Fixed the computing of basic blocks and used them in graphic views.

git-svn-id: svn://svn.gna.org/svn/chrysalide/trunk@316 abbe820e-26c8-41b2-8c08-b7b2b41f8b0a

2 files changed, 206 insertions, 16 deletions

diff --git a/src/analysis/disass/disassembler.c b/src/analysis/disass/disassembler.c
index 6147f3b..e7c972d 100644
--- a/src/analysis/disass/disassembler.c
+++ b/src/analysis/disass/disassembler.c
@@ -257,7 +257,7 @@ static void g_delayed_disassembly_process(GDelayedDisassembly *disass, GtkExtSta
     run_plugins_on_binary(disass->binary, PGA_BINARY_BOUNDED, true);
 
     /* Quatrième étape */
-#if 0
+
     id = gtk_extended_status_bar_push(statusbar, _("Grouping routines instructions..."), true);
 
     qsort(routines, routines_count, sizeof(GBinRoutine *), (__compar_fn_t)g_binary_routine_rcompare);
@@ -267,7 +267,7 @@ static void g_delayed_disassembly_process(GDelayedDisassembly *disass, GtkExtSta
     gtk_extended_status_bar_remove(statusbar, id);
 
     run_plugins_on_binary(disass->binary, PGA_BINARY_GROUPED, true);
-#endif
+
     /* Cinquième étape */
 
     id = gtk_extended_status_bar_push(statusbar, _("Printing disassembled code..."), true);
diff --git a/src/analysis/disass/macro.c b/src/analysis/disass/macro.c
index 7fda6b9..46dc0d3 100644
--- a/src/analysis/disass/macro.c
+++ b/src/analysis/disass/macro.c
@@ -41,6 +41,16 @@ typedef struct _branch_info
 } branch_info;
 
 
+/**
+ * Macros pour le marquage des instructions traitées.
+ * Dans un soucis d'optimisation, on ne traite que les instructions
+ * démarrant un bloc.
+ */
+#define MACRO_MARK_AS_PROCESSED(_instr) g_object_set_data(G_OBJECT(_instr), "macro_done", _instr)
+#define MACRO_IS_PROCESSED(_instr) (g_object_get_data(G_OBJECT(_instr), "macro_done") != NULL)
+#define MACRO_CLEAR_PROCESSED(_instr) g_object_set_data(G_OBJECT(_instr), "macro_done", NULL)
+
+
 /* Indique si une adresse est retenue comme point de passage. */
 static bool is_addr_in_branch(const branch_info *, const vmpa_t *, bool);
 
@@ -50,6 +60,9 @@ static void find_next_jumps(GArchInstruction *, vmpa_t, vmpa_t, branch_info *);
 /* Retrouve le point de ralliement entre deux branches. */
 static vmpa_t compute_first_common_addr(branch_info *, branch_info *);
 
+/* Retrouve le point de ralliement entre un groupe de branches. */
+static vmpa_t compute_first_common_addr_in_group(const branch_info *, size_t);
+
 /* Procède à la définition de bloc regroupant des instructions. */
 static GInstrBlock *build_instruction_block(GArchInstruction *, vmpa_t, vmpa_t, vmpa_t);
 
@@ -144,6 +157,7 @@ static void find_next_jumps(GArchInstruction *instrs, vmpa_t start, vmpa_t end,
             {
                 case ILT_EXEC_FLOW:
                 case ILT_JUMP:
+                case ILT_CASE_JUMP:
                 case ILT_JUMP_IF_TRUE:
                 case ILT_JUMP_IF_FALSE:
                     g_arch_instruction_get_location(dests[i], NULL, NULL, &addr);
@@ -204,6 +218,49 @@ static vmpa_t compute_first_common_addr(branch_info *a, branch_info *b)
 
 /******************************************************************************
 *                                                                             *
+*  Paramètres  : list  = liste d'ensembles de jalons à parcourir.             *
+*                count = taille de cette liste.                               *
+*                                                                             *
+*  Description : Retrouve le point de ralliement entre un groupe de branches. *
+*                                                                             *
+*  Retour      : Adresse commune aux branches.                                *
+*                                                                             *
+*  Remarques   : -                                                            *
+*                                                                             *
+******************************************************************************/
+
+static vmpa_t compute_first_common_addr_in_group(const branch_info *list, size_t count)
+{
+    vmpa_t result;                          /* Adresse trouvée à retourner */
+    size_t i;                               /* Boucle de parcours #1       */
+    bool keep;                              /* Candidate à garder ?        */
+    size_t j;                               /* Boucle de parcours #2       */
+
+    /* Valeur conceptuellement impossible à renvoyer */
+    result = VMPA_MAX;
+
+    //qsort(a->jumps, a->count, sizeof(vmpa_t), (__compar_fn_t)compare_vmpa);
+    //qsort(b->jumps, b->count, sizeof(vmpa_t), (__compar_fn_t)compare_vmpa);
+
+    for (i = 0; i < list[0].count && result == VMPA_MAX; i++)
+    {
+        keep = true;
+
+        for (j = 1; j < count && keep; j++)
+            keep = is_addr_in_branch(&list[j], &list[0].jumps[i], false);
+
+        if (keep)
+            result = list[0].jumps[i];
+
+    }
+
+    return result;
+
+}
+
+
+/******************************************************************************
+*                                                                             *
 *  Paramètres  : instrs = ensemble des instructions d'assemblage.             *
 *                start  = adresse de début du bloc.                           *
 *                end    = adresse de fin du bloc (exclusive).                 *
@@ -220,6 +277,7 @@ static vmpa_t compute_first_common_addr(branch_info *a, branch_info *b)
 static GInstrBlock *build_instruction_block(GArchInstruction *instrs, vmpa_t start, vmpa_t end, vmpa_t stop)
 {
     GInstrBlock *result;                    /* Regroupement à retourner    */
+    branch_info main_branch;                /* Flot d'exécution complet    */
     GArchInstruction *first;                /* Première instruction        */
     GArchInstruction *last;                 /* Dernière instruction        */
     GArchInstruction *iter;                 /* Boucle de parcours          */
@@ -227,17 +285,26 @@ static GInstrBlock *build_instruction_block(GArchInstruction *instrs, vmpa_t sta
     GArchInstruction **dests;               /* Instr. visée par une autre  */
     InstructionLinkType *types;             /* Type de lien entre lignes   */
     size_t dcount;                          /* Nombre de liens de dest.    */
-    size_t i;                               /* Boucle de parcours          */
+    size_t i;                               /* Boucle de parcours #1       */
     GInstrBlock *block;                     /* Nouveau bloc mis en place   */
+    branch_info *cases_branches;            /* Branches d'un aiguillage    */
+    size_t cases_count;                     /* Nombre d'aiguillages        */
     branch_info true_branch;                /* Branche 'condition vraie'   */
     branch_info false_branch;               /* Branche 'condition fausse'  */
+    branch_info *excep_branches;            /* Branches d'exceptions       */
+    size_t excep_count;                     /* Nombre d'exceptions         */
     vmpa_t next_addr;                       /* Prochaine instruction visée */
+    size_t j;                               /* Boucle de parcours #2       */
+    vmpa_t stop_addr;                       /* Adresse de fin de bloc      */
 
     result = NULL;
 
     first = NULL;
     last = NULL;
 
+    memset(&main_branch, 0, sizeof(branch_info));
+    find_next_jumps(instrs, start, end, &main_branch);
+
     //printf("[+] blocking 0x%08llx -> 0x%08llx... stop @ 0x%08llx\n", start, end, stop);
 
     for (iter = g_arch_instruction_find_by_address(instrs, start, true);
@@ -247,15 +314,14 @@ static GInstrBlock *build_instruction_block(GArchInstruction *instrs, vmpa_t sta
         g_arch_instruction_get_location(iter, NULL, NULL, &addr);
         if (addr == stop) break;
 
+        /* On s'arrêter si l'instruction est déjà décompilée */
+        if (MACRO_IS_PROCESSED(iter)) break;
+
         if (first == NULL)
             first = iter;
 
         last = iter;
 
-        /* On s'arrêter si l'instruction est déjà décompilée */
-        if (g_object_get_data(G_OBJECT(iter), "decomp_done") != NULL) break;
-        g_object_set_data(G_OBJECT(iter), "decomp_done", iter);
-
         /* On n'approfondit que les chemins qui se séparent */
         if (!g_arch_instruction_has_destinations(iter))
         {
@@ -268,8 +334,12 @@ static GInstrBlock *build_instruction_block(GArchInstruction *instrs, vmpa_t sta
         dcount = g_arch_instruction_get_destinations(iter, &dests, &types);
 
         next_addr = 0;
+        cases_branches = NULL;
+        cases_count = 0;
         memset(&true_branch, 0, sizeof(branch_info));
         memset(&false_branch, 0, sizeof(branch_info));
+        excep_branches = NULL;
+        excep_count = 0;
 
         for (i = 0; i < dcount; i++)
             switch (types[i])
@@ -281,6 +351,7 @@ static GInstrBlock *build_instruction_block(GArchInstruction *instrs, vmpa_t sta
                         result = g_virtual_block_new();
 
                     block = g_flow_block_new(instrs, first, iter);
+                    MACRO_MARK_AS_PROCESSED(first);
                     g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
                     first = NULL;
 
@@ -288,6 +359,17 @@ static GInstrBlock *build_instruction_block(GArchInstruction *instrs, vmpa_t sta
 
                     break;
 
+                case ILT_CASE_JUMP:
+
+                    g_arch_instruction_get_location(dests[i], NULL, NULL, &addr);
+
+                    cases_branches = (branch_info *)realloc(cases_branches,
+                                                            ++cases_count * sizeof(branch_info));
+                    memset(&cases_branches[cases_count - 1], 0, sizeof(branch_info));
+                    find_next_jumps(instrs, addr, end, &cases_branches[cases_count - 1]);
+
+                    break;
+
                 case ILT_JUMP_IF_TRUE:
                     g_arch_instruction_get_location(dests[i], NULL, NULL, &addr);
                     find_next_jumps(instrs, addr, end, &true_branch);
@@ -298,12 +380,97 @@ static GInstrBlock *build_instruction_block(GArchInstruction *instrs, vmpa_t sta
                     find_next_jumps(instrs, addr, end, &false_branch);
                     break;
 
+                case ILT_CATCH_EXCEPTION:
+
+                    g_arch_instruction_get_location(dests[i], NULL, NULL, &addr);
+
+                    excep_branches = (branch_info *)realloc(excep_branches,
+                                                            ++excep_count * sizeof(branch_info));
+                    memset(&excep_branches[excep_count - 1], 0, sizeof(branch_info));
+                    find_next_jumps(instrs, addr, end, &excep_branches[excep_count - 1]);
+
+                    break;
+
                 default:
-                    next_addr = VMPA_MAX;
+                    if (next_addr == 0)
+                        next_addr = VMPA_MAX;
                     break;
 
             }
 
+        /* Post-traitements de ILT_CASE_JUMP */
+        if (cases_count > 0)
+        {
+            if (result == NULL)
+                result = g_virtual_block_new();
+
+            if (first != NULL)
+            {
+                block = g_flow_block_new(instrs, first, iter);
+                MACRO_MARK_AS_PROCESSED(first);
+                g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
+                first = NULL;
+            }
+
+            //printf(" --- cases --- start\n");
+
+            next_addr = compute_first_common_addr_in_group(cases_branches, cases_count);
+            //printf("    stop :: 0x%08llx\n", next_addr);
+
+            for (j = 0; j < cases_count; j++)
+            {
+                //printf(" ## %zu\n", j);
+
+                block = build_instruction_block(instrs, cases_branches[j].jumps[0], end, next_addr);
+
+                if (block != NULL)
+
+                g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
+
+                free(cases_branches[j].jumps);
+
+            }
+
+            //printf(" --- cases --- end\n");
+
+
+            free(cases_branches);
+
+        }
+
+        /* Post-traitements de ILT_CATCH_EXCEPTION */
+        if (excep_count > 0)
+        {
+            if (result == NULL)
+                result = g_virtual_block_new();
+
+            if (first != NULL)
+            {
+                block = g_flow_block_new(instrs, first, iter);
+                MACRO_MARK_AS_PROCESSED(first);
+                g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
+                first = NULL;
+            }
+
+            for (j = 0; j < excep_count; j++)
+            {
+                stop_addr = compute_first_common_addr(&main_branch, &excep_branches[j]);
+                //next_addr = MIN(next_addr, end);
+
+                block = build_instruction_block(instrs, excep_branches[j].jumps[0], end, stop_addr);
+
+                if (block != NULL)
+
+                g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
+
+                free(excep_branches[j].jumps);
+
+            }
+
+            free(excep_branches);
+
+        }
+
         if (next_addr == VMPA_MAX)
         {
             iter = g_arch_instruction_get_next_iter(instrs, iter, end);
@@ -318,14 +485,24 @@ static GInstrBlock *build_instruction_block(GArchInstruction *instrs, vmpa_t sta
             if (result == NULL)
                 result = g_virtual_block_new();
 
-            block = g_flow_block_new(instrs, first, iter);
-            g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
-            first = NULL;
+            if (first != NULL)
+            {
+                block = g_flow_block_new(instrs, first, iter);
+                MACRO_MARK_AS_PROCESSED(first);
+                g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
+                first = NULL;
+            }
 
             block = build_instruction_block(instrs, true_branch.jumps[0], end, next_addr);
+
+                if (block != NULL)
+
             g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
 
             block = build_instruction_block(instrs, false_branch.jumps[0], end, next_addr);
+
+                if (block != NULL)
+
             g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
 
             free(true_branch.jumps);
@@ -342,15 +519,27 @@ static GInstrBlock *build_instruction_block(GArchInstruction *instrs, vmpa_t sta
 
     if (first != NULL && last != NULL)
     {
-        block = g_flow_block_new(instrs, first, last);
+        if (!MACRO_IS_PROCESSED(first))
+        {
+            //printf("--close?--\n");
+            block = g_flow_block_new(instrs, first, last);
+            MACRO_MARK_AS_PROCESSED(first);
+            //printf("--close!--\n");
 
-        if (result == NULL)
-            result = block;
-        else
-            g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
+            if (result == NULL)
+                result = block;
+            else
+                g_virtual_block_add_child(G_VIRTUAL_BLOCK(result), block);
+        }
 
     }
 
+    if (result == NULL)
+    {
+        //printf("WARNING :: null !\n");
+        //exit(0);
+    }
+
     return result;
 
 }
@@ -389,6 +578,7 @@ void group_routines_instructions(GArchInstruction *list, GBinRoutine **routines,
 
 
         block = build_instruction_block(list, start, end, VMPA_MAX);
+        g_binary_routine_set_basic_blocks(routines[i], block);
 
         gtk_extended_status_bar_update_activity(statusbar, id, (i + 1) * 1.0 / count);