fix bugs in fsdev hcd, dcd and midi host stream write

- hcd_edpt_clear_stall: use ep_addr instead of hardcoded 0 (control endpoint)
- hcd: add OPT_MCU_STM32U3 to compile guard to enable host mode on STM32U3
- dcd_edpt_close_all: use FSDEV_EP_COUNT instead of CFG_TUD_ENDPPOINT_MAX for
  PMA btable offset to match handle_bus_reset
- midi host tuh_midi_stream_write: add missing cable_num to system messages,
  SysEx, and real-time MIDI packets. Add 0xF mask for SysEx CIN checks.
  Aligns with midi_device.c tud_midi_n_stream_write implementation.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: hathach

src/class/midi/midi_host.c

@@ -461,7 +461,7 @@ uint32_t tuh_midi_stream_write(uint8_t idx, uint8_t cable_num, uint8_t const *bu
     if (data >= MIDI_STATUS_SYSREAL_TIMING_CLOCK) {
       // real-time messages need to be sent right away
       midi_driver_stream_t streamrt;
-      streamrt.buffer[0] = MIDI_CIN_SYSEX_END_1BYTE;
+      streamrt.buffer[0] = (uint8_t)((cable_num << 4) | MIDI_CIN_SYSEX_END_1BYTE);
       streamrt.buffer[1] = data;
       streamrt.index = 2;
       streamrt.total = 2;
@@ -476,9 +476,9 @@ uint32_t tuh_midi_stream_write(uint8_t idx, uint8_t cable_num, uint8_t const *bu
       stream->buffer[1] = data;
 
       // Check to see if we're still in a SysEx transmit.
-      if (stream->buffer[0] == MIDI_CIN_SYSEX_START) {
+      if ((stream->buffer[0] & 0xF) == MIDI_CIN_SYSEX_START) {
         if (data == MIDI_STATUS_SYSEX_END) {
-          stream->buffer[0] = MIDI_CIN_SYSEX_END_1BYTE;
+          stream->buffer[0] = (uint8_t)((cable_num << 4) | MIDI_CIN_SYSEX_END_1BYTE);
           stream->total = 2;
         } else {
           stream->total = 4;
@@ -506,6 +506,7 @@ uint32_t tuh_midi_stream_write(uint8_t idx, uint8_t cable_num, uint8_t const *bu
           stream->buffer[0] = MIDI_CIN_SYSEX_END_1BYTE;
           stream->total = 2;
         }
+        stream->buffer[0] |= (uint8_t)(cable_num << 4);
       } else {
         // Pack individual bytes if we don't support packing them into words.
         stream->buffer[0] = (uint8_t) (cable_num << 4 | 0xf);
@@ -520,8 +521,8 @@ uint32_t tuh_midi_stream_write(uint8_t idx, uint8_t cable_num, uint8_t const *bu
       stream->buffer[stream->index] = data;
       stream->index++;
       // See if this byte ends a SysEx.
-      if (stream->buffer[0] == MIDI_CIN_SYSEX_START && data == MIDI_STATUS_SYSEX_END) {
-        stream->buffer[0] = MIDI_CIN_SYSEX_START + (stream->index - 1);
+      if ((stream->buffer[0] & 0xF) == MIDI_CIN_SYSEX_START && data == MIDI_STATUS_SYSEX_END) {
+        stream->buffer[0] = (uint8_t)((cable_num << 4) | (MIDI_CIN_SYSEX_START + (stream->index - 1)));
         stream->total = stream->index;
       }
     }

src/portable/st/stm32_fsdev/dcd_stm32_fsdev.c

@@ -612,7 +612,7 @@ void dcd_edpt_close_all(uint8_t rhport) {
   dcd_int_enable(rhport);
 
   // Reset PMA allocation
-  ep_buf_ptr = FSDEV_BTABLE_BASE + 8 * CFG_TUD_ENDPPOINT_MAX + 2 * CFG_TUD_ENDPOINT0_SIZE;
+  ep_buf_ptr = FSDEV_BTABLE_BASE + 8 * FSDEV_EP_COUNT + 2 * CFG_TUD_ENDPOINT0_SIZE;
 }
 
 bool dcd_edpt_iso_alloc(uint8_t rhport, uint8_t ep_addr, uint16_t largest_packet_size) {

src/portable/st/stm32_fsdev/hcd_stm32_fsdev.c

@@ -39,7 +39,7 @@
 #include "tusb_option.h"
 
 #if CFG_TUH_ENABLED && defined(TUP_USBIP_FSDEV) && \
-    TU_CHECK_MCU(OPT_MCU_STM32C0, OPT_MCU_STM32G0, OPT_MCU_STM32H5, OPT_MCU_STM32U5)
+    TU_CHECK_MCU(OPT_MCU_STM32C0, OPT_MCU_STM32G0, OPT_MCU_STM32H5, OPT_MCU_STM32U3, OPT_MCU_STM32U5)
 
 #include "host/hcd.h"
 #include "host/usbh.h"
@@ -672,10 +672,8 @@ bool hcd_setup_send(uint8_t rhport, uint8_t dev_addr, uint8_t const setup_packet
 // Clear stall, data toggle is also reset to DATA0
 bool hcd_edpt_clear_stall(uint8_t rhport, uint8_t dev_addr, uint8_t ep_addr) {
   (void) rhport;
-  (void) dev_addr;
-  (void) ep_addr;
 
-  uint8_t const ep_id = endpoint_find(dev_addr, 0);
+  uint8_t const ep_id = endpoint_find(dev_addr, ep_addr);
   TU_ASSERT(ep_id != TUSB_INDEX_INVALID_8);
 
   hcd_endpoint_t *edpt = &_hcd_data.edpt[ep_id];