feat: migrate interpreter symbolization from agent to server

agent/crates/trace-utils/src/lib.rs

@@ -575,15 +575,6 @@ extern "C" {
     pub fn is_v8_process(pid: u32) -> bool;
 
     pub fn lua_detect(pid: u32, out: *mut LuaRuntimeInfo) -> i32;
-    pub fn lua_format_folded_stack_trace(
-        tracer: *mut libc::c_void,
-        pid: u32,
-        frames: *const u64,
-        frame_count: u32,
-        new_cache: bool,
-        info_p: *mut libc::c_void,
-        err_tag: *const libc::c_char,
-    ) -> *mut libc::c_char;
     pub fn lua_set_map_fds(
         lang_flags_fd: i32,
         unwind_info_fd: i32,
@@ -622,31 +613,6 @@ extern "C" {
     pub fn v8_unwind_table_load(table: *mut V8UnwindTable, pid: u32);
     pub fn v8_unwind_table_unload(table: *mut V8UnwindTable, pid: u32);
 
-    pub fn merge_lua_stacks(
-        trace_str: *mut libc::c_void,
-        len: usize,
-        u_trace: *const libc::c_void,
-        i_trace: *const libc::c_void,
-    ) -> usize;
-    pub fn merge_python_stacks(
-        trace_str: *mut libc::c_void,
-        len: usize,
-        i_trace: *const libc::c_void,
-        u_trace: *const libc::c_void,
-    ) -> usize;
-    pub fn merge_php_stacks(
-        trace_str: *mut libc::c_void,
-        len: usize,
-        i_trace: *const libc::c_void,
-        u_trace: *const libc::c_void,
-    ) -> usize;
-    pub fn merge_v8_stacks(
-        trace_str: *mut libc::c_void,
-        len: usize,
-        i_trace: *const libc::c_void,
-        u_trace: *const libc::c_void,
-    ) -> usize;
-
     pub fn resolve_php_frame(
         pid: u32,
         zend_function_ptr: u64,

agent/crates/trace-utils/src/trace_utils.h

@@ -608,16 +608,6 @@ extern bool is_v8_process(uint32_t pid);
 extern int32_t lua_detect(uint32_t pid, lua_runtime_info_t *out);
 #endif
 
-#if defined(DF_ENTERPRISE)
-extern char *lua_format_folded_stack_trace(void *tracer,
-                                           uint32_t pid,
-                                           const uint64_t *frames,
-                                           uint32_t frame_count,
-                                           bool new_cache,
-                                           void *info_p,
-                                           const char *err_tag);
-#endif
-
 #if defined(DF_ENTERPRISE)
 extern void lua_set_map_fds(int32_t lang_flags_fd,
                             int32_t unwind_info_fd,
@@ -644,34 +634,6 @@ extern void lua_unwind_table_load(lua_unwind_table_t *table, uint32_t pid);
 extern void lua_unwind_table_unload(lua_unwind_table_t *table, uint32_t pid);
 #endif
 
-#if defined(DF_ENTERPRISE)
-extern size_t merge_lua_stacks(void *trace_str,
-                               size_t len,
-                               const void *u_trace,
-                               const void *i_trace);
-#endif
-
-#if defined(DF_ENTERPRISE)
-extern size_t merge_php_stacks(void *trace_str,
-                               size_t len,
-                               const void *i_trace,
-                               const void *u_trace);
-#endif
-
-#if defined(DF_ENTERPRISE)
-extern size_t merge_python_stacks(void *trace_str,
-                                  size_t len,
-                                  const void *i_trace,
-                                  const void *u_trace);
-#endif
-
-#if defined(DF_ENTERPRISE)
-extern size_t merge_v8_stacks(void *trace_str,
-                              size_t len,
-                              const void *i_trace,
-                              const void *u_trace);
-#endif
-
 #if defined(DF_ENTERPRISE)
 extern php_unwind_table_t *php_unwind_table_create(int32_t unwind_info_map_fd,
                                                    int32_t offsets_map_fd);

agent/src/ebpf/mod.rs

@@ -544,6 +544,9 @@ pub struct stack_profile_data {
     pub comm: [u8; PACKET_KNAME_MAX_PADDING + 1],
     pub process_name: [u8; PACKET_KNAME_MAX_PADDING + 1], // process name
     pub container_id: [u8; CONTAINER_ID_SIZE],            // container id
+    pub interp_frame_count: u32,                          // number of structured interpreter frames
+    pub interp_frames_ptr: u64,                           // pointer to CSymbolInfo array
+    pub raw_interpreter_data: u8,                         // 1 = has structured interpreter data
     pub stack_data_len: u32,                              // stack data length
 
     /*
@@ -558,6 +561,22 @@ pub struct stack_profile_data {
     pub stack_data: *mut c_char,
 }
 
+/// C-compatible interpreter frame info struct.
+/// Must match the C `interp_symbol_info_t` layout defined in extended.h.
+#[repr(C)]
+#[derive(Debug, Copy, Clone)]
+pub struct CInterpreterFrameInfo {
+    pub frame_type: u32,
+    pub function_name: *mut c_char,
+    pub class_name: *mut c_char,
+    pub lineno: u32,
+    pub file_name: *mut c_char,
+    pub sub_type: u32,
+    pub is_jit: u8,
+    pub raw_addr: u64,
+    pub resolve_failed: u8,
+}
+
 extern "C" {
     /*
      * Set maximum amount of data passed to the agent by eBPF program.

agent/src/ebpf/user/extended/extended.c

@@ -26,6 +26,7 @@
 #include "../tracer.h"
 #include "../socket.h"
 #include "../proc.h"
+#include "extended.h"
 
 int __attribute__ ((weak)) extended_reader_create(struct bpf_tracer *tracer)
 {
@@ -91,19 +92,31 @@ int __attribute__ ((weak)) print_extra_pkt_info(bool datadump_enable,
 	return 0;
 }
 
-char * __attribute__ ((weak)) extended_resolve_frame(int pid, u64 addr, u8 frame_type, u64 extra_a, u64 extra_b)
+int __attribute__ ((weak)) extended_extract_interpreter_frames(int pid,
+                                                               const u8 *frame_types,
+                                                               const u64 *addrs,
+                                                               const u64 *extra_data_a,
+                                                               const u64 *extra_data_b,
+                                                               int frame_count,
+                                                               void *tracer,
+                                                               bool new_cache,
+                                                               void *info_p,
+                                                               interp_symbol_info_t *out_frames,
+                                                               int max_out)
 {
-	return NULL;
+	return 0;
 }
 
-int __attribute__ ((weak)) extended_merge_stacks(char *dst, int len, const char *i_trace, const char *u_trace, int pid)
+void __attribute__ ((weak)) extended_free_interp_frames(interp_symbol_info_t *frames, int count)
 {
-	return 0;
 }
 
-char * __attribute__ ((weak)) extended_format_lua_stack(void *tracer, int pid, int stack_id,
-                                                        const char *stack_map_name, void *h,
-                                                        bool new_cache, void *info_p)
+int __attribute__ ((weak)) extended_extract_structured_frames(void *tracer, int tgid,
+                                                              int user_stack_id, int interp_stack_id,
+                                                              const char *custom_stack_map_name,
+                                                              bool new_cache, void *info_p,
+                                                              interp_symbol_info_t *out_frames,
+                                                              int max_out)
 {
-	return NULL;
+	return 0;
 }

agent/src/ebpf/user/extended/extended.h

@@ -110,40 +110,80 @@ int print_extra_pkt_info(bool datadump_enable, const char *pkt_data, int len,
 			 char *buf, int buf_len, u8 direction);
 
 /**
- * @brief **extended_resolve_frame()** Resolve a custom/interpreter frame
- * @param pid Process ID
- * @param addr Frame address/ID
- * @param frame_type Frame type identifier
- * @param extra_a Extra data A from stack map
- * @param extra_b Extra data B from stack map
- * @return Resolved symbol string (must be freed) or NULL
+ * @brief Structured interpreter symbol info for per-frame extraction.
+ * Matches the Rust CSymbolInfo layout (#[repr(C)]).
  */
-char *extended_resolve_frame(int pid, u64 addr, u8 frame_type, u64 extra_a, u64 extra_b);
+#ifndef INTERP_SYMBOL_INFO_DEFINED
+#define INTERP_SYMBOL_INFO_DEFINED
+typedef struct {
+	u32 frame_type;		// FRAME_TYPE_PHP/V8/LUA/PYTHON
+	char *function_name;	// allocated via clib_mem_alloc
+	char *class_name;	// allocated via clib_mem_alloc (or NULL)
+	u32 lineno;
+	char *file_name;	// allocated via clib_mem_alloc (or NULL)
+	u32 sub_type;		// language-specific sub-type
+	u8 is_jit;		// 1 = JIT-compiled frame
+	u64 raw_addr;		// original address
+	u8 resolve_failed;	// 1 = resolution failed
+} interp_symbol_info_t;
+#endif
 
 /**
- * @brief **extended_merge_stacks()** Merge interpreter and user stacks
- * @param dst Destination buffer
- * @param len Buffer length
- * @param i_trace Interpreter stack string
- * @param u_trace User stack string
+ * @brief **extended_extract_interpreter_frames()** Extract structured interpreter frames
  * @param pid Process ID
- * @return Bytes written
+ * @param frame_types Array of frame types from BPF map
+ * @param addrs Array of frame addresses
+ * @param extra_data_a Array of extra data A values
+ * @param extra_data_b Array of extra data B values
+ * @param frame_count Number of frames in arrays
+ * @param tracer BPF tracer handle (for Lua)
+ * @param new_cache Whether this is a new cache entry (for Lua)
+ * @param info_p Process info pointer (for Lua)
+ * @param out_frames Output array of interp_symbol_info_t (caller-allocated)
+ * @param max_out Maximum output frames
+ * @return Number of frames written to out_frames
+ */
+int extended_extract_interpreter_frames(int pid,
+                                        const u8 *frame_types,
+                                        const u64 *addrs,
+                                        const u64 *extra_data_a,
+                                        const u64 *extra_data_b,
+                                        int frame_count,
+                                        void *tracer,
+                                        bool new_cache,
+                                        void *info_p,
+                                        interp_symbol_info_t *out_frames,
+                                        int max_out);
+
+/**
+ * @brief **extended_free_interp_frames()** Free memory owned by interp_symbol_info_t array
+ * @param frames Array of interp_symbol_info_t
+ * @param count Number of entries
  */
-int extended_merge_stacks(char *dst, int len, const char *i_trace, const char *u_trace, int pid);
+void extended_free_interp_frames(interp_symbol_info_t *frames, int count);
 
 /**
- * @brief **extended_format_lua_stack()** Format Lua interpreter stack frames
+ * @brief **extended_extract_structured_frames()** High-level extraction for a stack trace
+ *
+ * Reads user and interpreter BPF stack maps, extracts structured interpreter
+ * frame symbols via per-symbol cache + extract functions.
+ *
  * @param tracer BPF tracer handle
- * @param pid Process ID
- * @param stack_id Interpreter stack ID from BPF map
- * @param stack_map_name Name of the stack map
- * @param h Stack string hash table
- * @param new_cache Whether to create new cache entry
+ * @param tgid Process ID (for cache lookup and process type detection)
+ * @param user_stack_id User stack ID from BPF map (-1 if none)
+ * @param interp_stack_id Interpreter stack ID from BPF map (-1 if none)
+ * @param custom_stack_map_name Name of the custom stack map
+ * @param new_cache Whether cache entry is new
  * @param info_p Process info pointer
- * @return Formatted stack string (caller must free) or NULL
+ * @param out_frames Caller-allocated output array
+ * @param max_out Maximum output frame count
+ * @return Number of frames written to out_frames
  */
-char *extended_format_lua_stack(void *tracer, int pid, int stack_id,
-                                const char *stack_map_name, void *h,
-                                bool new_cache, void *info_p);
+int extended_extract_structured_frames(void *tracer, int tgid,
+                                       int user_stack_id, int interp_stack_id,
+                                       const char *custom_stack_map_name,
+                                       bool new_cache, void *info_p,
+                                       interp_symbol_info_t *out_frames,
+                                       int max_out);
 
 #endif /* DF_EXTENDED_H */

agent/src/ebpf/user/profile/perf_profiler.c

@@ -257,14 +257,8 @@ static void oncpu_reader_work(void *arg)
 exit:
 	print_cp_tracer_status();
 
-	print_hash_stack_str(&oncpu_ctx.stack_str_hash);
-	/* free stack_str_hash */
-	if (likely(oncpu_ctx.stack_str_hash.buckets != NULL)) {
-		release_stack_str_hash(&oncpu_ctx.stack_str_hash);
-	}
-
 	print_hash_stack_trace_msg(&oncpu_ctx.msg_hash);
-	/* free stack_str_hash */
+	/* free msg_hash */
 	if (likely(oncpu_ctx.msg_hash.buckets != NULL)) {
 		/* Ensure that all elements are released properly/cleanly */
 		push_and_release_stack_trace_msg(&oncpu_ctx,

agent/src/ebpf/user/profile/perf_profiler.h

@@ -186,6 +186,9 @@ typedef struct {
 	u8 comm[TASK_COMM_LEN];
 	u8 process_name[TASK_COMM_LEN];
 	u8 container_id[CONTAINER_ID_SIZE];
+	u32 interp_frame_count;	// number of structured interpreter frames
+	u64 interp_frames_ptr;	// pointer to interp_symbol_info_t array
+	u8 raw_interpreter_data;	// 1 = has structured interpreter data
 	u32 data_len;
 	u64 data_ptr;
 	u8 data[0];