diff --git a/kernel/include/kmalloc.h b/kernel/include/kmalloc.h
index 8573c7c..d959f59 100644
--- a/kernel/include/kmalloc.h
+++ b/kernel/include/kmalloc.h
@@ -15,6 +15,32 @@ void* kzalloc(size_t size);
 // zeroed-out memory. Returns a pointer to allocated memory or NULL.
 void* kcalloc(size_t count, size_t size);
 
+// 9 bytes in 8 bytes
+#define TAG_SHIFT 56
+#define TAG_MASK ((uintptr_t)0xFF << TAG_SHIFT)
+#define SIGN_BIT ((uintptr_t)1 << 55)
+
+typedef struct {
+	uintptr_t bits;
+} tagged_ptr;
+
+static inline tagged_ptr tp_make(void* p, uint8_t tag) {
+	uintptr_t addr = (uintptr_t)p & ~TAG_MASK;
+	return (tagged_ptr){addr | ((uintptr_t)tag << TAG_SHIFT)};
+}
+
+static inline uint8_t tp_tag(tagged_ptr t) {
+	return (uint8_t)(t.bits >> TAG_SHIFT);
+}
+
+static inline void* tp_ptr(tagged_ptr t) {
+	/* Bits 48..55 were untouched and still equal the canonical sign.
+	   Copy bit 55 into bits 56..63. */
+	uintptr_t addr =
+	    (t.bits & SIGN_BIT) ? (t.bits | TAG_MASK) : (t.bits & ~TAG_MASK);
+	return (void*)addr;
+}
+
 // powers of 2
 #define KMALLOC_MAX_SIZE 17
 #define LOWEST_POWER 3
@@ -28,6 +54,12 @@ struct partialy_freelist {
 	struct partialy_freelist* partialy_freelist_nextnode;
 };
 
+// this is used when object is too small to hold the big one in this case we use
+// first object as pointer storage and dont allocate it
+struct partialy_freelist_short {
+	struct partialy_freelist_short* partialy_freelist_nextnode;
+};
+
 union freelists_list {
 	struct freelist* freelists[KMALLOC_MAX_SIZE];
 	struct {
diff --git a/kernel/src/kmalloc.c b/kernel/src/kmalloc.c
index b59b730..7d7e7f8 100644
--- a/kernel/src/kmalloc.c
+++ b/kernel/src/kmalloc.c
@@ -58,6 +58,7 @@ void* kmalloc(size_t size) {
 
 	size_t alloc_index;
 	{
+		// TODO remove buildin
 		uint8_t leading_zeros = __builtin_clzll(size);
 		if (popcountll(size) == 1) {
 			alloc_index = leading_zeros;
@@ -73,6 +74,8 @@ void* kmalloc(size_t size) {
 		}
 		alloc_index -= LOWEST_POWER;
 	}
+	size_t object_size = (1 << (alloc_index + LOWEST_POWER));
+
 	struct freelist* selected_freelist =
 	    freelists_all_cpus[core_id].freelists[alloc_index];
 	if (selected_freelist != NULL) {
@@ -100,14 +103,16 @@ void* kmalloc(size_t size) {
 				// only one free object available
 				if (selected_partialy_freelist
 					->partialy_freelist_nextnode != NULL) {
-					// next partialy free list available
+					// next partialy free list
+					// available
 					partial_free_lists[core_id]
 					    .partialy_freelists[alloc_index] =
 					    selected_partialy_freelist
 						->partialy_freelist_nextnode;
 
 				} else {
-					// no other partialy free list available
+					// no other partialy free list
+					// available
 					partial_free_lists[core_id]
 					    .partialy_freelists[alloc_index] =
 					    NULL;
@@ -129,35 +134,44 @@ void* kmalloc(size_t size) {
 
 		} else {
 			// slowest path
-			size_t object_size = (1 << (alloc_index+LOWEST_POWER));
 
-			uint8_t page_count = object_size < PAGE_SIZE ? 1 : (object_size*2)/PAGE_SIZE;
+			uint8_t page_count =
+			    object_size < PAGE_SIZE
+				? 1
+				: (object_size * 2) / PAGE_SIZE;
 
 			void** alloc_addr = 0;
 			{
 				int ret;
-				ret = vmm_alloc((void**)&alloc_addr, page_count, 0);
+				ret = vmm_alloc((void**)&alloc_addr, page_count,
+						0);
 				if (ret != SUCCESS) {
 					return NULL;
 				}
 			}
 
 			// construct the linked list
-			// start at offset if object too small for partial free list pointer 
-			for (size_t i = object_size < sizeof(struct partialy_freelist) ? object_size : 0; i<page_count*PAGE_SIZE ;i += object_size) {
-				if ((i+1)*object_size>=page_count*PAGE_SIZE) {
-					alloc_addr[i*object_size] = NULL;
+			// start at offset if object too small for partial free
+			// list pointer
+			for (size_t i =
+				 object_size < sizeof(struct partialy_freelist)
+				     ? object_size
+				     : 0;
+			     i < page_count * PAGE_SIZE; i += object_size) {
+				if ((i + 1) * object_size >=
+				    page_count * PAGE_SIZE) {
+					alloc_addr[i * object_size] = NULL;
 				} else {
-					alloc_addr[i*object_size] = alloc_addr + ((i+1)*object_size);
+					alloc_addr[i * object_size] =
+					    alloc_addr +
+					    ((i + 1) * object_size);
 				}
 			}
-			if (object_size < sizeof(struct partialy_freelist)) {
-				freelists_all_cpus[core_id].freelists[alloc_index] = (struct freelist*)(alloc_addr+(object_size*2));
-				return alloc_addr+object_size;
-			} else  {
-				freelists_all_cpus[core_id].freelists[alloc_index] = (struct freelist*)(alloc_addr+object_size);
-				return alloc_addr;
-			}
+
+			freelists_all_cpus[core_id].freelists[alloc_index] =
+			    (struct freelist*)(alloc_addr + object_size);
+			return alloc_addr;
 		}
 	}
-}
\ No newline at end of file
+}
+void* kfree(void* p) {}
\ No newline at end of file