diff --git a/CMakeLists.txt b/CMakeLists.txt
index 3f7e0956..301fdea7 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -750,8 +750,8 @@ if (MI_BUILD_TESTS)
     endif()
     target_compile_options(mimalloc-test-stress-dynamic PRIVATE ${mi_cflags})
     target_include_directories(mimalloc-test-stress-dynamic PRIVATE include)
-    target_link_libraries(mimalloc-test-stress-dynamic PRIVATE mimalloc ${mi_libraries})  # mi_version
     if(WIN32)
+      target_link_libraries(mimalloc-test-stress-dynamic PRIVATE mimalloc ${mi_libraries})  # mi_version    
       add_test(NAME test-stress-dynamic COMMAND ${CMAKE_COMMAND} -E env MIMALLOC_SHOW_STATS=1 $<TARGET_FILE:mimalloc-test-stress-dynamic>)
     else()
       if(APPLE)
diff --git a/bin/mimalloc-redirect-arm64.dll b/bin/mimalloc-redirect-arm64.dll
old mode 100644
new mode 100755
index e6360285..27172d2c
Binary files a/bin/mimalloc-redirect-arm64.dll and b/bin/mimalloc-redirect-arm64.dll differ
diff --git a/bin/mimalloc-redirect-arm64.lib b/bin/mimalloc-redirect-arm64.lib
old mode 100644
new mode 100755
index 11d71ef9..dca80b9b
Binary files a/bin/mimalloc-redirect-arm64.lib and b/bin/mimalloc-redirect-arm64.lib differ
diff --git a/bin/mimalloc-redirect-arm64ec.dll b/bin/mimalloc-redirect-arm64ec.dll
old mode 100644
new mode 100755
index f5ee4e47..a228af39
Binary files a/bin/mimalloc-redirect-arm64ec.dll and b/bin/mimalloc-redirect-arm64ec.dll differ
diff --git a/bin/mimalloc-redirect-arm64ec.lib b/bin/mimalloc-redirect-arm64ec.lib
old mode 100644
new mode 100755
index b88e8fc1..0ce77436
Binary files a/bin/mimalloc-redirect-arm64ec.lib and b/bin/mimalloc-redirect-arm64ec.lib differ
diff --git a/bin/mimalloc-redirect.dll b/bin/mimalloc-redirect.dll
old mode 100644
new mode 100755
index 9e015cc6..ec035f1b
Binary files a/bin/mimalloc-redirect.dll and b/bin/mimalloc-redirect.dll differ
diff --git a/bin/mimalloc-redirect.lib b/bin/mimalloc-redirect.lib
old mode 100644
new mode 100755
index 1d710c01..785fa475
Binary files a/bin/mimalloc-redirect.lib and b/bin/mimalloc-redirect.lib differ
diff --git a/bin/mimalloc-redirect32.dll b/bin/mimalloc-redirect32.dll
old mode 100644
new mode 100755
index 32799ffe..92578f24
Binary files a/bin/mimalloc-redirect32.dll and b/bin/mimalloc-redirect32.dll differ
diff --git a/bin/mimalloc-redirect32.lib b/bin/mimalloc-redirect32.lib
old mode 100644
new mode 100755
index e2927250..bf649787
Binary files a/bin/mimalloc-redirect32.lib and b/bin/mimalloc-redirect32.lib differ
diff --git a/cmake/mimalloc-config-version.cmake b/cmake/mimalloc-config-version.cmake
index bd4adb2a..ad066d09 100644
--- a/cmake/mimalloc-config-version.cmake
+++ b/cmake/mimalloc-config-version.cmake
@@ -1,6 +1,6 @@
 set(mi_version_major 3)
 set(mi_version_minor 0)
-set(mi_version_patch 3)
+set(mi_version_patch 4)
 set(mi_version ${mi_version_major}.${mi_version_minor})
 
 set(PACKAGE_VERSION ${mi_version})
diff --git a/include/mimalloc.h b/include/mimalloc.h
index 3be9f619..c4a4eb19 100644
--- a/include/mimalloc.h
+++ b/include/mimalloc.h
@@ -8,7 +8,7 @@ terms of the MIT license. A copy of the license can be found in the file
 #ifndef MIMALLOC_H
 #define MIMALLOC_H
 
-#define MI_MALLOC_VERSION 303   // major + 2 digits minor
+#define MI_MALLOC_VERSION 304   // major + 2 digits minor
 
 // ------------------------------------------------------
 // Compiler specific attributes
diff --git a/include/mimalloc/atomic.h b/include/mimalloc/atomic.h
index 2d890954..e1a6d76d 100644
--- a/include/mimalloc/atomic.h
+++ b/include/mimalloc/atomic.h
@@ -376,8 +376,9 @@ static inline void mi_atomic_yield(void) {
   _mm_pause();
 }
 #elif (defined(__GNUC__) || defined(__clang__)) && \
-      (defined(__x86_64__) || defined(__i386__) || defined(__arm__) || defined(__armel__) || defined(__ARMEL__) || \
-       defined(__aarch64__) || defined(__powerpc__) || defined(__ppc__) || defined(__PPC__)) || defined(__POWERPC__)
+      (defined(__x86_64__) || defined(__i386__) || \
+       defined(__aarch64__) || defined(__arm__) || \
+       defined(__powerpc__) || defined(__ppc__) || defined(__PPC__) || defined(__POWERPC__))
 #if defined(__x86_64__) || defined(__i386__)
 static inline void mi_atomic_yield(void) {
   __asm__ volatile ("pause" ::: "memory");
@@ -386,10 +387,16 @@ static inline void mi_atomic_yield(void) {
 static inline void mi_atomic_yield(void) {
   __asm__ volatile("wfe");
 }
-#elif (defined(__arm__) && __ARM_ARCH >= 7)
+#elif defined(__arm__)
+#if __ARM_ARCH >= 7
 static inline void mi_atomic_yield(void) {
   __asm__ volatile("yield" ::: "memory");
 }
+#else
+static inline void mi_atomic_yield(void) {
+  __asm__ volatile ("nop" ::: "memory");
+}
+#endif
 #elif defined(__powerpc__) || defined(__ppc__) || defined(__PPC__) || defined(__POWERPC__)
 #ifdef __APPLE__
 static inline void mi_atomic_yield(void) {
@@ -400,10 +407,6 @@ static inline void mi_atomic_yield(void) {
   __asm__ __volatile__ ("or 27,27,27" ::: "memory");
 }
 #endif
-#elif defined(__armel__) || defined(__ARMEL__)
-static inline void mi_atomic_yield(void) {
-  __asm__ volatile ("nop" ::: "memory");
-}
 #endif
 #elif defined(__sun)
 // Fallback for other archs
diff --git a/include/mimalloc/internal.h b/include/mimalloc/internal.h
index 88404f98..9ec28503 100644
--- a/include/mimalloc/internal.h
+++ b/include/mimalloc/internal.h
@@ -339,10 +339,6 @@ typedef struct mi_option_desc_s {
   const char*       legacy_name; // potential legacy option name
 } mi_option_desc_t;
 
-// the static options
-extern mi_decl_hidden mi_option_desc_t mi_options[_mi_option_last];
-
-
 /* -----------------------------------------------------------
   Inlined definitions
 ----------------------------------------------------------- */
@@ -570,14 +566,14 @@ static inline size_t _mi_page_map_index(const void* p, size_t* sub_idx) {
 static inline mi_page_t* _mi_unchecked_ptr_page(const void* p) {
   size_t sub_idx;
   const size_t idx = _mi_page_map_index(p, &sub_idx);
-  return _mi_page_map[idx][sub_idx];
+  return _mi_page_map[idx][sub_idx];  // NULL if p==NULL
 }
 
 static inline mi_page_t* _mi_checked_ptr_page(const void* p) {
   size_t sub_idx;
   const size_t idx = _mi_page_map_index(p, &sub_idx);
   mi_page_t** const sub = _mi_page_map[idx];
-  if mi_unlikely(sub == NULL) return (mi_page_t*)&_mi_page_empty;
+  if mi_unlikely(sub == NULL) return NULL;
   return sub[sub_idx];
 }
 
diff --git a/readme.md b/readme.md
index 81f2057e..5a495275 100644
--- a/readme.md
+++ b/readme.md
@@ -12,9 +12,9 @@ is a general purpose allocator with excellent [performance](#performance) charac
 Initially developed by Daan Leijen for the runtime systems of the
 [Koka](https://koka-lang.github.io) and [Lean](https://github.com/leanprover/lean) languages.
 
-Latest release   : `v3.0.2` (beta) (2025-03-06).  
-Latest v2 release: `v2.2.2` (2025-03-06).  
-Latest v1 release: `v1.9.2` (2024-03-06).
+Latest release   : `v3.0.3` (beta) (2025-03-28).  
+Latest v2 release: `v2.2.3` (2025-03-28).  
+Latest v1 release: `v1.9.3` (2024-03-28).
 
 mimalloc is a drop-in replacement for `malloc` and can be used in other programs
 without code changes, for example, on dynamically linked ELF-based systems (Linux, BSD, etc.) you can use it as:
@@ -84,6 +84,9 @@ Enjoy!
 
 ### Releases
 
+* 2025-03-28, `v1.9.3`, `v2.2.3`, `v3.0.3` (beta): Various small bug and build fixes, including:
+  fix arm32 pre v7 builds, fix mingw build, get runtime statistics, improve statistic commit counts, 
+  fix execution on non BMI1 x64 systems. 
 * 2025-03-06, `v1.9.2`, `v2.2.2`, `v3.0.2-beta`: Various small bug and build fixes. 
   Add `mi_options_print`, `mi_arenas_print`, and the experimental `mi_stat_get` and `mi_stat_get_json`. 
   Add `mi_thread_set_in_threadpool` and `mi_heap_set_numa_affinity` (v3 only). Add vcpkg portfile. 
diff --git a/src/free.c b/src/free.c
index 00ecf34d..6b9d3431 100644
--- a/src/free.c
+++ b/src/free.c
@@ -115,6 +115,7 @@ static inline void mi_block_check_unguard(mi_page_t* page, mi_block_t* block, vo
 
 // free a local pointer  (page parameter comes first for better codegen)
 static void mi_decl_noinline mi_free_generic_local(mi_page_t* page, void* p) mi_attr_noexcept {
+  mi_assert_internal(p!=NULL && page != NULL);
   mi_block_t* const block = (mi_page_has_aligned(page) ? _mi_page_ptr_unalign(page, p) : (mi_block_t*)p);
   mi_block_check_unguard(page, block, p);
   mi_free_block_local(page, block, true /* track stats */, true /* check for a full page */);
@@ -122,11 +123,7 @@ static void mi_decl_noinline mi_free_generic_local(mi_page_t* page, void* p) mi_
 
 // free a pointer owned by another thread (page parameter comes first for better codegen)
 static void mi_decl_noinline mi_free_generic_mt(mi_page_t* page, void* p) mi_attr_noexcept {
-  if (p==NULL) return;  // a NULL pointer is seen as abandoned (tid==0) with a full flag set
-  #if !MI_PAGE_MAP_FLAT
-  if (page==&_mi_page_empty) return;  // an invalid pointer may lead to using the empty page
-  #endif
-  mi_assert_internal(p!=NULL && page != NULL && page != &_mi_page_empty);
+  mi_assert_internal(p!=NULL && page != NULL);
   mi_block_t* const block = _mi_page_ptr_unalign(page, p); // don't check `has_aligned` flag to avoid a race (issue #865)
   mi_block_check_unguard(page, block, p);
   mi_free_block_mt(page, block);
@@ -150,13 +147,8 @@ static inline mi_page_t* mi_validate_ptr_page(const void* p, const char* msg)
     return NULL;
   }
   mi_page_t* page = _mi_safe_ptr_page(p);
-  if (page == NULL) {
-    if (p != NULL) {
-      _mi_error_message(EINVAL, "%s: invalid pointer: %p\n", msg, p);
-    }
-    #if !MI_PAGE_MAP_FLAT
-    page = (mi_page_t*)&_mi_page_empty;
-    #endif
+  if (p != NULL && page == NULL) {
+    _mi_error_message(EINVAL, "%s: invalid pointer: %p\n", msg, p);    
   }
   return page;
   #else
@@ -169,12 +161,9 @@ static inline mi_page_t* mi_validate_ptr_page(const void* p, const char* msg)
 void mi_free(void* p) mi_attr_noexcept
 {
   mi_page_t* const page = mi_validate_ptr_page(p,"mi_free");
-
-  #if MI_PAGE_MAP_FLAT  // if not flat, p==NULL leads to `_mi_page_empty` which leads to `mi_free_generic_mt`
   if mi_unlikely(page==NULL) return;
-  #endif
-  mi_assert_internal(page!=NULL);
-
+  mi_assert_internal(p!=NULL && page!=NULL);
+  
   const mi_threadid_t xtid = (_mi_prim_thread_id() ^ mi_page_xthread_id(page));
   if mi_likely(xtid == 0) {                        // `tid == mi_page_thread_id(page) && mi_page_flags(page) == 0`
     // thread-local, aligned, and not a full page
diff --git a/src/init.c b/src/init.c
index 6d4ce65e..892f4988 100644
--- a/src/init.c
+++ b/src/init.c
@@ -16,7 +16,7 @@ terms of the MIT license. A copy of the license can be found in the file
 
 // Empty page used to initialize the small free pages array
 const mi_page_t _mi_page_empty = {
-  MI_ATOMIC_VAR_INIT(MI_PAGE_IN_FULL_QUEUE),  // xthread_id  (must set flag to catch NULL on a free)
+  MI_ATOMIC_VAR_INIT(0),  // xthread_id 
   NULL,                   // free
   0,                      // used
   0,                      // capacity
diff --git a/src/options.c b/src/options.c
index 2657cb6b..f1d16d6b 100644
--- a/src/options.c
+++ b/src/options.c
@@ -98,8 +98,8 @@ int mi_version(void) mi_attr_noexcept {
 #endif
 #endif
 
-// Static options (only exposed for the debugger)
-mi_option_desc_t mi_options[_mi_option_last] =
+// Static options 
+static mi_option_desc_t mi_options[_mi_option_last] =
 {
   // stable options
 #if MI_DEBUG || defined(MI_SHOW_ERRORS)
@@ -329,7 +329,7 @@ static void mi_cdecl mi_out_stderr(const char* msg, void* arg) {
 #ifndef MI_MAX_DELAY_OUTPUT
 #define MI_MAX_DELAY_OUTPUT ((size_t)(16*1024))
 #endif
-static char out_buf[MI_MAX_DELAY_OUTPUT+1];
+static char mi_output_buffer[MI_MAX_DELAY_OUTPUT+1];
 static _Atomic(size_t) out_len;
 
 static void mi_cdecl mi_out_buf(const char* msg, void* arg) {
@@ -345,7 +345,8 @@ static void mi_cdecl mi_out_buf(const char* msg, void* arg) {
   if (start+n >= MI_MAX_DELAY_OUTPUT) {
     n = MI_MAX_DELAY_OUTPUT-start-1;
   }
-  _mi_memcpy(&out_buf[start], msg, n);
+  mi_assert_internal(start + n <= MI_MAX_DELAY_OUTPUT);
+  _mi_memcpy(&mi_output_buffer[start], msg, n);
 }
 
 static void mi_out_buf_flush(mi_output_fun* out, bool no_more_buf, void* arg) {
@@ -354,10 +355,10 @@ static void mi_out_buf_flush(mi_output_fun* out, bool no_more_buf, void* arg) {
   size_t count = mi_atomic_add_acq_rel(&out_len, (no_more_buf ? MI_MAX_DELAY_OUTPUT : 1));
   // and output the current contents
   if (count>MI_MAX_DELAY_OUTPUT) count = MI_MAX_DELAY_OUTPUT;
-  out_buf[count] = 0;
-  out(out_buf,arg);
+  mi_output_buffer[count] = 0;
+  out(mi_output_buffer,arg);
   if (!no_more_buf) {
-    out_buf[count] = '\n'; // if continue with the buffer, insert a newline
+    mi_output_buffer[count] = '\n'; // if continue with the buffer, insert a newline
   }
 }
 
diff --git a/src/page-map.c b/src/page-map.c
index 44d6de4a..78e8ab8e 100644
--- a/src/page-map.c
+++ b/src/page-map.c
@@ -210,9 +210,7 @@ bool _mi_page_map_init(void) {
   if (!mi_page_map_memid.initially_committed) {
     _mi_os_commit(_mi_page_map[0], os_page_size, NULL);   // only first OS page
   }
-  _mi_page_map[0][0] = (mi_page_t*)&_mi_page_empty;       // caught in `mi_free`
-  
-  mi_assert_internal(_mi_ptr_page(NULL)==&_mi_page_empty);
+  mi_assert_internal(_mi_ptr_page(NULL)==NULL);
   return true;
 }
 
diff --git a/src/stats.c b/src/stats.c
index 7a045386..3b53d5a9 100644
--- a/src/stats.c
+++ b/src/stats.c
@@ -227,12 +227,14 @@ static void mi_stat_peak_print(const mi_stat_count_t* stat, const char* msg, int
   _mi_fprintf(out, arg, "\n");
 }
 
+#if MI_STAT>1
 static void mi_stat_total_print(const mi_stat_count_t* stat, const char* msg, int64_t unit, mi_output_fun* out, void* arg) {
   _mi_fprintf(out, arg, "%10s:", msg);
   _mi_fprintf(out, arg, "%12s", " ");  // no peak
   mi_print_amount(stat->total, unit, out, arg);
   _mi_fprintf(out, arg, "\n");
 }
+#endif
 
 static void mi_stat_counter_print(const mi_stat_counter_t* stat, const char* msg, mi_output_fun* out, void* arg ) {
   _mi_fprintf(out, arg, "%10s:", msg);