diff --git a/include/lpool.h b/include/lpool.h
index cfd46cd..5b29ad1 100644
--- a/include/lpool.h
+++ b/include/lpool.h
@@ -31,9 +31,9 @@ typedef struct LPool LPool;
 struct LPool {
     LPoolBlock *free_list;  // free list
     uint8_t *start;         // start of user buffer
-    uint8_t *end;           // one past end of user buffer
     uint32_t block_size;    // aligned size of each block
-    uint32_t capacity;      // number of blocks
+    uint32_t capacity;      // total number of blocks
+    uint32_t used;          // used number of blocks
     uint32_t align;         // alignment
 };
 
diff --git a/src/lpool.c b/src/lpool.c
index d0e4d1d..61a00be 100644
--- a/src/lpool.c
+++ b/src/lpool.c
@@ -36,34 +36,34 @@ void lpool_init(LPool *pool, void *buffer, size_t buffer_size, uint32_t block_si
         block_size = sizeof(LPoolBlock);
     }
     // ensure initial alignment
-    uint32_t stride = ALIGN_UP(block_size, align);
+    block_size = ALIGN_UP(block_size, align);
 
     // align first usable block
-    uintptr_t start_addr = (uintptr_t)buffer;
-    uintptr_t first_addr = ALIGN_UP(start_addr, align);
-    size_t padding = first_addr - start_addr;
+    uintptr_t buffer_addr = (uintptr_t)buffer;
+    uintptr_t first_addr = ALIGN_UP(buffer_addr, align);
+    size_t padding = first_addr - buffer_addr;
 
-    if (padding + stride > buffer_size) {
+    if (padding + block_size > buffer_size) {
         pool->capacity = 0;
         pool->free_list = NULL;
     } else {
         size_t usable = buffer_size - padding;
-        pool->capacity = usable / stride;
+        pool->capacity = usable / block_size;
 
         // build intrusive free list
         LPoolBlock *block = (LPoolBlock *)first_addr;
         pool->free_list = block;
-        for (uint32_t i = 0; i < pool->capacity - 1; i++) {
-            uint32_t next = (i + 1) * stride;
+        for (uint32_t i = 1; i < pool->capacity; i++) {
+            uint32_t next = i * block_size;
             block->next = next;
             block = (LPoolBlock *)(first_addr + next);
         }
         block->next = NULL_BLOCK;  // last element
     }
-    pool->block_size = stride;
+    pool->block_size = block_size;
     pool->align = align;
     pool->start = buffer;
-    pool->end = buffer + buffer_size;
+    pool->used = 0;
 }
 
 void *lpool_alloc(LPool *pool)
@@ -78,48 +78,52 @@ void *lpool_alloc(LPool *pool)
 #ifdef DEBUG
     memset(block, 0xCB, pool->block_size);
 #endif
+    pool->used++;
     return block->data;
 }
 
-void lpool_free(LPool *pool, void *p)
+void lpool_free(LPool *pool, void *ptr)
 {
-    if (p == NULL) {
+    if (ptr == NULL) {
         return;
     }
-    LPoolBlock *block = p;
 #ifdef DEBUG
-    assert((uint8_t *)p >= pool->start && (uint8_t *)p < pool->end);
-    assert(((uintptr_t)p - (uintptr_t)pool->start) % pool->block_size == 0);
+    assert(pool->used > 0);
+    uint8_t *end = pool->start + (pool->capacity * pool->block_size);
+    assert((uint8_t *)ptr >= pool->start && (uint8_t *)ptr < end);
+    assert((((uintptr_t)ptr - (uintptr_t)pool->start) & (pool->block_size - 1)) == 0);
 
     // double-free detection
-    for (LPoolBlock *f = pool->free_list; f; f = lpool_next(pool, f)) {
-        if (f == p) {
+    for (LPoolBlock *block = pool->free_list; block != NULL; block = lpool_next(pool, block)) {
+        if (block == ptr) {
             assert(0 && "Double free detected in pool!");
             return;
         }
     }
-    memset(p, 0xDD, pool->block_size);
+    memset(ptr, 0xDD, pool->block_size);
 #endif
+    LPoolBlock *block = ptr;
     // When not in use, store pointer to next free element inline in the block
     block->next = lpool_offset(pool, pool->free_list);
     pool->free_list = block;
+    pool->used--;
 }
 
 void lpool_reset(LPool *pool)
 {
     if (pool->capacity == 0) return;
 
-    uintptr_t start_addr = (uintptr_t)pool->start;
-    uintptr_t first_addr = ALIGN_UP(start_addr, pool->align);
+    uintptr_t first_addr = (uintptr_t)pool->start;
 
     // build intrusive free list
     LPoolBlock *block = (LPoolBlock *)first_addr;
     pool->free_list = block;
-    for (size_t i = 0; i < pool->capacity - 1; ++i) {
+    for (size_t i = 1; i < pool->capacity; ++i) {
         uint32_t next = i * pool->block_size;
         block->next = next;
         block = (LPoolBlock *)(first_addr + next);
     }
 
     block->next = NULL_BLOCK;  // last element
+    pool->used = 0;
 }
diff --git a/test/lpool_test.c b/test/lpool_test.c
index 5532f56..28d0ba8 100644
--- a/test/lpool_test.c
+++ b/test/lpool_test.c
@@ -1,4 +1,6 @@
 #include <stdio.h>
+#include <stdlib.h>
+
 #include "lpool.h"
 
 struct TestData {
@@ -10,9 +12,11 @@ struct TestData {
 
 int main(int argc, char *argv[])
 {
-    _Alignas(16) uint8_t store[2048];
+    //_Alignas(16) uint8_t store[2048];
+    uint8_t *store = malloc(2048);
+
     LPool pool;
-    lpool_init(&pool, store, sizeof store, sizeof(struct TestData), _Alignof(struct TestData));
+    lpool_init(&pool, store, 2048, sizeof(struct TestData), _Alignof(struct TestData));
     printf("align=%u capacity=%u block_size=%u\n", pool.align, pool.capacity, pool.block_size);
     struct TestData *data = lpool_alloc(&pool);
     size_t allocs = 1;
@@ -23,11 +27,21 @@ int main(int argc, char *argv[])
         }
         allocs++;
     }
-    printf("Total allocs = %zu\n", allocs);
-    printf("alloc succeded = %d\n", lpool_alloc(&pool) != NULL);
+    printf("Total allocs = %zu, used = %u\n", allocs, pool.used);
+    printf("alloc succeded = %d used  = %u\n", lpool_alloc(&pool) != NULL, pool.used);
     lpool_free(&pool, data);
+    printf("alloc succeded = %d used  = %u\n", lpool_alloc(&pool) != NULL, pool.used);
 
-    printf("alloc succeded = %d\n", lpool_alloc(&pool) != NULL);
+    lpool_reset(&pool);
+    allocs = 0;
+    for (;;) {
+        struct TestData *t = lpool_alloc(&pool);
+        if (t == NULL) {
+            break;
+        }
+        allocs++;
+    }
+    printf("Total allocs = %zu, used = %u\n", allocs, pool.used);
 
     return 0;
 }