Merge pull request #30 from parca-dev/add_tls_and_312_support

python: Add TLS, 3.12 and 3.13 Offsets

cmd/structlayout/structlayout.go

@@ -58,11 +58,15 @@ func main() {
 
 	var (
 		layoutMap       runtimedata.LayoutMap
-		initialStateMap any
+		initialStateMap runtimedata.InitialStateMap
 		outputDir       = givenOutputDir
 	)
 	switch runtime {
 	case "python":
+		if strings.Contains(version, "a") {
+			// Alpha version detected.
+			version = strings.ReplaceAll(version, "a", "-alpha.")
+		}
 		layoutMap = python.DataMapForLayout(version)
 		initialStateMap = python.DataMapForInitialState(version)
 		if outputDir == "" {
@@ -89,11 +93,6 @@ func main() {
 		os.Exit(1)
 	}
 
-	if layoutMap == nil {
-		logger.Error("unknown version", "version", version)
-		os.Exit(1)
-	}
-
 	var (
 		input = fSet.Arg(0)
 	)
@@ -103,19 +102,23 @@ func main() {
 		os.Exit(1)
 	}
 
-	output := filepath.Join(outputDir, "layout", fmt.Sprintf("%s_%s.yaml", runtime, sanitizeIdentifier(version)))
-	if err := processAndWriteLayout(dwarfData, output, version, layoutMap); err != nil {
-		logger.Error("failed to write layout", "err", err)
-		os.Exit(1)
+	if !isNil(layoutMap) {
+		output := filepath.Join(outputDir, "layout", fmt.Sprintf("%s_%s.yaml", runtime, sanitizeIdentifier(version)))
+		if err := processAndWriteLayout(dwarfData, output, version, layoutMap); err != nil {
+			logger.Error("failed to write layout", "err", err)
+			os.Exit(1)
+		}
+		logger.Info("layout file written", "file", output)
+	} else {
+		logger.Info("no layout map found, skipping layout generation")
 	}
-	logger.Info("layout file written", "file", output)
 
 	if isNil(initialStateMap) {
 		logger.Info("no initial state map found, skipping initial state generation")
 		os.Exit(0)
 	}
 
-	output = filepath.Join(outputDir, "initialstate", fmt.Sprintf("%s_%s.yaml", runtime, sanitizeIdentifier(version)))
+	output := filepath.Join(outputDir, "initialstate", fmt.Sprintf("%s_%s.yaml", runtime, sanitizeIdentifier(version)))
 	if err := processAndWriteInitialState(dwarfData, output, version, initialStateMap); err != nil {
 		logger.Error("failed to write initial state", "err", err)
 		os.Exit(1)
@@ -161,7 +164,7 @@ func processAndWriteLayout(dwarfData *dwarf.Data, output string, version string,
 }
 
 // processAndWriteInitialState processes the given ELF file and writes the initial state to the given output file.
-func processAndWriteInitialState(dwarfData *dwarf.Data, output string, version string, initialStateMap any) error {
+func processAndWriteInitialState(dwarfData *dwarf.Data, output string, version string, initialStateMap runtimedata.InitialStateMap) error {
 	dm, err := datamap.New(initialStateMap)
 	if err != nil {
 		return fmt.Errorf("failed to create data map: %w", err)
@@ -181,7 +184,7 @@ func processAndWriteInitialState(dwarfData *dwarf.Data, output string, version s
 	}
 
 	// Extremely in-efficient and hacky but it should work for now.
-	withVersion, err := runtimedata.WithVersion(version, convertToMapOfAny(initialStateMap))
+	withVersion, err := runtimedata.WithVersion(version, convertToMapOfAny(initialStateMap.InitialState()))
 	if err != nil {
 		return fmt.Errorf("failed to wrap layout with version: %w", err)
 	}

pkg/libc/glibc/datamap.go

@@ -11,10 +11,10 @@ type glibc struct {
 
 func (g *glibc) Layout() runtimedata.RuntimeData {
 	return &Layout{
-		PthreadSpecific1stblock: g.PThreadSpecific1stblock,
-		PthreadSize:             g.PThreadSize,
-		PthreadKeyData:          g.PThreadKeyData,
-		PthreadKeyDataSize:      g.PThreadKeyDataSize,
+		PThreadSpecific1stblock: g.PThreadSpecific1stblock,
+		PThreadSize:             g.PThreadSize,
+		PThreadKeyData:          g.PThreadKeyData,
+		PThreadKeyDataSize:      g.PThreadKeyDataSize,
 	}
 }
 

pkg/libc/glibc/glibc_test.go

@@ -21,43 +21,43 @@ func Test_getLayoutForArch(t *testing.T) {
 			v:    semver.MustParse("2.29.0"),
 			arch: "amd64",
 			want: &Layout{
-				PthreadSpecific1stblock: 784,
-				PthreadSize:             2304,
-				PthreadKeyData:          8,
-				PthreadKeyDataSize:      16,
+				PThreadSpecific1stblock: 784,
+				PThreadSize:             2304,
+				PThreadKeyData:          0x08,
+				PThreadKeyDataSize:      0x10,
 			},
 		},
 		{
 			name: "2.29.0",
 			v:    semver.MustParse("2.29.0"),
 			arch: "arm64",
 			want: &Layout{
-				PthreadSpecific1stblock: 272,
-				PthreadSize:             1792,
-				PthreadKeyData:          8,
-				PthreadKeyDataSize:      16,
+				PThreadSpecific1stblock: 272,
+				PThreadSize:             1792,
+				PThreadKeyData:          0x08,
+				PThreadKeyDataSize:      0x10,
 			},
 		},
 		{
 			name: "2.37.0",
 			v:    semver.MustParse("2.37.0"),
 			arch: "amd64",
 			want: &Layout{
-				PthreadSpecific1stblock: 784,
-				PthreadSize:             2368,
-				PthreadKeyData:          8,
-				PthreadKeyDataSize:      16,
+				PThreadSpecific1stblock: 784,
+				PThreadSize:             2368,
+				PThreadKeyData:          0x08,
+				PThreadKeyDataSize:      0x10,
 			},
 		},
 		{
 			name: "2.37.0",
 			v:    semver.MustParse("2.37.0"),
 			arch: "arm64",
 			want: &Layout{
-				PthreadSpecific1stblock: 272,
-				PthreadSize:             1856,
-				PthreadKeyData:          8,
-				PthreadKeyDataSize:      16,
+				PThreadSpecific1stblock: 272,
+				PThreadSize:             1856,
+				PThreadKeyData:          0x08,
+				PThreadKeyDataSize:      0x10,
 			},
 		},
 	}

pkg/libc/glibc/layout.go

@@ -9,10 +9,10 @@ import (
 )
 
 type Layout struct {
-	PthreadSpecific1stblock int64 `yaml:"pthread_specific_1stblock"`
-	PthreadSize             int64 `yaml:"pthread_size"`
-	PthreadKeyData          int64 `yaml:"pthread_key_data"`
-	PthreadKeyDataSize      int64 `yaml:"pthread_key_data_size"`
+	PThreadSpecific1stblock int64 `yaml:"pthread_specific_1stblock"`
+	PThreadSize             int64 `yaml:"pthread_size"`
+	PThreadKeyData          int64 `yaml:"pthread_key_data"`
+	PThreadKeyDataSize      int64 `yaml:"pthread_key_data_size"`
 }
 
 func (l Layout) Data() ([]byte, error) {

pkg/python/datamap.go

@@ -17,6 +17,7 @@ func DataMapForLayout(v string) runtimedata.LayoutMap {
 		version.MustParseConstraints("3.3.x - 3.9.x"): &python33_39{},
 		version.MustParseConstraints("3.10.x"):        &python310{},
 		version.MustParseConstraints("3.11.x"):        &python311{},
+		version.MustParseConstraints("3.12.x"):        &python312{},
 	}
 
 	lookupVersion := semver.MustParse(v)
@@ -302,3 +303,73 @@ func (p python311) Layout() runtimedata.RuntimeData {
 		},
 	}
 }
+
+type python312 struct {
+	PyObjectObType                    int64 `offsetof:"PyObject.ob_type"`
+	PyStringData                      int64 `sizeof:"PyASCIIObject"`
+	PyTypeObjectTpName                int64 `offsetof:"PyTypeObject.tp_name"`
+	PyThreadStateInterp               int64 `offsetof:"PyThreadState.interp"`
+	PyThreadStateNext                 int64 `offsetof:"PyThreadState.next"`
+	PyThreadStateThreadID             int64 `offsetof:"PyThreadState.thread_id"`
+	PyThreadStateNativeThreadID       int64 `offsetof:"PyThreadState.native_thread_id"`
+	PyThreadStateCFrame               int64 `offsetof:"PyThreadState.cframe"`
+	PyInterpreterStateTstateHead      int64 `offsetof:"PyInterpreterState.threads"`
+	PyInterpreterStateIsPythreadsHead int64 `offsetof:"pythreads.head"`
+	PyRuntimeStateInterpreters        int64 `offsetof:"pyruntimestate.interpreters"`
+	PyRuntimeStatePyInterpretersMain  int64 `offsetof:"pyinterpreters.main"`
+	PyFrameObjectFBack                int64 `offsetof:"_PyInterpreterFrame.previous"`
+	PyFrameObjectFCode                int64 `offsetof:"_PyInterpreterFrame.f_code"`
+	PyFrameObjectFLocalsplus          int64 `offsetof:"_PyInterpreterFrame.localsplus"`
+	PyCodeObjectCoFilename            int64 `offsetof:"PyCodeObject.co_filename"`
+	PyCodeObjectCoName                int64 `offsetof:"PyCodeObject.co_name"`
+	PyCodeObjectCoVarNames            int64 `offsetof:"PyCodeObject.co_localsplusnames"`
+	PyCodeObjectCoFirstlineno         int64 `offsetof:"PyCodeObject.co_firstlineno"`
+	PyTupleObjectObItem               int64 `offsetof:"PyTupleObject.ob_item"`
+}
+
+func (p python312) Layout() runtimedata.RuntimeData {
+	return &Layout{
+		PyObject: PyObject{
+			ObType: p.PyObjectObType,
+		},
+		PyString: PyString{
+			Data: p.PyStringData,
+			Size: doesNotExist,
+		},
+		PyTypeObject: PyTypeObject{
+			TPName: p.PyTypeObjectTpName,
+		},
+		PyThreadState: PyThreadState{
+			Interp:         p.PyThreadStateInterp,
+			Next:           p.PyThreadStateNext,
+			Frame:          doesNotExist,
+			ThreadID:       p.PyThreadStateThreadID,
+			NativeThreadID: p.PyThreadStateNativeThreadID,
+			CFrame:         p.PyThreadStateCFrame,
+		},
+		PyCFrame: PyCFrame{
+			CurrentFrame: doesNotExist,
+		},
+		PyInterpreterState: PyInterpreterState{
+			TStateHead: p.PyInterpreterStateTstateHead + p.PyInterpreterStateIsPythreadsHead,
+		},
+		PyRuntimeState: PyRuntimeState{
+			InterpMain: p.PyRuntimeStateInterpreters + p.PyRuntimeStatePyInterpretersMain,
+		},
+		PyFrameObject: PyFrameObject{
+			FBack:       p.PyFrameObjectFBack,
+			FCode:       p.PyFrameObjectFCode,
+			FLineno:     doesNotExist,
+			FLocalsplus: p.PyFrameObjectFLocalsplus,
+		},
+		PyCodeObject: PyCodeObject{
+			CoFilename:    p.PyCodeObjectCoFilename,
+			CoName:        p.PyCodeObjectCoName,
+			CoVarnames:    p.PyCodeObjectCoVarNames,
+			CoFirstlineno: p.PyCodeObjectCoFirstlineno,
+		},
+		PyTupleObject: PyTupleObject{
+			ObItem: p.PyTupleObjectObItem,
+		},
+	}
+}

pkg/python/initialstate.go

@@ -1,31 +1,33 @@
 package python
 
 import (
+	"bytes"
+	"encoding/binary"
 	"fmt"
+	"unsafe"
 
 	"github.com/Masterminds/semver/v3"
 
+	"github.com/parca-dev/runtime-data/pkg/byteorder"
 	"github.com/parca-dev/runtime-data/pkg/runtimedata"
 	"github.com/parca-dev/runtime-data/pkg/version"
 )
 
-type InitialState struct {
-	InterpreterHead    int64 `offsetof:"_PyRuntimeState.interpreters.head" yaml:"interpreter_head"`
-	ThreadStateCurrent int64 `offsetof:"_PyRuntimeState.gilstate.tstate_current" yaml:"tstate_current"`
-}
-
-var above3_7 = version.MustParseConstraints(">=3.7.0")
+var (
+	above3_7  = version.MustParseConstraints("3.7.x - 3.11.x")
+	above3_12 = version.MustParseConstraints(">=3.12.0-0")
+)
 
-func DataMapForInitialState(v string) *InitialState {
+func DataMapForInitialState(v string) runtimedata.InitialStateMap {
 	lookupVersion := semver.MustParse(v)
-	if above3_7.Check(lookupVersion) {
-		return &InitialState{}
+	switch {
+	case above3_7.Check(lookupVersion):
+		return &initialState38{}
+	case above3_12.Check(lookupVersion):
+		return &initialState312{}
 	}
 	// There is no offsets to be extracted version below 3.7
 	// so we return nil
-	// The state kept in the static global variables,
-	// we read them from the memory directly.
-	// Check the agent for more details.
 	return nil
 }
 
@@ -49,3 +51,72 @@ func WithVersion(version string, is InitialState) (InitialStateWithVersion, erro
 		InitialState: is,
 	}, nil
 }
+
+// The state kept in the static global variables,
+// we read them from the memory directly.
+// Check the agent for more details.
+type InitialState struct {
+	InterpreterHead    int64    `yaml:"interpreter_head"`
+	ThreadStateCurrent int64    `yaml:"tstate_current"`
+	AutoTSSKey         int64    `yaml:"auto_tss_key"`
+	PyTSS              PyTSSKey `yaml:"tss"`
+}
+
+type PyTSSKey struct {
+	Key  int64 `yaml:"key"`
+	Size int64 `yaml:"size"`
+}
+
+func (i InitialState) Data() ([]byte, error) {
+	buf := new(bytes.Buffer)
+	buf.Grow(int(unsafe.Sizeof(&i)))
+
+	if err := binary.Write(buf, byteorder.GetHostByteOrder(), &i); err != nil {
+		return nil, err
+	}
+
+	return buf.Bytes(), nil
+}
+
+type initialState312 struct {
+	InterpreterHead int64 `offsetof:"_PyRuntimeState.interpreters.head"`
+	AutoTSSKey      int64 `offsetof:"_PyRuntimeState.autoTSSkey"`
+	PyTSSKey        int64 `offsetof:"_Py_tss_t._key"`
+	PyTSSSize       int64 `sizeof:"_Py_tss_t"`
+}
+
+func (i initialState312) InitialState() runtimedata.RuntimeData {
+	return &InitialState{
+		InterpreterHead: i.InterpreterHead,
+		AutoTSSKey:      i.AutoTSSKey,
+		// https://github.com/python/cpython/issues/103323
+		ThreadStateCurrent: doesNotExist,
+		PyTSS: PyTSSKey{
+			Key:  i.PyTSSKey,
+			Size: i.PyTSSSize,
+		},
+	}
+}
+
+// https://pythondev.readthedocs.io/pystate.html
+// Python 3.7: PyThreadState_GET() reads _PyThreadState_Current (atomic variable).
+// Python 3.8: _PyThreadState_Current becomes _PyRuntime.gilstate.tstate_current.
+type initialState38 struct {
+	InterpreterHead    int64 `offsetof:"_PyRuntimeState.interpreters.head"`
+	ThreadStateCurrent int64 `offsetof:"_PyRuntimeState.gilstate.tstate_current"`
+	AutoTSSKey         int64 `offsetof:"_PyRuntimeState.gilstate.autoTSSkey"`
+	PyTSSKey           int64 `offsetof:"_Py_tss_t._key"`
+	PyTSSSize          int64 `sizeof:"_Py_tss_t"`
+}
+
+func (i initialState38) InitialState() runtimedata.RuntimeData {
+	return &InitialState{
+		InterpreterHead:    i.InterpreterHead,
+		ThreadStateCurrent: i.ThreadStateCurrent,
+		AutoTSSKey:         i.AutoTSSKey,
+		PyTSS: PyTSSKey{
+			Key:  i.PyTSSKey,
+			Size: i.PyTSSSize,
+		},
+	}
+}

pkg/python/initialstate/amd64/3.12.0 - 3.12.2.yaml

@@ -0,0 +1,6 @@
+auto_tss_key: 1544
+interpreter_head: 40
+tss:
+    key: 4
+    size: 8
+tstate_current: -1

pkg/python/initialstate/amd64/3.7.0 - 3.7.2.yaml

@@ -0,0 +1,6 @@
+auto_tss_key: 1416
+interpreter_head: 24
+tss:
+    key: 4
+    size: 8
+tstate_current: 1392

pkg/python/initialstate/amd64/3.8.0 - 3.8.1.yaml

@@ -0,0 +1,6 @@
+auto_tss_key: 1392
+interpreter_head: 32
+tss:
+    key: 4
+    size: 8
+tstate_current: 1368