fix(verify): fix flaky JIT tests with global buffers and KeepAlive

verify/abi_test.go

@@ -4,6 +4,7 @@
 package verify
 
 import (
+	"runtime"
 	"testing"
 	"unsafe"
 )
@@ -100,6 +101,8 @@ func TestABILZ4Kernels(t *testing.T) {
 	PutUint64(args, 40, 128)
 
 	_, report, err := k.CallFuncChecked("wideCopyAVX2", args)
+	runtime.KeepAlive(dst)
+	runtime.KeepAlive(src)
 	if err != nil {
 		t.Fatalf("CallFuncChecked(wideCopyAVX2): %v", err)
 	}
@@ -120,6 +123,8 @@ func TestABILZ4Kernels(t *testing.T) {
 	PutUint64(decArgs, 40, uint64(cap(decDst)))
 
 	_, report, err = k.CallFuncChecked("decodeBlockAVX2", decArgs)
+	runtime.KeepAlive(decSrc)
+	runtime.KeepAlive(decDst)
 	if err != nil {
 		t.Fatalf("CallFuncChecked(decodeBlockAVX2): %v", err)
 	}

verify/avx512_test.go

@@ -7,6 +7,7 @@ import (
 	"bytes"
 	"math/rand"
 	"os"
+	"runtime"
 	"testing"
 	"unsafe"
 )
@@ -53,6 +54,8 @@ func TestAVX512DecodeKnownAnswers(t *testing.T) {
 			PutUint64(args, 40, uint64(cap(dst)))
 
 			out, err := k.CallFunc("decodeBlockAVX512", args)
+			runtime.KeepAlive(tt.src)
+			runtime.KeepAlive(dst)
 			if err != nil {
 				t.Fatalf("CallFunc: %v", err)
 			}
@@ -91,6 +94,8 @@ func TestAVX512DifferentialFuzz(t *testing.T) {
 		PutUint64(args, 40, uint64(cap(jitDst)))
 
 		out, err := k.CallFunc("decodeBlockAVX512", args)
+		runtime.KeepAlive(src)
+		runtime.KeepAlive(jitDst)
 		if err != nil {
 			t.Fatalf("iter %d: %v", i, err)
 		}
@@ -134,6 +139,8 @@ func TestAVX512WideCopy(t *testing.T) {
 		PutUint64(args, 40, uint64(n))
 
 		_, err := k.CallFunc("wideCopyAVX512", args)
+		runtime.KeepAlive(dst)
+		runtime.KeepAlive(src)
 		if err != nil {
 			t.Fatalf("wideCopyAVX512(n=%d): %v", n, err)
 		}

verify/coverage_test.go

@@ -4,6 +4,7 @@
 package verify
 
 import (
+	"runtime"
 	"testing"
 	"unsafe"
 )
@@ -44,6 +45,7 @@ func TestProfilePaths(t *testing.T) {
 
 	// Build a corpus of varied LZ4 blocks.
 	var argSets [][]byte
+	var liveDsts [][]byte // keep dst buffers alive across ProfilePaths
 	blocks := []struct {
 		src     []byte
 		dstSize int
@@ -63,6 +65,7 @@ func TestProfilePaths(t *testing.T) {
 		PutUint64(args, 8, uint64(len(b.src)))
 		PutUint64(args, 16, uint64(cap(b.src)))
 		dst := make([]byte, b.dstSize)
+		liveDsts = append(liveDsts, dst)
 		if len(dst) > 0 {
 			PutPtr(args, 24, unsafe.Pointer(&dst[0]))
 		}
@@ -73,6 +76,8 @@ func TestProfilePaths(t *testing.T) {
 
 	// Result offsets: n+48 and code+56.
 	paths, err := k.ProfilePaths("decodeBlockAVX2", argSets, []int{48, 56})
+	runtime.KeepAlive(blocks)
+	runtime.KeepAlive(liveDsts)
 	if err != nil {
 		t.Fatalf("ProfilePaths: %v", err)
 	}

verify/differential_test.go

@@ -6,6 +6,7 @@ package verify
 import (
 	"bytes"
 	"math/rand"
+	"runtime"
 	"testing"
 	"unsafe"
 )
@@ -288,6 +289,8 @@ func callDecodeBlockAVX2Raw(t *testing.T, k *Kernel, src, dst []byte) (int, int)
 	PutUint64(args, 40, uint64(cap(dst)))
 
 	out, err := k.CallFunc("decodeBlockAVX2", args)
+	runtime.KeepAlive(src)
+	runtime.KeepAlive(dst)
 	if err != nil {
 		t.Fatalf("CallFunc(decodeBlockAVX2): %v", err)
 	}

verify/flac_test.go

@@ -7,6 +7,7 @@ import (
 	"bytes"
 	"math/rand"
 	"os"
+	"runtime"
 	"testing"
 	"unsafe"
 )
@@ -157,19 +158,51 @@ func decodeStereo16Go(src []byte, left, right []int32) {
 
 // --- Differential tests ---
 
+// Buffers whose addresses are passed to JIT code via unsafe.Pointer are
+// package-level globals: their addresses are stable and the GC never moves
+// or collects them, unlike per-iteration make() buffers (which caused flaky
+// stale reads, especially under -race).  Sizes cover the maximum n from the
+// rng.Intn(N) call in each test, rounded up.
+var (
+	mono16Src    [512]byte  // TestFLACDecodeMono16: max n = 255 → 2*n = 510
+	mono16JitDst [256]int32 // max n = 255
+
+	pack16Src    [256]int32 // TestFLACPack16: max n = 255
+	pack16JitDst [512]byte  // max 2*n = 510
+
+	decorLeft   [128]int32 // TestFLACDecorrelate: max n = 127
+	decorRight  [128]int32 // max n = 127
+	decorJitOut [256]int32 // max 2*n = 254
+
+	mono24Src    [768]byte  // TestFLACDecodeMono24: max n = 255 → 3*n = 765
+	mono24JitDst [256]int32 // max n = 255
+
+	stereo16Src      [1024]byte // TestFLACDecodeStereo16: max n = 255 → 4*n = 1020
+	stereo16JitLeft  [256]int32 // max n = 255
+	stereo16JitRight [256]int32 // max n = 255
+
+	fastStereoLeft  [256]int32 // TestFLACFastStereoSums: max n = 255
+	fastStereoRight [256]int32 // max n = 255
+)
+
 func TestFLACDecodeMono16(t *testing.T) {
 	k := loadFLACKernel(t)
 	rng := rand.New(rand.NewSource(7))
 
+	src := mono16Src[:0]
+	jitDst := mono16JitDst[:0]
 	for iter := 0; iter < 500; iter++ {
 		n := rng.Intn(256)
-		src := make([]byte, 2*n)
+		src = src[:2*n]
 		rng.Read(src)
 
 		goDst := make([]int32, n)
 		decodeMono16Go(src, goDst)
 
-		jitDst := make([]int32, n)
+		jitDst = jitDst[:n]
+		for i := range jitDst {
+			jitDst[i] = 0
+		}
 		args := make([]byte, 48)
 		if len(src) > 0 {
 			PutPtr(args, 0, unsafe.Pointer(&src[0]))
@@ -183,6 +216,8 @@ func TestFLACDecodeMono16(t *testing.T) {
 		PutUint64(args, 40, uint64(cap(jitDst)))
 
 		_, err := k.CallFunc("decodeMono16AVX2", args)
+		runtime.KeepAlive(src)
+		runtime.KeepAlive(jitDst)
 		if err != nil {
 			t.Fatalf("iter %d: %v", iter, err)
 		}
@@ -198,9 +233,11 @@ func TestFLACPack16(t *testing.T) {
 	k := loadFLACKernel(t)
 	rng := rand.New(rand.NewSource(13))
 
+	src := pack16Src[:0]
+	jitDst := pack16JitDst[:0]
 	for iter := 0; iter < 500; iter++ {
 		n := rng.Intn(256)
-		src := make([]int32, n)
+		src = src[:n]
 		for i := range src {
 			src[i] = int32(rng.Intn(65536) - 32768)
 		}
@@ -208,7 +245,10 @@ func TestFLACPack16(t *testing.T) {
 		goDst := make([]byte, 2*n)
 		pack16Go(goDst, src)
 
-		jitDst := make([]byte, 2*n)
+		jitDst = jitDst[:2*n]
+		for i := range jitDst {
+			jitDst[i] = 0
+		}
 		args := make([]byte, 48)
 		if len(jitDst) > 0 {
 			PutPtr(args, 0, unsafe.Pointer(&jitDst[0]))
@@ -222,6 +262,8 @@ func TestFLACPack16(t *testing.T) {
 		PutUint64(args, 40, uint64(cap(src)))
 
 		_, err := k.CallFunc("pack16AVX2", args)
+		runtime.KeepAlive(jitDst)
+		runtime.KeepAlive(src)
 		if err != nil {
 			t.Fatalf("iter %d: %v", iter, err)
 		}
@@ -247,10 +289,13 @@ func TestFLACDecorrelate(t *testing.T) {
 
 	for _, kk := range kernels {
 		t.Run(kk.name, func(t *testing.T) {
+			left := decorLeft[:0]
+			right := decorRight[:0]
+			jitOut := decorJitOut[:0]
 			for iter := 0; iter < 200; iter++ {
 				n := rng.Intn(128)
-				left := make([]int32, n)
+				left = left[:n]
-				right := make([]int32, n)
+				right = right[:n]
 				for i := range left {
 					left[i] = int32(rng.Intn(1<<24) - 1<<23)
 					right[i] = int32(rng.Intn(1<<24) - 1<<23)
@@ -259,7 +304,10 @@ func TestFLACDecorrelate(t *testing.T) {
 				goOut := make([]int32, 2*n)
 				kk.ref(left, right, goOut)
 
-				jitOut := make([]int32, 2*n)
+				jitOut = jitOut[:2*n]
+				for i := range jitOut {
+					jitOut[i] = 0
+				}
 				args := make([]byte, 72)
 				if n > 0 {
 					PutPtr(args, 0, unsafe.Pointer(&left[0]))
@@ -274,6 +322,9 @@ func TestFLACDecorrelate(t *testing.T) {
 				PutUint64(args, 64, uint64(cap(jitOut)))
 
 				_, err := k.CallFunc(kk.name, args)
+				runtime.KeepAlive(left)
+				runtime.KeepAlive(right)
+				runtime.KeepAlive(jitOut)
 				if err != nil {
 					t.Fatalf("iter %d: %v", iter, err)
 				}
@@ -291,9 +342,11 @@ func TestFLACAnalyzeO1Range(t *testing.T) {
 	k := loadFLACKernel(t)
 	rng := rand.New(rand.NewSource(33))
 
+	swin := analyzeSwin[:0]
+	jitDstP := analyzeJitDstP[:0]
 	for iter := 0; iter < 300; iter++ {
 		n := 1 + rng.Intn(128) // partition size
-		swin := make([]int32, n+1)
+		swin = swin[:n+1]
 		for i := range swin {
 			swin[i] = int32(rng.Intn(1<<20) - 1<<19)
 		}
@@ -302,8 +355,12 @@ func TestFLACAnalyzeO1Range(t *testing.T) {
 		var goHist [32]uint16
 		goSum, goOvf := analyzeO1RangeGo(swin, goDstP, &goHist)
 
-		jitDstP := make([]uint32, n)
+		jitDstP = jitDstP[:n]
-		var jitHist [32]uint16
+		for i := range jitDstP {
+			jitDstP[i] = 0
+		}
+		jitHist := &analyzeJitHist
+		*jitHist = [32]uint16{}  // reset for this iteration
 		args := make([]byte, 72) // 65 rounded up
 		PutPtr(args, 0, unsafe.Pointer(&swin[0]))
 		PutUint64(args, 8, uint64(len(swin)))
@@ -314,6 +371,9 @@ func TestFLACAnalyzeO1Range(t *testing.T) {
 		PutPtr(args, 48, unsafe.Pointer(&jitHist[0]))
 
 		out, err := k.CallFunc("analyzeO1RangeAVX2", args)
+		runtime.KeepAlive(swin)
+		runtime.KeepAlive(jitDstP)
+		runtime.KeepAlive(jitHist)
 		if err != nil {
 			t.Fatalf("iter %d: %v", iter, err)
 		}
@@ -331,8 +391,8 @@ func TestFLACAnalyzeO1Range(t *testing.T) {
 				t.Fatalf("iter %d: dstP[%d] mismatch: JIT=%d Go=%d", iter, i, jitDstP[i], goDstP[i])
 			}
 		}
-		if jitHist != goHist {
+		if *jitHist != goHist {
-			t.Fatalf("iter %d: hist mismatch: JIT=%v Go=%v", iter, jitHist, goHist)
+			t.Fatalf("iter %d: hist mismatch: JIT=%v Go=%v", iter, *jitHist, goHist)
 		}
 	}
 }
@@ -341,10 +401,12 @@ func TestFLACFastStereoSums(t *testing.T) {
 	k := loadFLACKernel(t)
 	rng := rand.New(rand.NewSource(44))
 
+	left := fastStereoLeft[:0]
+	right := fastStereoRight[:0]
 	for iter := 0; iter < 300; iter++ {
 		n := 1 + rng.Intn(256)
-		left := make([]int32, n)
+		left = left[:n]
-		right := make([]int32, n)
+		right = right[:n]
 		for i := range left {
 			left[i] = int32(rng.Intn(1<<24) - 1<<23)
 			right[i] = int32(rng.Intn(1<<24) - 1<<23)
@@ -363,7 +425,8 @@ func TestFLACFastStereoSums(t *testing.T) {
 			goSums[3] += foldAbs(mid) + foldAbs(side)
 		}
 
-		var jitSums [4]uint64
+		jitSums := &analyzeJitSums
+		*jitSums = [4]uint64{} // reset for this iteration
 		args := make([]byte, 56)
 		PutPtr(args, 0, unsafe.Pointer(&left[0]))
 		PutUint64(args, 8, uint64(n))
@@ -374,11 +437,14 @@ func TestFLACFastStereoSums(t *testing.T) {
 		PutPtr(args, 48, unsafe.Pointer(&jitSums[0]))
 
 		_, err := k.CallFunc("fastStereoSumsAVX2", args)
+		runtime.KeepAlive(left)
+		runtime.KeepAlive(right)
+		runtime.KeepAlive(jitSums)
 		if err != nil {
 			t.Fatalf("iter %d: %v", iter, err)
 		}
-		if jitSums != goSums {
+		if *jitSums != goSums {
-			t.Fatalf("iter %d: sums mismatch:\n  JIT=%v\n  Go =%v", iter, jitSums, goSums)
+			t.Fatalf("iter %d: sums mismatch:\n  JIT=%v\n  Go =%v", iter, *jitSums, goSums)
 		}
 	}
 }
@@ -388,12 +454,27 @@ func foldAbs(v int32) uint64 {
 }
 
 // runAnalyzeTest is the shared harness for the analyzeO*Range family.
+// analyzeJitHist is a shared histogram buffer for the analyze tests.
+// It is package-level so its address is stable across all calls.
+var analyzeJitHist [32]uint16
+
+// analyzeJitSums is a shared sums buffer for the fastStereoSums test.
+var analyzeJitSums [4]uint64
+
+// analyzeSwin and analyzeJitDstP are shared input/output buffers for the
+// analyze tests.  Package-level so their addresses are stable.
+var analyzeSwin [136]int32 // max n(128) + max order(4) = 132, rounded up
+var analyzeJitDstP [128]uint32
+
 func runAnalyzeTest(t *testing.T, k *Kernel, name string, order int, ref func([]int32, []uint32, *[32]uint16) (uint64, bool)) {
 	t.Helper()
 	rng := rand.New(rand.NewSource(int64(order)*100 + 7))
+	jitHist := &analyzeJitHist
+	swin := analyzeSwin[:0]
+	jitDstP := analyzeJitDstP[:0]
 	for iter := 0; iter < 200; iter++ {
 		n := 1 + rng.Intn(128)
-		swin := make([]int32, n+order)
+		swin = swin[:n+order]
 		for i := range swin {
 			swin[i] = int32(rng.Intn(1<<20) - 1<<19)
 		}
@@ -402,8 +483,11 @@ func runAnalyzeTest(t *testing.T, k *Kernel, name string, order int, ref func([]
 		var goHist [32]uint16
 		goSum, goOvf := ref(swin, goDstP, &goHist)
 
-		jitDstP := make([]uint32, n)
+		jitDstP = jitDstP[:n]
-		var jitHist [32]uint16
+		for i := range jitDstP {
+			jitDstP[i] = 0
+		}
+		*jitHist = [32]uint16{} // reset for this iteration
 		args := make([]byte, 72)
 		PutPtr(args, 0, unsafe.Pointer(&swin[0]))
 		PutUint64(args, 8, uint64(len(swin)))
@@ -413,7 +497,13 @@ func runAnalyzeTest(t *testing.T, k *Kernel, name string, order int, ref func([]
 		PutUint64(args, 40, uint64(cap(jitDstP)))
 		PutPtr(args, 48, unsafe.Pointer(&jitHist[0]))
 
+		// Memory barrier: ensure all writes above are visible to the JIT code.
+		runtime.KeepAlive(args)
+
 		out, err := k.CallFunc(name, args)
+		runtime.KeepAlive(swin)
+		runtime.KeepAlive(jitDstP)
+		runtime.KeepAlive(jitHist)
 		if err != nil {
 			t.Fatalf("iter %d: %v", iter, err)
 		}
@@ -431,8 +521,8 @@ func runAnalyzeTest(t *testing.T, k *Kernel, name string, order int, ref func([]
 				t.Fatalf("iter %d: dstP[%d]: JIT=%d Go=%d", iter, i, jitDstP[i], goDstP[i])
 			}
 		}
-		if jitHist != goHist {
+		if *jitHist != goHist {
-			t.Fatalf("iter %d: hist mismatch", iter)
+			t.Fatalf("iter %d: hist mismatch\n  JIT=%v\n  Go =%v\n  n=%d", iter, *jitHist, goHist, n)
 		}
 	}
 }
@@ -508,15 +598,20 @@ func TestFLACDecodeMono24(t *testing.T) {
 	k := loadFLACKernel(t)
 	rng := rand.New(rand.NewSource(55))
 
+	src := mono24Src[:0]
+	jitDst := mono24JitDst[:0]
 	for iter := 0; iter < 500; iter++ {
 		n := rng.Intn(256)
-		src := make([]byte, 3*n)
+		src = src[:3*n]
 		rng.Read(src)
 
 		goDst := make([]int32, n)
 		decodeMono24Go(src, goDst)
 
-		jitDst := make([]int32, n)
+		jitDst = jitDst[:n]
+		for i := range jitDst {
+			jitDst[i] = 0
+		}
 		args := make([]byte, 48)
 		if len(src) > 0 {
 			PutPtr(args, 0, unsafe.Pointer(&src[0]))
@@ -530,6 +625,8 @@ func TestFLACDecodeMono24(t *testing.T) {
 		PutUint64(args, 40, uint64(cap(jitDst)))
 
 		_, err := k.CallFunc("decodeMono24AVX2", args)
+		runtime.KeepAlive(src)
+		runtime.KeepAlive(jitDst)
 		if err != nil {
 			t.Fatalf("iter %d: %v", iter, err)
 		}
@@ -545,17 +642,24 @@ func TestFLACDecodeStereo16(t *testing.T) {
 	k := loadFLACKernel(t)
 	rng := rand.New(rand.NewSource(66))
 
+	src := stereo16Src[:0]
+	jitLeft := stereo16JitLeft[:0]
+	jitRight := stereo16JitRight[:0]
 	for iter := 0; iter < 500; iter++ {
 		n := rng.Intn(256)
-		src := make([]byte, 4*n) // [L0,R0,L1,R1,...]
+		src = src[:4*n] // [L0,R0,L1,R1,...]
 		rng.Read(src)
 
 		goLeft := make([]int32, n)
 		goRight := make([]int32, n)
 		decodeStereo16Go(src, goLeft, goRight)
 
-		jitLeft := make([]int32, n)
+		jitLeft = jitLeft[:n]
-		jitRight := make([]int32, n)
+		jitRight = jitRight[:n]
+		for i := range jitLeft {
+			jitLeft[i] = 0
+			jitRight[i] = 0
+		}
 		args := make([]byte, 72)
 		if len(src) > 0 {
 			PutPtr(args, 0, unsafe.Pointer(&src[0]))
@@ -572,6 +676,9 @@ func TestFLACDecodeStereo16(t *testing.T) {
 		PutUint64(args, 64, uint64(cap(jitRight)))
 
 		_, err := k.CallFunc("decodeStereo16AVX2", args)
+		runtime.KeepAlive(src)
+		runtime.KeepAlive(jitLeft)
+		runtime.KeepAlive(jitRight)
 		if err != nil {
 			t.Fatalf("iter %d: %v", iter, err)
 		}

verify/fuzz.go

@@ -7,6 +7,7 @@ import (
 	"fmt"
 	"math/rand"
 	"regexp"
+	"runtime"
 	"strconv"
 	"strings"
 	"unsafe"
@@ -192,7 +193,7 @@ func (k *Kernel) FuzzFunc(name string, sig funcSig, goCode []byte, iterations in
 			if result.FirstFail == "" {
 				result.FirstFail = fmt.Sprintf("iter %d: gasm call: %v", i, err)
 			}
-			releaseBufs(bufs)
+			runtime.KeepAlive(bufs)
 			continue
 		}
 
@@ -203,7 +204,7 @@ func (k *Kernel) FuzzFunc(name string, sig funcSig, goCode []byte, iterations in
 			if result.FirstFail == "" {
 				result.FirstFail = fmt.Sprintf("iter %d: go call: %v", i, err)
 			}
-			releaseBufs(bufs)
+			runtime.KeepAlive(bufs)
 			continue
 		}
 
@@ -221,7 +222,7 @@ func (k *Kernel) FuzzFunc(name string, sig funcSig, goCode []byte, iterations in
 		} else {
 			result.Matches++
 		}
-		releaseBufs(bufs)
+		runtime.KeepAlive(bufs)
 	}
 	return result
 }
@@ -384,9 +385,3 @@ func equalBytes(a, b []byte) bool {
 	}
 	return true
 }
-
-func releaseBufs(bufs [][]byte) {
-	// Keep buffers alive until after the call; nothing to free in Go,
-	// but this prevents the compiler from collecting them too early.
-	_ = bufs
-}

verify/jit_test.go

@@ -5,6 +5,7 @@ package verify
 
 import (
 	"bytes"
+	"runtime"
 	"testing"
 	"unsafe"
 )
@@ -68,6 +69,7 @@ func TestJITSum(t *testing.T) {
 		PutUint64(args, 16, uint64(cap(tt.data)))
 
 		out, err := k.CallFunc("sum", args)
+		runtime.KeepAlive(tt.data)
 		if err != nil {
 			t.Fatalf("CallFunc(sum, %v): %v", tt.data, err)
 		}
@@ -111,6 +113,8 @@ func TestJITWideCopy(t *testing.T) {
 			PutUint64(args, 40, uint64(tt.n)) // src_cap
 
 			_, err := k.CallFunc("wideCopy", args)
+			runtime.KeepAlive(dst)
+			runtime.KeepAlive(src)
 			if err != nil {
 				t.Fatalf("CallFunc(wideCopy): %v", err)
 			}

verify/lz4_test.go

@@ -6,6 +6,7 @@ package verify
 import (
 	"bytes"
 	"os"
+	"runtime"
 	"testing"
 	"unsafe"
 )
@@ -45,6 +46,8 @@ func callDecodeBlockAVX2(t *testing.T, k *Kernel, src, dst []byte) (int, int) {
 	PutUint64(args, 40, uint64(cap(dst)))
 
 	out, err := k.CallFunc("decodeBlockAVX2", args)
+	runtime.KeepAlive(src)
+	runtime.KeepAlive(dst)
 	if err != nil {
 		t.Fatalf("CallFunc(decodeBlockAVX2): %v", err)
 	}
@@ -150,6 +153,8 @@ func TestLZ4WideCopyAVX2(t *testing.T) {
 		PutUint64(args, 40, uint64(n))
 
 		_, err := k.CallFunc("wideCopyAVX2", args)
+		runtime.KeepAlive(dst)
+		runtime.KeepAlive(src)
 		if err != nil {
 			t.Fatalf("wideCopyAVX2(n=%d): %v", n, err)
 		}

verify/sigparse_test.go

@@ -0,0 +1,153 @@
+// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
+// SPDX-License-Identifier: BSD-3-Clause
+
+package verify
+
+import (
+	"testing"
+)
+
+func TestParseFuncSigExported(t *testing.T) {
+	tests := []struct {
+		comment  string
+		wantName string
+		nParams  int
+		nResults int
+	}{
+		{"// func add(a int64, b int64) int64", "add", 2, 1},
+		{"// func wideCopy(dst []byte, src []byte)", "wideCopy", 2, 0},
+		{"// func decodeBlockAVX2(src, dst []byte) (n int, code int)", "decodeBlockAVX2", 2, 2},
+		{"// func analyzeO1RangeAVX2(swin []int32, dstP []uint32, hist *[32]uint16) (partSum uint64, overflow bool)", "analyzeO1RangeAVX2", 3, 2},
+		{"// not a func", "", 0, 0},
+		{"// func", "", 0, 0},
+	}
+	for _, tt := range tests {
+		sig, ok := ParseFuncSig(tt.comment)
+		if tt.wantName == "" {
+			if ok {
+				t.Errorf("ParseFuncSig(%q): expected not ok", tt.comment)
+			}
+			continue
+		}
+		if !ok {
+			t.Errorf("ParseFuncSig(%q): expected ok", tt.comment)
+			continue
+		}
+		if sig.Name != tt.wantName {
+			t.Errorf("ParseFuncSig(%q).Name = %q, want %q", tt.comment, sig.Name, tt.wantName)
+		}
+		if len(sig.Params) != tt.nParams {
+			t.Errorf("ParseFuncSig(%q): %d params, want %d", tt.comment, len(sig.Params), tt.nParams)
+		}
+		if len(sig.Results) != tt.nResults {
+			t.Errorf("ParseFuncSig(%q): %d results, want %d", tt.comment, len(sig.Results), tt.nResults)
+		}
+	}
+}
+
+func TestExtractFuncSig(t *testing.T) {
+	src := `// func add(a int64, b int64) int64
+TEXT ·add(SB), NOSPLIT, $0-24
+
+// func wideCopyAVX2(dst, src []byte)
+TEXT ·wideCopyAVX2(SB), NOSPLIT, $0-48
+`
+	sig, ok := ExtractFuncSig(src, "add")
+	if !ok {
+		t.Fatal("ExtractFuncSig(add): not found")
+	}
+	if sig.Name != "add" {
+		t.Errorf("Name = %q, want %q", sig.Name, "add")
+	}
+	if len(sig.Params) != 2 {
+		t.Errorf("params = %d, want 2", len(sig.Params))
+	}
+
+	sig, ok = ExtractFuncSig(src, "wideCopyAVX2")
+	if !ok {
+		t.Fatal("ExtractFuncSig(wideCopyAVX2): not found")
+	}
+	if len(sig.Params) != 2 {
+		t.Errorf("params = %d, want 2", len(sig.Params))
+	}
+
+	_, ok = ExtractFuncSig(src, "nonexistent")
+	if ok {
+		t.Error("ExtractFuncSig(nonexistent): expected not found")
+	}
+}
+
+func TestArgLayout(t *testing.T) {
+	sig := FuncSig{
+		Name: "wideCopyAVX2",
+		Params: []Param{
+			{Name: "dst", Typ: "[]byte"},
+			{Name: "src", Typ: "[]byte"},
+		},
+	}
+	layout := ArgLayout(sig)
+	if len(layout) != 2 {
+		t.Fatalf("ArgLayout: %d entries, want 2", len(layout))
+	}
+	// dst: offset 0, size 24 (slice)
+	if layout[0].Name != "dst" || layout[0].Offset != 0 || layout[0].Size != 24 || !layout[0].IsPtr {
+		t.Errorf("layout[0] = %+v", layout[0])
+	}
+	// src: offset 24, size 24 (slice)
+	if layout[1].Name != "src" || layout[1].Offset != 24 || layout[1].Size != 24 || !layout[1].IsPtr {
+		t.Errorf("layout[1] = %+v", layout[1])
+	}
+}
+
+func TestArgLayoutMixed(t *testing.T) {
+	sig := FuncSig{
+		Name: "decodeBlockAVX2",
+		Params: []Param{
+			{Name: "src", Typ: "[]byte"},
+			{Name: "dst", Typ: "[]byte"},
+		},
+		Results: []Param{
+			{Name: "n", Typ: "int"},
+			{Name: "code", Typ: "int"},
+		},
+	}
+	layout := ArgLayout(sig)
+	if len(layout) != 2 {
+		t.Fatalf("ArgLayout: %d entries, want 2", len(layout))
+	}
+	// Two slices: 24 + 24 = 48 bytes of params.
+	if layout[1].Offset != 24 {
+		t.Errorf("src offset = %d, want 24", layout[1].Offset)
+	}
+}
+
+func TestParamSize(t *testing.T) {
+	tests := []struct {
+		typ  string
+		want int
+	}{
+		{"[]byte", 24},
+		{"[]int32", 24},
+		{"*uint16", 8},
+		{"*[32]uint16", 8},
+		{"int", 8},
+		{"int64", 8},
+		{"uint64", 8},
+		{"uintptr", 8},
+		{"int32", 4},
+		{"uint32", 4},
+		{"float32", 4},
+		{"int16", 2},
+		{"uint16", 2},
+		{"int8", 1},
+		{"uint8", 1},
+		{"byte", 1},
+		{"bool", 1},
+		{"string", 8}, // unknown type defaults to 8
+	}
+	for _, tt := range tests {
+		if got := paramSize(tt.typ); got != tt.want {
+			t.Errorf("paramSize(%q) = %d, want %d", tt.typ, got, tt.want)
+		}
+	}
+}

verify/verify.go

@@ -7,6 +7,7 @@ import (
 	"fmt"
 	"math/rand"
 	"os"
+	"runtime"
 
 	"sourcedock.dev/petrbalvin/gasm-devkit/asm"
 	"sourcedock.dev/petrbalvin/gasm-devkit/ast"
@@ -153,7 +154,7 @@ func (k *Kernel) FuzzFuncChecked(name string, sig funcSig, iterations int, seed
 			if result.FirstFail == "" {
 				result.FirstFail = fmt.Sprintf("iter %d: call: %v", i, err)
 			}
-			releaseBufs(bufs)
+			runtime.KeepAlive(bufs)
 			continue
 		}
 
@@ -166,7 +167,7 @@ func (k *Kernel) FuzzFuncChecked(name string, sig funcSig, iterations int, seed
 		} else {
 			result.Matches++
 		}
-		releaseBufs(bufs)
+		runtime.KeepAlive(bufs)
 	}
 
 	if violations > 0 && result.FirstFail == "" {