diff --git a/include/sound/sof/ipc4/header.h b/include/sound/sof/ipc4/header.h
index 78568abe26735e..c58f00ef054a78 100644
--- a/include/sound/sof/ipc4/header.h
+++ b/include/sound/sof/ipc4/header.h
@@ -106,12 +106,19 @@ enum sof_ipc4_global_msg {
 	SOF_IPC4_GLB_SAVE_PIPELINE,
 	SOF_IPC4_GLB_RESTORE_PIPELINE,
 
-	/* Loads library (using Code Load or HD/A Host Output DMA) */
+	/*
+	 * library loading
+	 *
+	 * Loads library (using Code Load or HD/A Host Output DMA)
+	 */
 	SOF_IPC4_GLB_LOAD_LIBRARY,
+	/*
+	 * Prepare the host DMA channel for library loading, must be followed by
+	 * a SOF_IPC4_GLB_LOAD_LIBRARY message as the library loading step
+	 */
+	SOF_IPC4_GLB_LOAD_LIBRARY_PREPARE,
 
-	/* 25: RESERVED - do not use */
-
-	SOF_IPC4_GLB_INTERNAL_MESSAGE = 26,
+	SOF_IPC4_GLB_INTERNAL_MESSAGE,
 
 	/* Notification (FW to SW driver) */
 	SOF_IPC4_GLB_NOTIFICATION,
diff --git a/sound/soc/sof/intel/hda-loader.c b/sound/soc/sof/intel/hda-loader.c
index 769697bdc3951a..46fb2d1425e9f2 100644
--- a/sound/soc/sof/intel/hda-loader.c
+++ b/sound/soc/sof/intel/hda-loader.c
@@ -545,11 +545,40 @@ int hda_dsp_ipc4_load_library(struct snd_sof_dev *sdev,
 
 	memcpy(dmab.area, stripped_firmware.data, stripped_firmware.size);
 
+	/*
+	 * 1st stage: SOF_IPC4_GLB_LOAD_LIBRARY_PREPARE
+	 * Message includes the dma_id to be prepared for the library loading.
+	 * If the firmware does not have support for the message, we will
+	 * receive -EOPNOTSUPP. In this case we will use single step library
+	 * loading and proceed to send the LOAD_LIBRARY message.
+	 */
 	msg.primary = hext_stream->hstream.stream_tag - 1;
-	msg.primary |= SOF_IPC4_MSG_TYPE_SET(SOF_IPC4_GLB_LOAD_LIBRARY);
+	msg.primary |= SOF_IPC4_MSG_TYPE_SET(SOF_IPC4_GLB_LOAD_LIBRARY_PREPARE);
 	msg.primary |= SOF_IPC4_MSG_DIR(SOF_IPC4_MSG_REQUEST);
 	msg.primary |= SOF_IPC4_MSG_TARGET(SOF_IPC4_FW_GEN_MSG);
-	msg.primary |= SOF_IPC4_GLB_LOAD_LIBRARY_LIB_ID(fw_lib->id);
+	ret = sof_ipc_tx_message_no_reply(sdev->ipc, &msg, 0);
+	if (!ret) {
+		int sd_offset = SOF_STREAM_SD_OFFSET(&hext_stream->hstream);
+		unsigned int status;
+
+		/*
+		 * Make sure that the FIFOS value is not 0 in SDxFIFOS register
+		 * which indicates that the firmware set the GEN bit and we can
+		 * continue to start the DMA
+		 */
+		ret = snd_sof_dsp_read_poll_timeout(sdev, HDA_DSP_HDA_BAR,
+					sd_offset + SOF_HDA_ADSP_REG_SD_FIFOSIZE,
+					status,
+					status & SOF_HDA_SD_FIFOSIZE_FIFOS_MASK,
+					HDA_DSP_REG_POLL_INTERVAL_US,
+					HDA_DSP_BASEFW_TIMEOUT_US);
+
+		if (ret < 0)
+			dev_warn(sdev->dev,
+				 "%s: timeout waiting for FIFOS\n", __func__);
+	} else if (ret != -EOPNOTSUPP) {
+		goto cleanup;
+	}
 
 	ret = cl_trigger(sdev, hext_stream, SNDRV_PCM_TRIGGER_START);
 	if (ret < 0) {
@@ -557,8 +586,17 @@ int hda_dsp_ipc4_load_library(struct snd_sof_dev *sdev,
 		goto cleanup;
 	}
 
+	/*
+	 * 2nd stage: LOAD_LIBRARY
+	 * Message includes the dma_id and the lib_id, the dma_id must be
+	 * identical to the one sent via LOAD_LIBRARY_PREPARE
+	 */
+	msg.primary &= ~SOF_IPC4_MSG_TYPE_MASK;
+	msg.primary |= SOF_IPC4_MSG_TYPE_SET(SOF_IPC4_GLB_LOAD_LIBRARY);
+	msg.primary |= SOF_IPC4_GLB_LOAD_LIBRARY_LIB_ID(fw_lib->id);
 	ret = sof_ipc_tx_message_no_reply(sdev->ipc, &msg, 0);
 
+	/* Stop the DMA channel */
 	ret1 = cl_trigger(sdev, hext_stream, SNDRV_PCM_TRIGGER_STOP);
 	if (ret1 < 0) {
 		dev_err(sdev->dev, "%s: DMA trigger stop failed\n", __func__);
diff --git a/sound/soc/sof/intel/hda-stream.c b/sound/soc/sof/intel/hda-stream.c
index 0b0087abcc50ed..65e9242365bedb 100644
--- a/sound/soc/sof/intel/hda-stream.c
+++ b/sound/soc/sof/intel/hda-stream.c
@@ -668,7 +668,7 @@ int hda_dsp_stream_hw_params(struct snd_sof_dev *sdev,
 			snd_sof_dsp_read(sdev, HDA_DSP_HDA_BAR,
 					 sd_offset +
 					 SOF_HDA_ADSP_REG_SD_FIFOSIZE);
-		hstream->fifo_size &= 0xffff;
+		hstream->fifo_size &= SOF_HDA_SD_FIFOSIZE_FIFOS_MASK;
 		hstream->fifo_size += 1;
 	} else {
 		hstream->fifo_size = 0;
diff --git a/sound/soc/sof/intel/hda.h b/sound/soc/sof/intel/hda.h
index 5c517ec57d4a20..2b228c63905bfd 100644
--- a/sound/soc/sof/intel/hda.h
+++ b/sound/soc/sof/intel/hda.h
@@ -135,6 +135,9 @@
 #define SOF_HDA_ADSP_REG_SD_BDLPU		0x1C
 #define SOF_HDA_ADSP_SD_ENTRY_SIZE		0x20
 
+/* SDxFIFOS FIFOS */
+#define SOF_HDA_SD_FIFOSIZE_FIFOS_MASK		GENMASK(15, 0)
+
 /* CL: Software Position Based FIFO Capability Registers */
 #define SOF_DSP_REG_CL_SPBFIFO \
 	(SOF_HDA_ADSP_LOADER_BASE + 0x20)
diff --git a/sound/soc/sof/ipc4.c b/sound/soc/sof/ipc4.c
index ab6eddd91bb771..e14924048eb5b9 100644
--- a/sound/soc/sof/ipc4.c
+++ b/sound/soc/sof/ipc4.c
@@ -99,6 +99,10 @@ static int sof_ipc4_check_reply_status(struct snd_sof_dev *sdev, u32 status)
 
 to_errno:
 	switch (status) {
+	case 2:
+	case 15:
+		ret = -EOPNOTSUPP;
+		break;
 	case 8:
 	case 11:
 	case 105 ... 109:
@@ -153,6 +157,7 @@ static const char * const ipc4_dbg_glb_msg_type[] = {
 	DBG_IPC4_MSG_TYPE_ENTRY(GLB_SAVE_PIPELINE),
 	DBG_IPC4_MSG_TYPE_ENTRY(GLB_RESTORE_PIPELINE),
 	DBG_IPC4_MSG_TYPE_ENTRY(GLB_LOAD_LIBRARY),
+	DBG_IPC4_MSG_TYPE_ENTRY(GLB_LOAD_LIBRARY_PREPARE),
 	DBG_IPC4_MSG_TYPE_ENTRY(GLB_INTERNAL_MESSAGE),
 	DBG_IPC4_MSG_TYPE_ENTRY(GLB_NOTIFICATION),
 };