aforementioned case
correctly, by not calling net.splice directly, but using a higher
level check.
Fixes#27355.
Change-Id: I0d5606b4775213f2dbbb84ef82ddfc3bab662a31
Reviewed-on: https://go-review.googlesource.com/132096
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The errors package has an example for Errorf, but the fmt
package does not. Copy the Errorf example from errors to
fmt. Move existing Stringer example into separate file, so as
not to break the assumption that the entire file will be
presented as the example.
Change-Id: I8a210a69362017fa08615a8c3feccdeee8427e22
Reviewed-on: https://go-review.googlesource.com/132239
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Change-Id: Ide50aba940727a7b32cd33dea5315050f1a34717
Reviewed-on: https://go-review.googlesource.com/132237
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Change-Id: Id0e2fb2abad5b776ac0ed76e55e36c6b774b5b7a
Reviewed-on: https://go-review.googlesource.com/132278
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
runtime.wbBufFlush must not modify its arguments, because the
argument slots are also used as spill slots in runtime.gcWriteBarrier.
So, GOEXPERIMENT=clobberdead must not clobber them.
Updates #27326.
Change-Id: Id02bb22a45201eecee748d89e7bdb3df7e4940e4
Reviewed-on: https://go-review.googlesource.com/131957
Reviewed-by: Keith Randall <khr@golang.org>
We now have safepoints at nearly all the instructions. When
GOEXPERIMENT=clobberdead is on, it inserts clobbers nearly at
every instruction. Currently this doesn't work. (Maybe the stack
maps at non-call safepoints are still imprecise. I haven't
investigated.) For now, only use call-based safepoints if the
experiment is on.
Updates #27326.
Change-Id: I72cda9b422d9637cc5738e681502035af7a5c02d
Reviewed-on: https://go-review.googlesource.com/131956
Reviewed-by: Keith Randall <khr@golang.org>
The current implementation crashes when someone writes a panic outside of
a function, which makes sense since that is broken code. This fix allows
one to type-check broken code.
Updates #22467
Change-Id: I81b90dbd918162a20c60a821340898eaf02e648d
Reviewed-on: https://go-review.googlesource.com/132235
Reviewed-by: Alan Donovan <adonovan@google.com>
The sym.Symbol 'ElfType' field is used only for symbols corresponding
to things in imported shared libraries, hence is not needed in the
common case. Relocate it to sym.AuxSymbol so as to shrink the main
Symbol struct.
Updates #26186
Change-Id: I803efc561c31a0ca1d93eca434fda1c862a7b2c5
Reviewed-on: https://go-review.googlesource.com/125479
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The sym.Symbol 'Plt' and 'Got' field are used only with cgo and/or
external linking and are not needed for most symbols. Relocate them to
sym.AuxSymbol so as to shrink the main Symbol struct.
Updates #26186
Change-Id: I170d628a760be300a0c1f738f0998970e91ce3d6
Reviewed-on: https://go-review.googlesource.com/125478
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The sym.Symbol 'Localentry' field is used only with cgo and/or
external linking on MachoPPC. Relocate it to sym.AuxSymbol since it is
infrequently used, so as to shrink the main Symbol struct.
Updates #26186
Change-Id: I5872aa3f059270c2a091016d235a1a732695e411
Reviewed-on: https://go-review.googlesource.com/125477
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Create a new "AuxSymbol" struct into which 'cold' or 'infrequently
set' symbol fields are located. Move the Extname field from the
main Symbol struct to AuxSymbol.
Updates #26186
Change-Id: I9e795fb0cc48f978e2818475fa073ed9f2db202d
Reviewed-on: https://go-review.googlesource.com/125476
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The system call geteuid can not work properly on android, which
causes a test case failed on rooted android/arm.
This CL disables the test case on android.
Fixes#27364
Change-Id: Ibfd33ef8cc1dfe8822c8be4280eae12ee30929c1
Reviewed-on: https://go-review.googlesource.com/132175
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Tobias Klauser <tobias.klauser@gmail.com>
