Created an instruction database for Chrysalide.

4 files changed, 108 insertions, 15 deletions

diff --git a/src/analysis/binary.c b/src/analysis/binary.c
index 4bb9e43..f028cd6 100644
--- a/src/analysis/binary.c
+++ b/src/analysis/binary.c
@@ -40,6 +40,7 @@
 #include "db/client.h"
 //#include "decomp/decompiler.h"
 #include "disass/disassembler.h"
+#include "../arch/storage.h"
 #include "../common/extstr.h"
 #include "../common/cpp.h"
 #include "../common/xdg.h"
@@ -1025,6 +1026,53 @@ static bool g_loaded_binary_connect_remote(GLoadedBinary *binary)
 }
 
 
+/******************************************************************************
+*                                                                             *
+*  Paramètres  : binary = élément binaire à manipuler.                        *
+*                                                                             *
+*  Description : Sauvegarde le cache des instructions désassemblées.          *
+*                                                                             *
+*  Retour      : Bilan préliminaire de l'opération.                           *
+*                                                                             *
+*  Remarques   : -                                                            *
+*                                                                             *
+******************************************************************************/
+
+bool g_loaded_binary_save_cache(const GLoadedBinary *binary)
+{
+    bool result;                            /* Bilan à faire remonter      */
+    GArchProcessor *proc;                   /* Processeur concerné         */
+    GBinContent *content;                   /* Contenu brut représenté     */
+    const gchar *id;                        /* Identifiant court et unique */
+    GAsmStorage *storage;                   /* Cache propre à constituer   */
+
+    proc = g_loaded_binary_get_processor(binary);
+    content = g_loaded_binary_get_content(binary);
+
+    id = g_binary_content_get_checksum(content);
+
+    storage = g_asm_storage_new_compressed(proc, id);
+
+    g_object_unref(G_OBJECT(content));
+    g_object_unref(G_OBJECT(proc));
+
+    if (storage != NULL)
+    {
+        g_signal_connect(G_OBJECT(storage), "saved", G_CALLBACK(g_object_unref), NULL);
+
+        g_asm_storage_save(storage);
+
+        result = true;
+
+    }
+    else
+        result = false;
+
+    return result;
+
+}
+
+
 
 /* ---------------------------------------------------------------------------------- */
 /*                            MANIPULATION DES COLLECTIONS                            */
@@ -1491,9 +1539,14 @@ static bool g_loaded_binary_save(const GLoadedBinary *binary, xmlDoc *xdoc, xmlX
 {
     bool result;                            /* Bilan à faire remonter      */
 
+    /* Mise en cache des instructions */
+
+    result = g_loaded_binary_save_cache(binary);
+
     /* Elément divers associés au binaire */
 
-    result = g_loaded_binary_save_storage(binary, xdoc, context, path);
+    if (result)
+        result = g_loaded_binary_save_storage(binary, xdoc, context, path);
 
     /* Sauvegarde côté serveur */
 
diff --git a/src/analysis/binary.h b/src/analysis/binary.h
index b57e072..fd25210 100644
--- a/src/analysis/binary.h
+++ b/src/analysis/binary.h
@@ -111,6 +111,9 @@ DBStorage g_loaded_binary_get_storage(const GLoadedBinary *, DBFeatures);
 /* Définit la forme d'enregistrement d'une fonctionnalité. */
 void g_loaded_binary_set_storage(GLoadedBinary *, DBFeatures, DBStorage);
 
+/* Sauvegarde le cache des instructions désassemblées. */
+bool g_loaded_binary_save_cache(const GLoadedBinary *);
+
 
 
 /* -------------------------- MANIPULATION DES COLLECTIONS -------------------------- */
diff --git a/src/analysis/disass/disassembler.c b/src/analysis/disass/disassembler.c
index ca5e565..efe50e1 100644
--- a/src/analysis/disass/disassembler.c
+++ b/src/analysis/disass/disassembler.c
@@ -36,6 +36,7 @@
 #include "instructions.h"
 #include "output.h"
 #include "routines.h"
+#include "../../arch/storage.h"
 #include "../../core/global.h"
 #include "../../glibext/generators/prologue.h"
 #include "../../plugins/pglist.h"
@@ -216,6 +217,10 @@ static void compute_disassembly(GLoadedBinary *binary, GProcContext *context, wg
 {
     GArchProcessor *proc;                   /* Architecture du binaire     */
     GExeFormat *format;                     /* Format du binaire représenté*/
+    GBinContent *content;                   /* Contenu brut représenté     */
+    const gchar *id;                        /* Identifiant court et unique */
+    GAsmStorage *storage;                   /* Cache propre à constituer   */
+    bool cached;                            /* Instructions en cache       */
     GArchInstruction **instrs;              /* Instructions résultantes    */
     size_t count;                           /* Quantité de ces instructions*/
 
@@ -227,25 +232,54 @@ static void compute_disassembly(GLoadedBinary *binary, GProcContext *context, wg
 
     format = g_loaded_binary_get_format(binary);
 
+    g_binary_format_preload_disassembling_context(G_BIN_FORMAT(format), context, status);
+
+    /**
+     * Etape zéro : récupération des instructions depuis un cache, si ce dernier exitste.
+     */
+
+    content = g_loaded_content_get_content(G_LOADED_CONTENT(binary));
+
+    id = g_binary_content_get_checksum(content);
+
+    storage = g_asm_storage_new_compressed(proc, id);
+
+    g_object_unref(G_OBJECT(content));
+
+    cached = g_asm_storage_has_cache(storage);
+
+    if (cached)
+        cached = g_asm_storage_open(storage, G_BIN_FORMAT(format), gid);
+
+    g_object_unref(G_OBJECT(storage));
+
     /**
      * Première étape : collecte des instructions.
      */
 
-    instrs = disassemble_binary_content(binary, context, gid, status, &count);
+    if (!cached)
+    {
+        instrs = disassemble_binary_content(binary, context, gid, status, &count);
+
+        g_arch_processor_set_instructions(proc, instrs, count);
 
-    g_arch_processor_set_instructions(proc, instrs, count);
+        process_disassembly_event(PGA_DISASSEMBLY_RAW, binary);
 
-    process_disassembly_event(PGA_DISASSEMBLY_RAW, binary);
+    }
 
     /**
      * Seconde étape : liaisons des instructions.
      */
 
-    process_all_instructions(gid, status, _("Calling 'link' hook on all instructions..."),
-                             g_instructions_study_do_link_operation,
-                             proc, context, format);
+    if (!cached)
+    {
+        process_all_instructions(gid, status, _("Calling 'link' hook on all instructions..."),
+                                 g_instructions_study_do_link_operation,
+                                 proc, context, format);
 
-    process_disassembly_event(PGA_DISASSEMBLY_HOOKED_LINK, binary);
+        process_disassembly_event(PGA_DISASSEMBLY_HOOKED_LINK, binary);
+
+    }
 
     /**
      * Troisième étape : exécution d'éventuels post-traitements.
@@ -271,11 +305,15 @@ static void compute_disassembly(GLoadedBinary *binary, GProcContext *context, wg
      * Cinquième étape : liaisons entre instructions.
      */
 
-    process_all_instructions(gid, status, _("Establishing links betweek all instructions..."),
-                             g_instructions_study_establish_links,
-                             proc, context, format);
+    if (!cached)
+    {
+        process_all_instructions(gid, status, _("Establishing links betweek all instructions..."),
+                                 g_instructions_study_establish_links,
+                                 proc, context, format);
 
-    process_disassembly_event(PGA_DISASSEMBLY_LINKED, binary);
+        process_disassembly_event(PGA_DISASSEMBLY_LINKED, binary);
+
+    }
 
     /**
      * Sixième étape : regroupement en blocs basiques.
diff --git a/src/analysis/disass/fetch.c b/src/analysis/disass/fetch.c
index 934c755..03b3ff0 100644
--- a/src/analysis/disass/fetch.c
+++ b/src/analysis/disass/fetch.c
@@ -431,11 +431,10 @@ GArchInstruction **disassemble_binary_content(GLoadedBinary *binary, GProcContex
     g_proc_context_attach_counter(template.ctx, &remaining_counter);
 
     /**
-     * Première phase de désassemblage : intégration des infos du format.
+     * Première phase de désassemblage : intégration des infos du format,
+     * récupérées dans le contexte via un appel à g_binary_format_preload_disassembling_context().
      */
 
-    g_binary_format_preload_disassembling_context(format, template.ctx, status);
-
     populate_fresh_memory_areas(gid, status, template.areas, template.count, G_PRELOAD_INFO(ctx));
 
     g_work_queue_wait_for_completion(queue, gid);