fix(ebpf): python backport fixes (#3268)

* fix(ebpf): fix glibc detection when the file name is libc-2.31.so instead of libc.so.6 * fix(ebpf): elf build id
grafana · May 6, 2024 · 5aed1e5 · 5aed1e5
1 parent e23c664
commit 5aed1e5
Show file tree

Hide file tree

Showing 10 changed files with 46 additions and 59 deletions.
diff --git a/ebpf/Makefile b/ebpf/Makefile
@@ -1,6 +1,6 @@
 GO ?= go
-RIDESHARE_REPO ?= korniltsev
-RIDESHARE="testdata/rideshare-flask-no-pip"
+RIDESHARE_REPO ?= pyroscope
+RIDESHARE=testdata/rideshare-flask-no-pip
 
 ifeq ($(shell uname -s),Linux)
 ifeq ($(shell uname -m),x86_64)
@@ -29,12 +29,12 @@ python/dwarfdump:
 	git submodule update --init --recursive
 
 	echo "//go:build amd64 && linux" > python/python_offsets_gen_amd64.go
-	go run cmd/python_dwarfdump/main.go $(shell find testdata/python-x64 -name libpy\*.so\*) \
+	go run cmd/python_dwarfdump/main.go $(shell find testdata/python-x64 -name libpy\*.so\* | grep -v pyston) \
 	 	$(shell find testdata/python-x64  | grep -E "/python3\\.[0-9]+") >> python/python_offsets_gen_amd64.go
 	go fmt python/python_offsets_gen_amd64.go
 
 	echo "//go:build arm64 && linux" > python/python_offsets_gen_arm64.go
-	go run cmd/python_dwarfdump/main.go $(shell find testdata/python-arm64 -name libpy\*.so\*) \
+	go run cmd/python_dwarfdump/main.go $(shell find testdata/python-arm64 -name libpy\*.so\* | grep -v pyston) \
 	 	$(shell find testdata/python-arm64  | grep -E "/python3\\.[0-9]+") >> python/python_offsets_gen_arm64.go
 	go fmt python/python_offsets_gen_arm64.go
 
@@ -104,4 +104,6 @@ rideshare/gen:
 	docker buildx build --platform=linux/amd64,linux/arm64 --push -t $(RIDESHARE_REPO)/ebpf-testdata-rideshare:3.10-alpine     --build-arg="PYTHON_VERSION=3.10-alpine"    $(RIDESHARE)
 	docker buildx build --platform=linux/amd64,linux/arm64 --push -t $(RIDESHARE_REPO)/ebpf-testdata-rideshare:3.11-alpine     --build-arg="PYTHON_VERSION=3.11-alpine"    $(RIDESHARE)
 	docker buildx build --platform=linux/amd64,linux/arm64 --push -t $(RIDESHARE_REPO)/ebpf-testdata-rideshare:3.12-alpine     --build-arg="PYTHON_VERSION=3.12-alpine"    $(RIDESHARE)
-	docker buildx build --platform=linux/amd64,linux/arm64 --push -t $(RIDESHARE_REPO)/ebpf-testdata-rideshare:3.13-rc-alpine  --build-arg="PYTHON_VERSION=3.13-rc-alpine" $(RIDESHARE)
+	docker buildx build --platform=linux/amd64,linux/arm64 --push -t $(RIDESHARE_REPO)/ebpf-testdata-rideshare:3.13-rc-alpine  --build-arg="PYTHON_VERSION=3.13-rc-alpine" $(RIDESHARE)
+	docker buildx build --platform=linux/amd64,linux/arm64 --push -t $(RIDESHARE_REPO)/ebpf-testdata-rideshare:ubuntu-20.04  --build-arg="BASE=ubuntu:20.04" --build-arg="FLASK_VERSION=3.0.3" -f $(RIDESHARE)/ubuntu.Dockerfile $(RIDESHARE)
+	docker buildx build --platform=linux/amd64,linux/arm64 --push -t $(RIDESHARE_REPO)/ebpf-testdata-rideshare:ubuntu-22.04  --build-arg="BASE=ubuntu:22.04" --build-arg="FLASK_VERSION=3.0.3" -f $(RIDESHARE)/ubuntu.Dockerfile $(RIDESHARE)
diff --git a/ebpf/cmd/python_dwarfdump/main.go b/ebpf/cmd/python_dwarfdump/main.go
@@ -66,7 +66,7 @@ var pythonFields = []dwarfdump.Need{
 	{Name: "_typeobject", PrettyName: "PyTypeObject", Fields: []dwarfdump.NeedField{
 		{"tp_name", "PyTypeObject_tp_name"},
 	}},
-	{Name: "PyThreadState", Fields: []dwarfdump.NeedField{
+	{Name: "_ts", Fields: []dwarfdump.NeedField{
 		{"frame", "PyThreadState_frame"},
 		{"cframe", "PyThreadState_cframe"},
 		{"current_frame", "PyThreadState_current_frame"},

diff --git a/ebpf/dwarfdump/dwarfdump.go b/ebpf/dwarfdump/dwarfdump.go
@@ -4,6 +4,7 @@ import (
 	"debug/dwarf"
 	"debug/elf"
 	"fmt"
+	"os"
 	"reflect"
 	"strings"
 )
@@ -193,6 +194,7 @@ type FieldDump struct {
 }
 
 func Dump(elfPath string, fields []Need) []FieldDump {
+	fmt.Fprintf(os.Stderr, "Dumping %s\n", elfPath)
 	var err error
 
 	f, err := elf.Open(elfPath)

diff --git a/ebpf/python/procinfo.go b/ebpf/python/procinfo.go
@@ -3,6 +3,7 @@ package python
 import (
 	"bufio"
 	"fmt"
+	"path/filepath"
 	"regexp"
 	"strconv"
 	"strings"
@@ -58,7 +59,7 @@ func GetProcInfo(s *bufio.Scanner) (ProcInfo, error) {
 				strings.Contains(m.Pathname, "/lib/ld-musl-aarch64.so.1") {
 				res.Musl = append(res.Musl, m)
 			}
-			if strings.HasSuffix(m.Pathname, "/libc.so.6") || strings.HasSuffix(m.Pathname, "/libc-2") {
+			if strings.HasSuffix(m.Pathname, "/libc.so.6") || strings.HasPrefix(filepath.Base(m.Pathname), "libc-2.") {
 				res.Glibc = append(res.Glibc, m)
 			}
 		}

diff --git a/ebpf/python/procinfo_test.go b/ebpf/python/procinfo_test.go
@@ -56,6 +56,7 @@ ffffffffff600000-ffffffffff601000 --xp 00000000 00:00 0                  [vsysca
 	info, err := GetProcInfo(bufio.NewScanner(bytes.NewReader([]byte(maps))))
 	require.NoError(t, err)
 	require.Nil(t, info.Musl)
+	require.NotNil(t, info.Glibc)
 	require.Equal(t, info.Version, Version{3, 6, 0})
 	require.NotNil(t, info.PythonMaps)
 	require.NotNil(t, info.LibPythonMaps)
@@ -116,6 +117,7 @@ ffffffffff600000-ffffffffff601000 --xp 00000000 00:00 0                  [vsysca
 	info, err = GetProcInfo(bufio.NewScanner(bytes.NewReader([]byte(maps))))
 	require.NoError(t, err)
 	require.NotNil(t, info.Musl)
+	require.Nil(t, info.Glibc)
 	require.Equal(t, info.Version, Version{3, 11, 0})
 	require.NotNil(t, info.PythonMaps)
 	require.NotNil(t, info.LibPythonMaps)
@@ -128,6 +130,7 @@ ffffffffff600000-ffffffffff601000 --xp 00000000 00:00 0                  [vsysca
 	info, err = GetProcInfo(bufio.NewScanner(bytes.NewReader([]byte(maps))))
 	require.NoError(t, err)
 	require.Nil(t, info.Musl)
+	require.Nil(t, info.Glibc)
 	require.Equal(t, info.Version, Version{3, 7, 0})
 	require.NotNil(t, info.PythonMaps)
 	require.Nil(t, info.LibPythonMaps)
@@ -160,6 +163,7 @@ fffff8c25000-fffff8c53000 rw-p 00000000 00:00 0                          [stack]
 	info, err = GetProcInfo(bufio.NewScanner(bytes.NewReader([]byte(maps))))
 	require.NoError(t, err)
 	require.Nil(t, info.Musl)
+	require.NotNil(t, info.Glibc)
 	require.Equal(t, info.Version, Version{3, 8, 0})
 	require.Nil(t, info.PythonMaps)
 	require.NotNil(t, info.LibPythonMaps)

diff --git a/ebpf/python/python_offsets_gen_amd64.go b/ebpf/python/python_offsets_gen_amd64.go
diff --git a/ebpf/python/python_offsets_gen_arm64.go b/ebpf/python/python_offsets_gen_arm64.go
diff --git a/ebpf/python3_ebpf_expected.txt b/ebpf/python3_ebpf_expected.txt
@@ -0,0 +1,4 @@
+python3;server.py <module>;app.py Flask.run;serving.py run_simple;serving.py BaseWSGIServer.serve_forever;socketserver.py BaseWSGIServer.serve_forever;socketserver.py BaseWSGIServer._handle_request_noblock;socketserver.py BaseWSGIServer.process_request;socketserver.py BaseWSGIServer.finish_request;socketserver.py WSGIRequestHandler.__init__;serving.py WSGIRequestHandler.handle;server.py WSGIRequestHandler.handle;server.py WSGIRequestHandler.handle_one_request;serving.py NullCls.run_wsgi;serving.py execute;app.py Flask.__call__;app.py Flask.wsgi_app;app.py Flask.full_dispatch_request;app.py Flask.dispatch_request;server.py bike;bike.py order_bike;utility.py find_nearest_vehicle
+python3;server.py <module>;app.py Flask.run;serving.py run_simple;serving.py BaseWSGIServer.serve_forever;socketserver.py BaseWSGIServer.serve_forever;socketserver.py BaseWSGIServer._handle_request_noblock;socketserver.py BaseWSGIServer.process_request;socketserver.py BaseWSGIServer.finish_request;socketserver.py WSGIRequestHandler.__init__;serving.py WSGIRequestHandler.handle;server.py WSGIRequestHandler.handle;server.py WSGIRequestHandler.handle_one_request;serving.py NullCls.run_wsgi;serving.py execute;app.py Flask.__call__;app.py Flask.wsgi_app;app.py Flask.full_dispatch_request;app.py Flask.dispatch_request;server.py car;car.py order_car;utility.py find_nearest_vehicle
+python3;server.py <module>;app.py Flask.run;serving.py run_simple;serving.py BaseWSGIServer.serve_forever;socketserver.py BaseWSGIServer.serve_forever;socketserver.py BaseWSGIServer._handle_request_noblock;socketserver.py BaseWSGIServer.process_request;socketserver.py BaseWSGIServer.finish_request;socketserver.py WSGIRequestHandler.__init__;serving.py WSGIRequestHandler.handle;server.py WSGIRequestHandler.handle;server.py WSGIRequestHandler.handle_one_request;serving.py NullCls.run_wsgi;serving.py execute;app.py Flask.__call__;app.py Flask.wsgi_app;app.py Flask.full_dispatch_request;app.py Flask.dispatch_request;server.py car;car.py order_car;utility.py find_nearest_vehicle;utility.py check_driver_availability
+python3;server.py <module>;app.py Flask.run;serving.py run_simple;serving.py BaseWSGIServer.serve_forever;socketserver.py BaseWSGIServer.serve_forever;socketserver.py BaseWSGIServer._handle_request_noblock;socketserver.py BaseWSGIServer.process_request;socketserver.py BaseWSGIServer.finish_request;socketserver.py WSGIRequestHandler.__init__;serving.py WSGIRequestHandler.handle;server.py WSGIRequestHandler.handle;server.py WSGIRequestHandler.handle_one_request;serving.py NullCls.run_wsgi;serving.py execute;app.py Flask.__call__;app.py Flask.wsgi_app;app.py Flask.full_dispatch_request;app.py Flask.dispatch_request;server.py scooter;scooter.py order_scooter;utility.py find_nearest_vehicle
diff --git a/ebpf/python_ebpf_test.go b/ebpf/python_ebpf_test.go
@@ -25,23 +25,28 @@ var pythonEBPFExpected []byte
 //go:embed python_ebpf_expected_3.11.txt
 var pythonEBPFExpected311 []byte
 
+//go:embed python3_ebpf_expected.txt
+var python3EBPFExpected []byte
+
 func TestEBPFPythonProfiler(t *testing.T) {
 	var testdata = []struct {
 		image    string
 		expected []byte
 	}{
-		{"korniltsev/ebpf-testdata-rideshare:3.8-slim", pythonEBPFExpected},
-		{"korniltsev/ebpf-testdata-rideshare:3.9-slim", pythonEBPFExpected},
-		{"korniltsev/ebpf-testdata-rideshare:3.10-slim", pythonEBPFExpected},
-		{"korniltsev/ebpf-testdata-rideshare:3.11-slim", pythonEBPFExpected311},
-		{"korniltsev/ebpf-testdata-rideshare:3.12-slim", pythonEBPFExpected311},
-		{"korniltsev/ebpf-testdata-rideshare:3.13-rc-slim", pythonEBPFExpected311},
-		{"korniltsev/ebpf-testdata-rideshare:3.8-alpine", pythonEBPFExpected},
-		{"korniltsev/ebpf-testdata-rideshare:3.9-alpine", pythonEBPFExpected},
-		{"korniltsev/ebpf-testdata-rideshare:3.10-alpine", pythonEBPFExpected},
-		{"korniltsev/ebpf-testdata-rideshare:3.11-alpine", pythonEBPFExpected311},
-		{"korniltsev/ebpf-testdata-rideshare:3.12-alpine", pythonEBPFExpected311},
-		{"korniltsev/ebpf-testdata-rideshare:3.13-rc-alpine", pythonEBPFExpected311},
+		{"pyroscope/ebpf-testdata-rideshare:3.8-slim", pythonEBPFExpected},
+		{"pyroscope/ebpf-testdata-rideshare:3.9-slim", pythonEBPFExpected},
+		{"pyroscope/ebpf-testdata-rideshare:3.10-slim", pythonEBPFExpected},
+		{"pyroscope/ebpf-testdata-rideshare:3.11-slim", pythonEBPFExpected311},
+		{"pyroscope/ebpf-testdata-rideshare:3.12-slim", pythonEBPFExpected311},
+		{"pyroscope/ebpf-testdata-rideshare:3.13-rc-slim", pythonEBPFExpected311},
+		{"pyroscope/ebpf-testdata-rideshare:3.8-alpine", pythonEBPFExpected},
+		{"pyroscope/ebpf-testdata-rideshare:3.9-alpine", pythonEBPFExpected},
+		{"pyroscope/ebpf-testdata-rideshare:3.10-alpine", pythonEBPFExpected},
+		{"pyroscope/ebpf-testdata-rideshare:3.11-alpine", pythonEBPFExpected311},
+		{"pyroscope/ebpf-testdata-rideshare:3.12-alpine", pythonEBPFExpected311},
+		{"pyroscope/ebpf-testdata-rideshare:3.13-rc-alpine", pythonEBPFExpected311},
+		{"pyroscope/ebpf-testdata-rideshare:ubuntu-20.04", python3EBPFExpected},
+		{"pyroscope/ebpf-testdata-rideshare:ubuntu-22.04", python3EBPFExpected},
 	}
 
 	const ridesharePort = "5000"

diff --git a/ebpf/symtab/elf/buildid.go b/ebpf/symtab/elf/buildid.go
@@ -93,7 +93,7 @@ func (f *MMapedElfFile) GNUBuildID() (BuildID, error) {
 		return BuildID{}, fmt.Errorf(".note.gnu.build-id is not a GNU build-id : %s", hex.EncodeToString(data))
 	}
 	rawBuildID := data[16:]
-	if len(rawBuildID) != 20 && len(rawBuildID) != 8 { // 8 is xxhash, for example in Container-Optimized OS
+	if len(rawBuildID) < 8 {
 		return BuildID{}, fmt.Errorf(".note.gnu.build-id has wrong size %s : %s ", f.fpath, hex.EncodeToString(data))
 	}
 	buildIDHex := hex.EncodeToString(rawBuildID)