test(asm): seed the riscv64 and loong64 assemble fuzz targets

Assisted-by: GLM 5.3

asm/assembler_fuzz_test.go

@@ -37,6 +37,22 @@ func corpusSeeds(f *testing.F) {
 	}
 }
 
+// archCorpusSeeds seeds a target with every assembly file of its architecture
+// seed directory, testdata/seeds/<dir>.  The files hold the target-specific
+// corpus: the instruction families GOROOT's own assembler corpus exercises for
+// the architecture, minimalised, plus the boundary shapes the encoder's range
+// gates live on.  They are committed assembly, so `gasm fmt` and `gasm lint`
+// gate them the way they gate every other .s file.  A plain `go test` run
+// replays each one as a regression case.
+func archCorpusSeeds(f *testing.F, dir string) {
+	files, _ := filepath.Glob(filepath.Join("testdata", "seeds", dir, "*.s"))
+	for _, path := range files {
+		if b, err := os.ReadFile(path); err == nil {
+			f.Add(string(b))
+		}
+	}
+}
+
 // fuzzAssemble is the whole fuzz body, shared by every target and one line
 // apart between them: parse with macro and include expansion, assemble
 // through the target's file-level entry, and hold the invariants.  The
@@ -179,3 +195,86 @@ func FuzzAssembleARM64(f *testing.F) {
 		fuzzAssemble(t, "fuzz_arm64.s", src, AssembleFileARM64)
 	})
 }
+
+// FuzzAssembleRISCV64 hammers the same pipeline for the fixed riscv64 target.
+// The seed corpus lives in testdata/seeds/riscv64: the instruction families
+// GOROOT's riscv64 assembler corpus exercises (the immediate-range ladder of
+// the I-type arithmetic, the load/store and branch offsets, the atomics, the
+// FP conversions and the fused multiply-adds), the RVV configuration and
+// arithmetic classes with their mask forms, the RVC-compressible shapes, the
+// CSR instructions and the Zbb/Zba/Zbs bit-manipulation set, plus one file per
+// rejection shape so each diagnostic path replays on its own.
+func FuzzAssembleRISCV64(f *testing.F) {
+	corpusSeeds(f)
+	archCorpusSeeds(f, "riscv64")
+	// Minimal seeds for the shared file-level surface, spelled the riscv64
+	// way: the guard classes, the macro and include expansion, and the data
+	// section with its symbol-valued fields.
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tRET\n")
+	f.Add("TEXT ·f(SB), $16-8\n\tMOV x+0(FP), X10\n\tMOV X10, ret+8(FP)\n\tRET\n")
+	f.Add("TEXT ·f(SB), $256-0\n\tCALL ·helper(SB)\n\tRET\nTEXT ·helper(SB), NOSPLIT, $0\n\tRET\n")
+	f.Add("#define L(n) ADDI $n, X10, X10\nTEXT ·f(SB), NOSPLIT, $0\n\tL(7)\n\tRET\n")
+	f.Add("#include \"textflag.h\"\nTEXT ·f(SB), NOSPLIT, $0\n\tRET\n")
+	f.Add("#include \"fuzzdefs.h\"\nTEXT ·f(SB), $16-8\n\tMOV KONST, X10\n\tMOV ARG(x), X11\n\tRET\n")
+	f.Add("DATA d<>+0(SB)/8, $0xf4f8fcff\nDATA d<>+4(SB)/4, $1\nGLOBL d<>(SB), RODATA, $8\n" +
+		"TEXT ·f(SB), NOSPLIT, $0\n\tMOV d<>(SB), X10\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tPCALIGN $16\n\tADD X11, X10, X10\n\tPCALIGN $2048\n\tRET\n")
+	f.Add("TEXT ·f(SB), $0\n\tPCDATA $0, $1\n\tFUNCDATA $0, ·meta(SB)\n\tRET\n")
+	// Shapes that must be rejected: each pins a diagnostic path the accepted
+	// seeds never reach.
+	f.Add("TEXT ·f(SB), $0\n\tBOGUSINSTR X10, X11\n\tRET\n")
+	f.Add("GLOBL d(SB), $-8\n")
+	f.Add("GLOBL d(SB), $0x4000001\n")
+	f.Add("DATA d+0(SB)/9, $1\nGLOBL d(SB), $8\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tADD X11, X32\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tADDI $2048, X5, X6\n\tADD X11, X40, X6\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tCSRRW $0x1000, X5, X6\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tSLLI $64, X5, X6\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tVLE8V (X10), V32\n\tRET\n")
+	f.Add("#define A A\nA\n")
+
+	f.Fuzz(func(t *testing.T, src string) {
+		fuzzAssemble(t, "fuzz_riscv64.s", src, AssembleFileRISCV)
+	})
+}
+
+// FuzzAssembleLOONG64 hammers the same pipeline for the fixed loong64 target.
+// The seed corpus lives in testdata/seeds/loong64: the LSX and LASX register
+// banks with their immediate forms and range gates (the si5 compares, the
+// biased shifts, the VSHUF4I/VPERMI/VEXTRINS immediates), the ll/sc offset
+// ladder with its three encoding spans, the pointer loads and stores, the
+// atomics with their dbar forms, the branches, the bit-field instructions and
+// the register-class moves, plus one file per rejection shape.
+func FuzzAssembleLOONG64(f *testing.F) {
+	corpusSeeds(f)
+	archCorpusSeeds(f, "loong64")
+	// Minimal seeds for the shared file-level surface, spelled the loong64
+	// way.
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tRET\n")
+	f.Add("TEXT ·f(SB), $16-8\n\tMOVV x+0(FP), R4\n\tMOVV R4, ret+8(FP)\n\tRET\n")
+	f.Add("TEXT ·f(SB), $256-0\n\tCALL ·helper(SB)\n\tRET\nTEXT ·helper(SB), NOSPLIT, $0\n\tRET\n")
+	f.Add("#define L(n) ADDV $n, R4, R4\nTEXT ·f(SB), NOSPLIT, $0\n\tL(7)\n\tRET\n")
+	f.Add("#include \"textflag.h\"\nTEXT ·f(SB), NOSPLIT, $0\n\tRET\n")
+	f.Add("#include \"fuzzdefs.h\"\nTEXT ·f(SB), $16-8\n\tMOVV KONST, R4\n\tMOVV ARG(x), R5\n\tRET\n")
+	f.Add("DATA d<>+0(SB)/8, $0xf4f8fcff\nDATA d<>+4(SB)/4, $1\nGLOBL d<>(SB), RODATA, $8\n" +
+		"TEXT ·f(SB), NOSPLIT, $0\n\tMOVV d<>(SB), R4\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tPCALIGN $16\n\tADDV R5, R4, R4\n\tPCALIGN $2048\n\tRET\n")
+	f.Add("TEXT ·f(SB), $0\n\tPCDATA $0, $1\n\tFUNCDATA $0, ·meta(SB)\n\tRET\n")
+	// Shapes that must be rejected: each pins a diagnostic path the accepted
+	// seeds never reach.
+	f.Add("TEXT ·f(SB), $0\n\tBOGUSINSTR R4, R5\n\tRET\n")
+	f.Add("GLOBL d(SB), $-8\n")
+	f.Add("GLOBL d(SB), $0x4000001\n")
+	f.Add("DATA d+0(SB)/9, $1\nGLOBL d(SB), $8\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tADD R5, R32\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tBEQZ V0, L1\nL1:\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tVADDV V1, V2, X3\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tVSEQV $32, V2, V3\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tVSHUF4IV $16, V2, V1\n\tRET\n")
+	f.Add("TEXT ·f(SB), NOSPLIT, $0\n\tBSTRPICKV $64, R4, $5, R6\n\tRET\n")
+	f.Add("#define A A\nA\n")
+
+	f.Fuzz(func(t *testing.T, src string) {
+		fuzzAssemble(t, "fuzz_loong64.s", src, AssembleFileLOONG64)
+	})
+}