diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml index ba31cc9..efe0baa 100644 --- a/.github/workflows/test.yml +++ b/.github/workflows/test.yml @@ -61,8 +61,13 @@ jobs: - name: Test tsvc (amd64, SIMD) run: make tsvc-test - - name: Install qemu-user-static - run: sudo apt-get update && sudo apt-get install -y qemu-user-static + # ubuntu-26.04 has no qemu-user-static: qemu-user ships the static + # emulators there, without the -static names the Makefile runs. + - name: Install qemu-user + run: | + sudo apt-get update && sudo apt-get install -y qemu-user + command -v qemu-aarch64-static || sudo ln -s "$(command -v qemu-aarch64)" /usr/local/bin/qemu-aarch64-static + qemu-aarch64-static --version - name: Test tsvc (arm64, qemu) run: make tsvc-test-qemu-arm64 diff --git a/BUGS.md b/BUGS.md index 843087e..55231c5 100644 --- a/BUGS.md +++ b/BUGS.md @@ -2,11 +2,11 @@ loopvec rewrites code to use `GOEXPERIMENT=simd`, which is new, so most of what it hits is in the Go toolchain. This file lists the Go issues that affect loopvec. "tip" means -gotip, "1.27.1" is the release toolchain loopvec is tested against. +gotip, "1.27.3" is the release toolchain loopvec is tested against. | Issue | Status | Affects | loopvec's answer | |---|---|---|---| -| [#80657](https://github.com/golang/go/issues/80657) simd in a method: internal compiler error | fixed on tip (CL 839405), still crashes on 1.27.1 | `-methods` | methods are skipped unless `-methods` is passed | +| [#80657](https://github.com/golang/go/issues/80657) simd in a method: internal compiler error | fixed on tip (CL 839405), still crashes on 1.27.3 | `-methods` | methods are skipped unless `-methods` is passed | | [#80835](https://github.com/golang/go/issues/80835) legacy SSE encodings in functions using simd cause AVX-SSE transition penalties | open, fix pending (CLs 825145, 825146, in tip) | speed of any scalar float code that runs after a rewritten loop, in the same function or another | none; the portable `simd` package has no `ClearAVXUpperBits` | | [#81844](https://github.com/golang/go/issues/81844) blank identifier parameters rejected | fixed on tip (CL 841145) | every rewrite of a function with a `_` parameter | the generated file renames them `_p0`, `_p1`, ... | | [#81845](https://github.com/golang/go/issues/81845) calls left unresolved when inlined into a package initializer | closed as a duplicate of [#80689](https://github.com/golang/go/issues/80689), which is open | loops in package-level `var x = func() {...}()` | not rewritten (the generated initializer fails to link), see `package_initializer_no_rewrite.txt` and `toolexec_package_initializer.txt` | @@ -19,7 +19,7 @@ gotip, "1.27.1" is the release toolchain loopvec is tested against. A function with a receiver that contains simd code failed to compile with `internal compiler error: missing Types entry`. loopvec does not rewrite loops in methods -by default; `-methods` turns it on. The fix is on tip. On 1.27.1 the crash is still there, +by default; `-methods` turns it on. The fix is on tip. On 1.27.3 the crash is still there, and `flag_methods_ice.txt` pins it, so that test starts failing when the pinned toolchain gets the fix. Reported by someone else; the fix is [CL 839405](https://go.dev/cl/839405). diff --git a/Makefile b/Makefile index b4226b3..3804653 100644 --- a/Makefile +++ b/Makefile @@ -1,4 +1,4 @@ -export GOTOOLCHAIN ?= go1.27.1 +export GOTOOLCHAIN ?= go1.27.3 .DEFAULT_GOAL := build diff --git a/go.mod b/go.mod index ed04302..9457290 100644 --- a/go.mod +++ b/go.mod @@ -3,7 +3,7 @@ module github.com/mauri870/loopvec go 1.27.1 require ( - golang.org/x/tools v0.51.0 + golang.org/x/tools v0.52.1-0.20261030190145-0f433775b36d rsc.io/script v0.0.2 ) @@ -207,10 +207,10 @@ require ( go.uber.org/zap v1.27.0 // indirect go.yaml.in/yaml/v3 v3.0.5 // indirect golang.org/x/exp/typeparams v0.0.0-20260811152304-ee035b5b010f // indirect - golang.org/x/mod v0.41.0 // indirect - golang.org/x/sync v0.23.0 // indirect - golang.org/x/sys v0.48.0 // indirect - golang.org/x/telemetry v0.0.0-20260924152758-ed294f943157 // indirect + golang.org/x/mod v0.42.0 // indirect + golang.org/x/sync v0.24.0 // indirect + golang.org/x/sys v0.49.0 // indirect + golang.org/x/telemetry v0.0.0-20261009151841-97e8ea77aa2a // indirect golang.org/x/text v0.41.0 // indirect google.golang.org/protobuf v1.36.12 // indirect gopkg.in/ini.v1 v1.67.0 // indirect diff --git a/go.sum b/go.sum index 04c2946..bb2b241 100644 --- a/go.sum +++ b/go.sum @@ -688,8 +688,8 @@ golang.org/x/mod v0.4.1/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA= golang.org/x/mod v0.6.0-dev.0.20220106191415-9b9b3d81d5e3/go.mod h1:3p9vT2HGsQu2K1YbXdKPJLVgG5VJdoTa1poYQBtP1AY= golang.org/x/mod v0.6.0-dev.0.20220419223038-86c51ed26bb4/go.mod h1:jJ57K6gSWd91VN4djpZkiMVwK6gcyfeH4XE8wZrZaV4= golang.org/x/mod v0.8.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs= -golang.org/x/mod v0.41.0 h1:qJmnOUb4YB+FsEuM3HcWucdZASCPGhsX6uljO6pog0c= -golang.org/x/mod v0.41.0/go.mod h1:Ek9pY8RKWXwsWvd3rQiHYtMqkjSUV+s1Rj7j4H5Ur6o= +golang.org/x/mod v0.42.0 h1:ICeL3t2L30ljSPHHxDnR4fqVcBt/zR2zJAb9juavmQg= +golang.org/x/mod v0.42.0/go.mod h1:u5fzIrRInkL8MaO1cbRkUHT64v303YIT5KEqpNr7jRs= golang.org/x/net v0.0.0-20180724234803-3673e40ba225/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4= golang.org/x/net v0.0.0-20180826012351-8a410e7b638d/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4= golang.org/x/net v0.0.0-20181114220301-adae6a3d119a/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4= @@ -724,8 +724,8 @@ golang.org/x/net v0.0.0-20210525063256-abc453219eb5/go.mod h1:9nx3DQGgdP8bBQD5qx golang.org/x/net v0.0.0-20211015210444-4f30a5c0130f/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y= golang.org/x/net v0.0.0-20220722155237-a158d28d115b/go.mod h1:XRhObCWvk6IyKnWLug+ECip1KBveYUHfp+8e9klMJ9c= golang.org/x/net v0.6.0/go.mod h1:2Tu9+aMcznHK/AK1HMvgo6xiTLG5rD5rZLDS+rp2Bjs= -golang.org/x/net v0.59.0 h1:5zfYln+w5XCxwrnMMJPufRgNoXEaGxl0wo5GqPXyues= -golang.org/x/net v0.59.0/go.mod h1:2DA/G1UfVbCpQPeWTmMPGY7Cs2PkBkwu743bVX5PIVg= +golang.org/x/net v0.61.0 h1:RR+6j/BTBCrggOLugxa+66R8FvoN6MHlzDlZIbN9YFw= +golang.org/x/net v0.61.0/go.mod h1:WgDCOTH8iwtB66T4/GfI94AHo/hhdb7Hah9NDseQfiI= golang.org/x/oauth2 v0.0.0-20180821212333-d2e6202438be/go.mod h1:N/0e6XlmueqKjAGxoOufVs8QHGRruUQn6yWY3a++T0U= golang.org/x/oauth2 v0.0.0-20190226205417-e64efc72b421/go.mod h1:gOpvHmFTYa4IltrdGE7lF6nIHvwfUNPOp7c8zoXwtLw= golang.org/x/oauth2 v0.0.0-20190604053449-0f29369cfe45/go.mod h1:gOpvHmFTYa4IltrdGE7lF6nIHvwfUNPOp7c8zoXwtLw= @@ -745,8 +745,8 @@ golang.org/x/sync v0.0.0-20201207232520-09787c993a3a/go.mod h1:RxMgew5VJxzue5/jJ golang.org/x/sync v0.0.0-20210220032951-036812b2e83c/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM= golang.org/x/sync v0.0.0-20220722155255-886fb9371eb4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM= golang.org/x/sync v0.1.0/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM= -golang.org/x/sync v0.23.0 h1:KameEIfc1IkluZyXWLn39Wd4tURc6GbCiISGiZm2bQk= -golang.org/x/sync v0.23.0/go.mod h1:sUUOizhqBxiL6pEWpqNLUiaJn1ShEbZ6BBqskPbjZm0= +golang.org/x/sync v0.24.0 h1:iqiDrGoDgyuvAKrhLQthUEzU/R378N78gXbBjxed7oA= +golang.org/x/sync v0.24.0/go.mod h1:sUUOizhqBxiL6pEWpqNLUiaJn1ShEbZ6BBqskPbjZm0= golang.org/x/sys v0.0.0-20180830151530-49385e6e1522/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY= golang.org/x/sys v0.0.0-20180905080454-ebe1bf3edb33/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY= golang.org/x/sys v0.0.0-20181116152217-5ac8a444bdc5/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY= @@ -790,10 +790,10 @@ golang.org/x/sys v0.0.0-20220114195835-da31bd327af9/go.mod h1:oPkhp1MJrh7nUepCBc golang.org/x/sys v0.0.0-20220520151302-bc2c85ada10a/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= golang.org/x/sys v0.0.0-20220722155257-8c9f86f7a55f/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= golang.org/x/sys v0.5.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= -golang.org/x/sys v0.48.0 h1:bbX/i/6MgT9BVLM9RT1thmxL04yeTAhbEz4SyadbXoo= -golang.org/x/sys v0.48.0/go.mod h1:hNLxWAXmnKAxqDtdwIYC4bM9oQPEecfsnNMuSxOs3og= -golang.org/x/telemetry v0.0.0-20260924152758-ed294f943157 h1:DftxASHKl6JolFPqoN+2WrFO53JvkHwRXvFqf8gsivA= -golang.org/x/telemetry v0.0.0-20260924152758-ed294f943157/go.mod h1:i+ivNqjDnTF3WTElsdk5g9V5DTSBYgdNo7xTU9SDwYA= +golang.org/x/sys v0.49.0 h1:XbzkgYJHdqh/8m2Uu0W/dQv8nktxx4BFHp1M0gROTXA= +golang.org/x/sys v0.49.0/go.mod h1:hNLxWAXmnKAxqDtdwIYC4bM9oQPEecfsnNMuSxOs3og= +golang.org/x/telemetry v0.0.0-20261009151841-97e8ea77aa2a h1:wnveMBKHwx8/nIb8WYT5P9vtefdXOvIWxzycCkc5n6Q= +golang.org/x/telemetry v0.0.0-20261009151841-97e8ea77aa2a/go.mod h1:i+ivNqjDnTF3WTElsdk5g9V5DTSBYgdNo7xTU9SDwYA= golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo= golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuXm/mYQcufACuNUgVhRMnK/tPxf8= golang.org/x/term v0.5.0/go.mod h1:jMB1sMXY+tzblOD4FWmEbocvup2/aLOaQEp7JmGp78k= @@ -857,8 +857,8 @@ golang.org/x/tools v0.1.1-0.20210302220138-2ac05c832e1a/go.mod h1:9bzcO0MWcOuT0t golang.org/x/tools v0.1.10/go.mod h1:Uh6Zz+xoGYZom868N8YTex3t7RhtHDBrE8Gzo9bV56E= golang.org/x/tools v0.1.12/go.mod h1:hNGJHUnrk76NpqgfD5Aqm5Crs+Hm0VOH/i9J2+nxYbc= golang.org/x/tools v0.6.0/go.mod h1:Xwgl3UAJ/d3gWutnCtw505GrjyAbvKui8lOU390QaIU= -golang.org/x/tools v0.51.0 h1:k4Xc/1Om9jwkBJBo4NVLMSARBoWtK10mx+W5BnXCeAI= -golang.org/x/tools v0.51.0/go.mod h1:9eEncMayCV6zRMGhR5eZEC2iBx98qWcF1HZ9Z7wJOoA= +golang.org/x/tools v0.52.1-0.20261030190145-0f433775b36d h1:F5ot9G6RM6pgx3TTdKF6IZa2nocit4tcsrVQlq4Zh5w= +golang.org/x/tools v0.52.1-0.20261030190145-0f433775b36d/go.mod h1:ItcP5UNaVGG+Yc4eIYmaFLNJ4soEopkx5Y2d4lcEurs= golang.org/x/tools/go/expect v0.1.1-deprecated h1:jpBZDwmgPhXsKZC6WhL20P4b/wmnpsEAGHaNy0n/rJM= golang.org/x/tools/go/expect v0.1.1-deprecated/go.mod h1:eihoPOH+FgIqa3FpoTwguz/bVUSGBlGQU67vpBeOrBY= golang.org/x/tools/go/packages/packagestest v0.1.1-deprecated h1:1h2MnaIAIXISqTFKdENegdpAgUXz6NrPEsbIeWaBRvM= diff --git a/internal/loopir/caps_gen.go b/internal/loopir/caps_gen.go index bccacfa..ba7ddff 100644 --- a/internal/loopir/caps_gen.go +++ b/internal/loopir/caps_gen.go @@ -1,4 +1,4 @@ -// Code generated by capsgen from the simd package of go1.27.1. DO NOT EDIT. +// Code generated by capsgen from the simd package of go1.27.3. DO NOT EDIT. package loopir diff --git a/internal/loopir/lower.go b/internal/loopir/lower.go index c4a7719..8d6eb6e 100644 --- a/internal/loopir/lower.go +++ b/internal/loopir/lower.go @@ -286,6 +286,12 @@ func (l *lowerer) bound(e ast.Expr) (bound, bool) { } b := bound{text: types.ExprString(e)} b.value, b.isConst = l.constant(e) + if b.isConst { + // A named constant or a constant expression (s0, 1 - 1) is spelled as its + // value, so the checks that read a constant start or limit see it and the + // emitted code never indexes at a negative constant they would have caught. + b.text = strconv.FormatInt(b.value, 10) + } if call, ok := ast.Unparen(e).(*ast.CallExpr); ok { b.slice, _ = l.lenOf(call) } @@ -532,6 +538,25 @@ func (l *lowerer) body(node ast.Stmt, block *ast.BlockStmt, trip Trip, step int6 } } + // A constant offset that puts an index below zero makes the scalar loop panic + // when it runs: on its first iteration when the start is a constant and the + // first index is negative, or on every iteration when the limit is a constant + // and even the last index is. It is left alone, since the vector loop would + // index or slice at a negative constant, which does not compile. + start, errStart := strconv.ParseInt(trip.Start, 10, 64) + limit, errLimit := strconv.ParseInt(trip.Limit, 10, 64) + negative := func(offset string) bool { + off, err := strconv.ParseInt(offset, 10, 64) + return err == nil && (errStart == nil && start+off < 0 || errLimit == nil && limit+off <= 0) + } + for _, stmt := range body { + for _, leaf := range Leaves(stmtValue(stmt)) { + if load, ok := leaf.(*Load); ok && negative(load.Off) { + return nil, ReasonNegativeIndex + } + } + } + loop := &Loop{ Node: node, Ind: Induction{Var: l.iv, Step: step, Trip: trip}, diff --git a/internal/loopir/reason.go b/internal/loopir/reason.go index aaeca53..4757287 100644 --- a/internal/loopir/reason.go +++ b/internal/loopir/reason.go @@ -42,6 +42,12 @@ const ( // iteration writes (a[i] = a[i-1] + x): iteration i reads what iteration i-1 // wrote, so the iterations cannot run as a vector. ReasonCarriedDependence Reason = "an iteration reads an element an earlier iteration wrote" + // ReasonNegativeIndex covers a slice read at a constant offset that puts an + // index below zero: the first index, with a constant start (a[i-1] in a loop + // from 0), or every index, with a constant limit (a[i-20] in a loop that stops + // before 16). The scalar loop panics when it runs, and the vector loop would + // index or slice at a negative constant, which does not compile. + ReasonNegativeIndex Reason = "a constant offset puts an index below zero" // ReasonUnsupportedIf covers an if that is not a select: a condition that is // not a comparison of loop values (joined by && and ||), an init statement, or // a branch that is not one assignment to an element of the same slice. diff --git a/testdata/offset_negative_constant.txt b/testdata/offset_negative_constant.txt new file mode 100644 index 0000000..d23cc26 --- /dev/null +++ b/testdata/offset_negative_constant.txt @@ -0,0 +1,90 @@ +# A read at a negative constant offset is rewritten while every index it reads can +# be zero or more. When an index is a negative constant, the first one in a loop +# with a constant start (a[i-1] from 0) or every one in a loop with a constant limit +# (a[i-16] in a loop that stops before 16), the scalar loop panics when it runs and +# the vector loop would index at a negative constant, which does not compile, so the +# loop is left alone, whether the start and limit are written as literals, named +# constants or constant expressions. The rewritten package must build. + +loopvec -json . +cmpenv stdout stdout.want + +loopvec -split . +cmpenv stderr stderr.want +env GOEXPERIMENT=simd +exec go build . + +-- go.mod -- +module example.com/test + +go 1.27.1 +-- ops.go -- +package test + +func FirstIndexZeroConstant(dst, a []int32) { + for i := 15; i < 31; i++ { + dst[i] = a[i-15] + 1 + } +} + +func EveryIndexNegative(dst, a []int32) { + for i := 0; i < 16; i++ { + dst[i] = a[i-16] + 1 + } +} + +func FarBelow(dst, a []float32) { + for i := 0; i < 16; i++ { + dst[i] = a[i-20] * a[i] + } +} + +func FirstIndexNegative(dst, a []int32, n int) { + for i := 0; i < n; i++ { + dst[i] = a[i-1] * 3 + } +} + +func FirstIndexZero(dst, a []int32, n int) { + for i := 1; i < n; i++ { + dst[i] = a[i-1] * 3 + } +} + +const s0, lim = 0, 16 + +func NamedLimit(dst, a []int32, k int) { + for i := k; i < lim; i++ { + dst[i] = a[i-20] + 1 + } +} + +func NamedStart(dst, a []int32, n int) { + for i := s0; i < n; i++ { + dst[i] = a[i-1] * 3 + } +} + +func ExpressionStart(dst, a []int32, n int) { + for i := 1 - 1; i < n; i++ { + dst[i] = a[i-1] * 3 + } +} + +func NamedStartCopy(dst, a []int32, n int) { + for i := s0; i < n; i++ { + dst[i] = a[i-1] + } +} +-- stdout.want -- +{"file":"$WORK/ops.go","line":4,"func":"FirstIndexZeroConstant","vectorized":true} +{"file":"$WORK/ops.go","line":10,"func":"EveryIndexNegative","vectorized":false,"reason":"a constant offset puts an index below zero","stage":"lower"} +{"file":"$WORK/ops.go","line":16,"func":"FarBelow","vectorized":false,"reason":"a constant offset puts an index below zero","stage":"lower"} +{"file":"$WORK/ops.go","line":22,"func":"FirstIndexNegative","vectorized":false,"reason":"a constant offset puts an index below zero","stage":"lower"} +{"file":"$WORK/ops.go","line":28,"func":"FirstIndexZero","vectorized":true} +{"file":"$WORK/ops.go","line":36,"func":"NamedLimit","vectorized":false,"reason":"a constant offset puts an index below zero","stage":"lower"} +{"file":"$WORK/ops.go","line":42,"func":"NamedStart","vectorized":false,"reason":"a constant offset puts an index below zero","stage":"lower"} +{"file":"$WORK/ops.go","line":48,"func":"ExpressionStart","vectorized":false,"reason":"a constant offset puts an index below zero","stage":"lower"} +{"file":"$WORK/ops.go","line":54,"func":"NamedStartCopy","vectorized":false,"reason":"a constant offset puts an index below zero","stage":"lower"} +-- stderr.want -- +loopvec: rewrote 2 loop(s) in $WORK/ops.go