add file hole support

25 files changed, 428 insertions, 792 deletions

diff --git a/.gitignore b/.gitignore
index ab4c720..07f2f4b 100644
--- a/.gitignore
+++ b/.gitignore
@@ -3,4 +3,5 @@
 docs/html/
 docs/latex/
 ioriot/ioriot
+ioriot/tags
 systemtap/downloads/
diff --git a/README.md b/README.md
index be6aa8f..0ab83cd 100644
--- a/README.md
+++ b/README.md
@@ -339,6 +339,7 @@ Currently, these file I/O related syscalls are supported (as of CentOS 7):
 open

 openat

 lseek

+llseek

 fcntl

 creat

 write

diff --git a/ioriot/src/datas/btree.c b/ioriot/src/datas/btree.c
index b150066..a70c442 100644
--- a/ioriot/src/datas/btree.c
+++ b/ioriot/src/datas/btree.c
@@ -37,7 +37,7 @@ void btree_destroy2(btree_s* b)
     free(b);
 }
 
-int btree_insert(btree_s* b, int key, void *data)
+int btree_insert(btree_s* b, long key, void *data)
 {
     int ret = 0;
 
@@ -48,13 +48,13 @@ int btree_insert(btree_s* b, int key, void *data)
         ret = btreelem_insert_r(b->root, key, data);
     }
 
-    if (ret == 0)
+    if (ret == 1)
         b->size++;
 
     return ret;
 }
 
-void* btree_get(btree_s* b, int key)
+void* btree_get(btree_s* b, long key)
 {
     if (b->root == NULL)
         return NULL;
@@ -62,17 +62,45 @@ void* btree_get(btree_s* b, int key)
     return btreelem_get_r(b->root, key);
 }
 
+bool btree_has_range_l(btree_s* b, const long start, const long end)
+{
+    if (b->root == NULL)
+        return false;
+
+    return btreelem_has_range_lr(b->root, start, end);
+}
+
+long btree_get_l(btree_s* b, long key)
+{
+    void *data = btree_get(b, key);
+    if (data)
+        return (long)data;
+    else
+        return -1;
+}
+
+void btree_ensure_range_l(btree_s* b, const long start, const long end, const long threshold)
+{
+    if (b->root == NULL) {
+        btree_insert(b, start, (void*)end);
+    } else {
+        if (1 == btreelem_ensure_range_lr(b->root, start, end, threshold))
+            b->size++;
+    }
+}
+
 void btree_print(btree_s* b)
 {
-    btreelem_print_r(b->root, 0);
+    Put("btree:%p size:%d", (void*)b, b->size);
+    btreelem_print_r(b->root, 1);
 }
 
-void btree_run_cb2(btree_s* b, void (*cb)(void *data, void *data2))
+void btree_run_cb2(btree_s* b, void (*cb)(long key, void *data, void *data2), void *data2)
 {
-    btreelem_run_cb2_r(b->root, cb);
+    btreelem_run_cb2_r(b->root, cb, data2);
 }
 
-btreelem_s* btreelem_new(int key, void *data)
+btreelem_s* btreelem_new(long key, void *data)
 {
     btreelem_s *e = Malloc(btreelem_s);
 
@@ -105,7 +133,7 @@ void btreelem_destroy_r2(btreelem_s* e)
     free(e);
 }
 
-int btreelem_insert_r(btreelem_s* e, int key, void *data)
+int btreelem_insert_r(btreelem_s* e, long key, void *data)
 {
     int ret = 1;
 
@@ -130,19 +158,65 @@ int btreelem_insert_r(btreelem_s* e, int key, void *data)
     return ret;
 }
 
-void* btreelem_get_r(btreelem_s* e, int key)
+int btreelem_ensure_range_lr(btreelem_s *e, const long start, const long end, const long threshold)
+{
+    int ret = 0;
+    long value = (long) e->data;
+
+    //Debug("%ld %ld %ld", start, end, threshold);
+
+    if (e->key == start) {
+        if (value < end) {
+            e->data = (void*) end;
+        } else {
+            // Nothing to do, range already present
+        }
+
+    } else if (e->key > start) {
+        if (e->left == NULL) {
+            if (value <= end) {
+                e->key = start;
+                e->data = (void*)end;
+            } else {
+                e->left = btreelem_new(start, (void*)end);
+                ret = 1;
+            }
+        } else {
+            ret = btreelem_ensure_range_lr(e->left, start, end, threshold);
+        }
+
+    } else { // if (e->key < start)
+        if (value >= start) {
+            if (value < end) {
+                e->data = (void*) end;
+            } else {
+                // Nothing to do, range already present
+            }
+        } else {
+            if (e->right == NULL) {
+                e->right = btreelem_new(start, (void*)end);
+                ret = 1;
+            } else {
+                ret = btreelem_ensure_range_lr(e->right, start, end, threshold);
+            }
+        }
+    }
+
+    return ret;
+}
+
+void* btreelem_get_r(btreelem_s* e, long key)
 {
     void *data = NULL;
 
-    if (e->key == key)
+    if (e->key == key) {
         data = e->data;
 
-    else if (e->key > key) {
+    } else if (e->key > key) {
         if (e->left)
             data = btreelem_get_r(e->left, key);
-    }
 
-    else {
+    } else {
         if (e->right)
             data = btreelem_get_r(e->right, key);
     }
@@ -150,6 +224,27 @@ void* btreelem_get_r(btreelem_s* e, int key)
     return data;
 }
 
+bool btreelem_has_range_lr(btreelem_s* e, const long start, const long end)
+{
+    long value = (long)e->data;
+
+    if (e->key <= start && value >= end) {
+        return true;
+
+    } else if (e->key > start) {
+        if (e->left)
+            return btreelem_has_range_lr(e->left, start, end);
+        else
+            return false;
+
+    } else {
+        if (e->right)
+            return btreelem_has_range_lr(e->right, start, end);
+        else
+            return false;
+    }
+}
+
 void btreelem_print_r(btreelem_s* e, int depth)
 {
     if (!e)
@@ -157,7 +252,7 @@ void btreelem_print_r(btreelem_s* e, int depth)
 
     for (int i = 0; i < depth; ++i)
         Out(" ");
-    Put("key:%d data:%ld\n", e->key, (long) e->data);
+    Put("key:%ld data:%ld", e->key, (long) e->data);
 
     if (e->left)
         btreelem_print_r(e->left, depth);
@@ -166,18 +261,18 @@ void btreelem_print_r(btreelem_s* e, int depth)
         btreelem_print_r(e->right, depth+1);
 }
 
-void btreelem_run_cb2_r(btreelem_s* e, void (*cb)(void *data, void *data2))
+void btreelem_run_cb2_r(btreelem_s* e, void (*cb)(long key, void *data, void *data2), void *data2)
 {
     if (!e)
         return;
 
-    cb((void*)(long)e->key, e->data);
+    cb(e->key, e->data, data2);
 
     if (e->left)
-        btreelem_run_cb2_r(e->left, cb);
+        btreelem_run_cb2_r(e->left, cb, data2);
 
     if (e->right)
-        btreelem_run_cb2_r(e->right, cb);
+        btreelem_run_cb2_r(e->right, cb, data2);
 }
 
 void btree_test(void)
@@ -186,23 +281,68 @@ void btree_test(void)
     void* somedata = (void*)b;
 
     assert(1 == btree_insert(b, 1, (void*)1));
+    assert(1 == b->size);
     assert(1 == btree_insert(b, 2, (void*)2));
+    assert(2 == b->size);
     assert(1 == btree_insert(b, 3, (void*)3));
-    assert(1 == (long)btree_get(b, 1));
+    assert(3 == b->size);
+    assert(1 == btree_get_l(b, 1));
 
     assert(1 == btree_insert(b, 1234, somedata));
+    assert(4 == b->size);
     assert(1 == btree_insert(b, 13, somedata));
+    assert(5 == b->size);
     assert(1 == btree_insert(b, 666, somedata));
+    assert(6 == b->size);
     assert(0 == btree_insert(b, 13, somedata));
+    assert(6 == b->size);
 
     assert(NULL != btree_get(b, 666));
     assert(NULL == btree_get(b, 777));
 
     assert(0 == btree_insert(b, 666, somedata));
     assert(1 == btree_insert(b, 42, (void*)42));
-    assert(42 == (long)btree_get(b, 42));
+    assert(42 == btree_get_l(b, 42));
 
     btree_print(b);
+    btree_destroy(b);
 
+    b = btree_new();
+    assert(0 == b->size);
+
+    btree_ensure_range_l(b, 0, 23, 0);
+    assert(btree_get_l(b, 0) == (long) btree_get(b, 0));
+    assert(23 == btree_get_l(b, 0));
+    assert(btree_has_range_l(b, 2, 10));
+
+    assert(!btree_has_range_l(b, 300, 325));
+    btree_ensure_range_l(b, 300, 325, 0);
+    assert(2 == b->size);
+    assert(325 == btree_get_l(b, 300));
+    assert(btree_has_range_l(b, 300, 325));
+    assert(!btree_has_range_l(b, 300, 326));
+
+    btree_ensure_range_l(b, 200, 3321, 0);
+    assert(2 == b->size);
+
+    btree_ensure_range_l(b, 200, 1000, 0);
+    assert(2 == b->size);
+    assert(3321 == btree_get_l(b, 200));
+
+    btree_ensure_range_l(b, 0, 23, 0);
+    btree_ensure_range_l(b, 10, 10, 0);
+    btree_ensure_range_l(b, 22, 25, 0);
+    assert(25 == btree_get_l(b, 0));
+    assert(2 == b->size);
+
+    assert(!btree_has_range_l(b, 4000, 4000));
+    btree_ensure_range_l(b, 4000, 4000, 0);
+    assert(3 == b->size);
+    assert(btree_has_range_l(b, 4000, 4000));
+    assert(!btree_has_range_l(b, 4000, 4001));
+    assert(!btree_has_range_l(b, 3999, 4000));
+    btree_print(b);
+
+    btree_print(b);
     btree_destroy(b);
 }
diff --git a/ioriot/src/datas/btree.h b/ioriot/src/datas/btree.h
index acc5477..0a05cee 100644
--- a/ioriot/src/datas/btree.h
+++ b/ioriot/src/datas/btree.h
@@ -23,7 +23,7 @@
 typedef struct btreelem_ {
     struct btreelem_ *left; /**< The next element to the left */
     struct btreelem_ *right; /**< The next element to the right */
-    int key; /**< The key of the element */
+    long key; /**< The key of the element */
     void *data; /**< A pointer to the data stored in this element */
 } btreelem_s;
 
@@ -38,18 +38,23 @@ typedef struct btree_s_ {
 btree_s* btree_new();
 void btree_destroy(btree_s *b);
 void btree_destroy2(btree_s *b);
-int btree_insert(btree_s *b, int key, void *data);
-void* btree_get(btree_s *b, int key);
+int btree_insert(btree_s *b, long key, void *data);
+void* btree_get(btree_s *b, long key);
+long btree_get_l(btree_s *b, long key);
+bool btree_has_range_l(btree_s *b, const long start, const long end);
+void btree_ensure_range_l(btree_s *b, const long start, const long end, const long threshold);
 void btree_print(btree_s *b);
-void btree_run_cb2(btree_s* b, void (*cb)(void *data, void *data2));
+void btree_run_cb2(btree_s* b, void (*cb)(long key, void *data, void *data2), void *data2);
 void btree_test(void);
 
-btreelem_s* btreelem_new(int key, void *data);
+btreelem_s* btreelem_new(long key, void *data);
 void btreelem_destroy_r(btreelem_s *e);
 void btreelem_destroy_r2(btreelem_s *e);
-int btreelem_insert_r(btreelem_s *e, int key, void *data);
-void* btreelem_get_r(btreelem_s *e, int key);
+int btreelem_insert_r(btreelem_s *e, long key, void *data);
+void* btreelem_get_r(btreelem_s *e, long key);
+bool btreelem_has_range_lr(btreelem_s *e, const long start, const long end);
+int btreelem_ensure_range_lr(btreelem_s *e, const long start, const long end, const long threshold);
 void btreelem_print_r(btreelem_s *e, int depth);
-void btreelem_run_cb2_r(btreelem_s* e, void (*cb)(void *data, void *data2));
+void btreelem_run_cb2_r(btreelem_s* e, void (*cb)(long key, void *data, void *data2), void *data2);
 
 #endif // BTREE_H
diff --git a/ioriot/src/datas/list.c b/ioriot/src/datas/list.c
index 9cc78db..81e3e57 100644
--- a/ioriot/src/datas/list.c
+++ b/ioriot/src/datas/list.c
@@ -14,7 +14,6 @@
 
 #include "list.h"
 
-
 list_s *list_new()
 {
     list_s *l = Malloc(list_s);
diff --git a/ioriot/src/defaults.h b/ioriot/src/defaults.h
index c0833c8..d822423 100644
--- a/ioriot/src/defaults.h
+++ b/ioriot/src/defaults.h
@@ -18,9 +18,9 @@
 #include "utils/utils.h"
 
 /** Version of the supported .capture format */
-#define CAPTURE_VERSION 1
+#define CAPTURE_VERSION 2
 /** Version of the supported .replay format */
-#define REPLAY_VERSION 1
+#define REPLAY_VERSION 2
 /** Max amount of tokens per line in the .capture file */
 #define MAX_TOKENS 10
 /** Max line length in either .capture or .replay file */
@@ -30,11 +30,13 @@
 /** Version of I/O Riot */
 #define IORIOT_VERSION "0.4-develop"
 /** Copyright information */
-#define IORIOT_COPYRIGHT "Mimecast 2017, 2018 (c)"
+#define IORIOT_COPYRIGHT "(c) Mimecast 2018"
 /** Max open files resource user limit */
 #define SET_RLIMIT_NOFILE 369216
 /** Max processes resource user limit */
 #define SET_RLIMIT_NPROC 30768
+/** Ignore file hole size */
+#define IGNORE_FILE_HOLE_BYTES 1024*1024*10
 
 // The following are for debugging purposes only
 
diff --git a/ioriot/src/generate/generate.c b/ioriot/src/generate/generate.c
index ee7178c..53751de 100644
--- a/ioriot/src/generate/generate.c
+++ b/ioriot/src/generate/generate.c
@@ -68,7 +68,6 @@ void generate_destroy(generate_s *g)
     free(g);
 }
 
-
 status_e generate_run(options_s *opts)
 {
     generate_s *g = generate_new(opts);
@@ -189,14 +188,30 @@ status_e generate_run(options_s *opts)
     return SUCCESS;
 }
 
-void generate_write_init_cb(void *data)
+void _write_ranges_cb(long start, void *data, void *data2)
 {
-    vsize_s *l = data;
-    generate_s *g = l->generate;
+    vsize_s *v = data2;
+    generate_s *g = v->generate;
+    long end = (long) data;
+    long bytes = end-start;
+    if (bytes > 0) {
+        fprintf(g->replay_fd, "%d|%d|%ld|%ld|%s|\n",
+                v->is_dir, v->is_file, start, bytes, v->path);
+    }
+}
 
-    if (l->required && strlen(l->path) > 0) {
-        fprintf(g->replay_fd, "%d|%d|%ld|%s|\n",
-                l->is_dir, l->is_file, -l->vsize_deficit, l->path);
+void generate_write_init_cb(void *data)
+{
+    vsize_s *v = data;
+    generate_s *g = v->generate;
+
+    if (v->required && strlen(v->path) > 0) {
+        if (v->read_ranges) {
+            btree_run_cb2(v->read_ranges, _write_ranges_cb, data);
+        } else if (v->bytes >= 0) {
+            fprintf(g->replay_fd, "%d|%d|%ld|%ld|%s|\n",
+                    v->is_dir, v->is_file, 0L, v->bytes, v->path);
+        }
     }
 }
 
diff --git a/ioriot/src/generate/gioop.c b/ioriot/src/generate/gioop.c
index 01701bc..d5b9a28 100644
--- a/ioriot/src/generate/gioop.c
+++ b/ioriot/src/generate/gioop.c
@@ -27,7 +27,7 @@ status_e gioop_run(gwriter_s *w, gtask_s *t)
     generate_s *g = w->generate;
 
     // Get the virtual process data object from the virtual PID space and store
-    // a pointer to it to t->gprocess
+    // a pointer to it at t->gprocess
     generate_gprocess_by_realpid(g, t);
 
     // One of the open syscalls may openes a file handle succesfully
@@ -109,6 +109,9 @@ status_e gioop_run(gwriter_s *w, gtask_s *t)
     } else if (Eq(t->op, "lseek")) {
         Cleanup(gioop_lseek(w, t, g));
 
+    } else if (Eq(t->op, "llseek")) {
+        Cleanup(gioop_llseek(w, t, g));
+
     } else if (Eq(t->op, "getdents")) {
         Cleanup(gioop_getdents(w, t, g));
 
@@ -559,6 +562,22 @@ status_e gioop_lseek(gwriter_s *w, gtask_s *t, generate_s *g)
     return SUCCESS;
 }
 
+status_e gioop_llseek(gwriter_s *w, gtask_s *t, generate_s *g)
+{
+    if (!t->has_fd) {
+        return ERROR;
+    }
+
+    generate_vsize_by_path(g, t, t->vfd->path);
+    Gioop_write(LLSEEK, "%ld|%ld|%ld|%ld|llseek",
+                t->mapped_fd, t->offset, t->whence, t->bytes);
+
+    if (t->bytes >= 0)
+        vsize_seek(t->vsize, t->vfd, t->bytes);
+
+    return SUCCESS;
+}
+
 status_e gioop_getdents(gwriter_s *w, gtask_s *t, generate_s *g)
 {
     if (!t->has_fd) {
diff --git a/ioriot/src/generate/gioop.h b/ioriot/src/generate/gioop.h
index ad49713..1df57c1 100644
--- a/ioriot/src/generate/gioop.h
+++ b/ioriot/src/generate/gioop.h
@@ -79,6 +79,7 @@ status_e gioop_readlinkat(gwriter_s *w, gtask_s *t, generate_s *g);
 status_e gioop_write(gwriter_s *w, gtask_s *t, generate_s *g);
 status_e gioop_writev(gwriter_s *w, gtask_s *t, generate_s *g);
 status_e gioop_lseek(gwriter_s *w, gtask_s *t, generate_s *g);
+status_e gioop_llseek(gwriter_s *w, gtask_s *t, generate_s *g);
 status_e gioop_getdents(gwriter_s *w, gtask_s *t, generate_s *g);
 status_e gioop_mkdir(gwriter_s *w, gtask_s *t, generate_s *g);
 status_e gioop_rmdir(gwriter_s *w, gtask_s *t, generate_s *g);
diff --git a/ioriot/src/generate/vsize.c b/ioriot/src/generate/vsize.c
index f2d56ba..4bf8433 100644
--- a/ioriot/src/generate/vsize.c
+++ b/ioriot/src/generate/vsize.c
@@ -17,7 +17,6 @@
 #include "generate.h"
 
 // Helper macros
-
 #define _Set_file(v) v->is_file = true; v->unsure = v->is_dir = false
 #define _Set_dir(v) v->is_dir = true; v->unsure = v->is_file = false
 #define _Set_unsure(v) v->unsure = true
@@ -35,14 +34,14 @@ vsize_s* vsize_new(char *file_path, const unsigned long id,
     v->inserted = false;
     v->is_dir = false;
     v->is_file = false;
-    v->offset = -1;
     v->path = Clone(file_path);
     v->renamed = false;
     v->required = false;
     v->unsure = false;
     v->updates = 0;
-    v->vsize = 0;
-    v->vsize_deficit = 0;
+    v->bytes = 0;
+    v->read_ranges = NULL;
+    v->write_ranges = NULL;
 
     return v;
 }
@@ -52,6 +51,11 @@ void vsize_destroy(vsize_s *v)
     if (!v)
         return;
 
+    if (v->read_ranges)
+        btree_destroy(v->read_ranges);
+    if (v->write_ranges)
+        btree_destroy(v->write_ranges);
+
     free(v->path);
     free(v);
 }
@@ -91,9 +95,9 @@ void init_parent_dir(vsize_s *v, const char *path)
 void vsize_open(vsize_s *v, void *vfd, const char *path, const int flags)
 {
 
-    // v->first_encounter == false means, that this is the first occurance of
+    // v->updates == 0 means, that this is the first occurance of
     // this path and we didn't initialise it (means we didn't ensure that
-    // we want to create all parent directories etc.
+    // we want to create all parent directories etc.)
 
     if (v->updates == 0) {
         // We may use a recycled vfd object! When opening a file we always
@@ -161,56 +165,87 @@ void vsize_rename(vsize_s *v, vsize_s *v2,
         // We are not 100% sure that this is really a file,
         // the path might be still a directory though!
         _Set_unsure(v2);
+        v2->updates++;
 
         // For debugging purposes only
         _Set_renamed(v2);
-        v2->updates++;
     }
 }
 
-void vsize_adjust(vsize_s *v, vfd_s* vfd)
+void vsize_ensure_data_range(vsize_s *v, btree_s **ranges, const long offset, const long bytes)
 {
-    if (v->vsize >= vfd->offset)
-        return;
+    if (*ranges == NULL) {
+        if (offset == 0) {
+            if (v->bytes + IGNORE_FILE_HOLE_BYTES < bytes) {
+                // No file hole, just set the new (larger) vsize
+                v->bytes = bytes;
+            } else {
+                // Nothing to do, vsize is already sufficient
+            }
+
+        } else { // if (offset > 0)
+            if (offset <= v->bytes) {
+                // We won't add a hole to the file here either
+                if (v->bytes < bytes) {
+                    // No file hole, just set the new (larger) bytes
+                    v->bytes = bytes;
+                } else {
+                    // Nothing to do, bytes is already sufficient
+                }
+            } else { // if (offset > v->bytes)
+                // From offset larger than bytes, we have a hole in the file!
+                *ranges = btree_new();
+
+                // Insert root data range
+                btree_insert(*ranges, 0, (void*)v->bytes);
+
+                // This value is not used anymore, we use the btree instead from now
+                // to store all used data ranges of a file.
+                v->bytes = -1;
+
+                // Rerun this function with a data range btree initialised
+                vsize_ensure_data_range(v, ranges, offset, bytes);
+            }
+        }
 
-    long deficit = v->vsize - vfd->offset;
-    if (deficit < v->vsize_deficit) {
-        v->vsize_deficit = deficit;
-        _Set_required(v);
-        _Set_file(v);
+    } else { // if (*ranges != NULL)
+        btree_ensure_range_l(*ranges, offset, offset+bytes,
+                IGNORE_FILE_HOLE_BYTES);
     }
 }
 
 void vsize_read(vsize_s *v, void *vfd, const char *path, const int bytes)
 {
     vfd_s *vfd_ = vfd;
+
+    if (v->write_ranges == NULL ||
+            !btree_has_range_l(v->write_ranges, vfd_->offset, vfd_->offset+bytes)) {
+        vsize_ensure_data_range(v, &v->read_ranges, vfd_->offset, bytes);
+        _Set_required(v);
+        _Set_file(v);
+    }
+
     vfd_->offset += bytes;
-    vsize_adjust(v, vfd_);
     v->updates++;
 }
 
-void vsize_seek(vsize_s *v, void *vfd, const long new_offset)
+void vsize_write(vsize_s *v, void *vfd, const char *path, const int bytes)
 {
-    //vfd_s *vfd_ = vfd;
-
-    // The file's offset can be greater than the file's current size, in which
-    // case the next write to the file will extend the file. This is referred
-    // to as creating a hole in a file and is allowed. However, this behaviour
-    // does not suit the estimation of the file size before we want to run the
-    // test.
-
-    // TODO: Implement file hole support!
-    //v->updates++;
+    vfd_s *vfd_ = vfd;
+    vsize_ensure_data_range(v, &v->write_ranges, vfd_->offset, bytes);
+    vfd_->offset += bytes;
+    v->updates++;
 }
 
-void vsize_write(vsize_s *v, void *vfd, const char *path, const int bytes)
+void vsize_seek(vsize_s *v, void *vfd, const long new_offset)
 {
     vfd_s *vfd_ = vfd;
-    vfd_->offset += bytes;
 
-    if (v->vsize < vfd_->offset)
-        v->vsize = vfd_->offset;
+    // The file's offset can be greater than the file's current size, in which
+    // case the next write to the file will extend the file. This is referred
+    // to as creating a hole in a file.
 
+    vfd_->offset = new_offset;
     v->updates++;
 }
 
diff --git a/ioriot/src/generate/vsize.h b/ioriot/src/generate/vsize.h
index 8556ec0..ffda9e5 100644
--- a/ioriot/src/generate/vsize.h
+++ b/ioriot/src/generate/vsize.h
@@ -16,7 +16,7 @@
 #define VSIZE_H
 
 #include "../utils/utils.h"
-#include "../datas/hmap.h"
+#include "../datas/btree.h"
 #include "../vfd.h"
 
 /**
@@ -30,11 +30,11 @@
  */
 typedef struct vsize_s_ {
     char *path; /**< The path to the file/directory */
-    off_t offset; /**< The current file offset */
+    long bytes; /**< The virtual size */
+    btree_s *read_ranges; /**< Used to store used data ranges in a file with holes */
+    btree_s *write_ranges; /**< Used to store used data ranges in a file with holes */
     unsigned long id; /**< The vsize id */
     void *generate; /**< A pointer to the generate object */
-    long vsize; /**< The virtual size */
-    long vsize_deficit; /**< Size to use for file creating during init mode */
     bool renamed; /**< True if file/dir has been renamed */
     bool required; /**< True if init mode will create this file/dir */
     bool is_dir; /**< True if this file/dir is a directory */
@@ -72,19 +72,18 @@ void vsize_destroy(vsize_s *v);
 void init_parent_dir(vsize_s *v, const char *path);
 
 /**
- * @brief Adjusts the vsize
+ * @brief Ensure that we have a data range in the virtual file
  *
  * Compares the virtual file size of the file in the vsize
- * object to the the offset in the virtual file descriptor.
- * In case the offset is higher we have a size deficit and
- * we need to mark it. That way ioriot can ensure that
- * during init mode it will create a file with the correct
- * size prior of running the test!
+ * object to the the given offsets. This is required just in case
+ * we have holes in the file to be replayed.
  *
  * @param v The virtual size object
- * @param vfd The virtual file descriptor object
+ * @param ranges The ranges object to operaet on
+ * @param offset The start offset
+ * @param bytes The amount of bytes
  */
-void vsize_adjust(vsize_s *v, vfd_s* vfd);
+void vsize_ensure_data_range(vsize_s *v, btree_s **ranges, const long offset, const long bytes);
 
 /**
  * @brief Adjust vsize on open
diff --git a/ioriot/src/init/init.c b/ioriot/src/init/init.c
index e375379..3f9d8a0 100644
--- a/ioriot/src/init/init.c
+++ b/ioriot/src/init/init.c
@@ -21,7 +21,6 @@
 #include "../mounts.h"
 #include "../utils/futils.h"
 
-
 init_s *init_new(options_s *opts)
 {
     init_s *i = Malloc(init_s);
@@ -118,7 +117,8 @@ status_e init_run(options_s *opts)
     fseeko(i->replay_fd, init_offset, SEEK_SET);
 
     bool is_file = false, is_dir = false;
-    long vsize = 0;
+    long offset = 0;
+    long bytes = 0;
     char *path;
 
     // Stats
@@ -136,6 +136,7 @@ status_e init_run(options_s *opts)
     // Process the INIT section of the .replay file line by line.
 
     while ((read = getline(&line, &len, i->replay_fd)) != -1) {
+        Debug(line);
         char *tok = strtok_r(line, "|", &saveptr);
 
         for (int ntok = 0; tok; ntok++) {
@@ -147,12 +148,18 @@ status_e init_run(options_s *opts)
                 is_file = atoi(tok) == 1;
                 break;
             case 2:
-                vsize = atol(tok);
-                if (vsize < 0) {
-                    Error("Size overflow");
+                offset = atol(tok);
+                if (offset < 0) {
+                    Error("Offset overflow: '%ld'", offset);
                 }
                 break;
             case 3:
+                bytes = atol(tok);
+                if (bytes < 0) {
+                    Error("Size overflow: '%ld'", bytes);
+                }
+                break;
+            case 4:
                 path = tok;
                 break;
             default:
@@ -178,7 +185,8 @@ status_e init_run(options_s *opts)
 
         } else if (is_file) {
             task->is_file = true;
-            task->vsize = vsize;
+            task->bytes = bytes;
+            task->offset = offset;
         }
         task->path = Clone(path);
 
diff --git a/ioriot/src/init/itask.c b/ioriot/src/init/itask.c
index f04ce33..de7c551 100644
--- a/ioriot/src/init/itask.c
+++ b/ioriot/src/init/itask.c
@@ -35,7 +35,7 @@ void itask_destroy(itask_s *task)
 void itask_reset_stats(itask_s *task)
 {
     task->is_dir = task->is_file = false;
-    task->sizes_created = task->vsize = 0;
+    task->sizes_created = task->offset = task->bytes = 0;
     task->dirs_created = task->files_created = 0;
 
     if (task->path) {
@@ -60,7 +60,7 @@ void itask_extract_stats(itask_s *task, long* dirs_created, long *files_created,
 
 void itask_print(itask_s *task)
 {
-    Put("itask(%p): is_dir:%d is_file:%d vsize:%ld path:%s",
+    Put("itask(%p): is_dir:%d is_file:%d offset:%ld bytes:%ld path:%s",
         (void*)task, task->is_dir, task->is_file,
-        task->vsize, task->path);
+        task->offset, task->bytes, task->path);
 }
diff --git a/ioriot/src/init/itask.h b/ioriot/src/init/itask.h
index b10d515..21afba4 100644
--- a/ioriot/src/init/itask.h
+++ b/ioriot/src/init/itask.h
@@ -23,7 +23,8 @@
 typedef struct itask_s_ {
     bool is_dir;
     bool is_file;
-    long vsize;
+    long offset;
+    long bytes;
     char *path;
     long dirs_created;
     long files_created;
diff --git a/ioriot/src/init/ithread.c b/ioriot/src/init/ithread.c
index a580e70..bcc31f3 100644
--- a/ioriot/src/init/ithread.c
+++ b/ioriot/src/init/ithread.c
@@ -90,9 +90,13 @@ void ithread_run_task(ithread_s *t, itask_s *task)
     } else if (task->is_file) {
         if (!ensure_file_exists(task->path, &task->dirs_created)) {
             task->files_created++;
-            if (task->vsize > 0) {
-                append_random_to_file(task->path, task->vsize);
-                task->sizes_created += task->vsize;
+            if (task->bytes > 0) {
+                if (task->offset > 0) {
+                    write_random_to_file(task->path, task->bytes, task->offset);
+                } else {
+                    append_random_to_file(task->path, task->bytes);
+                }
+                task->sizes_created += task->bytes;
             }
         }
     }
diff --git a/ioriot/src/opcodes.h b/ioriot/src/opcodes.h
index 3d5c114..0dd060c 100644
--- a/ioriot/src/opcodes.h
+++ b/ioriot/src/opcodes.h
@@ -69,6 +69,7 @@ typedef enum {
     FCNTL = 70,
     GETDENTS,
     LSEEK,
+    LLSEEK,
 
     // mmap syscalls
     MMAP2 = 80,
diff --git a/ioriot/src/replay/rioop.c b/ioriot/src/replay/rioop.c
index 2e16c94..0ce19c5 100644
--- a/ioriot/src/replay/rioop.c
+++ b/ioriot/src/replay/rioop.c
@@ -151,6 +151,7 @@ void rioop_run(rprocess_s *p, rthread_s *t, rtask_s *task)
         rioop_getdents(p, t, task);
         break;
     case LSEEK:
+    case LLSEEK:
         rioop_lseek(p, t, task);
         break;
 
diff --git a/ioriot/src/utests.c b/ioriot/src/utests.c
index 0fc53c3..7cae7e8 100644
--- a/ioriot/src/utests.c
+++ b/ioriot