improve coff deplibs

2 files changed, 143 insertions, 93 deletions

diff --git a/core/src/xmake/binutils/coff/deplibs.c b/core/src/xmake/binutils/coff/deplibs.c
index 44809a535..969702e98 100644
--- a/core/src/xmake/binutils/coff/deplibs.c
+++ b/core/src/xmake/binutils/coff/deplibs.c
@@ -49,72 +49,78 @@ tb_bool_t xm_binutils_coff_deplibs(tb_stream_ref_t istream, tb_hize_t base_offse
         return tb_false;
     }
 
-    // read section headers
-    tb_hize_t section_offset = base_offset + sizeof(xm_coff_header_t) + header.opthdr;
-    if (!tb_stream_seek(istream, section_offset)) {
-        return tb_false;
-    }
-
-    // try to get import directory rva from optional header
+    // read optional header and import rva
     tb_uint32_t import_rva = 0;
     if (header.opthdr > 0) {
-        // save pos
-        tb_hize_t saved_pos = tb_stream_offset(istream);
-
-        // seek to optional header
-        if (tb_stream_seek(istream, base_offset + sizeof(xm_coff_header_t))) {
-            tb_uint16_t magic = 0;
-            if (tb_stream_bread(istream, (tb_byte_t*)&magic, 2)) {
-                // magic is little endian
-                magic = tb_bits_le_to_ne_u16(magic);
-
-                tb_uint32_t data_dir_offset = 0;
+        // save current offset
+        tb_hize_t opt_offset = tb_stream_offset(istream);
+        
+        // read magic
+        tb_uint16_t magic = 0;
+        if (tb_stream_bread(istream, (tb_byte_t*)&magic, sizeof(magic))) {
+            // seek back
+            if (tb_stream_seek(istream, opt_offset)) {
                 if (magic == XM_PE32_MAGIC) {
-                    data_dir_offset = 96;
+                    xm_pe32_opt_header_t opt_header = {0};
+                    if (tb_stream_bread(istream, (tb_byte_t*)&opt_header, tb_min(sizeof(opt_header), header.opthdr))) {
+                        if (opt_header.number_of_rva_and_sizes > 1) {
+                            import_rva = opt_header.data_directory[1].vaddr;
+                        }
+                    }
                 } else if (magic == XM_PE32P_MAGIC) {
-                    data_dir_offset = 112;
-                }
-
-                /* check if optional header is large enough to contain import directory entry (index 1)
-                 * export(0) + import(1) -> 2 entries -> 16 bytes
-                 */
-                if (data_dir_offset != 0 && header.opthdr >= data_dir_offset + 16) {
-                    /* seek to Import Directory (Index 1)
-                     * Data Directory Array starts at optional_header_start + data_dir_offset
-                     * Index 1 is at + 8 bytes (sizeof(IMAGE_DATA_DIRECTORY) * 1)
-                     */
-                    if (tb_stream_seek(istream, base_offset + sizeof(xm_coff_header_t) + data_dir_offset + 8)) {
-                        if (tb_stream_bread(istream, (tb_byte_t*)&import_rva, 4)) {
-                            import_rva = tb_bits_le_to_ne_u32(import_rva);
+                    xm_pe32p_opt_header_t opt_header = {0};
+                    if (tb_stream_bread(istream, (tb_byte_t*)&opt_header, tb_min(sizeof(opt_header), header.opthdr))) {
+                        if (opt_header.number_of_rva_and_sizes > 1) {
+                            import_rva = opt_header.data_directory[1].vaddr;
                         }
                     }
                 }
             }
         }
-        // restore pos
-        tb_stream_seek(istream, saved_pos);
     }
 
+    // read section headers
+    xm_coff_section_t* sections = tb_null;
+    if (header.nsects > 0) {
+        sections = tb_nalloc_type(header.nsects, xm_coff_section_t);
+        if (sections) {
+             tb_hize_t section_offset = base_offset + sizeof(xm_coff_header_t) + header.opthdr;
+             if (tb_stream_seek(istream, section_offset)) {
+                 if (!tb_stream_bread(istream, (tb_byte_t*)sections, header.nsects * sizeof(xm_coff_section_t))) {
+                     tb_free(sections);
+                     sections = tb_null;
+                 }
+             } else {
+                 tb_free(sections);
+                 sections = tb_null;
+             }
+        }
+    }
+    
+    if (!sections) {
+        if (header.nsects > 0) return tb_false;
+        lua_newtable(lua);
+        return tb_true;
+    }
+
+    lua_newtable(lua);
     tb_size_t result_count = 0;
     for (tb_uint16_t i = 0; i < header.nsects; i++) {
-        xm_coff_section_t section;
-        if (!tb_stream_bread(istream, (tb_byte_t*)&section, sizeof(section))) {
-            return tb_false;
-        }
+        xm_coff_section_t* section = &sections[i];
 
         // check if it is .idata section (import directory table)
         tb_bool_t found_idt = tb_false;
         tb_uint32_t idt_offset = 0;
         if (import_rva != 0) {
             // check if import rva is in this section
-            if (import_rva >= section.vaddr && import_rva < section.vaddr + section.vsize) {
-                idt_offset = section.ofs + (import_rva - section.vaddr);
+            if (import_rva >= section->vaddr && import_rva < section->vaddr + section->vsize) {
+                idt_offset = section->ofs + (import_rva - section->vaddr);
                 found_idt = tb_true;
             }
         } else {
             // fallback to check section name
-            if (tb_strncmp(section.name, ".idata", 6) == 0) {
-                idt_offset = section.ofs;
+            if (tb_strncmp(section->name, ".idata", 6) == 0) {
+                idt_offset = section->ofs;
                 found_idt = tb_true;
             }
         }
@@ -137,67 +143,37 @@ tb_bool_t xm_binutils_coff_deplibs(tb_stream_ref_t istream, tb_hize_t base_offse
              */
 
             if (!tb_stream_seek(istream, idt_offset)) {
-                return tb_false;
+                break;
             }
 
             while (1) {
-                tb_uint32_t original_first_thunk;
-                tb_uint32_t time_date_stamp;
-                tb_uint32_t forwarder_chain;
-                tb_uint32_t name_rva;
-                tb_uint32_t first_thunk;
-
-                if (!tb_stream_bread(istream, (tb_byte_t*)&original_first_thunk, 4)) { break; }
-                if (!tb_stream_bread(istream, (tb_byte_t*)&time_date_stamp, 4)) { break; }
-                if (!tb_stream_bread(istream, (tb_byte_t*)&forwarder_chain, 4)) { break; }
-                if (!tb_stream_bread(istream, (tb_byte_t*)&name_rva, 4)) { break; }
-                if (!tb_stream_bread(istream, (tb_byte_t*)&first_thunk, 4)) { break; }
+                xm_coff_import_directory_table_t entry;
+                if (!tb_stream_bread(istream, (tb_byte_t*)&entry, sizeof(entry))) {
+                    break;
+                }
 
                 // check for null entry (end of table)
-                if (original_first_thunk == 0 && name_rva == 0) {
+                if (entry.original_first_thunk == 0 && entry.name_rva == 0) {
                     break;
                 }
 
-                name_rva = tb_bits_le_to_ne_u32(name_rva);
+                tb_uint32_t name_rva = tb_bits_le_to_ne_u32(entry.name_rva);
 
                 if (name_rva != 0) {
                     /* map RVA to file offset to read the name
                      * We need to find the section that contains this RVA.
-                     * Since we are iterating sections, we might need to seek back to read section headers again or cache them.
-                     * For simplicity, we assume the name string is within the same .idata section or we can find it by scanning sections.
-                     */
-
-                    /* To do this correctly, we should scan all sections to find which one contains the RVA.
-                     * But here we are inside a loop iterating sections.
-                     * We can save current position and iterate sections from the beginning (or cached) to find the RVA.
                      */
 
-                    // Optimization: usually the name is in the same .idata section or a nearby .rdata section.
-
                     tb_hize_t saved_pos_inner = tb_stream_offset(istream);
-
-                    // Find the section containing name_rva
                     tb_uint32_t name_file_offset = 0;
 
-                    // check current section first
-                    if (name_rva >= section.vaddr && name_rva < section.vaddr + section.vsize) {
-                        name_file_offset = section.ofs + (name_rva - section.vaddr);
-                    } else {
-                        // Scan all sections to find the RVA
-                        tb_hize_t saved_pos_sections = tb_stream_offset(istream);
-                        if (tb_stream_seek(istream, section_offset)) {
-                            for (tb_uint16_t k = 0; k < header.nsects; k++) {
-                                xm_coff_section_t s;
-                                if (!tb_stream_bread(istream, (tb_byte_t*)&s, sizeof(s))) {
-                                    break;
-                                }
-                                if (name_rva >= s.vaddr && name_rva < s.vaddr + s.vsize) {
-                                    name_file_offset = s.ofs + (name_rva - s.vaddr);
-                                    break;
-                                }
-                            }
+                    // Find the section containing name_rva
+                    for (tb_uint16_t k = 0; k < header.nsects; k++) {
+                        xm_coff_section_t* s = &sections[k];
+                        if (name_rva >= s->vaddr && name_rva < s->vaddr + s->vsize) {
+                            name_file_offset = s->ofs + (name_rva - s->vaddr);
+                            break;
                         }
-                        tb_stream_seek(istream, saved_pos_sections); // restore to current descriptor
                     }
 
                     if (name_file_offset != 0) {
@@ -213,16 +189,10 @@ tb_bool_t xm_binutils_coff_deplibs(tb_stream_ref_t istream, tb_hize_t base_offse
                     tb_stream_seek(istream, saved_pos_inner);
                 }
             }
-            /* We found .idata and processed it. Usually there is only one import table.
-             * But we should continue just in case or break?
-             * Typically break is enough after processing the import table.
-             */
             break;
         }
     }
 
-    if (result_count == 0) {
-        lua_newtable(lua);
-    }
+    if (sections) tb_free(sections);
     return tb_true;
 }
diff --git a/core/src/xmake/binutils/coff/prefix.h b/core/src/xmake/binutils/coff/prefix.h
index f991ac26d..8017beeff 100644
--- a/core/src/xmake/binutils/coff/prefix.h
+++ b/core/src/xmake/binutils/coff/prefix.h
@@ -59,6 +59,86 @@ typedef struct __xm_coff_header_t {
     tb_uint16_t flags;
 } __tb_packed__ xm_coff_header_t;
 
+typedef struct __xm_pe32_data_directory_t {
+    tb_uint32_t vaddr;
+    tb_uint32_t size;
+} __tb_packed__ xm_pe32_data_directory_t;
+
+typedef struct __xm_pe32_opt_header_t {
+    tb_uint16_t magic;
+    tb_uint8_t  major_linker_version;
+    tb_uint8_t  minor_linker_version;
+    tb_uint32_t size_of_code;
+    tb_uint32_t size_of_initialized_data;
+    tb_uint32_t size_of_uninitialized_data;
+    tb_uint32_t address_of_entry_point;
+    tb_uint32_t base_of_code;
+    tb_uint32_t base_of_data;
+    tb_uint32_t image_base;
+    tb_uint32_t section_alignment;
+    tb_uint32_t file_alignment;
+    tb_uint16_t major_operating_system_version;
+    tb_uint16_t minor_operating_system_version;
+    tb_uint16_t major_image_version;
+    tb_uint16_t minor_image_version;
+    tb_uint16_t major_subsystem_version;
+    tb_uint16_t minor_subsystem_version;
+    tb_uint32_t win32_version_value;
+    tb_uint32_t size_of_image;
+    tb_uint32_t size_of_headers;
+    tb_uint32_t checksum;
+    tb_uint16_t subsystem;
+    tb_uint16_t dll_characteristics;
+    tb_uint32_t size_of_stack_reserve;
+    tb_uint32_t size_of_stack_commit;
+    tb_uint32_t size_of_heap_reserve;
+    tb_uint32_t size_of_heap_commit;
+    tb_uint32_t loader_flags;
+    tb_uint32_t number_of_rva_and_sizes;
+    xm_pe32_data_directory_t data_directory[16];
+} __tb_packed__ xm_pe32_opt_header_t;
+
+typedef struct __xm_pe32p_opt_header_t {
+    tb_uint16_t magic;
+    tb_uint8_t  major_linker_version;
+    tb_uint8_t  minor_linker_version;
+    tb_uint32_t size_of_code;
+    tb_uint32_t size_of_initialized_data;
+    tb_uint32_t size_of_uninitialized_data;
+    tb_uint32_t address_of_entry_point;
+    tb_uint32_t base_of_code;
+    tb_uint64_t image_base;
+    tb_uint32_t section_alignment;
+    tb_uint32_t file_alignment;
+    tb_uint16_t major_operating_system_version;
+    tb_uint16_t minor_operating_system_version;
+    tb_uint16_t major_image_version;
+    tb_uint16_t minor_image_version;
+    tb_uint16_t major_subsystem_version;
+    tb_uint16_t minor_subsystem_version;
+    tb_uint32_t win32_version_value;
+    tb_uint32_t size_of_image;
+    tb_uint32_t size_of_headers;
+    tb_uint32_t checksum;
+    tb_uint16_t subsystem;
+    tb_uint16_t dll_characteristics;
+    tb_uint64_t size_of_stack_reserve;
+    tb_uint64_t size_of_stack_commit;
+    tb_uint64_t size_of_heap_reserve;
+    tb_uint64_t size_of_heap_commit;
+    tb_uint32_t loader_flags;
+    tb_uint32_t number_of_rva_and_sizes;
+    xm_pe32_data_directory_t data_directory[16];
+} __tb_packed__ xm_pe32p_opt_header_t;
+
+typedef struct __xm_coff_import_directory_table_t {
+    tb_uint32_t original_first_thunk; // RVA to original unbound IAT (PIMAGE_THUNK_DATA)
+    tb_uint32_t time_date_stamp;      // 0 if not bound, -1 if bound, and real date\time stamp in IMAGE_DIRECTORY_ENTRY_BOUND_IMPORT (new BIND) O.W. date/time stamp of DLL bound to (Old BIND)
+    tb_uint32_t forwarder_chain;      // -1 if no forwarders
+    tb_uint32_t name_rva;             // RVA to name
+    tb_uint32_t first_thunk;          // RVA to IAT (if bound this IAT has actual addresses)
+} __tb_packed__ xm_coff_import_directory_table_t;
+
 typedef struct __xm_coff_import_header_t {
     tb_uint16_t sig1;     // 0
     tb_uint16_t sig2;     // 0xffff