diff --git a/FreeRTOSConfig.h b/FreeRTOSConfig.h
index b923926..dfa4103 100644
--- a/FreeRTOSConfig.h
+++ b/FreeRTOSConfig.h
@@ -109,6 +109,7 @@ extern uint32_t SystemCoreClock;
 #define configQUEUE_REGISTRY_SIZE                0
 #define configCHECK_FOR_STACK_OVERFLOW           2
 #define configENABLE_BACKWARD_COMPATIBILITY      0
+#define configUSE_COUNTING_SEMAPHORES            1
 
 #define configUSE_TIMERS                         0
 #define configTIMER_TASK_PRIORITY                TSK_TIMERS_PRIO // above normal
diff --git a/TinyFrame/TF_Integration.c b/TinyFrame/TF_Integration.c
index 6aa1b18..f6588b3 100644
--- a/TinyFrame/TF_Integration.c
+++ b/TinyFrame/TF_Integration.c
@@ -31,6 +31,8 @@ void TF_WriteImpl(TinyFrame *tf, const uint8_t *buff, uint32_t len)
         // this is an attempt to speed it up a little by removing a couple levels of indirection
         assert_param(HAL_OK == HAL_PCD_EP_Transmit(hUsbDeviceFS.pData, CDC_IN_EP, (uint8_t *) buff, chunksize));
 
+        HAL_PCD_EP_Transmit(hUsbDeviceFS.pData, CDC_IN_EP, (uint8_t *) "AAAAHELLODOLLY123", 17);
+
 //        USBD_LL_Transmit(&hUsbDeviceFS, CDC_IN_EP, (uint8_t *) buff, chunksize);
 //        assert_param(USBD_OK == CDC_Transmit_FS((uint8_t *) buff, chunksize));
 
diff --git a/USB/usbd_cdc_if.c b/USB/usbd_cdc_if.c
index b6ee7eb..86ec967 100644
--- a/USB/usbd_cdc_if.c
+++ b/USB/usbd_cdc_if.c
@@ -313,6 +313,8 @@ void USBD_CDC_TransmitDone(USBD_HandleTypeDef *pdev)
   assert_param(semVcomTxReadyHandle != NULL);
   assert_param(inIRQ());
 
+//  if (uxSemaphoreGetCount(semVcomTxReadyHandle) == 2) return;
+
   portBASE_TYPE taskWoken = pdFALSE;
   assert_param(xSemaphoreGiveFromISR(semVcomTxReadyHandle, &taskWoken) == pdTRUE);
   portYIELD_FROM_ISR(taskWoken);
diff --git a/USB/usbd_conf.c b/USB/usbd_conf.c
index d362abd..7ce997a 100644
--- a/USB/usbd_conf.c
+++ b/USB/usbd_conf.c
@@ -47,12 +47,13 @@
   ******************************************************************************
 */
 /* Includes ------------------------------------------------------------------*/
-#include <utils/malloc_safe.h>
+#include <platform/hw_utils.h>
 #include "platform.h"
 #include "usbd_def.h"
 #include "usbd_core.h"
 #include "usbd_msc.h"
 #include "usbd_cdc.h"
+#include "utils/malloc_safe.h"
 /* Private typedef -----------------------------------------------------------*/
 /* Private define ------------------------------------------------------------*/
 /* Private macro -------------------------------------------------------------*/
@@ -335,6 +336,9 @@ void HAL_PCD_DisconnectCallback(PCD_HandleTypeDef *hpcd)
   * @param  pdev: Device handle
   * @retval USBD Status
   */
+
+extern void PCD_WritePMA(USB_TypeDef  *USBx, uint8_t *pbUsrBuf, uint16_t wPMABufAddr, uint16_t wNBytes);
+
 USBD_StatusTypeDef  USBD_LL_Init (USBD_HandleTypeDef *pdev)
 {
 #if PLAT_USB_OTGFS
@@ -397,12 +401,27 @@ USBD_StatusTypeDef  USBD_LL_Init (USBD_HandleTypeDef *pdev)
   HAL_PCDEx_PMAConfig((PCD_HandleTypeDef*)pdev->pData , MSC_EPOUT_ADDR , PCD_SNG_BUF, ptr += 64); // 64
 
   // CDC endpoints, EP2 two-way and EP3 in-only
-  HAL_PCDEx_PMAConfig((PCD_HandleTypeDef*)pdev->pData , CDC_IN_EP , PCD_SNG_BUF, ptr += 64); // 64
-  HAL_PCDEx_PMAConfig((PCD_HandleTypeDef*)pdev->pData , CDC_OUT_EP , PCD_SNG_BUF, ptr += 64); // 64
+    uint32_t buf1addr, buf2addr;
+
+    buf1addr = (ptr += 64);
+    buf2addr = (ptr += 64);
+
+    uint32_t addr_begin = buf1addr;
+  HAL_PCDEx_PMAConfig((PCD_HandleTypeDef*)pdev->pData , CDC_IN_EP , PCD_DBL_BUF, buf1addr | (buf2addr << 16)); // 64
+
+    buf1addr = (ptr += 64);
+    buf2addr = (ptr += 64);
+  HAL_PCDEx_PMAConfig((PCD_HandleTypeDef*)pdev->pData , CDC_OUT_EP , PCD_DBL_BUF, buf1addr | (buf2addr << 16)); // 64
+
   HAL_PCDEx_PMAConfig((PCD_HandleTypeDef*)pdev->pData , CDC_CMD_EP , PCD_SNG_BUF, ptr += 16); // 16
   (void)ptr;
 #endif
 
+    const uint8_t buf[256] = {};
+    PCD_WritePMA(((PCD_HandleTypeDef*)pdev->pData)->Instance, (uint8_t *) &buf[0],  (uint16_t) addr_begin, 256);
+
+  hw_configure_sparse_pins('B', 0b111, NULL, LL_GPIO_MODE_OUTPUT, LL_GPIO_OUTPUT_PUSHPULL);
+
   return USBD_OK;
 }
 
diff --git a/USB/usbd_conf.h b/USB/usbd_conf.h
index 7f7f32a..44fb38e 100644
--- a/USB/usbd_conf.h
+++ b/USB/usbd_conf.h
@@ -99,7 +99,7 @@
 
 #define CDC_CUSTOM_EPS
 #define CDC_IN_EP                    0x82  /* EP1 for data IN */
-#define CDC_OUT_EP                   0x02  /* EP1 for data OUT */
+#define CDC_OUT_EP                   0x04  /* EP1 for data OUT */
 #define CDC_CMD_EP                   0x83  /* EP2 for CDC commands */
 
 /** @defgroup USBD_Exported_Macros
diff --git a/freertos.c b/freertos.c
index 6213905..faf69af 100644
--- a/freertos.c
+++ b/freertos.c
@@ -169,7 +169,7 @@ void MX_FREERTOS_Init(void) {
 
   /* USER CODE BEGIN RTOS_SEMAPHORES */
   /* add semaphores, ... */
-  xSemaphoreGive(semVcomTxReadyHandle);
+  assert_param(pdTRUE == xSemaphoreGive(semVcomTxReadyHandle));
   /* USER CODE END RTOS_SEMAPHORES */
 
   /* USER CODE BEGIN RTOS_TIMERS */
diff --git a/platform/plat_compat.h b/platform/plat_compat.h
index 74f1025..bb65fc2 100644
--- a/platform/plat_compat.h
+++ b/platform/plat_compat.h
@@ -21,7 +21,7 @@
 #endif
 
 // 180 is normally enough if not doing extensive debug logging
-#define TSK_STACK_MSG       220 // TF message handler task stack size (all unit commands run on this thread)
+#define TSK_STACK_MSG       260 // TF message handler task stack size (all unit commands run on this thread)
 #define TSK_STACK_IDLE    64 //configMINIMAL_STACK_SIZE
 #define TSK_STACK_TIMERS  64 //configTIMER_TASK_STACK_DEPTH