Merge branch 'master' into ncm_packet_filter

41 files changed, 270 insertions, 199 deletions

diff --git a/.gitignore b/.gitignore
index 162f9a019..b833191f8 100644
--- a/.gitignore
+++ b/.gitignore
@@ -55,3 +55,4 @@ Release
 BrowseInfo
 .cmake_build
 README_processed.rst
+.worktrees
diff --git a/README.rst b/README.rst
index 6be9d4873..04998abaa 100644
--- a/README.rst
+++ b/README.rst
@@ -157,6 +157,8 @@ Supported CPUs
 | Espressif    | S2, S3, H4                  | ✔      | ✔    | ✖         | dwc2                   |                    |
 | ESP32        +-----------------------------+--------+------+-----------+------------------------+--------------------+
 |              | P4                          | ✔      | ✔    | ✔         | dwc2                   |                    |
+|              +-----------------------------+--------+------+-----------+------------------------+--------------------+
+|              | S31                         | ✔      | ✔    | ✔         | dwc2                   |                    |
 +--------------+-----------------------------+--------+------+-----------+------------------------+--------------------+
 | GigaDevice   | GD32VF103                   | ✔      |      | ✖         | dwc2                   |                    |
 +--------------+-----------------------------+--------+------+-----------+------------------------+--------------------+
diff --git a/examples/device/audio_test_multi_rate/src/main.c b/examples/device/audio_test_multi_rate/src/main.c
index 952176997..a86beb415 100644
--- a/examples/device/audio_test_multi_rate/src/main.c
+++ b/examples/device/audio_test_multi_rate/src/main.c
@@ -532,7 +532,7 @@ static bool audio20_get_req_entity(uint8_t rhport, tusb_control_request_t const
 bool tud_audio_set_itf_cb(uint8_t rhport, tusb_control_request_t const *p_request) {
   (void) rhport;
   //uint8_t const itf = tu_u16_low(tu_le16toh(p_request->wIndex));
-  uint8_t const alt = tu_u16_low(tu_le16toh(p_request->wValue));
+  uint8_t const alt = tu_u16_low(p_request->wValue);
 
   // Clear buffer when streaming format is changed
   if (alt != 0) {
diff --git a/examples/device/cdc_uac2/src/uac2_app.c b/examples/device/cdc_uac2/src/uac2_app.c
index 7760c402b..6e9d1d9e3 100644
--- a/examples/device/cdc_uac2/src/uac2_app.c
+++ b/examples/device/cdc_uac2/src/uac2_app.c
@@ -263,8 +263,8 @@ bool tud_audio_set_itf_close_ep_cb(uint8_t rhport, tusb_control_request_t const
 {
   (void)rhport;
 
-  uint8_t const itf = tu_u16_low(tu_le16toh(p_request->wIndex));
-  uint8_t const alt = tu_u16_low(tu_le16toh(p_request->wValue));
+  uint8_t const itf = tu_u16_low(p_request->wIndex);
+  uint8_t const alt = tu_u16_low(p_request->wValue);
 
   if (ITF_NUM_AUDIO_STREAMING_SPK == itf && alt == 0) {
     // Audio streaming stop
@@ -277,8 +277,8 @@ bool tud_audio_set_itf_close_ep_cb(uint8_t rhport, tusb_control_request_t const
 bool tud_audio_set_itf_cb(uint8_t rhport, tusb_control_request_t const * p_request)
 {
   (void)rhport;
-  uint8_t const itf = tu_u16_low(tu_le16toh(p_request->wIndex));
-  uint8_t const alt = tu_u16_low(tu_le16toh(p_request->wValue));
+  uint8_t const itf = tu_u16_low(p_request->wIndex);
+  uint8_t const alt = tu_u16_low(p_request->wValue);
 
   TU_LOG2("Set interface %d alt %d\r\n", itf, alt);
   if (ITF_NUM_AUDIO_STREAMING_SPK == itf && alt != 0) {
diff --git a/examples/device/uac2_headset/src/main.c b/examples/device/uac2_headset/src/main.c
index 0ea63d8f7..779e927bc 100644
--- a/examples/device/uac2_headset/src/main.c
+++ b/examples/device/uac2_headset/src/main.c
@@ -522,8 +522,8 @@ bool tud_audio_set_req_entity_cb(uint8_t rhport, tusb_control_request_t const *p
 bool tud_audio_set_itf_close_ep_cb(uint8_t rhport, tusb_control_request_t const *p_request) {
   (void) rhport;
 
-  uint8_t const itf = tu_u16_low(tu_le16toh(p_request->wIndex));
-  uint8_t const alt = tu_u16_low(tu_le16toh(p_request->wValue));
+  uint8_t const itf = tu_u16_low(p_request->wIndex);
+  uint8_t const alt = tu_u16_low(p_request->wValue);
 
   if (ITF_NUM_AUDIO_STREAMING_SPK == itf && alt == 0) {
     blink_interval_ms = BLINK_MOUNTED;
@@ -534,8 +534,8 @@ bool tud_audio_set_itf_close_ep_cb(uint8_t rhport, tusb_control_request_t const
 
 bool tud_audio_set_itf_cb(uint8_t rhport, tusb_control_request_t const *p_request) {
   (void) rhport;
-  uint8_t const itf = tu_u16_low(tu_le16toh(p_request->wIndex));
-  uint8_t const alt = tu_u16_low(tu_le16toh(p_request->wValue));
+  uint8_t const itf = tu_u16_low(p_request->wIndex);
+  uint8_t const alt = tu_u16_low(p_request->wValue);
 
   TU_LOG2("Set interface %d alt %d\r\n", itf, alt);
   if (ITF_NUM_AUDIO_STREAMING_SPK == itf && alt != 0) {
diff --git a/examples/device/uac2_speaker_fb/src/main.c b/examples/device/uac2_speaker_fb/src/main.c
index c3e97bb28..402642162 100644
--- a/examples/device/uac2_speaker_fb/src/main.c
+++ b/examples/device/uac2_speaker_fb/src/main.c
@@ -457,8 +457,8 @@ static bool audio20_set_req_entity(tusb_control_request_t const *p_request, uint
 
 bool tud_audio_set_itf_cb(uint8_t rhport, tusb_control_request_t const *p_request) {
   (void) rhport;
-  uint8_t const itf = tu_u16_low(tu_le16toh(p_request->wIndex));
-  uint8_t const alt = tu_u16_low(tu_le16toh(p_request->wValue));
+  uint8_t const itf = tu_u16_low(p_request->wIndex);
+  uint8_t const alt = tu_u16_low(p_request->wValue);
 
   TU_LOG2("Set interface %d alt %d\r\n", itf, alt);
   if (ITF_NUM_AUDIO_STREAMING == itf && alt != 0)
@@ -531,8 +531,8 @@ bool tud_audio_get_req_entity_cb(uint8_t rhport, tusb_control_request_t const *p
 bool tud_audio_set_itf_close_ep_cb(uint8_t rhport, tusb_control_request_t const *p_request) {
   (void) rhport;
 
-  uint8_t const itf = tu_u16_low(tu_le16toh(p_request->wIndex));
-  uint8_t const alt = tu_u16_low(tu_le16toh(p_request->wValue));
+  uint8_t const itf = tu_u16_low(p_request->wIndex);
+  uint8_t const alt = tu_u16_low(p_request->wValue);
 
   if (ITF_NUM_AUDIO_STREAMING == itf && alt == 0) {
     blink_interval_ms = BLINK_MOUNTED;
diff --git a/hw/bsp/board.c b/hw/bsp/board.c
index ae58bb5fc..65b44e5f2 100644
--- a/hw/bsp/board.c
+++ b/hw/bsp/board.c
@@ -51,12 +51,12 @@ int sys_read(int fhdl, char *buf, size_t count) TU_ATTR_USED;
 
 int sys_write(int fhdl, const char *buf, size_t count) {
   (void) fhdl;
-  return (int) SEGGER_RTT_Write(0, buf, (int) count);
+  return (int) SEGGER_RTT_Write(0, buf, (unsigned) count);
 }
 
 int sys_read(int fhdl, char *buf, size_t count) {
   (void) fhdl;
-  int rd = (int) SEGGER_RTT_Read(0, buf, count);
+  int rd = (int) SEGGER_RTT_Read(0, buf, (unsigned) count);
   return (rd > 0) ? rd : -1;
 }
 #endif
diff --git a/hw/bsp/espressif/boards/family.c b/hw/bsp/espressif/boards/family.c
index 48b1253b6..4b28cecf2 100644
--- a/hw/bsp/espressif/boards/family.c
+++ b/hw/bsp/espressif/boards/family.c
@@ -49,7 +49,7 @@ static led_strip_handle_t led_strip;
 static void max3421_init(void);
 #endif
 
-#if TU_CHECK_MCU(OPT_MCU_ESP32S2, OPT_MCU_ESP32S3, OPT_MCU_ESP32H4, OPT_MCU_ESP32P4)
+#if TU_CHECK_MCU(OPT_MCU_ESP32S2, OPT_MCU_ESP32S3, OPT_MCU_ESP32H4, OPT_MCU_ESP32P4, OPT_MCU_ESP32S31)
 static bool usb_init(uint8_t rhport, bool is_host);
 #endif
 
@@ -111,10 +111,6 @@ void board_init(void) {
 #endif
 }
 
-#if TU_CHECK_MCU(OPT_MCU_ESP32S2, OPT_MCU_ESP32S3, OPT_MCU_ESP32H4)
-
-#endif
-
 //--------------------------------------------------------------------+
 // Board porting API
 //--------------------------------------------------------------------+
@@ -178,7 +174,7 @@ void board_reset_to_bootloader(void) {
 // PHY Init
 //--------------------------------------------------------------------
 
-#if TU_CHECK_MCU(OPT_MCU_ESP32S2, OPT_MCU_ESP32S3, OPT_MCU_ESP32H4, OPT_MCU_ESP32P4)
+#if TU_CHECK_MCU(OPT_MCU_ESP32S2, OPT_MCU_ESP32S3, OPT_MCU_ESP32H4, OPT_MCU_ESP32P4, OPT_MCU_ESP32S31)
 #if ESP_IDF_VERSION >= ESP_IDF_VERSION_VAL(5, 3, 0)
 
 #include "esp_private/usb_phy.h"
diff --git a/hw/bsp/espressif/family.cmake b/hw/bsp/espressif/family.cmake
index ca9eadaf6..30d5a6ac9 100644
--- a/hw/bsp/espressif/family.cmake
+++ b/hw/bsp/espressif/family.cmake
@@ -2,7 +2,10 @@
 include("${CMAKE_CURRENT_LIST_DIR}/boards/${BOARD}/board.cmake")
 string(TOUPPER ${IDF_TARGET} FAMILY_MCUS)
 
-# Device port default to Port1 for P4 (highspeed), Port0 for others (fullspeed)
+# Device/host port defaults:
+# - ESP32-P4 uses Port1 (highspeed)
+# - ESP32-S31 uses Port0 (highspeed)
+# - Other targets use Port0 and derive the default speed from RHPORT_SPEED
 set(RHPORT_SPEED OPT_MODE_FULL_SPEED OPT_MODE_HIGH_SPEED)
 
 if (NOT DEFINED RHPORT_DEVICE)
@@ -22,10 +25,18 @@ if (NOT DEFINED RHPORT_HOST)
 endif()
 
 if (NOT DEFINED RHPORT_DEVICE_SPEED)
-  list(GET RHPORT_SPEED ${RHPORT_DEVICE} RHPORT_DEVICE_SPEED)
+  if (IDF_TARGET STREQUAL "esp32s31")
+    set(RHPORT_DEVICE_SPEED OPT_MODE_HIGH_SPEED)
+  else ()
+    list(GET RHPORT_SPEED ${RHPORT_DEVICE} RHPORT_DEVICE_SPEED)
+  endif ()
 endif ()
 if (NOT DEFINED RHPORT_HOST_SPEED)
-  list(GET RHPORT_SPEED ${RHPORT_HOST} RHPORT_HOST_SPEED)
+  if (IDF_TARGET STREQUAL "esp32s31")
+    set(RHPORT_HOST_SPEED OPT_MODE_HIGH_SPEED)
+  else ()
+    list(GET RHPORT_SPEED ${RHPORT_HOST} RHPORT_HOST_SPEED)
+  endif ()
 endif ()
 
 # Add example src and bsp directories
diff --git a/hw/bsp/mm32/family.c b/hw/bsp/mm32/family.c
index 651c9496e..14a17f6c5 100644
--- a/hw/bsp/mm32/family.c
+++ b/hw/bsp/mm32/family.c
@@ -59,11 +59,11 @@ void OTG_FS_IRQHandler(void) {
 void USB_DeviceClockInit(void) {
   /* Select USBCLK source */
   //  RCC_USBCLKConfig(RCC_USBCLKSource_PLLCLK_Div1);
-  RCC->CFGR &= ~(0x3 << 22);
-  RCC->CFGR |= (0x1 << 22);
+  RCC->CFGR &= ~(0x3U << 22);
+  RCC->CFGR |= (0x1U << 22);
 
   /* Enable USB clock */
-  RCC->AHB2ENR |= 0x1 << 7;
+  RCC->AHB2ENR |= 0x1U << 7;
 }
 
 void board_init(void) {
diff --git a/hw/bsp/samd11/family.c b/hw/bsp/samd11/family.c
index 0c987b85a..0864aa3dd 100644
--- a/hw/bsp/samd11/family.c
+++ b/hw/bsp/samd11/family.c
@@ -34,6 +34,8 @@
 #ifdef __GNUC__
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wcast-qual"
+#pragma GCC diagnostic ignored "-Wconversion"
+#pragma GCC diagnostic ignored "-Wsign-conversion"
 #endif
 
 #include "hal/include/hal_gpio.h"
@@ -66,7 +68,7 @@ void USB_Handler(void)
 //--------------------------------------------------------------------+
 
 /* Referenced GCLKs, should be initialized firstly */
-#define _GCLK_INIT_1ST (1 << 0 | 1 << 1)
+#define _GCLK_INIT_1ST (1u << 0 | 1u << 1)
 
 /* Not referenced GCLKs, initialized last */
 #define _GCLK_INIT_LAST (~_GCLK_INIT_1ST)
diff --git a/hw/bsp/samd2x_l2x/family.c b/hw/bsp/samd2x_l2x/family.c
index 11b6343cd..306ea2308 100644
--- a/hw/bsp/samd2x_l2x/family.c
+++ b/hw/bsp/samd2x_l2x/family.c
@@ -79,7 +79,7 @@
 
 #ifdef SAMD21_FAMILY
 /* Referenced GCLKs, should be initialized firstly */
-#define _GCLK_INIT_1ST (1 << 0 | 1 << 1)
+#define _GCLK_INIT_1ST (1u << 0 | 1u << 1)
 /* Not referenced GCLKs, initialized last */
 #define _GCLK_INIT_LAST (~_GCLK_INIT_1ST)
 #endif
diff --git a/hw/bsp/samg/family.c b/hw/bsp/samg/family.c
index 519068986..9edc43b51 100644
--- a/hw/bsp/samg/family.c
+++ b/hw/bsp/samg/family.c
@@ -27,15 +27,17 @@
    manufacturer: Microchip
 */
 
-#include "sam.h"
 
 // Suppress warning caused by mcu driver
 #ifdef __GNUC__
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wcast-qual"
 #pragma GCC diagnostic ignored "-Wredundant-decls"
+#pragma GCC diagnostic ignored "-Wconversion"
+#pragma GCC diagnostic ignored "-Wsign-conversion"
 #endif
 
+#include "sam.h"
 #include "peripheral_clk_config.h"
 #include "hal/include/hal_init.h"
 #include "hal/include/hpl_usart_sync.h"
diff --git a/hw/bsp/tm4c/family.c b/hw/bsp/tm4c/family.c
index 6988a264e..fe4737e50 100644
--- a/hw/bsp/tm4c/family.c
+++ b/hw/bsp/tm4c/family.c
@@ -36,7 +36,7 @@ static void board_uart_init(void) {
   // BAUDRATE = 115200, with SystemCoreClock = 50 Mhz refer manual for calculation
   //  - BRDI = SystemCoreClock / (16* baud)
   //  - BRDF = int(fraction*64 + 0.5)
-  UART0->CTL &= ~(1 << 0);                     // Disable UART0 by clearing UARTEN bit in the UARTCTL register
+  UART0->CTL &= ~(1U << 0);                     // Disable UART0 by clearing UARTEN bit in the UARTCTL register
   UART0->IBRD = 27;                            // Write the integer portion of the BRD to the UARTIRD register
   UART0->FBRD = 8;                             // Write the fractional portion of the BRD to the UARTFBRD registerer
 
diff --git a/src/class/mtp/mtp_device.c b/src/class/mtp/mtp_device.c
index 59096e476..0da984f4a 100644
--- a/src/class/mtp/mtp_device.c
+++ b/src/class/mtp/mtp_device.c
@@ -321,7 +321,7 @@ bool mtpd_control_xfer_cb(uint8_t rhport, uint8_t stage, tusb_control_request_t
     .session_id = p_mtp->session_id,
     .request = request,
     .buf = p_mtp->control_buf,
-    .bufsize = tu_le16toh(request->wLength),
+    .bufsize = request->wLength,
   };
 
   switch (request->bRequest) {
diff --git a/src/common/tusb_compiler.h b/src/common/tusb_compiler.h
index f20834cea..a8971c3df 100644
--- a/src/common/tusb_compiler.h
+++ b/src/common/tusb_compiler.h
@@ -66,6 +66,9 @@
 #define TU_LITTLE_ENDIAN (0x12u)
 #define TU_BIG_ENDIAN (0x21u)
 
+#define TU_BITFIELD_LE (0x34u)
+#define TU_BITFIELD_BE (0x43u)
+
 /*------------------------------------------------------------------*/
 /* Count number of arguments of __VA_ARGS__
  * - reference www.stackoverflow.com/questions/2124339/c-preprocessor-va-args-number-of-arguments
@@ -167,8 +170,10 @@
 // For TI ARM compiler, __BYTE_ORDER__ is not defined for MSP430 but still LE
   #if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ || defined(__MSP430__)
     #define TU_BYTE_ORDER TU_LITTLE_ENDIAN
+    #define TU_BITFIELD_ORDER TU_BITFIELD_LE
   #else
     #define TU_BYTE_ORDER TU_BIG_ENDIAN
+    #define TU_BITFIELD_ORDER TU_BITFIELD_BE
   #endif
 
   // Unfortunately XC16 doesn't provide builtins for 32bit endian conversion
@@ -212,8 +217,10 @@
   // Endian conversion use well-known host to network (big endian) naming
   #if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
     #define TU_BYTE_ORDER TU_LITTLE_ENDIAN
+    #define TU_BITFIELD_ORDER TU_BITFIELD_LE
   #else
     #define TU_BYTE_ORDER TU_BIG_ENDIAN
+    #define TU_BITFIELD_ORDER TU_BITFIELD_BE
   #endif
 
   #define TU_BSWAP16(u16) (__iar_builtin_REV16(u16))
@@ -239,8 +246,10 @@
   // Endian conversion use well-known host to network (big endian) naming
   #if defined(__LIT)
     #define TU_BYTE_ORDER TU_LITTLE_ENDIAN
+    #define TU_BITFIELD_ORDER TU_BITFIELD_LE
   #else
     #define TU_BYTE_ORDER TU_BIG_ENDIAN
+    #define TU_BITFIELD_ORDER TU_BITFIELD_BE
   #endif
 
   #define TU_BSWAP16(u16) ((unsigned short)_builtin_revw((unsigned long)u16))
diff --git a/src/common/tusb_fifo.c b/src/common/tusb_fifo.c
index 06d25d131..a8ac99fd2 100644
--- a/src/common/tusb_fifo.c
+++ b/src/common/tusb_fifo.c
@@ -281,14 +281,14 @@ static void hwff_push_n(const tu_fifo_t *f, const void *app_buf, uint16_t n, uin
     // Write full words to the linear part of the buffer
     const uint8_t  data_stride = access_mode->data_stride;
     const uint32_t odd_mask    = data_stride - 1;
-    uint16_t       lin_even    = lin_bytes & ~odd_mask;
+    uint16_t       lin_even    = (uint16_t)(lin_bytes & ~odd_mask);
     tu_hwfifo_read(hwfifo, ff_buf, lin_even, access_mode);
     HWFIFO_ADDR_NEXT_N(hwfifo, const, lin_even * HWFIFO_ADDR_DATA_RATIO);
     ff_buf += lin_even;
 
     // There could be an odd 1 byte (16bit) or 1-3 bytes (32bit) before the wrap-around boundary
     // combine it with the wrapped part to form a full word for data stride
-    const uint8_t lin_odd = lin_bytes & odd_mask;
+    const uint8_t lin_odd = (uint8_t)(lin_bytes & odd_mask);
     if (lin_odd > 0) {
       const uint8_t wrap_odd = (uint8_t)tu_min16(wrap_bytes, data_stride - lin_odd);
       uint8_t       buf_temp[4];
@@ -338,13 +338,13 @@ static void hwff_pull_n(const tu_fifo_t *f, void *app_buf, uint16_t n, uint16_t
     // Read full words from linear part
     const uint8_t  data_stride = access_mode->data_stride;
     const uint32_t odd_mask    = data_stride - 1;
-    uint16_t       lin_even    = lin_bytes & ~odd_mask;
+    uint16_t       lin_even    = (uint16_t)(lin_bytes & ~odd_mask);
     tu_hwfifo_write(hwfifo, ff_buf, lin_even, access_mode);
     HWFIFO_ADDR_NEXT_N(hwfifo, , lin_even * HWFIFO_ADDR_DATA_RATIO);
     ff_buf += lin_even;
 
     // There could be odd 1 byte (16bit) or 1-3 bytes (32bit) before the wrap-around boundary
-    const uint8_t lin_odd = lin_bytes & odd_mask;
+    const uint8_t lin_odd = (uint8_t)(lin_bytes & odd_mask);
     if (lin_odd > 0) {
       const uint8_t wrap_odd = (uint8_t)tu_min16(wrap_bytes, data_stride - lin_odd);
 
diff --git a/src/common/tusb_mcu.h b/src/common/tusb_mcu.h
index 77a0bbf1d..a17c76a3b 100644
--- a/src/common/tusb_mcu.h
+++ b/src/common/tusb_mcu.h
@@ -455,7 +455,7 @@
   #define CFG_TUSB_OS_INC_PATH_DEFAULT freertos/
   // clang-format on
 
-  #if CFG_TUSB_MCU == OPT_MCU_ESP32S3
+  #if CFG_TUSB_MCU == OPT_MCU_ESP32S3 || CFG_TUSB_MCU == OPT_MCU_ESP32H4
     #define TUP_MCU_MULTIPLE_CORE 1
   #endif
 
@@ -476,6 +476,22 @@
   #define CFG_TUH_MEM_DCACHE_ENABLE_DEFAULT     CFG_TUH_DWC2_DMA_ENABLE
   #define CFG_TUSB_MEM_DCACHE_LINE_SIZE_DEFAULT 64
 
+#elif TU_CHECK_MCU(OPT_MCU_ESP32S31)
+  #define TUP_USBIP_DWC2
+  #define TUP_USBIP_DWC2_ESP32
+  #define TUP_RHPORT_HIGHSPEED                  1
+  #define TUP_DCD_ENDPOINT_MAX                  16
+
+  // clang-format off
+  #define CFG_TUSB_OS_INC_PATH_DEFAULT          freertos/
+  // clang-format on
+
+  #define TUP_MCU_MULTIPLE_CORE                 1
+
+  // Disable slave if DMA is enabled
+  #define CFG_TUD_DWC2_SLAVE_ENABLE_DEFAULT     !CFG_TUD_DWC2_DMA_ENABLE
+  #define CFG_TUH_DWC2_SLAVE_ENABLE_DEFAULT     !CFG_TUH_DWC2_DMA_ENABLE
+
 #elif TU_CHECK_MCU(OPT_MCU_ESP32, OPT_MCU_ESP32C2, OPT_MCU_ESP32C3, OPT_MCU_ESP32C5, OPT_MCU_ESP32C6, \
                    OPT_MCU_ESP32C61, OPT_MCU_ESP32H2)
   #if (CFG_TUD_ENABLED || !(defined(CFG_TUH_MAX3421) && CFG_TUH_MAX3421))
diff --git a/src/common/tusb_types.h b/src/common/tusb_types.h
index a18f9feb7..8ae1407f8 100644
--- a/src/common/tusb_types.h
+++ b/src/common/tusb_types.h
@@ -100,8 +100,8 @@ typedef enum {
 } tusb_xfer_type_t;
 
 typedef enum {
-  TUSB_DIR_OUT = 0,
-  TUSB_DIR_IN  = 1,
+  TUSB_DIR_OUT = 0u,
+  TUSB_DIR_IN  = 1u,
 
   TUSB_EPNUM_MASK = 0x0F,
   TUSB_DIR_IN_MASK = 0x80
@@ -409,10 +409,19 @@ typedef struct TU_ATTR_PACKED {
   uint8_t  bEndpointAddress ; // The address of the endpoint
 
   struct TU_ATTR_PACKED {
+#if (TU_BITFIELD_ORDER == TU_BITFIELD_LE)
     uint8_t xfer  : 2;        // Control, ISO, Bulk, Interrupt
     uint8_t sync  : 2;        // None, Asynchronous, Adaptive, Synchronous
     uint8_t usage : 2;        // Data, Feedback, Implicit feedback
     uint8_t       : 2;
+#elif (TU_BITFIELD_ORDER == TU_BITFIELD_BE)
+    uint8_t       : 2;
+    uint8_t usage : 2;
+    uint8_t sync  : 2;
+    uint8_t xfer  : 2;
+#else
+  #error "Please define TU_BITFIELD_ORDER as TU_BITFIELD_LE or TU_BITFIELD_BE"
+#endif
   } bmAttributes;
 
   uint16_t wMaxPacketSize   ; // Bit 10..0 : max packet size, bit 12..11 additional transaction per highspeed micro-frame
@@ -522,9 +531,17 @@ typedef struct TU_ATTR_PACKED {
 typedef struct TU_ATTR_PACKED {
   union {
     struct TU_ATTR_PACKED {
+#if (TU_BITFIELD_ORDER == TU_BITFIELD_LE)
       uint8_t recipient :  5; ///< Recipient type tusb_request_recipient_t.
       uint8_t type      :  2; ///< Request type tusb_request_type_t.
       uint8_t direction :  1; ///< Direction type. tusb_dir_t
+#elif (TU_BITFIELD_ORDER == TU_BITFIELD_BE)
+      uint8_t direction :  1; ///< Direction type. tusb_dir_t
+      uint8_t type      :  2; ///< Request type tusb_request_type_t.
+      uint8_t recipient :  5; ///< Recipient type tusb_request_recipient_t.
+#else
+  #error "Please define TU_BITFIELD_ORDER as TU_BITFIELD_LE or TU_BITFIELD_BE"
+#endif
     } bmRequestType_bit;
 
     uint8_t bmRequestType;
diff --git a/src/device/dcd.h b/src/device/dcd.h
index 850c37bc2..f861eb258 100644
--- a/src/device/dcd.h
+++ b/src/device/dcd.h
@@ -219,6 +219,11 @@ TU_ATTR_ALWAYS_INLINE static inline void dcd_event_setup_received(uint8_t rhport
   event.rhport = rhport;
   event.event_id = DCD_EVENT_SETUP_RECEIVED;
   (void) memcpy(&event.setup_received, setup, sizeof(tusb_control_request_t));
+  // USB wire format is little-endian. Convert multi-byte fields to host byte order
+  // so the stack always sees correct values regardless of CPU endianness.
+  event.setup_received.wValue  = tu_le16toh(event.setup_received.wValue);
+  event.setup_received.wIndex  = tu_le16toh(event.setup_received.wIndex);
+  event.setup_received.wLength = tu_le16toh(event.setup_received.wLength);
   dcd_event_handler(&event, in_isr);
 }
 
diff --git a/src/device/usbd.c b/src/device/usbd.c
index 3c14175f6..da0ffb4c6 100644
--- a/src/device/usbd.c
+++ b/src/device/usbd.c
@@ -1212,7 +1212,7 @@ static bool process_get_descriptor(uint8_t rhport, tusb_control_request_t const
       TU_LOG_USBD(" String[%u]\r\n", desc_index);
 
       // String Descriptor always uses the desc set from user
-      uint8_t const* desc_str = (uint8_t const*) tud_descriptor_string_cb(desc_index, tu_le16toh(p_request->wIndex));
+      uint8_t const* desc_str = (uint8_t const*) tud_descriptor_string_cb(desc_index, p_request->wIndex);
       TU_VERIFY(desc_str);
 
       // first byte of descriptor is its size
diff --git a/src/portable/chipidea/ci_fs/dcd_ci_fs.c b/src/portable/chipidea/ci_fs/dcd_ci_fs.c
index 312a98299..62df1a6d5 100644
--- a/src/portable/chipidea/ci_fs/dcd_ci_fs.c
+++ b/src/portable/chipidea/ci_fs/dcd_ci_fs.c
@@ -360,7 +360,7 @@ static bool edpt_open(uint8_t rhport, uint8_t ep_addr, uint16_t max_packet_size,
   unsigned val = USB_ENDPT_EPCTLDIS_MASK;
   val |= (xfer != TUSB_XFER_ISOCHRONOUS) ? USB_ENDPT_EPHSHK_MASK : 0;
   val |= dir ? USB_ENDPT_EPTXEN_MASK : USB_ENDPT_EPRXEN_MASK;
-  CI_REG->EP[epn].CTL |= val;
+  CI_REG->EP[epn].CTL |= (uint8_t)val;
 
   if (xfer != TUSB_XFER_ISOCHRONOUS) {
     bd[odd].dts      = 1;
@@ -434,11 +434,11 @@ bool dcd_edpt_xfer(uint8_t rhport, uint8_t ep_addr, uint8_t * buffer, uint16_t t
     buffer_descriptor_t *next = ep->odd ? bd - 1: bd + 1;
     /* When total_bytes is greater than the max packet size,
      * it prepares to the next transfer to avoid NAK in advance. */
-    next->bc   = total_bytes >= 2 * mps ? mps: total_bytes - mps;
+    next->bc   = (total_bytes >= 2 * mps) ? mps : (total_bytes - mps);
     next->addr = buffer + mps;
     next->own  = 1;
   }
-  bd->bc   = total_bytes >= mps ? mps: total_bytes;
+  bd->bc   = (total_bytes >= mps ? mps : total_bytes);
   bd->addr = buffer;
   __DSB();
   bd->own  = 1; /* This bit must be set last */
@@ -506,16 +506,16 @@ void dcd_edpt_clear_stall(uint8_t rhport, uint8_t ep_addr)
 //--------------------------------------------------------------------+
 void dcd_int_handler(uint8_t rhport)
 {
-  uint32_t is  = CI_REG->INT_STAT;
-  uint32_t msk = CI_REG->INT_EN;
+  uint8_t is  = CI_REG->INT_STAT;
+  uint8_t msk = CI_REG->INT_EN;
 
   // clear non-enabled interrupts
-  CI_REG->INT_STAT = is & ~msk;
+  CI_REG->INT_STAT = (uint8_t)(is & ~msk);
   is &= msk;
 
   if (is & USB_ISTAT_ERROR_MASK) {
     /* TODO: */
-    uint32_t es = CI_REG->ERR_STAT;
+    uint8_t es = CI_REG->ERR_STAT;
     CI_REG->ERR_STAT = es;
     CI_REG->INT_STAT = is; /* discard any pending events */
   }
diff --git a/src/portable/dialog/da146xx/dcd_da146xx.c b/src/portable/dialog/da146xx/dcd_da146xx.c
index a283c8362..7d90b1f94 100644
--- a/src/portable/dialog/da146xx/dcd_da146xx.c
+++ b/src/portable/dialog/da146xx/dcd_da146xx.c
@@ -148,8 +148,8 @@ typedef struct
 #ifndef TU_DA146XX_DMA_RX_CHANNEL
 #define TU_DA146XX_DMA_RX_CHANNEL 6
 #endif
-#define DA146XX_DMA_USB_MUX       (0x6 << (TU_DA146XX_DMA_RX_CHANNEL * 2))
-#define DA146XX_DMA_USB_MUX_MASK  (0xF << (TU_DA146XX_DMA_RX_CHANNEL * 2))
+#define DA146XX_DMA_USB_MUX       (0x6u << (TU_DA146XX_DMA_RX_CHANNEL * 2))
+#define DA146XX_DMA_USB_MUX_MASK  (0xFu << (TU_DA146XX_DMA_RX_CHANNEL * 2))
 
 typedef struct
 {
@@ -311,12 +311,12 @@ static void fill_tx_fifo(xfer_ctl_t * xfer)
       // Max packet size is set to value greater then FIFO. Enable fifo level warning
       // to handle larger packets.
       regs->txc |= (3 << USB_USB_TXC1_REG_USB_TFWL_Pos);
-      USB->USB_FWMSK_REG |= 1 << (epnum - 1 + USB_USB_FWMSK_REG_USB_M_TXWARN31_Pos);
+      USB->USB_FWMSK_REG |= 1u << (epnum - 1 + USB_USB_FWMSK_REG_USB_M_TXWARN31_Pos);
     }
     else
     {
       regs->txc &= ~USB_USB_TXC1_REG_USB_TFWL_Msk;
-      USB->USB_FWMSK_REG &= ~(1 << (epnum - 1 + USB_USB_FWMSK_REG_USB_M_TXWARN31_Pos));
+      USB->USB_FWMSK_REG &= ~(1u << (epnum - 1 + USB_USB_FWMSK_REG_USB_M_TXWARN31_Pos));
       // Whole packet already in fifo, no need to refill it later.  Mark last.
       regs->txc |= USB_USB_TXC1_REG_USB_LAST_Msk;
     }
@@ -371,14 +371,14 @@ static void start_rx_packet(xfer_ctl_t *xfer)
       // For endpoint size greater than FIFO size enable FIFO level warning interrupt
       // when FIFO has less than 17 bytes free.
       regs->rxc |= USB_USB_RXC1_REG_USB_RFWL_Msk;
-      USB->USB_FWMSK_REG |= 1 << (epnum - 1 + USB_USB_FWMSK_REG_USB_M_RXWARN31_Pos);
+      USB->USB_FWMSK_REG |= 1u << (epnum - 1 + USB_USB_FWMSK_REG_USB_M_RXWARN31_Pos);
     }
   }
   else if (epnum != 0)
   {
     // If max_packet_size would fit in FIFO no need for FIFO level warning interrupt.
     regs->rxc &= ~USB_USB_RXC1_REG_USB_RFWL_Msk;
-    USB->USB_FWMSK_REG &= ~(1 << (epnum - 1 + USB_USB_FWMSK_REG_USB_M_RXWARN31_Pos));
+    USB->USB_FWMSK_REG &= ~(1u << (epnum - 1 + USB_USB_FWMSK_REG_USB_M_RXWARN31_Pos));
   }
   regs->rxc |= USB_USB_RXC1_REG_USB_RX_EN_Msk;
 }
@@ -388,7 +388,7 @@ static void start_tx_dma(void *src, volatile void *dst, uint16_t size)
   // Setup SRC and DST registers
   TX_DMA_REGS->DMAx_A_START_REG = (uint32_t)src;
   TX_DMA_REGS->DMAx_B_START_REG = (uint32_t)dst;
-  // Interrupt not needed
+  // Interrupt is not needed
   TX_DMA_REGS->DMAx_INT_REG = size;
   TX_DMA_REGS->DMAx_LEN_REG = size - 1;
   TX_DMA_REGS->DMAx_CTRL_REG = TX_DMA_START;
@@ -430,7 +430,9 @@ static uint16_t read_rx_fifo(xfer_ctl_t *xfer, uint16_t bytes_in_fifo)
 
   uint8_t *buf = xfer->buffer + xfer->transferred + xfer->last_packet_size;
 
-  for (int i = 0; i < receive_this_time; ++i) buf[i] = regs->rxd;
+  for (int i = 0; i < receive_this_time; ++i) {
+    buf[i] = (uint8_t)regs->rxd;
+  }
 
   xfer->last_packet_size += receive_this_time;
 
@@ -449,7 +451,9 @@ static void handle_ep0_rx(void)
   {
     xfer_ctl_t *xfer_in = XFER_CTL_BASE(0, TUSB_DIR_IN);
     // Setup packet is in
-    for (int i = 0; i < fifo_bytes; ++i) _setup_packet[i] = USB->USB_RXD0_REG;
+    for (int i = 0; i < fifo_bytes; ++i) {
+      _setup_packet[i] = (uint8_t)USB->USB_RXD0_REG;
+    }
 
     xfer->stall = 0;
     xfer->data1 = 1;
@@ -469,7 +473,7 @@ static void handle_ep0_rx(void)
     }
     else
     {
-      read_rx_fifo(xfer, fifo_bytes);
+      read_rx_fifo(xfer, (uint16_t)fifo_bytes);
       if (rxs0 & USB_USB_RXS0_REG_USB_RX_LAST_Msk)
       {
         xfer->transferred += xfer->last_packet_size;
@@ -553,7 +557,7 @@ static void handle_epx_rx_ev(uint8_t ep)
       {
         // Disable DMA and update last_packet_size with what DMA reported.
         RX_DMA_REGS->DMAx_CTRL_REG &= ~DMA_DMA0_CTRL_REG_DMA_ON_Msk;
-        xfer->last_packet_size = RX_DMA_REGS->DMAx_IDX_REG;
+        xfer->last_packet_size = (uint16_t)RX_DMA_REGS->DMAx_IDX_REG;
         // When DMA did not finished (packet was smaller then MPS), DMAx_IDX_REG holds exact number of bytes transmitted.
         // When DMA finished value in DMAx_IDX_REG is one less then actual number of transmitted bytes.
         if (xfer->last_packet_size == RX_DMA_REGS->DMAx_LEN_REG) xfer->last_packet_size++;
@@ -564,7 +568,7 @@ static void handle_epx_rx_ev(uint8_t ep)
       // FIFO maybe empty if DMA read it before or it's final iteration and function already read all that was to read.
       if (fifo_bytes > 0)
       {
-        fifo_bytes = read_rx_fifo(xfer, fifo_bytes);
+        fifo_bytes = read_rx_fifo(xfer, (uint16_t)fifo_bytes);
       }
       if (GET_BIT(rxs, USB_USB_RXS1_REG_USB_RX_LAST))
       {
@@ -624,7 +628,7 @@ static void handle_epx_tx_ev(xfer_ctl_t *xfer)
     {
       // Disable DMA and update last_packet_size with what DMA reported.
       TX_DMA_REGS->DMAx_CTRL_REG &= ~DMA_DMA1_CTRL_REG_DMA_ON_Msk;
-      xfer->last_packet_size = TX_DMA_REGS->DMAx_IDX_REG + 1;
+      xfer->last_packet_size = (uint16_t)(TX_DMA_REGS->DMAx_IDX_REG + 1);
       // Release DMA to used by other endpoints.
       _dcd.dma_ep[TUSB_DIR_IN] = 0;
     }
@@ -954,13 +958,13 @@ bool dcd_edpt_open(uint8_t rhport, tusb_desc_endpoint_t const * desc_edpt)
     if (dir == TUSB_DIR_OUT)
     {
       regs->epc_out = epnum | USB_USB_EPC1_REG_USB_EP_EN_Msk | iso_mask;
-      USB->USB_RXMSK_REG |= 0x11 << (epnum - 1);
+      USB->USB_RXMSK_REG |= 0x11u << (epnum - 1);
       REG_SET_BIT(USB_MAMSK_REG, USB_M_RX_EV);
     }
     else
     {
       regs->epc_in = epnum | USB_USB_EPC1_REG_USB_EP_EN_Msk | iso_mask;
-      USB->USB_TXMSK_REG |= 0x11 << (epnum - 1);
+      USB->USB_TXMSK_REG |= 0x11u << (epnum - 1);
       REG_SET_BIT(USB_MAMSK_REG, USB_M_TX_EV);
     }
   }
@@ -974,8 +978,8 @@ void dcd_edpt_close_all (uint8_t rhport)
 
   for (int epnum = 1; epnum < EP_MAX; ++epnum)
   {
-    dcd_edpt_close(0, epnum | TUSB_DIR_OUT);
-    dcd_edpt_close(0, epnum | TUSB_DIR_IN);
+    dcd_edpt_close(0, (uint8_t)(epnum | TUSB_DIR_OUT));
+    dcd_edpt_close(0, (uint8_t)(epnum | TUSB_DIR_IN));
   }
 }
 
@@ -1001,7 +1005,7 @@ void dcd_edpt_close(uint8_t rhport, uint8_t ep_addr)
     {
       regs->rxc = USB_USB_RXC1_REG_USB_FLUSH_Msk;
       regs->epc_out = 0;
-      USB->USB_RXMSK_REG &= ~(0x11 << (epnum - 1));
+      USB->USB_RXMSK_REG &= ~(0x11u << (epnum - 1));
       // Release DMA if needed
       if (_dcd.dma_ep[TUSB_DIR_OUT] == epnum)
       {
@@ -1013,7 +1017,7 @@ void dcd_edpt_close(uint8_t rhport, uint8_t ep_addr)
     {
       regs->txc = USB_USB_TXC1_REG_USB_FLUSH_Msk;
       regs->epc_in = 0;
-      USB->USB_TXMSK_REG &= ~(0x11 << (epnum - 1));
+      USB->USB_TXMSK_REG &= ~(0x11u << (epnum - 1));
       // Release DMA if needed
       if (_dcd.dma_ep[TUSB_DIR_IN] == epnum)
       {
diff --git a/src/portable/mentor/musb/dcd_musb.c b/src/portable/mentor/musb/dcd_musb.c
index 64f9ebacf..2f6f231ef 100644
--- a/src/portable/mentor/musb/dcd_musb.c
+++ b/src/portable/mentor/musb/dcd_musb.c
@@ -111,8 +111,8 @@ TU_ATTR_ALWAYS_INLINE static inline void hwfifo_reset(musb_regs_t* musb, unsigne
 TU_ATTR_ALWAYS_INLINE static inline bool hwfifo_config(musb_regs_t* musb, unsigned epnum, unsigned is_rx, unsigned mps,
                                                        bool double_packet) {
   (void) epnum;
-  uint8_t ffsize = hwfifo_byte2size(mps);
-  mps = 8 << ffsize; // round up to the next power of 2
+  uint8_t ffsize = hwfifo_byte2size((uint16_t)mps);
+  mps = 8u << ffsize; // round up to the next power of 2
 
   if (double_packet) {
     ffsize |= MUSB_FIFOSZ_DOUBLE_PACKET;
@@ -120,7 +120,7 @@ TU_ATTR_ALWAYS_INLINE static inline bool hwfifo_config(musb_regs_t* musb, unsign
   }
 
   TU_ASSERT(alloced_fifo_bytes + mps <= MUSB_CFG_DYNAMIC_FIFO_SIZE);
-  musb->fifo_addr[is_rx] = alloced_fifo_bytes / 8;
+  musb->fifo_addr[is_rx] = (uint16_t)(alloced_fifo_bytes / 8);
   musb->fifo_size[is_rx] = ffsize;
 
   alloced_fifo_bytes += mps;
@@ -157,12 +157,12 @@ TU_ATTR_ALWAYS_INLINE static inline bool hwfifo_config(musb_regs_t* musb, unsign
 // Flush FIFO and clear data toggle
 TU_ATTR_ALWAYS_INLINE static inline void hwfifo_flush(musb_regs_t* musb, unsigned epnum, unsigned is_rx, bool clear_dtog) {
   (void) epnum;
-  const uint8_t csrl_dtog = clear_dtog ? MUSB_CSRL_CLEAR_DATA_TOGGLE(is_rx) : 0;
+  const uint8_t csrl_dtog = clear_dtog ? (uint8_t)MUSB_CSRL_CLEAR_DATA_TOGGLE(is_rx) : 0;
   musb_ep_maxp_csr_t* maxp_csr = &musb->indexed_csr.maxp_csr[is_rx];
   // may need to flush twice for double packet
   for (unsigned i=0; i<2; i++) {
     if (maxp_csr->csrl & MUSB_CSRL_PACKET_READY(is_rx)) {
-      maxp_csr->csrl = MUSB_CSRL_FLUSH_FIFO(is_rx) | csrl_dtog;
+      maxp_csr->csrl = (uint8_t)(MUSB_CSRL_FLUSH_FIFO(is_rx) | csrl_dtog);
     }
   }
 }
@@ -180,7 +180,7 @@ static void process_setup_packet(uint8_t rhport) {
   dcd_event_setup_received(rhport, (const uint8_t*)(uintptr_t)&_dcd.setup_packet, true);
 
   const unsigned len    = _dcd.setup_packet.wLength;
-  _dcd.remaining_ctrl   = len;
+  _dcd.remaining_ctrl   = (uint16_t)len;
   const unsigned dir_in = tu_edpt_dir(_dcd.setup_packet.bmRequestType);
   /* Clear RX FIFO and reverse the transaction direction */
   if (len && dir_in) {
@@ -448,14 +448,14 @@ static void process_edpt_n(uint8_t rhport, uint_fast8_t ep_addr)
   if (dir_in) {
     // TU_LOG1(" TX CSRL%d = %x\r\n", epn, ep_csr->tx_csrl);
     if (ep_csr->tx_csrl & MUSB_TXCSRL1_STALLED) {
-      ep_csr->tx_csrl &= ~(MUSB_TXCSRL1_STALLED | MUSB_TXCSRL1_UNDRN);
+      ep_csr->tx_csrl = (uint8_t)(ep_csr->tx_csrl & ~(MUSB_TXCSRL1_STALLED | MUSB_TXCSRL1_UNDRN));
       return;
     }
     completed = handle_xfer_in(rhport, ep_addr);
   } else {
     // TU_LOG1(" RX CSRL%d = %x\r\n", epn, ep_csr->rx_csrl);
     if (ep_csr->rx_csrl & MUSB_RXCSRL1_STALLED) {
-      ep_csr->rx_csrl &= ~(MUSB_RXCSRL1_STALLED | MUSB_RXCSRL1_OVER);
+      ep_csr->rx_csrl = (uint8_t)(ep_csr->rx_csrl & ~(MUSB_RXCSRL1_STALLED | MUSB_RXCSRL1_OVER));
       return;
     }
     completed = handle_xfer_out(rhport, ep_addr);
@@ -785,7 +785,7 @@ void dcd_edpt_clear_stall(uint8_t rhport, uint8_t ep_addr)
   musb_ep_csr_t* ep_csr = get_ep_csr(musb_regs, epn);
   const uint8_t is_rx = 1 - tu_edpt_dir(ep_addr);
 
-  ep_csr->maxp_csr[is_rx].csrl = MUSB_CSRL_CLEAR_DATA_TOGGLE(is_rx);
+  ep_csr->maxp_csr[is_rx].csrl = (uint8_t)MUSB_CSRL_CLEAR_DATA_TOGGLE(is_rx);
 
   if (ie) musb_dcd_int_enable(rhport);
 }
@@ -801,8 +801,8 @@ void dcd_int_handler(uint8_t rhport) {
   musb_dcd_int_handler_enter(rhport);
 
   uint_fast8_t intr_usb = musb_regs->intr_usb; // a read will clear this interrupt status
-  uint_fast8_t intr_tx = musb_regs->intr_tx; // a read will clear this interrupt status
-  uint_fast8_t intr_rx = musb_regs->intr_rx; // a read will clear this interrupt status
+  uint_fast16_t intr_tx = musb_regs->intr_tx; // a read will clear this interrupt status
+  uint_fast16_t intr_rx = musb_regs->intr_rx; // a read will clear this interrupt status
   // TU_LOG1("D%2x T%2x R%2x\r\n", is, txis, rxis);
 
   intr_usb &= musb_regs->intr_usben; /* Clear disabled interrupts */
diff --git a/src/portable/mentor/musb/musb_type.h b/src/portable/mentor/musb/musb_type.h
index b2f6492fa..176504a2f 100644
--- a/src/portable/mentor/musb/musb_type.h
+++ b/src/portable/mentor/musb/musb_type.h
@@ -300,7 +300,7 @@ TU_VERIFY_STATIC(sizeof(musb_regs_t) == 0x350, "size is not correct");
 // Helper
 //--------------------------------------------------------------------+
 TU_ATTR_ALWAYS_INLINE static inline musb_ep_csr_t* get_ep_csr(musb_regs_t* musb_regs, unsigned epnum) {
-  musb_regs->index = epnum;
+  musb_regs->index = (uint8_t)epnum;
   return &musb_regs->indexed_csr;
 }
 
diff --git a/src/portable/microchip/samg/dcd_samg.c b/src/portable/microchip/samg/dcd_samg.c
index 4115eecc5..f8980b775 100644
--- a/src/portable/microchip/samg/dcd_samg.c
+++ b/src/portable/microchip/samg/dcd_samg.c
@@ -352,8 +352,8 @@ void dcd_edpt_clear_stall (uint8_t rhport, uint8_t ep_addr)
   csr_clear(epnum, UDP_CSR_FORCESTALL_Msk);
 
   // must also reset EP to clear data toggle
-  UDP->UDP_RST_EP |= (1 << epnum);
-  UDP->UDP_RST_EP &= ~(1 << epnum);
+  UDP->UDP_RST_EP |= (1u << epnum);
+  UDP->UDP_RST_EP &= ~(1u << epnum);
 }
 
 //--------------------------------------------------------------------+
diff --git a/src/portable/nordic/nrf5x/dcd_nrf5x.c b/src/portable/nordic/nrf5x/dcd_nrf5x.c
index 8a41c4790..6ed5fde8e 100644
--- a/src/portable/nordic/nrf5x/dcd_nrf5x.c
+++ b/src/portable/nordic/nrf5x/dcd_nrf5x.c
@@ -36,6 +36,8 @@
 #pragma GCC diagnostic ignored "-Wcast-qual"
 #pragma GCC diagnostic ignored "-Wcast-align"
 #pragma GCC diagnostic ignored "-Wunused-parameter"
+#pragma GCC diagnostic ignored "-Wconversion"
+#pragma GCC diagnostic ignored "-Wsign-conversion"
 #endif
 
 #include "nrf.h"
@@ -461,7 +463,7 @@ bool dcd_edpt_xfer(uint8_t rhport, uint8_t ep_addr, uint8_t* buffer, uint16_t to
   xfer->actual_len = 0;
 
   // Control endpoint with zero-length packet and opposite direction to 1st request byte --> status stage
-  bool const control_status = (epnum == 0 && total_bytes == 0 && dir != tu_edpt_dir(NRF_USBD->BMREQUESTTYPE));
+  bool const control_status = (epnum == 0 && total_bytes == 0 && dir != tu_edpt_dir((uint8_t)NRF_USBD->BMREQUESTTYPE));
 
   if (control_status) {
     // The nRF doesn't interrupt on status transmit so we queue up a success response.
diff --git a/src/portable/nuvoton/nuc120/dcd_nuc120.c b/src/portable/nuvoton/nuc120/dcd_nuc120.c
index d9a0e3fa8..2edb1bc7a 100644
--- a/src/portable/nuvoton/nuc120/dcd_nuc120.c
+++ b/src/portable/nuvoton/nuc120/dcd_nuc120.c
@@ -253,13 +253,13 @@ bool dcd_edpt_open(uint8_t rhport, tusb_desc_endpoint_t const * p_endpoint_desc)
 
   /* mine the data for the information we need */
   int const dir = tu_edpt_dir(p_endpoint_desc->bEndpointAddress);
-  int const size = tu_edpt_packet_size(p_endpoint_desc);
+  uint16_t const size = tu_edpt_packet_size(p_endpoint_desc);
   tusb_xfer_type_t const type = (tusb_xfer_type_t) p_endpoint_desc->bmAttributes.xfer;
   struct xfer_ctl_t *xfer = &xfer_table[ep - USBD->EP];
 
   /* allocate buffer from USB RAM */
   ep->BUFSEG = bufseg_addr;
-  bufseg_addr += size;
+  bufseg_addr += (uint32_t)size;
   TU_ASSERT(bufseg_addr <= USBD_BUF_SIZE);
 
   /* construct USB Configuration Register value and then write it */
@@ -435,7 +435,7 @@ void dcd_int_handler(uint8_t rhport)
       /* given ACK from host has happened, we can now set the address (if not already done) */
       if((USBD->FADDR != assigned_address) && (USBD->FADDR == 0)) USBD->FADDR = assigned_address;
 
-      uint16_t const available_bytes = USBD->EP[PERIPH_EP0].MXPLD;
+      uint16_t const available_bytes = (uint16_t)USBD->EP[PERIPH_EP0].MXPLD;
 
       active_ep0_xfer = (available_bytes == xfer_table[PERIPH_EP0].max_packet_size);
 
@@ -453,7 +453,7 @@ void dcd_int_handler(uint8_t rhport)
       {
         USBD->INTSTS = mask;
 
-        uint16_t const available_bytes = ep->MXPLD;
+        uint16_t const available_bytes = (uint16_t)ep->MXPLD;
         uint8_t const ep_addr = decode_ep_addr(ep);
         bool const out_ep = !(ep_addr & TUSB_DIR_IN_MASK);
 
diff --git a/src/portable/nuvoton/nuc121/dcd_nuc121.c b/src/portable/nuvoton/nuc121/dcd_nuc121.c
index 42fb58a0a..008c9df6b 100644
--- a/src/portable/nuvoton/nuc121/dcd_nuc121.c
+++ b/src/portable/nuvoton/nuc121/dcd_nuc121.c
@@ -42,6 +42,8 @@
 #ifdef __GNUC__
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wredundant-decls"
+#pragma GCC diagnostic ignored "-Wconversion"
+#pragma GCC diagnostic ignored "-Wsign-conversion"
 #endif
 
 #include "NuMicro.h"
@@ -291,7 +293,7 @@ bool dcd_edpt_open(uint8_t rhport, tusb_desc_endpoint_t const * p_endpoint_desc)
 
   /* mine the data for the information we need */
   int const dir = tu_edpt_dir(p_endpoint_desc->bEndpointAddress);
-  int const size = tu_edpt_packet_size(p_endpoint_desc);
+  uint16_t const size = tu_edpt_packet_size(p_endpoint_desc);
   tusb_xfer_type_t const type = (tusb_xfer_type_t) p_endpoint_desc->bmAttributes.xfer;
   struct xfer_ctl_t *xfer = &xfer_table[ep - USBD->EP];
 
@@ -478,7 +480,7 @@ void dcd_int_handler(uint8_t rhport)
   {
     if (status & USBD_INTSTS_EPEVT0_Msk) /* PERIPH_EP0 (EP0_IN) event: this is treated separately from the rest */
     {
-      uint16_t const available_bytes = USBD->EP[PERIPH_EP0].MXPLD;
+      uint16_t const available_bytes = (uint16_t)USBD->EP[PERIPH_EP0].MXPLD;
 
       active_ep0_xfer = (available_bytes == xfer_table[PERIPH_EP0].max_packet_size);
 
@@ -496,7 +498,7 @@ void dcd_int_handler(uint8_t rhport)
       {
         USBD->INTSTS = mask;
 
-        uint16_t const available_bytes = ep->MXPLD;
+        uint16_t const available_bytes = (uint16_t)ep->MXPLD;
         uint8_t const ep_addr = decode_ep_addr(ep);
         bool const out_ep = !(ep_addr & TUSB_DIR_IN_MASK);
 
diff --git a/src/portable/nuvoton/nuc505/dcd_nuc505.c b/src/portable/nuvoton/nuc505/dcd_nuc505.c
index ca17d6251..a0f3d4c3f 100644
--- a/src/portable/nuvoton/nuc505/dcd_nuc505.c
+++ b/src/portable/nuvoton/nuc505/dcd_nuc505.c
@@ -42,6 +42,8 @@
 #ifdef __GNUC__
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wredundant-decls"
+#pragma GCC diagnostic ignored "-Wconversion"
+#pragma GCC diagnostic ignored "-Wsign-conversion"
 #endif
 
 #include "NUC505Series.h"
@@ -338,13 +340,13 @@ bool dcd_edpt_open(uint8_t rhport, tusb_desc_endpoint_t const * p_endpoint_desc)
 
   /* mine the data for the information we need */
   int const dir = tu_edpt_dir(p_endpoint_desc->bEndpointAddress);
-  int const size = tu_edpt_packet_size(p_endpoint_desc);
+  uint16_t const size = tu_edpt_packet_size(p_endpoint_desc);
   tusb_xfer_type_t const type = p_endpoint_desc->bmAttributes.xfer;
   struct xfer_ctl_t *xfer = &xfer_table[ep - USBD->EP];
 
   /* allocate buffer from USB RAM */
   ep->EPBUFSTART = bufseg_addr;
-  bufseg_addr += size;
+  bufseg_addr += (uint32_t)size;
   ep->EPBUFEND = bufseg_addr - 1;
   TU_ASSERT(bufseg_addr <= USBD_BUF_SIZE);
 
diff --git a/src/portable/nxp/lpc17_40/dcd_lpc17_40.c b/src/portable/nxp/lpc17_40/dcd_lpc17_40.c
index 349229c8d..2840c6d5e 100644
--- a/src/portable/nxp/lpc17_40/dcd_lpc17_40.c
+++ b/src/portable/nxp/lpc17_40/dcd_lpc17_40.c
@@ -131,7 +131,7 @@ static uint8_t sie_read (uint8_t cmd_code)
 //--------------------------------------------------------------------+
 static inline uint8_t ep_addr2idx(uint8_t ep_addr)
 {
-  return 2*(ep_addr & 0x0F) + ((ep_addr & TUSB_DIR_IN_MASK) ? 1 : 0);
+  return (uint8_t)(2*(ep_addr & 0x0F) + ((ep_addr & TUSB_DIR_IN_MASK) ? 1 : 0));
 }
 
 static void set_ep_size(uint8_t ep_id, uint16_t max_packet_size)
@@ -243,7 +243,7 @@ void dcd_sof_enable(uint8_t rhport, bool en)
 //--------------------------------------------------------------------+
 static inline uint8_t byte2dword(uint8_t bytes)
 {
-  return (bytes + 3) / 4; // length in dwords
+  return (uint8_t)((bytes + 3) / 4); // length in dwords
 }
 
 static void control_ep_write(void const * buffer, uint8_t len)
diff --git a/src/portable/nxp/lpc_ip3511/dcd_lpc_ip3511.c b/src/portable/nxp/lpc_ip3511/dcd_lpc_ip3511.c
index 5f4a441dc..8adf0f840 100644
--- a/src/portable/nxp/lpc_ip3511/dcd_lpc_ip3511.c
+++ b/src/portable/nxp/lpc_ip3511/dcd_lpc_ip3511.c
@@ -243,7 +243,7 @@ TU_ATTR_ALWAYS_INLINE static inline uint16_t get_buf_offset(void const * buffer)
 }
 
 TU_ATTR_ALWAYS_INLINE static inline uint8_t ep_addr2id(uint8_t ep_addr) {
-  return 2*(ep_addr & 0x0F) + ((ep_addr & TUSB_DIR_IN_MASK) ? 1 : 0);
+  return (uint8_t)(2*(ep_addr & 0x0F) + ((ep_addr & TUSB_DIR_IN_MASK) ? 1 : 0));
 }
 
 TU_ATTR_ALWAYS_INLINE static inline bool ep_is_iso(ep_cmd_sts_t* ep_cs, bool is_highspeed) {
@@ -539,8 +539,8 @@ static void process_xfer_isr(uint8_t rhport, uint32_t int_status) {
       uint16_t buf_nbytes;
 
       if ( rhport_is_highspeed(rhport) ) {
-        buf_offset = ep_cs->buffer_hs.offset;
-        buf_nbytes = ep_cs->buffer_hs.nbytes;
+        buf_offset = (uint16_t)ep_cs->buffer_hs.offset;
+        buf_nbytes = (uint16_t)ep_cs->buffer_hs.nbytes;
 
         #if TU_CHECK_MCU(OPT_MCU_LPC54)
         // LPC54 Errata USB.2: In USB high-speed device mode, the NBytes field is not correct after BULK IN transfer
@@ -550,8 +550,8 @@ static void process_xfer_isr(uint8_t rhport, uint32_t int_status) {
         }
         #endif
       } else {
-        buf_offset = ep_cs->buffer_fs.offset;
-        buf_nbytes = ep_cs->buffer_fs.nbytes;
+        buf_offset = (uint16_t)ep_cs->buffer_fs.offset;
+        buf_nbytes = (uint16_t)ep_cs->buffer_fs.nbytes;
       }
 
       xfer_dma->xferred_bytes += xfer_dma->nbytes - buf_nbytes;
diff --git a/src/portable/renesas/rusb2/dcd_rusb2.c b/src/portable/renesas/rusb2/dcd_rusb2.c
index e2a51a5ca..adbb53787 100644
--- a/src/portable/renesas/rusb2/dcd_rusb2.c
+++ b/src/portable/renesas/rusb2/dcd_rusb2.c
@@ -93,7 +93,9 @@ static unsigned find_pipe(unsigned xfer_type) {
   const uint8_t idx_last  = pipe_idx_arr[xfer_type][1];
 
   for (int i = idx_last; i >= idx_first; i--) {
-    if (0 == _dcd.pipe[i].ep) return i;
+    if (0 == _dcd.pipe[i].ep) {
+      return (unsigned)i;
+    }
   }
 
   return 0;
@@ -117,10 +119,10 @@ static volatile reg_pipetre_t* get_pipetre(rusb2_reg_t *rusb, unsigned num) {
 
 static volatile uint16_t* ep_addr_to_pipectr(uint8_t rhport, unsigned ep_addr) {
   rusb2_reg_t *rusb = RUSB2_REG(rhport);
-  const unsigned epn = tu_edpt_number(ep_addr);
+  const unsigned epn = tu_edpt_number((uint8_t)ep_addr);
 
   if (epn) {
-    const unsigned dir = tu_edpt_dir(ep_addr);
+    const unsigned dir = tu_edpt_dir((uint8_t)ep_addr);
     const unsigned num = _dcd.ep[dir][epn];
     return get_pipectr(rusb, num);
   } else {
@@ -129,11 +131,11 @@ static volatile uint16_t* ep_addr_to_pipectr(uint8_t rhport, unsigned ep_addr) {
 }
 
 static uint16_t edpt0_max_packet_size(rusb2_reg_t* rusb) {
-  return rusb->DCPMAXP_b.MXPS;
+  return (uint16_t)rusb->DCPMAXP_b.MXPS;
 }
 
 static uint16_t edpt_max_packet_size(rusb2_reg_t *rusb, unsigned num) {
-  rusb->PIPESEL = num;
+  rusb->PIPESEL = (uint16_t)num;
   return rusb->PIPEMAXP;
 }
 
@@ -285,7 +287,7 @@ static bool pipe_xfer_out(rusb2_reg_t* rusb, unsigned num)
   const uint16_t mps = edpt_max_packet_size(rusb, num);
   pipe_wait_for_ready(rusb, num);
 
-  const uint16_t vld  = rusb->D0FIFOCTR_b.DTLN;
+  const uint16_t vld  = (uint16_t)rusb->D0FIFOCTR_b.DTLN;
   const uint16_t len  = tu_min16(tu_min16(rem, mps), vld);
   void          *buf  = pipe->buf;
 
@@ -498,7 +500,7 @@ static void process_bus_reset(uint8_t rhport)
   volatile uint16_t *ctr = (volatile uint16_t*)((uintptr_t) (&rusb->PIPE_CTR[0]));
   volatile uint16_t *tre = (volatile uint16_t*)((uintptr_t) (&rusb->PIPE_TR[0].E));
 
-  for (int i = 1; i <= 5; ++i) {
+  for (uint16_t i = 1; i <= 5; ++i) {
     rusb->PIPESEL = i;
     rusb->PIPECFG = 0;
     *ctr = RUSB2_PIPE_CTR_ACLRM_Msk;
@@ -508,7 +510,7 @@ static void process_bus_reset(uint8_t rhport)
     tre += 2;
   }
 
-  for (int i = 6; i <= 9; ++i) {
+  for (uint16_t i = 6; i <= 9; ++i) {
     rusb->PIPESEL = i;
     rusb->PIPECFG = 0;
     *ctr = RUSB2_PIPE_CTR_ACLRM_Msk;
@@ -542,7 +544,7 @@ static void process_bus_reset(uint8_t rhport)
 static void process_set_address(uint8_t rhport)
 {
   rusb2_reg_t* rusb = RUSB2_REG(rhport);
-  const uint16_t addr = rusb->USBADDR_b.USBADDR;
+  const uint16_t addr = (uint16_t)rusb->USBADDR_b.USBADDR;
   if (!addr) {
     return;
   }
@@ -706,7 +708,7 @@ bool dcd_edpt_open(uint8_t rhport, tusb_desc_endpoint_t const * ep_desc)
   (void)rhport;
 
   rusb2_reg_t * rusb = RUSB2_REG(rhport);
-  const unsigned ep_addr = ep_desc->bEndpointAddress;
+  const uint8_t  ep_addr = ep_desc->bEndpointAddress;
   const unsigned epn     = tu_edpt_number(ep_addr);
   const unsigned dir     = tu_edpt_dir(ep_addr);
   const unsigned xfer    = ep_desc->bmAttributes.xfer;
@@ -770,8 +772,10 @@ void dcd_edpt_close_all(uint8_t rhport)
   dcd_int_disable(rhport);
   while (--i) { /* Close all pipes except 0 */
     const unsigned ep_addr = _dcd.pipe[i].ep;
-    if (!ep_addr) continue;
-    dcd_edpt_close(rhport, ep_addr);
+    if (!ep_addr) {
+      continue;
+    }
+    dcd_edpt_close(rhport, (uint8_t)ep_addr);
   }
   dcd_int_enable(rhport);
 }
@@ -783,10 +787,10 @@ void dcd_edpt_close(uint8_t rhport, uint8_t ep_addr)
   const unsigned dir = tu_edpt_dir(ep_addr);
   const unsigned num = _dcd.ep[dir][epn];
 
-  rusb->BRDYENB &= ~TU_BIT(num);
+  rusb->BRDYENB &= (uint16_t)~TU_BIT(num);
   volatile uint16_t *ctr = get_pipectr(rusb, num);
   *ctr = 0;
-  rusb->PIPESEL = num;
+  rusb->PIPESEL = (uint16_t)num;
   rusb->PIPECFG = 0;
   _dcd.pipe[num].ep = 0;
   _dcd.ep[dir][epn] = 0;
@@ -860,7 +864,7 @@ void dcd_edpt_clear_stall(uint8_t rhport, uint8_t ep_addr)
     *ctr = RUSB2_PIPE_CTR_PID_BUF;
   } else {
     const unsigned num = _dcd.ep[0][tu_edpt_number(ep_addr)];
-    rusb->PIPESEL = num;
+    rusb->PIPESEL = (uint16_t)num;
     if (rusb->PIPECFG_b.TYPE != 1) {
       *ctr = RUSB2_PIPE_CTR_PID_BUF;
     }
diff --git a/src/portable/st/stm32_fsdev/dcd_stm32_fsdev.c b/src/portable/st/stm32_fsdev/dcd_stm32_fsdev.c
index 8b4719b21..41da3ddd0 100644
--- a/src/portable/st/stm32_fsdev/dcd_stm32_fsdev.c
+++ b/src/portable/st/stm32_fsdev/dcd_stm32_fsdev.c
@@ -259,7 +259,7 @@ static void handle_ctr_tx(uint32_t ep_id) {
   }
 
   if (xfer->total_len != xfer->queued_len) {
-    dcd_transmit_packet(xfer, ep_id);
+    dcd_transmit_packet(xfer, (uint16_t)ep_id);
   } else {
     dcd_event_xfer_complete(0, ep_num | TUSB_DIR_IN_MASK, xfer->queued_len, XFER_RESULT_SUCCESS, true);
   }
@@ -267,7 +267,7 @@ static void handle_ctr_tx(uint32_t ep_id) {
 
 static void handle_ctr_setup(uint32_t ep_id) {
   uint16_t rx_count = btable_get_count(ep_id, BTABLE_BUF_RX);
-  uint16_t rx_addr  = btable_get_addr(ep_id, BTABLE_BUF_RX);
+  uint16_t rx_addr  = (uint16_t)btable_get_addr(ep_id, BTABLE_BUF_RX);
   uint8_t  setup_packet[8] TU_ATTR_ALIGNED(4);
 
   tu_hwfifo_read(PMA_BUF_AT(rx_addr), setup_packet, rx_count, NULL);
@@ -393,26 +393,8 @@ void dcd_int_handler(uint8_t rhport) {
     const uint32_t ep_reg = ep_read(ep_id);
 
     if (ep_reg & U_EP_CTR_RX) {
-  #ifdef  CFG_TUSB_FSDEV_32BIT
-      /* https://www.st.com/resource/en/errata_sheet/es0561-stm32h503cbebkbrb-device-errata-stmicroelectronics.pdf
-       * https://www.st.com/resource/en/errata_sheet/es0587-stm32u535xx-and-stm32u545xx-device-errata-stmicroelectronics.pdf
-       * From H503/U535 errata: Buffer description table update completes after CTR interrupt triggers
-       * Description:
-       * - During OUT transfers, the correct transfer interrupt (CTR) is triggered a little before the last USB SRAM
-       * accesses have completed. If the software responds quickly to the interrupt, the full buffer contents may not be
-       * correct. Workaround:
-       * - Software should ensure that a small delay is included before accessing the SRAM contents. This delay
-       * should be 800 ns in Full Speed mode and 6.4 μs in Low Speed mode
-       * - Since H5 can run up to 250Mhz -> 1 cycle = 4ns. Per errata, we need to wait 200 cycles. Though executing code
-       * also takes time, so we'll wait 60 cycles (count = 20).
-       * - Since Low Speed mode is not supported/popular, we will ignore it for now.
-       *
-       * Note: this errata may also apply to G0, U5, H5 etc.
-       */
-      volatile uint32_t cycle_count = 20; // defined as PCD_RX_PMA_CNT in stm32 hal_driver
-      while (cycle_count > 0U) {
-        cycle_count--;                    // each count take 3 cycles (1 for sub, jump, and compare)
-      }
+  #if defined(TUP_USBIP_FSDEV_STM32) && defined(CFG_TUSB_FSDEV_32BIT)
+      fsdev_btable_workaround_delay(false);
   #endif
 
       if (ep_reg & U_EP_SETUP) {
@@ -531,8 +513,8 @@ void edpt0_open(uint8_t rhport) {
   xfer_status[0][1].max_packet_size = CFG_TUD_ENDPOINT0_SIZE;
   xfer_status[0][1].ep_idx          = 0;
 
-  uint16_t pma_addr0 = dcd_pma_alloc(CFG_TUD_ENDPOINT0_SIZE, false);
-  uint16_t pma_addr1 = dcd_pma_alloc(CFG_TUD_ENDPOINT0_SIZE, false);
+  uint16_t pma_addr0 = (uint16_t)dcd_pma_alloc(CFG_TUD_ENDPOINT0_SIZE, false);
+  uint16_t pma_addr1 = (uint16_t)dcd_pma_alloc(CFG_TUD_ENDPOINT0_SIZE, false);
 
   btable_set_addr(0, BTABLE_BUF_RX, pma_addr0);
   btable_set_addr(0, BTABLE_BUF_TX, pma_addr1);
@@ -574,7 +556,7 @@ bool dcd_edpt_open(uint8_t rhport, const tusb_desc_endpoint_t *desc_ep) {
   }
 
   /* Create a packet memory buffer area. */
-  uint16_t pma_addr = dcd_pma_alloc(packet_size, false);
+  uint16_t pma_addr = (uint16_t)dcd_pma_alloc(packet_size, false);
   btable_set_addr(ep_idx, dir == TUSB_DIR_IN ? BTABLE_BUF_TX : BTABLE_BUF_RX, pma_addr);
 
   xfer_ctl_t *xfer      = xfer_ctl_ptr(ep_num, dir);
@@ -624,17 +606,17 @@ bool dcd_edpt_iso_alloc(uint8_t rhport, uint8_t ep_addr, uint16_t largest_packet
 
   #if CFG_TUD_FSDEV_DOUBLE_BUFFERED_ISO_EP != 0
   uint32_t pma_addr  = dcd_pma_alloc(largest_packet_size, true);
-  uint16_t pma_addr2 = pma_addr >> 16;
+  uint16_t pma_addr2 = (uint16_t)(pma_addr >> 16);
   #else
   uint32_t pma_addr  = dcd_pma_alloc(largest_packet_size, false);
-  uint16_t pma_addr2 = pma_addr;
+  uint16_t pma_addr2 = (uint16_t)pma_addr;
   #endif
 
   #if FSDEV_USE_SBUF_ISO == 0
-  btable_set_addr(ep_idx, 0, pma_addr);
+  btable_set_addr(ep_idx, 0, (uint16_t)pma_addr);
   btable_set_addr(ep_idx, 1, pma_addr2);
   #else
-  btable_set_addr(ep_idx, dir == TUSB_DIR_IN ? BTABLE_BUF_TX : BTABLE_BUF_RX, pma_addr);
+  btable_set_addr(ep_idx, dir == TUSB_DIR_IN ? BTABLE_BUF_TX : BTABLE_BUF_RX, (uint16_t)pma_addr);
   (void)pma_addr2;
   #endif
 
diff --git a/src/portable/st/stm32_fsdev/fsdev_common.c b/src/portable/st/stm32_fsdev/fsdev_common.c
index 003bcd069..7c4572a1e 100644
--- a/src/portable/st/stm32_fsdev/fsdev_common.c
+++ b/src/portable/st/stm32_fsdev/fsdev_common.c
@@ -81,11 +81,11 @@ uint16_t pma_align_buffer_size(uint16_t size, uint8_t* blsize, uint8_t* num_bloc
   if (size > 62) {
     block_in_bytes = 32;
     *blsize = 1;
-    *num_block = tu_div_ceil(size, 32);
+    *num_block = (uint8_t)tu_div_ceil(size, 32);
   } else {
     block_in_bytes = 2;
     *blsize = 0;
-    *num_block = tu_div_ceil(size, 2);
+    *num_block = (uint8_t)tu_div_ceil(size, 2);
   }
 
   return (*num_block) * block_in_bytes;
diff --git a/src/portable/st/stm32_fsdev/fsdev_common.h b/src/portable/st/stm32_fsdev/fsdev_common.h
index 140ff1d61..af84b8b97 100644
--- a/src/portable/st/stm32_fsdev/fsdev_common.h
+++ b/src/portable/st/stm32_fsdev/fsdev_common.h
@@ -345,7 +345,7 @@ TU_ATTR_ALWAYS_INLINE static inline void ep_write_clear_ctr(uint32_t ep_id, tusb
   uint32_t reg = FSDEV_REG->ep[ep_id].reg;
   reg |= U_EP_CTR_TX | U_EP_CTR_RX;
   reg &= U_EPREG_MASK;
-  reg &= ~(1 << (U_EP_CTR_TX_Pos + (dir == TUSB_DIR_IN ? 0 : 8)));
+  reg &= ~(1u << (U_EP_CTR_TX_Pos + (dir == TUSB_DIR_IN ? 0u : 8u)));
   ep_write(ep_id, reg, false);
 }
 
@@ -378,7 +378,7 @@ TU_ATTR_ALWAYS_INLINE static inline void ch_write_clear_ctr(uint32_t ch_id, tusb
   uint32_t reg = FSDEV_REG->ep[ch_id].reg;
   reg |= U_EP_CTR_TX | U_EP_CTR_RX;
   reg &= U_EPREG_MASK;
-  reg &= ~(1 << (U_EP_CTR_TX_Pos + (dir == TUSB_DIR_IN ? 8 : 0)));
+  reg &= ~(1u << (U_EP_CTR_TX_Pos + (dir == TUSB_DIR_IN ? 8u : 0u)));
   ep_write(ch_id, reg, false);
 }
 
diff --git a/src/portable/st/stm32_fsdev/fsdev_stm32.h b/src/portable/st/stm32_fsdev/fsdev_stm32.h
index a63592c5d..070aa00ec 100644
--- a/src/portable/st/stm32_fsdev/fsdev_stm32.h
+++ b/src/portable/st/stm32_fsdev/fsdev_stm32.h
@@ -252,6 +252,49 @@ TU_ATTR_ALWAYS_INLINE static inline void fsdev_int_disable(uint8_t rhport) {
 }
 
 //--------------------------------------------------------------------+
+// STM32 FSDEV PMA Buffer Description Table errata workaround
+//--------------------------------------------------------------------+
+
+#ifdef CFG_TUSB_FSDEV_32BIT
+/* Errata: Buffer description table update completes after CTR interrupt triggers
+ * https://www.st.com/resource/en/errata_sheet/es0561-stm32h503cbebkbrb-device-errata-stmicroelectronics.pdf
+ * https://www.st.com/resource/en/errata_sheet/es0587-stm32u535xx-and-stm32u545xx-device-errata-stmicroelectronics.pdf
+ *
+ * CTR may trigger before final PMA SRAM accesses complete on OUT transfers.
+ * Insert delay before reading PMA count/data.
+ * Max CPU frequency in MHz, used to derive conservative FSDEV PMA delay defaults.
+ */
+#if CFG_TUSB_MCU == OPT_MCU_STM32H5
+  #define FSDEV_STM32_CPU_MHZ 250U
+#elif CFG_TUSB_MCU == OPT_MCU_STM32U5
+  #define FSDEV_STM32_CPU_MHZ 160U
+#elif CFG_TUSB_MCU == OPT_MCU_STM32U3
+  #define FSDEV_STM32_CPU_MHZ 96U
+#elif CFG_TUSB_MCU == OPT_MCU_STM32U0
+  #define FSDEV_STM32_CPU_MHZ 56U
+#elif CFG_TUSB_MCU == OPT_MCU_STM32G0
+  #define FSDEV_STM32_CPU_MHZ 64U
+#elif CFG_TUSB_MCU == OPT_MCU_STM32C0
+  #define FSDEV_STM32_CPU_MHZ 48U
+#endif
+
+#ifndef CFG_TUSB_FSDEV_BTABLE_FS_DELAY_COUNT
+  #define CFG_TUSB_FSDEV_BTABLE_FS_DELAY_COUNT (FSDEV_STM32_CPU_MHZ / 4U)
+#endif
+
+#ifndef CFG_TUSB_FSDEV_BTABLE_LS_DELAY_COUNT
+  #define CFG_TUSB_FSDEV_BTABLE_LS_DELAY_COUNT (FSDEV_STM32_CPU_MHZ * 2U)
+#endif
+
+TU_ATTR_ALWAYS_INLINE static inline void fsdev_btable_workaround_delay(bool low_speed) {
+  volatile uint32_t cycle_count = low_speed ? CFG_TUSB_FSDEV_BTABLE_LS_DELAY_COUNT : CFG_TUSB_FSDEV_BTABLE_FS_DELAY_COUNT;
+  while (cycle_count > 0U) {
+    cycle_count--;
+  }
+}
+#endif
+
+//--------------------------------------------------------------------+
 // Connect / Disconnect
 //--------------------------------------------------------------------+
 
diff --git a/src/portable/st/stm32_fsdev/hcd_stm32_fsdev.c b/src/portable/st/stm32_fsdev/hcd_stm32_fsdev.c
index 18685dbdc..f9201651a 100644
--- a/src/portable/st/stm32_fsdev/hcd_stm32_fsdev.c
+++ b/src/portable/st/stm32_fsdev/hcd_stm32_fsdev.c
@@ -58,20 +58,6 @@
 
 TU_VERIFY_STATIC(CFG_TUH_FSDEV_ENDPOINT_MAX <= 255, "currently only use 8-bit for index");
 
-#if CFG_TUSB_MCU == OPT_MCU_STM32H5
-  #define CPU_FREQUENCY_MHZ 250U
-#elif CFG_TUSB_MCU == OPT_MCU_STM32U5
-  #define CPU_FREQUENCY_MHZ 160U
-#elif CFG_TUSB_MCU == OPT_MCU_STM32U3
-  #define CPU_FREQUENCY_MHZ 96U
-#elif CFG_TUSB_MCU == OPT_MCU_STM32G0
-  #define CPU_FREQUENCY_MHZ 64U
-#elif CFG_TUSB_MCU == OPT_MCU_STM32C0
-  #define CPU_FREQUENCY_MHZ 48U
-#else
-  #error "CPU_FREQUENCY_MHZ not defined for this STM32 MCU"
-#endif
-
 enum {
   HCD_XFER_ERROR_MAX = 3,
   HCD_XFER_NAK_MAX = 15,
@@ -165,35 +151,9 @@ static inline void channel_write_status(uint8_t ch_id, uint32_t ch_reg, tusb_dir
 }
 
 static inline uint16_t channel_get_rx_count(uint8_t ch_id) {
-  /* https://www.st.com/resource/en/errata_sheet/es0561-stm32h503cbebkbrb-device-errata-stmicroelectronics.pdf
-  * https://www.st.com/resource/en/errata_sheet/es0587-stm32u535xx-and-stm32u545xx-device-errata-stmicroelectronics.pdf
-  * From H503/U535 errata: Buffer description table update completes after CTR interrupt triggers
-  * Description:
-  * - During OUT transfers, the correct transfer interrupt (CTR) is triggered a little before the last USB SRAM accesses
-  * have completed. If the software responds quickly to the interrupt, the full buffer contents may not be correct.
-  * Workaround:
-  * - Software should ensure that a small delay is included before accessing the SRAM contents. This delay
-  * should be 800 ns in Full Speed mode and 6.4 μs in Low Speed mode
-  *
-  * Note: this errata may also apply to G0, U5, H5 etc.
-  *
-  * We choose the delay count based on max CPU frequency (in MHz) to ensure the delay is at least the required time.
-  */
-
   uint32_t ch_reg = ch_read(ch_id);
-  if (FSDEV_REG->ISTR & U_ISTR_LS_DCONN || ch_reg & U_EP_LSEP) {
-    // Low speed mode: 6.4 us delay -> about 2 cycles per MHz
-    volatile uint32_t cycle_count = CPU_FREQUENCY_MHZ * 2U;
-    while (cycle_count > 0U) {
-      cycle_count--; // each count take 3 cycles (1 for sub, jump, and compare)
-    }
-  } else {
-    // Full speed mode: 800 ns delay -> about 0.25 cycles per MHz
-    volatile uint32_t cycle_count = CPU_FREQUENCY_MHZ / 4U;
-    while (cycle_count > 0U) {
-      cycle_count--; // each count take 3 cycles (1 for sub, jump, and compare)
-    }
-  }
+  const bool is_low_speed = (FSDEV_REG->ISTR & U_ISTR_LS_DCONN) || (ch_reg & U_EP_LSEP);
+  fsdev_btable_workaround_delay(is_low_speed);
 
   return btable_get_count(ch_id, BTABLE_BUF_RX);
 }
@@ -237,11 +197,7 @@ bool hcd_init(uint8_t rhport, const tusb_rhport_init_t* rh_init) {
 
   // If DCON_STAT is already set, the controller sometimes misses the initial connection interrupt
   if (FSDEV_REG->ISTR & U_ISTR_DCON_STAT) {
-    // Wait DP/DM stabilize time
-    volatile uint32_t cycle_count = CPU_FREQUENCY_MHZ / 4U;
-    while (cycle_count > 0U) {
-      cycle_count--;
-    }
+    tusb_time_delay_ms_api(2);
     port_status_handler(rhport, false);
   }
 
diff --git a/src/portable/synopsys/dwc2/dcd_dwc2.c b/src/portable/synopsys/dwc2/dcd_dwc2.c
index 9a9c734a0..30e24a9ad 100644
--- a/src/portable/synopsys/dwc2/dcd_dwc2.c
+++ b/src/portable/synopsys/dwc2/dcd_dwc2.c
@@ -191,7 +191,7 @@ static void dma_setup_prepare(uint8_t rhport) {
 */
 
 TU_ATTR_ALWAYS_INLINE static inline uint16_t calc_device_grxfsiz(uint16_t largest_ep_size, uint8_t ep_count) {
-  return 13 + 1 + 2 * ((largest_ep_size / 4) + 1) + 2 * ep_count;
+  return (uint16_t)(13 + 1 + 2 * ((largest_ep_size / 4) + 1) + 2 * ep_count);
 }
 
 static bool dfifo_alloc(uint8_t rhport, uint8_t ep_addr, uint16_t packet_size, bool is_bulk) {
@@ -203,7 +203,7 @@ static bool dfifo_alloc(uint8_t rhport, uint8_t ep_addr, uint16_t packet_size, b
 
   TU_ASSERT(epnum < ep_count);
 
-  uint16_t fifo_size = tu_div_ceil(packet_size, 4);
+  uint16_t fifo_size = (uint16_t)tu_div_ceil(packet_size, 4);
   if (dir == TUSB_DIR_OUT) {
     // Calculate required size of RX FIFO
     const uint16_t new_sz = calc_device_grxfsiz(4 * fifo_size, ep_count);
@@ -371,7 +371,7 @@ static void edpt_schedule_packets(uint8_t rhport, const uint8_t epnum, const uin
     num_packets = 1;
   } else {
     total_bytes = xfer->total_len;
-    num_packets = tu_div_ceil(total_bytes, xfer->max_size);
+    num_packets = (uint16_t)tu_div_ceil(total_bytes, xfer->max_size);
     if (num_packets == 0) {
       num_packets = 1; // zero length packet still count as 1
     }
@@ -541,8 +541,9 @@ void dcd_remote_wakeup(uint8_t rhport) {
 void dcd_connect(uint8_t rhport) {
   dwc2_regs_t* dwc2 = DWC2_REG(rhport);
 
-#ifdef TUP_USBIP_DWC2_ESP32
-  // On ESP32-P4 HS PHY, do not write to USB_WRAP register which belongs to FS PHY
+#if defined(TUP_USBIP_DWC2_ESP32) && !TU_CHECK_MCU(OPT_MCU_ESP32S31)
+  // S31 is excluded at compile time (no USB_WRAP peripheral).
+  // On P4, the HS PHY (port 1) must not touch USB_WRAP which belongs to the FS PHY.
   if (rhport == 0) {
     usb_wrap_otg_conf_reg_t conf = USB_WRAP.otg_conf;
     conf.pad_pull_override = 0;
@@ -560,8 +561,9 @@ void dcd_connect(uint8_t rhport) {
 void dcd_disconnect(uint8_t rhport) {
   dwc2_regs_t* dwc2 = DWC2_REG(rhport);
 
-#ifdef TUP_USBIP_DWC2_ESP32
-  // On ESP32-P4 HS PHY, do not write to USB_WRAP register which belongs to FS PHY
+#if defined(TUP_USBIP_DWC2_ESP32) && !TU_CHECK_MCU(OPT_MCU_ESP32S31)
+  // S31 is excluded at compile time (no USB_WRAP peripheral).
+  // On P4, the HS PHY (port 1) must not touch USB_WRAP which belongs to the FS PHY.
   if (rhport == 0) {
     usb_wrap_otg_conf_reg_t conf = USB_WRAP.otg_conf;
     conf.pad_pull_override = 1;
diff --git a/src/portable/synopsys/dwc2/dwc2_esp32.h b/src/portable/synopsys/dwc2/dwc2_esp32.h
index 6a10dc7f8..436f8dc30 100644
--- a/src/portable/synopsys/dwc2/dwc2_esp32.h
+++ b/src/portable/synopsys/dwc2/dwc2_esp32.h
@@ -37,7 +37,11 @@
 
 #include "esp_intr_alloc.h"
 #include "soc/periph_defs.h"
+
+// ESP32-S31 does not have USB_WRAP peripheral (HS-only with UTMI PHY)
+#if !TU_CHECK_MCU(OPT_MCU_ESP32S31)
 #include "soc/usb_wrap_struct.h"
+#endif
 
 #if TU_CHECK_MCU(OPT_MCU_ESP32S2, OPT_MCU_ESP32S3)
 #define DWC2_FS_REG_BASE   0x60080000UL
@@ -75,6 +79,14 @@ static const dwc2_controller_t _dwc2_controller[] = {
   { .reg_base = DWC2_FS_REG_BASE, .irqnum = ETS_USB_OTG11_CH0_INTR_SOURCE, .ep_count = 7, .ep_in_count = 5, .otg_dfifo_depth = 256 },
   { .reg_base = DWC2_HS_REG_BASE, .irqnum = ETS_USB_OTG_INTR_SOURCE, .ep_count = 16, .ep_in_count = 8, .otg_dfifo_depth = 1024 }
 };
+
+#elif TU_CHECK_MCU(OPT_MCU_ESP32S31)
+#define DWC2_HS_REG_BASE   0x20300000UL
+#define DWC2_EP_MAX        16
+
+static const dwc2_controller_t _dwc2_controller[] = {
+  { .reg_base = DWC2_HS_REG_BASE, .irqnum = ETS_USB_OTGHS_INTR_SOURCE, .ep_count = 16, .ep_in_count = 8, .otg_dfifo_depth = 1024 }
+};
 #endif
 
 //--------------------------------------------------------------------+
diff --git a/src/tusb_option.h b/src/tusb_option.h
index dd7af76f6..154f8e2a4 100644
--- a/src/tusb_option.h
+++ b/src/tusb_option.h
@@ -134,6 +134,7 @@
 #define OPT_MCU_ESP32C5           908 ///< Espressif ESP32-C5
 #define OPT_MCU_ESP32C61          909 ///< Espressif ESP32-C61
 #define OPT_MCU_ESP32H4           910 ///< Espressif ESP32-H4
+#define OPT_MCU_ESP32S31          911 ///< Espressif ESP32-S31
 
 // Dialog
 #define OPT_MCU_DA1469X          1000 ///< Dialog Semiconductor DA1469x