diff --git a/src/device/usbd.c b/src/device/usbd.c
index 4ab60569b..3043fc7bc 100644
--- a/src/device/usbd.c
+++ b/src/device/usbd.c
@@ -1041,6 +1041,7 @@ static bool process_get_descriptor(uint8_t rhport, tusb_control_request_t const
     break;
 
     case TUSB_DESC_DEVICE_QUALIFIER:
+    {
       TU_LOG2(" Device Qualifier\r\n");
 
       TU_VERIFY( tud_descriptor_device_qualifier_cb );
@@ -1050,6 +1051,7 @@ static bool process_get_descriptor(uint8_t rhport, tusb_control_request_t const
 
       // first byte of descriptor is its size
       return tud_control_xfer(rhport, p_request, (void*) desc_qualifier, desc_qualifier[0]);
+    }
     break;
 
     default: return false;
@@ -1377,7 +1379,13 @@ void usbd_edpt_close(uint8_t rhport, uint8_t ep_addr)
   TU_ASSERT(dcd_edpt_close, /**/);
   TU_LOG2("  CLOSING Endpoint: 0x%02X\r\n", ep_addr);
 
+  uint8_t const epnum = tu_edpt_number(ep_addr);
+  uint8_t const dir   = tu_edpt_dir(ep_addr);
+
   dcd_edpt_close(rhport, ep_addr);
+  _usbd_dev.ep_status[epnum][dir].stalled = false;
+  _usbd_dev.ep_status[epnum][dir].busy = false;
+  _usbd_dev.ep_status[epnum][dir].claimed = false;
 
   return;
 }
diff --git a/src/portable/nxp/transdimension/dcd_transdimension.c b/src/portable/nxp/transdimension/dcd_transdimension.c
index 07009ff50..a7c4545c2 100644
--- a/src/portable/nxp/transdimension/dcd_transdimension.c
+++ b/src/portable/nxp/transdimension/dcd_transdimension.c
@@ -134,7 +134,8 @@ typedef struct
   // QHD is 64 bytes aligned but occupies only 48 bytes
   // Therefore there are 16 bytes padding that we can use.
   //--------------------------------------------------------------------+
-  uint8_t reserved[16];
+  tu_fifo_t * ff;
+  uint8_t reserved[12];
 } dcd_qhd_t;
 
 TU_VERIFY_STATIC( sizeof(dcd_qhd_t) == 64, "size is not correct");
@@ -240,8 +241,9 @@ void dcd_init(uint8_t rhport)
   dcd_reg->USBMODE = USBMODE_CM_DEVICE;
   dcd_reg->OTGSC = OTGSC_VBUS_DISCHARGE | OTGSC_OTG_TERMINATION;
 
-  // TODO Force fullspeed on non-highspeed port
-  // dcd_reg->PORTSC1 = PORTSC1_FORCE_FULL_SPEED;
+#if !TUD_OPT_HIGH_SPEED
+  dcd_reg->PORTSC1 = PORTSC1_FORCE_FULL_SPEED;
+#endif
 
   CleanInvalidateDCache_by_Addr((uint32_t*) &_dcd_data, sizeof(dcd_data_t));
 
@@ -296,18 +298,30 @@ void dcd_disconnect(uint8_t rhport)
 
 static void qtd_init(dcd_qtd_t* p_qtd, void * data_ptr, uint16_t total_bytes)
 {
+  // Force the CPU to flush the buffer. We increase the size by 31 because the call aligns the
+  // address to 32-byte boundaries. Buffer must be word aligned
+  CleanInvalidateDCache_by_Addr((uint32_t*) tu_align((uint32_t) data_ptr, 4), total_bytes + 31);
+
   tu_memclr(p_qtd, sizeof(dcd_qtd_t));
 
-  p_qtd->next        = QTD_NEXT_INVALID;
-  p_qtd->active      = 1;
-  p_qtd->total_bytes = p_qtd->expected_bytes = total_bytes;
+  p_qtd->next            = QTD_NEXT_INVALID;
+  p_qtd->active          = 1;
+  p_qtd->total_bytes     = p_qtd->expected_bytes = total_bytes;
+  p_qtd->int_on_complete = true;
 
   if (data_ptr != NULL)
   {
-    p_qtd->buffer[0]   = (uint32_t) data_ptr;
+    p_qtd->buffer[0] = (uint32_t) data_ptr;
+
+    uint32_t const bufend = p_qtd->buffer[0] + total_bytes;
     for(uint8_t i=1; i<5; i++)
     {
-      p_qtd->buffer[i] |= tu_align4k( p_qtd->buffer[i-1] ) + 4096;
+      uint32_t const next_page = tu_align4k( p_qtd->buffer[i-1] ) + 4096;
+      if ( bufend <= next_page ) break;
+
+      p_qtd->buffer[i] = next_page;
+
+      // TODO page[1] FRAME_N for ISO transfer
     }
   }
 }
@@ -340,9 +354,6 @@ void dcd_edpt_clear_stall(uint8_t rhport, uint8_t ep_addr)
 
 bool dcd_edpt_open(uint8_t rhport, tusb_desc_endpoint_t const * p_endpoint_desc)
 {
-  // TODO not support ISO yet
-  TU_VERIFY ( p_endpoint_desc->bmAttributes.xfer != TUSB_XFER_ISOCHRONOUS);
-
   uint8_t const epnum = tu_edpt_number(p_endpoint_desc->bEndpointAddress);
   uint8_t const dir   = tu_edpt_dir(p_endpoint_desc->bEndpointAddress);
 
@@ -355,13 +366,27 @@ bool dcd_edpt_open(uint8_t rhport, tusb_desc_endpoint_t const * p_endpoint_desc)
 
   p_qhd->zero_length_termination = 1;
   p_qhd->max_packet_size         = p_endpoint_desc->wMaxPacketSize.size;
+  if (p_endpoint_desc->bmAttributes.xfer == TUSB_XFER_ISOCHRONOUS)
+  {
+    p_qhd->iso_mult = 1;
+  }
+
   p_qhd->qtd_overlay.next        = QTD_NEXT_INVALID;
 
   CleanInvalidateDCache_by_Addr((uint32_t*) &_dcd_data, sizeof(dcd_data_t));
 
   // Enable EP Control
   dcd_registers_t* dcd_reg = _dcd_controller[rhport].regs;
-  dcd_reg->ENDPTCTRL[epnum] |= ((p_endpoint_desc->bmAttributes.xfer << 2) | ENDPTCTRL_ENABLE | ENDPTCTRL_TOGGLE_RESET) << (dir ? 16 : 0);
+
+  uint32_t const epctrl = (p_endpoint_desc->bmAttributes.xfer << ENDPTCTRL_TYPE_POS) | ENDPTCTRL_ENABLE | ENDPTCTRL_TOGGLE_RESET;
+
+  if ( dir == TUSB_DIR_OUT )
+  {
+    dcd_reg->ENDPTCTRL[epnum] = (dcd_reg->ENDPTCTRL[epnum] & 0xFFFF0000u) | epctrl;
+  }else
+  {
+    dcd_reg->ENDPTCTRL[epnum] = (dcd_reg->ENDPTCTRL[epnum] & 0x0000FFFFu) | (epctrl << 16);
+  }
 
   return true;
 }
@@ -381,11 +406,35 @@ void dcd_edpt_close_all (uint8_t rhport)
   }
 }
 
-bool dcd_edpt_xfer(uint8_t rhport, uint8_t ep_addr, uint8_t * buffer, uint16_t total_bytes)
+void dcd_edpt_close(uint8_t rhport, uint8_t ep_addr)
+{
+  uint8_t const epnum  = tu_edpt_number(ep_addr);
+  uint8_t const dir    = tu_edpt_dir(ep_addr);
+
+  dcd_registers_t* dcd_reg = _dcd_controller[rhport].regs;
+
+  _dcd_data.qhd[epnum][dir].qtd_overlay.halted = 1;
+
+  // Flush EP
+  uint32_t const flush_mask = TU_BIT(epnum + (dir ? 16 : 0));
+  dcd_reg->ENDPTFLUSH = flush_mask;
+  while(dcd_reg->ENDPTFLUSH & flush_mask);
+
+  // Clear EP enable
+  dcd_reg->ENDPTCTRL[epnum] &=~(ENDPTCTRL_ENABLE << (dir ? 16 : 0));
+}
+
+static void qhd_start_xfer(uint8_t rhport, uint8_t epnum, uint8_t dir)
 {
   dcd_registers_t* dcd_reg = _dcd_controller[rhport].regs;
-  uint8_t const epnum = tu_edpt_number(ep_addr);
-  uint8_t const dir   = tu_edpt_dir(ep_addr);
+  dcd_qhd_t* p_qhd = &_dcd_data.qhd[epnum][dir];
+  dcd_qtd_t* p_qtd = &_dcd_data.qtd[epnum][dir];
+
+  p_qhd->qtd_overlay.halted = false;            // clear any previous error
+  p_qhd->qtd_overlay.next   = (uint32_t) p_qtd; // link qtd to qhd
+
+  // flush cache
+  CleanInvalidateDCache_by_Addr((uint32_t*) &_dcd_data, sizeof(dcd_data_t));
 
   if ( epnum == 0 )
   {
@@ -394,25 +443,87 @@ bool dcd_edpt_xfer(uint8_t rhport, uint8_t ep_addr, uint8_t * buffer, uint16_t t
     while(dcd_reg->ENDPTSETUPSTAT & TU_BIT(0)) {}
   }
 
+  // start transfer
+  dcd_reg->ENDPTPRIME = TU_BIT(epnum + (dir ? 16 : 0));
+}
+
+bool dcd_edpt_xfer(uint8_t rhport, uint8_t ep_addr, uint8_t * buffer, uint16_t total_bytes)
+{
+  uint8_t const epnum = tu_edpt_number(ep_addr);
+  uint8_t const dir   = tu_edpt_dir(ep_addr);
+
+  dcd_qhd_t* p_qhd = &_dcd_data.qhd[epnum][dir];
+  dcd_qtd_t* p_qtd = &_dcd_data.qtd[epnum][dir];
+
+  // Prepare qtd
+  qtd_init(p_qtd, buffer, total_bytes);
+
+  // Start qhd transfer
+  p_qhd->ff = NULL;
+  qhd_start_xfer(rhport, epnum, dir);
+
+  return true;
+}
+
+// fifo has to be aligned to 4k boundary
+bool dcd_edpt_xfer_fifo (uint8_t rhport, uint8_t ep_addr, tu_fifo_t * ff, uint16_t total_bytes)
+{
+  uint8_t const epnum = tu_edpt_number(ep_addr);
+  uint8_t const dir   = tu_edpt_dir(ep_addr);
+
   dcd_qhd_t * p_qhd = &_dcd_data.qhd[epnum][dir];
   dcd_qtd_t * p_qtd = &_dcd_data.qtd[epnum][dir];
 
-  // Force the CPU to flush the buffer. We increase the size by 32 because the call aligns the
-  // address to 32-byte boundaries.
-  // void* cast to suppress cast-align warning, buffer must be
-  CleanInvalidateDCache_by_Addr((uint32_t*) tu_align((uint32_t) buffer, 4), total_bytes + 31);
+  tu_fifo_buffer_info_t fifo_info;
 
-  //------------- Prepare qtd -------------//
-  qtd_init(p_qtd, buffer, total_bytes);
-  p_qtd->int_on_complete = true;
+  if (dir)
+  {
+    tu_fifo_get_read_info(ff, &fifo_info);
+  } else
+  {
+    tu_fifo_get_write_info(ff, &fifo_info);
+  }
 
-  p_qhd->qtd_overlay.halted = false;            // clear any previous error
-  p_qhd->qtd_overlay.next   = (uint32_t) p_qtd; // activate by linking qtd to qhd
+  if ( fifo_info.len_lin >= total_bytes )
+  {
+    // Linear length is enough for this transfer
+    qtd_init(p_qtd, fifo_info.ptr_lin, total_bytes);
+  }
+  else
+  {
+    // linear part is not enough
 
-  CleanInvalidateDCache_by_Addr((uint32_t*) &_dcd_data, sizeof(dcd_data_t));
+    // prepare TD up to linear length
+    qtd_init(p_qtd, fifo_info.ptr_lin, fifo_info.len_lin);
 
-  // start transfer
-  dcd_reg->ENDPTPRIME = TU_BIT(epnum + (dir ? 16 : 0));
+    if ( !tu_offset4k((uint32_t) fifo_info.ptr_wrap) && !tu_offset4k(tu_fifo_depth(ff)) )
+    {
+      // If buffer is aligned to 4K & buffer size is multiple of 4K
+      // We can make use of buffer page array to also combine the linear + wrapped length
+      p_qtd->total_bytes = p_qtd->expected_bytes = total_bytes;
+
+      for(uint8_t i = 1, page = 0; i < 5; i++)
+      {
+        // pick up buffer array where linear ends
+        if (p_qtd->buffer[i] == 0)
+        {
+          p_qtd->buffer[i] = (uint32_t) fifo_info.ptr_wrap + 4096 * page;
+          page++;
+        }
+      }
+
+      CleanInvalidateDCache_by_Addr((uint32_t*) tu_align((uint32_t) fifo_info.ptr_wrap, 4), total_bytes - fifo_info.len_wrap + 31);
+    }
+    else
+    {
+      // TODO we may need to carry the wrapped length after the linear part complete
+      // for now only transfer up to linear part
+    }
+  }
+
+  // Start qhd transfer
+  p_qhd->ff = ff;
+  qhd_start_xfer(rhport, epnum, dir);
 
   return true;
 }
@@ -423,6 +534,7 @@ bool dcd_edpt_xfer(uint8_t rhport, uint8_t ep_addr, uint8_t * buffer, uint16_t t
 
 static void process_edpt_complete_isr(uint8_t rhport, uint8_t epnum, uint8_t dir)
 {
+  dcd_qhd_t * p_qhd = &_dcd_data.qhd[epnum][dir];
   dcd_qtd_t * p_qtd = &_dcd_data.qtd[epnum][dir];
 
   uint8_t result = p_qtd->halted ? XFER_RESULT_STALLED :
@@ -435,8 +547,21 @@ static void process_edpt_complete_isr(uint8_t rhport, uint8_t epnum, uint8_t dir
     dcd_reg->ENDPTFLUSH = TU_BIT(epnum + (dir ? 16 : 0));
   }
 
+  uint16_t const xferred_bytes = p_qtd->expected_bytes - p_qtd->total_bytes;
+
+  if (p_qhd->ff)
+  {
+    if (dir == TUSB_DIR_IN)
+    {
+      tu_fifo_advance_read_pointer(p_qhd->ff, xferred_bytes);
+    } else
+    {
+      tu_fifo_advance_write_pointer(p_qhd->ff, xferred_bytes);
+    }
+  }
+  
   // only number of bytes in the IOC qtd
-  dcd_event_xfer_complete(rhport, tu_edpt_addr(epnum, dir), p_qtd->expected_bytes - p_qtd->total_bytes, result, true);
+  dcd_event_xfer_complete(rhport, tu_edpt_addr(epnum, dir), xferred_bytes, result, true);
 }
 
 void dcd_int_handler(uint8_t rhport)