Merge pull request #1740 from AFLplusplus/dev

push to stable

6 files changed, 541 insertions, 5 deletions

diff --git a/custom_mutators/aflpp_tritondse/README.md b/custom_mutators/aflpp_tritondse/README.md
index 8a5dd02b..033655d2 100644
--- a/custom_mutators/aflpp_tritondse/README.md
+++ b/custom_mutators/aflpp_tritondse/README.md
@@ -10,8 +10,13 @@
 ../../afl-cc -o ../../test-instr ../../test-instr.c
 mkdir -p in
 echo aaaa > in/in
-TRITON_DSE_TARGET=../../test-instr AFL_CUSTOM_MUTATOR_ONLY=1 AFL_SYNC_TIME=1 AFL_PYTHON_MODULE=aflpp_tritondse PYTHONPATH=. ../../afl-fuzz -i in -o out -- ../../test-instr
+AFL_DISABLE_TRIM=1 AFL_CUSTOM_MUTATOR_ONLY=1 AFL_SYNC_TIME=1 AFL_PYTHON_MODULE=aflpp_tritondse PYTHONPATH=. ../../afl-fuzz -i in -o out -- ../../test-instr
 ```
 
 Note that this custom mutator works differently, new finds are synced
-after 10-60 seconds to the fuzzing instance.
+after 10-60 seconds to the fuzzing instance. This is necessary because only
+C/C++ custom mutators have access to the internal AFL++ state.
+
+Note that you should run first with `AFL_DEBUG` for 5-10 minutes and see if
+all important libraries and syscalls are hooked (look at `WARNING` and `CRITICAL`
+output during the run, best use with `AFL_NO_UI=1`)
diff --git a/custom_mutators/aflpp_tritondse/aflpp_tritondse.py b/custom_mutators/aflpp_tritondse/aflpp_tritondse.py
index e0219f0b..58b506b6 100644
--- a/custom_mutators/aflpp_tritondse/aflpp_tritondse.py
+++ b/custom_mutators/aflpp_tritondse/aflpp_tritondse.py
@@ -22,14 +22,17 @@ config = None
 dse = None
 cycle = 0
 count = 0
+finding = 0
 hashes = set()
 format = SeedFormat.RAW
 
 def pre_exec_hook(se: SymbolicExecutor, state: ProcessState):
     global count
     global hashes
+    global finding
     if se.seed.hash not in hashes:
         hashes.add(se.seed.hash)
+        finding = 1
         filename = out_path + "/id:" + f"{count:06}" + "," + se.seed.hash
         if not os.path.exists(filename):
             if is_debug:
@@ -47,6 +50,59 @@ def pre_exec_hook(se: SymbolicExecutor, state: ProcessState):
     #        file.write(se.seed.content)
 
 
+#def rtn_open(se: SymbolicExecutor, pstate: ProcessState, pc):
+#    """
+#    The open behavior.
+#    """
+#    logging.debug('open hooked')
+#
+#    # Get arguments
+#    arg0 = pstate.get_argument_value(0)  # const char *pathname
+#    flags = pstate.get_argument_value(1)  # int flags
+#    mode = pstate.get_argument_value(2)  # int mode
+#    arg0s = pstate.memory.read_string(arg0)
+#
+#    # Concretize the whole path name
+#    pstate.concretize_memory_bytes(arg0, len(arg0s)+1)  # Concretize the whole string + \0
+#
+#    # We use flags as concrete value
+#    pstate.concretize_argument(1)
+#
+#    # Use the flags to open the file in the write mode.
+#    mode = ""
+#    if (flags & 0xFF) == 0x00:   # O_RDONLY
+#        mode = "r"
+#    elif (flags & 0xFF) == 0x01: # O_WRONLY
+#        mode = "w"
+#    elif (flags & 0xFF) == 0x02: # O_RDWR
+#        mode = "r+"
+#
+#    if (flags & 0x0100): # O_CREAT
+#        mode += "x"
+#    if (flags & 0x0200): # O_APPEND
+#        mode = "a"  # replace completely value
+#
+#    if se.seed.is_file_defined(arg0s) and "r" in mode:  # input file and opened in reading
+#        logging.info(f"opening an input file: {arg0s}")
+#        # Program is opening an input
+#        data = se.seed.get_file_input(arg0s)
+#        filedesc = pstate.create_file_descriptor(arg0s, io.BytesIO(data))
+#        fd = filedesc.id
+#    else:
+#        # Try to open it as a regular file
+#        try:
+#            fd = open(arg0s, mode)  # use the mode here
+#            filedesc = pstate.create_file_descriptor(arg0s, fd)
+#            fd = filedesc.id
+#        except Exception as e:
+#            logging.debug(f"Failed to open {arg0s} {e}")
+#            fd = pstate.minus_one
+#
+#    pstate.write_register("rax", fd)  # write the return value
+#    pstate.cpu.program_counter = pstate.pop_stack_value()  # pop the return value
+#    se.skip_instruction()  # skip the current instruction so that the engine go straight fetching the next instruction
+
+
 def init(seed):
     global config
     global dse
@@ -64,6 +120,10 @@ def init(seed):
         is_debug = True
     except KeyError:
         pass
+    if is_debug:
+        logging.basicConfig(level=logging.WARNING)
+    else:
+        logging.basicConfig(level=logging.CRITICAL)
     try:
         foo = os.environ['AFL_CUSTOM_INFO_OUT']
         out_path = foo + '/../tritondse/queue'
@@ -115,10 +175,16 @@ def init(seed):
     dse = SymbolicExplorator(config, prog)
     # Add callbacks.
     dse.callback_manager.register_pre_execution_callback(pre_exec_hook)
+    #dse.callback_manager.register_function_callback("open", rtn_open)
 
 
-#def fuzz(buf, add_buf, max_size):
-#    return b""
+def fuzz(buf, add_buf, max_size):
+    global finding
+    finding = 1
+    while finding == 1:
+      finding = 0
+      dse.step()
+    return b""
 
 
 def queue_new_entry(filename_new_queue, filename_orig_queue):
@@ -141,8 +207,14 @@ def queue_new_entry(filename_new_queue, filename_orig_queue):
         dse.add_input_seed(seed)
         # Start exploration!
         #dse.step()
-        dse.explore()
+        #dse.explore()
     pass
 
+
+# we simulate just doing one single fuzz in the custom mutator
+def fuzz_count(buf):
+    return 1
+
+
 def splice_optout():
     pass
diff --git a/custom_mutators/symcc/README.md b/custom_mutators/symcc/README.md
index 364a348e..a6839a37 100644
--- a/custom_mutators/symcc/README.md
+++ b/custom_mutators/symcc/README.md
@@ -5,6 +5,8 @@ This uses the symcc to find new paths into the target.
 Note that this is a just a proof of concept example! It is better to use
 the fuzzing helpers of symcc, symqemu, Fuzzolic, etc. rather than this.
 
+Also the symqemu custom mutator is better than this.
+
 To use this custom mutator follow the steps in the symcc repository 
 [https://github.com/eurecom-s3/symcc/](https://github.com/eurecom-s3/symcc/) 
 on how to build symcc and how to instrument a target binary (the same target
diff --git a/custom_mutators/symqemu/Makefile b/custom_mutators/symqemu/Makefile
new file mode 100644
index 00000000..958aec19
--- /dev/null
+++ b/custom_mutators/symqemu/Makefile
@@ -0,0 +1,14 @@
+
+ifdef DEBUG
+  CFLAGS += -DDEBUG
+endif
+
+all: symqemu-mutator.so
+
+CFLAGS	+= -O3 -funroll-loops
+
+symqemu-mutator.so: symqemu.c
+	$(CC) -g $(CFLAGS) $(CPPFLAGS) -g -I../../include -shared -fPIC -o symqemu-mutator.so symqemu.c
+
+clean:
+	rm -f symqemu-mutator.so *.o *~ core
diff --git a/custom_mutators/symqemu/README.md b/custom_mutators/symqemu/README.md
new file mode 100644
index 00000000..c3071afc
--- /dev/null
+++ b/custom_mutators/symqemu/README.md
@@ -0,0 +1,19 @@
+# custum mutator: symqemu
+
+This uses the symcc to find new paths into the target.
+
+## How to build and use
+
+To use this custom mutator follow the steps in the symqemu repository 
+[https://github.com/eurecom-s3/symqemu/](https://github.com/eurecom-s3/symqemu/) 
+on how to build symqemu-x86_x64 and put it in your `PATH`.
+
+Just type `make` to build this custom mutator.
+
+```AFL_CUSTOM_MUTATOR_LIBRARY=custom_mutators/symqemu/symqemu-mutator.so AFL_DISABLE_TRIM=1 afl-fuzz ...```
+
+## Options
+
+`SYMQEMU_ALL=1` - use concolic solving on **all** queue items, not only interesting/favorite ones.
+
+`SYMQEMU_LATE=1` - use concolic solving only after there have been no finds for 5 minutes.
diff --git a/custom_mutators/symqemu/symqemu.c b/custom_mutators/symqemu/symqemu.c
new file mode 100644
index 00000000..73a1640a
--- /dev/null
+++ b/custom_mutators/symqemu/symqemu.c
@@ -0,0 +1,424 @@
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <ctype.h>
+#include "config.h"
+#include "debug.h"
+#include "afl-fuzz.h"
+#include "common.h"
+
+afl_state_t *afl_struct;
+static u32   debug = 0;
+static u32   found_items = 0;
+
+#define SYMQEMU_LOCATION "symqemu"
+
+#define DBG(x...) \
+  if (debug) { fprintf(stderr, x); }
+
+typedef struct my_mutator {
+
+  afl_state_t *afl;
+  u32          all;
+  u32          late;
+  u8          *mutator_buf;
+  u8          *out_dir;
+  u8          *target;
+  u8          *symqemu;
+  u8          *input_file;
+  u32          counter;
+  u32          seed;
+  u32          argc;
+  u8         **argv;
+
+} my_mutator_t;
+
+my_mutator_t *afl_custom_init(afl_state_t *afl, unsigned int seed) {
+
+  if (getenv("AFL_DEBUG")) debug = 1;
+
+  my_mutator_t *data = calloc(1, sizeof(my_mutator_t));
+  if (!data) {
+
+    perror("afl_custom_init alloc");
+    return NULL;
+
+  }
+
+  char *path = getenv("PATH");
+  char *exec_name = "symqemu-x86_64";
+  char *token = strtok(path, ":");
+  char  exec_path[4096];
+
+  while (token != NULL && data->symqemu == NULL) {
+
+    snprintf(exec_path, sizeof(exec_path), "%s/%s", token, exec_name);
+    if (access(exec_path, X_OK) == 0) {
+
+      data->symqemu = (u8 *)strdup(exec_path);
+      break;
+
+    }
+
+    token = strtok(NULL, ":");
+
+  }
+
+  if (!data->symqemu) FATAL("symqemu binary %s not found", exec_name);
+  DBG("Found %s\n", data->symqemu);
+
+  if (getenv("AFL_CUSTOM_MUTATOR_ONLY")) {
+
+    WARNF(
+        "the symqemu module is not very effective with "
+        "AFL_CUSTOM_MUTATOR_ONLY.");
+
+  }
+
+  if ((data->mutator_buf = malloc(MAX_FILE)) == NULL) {
+
+    free(data);
+    perror("mutator_buf alloc");
+    return NULL;
+
+  }
+
+  data->target = getenv("AFL_CUSTOM_INFO_PROGRAM");
+
+  u8 *path_tmp = getenv("AFL_CUSTOM_INFO_OUT");
+  u32 len = strlen(path_tmp) + 32;
+  u8 *symqemu_path = malloc(len);
+  data->out_dir = malloc(len);
+  snprintf(symqemu_path, len, "%s/%s", path_tmp, SYMQEMU_LOCATION);
+  snprintf(data->out_dir, len, "%s/out", symqemu_path, path_tmp);
+
+  (void)mkdir(symqemu_path, 0755);
+  (void)mkdir(data->out_dir, 0755);
+
+  setenv("SYMCC_OUTPUT_DIR", data->out_dir, 1);
+
+  data->input_file = getenv("AFL_CUSTOM_INFO_PROGRAM_INPUT");
+
+  u8 *tmp = NULL;
+  if ((tmp = getenv("AFL_CUSTOM_INFO_PROGRAM_ARGV")) && *tmp) {
+
+    int argc = 0, index = 2;
+    for (u32 i = 0; i < strlen(tmp); ++i)
+      if (isspace(tmp[i])) ++argc;
+
+    data->argv = (u8 **)malloc((argc + 4) * sizeof(u8 **));
+    u8 *p = strdup(tmp);
+
+    do {
+
+      data->argv[index] = p;
+      while (*p && !isspace(*p))
+        ++p;
+      if (*p) {
+
+        *p++ = 0;
+        while (isspace(*p))
+          ++p;
+
+      }
+
+      if (strcmp(data->argv[index], "@@") == 0) {
+
+        if (!data->input_file) {
+
+          u32 ilen = strlen(symqemu_path) + 32;
+          data->input_file = malloc(ilen);
+          snprintf(data->input_file, ilen, "%s/.input", symqemu_path);
+
+        }
+
+        data->argv[index] = data->input_file;
+
+      }
+
+      DBG("%d: %s\n", index, data->argv[index]);
+      index++;
+
+    } while (*p);
+
+    data->argv[index] = NULL;
+    data->argc = index;
+
+  } else {
+
+    data->argv = (u8 **)malloc(8 * sizeof(u8 **));
+    data->argc = 2;
+    data->argv[2] = NULL;
+
+  }
+
+  data->argv[0] = data->symqemu;
+  data->argv[1] = data->target;
+  data->afl = afl;
+  data->seed = seed;
+  afl_struct = afl;
+
+  if (getenv("SYMQEMU_ALL")) { data->all = 1; }
+  if (getenv("SYMQEMU_LATE")) { data->late = 1; }
+  if (data->input_file) { setenv("SYMCC_INPUT_FILE", data->input_file, 1); }
+
+  DBG("out_dir=%s, target=%s, input_file=%s, argc=%u\n", data->out_dir,
+      data->target,
+      data->input_file ? (char *)data->input_file : (char *)"<stdin>",
+      data->argc);
+
+  if (debug) {
+
+    fprintf(stderr, "[");
+    for (u32 i = 0; i <= data->argc; ++i)
+      fprintf(stderr, " \"%s\"",
+              data->argv[i] ? (char *)data->argv[i] : "<NULL>");
+    fprintf(stderr, " ]\n");
+
+  }
+
+  return data;
+
+}
+
+/* No need to receive a splicing item */
+void afl_custom_splice_optout(void *data) {
+
+  (void)(data);
+
+}
+
+/* Get unix time in milliseconds */
+
+inline u64 get_cur_time(void) {
+
+  struct timeval  tv;
+  struct timezone tz;
+
+  gettimeofday(&tv, &tz);
+
+  return (tv.tv_sec * 1000ULL) + (tv.tv_usec / 1000);
+
+}
+
+u32 afl_custom_fuzz_count(my_mutator_t *data, const u8 *buf, size_t buf_size) {
+
+  if (likely((!afl_struct->queue_cur->favored && !data->all) ||
+             afl_struct->queue_cur->was_fuzzed)) {
+
+    return 0;
+
+  }
+
+  if (likely(data->late)) {
+
+    if (unlikely(get_cur_time() - afl_struct->last_find_time <=
+                 10 * 60 * 1000)) {
+
+      return 0;
+
+    }
+
+  }
+
+  int         pipefd[2];
+  struct stat st;
+
+  if (afl_struct->afl_env.afl_no_ui) {
+
+    ACTF("Sending to symqemu: %s", afl_struct->queue_cur->fname);
+
+  }
+
+  if (!(stat(afl_struct->queue_cur->fname, &st) == 0 && S_ISREG(st.st_mode) &&
+        st.st_size)) {
+
+    PFATAL("Couldn't find enqueued file: %s", afl_struct->queue_cur->fname);
+
+  }
+
+  if (afl_struct->fsrv.use_stdin) {
+
+    if (pipe(pipefd) == -1) {
+
+      PFATAL(
+          "Couldn't create a pipe for interacting with symqemu child process");
+
+    }
+
+  }
+
+  if (data->input_file) {
+
+    int     fd = open(data->input_file, O_WRONLY | O_CREAT | O_TRUNC, 0644);
+    ssize_t s = write(fd, buf, buf_size);
+    close(fd);
+    DBG("wrote %zd/%zd to %s\n", s, buf_size, data->input_file);
+
+  }
+
+  int pid = fork();
+
+  if (pid == -1) return 0;
+
+  if (likely(pid)) {
+
+    if (!data->input_file || afl_struct->fsrv.use_stdin) {
+
+      close(pipefd[0]);
+
+      if (fcntl(pipefd[1], F_GETPIPE_SZ)) {
+
+        fcntl(pipefd[1], F_SETPIPE_SZ, MAX_FILE);
+
+      }
+
+      ck_write(pipefd[1], buf, buf_size, data->input_file);
+
+      close(pipefd[1]);
+
+    }
+
+    pid = waitpid(pid, NULL, 0);
+    DBG("symqemu finished executing!\n");
+
+  } else /* (pid == 0) */ {  // child
+
+    if (afl_struct->fsrv.use_stdin) {
+
+      close(pipefd[1]);
+      dup2(pipefd[0], 0);
+
+    }
+
+    DBG("exec=%s\n", data->target);
+    if (!debug) {
+
+      close(1);
+      close(2);
+      dup2(afl_struct->fsrv.dev_null_fd, 1);
+      dup2(afl_struct->fsrv.dev_null_fd, 2);
+
+    }
+
+    execvp((char *)data->argv[0], (char **)data->argv);
+    fprintf(stderr, "Executing: [");
+    for (u32 i = 0; i <= data->argc; ++i)
+      fprintf(stderr, " \"%s\"",
+              data->argv[i] ? (char *)data->argv[i] : "<NULL>");
+    fprintf(stderr, " ]\n");
+    FATAL("Failed to execute %s %s\n", data->argv[0], data->argv[1]);
+    exit(-1);
+
+  }
+
+  /* back in mother process */
+
+  struct dirent **nl;
+  s32             i, items = scandir(data->out_dir, &nl, NULL, NULL);
+  found_items = 0;
+  char source_name[4096];
+
+  if (items > 0) {
+
+    for (i = 0; i < (u32)items; ++i) {
+
+      // symqemu output files start with a digit
+      if (!isdigit(nl[i]->d_name[0])) continue;
+
+      struct stat st;
+      snprintf(source_name, sizeof(source_name), "%s/%s", data->out_dir,
+               nl[i]->d_name);
+      DBG("file=%s\n", source_name);
+
+      if (stat(source_name, &st) == 0 && S_ISREG(st.st_mode) && st.st_size) {
+
+        ++found_items;
+
+      }
+
+      free(nl[i]);
+
+    }
+
+    free(nl);
+
+  }
+
+  DBG("Done, found %u items!\n", found_items);
+
+  return found_items;
+
+}
+
+size_t afl_custom_fuzz(my_mutator_t *data, u8 *buf, size_t buf_size,
+                       u8 **out_buf, u8 *add_buf, size_t add_buf_size,
+                       size_t max_size) {
+
+  struct dirent **nl;
+  s32             done = 0, i, items = scandir(data->out_dir, &nl, NULL, NULL);
+  char            source_name[4096];
+
+  if (items > 0) {
+
+    for (i = 0; i < (u32)items; ++i) {
+
+      // symqemu output files start with a digit
+      if (!isdigit(nl[i]->d_name[0])) continue;
+
+      struct stat st;
+      snprintf(source_name, sizeof(source_name), "%s/%s", data->out_dir,
+               nl[i]->d_name);
+      DBG("file=%s\n", source_name);
+
+      if (stat(source_name, &st) == 0 && S_ISREG(st.st_mode) && st.st_size) {
+
+        int fd = open(source_name, O_RDONLY);
+        if (fd < 0) { goto got_an_issue; }
+
+        ssize_t r = read(fd, data->mutator_buf, MAX_FILE);
+        close(fd);
+
+        DBG("fn=%s, fd=%d, size=%ld\n", source_name, fd, r);
+
+        if (r < 1) { goto got_an_issue; }
+
+        done = 1;
+        --found_items;
+        unlink(source_name);
+
+        *out_buf = data->mutator_buf;
+        return (u32)r;
+
+      }
+
+      free(nl[i]);
+
+    }
+
+    free(nl);
+
+  }
+
+got_an_issue:
+  *out_buf = NULL;
+  return 0;
+
+}
+
+/**
+ * Deinitialize everything
+ *
+ * @param data The data ptr from afl_custom_init
+ */
+void afl_custom_deinit(my_mutator_t *data) {
+
+  free(data->mutator_buf);
+  free(data);
+
+}
+