Read TLS DTV

interpreter/instancestubs.go

@@ -5,11 +5,11 @@ package interpreter // import "go.opentelemetry.io/ebpf-profiler/interpreter"
 
 import (
 	"go.opentelemetry.io/ebpf-profiler/host"
+	"go.opentelemetry.io/ebpf-profiler/libc"
 	"go.opentelemetry.io/ebpf-profiler/libpf"
 	"go.opentelemetry.io/ebpf-profiler/metrics"
 	"go.opentelemetry.io/ebpf-profiler/process"
 	"go.opentelemetry.io/ebpf-profiler/reporter"
-	"go.opentelemetry.io/ebpf-profiler/tpbase"
 )
 
 // InstanceStubs provides empty implementations of Instance hooks that are
@@ -22,7 +22,7 @@ func (is *InstanceStubs) SynchronizeMappings(EbpfHandler, reporter.ExecutableRep
 	return nil
 }
 
-func (is *InstanceStubs) UpdateTSDInfo(EbpfHandler, libpf.PID, tpbase.TSDInfo) error {
+func (is *InstanceStubs) UpdateLibcInfo(EbpfHandler, libpf.PID, libc.LibcInfo) error {
 	return nil
 }
 

interpreter/multi.go

@@ -8,12 +8,12 @@ import (
 
 	"go.opentelemetry.io/ebpf-profiler/host"
 	"go.opentelemetry.io/ebpf-profiler/internal/log"
+	"go.opentelemetry.io/ebpf-profiler/libc"
 	"go.opentelemetry.io/ebpf-profiler/libpf"
 	"go.opentelemetry.io/ebpf-profiler/metrics"
 	"go.opentelemetry.io/ebpf-profiler/process"
 	"go.opentelemetry.io/ebpf-profiler/remotememory"
 	"go.opentelemetry.io/ebpf-profiler/reporter"
-	"go.opentelemetry.io/ebpf-profiler/tpbase"
 )
 
 // MultiData implements the Data interface for multiple interpreters.
@@ -104,11 +104,11 @@ func (m *MultiInstance) SynchronizeMappings(ebpf EbpfHandler,
 	return errors.Join(errs...)
 }
 
-// UpdateTSDInfo updates TSD info for all interpreter instances.
-func (m *MultiInstance) UpdateTSDInfo(ebpf EbpfHandler, pid libpf.PID, info tpbase.TSDInfo) error {
+// UpdateLibcInfo updates libc info for all interpreter instances.
+func (m *MultiInstance) UpdateLibcInfo(ebpf EbpfHandler, pid libpf.PID, info libc.LibcInfo) error {
 	var errs []error
 	for _, instance := range m.instances {
-		if err := instance.UpdateTSDInfo(ebpf, pid, info); err != nil {
+		if err := instance.UpdateLibcInfo(ebpf, pid, info); err != nil {
 			errs = append(errs, err)
 		}
 	}

interpreter/perl/instance.go

@@ -15,14 +15,14 @@ import (
 
 	"go.opentelemetry.io/ebpf-profiler/host"
 	"go.opentelemetry.io/ebpf-profiler/interpreter"
+	"go.opentelemetry.io/ebpf-profiler/libc"
 	"go.opentelemetry.io/ebpf-profiler/libpf"
 	"go.opentelemetry.io/ebpf-profiler/libpf/pfunsafe"
 	"go.opentelemetry.io/ebpf-profiler/metrics"
 	npsr "go.opentelemetry.io/ebpf-profiler/nopanicslicereader"
 	"go.opentelemetry.io/ebpf-profiler/remotememory"
 	"go.opentelemetry.io/ebpf-profiler/successfailurecounter"
 	"go.opentelemetry.io/ebpf-profiler/support"
-	"go.opentelemetry.io/ebpf-profiler/tpbase"
 	"go.opentelemetry.io/ebpf-profiler/util"
 )
 
@@ -73,9 +73,8 @@ func hashCOPKey(k copKey) uint32 {
 	return uint32(h ^ xxh3.HashString128(k.funcName.String()).Lo)
 }
 
-func (i *perlInstance) UpdateTSDInfo(ebpf interpreter.EbpfHandler, pid libpf.PID,
-	tsdInfo tpbase.TSDInfo,
-) error {
+func (i *perlInstance) UpdateLibcInfo(ebpf interpreter.EbpfHandler, pid libpf.PID,
+	libcInfo libc.LibcInfo) error {
 	d := i.d
 	stateInTSD := uint8(0)
 	if d.stateInTSD {
@@ -88,9 +87,9 @@ func (i *perlInstance) UpdateTSDInfo(ebpf interpreter.EbpfHandler, pid libpf.PID
 		StateInTSD: stateInTSD,
 
 		TsdInfo: support.TSDInfo{
-			Offset:     tsdInfo.Offset,
-			Multiplier: tsdInfo.Multiplier,
-			Indirect:   tsdInfo.Indirect,
+			Offset:     libcInfo.TSDInfo.Offset,
+			Multiplier: libcInfo.TSDInfo.Multiplier,
+			Indirect:   libcInfo.TSDInfo.Indirect,
 		},
 
 		Interpreter_curcop:       uint16(vms.interpreter.curcop),

interpreter/python/python.go

@@ -26,14 +26,14 @@ import (
 
 	"go.opentelemetry.io/ebpf-profiler/host"
 	"go.opentelemetry.io/ebpf-profiler/interpreter"
+	"go.opentelemetry.io/ebpf-profiler/libc"
 	"go.opentelemetry.io/ebpf-profiler/libpf"
 	"go.opentelemetry.io/ebpf-profiler/libpf/pfelf"
 	"go.opentelemetry.io/ebpf-profiler/metrics"
 	npsr "go.opentelemetry.io/ebpf-profiler/nopanicslicereader"
 	"go.opentelemetry.io/ebpf-profiler/remotememory"
 	"go.opentelemetry.io/ebpf-profiler/successfailurecounter"
 	"go.opentelemetry.io/ebpf-profiler/support"
-	"go.opentelemetry.io/ebpf-profiler/tpbase"
 	"go.opentelemetry.io/ebpf-profiler/util"
 )
 
@@ -371,19 +371,18 @@ func (p *pythonInstance) GetAndResetMetrics() ([]metrics.Metric, error) {
 	}, nil
 }
 
-func (p *pythonInstance) UpdateTSDInfo(ebpf interpreter.EbpfHandler, pid libpf.PID,
-	tsdInfo tpbase.TSDInfo,
-) error {
+func (p *pythonInstance) UpdateLibcInfo(ebpf interpreter.EbpfHandler, pid libpf.PID,
+	libcInfo libc.LibcInfo) error {
 	d := p.d
 	vm := &d.vmStructs
 	cdata := support.PyProcInfo{
 		AutoTLSKeyAddr: uint64(d.autoTLSKey) + uint64(p.bias),
 		Version:        d.version,
 
 		TsdInfo: support.TSDInfo{
-			Offset:     tsdInfo.Offset,
-			Multiplier: tsdInfo.Multiplier,
-			Indirect:   tsdInfo.Indirect,
+			Offset:     libcInfo.TSDInfo.Offset,
+			Multiplier: libcInfo.TSDInfo.Multiplier,
+			Indirect:   libcInfo.TSDInfo.Indirect,
 		},
 
 		PyThreadState_frame:            uint8(vm.PyThreadState.Frame),

interpreter/ruby/ruby.go

@@ -4,6 +4,7 @@
 package ruby // import "go.opentelemetry.io/ebpf-profiler/interpreter/ruby"
 
 import (
+	"debug/elf"
 	"encoding/binary"
 	"errors"
 	"fmt"
@@ -92,6 +93,12 @@ type rubyData struct {
 	// Address to the ruby_current_ec variable in TLS, as an offset from tpbase
 	currentEcTpBaseTlsOffset libpf.Address
 
+	// Offset of the current EC within TLS for this module
+	currentEcTlsOffset libpf.Address
+
+	// TLS Module ID offset in ELF, to read mod ID after loaded and written by linker
+	tlsModuleIdOffset libpf.Address
+
 	// version of the currently used Ruby interpreter.
 	// major*0x10000 + minor*0x100 + release (e.g. 3.0.1 -> 0x30001)
 	version uint32
@@ -192,11 +199,19 @@ func (r *rubyData) Attach(ebpf interpreter.EbpfHandler, pid libpf.PID, bias libp
 		tlsOffset = rm.Uint64(bias + r.currentEcTpBaseTlsOffset + 8)
 	}
 
+	var modId uint64
+	if r.tlsModuleIdOffset != 0 {
+		modId = rm.Uint64(bias + r.tlsModuleIdOffset)
+		log.Debugf("Read TLS module as %d", modId)
+	}
+
 	cdata := support.RubyProcInfo{
 		Version: r.version,
 
 		Current_ctx_ptr:              uint64(r.currentCtxPtr + bias),
 		Current_ec_tpbase_tls_offset: tlsOffset,
+		Current_ec_tls_offset:        uint64(r.currentEcTlsOffset),
+		Tls_module_id:                modId,
 
 		Vm_stack:      r.vmStructs.execution_context_struct.vm_stack,
 		Vm_stack_size: r.vmStructs.execution_context_struct.vm_stack_size,
@@ -853,11 +868,28 @@ func Loader(ebpf interpreter.EbpfHandler, info *interpreter.LoaderInfo) (interpr
 		log.Warnf("failed to locate TLS descriptor: %v", err)
 	}
 
+	var tlsModuleIdOffset libpf.Address
+	if err = ef.VisitRelocations(func(r pfelf.ElfReloc, symName string) bool {
+		// We've already verified that the relocation is DTPMOD type
+		// so if we get here, we should be able to read the module ID from this offset
+		log.Debugf("Found module id at %x", r.Off)
+		tlsModuleIdOffset = libpf.Address(r.Off)
+		return false
+	}, func(rela pfelf.ElfReloc) bool {
+		ty := rela.Info & 0xffff
+		return (ef.Machine == elf.EM_AARCH64 && elf.R_AARCH64(ty) == elf.R_AARCH64_TLS_DTPMOD64) ||
+			(ef.Machine == elf.EM_X86_64 && elf.R_X86_64(ty) == elf.R_X86_64_DTPMOD64)
+	}); err != nil {
+		log.Warnf("failed to mod offset: %v", err)
+	}
+
 	log.Debugf("Discovered EC tls tpbase offset %x, fallback ctx %x, interp ranges: %v", currentEcTpBaseTlsOffset, currentCtxPtr, interpRanges)
 
 	rid := &rubyData{
 		version:                  version,
 		currentEcTpBaseTlsOffset: libpf.Address(currentEcTpBaseTlsOffset),
+		tlsModuleIdOffset:        tlsModuleIdOffset,
+		currentEcTlsOffset:       libpf.Address(currentEcSymbolAddress),
 		currentCtxPtr:            libpf.Address(currentCtxPtr),
 	}
 

interpreter/types.go

@@ -8,13 +8,13 @@ import (
 	"unsafe"
 
 	"go.opentelemetry.io/ebpf-profiler/host"
+	"go.opentelemetry.io/ebpf-profiler/libc"
 	"go.opentelemetry.io/ebpf-profiler/libpf"
 	"go.opentelemetry.io/ebpf-profiler/lpm"
 	"go.opentelemetry.io/ebpf-profiler/metrics"
 	"go.opentelemetry.io/ebpf-profiler/process"
 	"go.opentelemetry.io/ebpf-profiler/remotememory"
 	"go.opentelemetry.io/ebpf-profiler/reporter"
-	"go.opentelemetry.io/ebpf-profiler/tpbase"
 	"go.opentelemetry.io/ebpf-profiler/util"
 )
 
@@ -143,9 +143,9 @@ type Instance interface {
 	SynchronizeMappings(ebpf EbpfHandler, exeReporter reporter.ExecutableReporter,
 		pr process.Process, mappings []process.Mapping) error
 
-	// UpdateTSDInfo is called when the process C-library Thread Specific Data related
+	// UpdateLibcInfo is called when the process C-library related
 	// introspection data has been updated.
-	UpdateTSDInfo(ebpf EbpfHandler, pid libpf.PID, info tpbase.TSDInfo) error
+	UpdateLibcInfo(ebpf EbpfHandler, pid libpf.PID, info libc.LibcInfo) error
 
 	// Symbolize converts one ebpf frame to one or more (if inlining was expanded) libpf.Frame.
 	// The resulting libpf.Frame values are appended to frames.

tpbase/assembly_decode_aarch64.go → libc/assembly_decode_aarch64.go

@@ -1,7 +1,7 @@
 // Copyright The OpenTelemetry Authors
 // SPDX-License-Identifier: Apache-2.0
 
-package tpbase // import "go.opentelemetry.io/ebpf-profiler/tpbase"
+package libc // import "go.opentelemetry.io/ebpf-profiler/libc"
 
 import (
 	"errors"

tpbase/assembly_decode_test.go → libc/assembly_decode_test.go

@@ -1,7 +1,7 @@
 // Copyright The OpenTelemetry Authors
 // SPDX-License-Identifier: Apache-2.0
 
-package tpbase
+package libc
 
 import (
 	"debug/elf"

tpbase/assembly_decode_x86.go → libc/assembly_decode_x86.go

@@ -1,7 +1,7 @@
 // Copyright The OpenTelemetry Authors
 // SPDX-License-Identifier: Apache-2.0
 
-package tpbase // import "go.opentelemetry.io/ebpf-profiler/tpbase"
+package libc // import "go.opentelemetry.io/ebpf-profiler/libc"
 
 import (
 	"bytes"

tpbase/libc.go → libc/libc.go

@@ -1,7 +1,7 @@
 // Copyright The OpenTelemetry Authors
 // SPDX-License-Identifier: Apache-2.0
 
-package tpbase // import "go.opentelemetry.io/ebpf-profiler/tpbase"
+package libc // import "go.opentelemetry.io/ebpf-profiler/libc"
 
 import (
 	"debug/elf"
@@ -11,6 +11,17 @@ import (
 	"go.opentelemetry.io/ebpf-profiler/libpf/pfelf"
 )
 
+type LibcInfo struct {
+	DTVInfo *DTVInfo
+	TSDInfo *TSDInfo
+}
+
+type DTVInfo struct {
+	Offset     int32  // Offset of DTV from FS base (or from thread pointer)
+	EntryWidth uint32 // Width of each DTV entry in bytes
+	Indirect   uint8  // 0 if DTV is at FS+offset, 1 if at [FS+0]+offset
+}
+
 // TSDInfo contains information to access C-library's Thread Specific Data from eBPF
 type TSDInfo struct {
 	// Offset is the pointer difference from "tpbase" pointer to the C-library
@@ -84,6 +95,23 @@ func IsPotentialTSDDSO(filename string) bool {
 	return libcRegex.MatchString(filename)
 }
 
+func ExtractLibcInfo(ef *pfelf.File) (*LibcInfo, error) {
+	tsdinfo, err := ExtractTSDInfo(ef)
+	if err != nil {
+		return nil, err
+	}
+
+	dtvinfo, err := extractDTVInfo(ef)
+	if err != nil {
+		return nil, err
+	}
+
+	return &LibcInfo{
+		TSDInfo: tsdinfo,
+		DTVInfo: dtvinfo,
+	}, nil
+}
+
 // ExtractTSDInfo extracts the introspection data for pthread thread specific data.
 func ExtractTSDInfo(ef *pfelf.File) (*TSDInfo, error) {
 	_, code, err := ef.SymbolData("__pthread_getspecific", 2048)
@@ -111,3 +139,28 @@ func ExtractTSDInfo(ef *pfelf.File) (*TSDInfo, error) {
 	}
 	return &info, nil
 }
+
+// extractDTVInfo extracts the introspection data for the DTV to access TLS vars
+func extractDTVInfo(ef *pfelf.File) (*DTVInfo, error) {
+	_, code, err := ef.SymbolData("__tls_get_addr", 2048)
+	if err != nil {
+		return nil, fmt.Errorf("unable to read '__tls_get_addr': %s", err)
+	}
+	if len(code) < 8 {
+		return nil, fmt.Errorf("__tls_get_addr function size is %d", len(code))
+	}
+
+	var info DTVInfo
+	switch ef.Machine {
+	case elf.EM_AARCH64:
+		info, err = extractDTVInfoARM(code)
+	case elf.EM_X86_64:
+		info, err = extractDTVInfoX86(code)
+	default:
+		return nil, fmt.Errorf("unsupported arch %s", ef.Machine.String())
+	}
+	if err != nil {
+		return nil, fmt.Errorf("failed to extract DTV data: %s", err)
+	}
+	return &info, nil
+}