pkg/compiler: support size attribute for unions

10 files changed, 89 insertions, 35 deletions

diff --git a/executor/syscalls_test.h b/executor/syscalls_test.h
index 9468387e1..d5be16213 100644
--- a/executor/syscalls_test.h
+++ b/executor/syscalls_test.h
@@ -2,11 +2,11 @@
 
 #if 0
 #define GOARCH "32"
-#define SYZ_REVISION "035aa7331fff3c8bb13bc8925ae43befcbc94a71"
+#define SYZ_REVISION "d4cbc2aec52e291fefa8155c6c6ca397ba3c9e49"
 #define SYZ_PAGE_SIZE 8192
 #define SYZ_NUM_PAGES 2048
 #define SYZ_DATA_OFFSET 536870912
-unsigned syscall_count = 91;
+unsigned syscall_count = 92;
 call_t syscalls[] = {
 	{"foo$any0", 0, (syscall_t)foo},
 	{"foo$anyres", 0, (syscall_t)foo},
@@ -68,6 +68,7 @@ call_t syscalls[] = {
 	{"syz_test$length25", 0, (syscall_t)syz_test},
 	{"syz_test$length26", 0, (syscall_t)syz_test},
 	{"syz_test$length27", 0, (syscall_t)syz_test},
+	{"syz_test$length28", 0, (syscall_t)syz_test},
 	{"syz_test$length3", 0, (syscall_t)syz_test},
 	{"syz_test$length4", 0, (syscall_t)syz_test},
 	{"syz_test$length5", 0, (syscall_t)syz_test},
@@ -105,11 +106,11 @@ call_t syscalls[] = {
 
 #if 0
 #define GOARCH "64"
-#define SYZ_REVISION "7106e9ab2b7a52203a8ad681140f3326e8b3d7ce"
+#define SYZ_REVISION "cabe73c3b49273a8b0cd13bff35d40897dd88138"
 #define SYZ_PAGE_SIZE 4096
 #define SYZ_NUM_PAGES 4096
 #define SYZ_DATA_OFFSET 536870912
-unsigned syscall_count = 91;
+unsigned syscall_count = 92;
 call_t syscalls[] = {
 	{"foo$any0", 0, (syscall_t)foo},
 	{"foo$anyres", 0, (syscall_t)foo},
@@ -171,6 +172,7 @@ call_t syscalls[] = {
 	{"syz_test$length25", 0, (syscall_t)syz_test},
 	{"syz_test$length26", 0, (syscall_t)syz_test},
 	{"syz_test$length27", 0, (syscall_t)syz_test},
+	{"syz_test$length28", 0, (syscall_t)syz_test},
 	{"syz_test$length3", 0, (syscall_t)syz_test},
 	{"syz_test$length4", 0, (syscall_t)syz_test},
 	{"syz_test$length5", 0, (syscall_t)syz_test},
diff --git a/pkg/compiler/check.go b/pkg/compiler/check.go
index 2a6afae3a..4dac37260 100644
--- a/pkg/compiler/check.go
+++ b/pkg/compiler/check.go
@@ -240,10 +240,11 @@ func (comp *compiler) checkAttributeValues() {
 		switch n := decl.(type) {
 		case *ast.Struct:
 			for _, attr := range n.Attrs {
-				if !n.IsUnion && attr.Ident == "size" {
+				if attr.Ident == "size" {
+					_, typ, name := n.Info()
 					if comp.structIsVarlen(n.Name.Name) {
-						comp.error(attr.Pos, "varlen struct %v has size attribute",
-							n.Name.Name)
+						comp.error(attr.Pos, "varlen %v %v has size attribute",
+							typ, name)
 					}
 					sz := attr.Args[0].Value
 					if sz == 0 || sz > 1<<20 {
@@ -906,7 +907,7 @@ func (comp *compiler) checkVarlen(n *ast.Struct) {
 	// Non-varlen unions can't have varlen fields.
 	// Non-packed structs can't have varlen fields in the middle.
 	if n.IsUnion {
-		if varlen := comp.parseUnionAttrs(n); varlen {
+		if varlen, _ := comp.parseUnionAttrs(n); varlen {
 			return
 		}
 	} else {
diff --git a/pkg/compiler/compiler.go b/pkg/compiler/compiler.go
index 314cbef41..f0d1176e7 100644
--- a/pkg/compiler/compiler.go
+++ b/pkg/compiler/compiler.go
@@ -145,9 +145,11 @@ func (comp *compiler) structIsVarlen(name string) bool {
 		return varlen
 	}
 	s := comp.structs[name]
-	if s.IsUnion && comp.parseUnionAttrs(s) {
-		comp.structVarlen[name] = true
-		return true
+	if s.IsUnion {
+		if varlen, _ := comp.parseUnionAttrs(s); varlen {
+			comp.structVarlen[name] = true
+			return true
+		}
 	}
 	comp.structVarlen[name] = false // to not hang on recursive types
 	varlen := false
@@ -161,18 +163,21 @@ func (comp *compiler) structIsVarlen(name string) bool {
 	return varlen
 }
 
-func (comp *compiler) parseUnionAttrs(n *ast.Struct) (varlen bool) {
+func (comp *compiler) parseUnionAttrs(n *ast.Struct) (varlen bool, size uint64) {
+	size = sizeUnassigned
 	for _, attr := range n.Attrs {
 		switch attr.Ident {
 		case "varlen":
+			if len(attr.Args) != 0 {
+				comp.error(attr.Pos, "%v attribute has args", attr.Ident)
+			}
 			varlen = true
+		case "size":
+			size = comp.parseSizeAttr(n, attr)
 		default:
 			comp.error(attr.Pos, "unknown union %v attribute %v",
 				n.Name.Name, attr.Ident)
 		}
-		if len(attr.Args) != 0 {
-			comp.error(attr.Pos, "%v attribute has args", attr.Ident)
-		}
 	}
 	return
 }
@@ -207,19 +212,7 @@ func (comp *compiler) parseStructAttrs(n *ast.Struct) (packed bool, size, align
 			}
 			align = a
 		case attr.Ident == "size":
-			if len(attr.Args) != 1 {
-				comp.error(attr.Pos, "%v attribute is expected to have 1 argument", attr.Ident)
-			}
-			sz := attr.Args[0]
-			if unexpected, _, ok := checkTypeKind(sz, kindInt); !ok {
-				comp.error(sz.Pos, "unexpected %v, expect int", unexpected)
-				return
-			}
-			if sz.HasColon || len(sz.Args) != 0 {
-				comp.error(sz.Pos, "size attribute has colon or args")
-				return
-			}
-			size = sz.Value
+			size = comp.parseSizeAttr(n, attr)
 		default:
 			comp.error(attr.Pos, "unknown struct %v attribute %v",
 				n.Name.Name, attr.Ident)
@@ -228,6 +221,23 @@ func (comp *compiler) parseStructAttrs(n *ast.Struct) (packed bool, size, align
 	return
 }
 
+func (comp *compiler) parseSizeAttr(n *ast.Struct, attr *ast.Type) uint64 {
+	if len(attr.Args) != 1 {
+		comp.error(attr.Pos, "%v attribute is expected to have 1 argument", attr.Ident)
+		return sizeUnassigned
+	}
+	sz := attr.Args[0]
+	if unexpected, _, ok := checkTypeKind(sz, kindInt); !ok {
+		comp.error(sz.Pos, "unexpected %v, expect int", unexpected)
+		return sizeUnassigned
+	}
+	if sz.HasColon || len(sz.Args) != 0 {
+		comp.error(sz.Pos, "size attribute has colon or args")
+		return sizeUnassigned
+	}
+	return sz.Value
+}
+
 func (comp *compiler) getTypeDesc(t *ast.Type) *typeDesc {
 	if desc := builtinTypes[t.Ident]; desc != nil {
 		return desc
diff --git a/pkg/compiler/consts.go b/pkg/compiler/consts.go
index 789674578..1b008bd4f 100644
--- a/pkg/compiler/consts.go
+++ b/pkg/compiler/consts.go
@@ -87,7 +87,7 @@ func (comp *compiler) extractConsts() map[string]*ConstInfo {
 		switch n := decl.(type) {
 		case *ast.Struct:
 			for _, attr := range n.Attrs {
-				if !n.IsUnion && attr.Ident == "size" {
+				if attr.Ident == "size" {
 					info := getConstInfo(infos, attr.Pos)
 					info.consts[attr.Args[0].Ident] = true
 				}
@@ -199,7 +199,7 @@ func (comp *compiler) patchConsts(consts map[string]uint64) {
 			}
 			if n, ok := decl.(*ast.Struct); ok {
 				for _, attr := range n.Attrs {
-					if !n.IsUnion && attr.Ident == "size" {
+					if attr.Ident == "size" {
 						sz := attr.Args[0]
 						comp.patchIntConst(sz.Pos, &sz.Value,
 							&sz.Ident, consts, &missing)
diff --git a/pkg/compiler/gen.go b/pkg/compiler/gen.go
index 74d1b5d78..5bf239e79 100644
--- a/pkg/compiler/gen.go
+++ b/pkg/compiler/gen.go
@@ -165,14 +165,24 @@ func (comp *compiler) genStructDescs(syscalls []*prog.Syscall) []*prog.KeyedStru
 			if !checkStruct(t.Key, &t.StructDesc) {
 				return
 			}
+			structNode := comp.structNodes[t.StructDesc]
+			varlen, sizeAttr := comp.parseUnionAttrs(structNode)
 			t.TypeSize = 0
-			varlen := comp.parseUnionAttrs(comp.structNodes[t.StructDesc])
 			if !varlen {
 				for _, fld := range t.Fields {
-					if t.TypeSize < fld.Size() {
-						t.TypeSize = fld.Size()
+					sz := fld.Size()
+					if sizeAttr != sizeUnassigned && sz > sizeAttr {
+						comp.error(structNode.Pos, "union %v has size attribute %v"+
+							" which is less than field %v size %v",
+							structNode.Name.Name, sizeAttr, fld.Name(), sz)
+					}
+					if t.TypeSize < sz {
+						t.TypeSize = sz
 					}
 				}
+				if sizeAttr != sizeUnassigned {
+					t.TypeSize = sizeAttr
+				}
 			}
 		}
 	}
diff --git a/pkg/compiler/testdata/errors2.txt b/pkg/compiler/testdata/errors2.txt
index a8d29789a..86148a563 100644
--- a/pkg/compiler/testdata/errors2.txt
+++ b/pkg/compiler/testdata/errors2.txt
@@ -91,6 +91,14 @@ u1 [
 	f2	len[f1, int32]	### len target f1 does not exist
 ]
 
+u2 [
+	f1	int8
+] [size[0]]			### size attribute has bad value 0, expect [1, 1<<20]
+
+u3 [
+	f1	int8
+] [varlen, size[8]]		### varlen union u3 has size attribute
+
 foo$200(a ptr[in, s2])
 foo$201(a ptr[in, s1])
 foo$202(a u1)
diff --git a/prog/size_test.go b/prog/size_test.go
index f5ac9a958..08b7cba92 100644
--- a/prog/size_test.go
+++ b/prog/size_test.go
@@ -142,6 +142,10 @@ func TestAssignSize(t *testing.T) {
 			"syz_test$length27(&(0x7f0000000000), 0x0)",
 			"syz_test$length27(&(0x7f0000000000), 0x2a)",
 		},
+		{
+			"syz_test$length28(&(0x7f0000000000), 0x0)",
+			"syz_test$length28(&(0x7f0000000000)=@f1, 0x2a)",
+		},
 	}
 
 	for i, test := range tests {
diff --git a/sys/test/32.go b/sys/test/32.go
index 2afb57717..7425c8bd9 100644
--- a/sys/test/32.go
+++ b/sys/test/32.go
@@ -54,6 +54,9 @@ var structDescs_32 = []*KeyedStruct{
 		&IntType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "int8", FldName: "f1", TypeSize: 1}}},
 		&ConstType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "pad", TypeSize: 41}}, IsPad: true},
 	}}},
+	{Key: StructKey{Name: "explicitly_sized_union"}, Desc: &StructDesc{TypeCommon: TypeCommon{TypeName: "explicitly_sized_union", TypeSize: 42}, Fields: []Type{
+		&IntType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "int8", FldName: "f1", TypeSize: 1}}},
+	}}},
 	{Key: StructKey{Name: "len_nontemp4"}, Desc: &StructDesc{TypeCommon: TypeCommon{TypeName: "len_nontemp4", TypeSize: 4}, Fields: []Type{
 		&LenType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "len", FldName: "f1", TypeSize: 4}}, Buf: "len_temp3"},
 	}}},
@@ -683,6 +686,10 @@ var syscalls_32 = []*Syscall{
 		&PtrType{TypeCommon: TypeCommon{TypeName: "ptr", FldName: "a0", TypeSize: 4}, Type: &StructType{Key: StructKey{Name: "explicitly_sized"}}},
 		&LenType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "len", FldName: "a1", TypeSize: 4}}, Buf: "a0"},
 	}},
+	{Name: "syz_test$length28", CallName: "syz_test", Args: []Type{
+		&PtrType{TypeCommon: TypeCommon{TypeName: "ptr", FldName: "a0", TypeSize: 4}, Type: &UnionType{Key: StructKey{Name: "explicitly_sized_union"}}},
+		&LenType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "len", FldName: "a1", TypeSize: 4}}, Buf: "a0"},
+	}},
 	{Name: "syz_test$length3", CallName: "syz_test", Args: []Type{
 		&PtrType{TypeCommon: TypeCommon{TypeName: "ptr", FldName: "a0", TypeSize: 4}, Type: &StructType{Key: StructKey{Name: "syz_length_len_struct"}}},
 	}},
@@ -790,4 +797,4 @@ var consts_32 = []ConstValue{
 	{Name: "ONLY_32BITS_CONST", Value: 1},
 }
 
-const revision_32 = "035aa7331fff3c8bb13bc8925ae43befcbc94a71"
+const revision_32 = "d4cbc2aec52e291fefa8155c6c6ca397ba3c9e49"
diff --git a/sys/test/64.go b/sys/test/64.go
index ac32ebbe8..4b48d9b05 100644
--- a/sys/test/64.go
+++ b/sys/test/64.go
@@ -54,6 +54,9 @@ var structDescs_64 = []*KeyedStruct{
 		&IntType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "int8", FldName: "f1", TypeSize: 1}}},
 		&ConstType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "pad", TypeSize: 41}}, IsPad: true},
 	}}},
+	{Key: StructKey{Name: "explicitly_sized_union"}, Desc: &StructDesc{TypeCommon: TypeCommon{TypeName: "explicitly_sized_union", TypeSize: 42}, Fields: []Type{
+		&IntType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "int8", FldName: "f1", TypeSize: 1}}},
+	}}},
 	{Key: StructKey{Name: "len_nontemp4"}, Desc: &StructDesc{TypeCommon: TypeCommon{TypeName: "len_nontemp4", TypeSize: 4}, Fields: []Type{
 		&LenType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "len", FldName: "f1", TypeSize: 4}}, Buf: "len_temp3"},
 	}}},
@@ -682,6 +685,10 @@ var syscalls_64 = []*Syscall{
 		&PtrType{TypeCommon: TypeCommon{TypeName: "ptr", FldName: "a0", TypeSize: 8}, Type: &StructType{Key: StructKey{Name: "explicitly_sized"}}},
 		&LenType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "len", FldName: "a1", TypeSize: 8}}, Buf: "a0"},
 	}},
+	{Name: "syz_test$length28", CallName: "syz_test", Args: []Type{
+		&PtrType{TypeCommon: TypeCommon{TypeName: "ptr", FldName: "a0", TypeSize: 8}, Type: &UnionType{Key: StructKey{Name: "explicitly_sized_union"}}},
+		&LenType{IntTypeCommon: IntTypeCommon{TypeCommon: TypeCommon{TypeName: "len", FldName: "a1", TypeSize: 8}}, Buf: "a0"},
+	}},
 	{Name: "syz_test$length3", CallName: "syz_test", Args: []Type{
 		&PtrType{TypeCommon: TypeCommon{TypeName: "ptr", FldName: "a0", TypeSize: 8}, Type: &StructType{Key: StructKey{Name: "syz_length_len_struct"}}},
 	}},
@@ -788,4 +795,4 @@ var consts_64 = []ConstValue{
 	{Name: "IPPROTO_UDP", Value: 17},
 }
 
-const revision_64 = "7106e9ab2b7a52203a8ad681140f3326e8b3d7ce"
+const revision_64 = "cabe73c3b49273a8b0cd13bff35d40897dd88138"
diff --git a/sys/test/test.txt b/sys/test/test.txt
index 063a9a2e2..ca548d073 100644
--- a/sys/test/test.txt
+++ b/sys/test/test.txt
@@ -405,10 +405,15 @@ explicitly_sized {
 	f1	int8
 } [size[42]]
 
+explicitly_sized_union [
+	f1	int8
+] [size[42]]
+
 syz_test$length24(a ptr[in, len_nontemp5])
 syz_test$length25(a0 ptr[in, array[array[int8]]], a1 len[a0])
 syz_test$length26(a ptr[in, len_unaligned], b bytesize[a])
 syz_test$length27(a0 ptr[in, explicitly_sized], a1 len[a0])
+syz_test$length28(a0 ptr[in, explicitly_sized_union], a1 len[a0])
 
 # Big endian