reading & organization pass over unwinding layer; deduplicate PE info with PE layer, move dwarf info to in-progress dwarf layer

2026-09-25 18:50:05 +00:00 · 2024-03-25 14:42:54 -07:00
parent 1c3792a3da
commit 55146b547b
4 changed files with 645 additions and 718 deletions
@@ -1260,6 +1260,167 @@ if (success__)                             \
 (x) = dwarf_leb128_decode(T,first__, (p)); \
 }while(0)

+
+////////////////////////////////
+//~ allen: ELF/DW Unwind Types
+//
+// TODO(rjf): OLD TYPES FROM UNWINDER CODE. bucketing this here, and deferring dwarf-based
+// unwinding info to future DWARF/linux work.
+//
+#if 0
+
+// * applies to (any X: unwind(ELF/DW, X))
+
+// EH: Exception Frames
+typedef U8 UNW_DW_EhPtrEnc;
+enum{
+  UNW_DW_EhPtrEnc_TYPE_MASK = 0x0F,
+  UNW_DW_EhPtrEnc_PTR     = 0x00, // Pointer sized unsigned value
+  UNW_DW_EhPtrEnc_ULEB128 = 0x01, // Unsigned LE base-128 value
+  UNW_DW_EhPtrEnc_UDATA2  = 0x02, // Unsigned 16-bit value
+  UNW_DW_EhPtrEnc_UDATA4  = 0x03, // Unsigned 32-bit value
+  UNW_DW_EhPtrEnc_UDATA8  = 0x04, // Unsigned 64-bit value
+  UNW_DW_EhPtrEnc_SIGNED  = 0x08, // Signed pointer
+  UNW_DW_EhPtrEnc_SLEB128 = 0x09, // Signed LE base-128 value
+  UNW_DW_EhPtrEnc_SDATA2  = 0x0A, // Signed 16-bit value
+  UNW_DW_EhPtrEnc_SDATA4  = 0x0B, // Signed 32-bit value
+  UNW_DW_EhPtrEnc_SDATA8  = 0x0C, // Signed 64-bit value
+};
+enum{
+  UNW_DW_EhPtrEnc_MODIF_MASK = 0x70,
+  UNW_DW_EhPtrEnc_PCREL   = 0x10, // Value is relative to the current program counter.
+  UNW_DW_EhPtrEnc_TEXTREL = 0x20, // Value is relative to the .text section.
+  UNW_DW_EhPtrEnc_DATAREL = 0x30, // Value is relative to the .got or .eh_frame_hdr section.
+  UNW_DW_EhPtrEnc_FUNCREL = 0x40, // Value is relative to the function.
+  UNW_DW_EhPtrEnc_ALIGNED = 0x50, // Value is aligned to an address unit sized boundary.
+};
+enum{
+  UNW_DW_EhPtrEnc_INDIRECT = 0x80, // This flag indicates that value is stored in virtual memory.
+  UNW_DW_EhPtrEnc_OMIT     = 0xFF,
+};
+
+typedef struct UNW_DW_EhPtrCtx{
+  U64 raw_base_vaddr; // address where pointer is being read
+  U64 text_vaddr;     // base address of section with instructions (used for encoding pointer on SH and IA64)
+  U64 data_vaddr;     // base address of data section (used for encoding pointer on x86-64)
+  U64 func_vaddr;     // base address of function where IP is located
+} UNW_DW_EhPtrCtx;
+
+// CIE: Common Information Entry
+typedef struct UNW_DW_CIEUnpacked{
+  U8 version;
+  UNW_DW_EhPtrEnc lsda_encoding;
+  UNW_DW_EhPtrEnc addr_encoding;
+  
+  B8 has_augmentation_size;
+  U64 augmentation_size;
+  String8 augmentation;
+  
+  U64 code_align_factor;
+  S64 data_align_factor;
+  U64 ret_addr_reg;
+  
+  U64 handler_ip;
+  
+  U64 cfi_range_min;
+  U64 cfi_range_max;
+} UNW_DW_CIEUnpacked;
+
+typedef struct UNW_DW_CIEUnpackedNode{
+  struct UNW_DW_CIEUnpackedNode *next;
+  UNW_DW_CIEUnpacked cie;
+  U64 offset;
+} UNW_DW_CIEUnpackedNode;
+
+// FDE: Frame Description Entry
+typedef struct UNW_DW_FDEUnpacked{
+  U64 ip_voff_min;
+  U64 ip_voff_max;
+  U64 lsda_ip;
+  
+  U64 cfi_range_min;
+  U64 cfi_range_max;
+} UNW_DW_FDEUnpacked;
+
+// CFI: Call Frame Information
+typedef struct UNW_DW_CFIRecords{
+  B32 valid;
+  UNW_DW_CIEUnpacked cie;
+  UNW_DW_FDEUnpacked fde;
+} UNW_DW_CFIRecords;
+
+typedef enum UNW_DW_CFICFARule{
+  UNW_DW_CFICFARule_REGOFF,
+  UNW_DW_CFICFARule_EXPR,
+} UNW_DW_CFICFARule;
+
+typedef struct UNW_DW_CFICFACell{
+  UNW_DW_CFICFARule rule;
+  union{
+    struct{
+      U64 reg_idx;
+      S64 offset;
+    };
+    U64 expr_min;
+    U64 expr_max;
+  };
+} UNW_DW_CFICFACell;
+
+typedef enum UNW_DW_CFIRegisterRule{
+  UNW_DW_CFIRegisterRule_SAME_VALUE,
+  UNW_DW_CFIRegisterRule_UNDEFINED,
+  UNW_DW_CFIRegisterRule_OFFSET,
+  UNW_DW_CFIRegisterRule_VAL_OFFSET,
+  UNW_DW_CFIRegisterRule_REGISTER,
+  UNW_DW_CFIRegisterRule_EXPRESSION,
+  UNW_DW_CFIRegisterRule_VAL_EXPRESSION,
+} UNW_DW_CFIRegisterRule;
+
+typedef struct UNW_DW_CFICell{
+  UNW_DW_CFIRegisterRule rule;
+  union{
+    S64 n;
+    struct{
+      U64 expr_min;
+      U64 expr_max;
+    };
+  };
+} UNW_DW_CFICell;
+
+typedef struct UNW_DW_CFIRow{
+  struct UNW_DW_CFIRow *next;
+  UNW_DW_CFICell *cells;
+  UNW_DW_CFICFACell cfa_cell;
+} UNW_DW_CFIRow;
+
+typedef struct UNW_DW_CFIMachine{
+  U64 cells_per_row;
+  UNW_DW_CIEUnpacked *cie;
+  UNW_DW_EhPtrCtx *ptr_ctx;
+  UNW_DW_CFIRow *initial_row;
+  U64 fde_ip;
+} UNW_DW_CFIMachine;
+
+typedef U8 UNW_DW_CFADecode;
+enum{
+  UNW_DW_CFADecode_NOP     = 0x0,
+  // 1,2,4,8 reserved for literal byte sizes
+  UNW_DW_CFADecode_ADDRESS = 0x9,
+  UNW_DW_CFADecode_ULEB128 = 0xA,
+  UNW_DW_CFADecode_SLEB128 = 0xB,
+};
+
+typedef U16 UNW_DW_CFAControlBits;
+enum{
+  UNW_DW_CFAControlBits_DEC1_MASK = 0x00F,
+  UNW_DW_CFAControlBits_DEC2_MASK = 0x0F0,
+  UNW_DW_CFAControlBits_IS_REG_0  = 0x100,
+  UNW_DW_CFAControlBits_IS_REG_1  = 0x200,
+  UNW_DW_CFAControlBits_IS_REG_2  = 0x400,
+  UNW_DW_CFAControlBits_NEW_ROW   = 0x800,
+};
+#endif
+
 ////////////////////////////////
 //~ Dwarf Parser Functions