Escape analysis is currently very naive about identifying calls to
known functions: it only recognizes direct calls to a declared
function, or direct calls to a closure.
This CL adds a new "staticValue" helper function that can trace back
through local variables that were initialized and never reassigned
based on a similar optimization already used by inlining. (And to be
used by inlining in a followup CL.)
Updates #41474.
Change-Id: I8204fd3b1e150ab77a27f583985cf099a8572b2e
Reviewed-on: https://go-review.googlesource.com/c/go/+/256458
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: David Chase <drchase@google.com>
The gofrontend code doesn't correctly handle inlining a function that
refers to a constant with methods.
For #35739
Change-Id: I6bd0b5cd4272dbe9969634b4821e668acacfdcf9
Reviewed-on: https://go-review.googlesource.com/c/go/+/261662
Trust: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
We are converting from using error-prone ad-hoc syntax // +build lines
to less error-prone, standard boolean syntax //go:build lines.
The timeline is:
Go 1.16: prepare for transition
- Builds still use // +build for file selection.
- Source files may not contain //go:build without // +build.
- Builds fail when a source file contains //go:build lines without // +build lines. <<<
Go 1.17: start transition
- Builds prefer //go:build for file selection, falling back to // +build
for files containing only // +build.
- Source files may contain //go:build without // +build (but they won't build with Go 1.16).
- Gofmt moves //go:build and // +build lines to proper file locations.
- Gofmt introduces //go:build lines into files with only // +build lines.
- Go vet rejects files with mismatched //go:build and // +build lines.
Go 1.18: complete transition
- Go fix removes // +build lines, leaving behind equivalent // +build lines.
This CL provides part of the <<< marked line above in the Go 1.16 step:
rejecting files containing //go:build but not // +build.
The standard go command checks only consider the top of the file.
This compiler check, along with a separate go vet check for ignored files,
handles the remainder of the file.
For #41184.
Change-Id: I014006eebfc84ab5943de18bc90449e534f150a2
Reviewed-on: https://go-review.googlesource.com/c/go/+/240601
Trust: Russ Cox <rsc@golang.org>
Run-TryBot: Russ Cox <rsc@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
We lost a sign extension that was necessary. The nonnegative comparison
didn't have the correct extension on it. If the larger constant is
positive, but its shorter sign extension is negative, the rule breaks.
Fixes#41872
Change-Id: I6592ef103f840fbb786bf8cb94fd8804c760c976
Reviewed-on: https://go-review.googlesource.com/c/go/+/260701
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Alberto Donizetti <alb.donizetti@gmail.com>
Backstop support for non-sse2 chips now that 387 is gone.
RELNOTE=yes
Change-Id: Ib10e69c4a3654c15a03568f93393437e1939e013
Reviewed-on: https://go-review.googlesource.com/c/go/+/260017
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Unlike iOS, macOS ARM64 is more of a fully featured OS. Enable
more tests.
Updates #38485.
Change-Id: I2e2240c848d21996db2b950a4a6856987f7a652c
Reviewed-on: https://go-review.googlesource.com/c/go/+/256919
Trust: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Two part fix:
1) bring the type "correction" forward from a later CL in the expand calls series
2) when a leaf-selwect is rewritten in place, update the type (it might have been
changed by the type correction in 1).
Fixes#41736.
Change-Id: Id097efd10481bf0ad92aaead81a7207221c144b5
Reviewed-on: https://go-review.googlesource.com/c/go/+/259203
Trust: David Chase <drchase@google.com>
Run-TryBot: David Chase <drchase@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
This change renames mustHeapAlloc to heapAllocReason, and changes it
to return the reason why the argument must escape, so we don't have to
re-deduce it in its callers just to print the escape reason. It also
embeds isSmallMakeSlice body in heapAllocReason, since the former was
only used by the latter, and deletes isSmallMakeSlice.
An outdated TODO to remove smallintconst, which the TODO claimed was
only used in one place, was also removed, since grepping shows we
currently call smallintconst in 11 different places.
Change-Id: I0bd11bf29b92c4126f5bb455877ff73217d5a155
Reviewed-on: https://go-review.googlesource.com/c/go/+/258678
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Alberto Donizetti <alb.donizetti@gmail.com>
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
My last 387 CL. So sad ... ... ... ... not!
Fixes#40255
Change-Id: I8d4ddb744b234b8adc735db2f7c3c7b6d8bbdfa4
Reviewed-on: https://go-review.googlesource.com/c/go/+/258957
Trust: Keith Randall <khr@golang.org>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
A recent change to improve shifts was generating some
invalid cases when the rule was based on an AND. The
extended mnemonics CLRLSLDI and CLRLSLWI only allow
certain values for the operands and in the mask case
those values were not being checked properly. This
adds a check to those rules to verify that the
'b' and 'n' values used when an AND was part of the rule
have correct values.
There was a bug in some diag messages in asm9. The
message expected 3 values but only provided 2. Those are
corrected here also.
The test/codegen/shift.go was updated to add a few more
cases to check for the case mentioned here.
Some of the comments that mention the order of operands
in these extended mnemonics were wrong and those have been
corrected.
Fixes#41683.
Change-Id: If5bb860acaa5051b9e0cd80784b2868b85898c31
Reviewed-on: https://go-review.googlesource.com/c/go/+/258138
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
Reviewed-by: Paul Murphy <murp@ibm.com>
Reviewed-by: Carlos Eduardo Seo <carlos.seo@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>
Similar to how we report initialization loops in initorder.go and type
alias loops in typecheck.go, this CL updates align.go to warn about
invalid recursive types. The code is based on the loop code from
initorder.go, with minimal changes to adapt from detecting
variable/function initialization loops to detecting type declaration
loops.
Thanks to Cuong Manh Le for investigating this, helping come up with
test cases, and exploring solutions.
Fixes#41575
Updates #41669.
Change-Id: Idb2cb8c5e1d645e62900e178fcb50af33e1700a1
Reviewed-on: https://go-review.googlesource.com/c/go/+/258177
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Trust: Matthew Dempsky <mdempsky@google.com>
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
As part of type checking make's arguments, we were converting untyped
float and complex constant arguments to integers. However, we were
doing this without concern for whether the argument was a declared
constant. Thus a call like "make([]T, n)" could change n from an
untyped float or untyped complex to an untyped integer.
The fix here is to simply change checkmake to not call SetVal, which
will be handled by defaultlit anyway. However, we also need to
properly return the defaultlit result value to the caller, so
checkmake's *Node parameter is also changed to **Node.
Fixes#41680.
Change-Id: I858927a052f384ec38684570d37b10a6906961f7
Reviewed-on: https://go-review.googlesource.com/c/go/+/257966
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
This adds support for the extswsli instruction which combines
extsw followed by a shift.
New benchmark demonstrates the improvement:
name old time/op new time/op delta
ExtShift 1.34µs ± 0% 1.30µs ± 0% -3.15% (p=0.057 n=4+3)
Change-Id: I21b410676fdf15d20e0cbbaa75d7c6dcd3bbb7b0
Reviewed-on: https://go-review.googlesource.com/c/go/+/257017
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Carlos Eduardo Seo <carlos.seo@gmail.com>
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>
When explaining why the slice from a make() call escapes for the -m -m
message, we print "non-const size" if any one of Isconst(n.Left) and
Isconst(n.Right) return false; but for OMAKESLICE nodes with no cap,
n.Right is nil, so Isconst(n.Right, CTINT) will be always false.
Only call Isconst on n.Right if it's not nil.
Fixes#41635
Change-Id: I8729801a9b234b68ae40adad64d66fa7653adf09
Reviewed-on: https://go-review.googlesource.com/c/go/+/257641
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Alberto Donizetti <alb.donizetti@gmail.com>
We grow the backing store on append by 2x for small sizes and 1.25x
for large sizes. The threshold we use for picking the growth factor
used to depend on the old length, not the old capacity. That's kind of
unfortunate, because then doing append(s, 0, 0) and append(append(s,
0), 0) do different things. (If s has one more spot available, then
the former expression chooses its growth based on len(s) and the
latter on len(s)+1.) If we instead use the old capacity, we get more
consistent behavior. (Both expressions use len(s)+1 == cap(s) to
decide.)
Fixes#41239
Change-Id: I40686471d256edd72ec92aef973a89b52e235d4b
Reviewed-on: https://go-review.googlesource.com/c/go/+/257338
Trust: Keith Randall <khr@golang.org>
Trust: Josh Bleecher Snyder <josharian@gmail.com>
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com>
The revised test now checks that unsafe-uintptr correctly works for
variadic uintptr parameters too, and the CL corrects the code so this
code compiles again.
The pointers are still not kept alive properly. That will be fixed by
a followup CL. But this CL at least allows programs not affected by
that to build again.
Updates #24991.
Updates #41460.
Change-Id: If4c39167b6055e602213fb7522c4f527c43ebda9
Reviewed-on: https://go-review.googlesource.com/c/go/+/255877
Trust: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
//go:notinheap
type T int
type U T
We already correctly propagate the notinheap-ness of T to U. But we
have an assertion in the typechecker that if there's no explicit
//go:notinheap associated with U, then report an error. Get rid of
that error so that implicit propagation is allowed.
Adjust the tests so that we make sure that uses of types like U
do correctly report an error when U is used in a context that might
cause a Go heap allocation.
Fixes#41451
Update #40954
Update #41432
Change-Id: I1692bc7cceff21ebb3f557f3748812a40887118d
Reviewed-on: https://go-review.googlesource.com/c/go/+/255637
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
This change adds rules to find pairs of instructions that can
be combined into a single shifts. These instruction sequences
are common in array addressing within loops. Improvements can
be seen in many crypto packages and the hash packages.
These are based on the extended mnemonics found in the ISA
sections C.8.1 and C.8.2.
Some rules in PPC64.rules were moved because the ordering prevented
some matching.
The following results were generated on power9.
hash/crc32:
CRC32/poly=Koopman/size=40/align=0 195ns ± 0% 163ns ± 0% -16.41%
CRC32/poly=Koopman/size=40/align=1 200ns ± 0% 163ns ± 0% -18.50%
CRC32/poly=Koopman/size=512/align=0 1.98µs ± 0% 1.67µs ± 0% -15.46%
CRC32/poly=Koopman/size=512/align=1 1.98µs ± 0% 1.69µs ± 0% -14.80%
CRC32/poly=Koopman/size=1kB/align=0 3.90µs ± 0% 3.31µs ± 0% -15.27%
CRC32/poly=Koopman/size=1kB/align=1 3.85µs ± 0% 3.31µs ± 0% -14.15%
CRC32/poly=Koopman/size=4kB/align=0 15.3µs ± 0% 13.1µs ± 0% -14.22%
CRC32/poly=Koopman/size=4kB/align=1 15.4µs ± 0% 13.1µs ± 0% -14.79%
CRC32/poly=Koopman/size=32kB/align=0 137µs ± 0% 105µs ± 0% -23.56%
CRC32/poly=Koopman/size=32kB/align=1 137µs ± 0% 105µs ± 0% -23.53%
crypto/rc4:
RC4_128 733ns ± 0% 650ns ± 0% -11.32% (p=1.000 n=1+1)
RC4_1K 5.80µs ± 0% 5.17µs ± 0% -10.89% (p=1.000 n=1+1)
RC4_8K 45.7µs ± 0% 40.8µs ± 0% -10.73% (p=1.000 n=1+1)
crypto/sha1:
Hash8Bytes 635ns ± 0% 613ns ± 0% -3.46% (p=1.000 n=1+1)
Hash320Bytes 2.30µs ± 0% 2.18µs ± 0% -5.38% (p=1.000 n=1+1)
Hash1K 5.88µs ± 0% 5.38µs ± 0% -8.62% (p=1.000 n=1+1)
Hash8K 42.0µs ± 0% 37.9µs ± 0% -9.75% (p=1.000 n=1+1)
There are other improvements found in golang.org/x/crypto which are all in the
range of 5-15%.
Change-Id: I193471fbcf674151ffe2edab212799d9b08dfb8c
Reviewed-on: https://go-review.googlesource.com/c/go/+/252097
Trust: Lynn Boger <laboger@linux.vnet.ibm.com>
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com>
"cannot assign to" compiler errors are very laconic: they never
explain why the lhs cannot be assigned to (with one exception, when
assigning to a struct field in a map).
This change makes them a little more specific, in two more cases: when
assigning to a string, or to a const; by giving a very brief reason
why the lhs cannot be assigned to.
Change-Id: I244cca7fc3c3814e00e0ccadeec62f747c293979
Reviewed-on: https://go-review.googlesource.com/c/go/+/255199
Trust: Alberto Donizetti <alb.donizetti@gmail.com>
Run-TryBot: Alberto Donizetti <alb.donizetti@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Update #40954
Change-Id: Ifaab7349631ccb12fc892882bbdf7f0ebf3d845f
Reviewed-on: https://go-review.googlesource.com/c/go/+/251158
Run-TryBot: Keith Randall <khr@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Keith Randall <khr@golang.org>
They can't reasonably be allocated on the heap. Not a huge deal, but
it has an interesting and useful side effect.
After CL 249917, the compiler and runtime treat pointers to
go:notinheap types as uintptrs instead of real pointers (no write
barrier, not processed during stack scanning, ...). That feature is
exactly what we want for cgo to fix#40954. All the cases we have of
pointers declared in C, but which might actually be filled with
non-pointer data, are of this form (JNI's jobject heirarch, Darwin's
CFType heirarchy, ...).
Fixes#40954
Change-Id: I44a3b9bc2513d4287107e39d0cbbd0efd46a3aae
Reviewed-on: https://go-review.googlesource.com/c/go/+/250940
Run-TryBot: Emmanuel Odeke <emm.odeke@gmail.com>
TryBot-Result: Go Bot <gobot@golang.org>
Trust: Keith Randall <khr@golang.org>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
The alias doesn't need to be marked go:notinheap. It gets its
notinheap-ness from the target type.
Without this change, the type alias test in the notinheap.go file
generates these two errors:
notinheap.go:62: misplaced compiler directive
notinheap.go:63: type nih must be go:notinheap
The first is a result of go:notinheap pragmas not applying
to type alias declarations.
The second is the result of then trying to match the notinheap-ness
of the alias and the target type.
Add a few more go:notinheap tests while we are here.
Update #40954
Change-Id: I067ec47698df6e9e593e080d67796fd05a1d480f
Reviewed-on: https://go-review.googlesource.com/c/go/+/250939
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Emmanuel Odeke <emm.odeke@gmail.com>
Trust: Keith Randall <khr@golang.org>
This allows the global initializers function to go through normal
mid-end optimizations (e.g., inlining, escape analysis) like any other
function.
Updates #33485.
Change-Id: I9bcfe98b8628d1aca09b4c238d8d3b74c69010a5
Reviewed-on: https://go-review.googlesource.com/c/go/+/254839
Reviewed-by: Keith Randall <khr@golang.org>
Trust: Matthew Dempsky <mdempsky@google.com>
runtime.GC() doesn't guarantee the finalizer has run, so use a channel
instead to make sure finalizer was run in call to "after()".
Fixes#41361
Change-Id: I69c801e29aea49757ea72c52e8db13239de19ddc
Reviewed-on: https://go-review.googlesource.com/c/go/+/254401
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Trust: Cuong Manh Le <cuong.manhle.vn@gmail.com>
So we can insert theses OVARLIVE nodes right after OpStaticCall in SSA.
This helps fixing issue that unsafe-uintptr arguments are not kept alive
during return statement, or can be kept alive longer than expected.
Fixes#24491
Change-Id: Ic04a5d1bbb5c90dcfae65bd95cdd1da393a66800
Reviewed-on: https://go-review.googlesource.com/c/go/+/254397
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
This CL changes "T literal.M" error message to "T{...}.M". It's clearer
expression and focusing user on actual issue.
Updates #38745
Change-Id: I84b455a86742f37e0bde5bf390aa02984eecc3c9
Reviewed-on: https://go-review.googlesource.com/c/go/+/253677
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Currently, the statement:
go g(uintptr(f()))
gets rewritten into:
tmp := f()
newproc(8, g, uintptr(tmp))
runtime.KeepAlive(tmp)
which doesn't guarantee that tmp is still alive by time the g call is
scheduled to run.
This CL fixes the issue, by wrapping g call in a closure:
go func(p unsafe.Pointer) {
g(uintptr(p))
}(f())
then this will be rewritten into:
tmp := f()
go func(p unsafe.Pointer) {
g(uintptr(p))
runtime.KeepAlive(p)
}(tmp)
runtime.KeepAlive(tmp) // superfluous, but harmless
So the unsafe.Pointer p will be kept alive at the time g call runs.
Updates #24491
Change-Id: Ic10821251cbb1b0073daec92b82a866c6ebaf567
Reviewed-on: https://go-review.googlesource.com/c/go/+/253457
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The order array was zero initialized by the compiler, but ends up being
overwritten by the runtime anyway.
So let the runtime takes full responsibility for initializing, save us
one instruction per select.
Fixes#40399
Change-Id: Iec1eca27ad7180d4fcb3cc9ef97348206b7fe6b8
Reviewed-on: https://go-review.googlesource.com/c/go/+/251517
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
The current command will run this entire set of tests, which takes a
noticeable amount of time. Contributors may wish to run only a subset of
these tests to save time/compute (e.g. when iterating on a CL that
failed tests in that subset). Listing file(s) as operands to the command
will run only those tests.
Change-Id: I1874c43681a594190bc40b61cee0b8d321be73f8
Reviewed-on: https://go-review.googlesource.com/c/go/+/242997
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
Flag constant Ops on arm and arm64 are under refactoring, this change adds
a couple of testcases that verify the behavior of 'noov' branches.
Updates #39505
Updates #38740
Updates #39303
Change-Id: I493344b52276900cd296c32da494d72932dfc9be
Reviewed-on: https://go-review.googlesource.com/c/go/+/238677
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
We're using sort.SliceStable, so no need to keep track of indexes as well.
Use a more robust test for whether a node is a call.
Add a test that we're actually reordering comparisons. This test fails
without the alg.go changes in this CL because eqstring uses OCALLFUNC
instead of OCALL for its data comparisons.
Update #8606
Change-Id: Ieeec33434c72e3aa328deb11cc415cfda05632e2
Reviewed-on: https://go-review.googlesource.com/c/go/+/237921
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
This merges an lis + subf into subfic, and for 32b constants
lwa + subf into oris + ori + subf.
The carry bit is no longer used in code generation, therefore
I think we can clobber it as needed. Note, lowered borrow/carry
arithmetic is self-contained and thus is not affected.
A few extra rules are added to ensure early transformations to
SUBFCconst don't trip up earlier rules, fold constant operations,
or otherwise simplify lowering. Likewise, tests are added to
ensure all rules are hit. Generic constant folding catches
trivial cases, however some lowering rules insert arithmetic
which can introduce new opportunities (e.g BitLen or Slicemask).
I couldn't find a specific benchmark to demonstrate noteworthy
improvements, but this is generating subfic in many of the default
bent test binaries, so we are at least saving a little code space.
Change-Id: Iad7c6e5767eaa9dc24dc1c989bd1c8cfe1982012
Reviewed-on: https://go-review.googlesource.com/c/go/+/249461
Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com>
Right now we just prevent such types from being on the heap. This CL
makes it so they cannot appear on the stack either. The distinction
between heap and stack is pretty vague at the language level (e.g. it
is affected by -N), and we don't need the flexibility anyway.
Once go:notinheap types cannot be in either place, we don't need to
consider pointers to such types to be pointers, at least according to
the garbage collector and stack copying. (This is the big win of this
CL, in my opinion.)
The distinction between HasPointers and HasHeapPointer no longer
exists. There is only HasPointers.
This CL is cleanup before possible use of go:notinheap to fix#40954.
Update #13386
Change-Id: Ibd895aadf001c0385078a6d4809c3f374991231a
Reviewed-on: https://go-review.googlesource.com/c/go/+/249917
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
With this patch, opt pass can expose more obvious constant-folding
opportunites.
Example:
func test(i int) int {return (i+8)-(i+4)}
The previous version:
MOVD "".i(FP), R0
ADD $8, R0, R1
ADD $4, R0, R0
SUB R0, R1, R0
MOVD R0, "".~r1+8(FP)
RET (R30)
The optimized version:
MOVD $4, R0
MOVD R0, "".~r1+8(FP)
RET (R30)
This patch removes some existing reassociation rules, such as "x+(z-C)",
because the current generic rewrite rules will canonicalize "x-const"
to "x+(-const)", making "x+(z-C)" equal to "x+(z+(-C))".
This patch also adds test cases.
Change-Id: I857108ba0b5fcc18a879eeab38e2551bc4277797
Reviewed-on: https://go-review.googlesource.com/c/go/+/237137
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
go1.14 drop nacl support, as go1.15 was released, go1.13 is not
supported anymore, nacl is absolutely gone.
Change-Id: I05efb46891ec875b08da8f2996751a8e9cb57d0c
Reviewed-on: https://go-review.googlesource.com/c/go/+/249977
Run-TryBot: Cuong Manh Le <cuong.manhle.vn@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
checkptr has code to recognize &^ expressions, but it didn't take into
account that "p &^ x" gets rewritten to "p & ^x" during walk, which
resulted in false positive diagnostics.
This CL changes walkexpr to mark OANDNOT expressions with Implicit
when they're rewritten to OAND, so that walkCheckPtrArithmetic can
still recognize them later.
It would be slightly more idiomatic to instead mark the OBITNOT
expression as Implicit (as it's a compiler-generated Node), but the
OBITNOT expression might get constant folded. It's not worth the extra
complexity/subtlety of relying on n.Right.Orig, so we set Implicit on
the OAND node instead.
To atone for this transgression, I add documentation for nodeImplicit.
Fixes#40917.
Change-Id: I386304171ad299c530e151e5924f179e9a5fd5b8
Reviewed-on: https://go-review.googlesource.com/c/go/+/249477
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Cuong Manh Le <cuong.manhle.vn@gmail.com>
Add a new lowering rule to match and replace such instances
with the MADDLD instruction available on power9 where
possible.
Likewise, this plumbs in a new ppc64 ssa opcode to house
the newly generated MADDLD instructions.
When testing ed25519, this reduced binary size by 936B.
Similarly, MADDLD combination occcurs in a few other less
obvious cases such as division by constant.
Testing of golang.org/x/crypto/ed25519 shows non-trivial
speedup during keygeneration:
name old time/op new time/op delta
KeyGeneration 65.2µs ± 0% 63.1µs ± 0% -3.19%
Signing 64.3µs ± 0% 64.4µs ± 0% +0.16%
Verification 147µs ± 0% 147µs ± 0% +0.11%
Similarly, this test binary has shrunk by 66488B.
Change-Id: I077aeda7943119b41f07e4e62e44a648f16e4ad0
Reviewed-on: https://go-review.googlesource.com/c/go/+/248723
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 of the existing optimizations aren't triggered because they
are handled by the generic rules so this CL removes them. Also
some constraints were copied without much thought from the amd64
rules and they don't make sense on s390x, so we remove those
constraints.
Finally, add a 'multiply by the sum of two powers of two'
optimization. This makes sense on s390x as shifts are low latency
and can also sometimes be optimized further (especially if we add
support for RISBG instructions).
name old time/op new time/op delta
IntMulByConst/3-8 1.70ns ±11% 1.10ns ± 5% -35.26% (p=0.000 n=10+10)
IntMulByConst/5-8 1.64ns ± 7% 1.10ns ± 4% -32.94% (p=0.000 n=10+9)
IntMulByConst/12-8 1.65ns ± 6% 1.20ns ± 4% -27.16% (p=0.000 n=10+9)
IntMulByConst/120-8 1.66ns ± 4% 1.22ns ±13% -26.43% (p=0.000 n=10+10)
IntMulByConst/-120-8 1.65ns ± 7% 1.19ns ± 4% -28.06% (p=0.000 n=9+10)
IntMulByConst/65537-8 0.86ns ± 9% 1.12ns ±12% +30.41% (p=0.000 n=10+10)
IntMulByConst/65538-8 1.65ns ± 5% 1.23ns ± 5% -25.11% (p=0.000 n=10+10)
Change-Id: Ib196e6bff1e97febfd266134d0a2b2a62897989f
Reviewed-on: https://go-review.googlesource.com/c/go/+/248937
Run-TryBot: Michael Munday <mike.munday@ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>
For an unsigned integer, it's useful to convert its order test with 0/1
to its equality test with 0. We can save a comparison instruction that
followed by a conditional branch on arm64 since it supports
compare-with-zero-and-branch instructions. For example,
if x > 0 { ... } else { ... }
the original version:
CMP $0, R0
BLS 9
the optimized version:
CBZ R0, 8
Updates #21439
Change-Id: Id1de6f865f6aa72c5d45b29f7894818857288425
Reviewed-on: https://go-review.googlesource.com/c/go/+/246857
Reviewed-by: Keith Randall <khr@golang.org>
