added support for R_X86_64_64/R_X86_64_RELATIVE relocations + added init/fini

3 files changed, 110 insertions, 42 deletions

diff --git a/04_dynld_nostd/dynld.c b/04_dynld_nostd/dynld.c
index 56d48b7..a7b0fe1 100644
--- a/04_dynld_nostd/dynld.c
+++ b/04_dynld_nostd/dynld.c
@@ -205,6 +205,41 @@ const Elf64Rela* get_reloca(const Dso* dso, const uint64_t idx) {
 }
 
 
+/// -----------
+/// Init & Fini
+/// -----------
+
+typedef void (*initfptr)();
+
+static void init(const Dso* dso) {
+    if (dso->dynamic[DT_INIT]) {
+        initfptr* fn = (initfptr*)(dso->base + dso->dynamic[DT_INIT]);
+        (*fn)();
+    }
+
+    size_t nfns = dso->dynamic[DT_INIT_ARRAYSZ] / sizeof(initfptr);
+    initfptr* fns = (initfptr*)(dso->base + dso->dynamic[DT_INIT_ARRAY]);
+    while (nfns--) {
+        (*fns++)();
+    }
+}
+
+typedef void (*finifptr)();
+
+static void fini(const Dso* dso) {
+    size_t nfns = dso->dynamic[DT_FINI_ARRAYSZ] / sizeof(finifptr);
+    finifptr* fns = (finifptr*)(dso->base + dso->dynamic[DT_FINI_ARRAY]) + nfns /* reverse destruction order */;
+    while (nfns--) {
+        (*--fns)();
+    }
+
+    if (dso->dynamic[DT_FINI]) {
+        finifptr* fn = (finifptr*)(dso->base + dso->dynamic[DT_FINI]);
+        (*fn)();
+    }
+}
+
+
 /// -------------
 /// Symbol lookup
 /// -------------
@@ -226,12 +261,12 @@ int strcmp(const char* s1, const char* s2) {
 //
 // `dso`          A handle to the dso which dynamic symbol table should be searched.
 // `sym_name`     Name of the symbol to look up.
-// `sym_type`     Type of the symbol to look up (STT_OBJECT | STT_FUNC).
-void* lookup_sym(const Dso* dso, const char* sym_name, unsigned sym_type) {
+void* lookup_sym(const Dso* dso, const char* sym_name) {
     for (unsigned i = 0; i < get_num_dynsyms(dso); ++i) {
         const Elf64Sym* sym = get_sym(dso, i);
 
-        if (ELF64_ST_TYPE(sym->info) == sym_type && ELF64_ST_BIND(sym->info) == STB_GLOBAL && sym->shndx != SHN_UNDEF) {
+        if ((ELF64_ST_TYPE(sym->info) == STT_OBJECT || ELF64_ST_TYPE(sym->info) == STT_FUNC) && ELF64_ST_BIND(sym->info) == STB_GLOBAL &&
+            sym->shndx != SHN_UNDEF) {
             if (strcmp(sym_name, get_str(dso, sym->name)) == 0) {
                 return dso->base + sym->value;
             }
@@ -367,7 +402,7 @@ typedef struct LinkMap LinkMap;
 // Resolve the relocation `reloc` by looking up the address of the symbol
 // referenced by the relocation. If the address of the symbol was found the
 // relocation is patched, if the address was not found the process exits.
-static void resolve_reloc(const Dso* dso, const LinkMap* map, const Elf64Rela* reloc, unsigned symtype) {
+static void resolve_reloc(const Dso* dso, const LinkMap* map, const Elf64Rela* reloc) {
     // Get symbol referenced by relocation.
     const int symidx = ELF64_R_SYM(reloc->info);
     const Elf64Sym* sym = get_sym(dso, symidx);
@@ -376,20 +411,28 @@ static void resolve_reloc(const Dso* dso, const LinkMap* map, const Elf64Rela* r
     // Get relocation typy.
     unsigned reloctype = ELF64_R_TYPE(reloc->info);
 
-    // Lookup symbol address.
+    // Find symbol address.
     void* symaddr = 0;
-    // TODO: Explain special handling of R_X86_64_COPY.
-    for (const LinkMap* lmap = (reloctype == R_X86_64_COPY ? map->next : map); lmap && symaddr == 0; lmap = lmap->next) {
-        symaddr = lookup_sym(lmap->dso, symname, symtype);
+    // FIXME: Should relocations of type `R_X86_64_64` only be looked up in `dso` directly?
+    if (reloctype == R_X86_64_RELATIVE) {
+        // Symbols address is computed by re-basing the relative address based on the DSOs base address.
+        symaddr = (void*)(dso->base + reloc->addend);
+    } else {
+        // TODO: Explain special handling of R_X86_64_COPY.
+        for (const LinkMap* lmap = (reloctype == R_X86_64_COPY ? map->next : map); lmap && symaddr == 0; lmap = lmap->next) {
+            symaddr = lookup_sym(lmap->dso, symname);
+        }
     }
     ERROR_ON(symaddr == 0, "Failed lookup symbol %s while resolving relocations!", symname);
 
-    pfmt("Resolved reloc %s to %p (base %p)\n", symname, symaddr, dso->base);
+    pfmt("Resolved reloc %s to %p (base %p)\n", reloctype == R_X86_64_RELATIVE ? "<relative>" : symname, symaddr, dso->base);
 
     // Perform relocation according to relocation type.
     switch (reloctype) {
         case R_X86_64_GLOB_DAT:  /* GOT entry for data objects. */
         case R_X86_64_JUMP_SLOT: /* PLT entry. */
+        case R_X86_64_64:        /* 64bit relocation (non-lazy). */
+        case R_X86_64_RELATIVE:  /* DSO base relative relocation. */
             // Patch storage unit of relocation with absolute address of the symbol.
             *(uint64_t*)(dso->base + reloc->offset) = (uint64_t)symaddr;
             break;
@@ -412,14 +455,14 @@ static void resolve_relocs(const Dso* dso, const LinkMap* map) {
     // variables).
     for (unsigned long relocidx = 0; relocidx < (dso->dynamic[DT_RELASZ] / sizeof(Elf64Rela)); ++relocidx) {
         const Elf64Rela* reloc = get_reloca(dso, relocidx);
-        resolve_reloc(dso, map, reloc, STT_OBJECT);
+        resolve_reloc(dso, map, reloc);
     }
 
     // Resolve all relocation from the PLT jump table found in `dso`. There is
     // typically one relocation per undefined dynamic function symbol.
     for (unsigned long relocidx = 0; relocidx < (dso->dynamic[DT_PLTRELSZ] / sizeof(Elf64Rela)); ++relocidx) {
         const Elf64Rela* reloc = get_pltreloca(dso, relocidx);
-        resolve_reloc(dso, map, reloc, STT_FUNC);
+        resolve_reloc(dso, map, reloc);
     }
 }
 
@@ -486,11 +529,15 @@ void dl_entry(const uint64_t* prctx) {
 
     // Resolve relocations of the library (dependency).
     resolve_relocs(&dso_lib, &map_prog);
-
     // Resolve relocations of the main program.
     resolve_relocs(&dso_prog, &map_prog);
 
-    // Install dynamic resolve handler.
+    // Initialize library.
+    init(&dso_lib);
+    // Initialize main program.
+    init(&dso_prog);
+
+    // Install dynamic resolve handler (lazy resolve).
     //
     // The dynamic resolve handler is used when binding symbols lazily. Hence
     // it should not be called in this example as we resolve all relocations
@@ -517,5 +564,10 @@ void dl_entry(const uint64_t* prctx) {
     // Transfer control to user program.
     dso_prog.entry();
 
+    // Finalize main program.
+    fini(&dso_prog);
+    // Finalize library.
+    fini(&dso_lib);
+
     _exit(0);
 }
diff --git a/04_dynld_nostd/libgreet.c b/04_dynld_nostd/libgreet.c
index 439e236..e697690 100644
--- a/04_dynld_nostd/libgreet.c
+++ b/04_dynld_nostd/libgreet.c
@@ -1,5 +1,7 @@
 // Copyright (c) 2020 Johannes Stoelp
 
+#include <io.h>
+
 int gCalled = 0;
 
 const char* get_greet() {
@@ -11,3 +13,11 @@ const char* get_greet2() {
     ++gCalled;
     return "Hello 2 from libgreet.so!";
 }
+
+__attribute__((constructor)) static void libinit() { /* static -> generates R_X86_64_RELATIVE relocation */
+    pfmt("libgreet.so: libinit\n");
+}
+
+__attribute__((destructor)) void libfini() { /* non static -> generates R_X86_64_64 relocation */
+    pfmt("libgreet.so: libfini\n");
+}
diff --git a/lib/include/elf.h b/lib/include/elf.h
index f0a0940..7a2e597 100644
--- a/lib/include/elf.h
+++ b/lib/include/elf.h
@@ -84,32 +84,36 @@ typedef struct {
 /// Dynamic Section
 /// ---------------
 
-#define DT_NULL     0  /* [ignored] Marks end of dynamic section */
-#define DT_NEEDED   1  /* [val] Name of needed library */
-#define DT_PLTRELSZ 2  /* [val] Size in bytes of PLT relocs */
-#define DT_PLTGOT   3  /* [ptr] Processor defined value */
-#define DT_HASH     4  /* [ptr] Address of symbol hash table */
-#define DT_STRTAB   5  /* [ptr] Address of string table */
-#define DT_SYMTAB   6  /* [ptr] Address of symbol table */
-#define DT_RELA     7  /* [ptr] Address of Rela relocs */
-#define DT_RELASZ   8  /* [val] Total size of Rela relocs */
-#define DT_RELAENT  9  /* [val] Size of one Rela reloc */
-#define DT_STRSZ    10 /* [val] Size of string table */
-#define DT_SYMENT   11 /* [val] Size of one symbol table entry */
-#define DT_INIT     12 /* [ptr] Address of init function */
-#define DT_FINI     13 /* [ptr] Address of termination function */
-#define DT_SONAME   14 /* [val] Name of shared object */
-#define DT_RPATH    15 /* [val] Library search path (deprecated) */
-#define DT_SYMBOLIC 16 /* [ignored] Start symbol search here */
-#define DT_REL      17 /* [ptr] Address of Rel relocs */
-#define DT_RELSZ    18 /* [val] Total size of Rel relocs */
-#define DT_RELENT   19 /* [val] Size of one Rel reloc */
-#define DT_PLTREL   20 /* [val] Type of reloc in PLT */
-#define DT_DEBUG    21 /* [ptr] For debugging; unspecified */
-#define DT_TEXTREL  22 /* [ignored] Reloc might modify .text */
-#define DT_JMPREL   23 /* [ptr] Address of PLT relocs */
-#define DT_BIND_NOW 24 /* [ignored] Process relocations of object */
-#define DT_MAX_CNT  25
+#define DT_NULL         0  /* [ignored] Marks end of dynamic section */
+#define DT_NEEDED       1  /* [val] Name of needed library */
+#define DT_PLTRELSZ     2  /* [val] Size in bytes of PLT relocs */
+#define DT_PLTGOT       3  /* [ptr] Processor defined value */
+#define DT_HASH         4  /* [ptr] Address of symbol hash table */
+#define DT_STRTAB       5  /* [ptr] Address of string table */
+#define DT_SYMTAB       6  /* [ptr] Address of symbol table */
+#define DT_RELA         7  /* [ptr] Address of Rela relocs */
+#define DT_RELASZ       8  /* [val] Total size of Rela relocs */
+#define DT_RELAENT      9  /* [val] Size of one Rela reloc */
+#define DT_STRSZ        10 /* [val] Size of string table */
+#define DT_SYMENT       11 /* [val] Size of one symbol table entry */
+#define DT_INIT         12 /* [ptr] Address of init function */
+#define DT_FINI         13 /* [ptr] Address of termination function */
+#define DT_SONAME       14 /* [val] Name of shared object */
+#define DT_RPATH        15 /* [val] Library search path (deprecated) */
+#define DT_SYMBOLIC     16 /* [ignored] Start symbol search here */
+#define DT_REL          17 /* [ptr] Address of Rel relocs */
+#define DT_RELSZ        18 /* [val] Total size of Rel relocs */
+#define DT_RELENT       19 /* [val] Size of one Rel reloc */
+#define DT_PLTREL       20 /* [val] Type of reloc in PLT */
+#define DT_DEBUG        21 /* [ptr] For debugging; unspecified */
+#define DT_TEXTREL      22 /* [ignored] Reloc might modify .text */
+#define DT_JMPREL       23 /* [ptr] Address of PLT relocs */
+#define DT_BIND_NOW     24 /* [ignored] Process relocations of object */
+#define DT_INIT_ARRAY   25 /* [ptr] Address of array of initialization functions */
+#define DT_FINI_ARRAY   26 /* [ptr] Address of array of termination functions */
+#define DT_INIT_ARRAYSZ 27 /* [val] Size in bytes of the initialization array */
+#define DT_FINI_ARRAYSZ 28 /* [val] Size in bytes of the termination array */
+#define DT_MAX_CNT      29
 
 typedef struct {
     uint64_t tag;
@@ -167,6 +171,8 @@ typedef struct {
 #define ELF64_R_TYPE(i) ((i)&0xffffffffL)
 
 // x86_64 relocation types.
-#define R_X86_64_COPY      5 /* Copy content from sym addr to relocation address */
-#define R_X86_64_GLOB_DAT  6 /* Address affected by relocation: `offset` (+ base) */
-#define R_X86_64_JUMP_SLOT 7 /* Address affected by relocation: `offset` (+ base) */
+#define R_X86_64_64        1 /* Absolute 64bit address, address affected by relocation: `base + offset` */
+#define R_X86_64_COPY      5 /* Copy content from sym addr to relocation address: `base + offset` */
+#define R_X86_64_GLOB_DAT  6 /* Address affected by relocation: `base + offset` */
+#define R_X86_64_JUMP_SLOT 7 /* Address affected by relocation: `base + offset` */
+#define R_X86_64_RELATIVE  8 /* Relative address *`base + offset` = `base + addend` */