When adding a large constant to a register we generate two adds,
we may generate two ADD instructions if the constant does not fit
in one ADD but does fit in two. This is generally fine except that
if the target register is SP (such as in function prologues or
epilogues for functions with large frames), this creates an
intermediate state that the SP is not 0 nor the full frame size.
For signal safety (preemption signal and profiling signal) we
require that the frame is either not created at all or fully
created, meaning that the SP must be written in a single
instruction. Splitting to two adds breaks the requirement. So not
splitting it.
(We could mark such instructions not async-preemptible. But
profiling signal can still cause problems.)
(We could generate "ADD $c1, SP, Rtmp; ADD $c2; Rtmp; SP" to save
an instruction if that is desired, while still ensuring that SP
is written in a single instruction.)
May fix flaky failures like https://build.golang.org/log/11537ec020a902b0ec0fc065f61161b729eb9880
Change-Id: I5cf38a6a028afe01aa3b6eeb163487bbd504b64f
Reviewed-on: https://go-review.googlesource.com/c/go/+/358436
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
Reviewed-by: Joel Sing <joel@sing.id.au>
TryBot-Result: Go Bot <gobot@golang.org>
Bump the export version to a new value iexportVersionGo1_18 (2). This
will give a better error message when old compilers/tools encounter the
new export format (that includes parameterized types and functions).
We are also making a breaking change in the format:
- a 'kind' byte is added to constant values
Also updated tinter() to pass the implicit bit through during type
substitution.
Tested that all tests still pass if the iexportVersionCurrent is changed
back to 1 in typecheck/iexport.go, iimporter/iimport.go, and
gcimporter/iimport.go
Updates #47654
Change-Id: I1dbeb167a97f6c7e0b7e0c011d6bada5db312b36
Reviewed-on: https://go-review.googlesource.com/c/go/+/357049
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Trust: Dan Scales <danscales@google.com>
The -asan option causes the compiler to add instrumentation for the
C/C++ address sanitizer. Every memory read/write will be replaced
by a call to asanread/asanwrite.
This CL also inserts asan instrumentation during SSA building.
This CL passes tests but is not usable by itself. The actual
implementation of asanread/asanwrite in the runtime package, and
support for -asan in the go tool and tests, will follow in subsequent
CLs.
Updates #44853.
Change-Id: Ia18c9c5d5c351857420d2f6835f0daec2ad31096
Reviewed-on: https://go-review.googlesource.com/c/go/+/298611
Trust: fannie zhang <Fannie.Zhang@arm.com>
Run-TryBot: fannie zhang <Fannie.Zhang@arm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
We want to set a finalizer on splicePipe, so make it large enough to
not use the tiny allocator. Otherwise the finalizer will not run until
the rest of the tiny allocation can be freed. This only matters on
32-bit systems.
Fixes#48968
Change-Id: I8eb3c9f48fdccab7dc79c5b918d4257b6151ee91
Reviewed-on: https://go-review.googlesource.com/c/go/+/358114
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
Generic function symbols sometimes have % in them, like:
main.B2[%2eshape.string_0].m2·f
Which confuses this code because it doesn't esacpe % when
using this string as a format string, instead of a format argument.
Or could we get rid of the . -> %2e rewrite somehow?
I think it comes from LinkString.
Change-Id: I3275501f44cf30485e9d4577e0dfa77996d4939e
Reviewed-on: https://go-review.googlesource.com/c/go/+/357837
Trust: Keith Randall <khr@golang.org>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
The method of using references to dictionaries to hold methods
live during linker deadcode elimination wasn't working very well.
I implemented a new scheme in the CL below this, so this CL strips
out the old method.
The new method has the added benefit of having 0 runtime overhead
(unlike the stuff we're ripping out here, which does have a small overhead).
Update #48047
Change-Id: I68ac57119792d53c58f1480f407de6ab2bb53211
Reviewed-on: https://go-review.googlesource.com/c/go/+/357836
Trust: Keith Randall <khr@golang.org>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
We don't need it any more, after CL 357835.
Change-Id: I1ff5f24b5540c3e80c4b35be8215a1c378952274
Reviewed-on: https://go-review.googlesource.com/c/go/+/357894
Trust: Keith Randall <khr@golang.org>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
To capture the fact that a method was called on a generic interface,
so we can make sure the linker doesn't throw away any implementations
that might be the method called.
See the comment in reflect.go for details.
Fixes#49049
Change-Id: I0be74b6e727c1ecefedae072b149f59d539dc1e9
Reviewed-on: https://go-review.googlesource.com/c/go/+/357835
Trust: Keith Randall <khr@golang.org>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Dan Scales <danscales@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
On wasm, the wasm_exec.js helper passes the command line arguments and
environment variables via a reserved space in the wasm linear memory.
Increase this reserved space from 4096 to 8192 bytes so more environment
variables can fit into the limit.
Later, after https://golang.org/cl/350737 landed, we can switch to the
WASI interface for getting the arguments and environment. This would
remove the limit entirely.
Fixes#49011
Change-Id: I48a6e952a97d33404ed692c98e9b49c5cd6b269b
Reviewed-on: https://go-review.googlesource.com/c/go/+/358194
Trust: Richard Musiol <neelance@gmail.com>
Run-TryBot: Richard Musiol <neelance@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Mui <cherryyz@google.com>
Previously, the test would panic if the short timeout
was reached before fuzzing began. Increasing the
timeout should stop this test from being flaky.
Fixes#49046
Change-Id: Iaa0b3b3e8ea29d9a42ab5fc1c801fc73fffe1675
Reviewed-on: https://go-review.googlesource.com/c/go/+/358055
Trust: Katie Hockman <katie@golang.org>
Run-TryBot: Katie Hockman <katie@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Bryan C. Mills <bcmills@google.com>
When doing a tail call the link register is live as the callee
will directly return to the caller (of the function that does the
tail call). Don't allocate or clobber the link register.
Fixes#49032.
Change-Id: I2d60f2354e5b6c14aa285c8983a9786687b90223
Reviewed-on: https://go-review.googlesource.com/c/go/+/358435
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
In preparation for capturing the implicit interface bit in export data,
thread through the IsImplicit property from types2 into typecheck.
Change-Id: I9b46fe73de102935a127e6ececaacd76738b557e
Reviewed-on: https://go-review.googlesource.com/c/go/+/357109
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Now that we've removed the necessity for subscripts in importers, we can
effectively eliminate them from the the type parameter API by removing
them from the type string.
Change-Id: Ic4491b0dea27b0e0ce2d0636dccfaf05168ef9e5
Reviewed-on: https://go-review.googlesource.com/c/go/+/357814
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
The Content-Length was incorrectly set to 0 for ill-formed and invalid
values. In these cases, return an error.
If the Content-Length header was omitted, it was incorrectly set to 0.
In this case, set the Content-Length value to -1.
Fixes#49108
Change-Id: I24fe9a31ed5b6ddb53f2b2bd10f2c84e428823e3
Reviewed-on: https://go-review.googlesource.com/c/go/+/358134
Run-TryBot: Johan Brandhorst-Satzkorn <johan.brandhorst@gmail.com>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Trust: Brad Fitzpatrick <bradfitz@golang.org>
Trust: David Crawshaw <crawshaw@golang.org>
CL 357649 fixes inlining labeled FOR/RANGE loops,
we should do same translation for inlined SWITCH's label
Fixes#49145
Change-Id: I9a6f365f57e974271a1eb279b38e81f9b5148788
Reviewed-on: https://go-review.googlesource.com/c/go/+/358315
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Trust: Dan Scales <danscales@google.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Dan Scales <danscales@google.com>
When running benchmarks with high -count and no tests (either at all or
filtered with -run), the time for runTests to check for tests -count
times can add a significant delay to starting benchmarks.
To avoid this delay, make runTests bail out on the second iteration if
the first found no tests to run. We expect the same tests to run every
time, so there is no reason to duplicate work.
One caveat: the testing docs do not explicitly require the same subtests
to be run on every call, so this could break tests that depend on
multiple runs to actually run all tests. I consider such tests invalid,
but some may exist.
Fixes#49050
Change-Id: I7b34f3832b31493cc089ee0555e231f4dc690154
Reviewed-on: https://go-review.googlesource.com/c/go/+/356669
Trust: Michael Pratt <mpratt@google.com>
Run-TryBot: Michael Pratt <mpratt@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
- change _TypeSet.hasTerms() to report if a type set has actual types
(excluding a "universe" term)
- handle empty type set type arguments correctly
- bring comments up-to-date in Checker.satisfies
Change-Id: I87f9a1096ebb21a1b08c87a9b59f400f3bc2f040
Reviewed-on: https://go-review.googlesource.com/c/go/+/358175
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
This updates the testing package documentation to reference the
specification for the benchmark format, and points users to our
standard tools for working with benchmark data. (It's somewhat
remarkable how widely used benchstat appears to be given that we don't
mention it anywhere!)
Change-Id: Idbb4416d2fde9606ea7e6c15595f3b9e6a38f3b0
Reviewed-on: https://go-review.googlesource.com/c/go/+/357589
Trust: Austin Clements <austin@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The existing DrawMask method is generic and is therefore calling the At().RGBA() method for every pixel of the mask and the source.
Do a specific implementation when the mask is *image.Alpha (which is
common) and use use the PixOffset method to increase performances.
name old time/op new time/op delta
RGBA2-12 1.60ms ± 0% 1.13ms ± 1% -29.16% (p=0.008 n=5+5)
GenericMaskOver-12 915µs ± 4% 926µs ± 1% ~ (p=0.190 n=5+4)
RGBA64Over-12 1.53ms ± 3% 1.21ms ± 2% -20.74% (p=0.008 n=5+5)
GrayOver-12 1.36ms ± 2% 1.01ms ± 7% -26.27% (p=0.008 n=5+5)
Fixes: #46395
Change-Id: Iaeaa8cfcc6a3fe93eb19b361f3bf076e41cac5b6
Reviewed-on: https://go-review.googlesource.com/c/go/+/323749
Reviewed-by: Nigel Tao <nigeltao@golang.org>
Trust: Nigel Tao <nigeltao@golang.org>
Trust: Andrew Gerrand <adg@golang.org>
Run-TryBot: Nigel Tao <nigeltao@golang.org>
CL 358117 fixed a bug that Phi's argument wasn't updated correctly after
removing a predecessor of Block. This CL factor out the code that
updates phi argument into a Block's method, so it's easier to use,
maintain and hopefully prevent that kind of bug in the future.
Change-Id: Ie9741e19ea28f56860425089b6093a381aa10f5b
Reviewed-on: https://go-review.googlesource.com/c/go/+/357964
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
It's only used in two places:
- The one in regalloc.go can be replaced with v.resetArgs()
- The one in rewrite.go can be open coded
and can cause wrong usage like the bug that CL 358117 fixed.
Change-Id: I125baf237db159d056fe4b1c73072331eea4d06a
Reviewed-on: https://go-review.googlesource.com/c/go/+/357965
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
removePred and removeArg do different things. removePred moves the last
predecessor to index k, whereas removeArg slides all the args k or
greater down by 1 index.
Kind of unfortunate different behavior in things named similarly.
Fixes#49122
Change-Id: I9ae409bdac744e713f4c121f948e43db6fdc8542
Reviewed-on: https://go-review.googlesource.com/c/go/+/358117
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
This generalizes range clauses. Removed some dead code and cleaned
up the surrounding bits.
Change-Id: Icd8384205afa3f52b7e7df9abed5de2bb556861d
Reviewed-on: https://go-review.googlesource.com/c/go/+/357778
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
This generalizes make the same way copy was generalized and eliminates
a use of optype.
Change-Id: I8221abd53d77dde8ead47c0075c13fd2a3221642
Reviewed-on: https://go-review.googlesource.com/c/go/+/357776
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Implement singleUnder[String] which determines a single underlying type
for a given type: either the underlying type, or the single underlying
type for a type parameter, if it exists. Use singleUnder[String] instead
of optype for copy built-in.
This CL removes a dependency on optype and also makes the copy built-in
slighty more general for generic arguments (the source argument may be
constrained by a slice or string simultaneously).
Change-Id: Ia329e96afc69a09d2ca3b1f82fe712d4f7ba1d9f
Reviewed-on: https://go-review.googlesource.com/c/go/+/357413
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
Since CL 282892, functions are always compiled before closures. To do
that, when walking the closure, it is added to its outer function queue
for scheduling compilation later. Thus, a closure may be added to queue
more than once, causing the ICE dues to being compiled twice.
To fix this, catching the re-walking of the closure expression and do
not add it to the compilation queue.
Fixes#49029
Change-Id: I7d188e8f5b4d5c4248a0d8e6389da26f1084e464
Reviewed-on: https://go-review.googlesource.com/c/go/+/357960
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
The -asan option causes the linker to link against the runtime/asan
package in order to use the C/C++ address sanitizer.
This CL passes tests but is not usable by itself. The actual
runtime/asan package, and support for -asan in the go tool and the
compiler, and tests, are in separate CLs.
Updates #44853.
Change-Id: Ifc6046c1f75ba52777cbb3d937a4b66e91d5798d
Reviewed-on: https://go-review.googlesource.com/c/go/+/298610
Trust: fannie zhang <Fannie.Zhang@arm.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
- slightly refactor convertibleTo and convertibleToImpl
- provide ability to return a conversion failure cause
- add detailed cause for generic conversions
For #47150.
Change-Id: Ie97d89be0234414ef4df22a6920e18acc944a102
Reviewed-on: https://go-review.googlesource.com/c/go/+/357249
Trust: Robert Griesemer <gri@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
There's no need to say "type parameter" for a type parameter. The
name is sufficient (they are always named), and the prose is followed
by "constrained by".
Change-Id: I98df8caa1432b8b7a874e58da6e3ed6be102b4a0
Reviewed-on: https://go-review.googlesource.com/c/go/+/357410
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Findley <rfindley@google.com>
CL 356254 fixed crawling of embeddable types during inline. However, we
are too agressive, since when we call markEmbed for every type seen
during inlining function body. That leads to false positive that for a
non-embedded type, its unexported methods are also marked inline.
Instead, we should only look at struct type that we seen during inlining
function body, and calling markEmbed for all of its embedded fields.
Fixes#49094
Change-Id: I6ef9a8bf1fc649ec6bf75e4883f6031ec8560ba1
Reviewed-on: https://go-review.googlesource.com/c/go/+/357232
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Same as CL 350153 did for Value.Elem to panic on bad notinheap pointers.
While at it, also add more tests for notinheap deref.
Change-Id: Id7d9d12ad8467de5926b6a7e8f9d659fea5fedb5
Reviewed-on: https://go-review.googlesource.com/c/go/+/357630
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Add a new interface method, MarkImplicit, to allow marking interfaces as
implicit from outside the type-checker. This is necessary so that we can
capture the implicit bit in export data, and use it from importers.
For #48424
For #49040
Change-Id: I999aba2a298f92432326d7ccbd87fe133a2e1a72
Reviewed-on: https://go-review.googlesource.com/c/go/+/357796
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
This introduces a number of new classifications which will make it
easier to generate functions to assemble the new instructions of
ISA 3.1, and potentially earlier versions.
No code generation changes should occur as a result of these. These
allow finer control over how an opcode is matched to an optab entry.
Literal values are now classified based on the smallest number of bits
needed to encode, and matching rules will accept a literal if it
can be zero/sign extended to fit a larger literal class.
Likewise, support classifying even register numbers for GPR, VSX, and
FPR instructions. Some instructions require and even/odd register pair,
and these are usually represented by specifying the even register, and
similarly encoded.
Likewise, add a unit test for the argument classifier function (aclass).
This caught an off-by-one bug in aclass which is also fixed.
Updates #44549
Change-Id: Ia03013aea8b56c4d59b7c3812cdd67ddb3b720b9
Reviewed-on: https://go-review.googlesource.com/c/go/+/350152
Reviewed-by: Cherry Mui <cherryyz@google.com>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
Run-TryBot: Cherry Mui <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
There is already a mechanism using inlgen to rename labels insided
inlined functions so that they are unique and don't clash with loops in
the outer function. This is used for OLABEL and OGOTO. Now that we are
doing inlining of OFOR loops, we need to do this translation for OBREAK,
OCONTINUE, and OFOR. I also added the translation for ORANGE loops, in
anticipation of a CL that will allow inlining of ORANGE for loops.
Fixes#49100
Change-Id: I2ccddc3350370825c386965f4a1e4bc54d3c369b
Reviewed-on: https://go-review.googlesource.com/c/go/+/357649
Run-TryBot: Dan Scales <danscales@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Trust: Dan Scales <danscales@google.com>
Currently gcPaceScavenger is called by gcControllerState.commit, but it
manipulates global state which precludes testing. This change detangles
the two.
Change-Id: I10d8ebdf426d99ba49d2f2cb4fb64891e9fd6091
Reviewed-on: https://go-review.googlesource.com/c/go/+/309272
Reviewed-by: Michael Pratt <mpratt@google.com>
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Currently gcController.gcPercent is read non-atomically by
gcControllerState.revise and gcTrigger.test, but these users may
execute concurrently with an update to gcPercent.
Although revise's results are best-effort, reading it directly in this
way is, generally speaking, unsafe.
This change makes gcPercent atomically updated for concurrent readers
and documents the complete synchronization semantics.
Because gcPercent otherwise only updated with the heap lock held or the
world stopped, all other reads can remain unsynchronized.
For #44167.
Change-Id: If09af103aae84a1e133e2d4fed8ab888d4b8f457
Reviewed-on: https://go-review.googlesource.com/c/go/+/308690
Trust: Michael Knyszek <mknyszek@google.com>
Run-TryBot: Michael Knyszek <mknyszek@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Michael Pratt <mpratt@google.com>
Optype should never return a defined type.
Change-Id: I37b29e0c958e127e75e834e71d6392ea80827773
Reviewed-on: https://go-review.googlesource.com/c/go/+/357694
Trust: Robert Findley <rfindley@google.com>
Run-TryBot: Robert Findley <rfindley@google.com>
Reviewed-by: Robert Griesemer <gri@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Change-Id: Id71f3d8d7c1ef7910d5d9497167dc677f2f0a2ef
Reviewed-on: https://go-review.googlesource.com/c/go/+/356535
Trust: Damien Neil <dneil@google.com>
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
Change-Id: I61011b75128478aa50308d84f4cba23b3e241b3f
Reviewed-on: https://go-review.googlesource.com/c/go/+/356536
Trust: Damien Neil <dneil@google.com>
Trust: Cherry Mui <cherryyz@google.com>
Run-TryBot: Damien Neil <dneil@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Damien Neil <dneil@google.com>
Modified condition in the ASM implementation of indexbody to
determine if separator length crosses 16 bytes to BGT from BGE
to avoid incorrectly crossing a page.
Also fixed IndexString to invoke indexbodyp9 when on the POWER9
platform
Change-Id: I0602a797cc75287990eea1972e9e473744f6f5a9
Reviewed-on: https://go-review.googlesource.com/c/go/+/356849
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Lynn Boger <laboger@linux.vnet.ibm.com>
Trust: Keith Randall <khr@golang.org>
This adds a few rules to PPC64 to eliminate some instructions:
- when an isel is used to generate a boolean value based on a
condition and followed by an xori to flip the result, it can
instead flip the operands in the isel and avoid the xori.
= when a neg follows a sub the operands to the sub can be
swapped and the neg avoided.
There are several opportunities in reflect.DeepEqual to omit
xori which improves some of its benchmarks by as much as
5%
Change-Id: I81bbc02c0f16995c65934b6f045867b731ab302b
Reviewed-on: https://go-review.googlesource.com/c/go/+/357509
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Cherry Mui <cherryyz@google.com>
