Merge tag 'asoc-v5.9' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus

ASoC: Updates for v5.9

The biggest changes here one again come from Mormioto-san who has
continued his dilligent work cleaning up long standing issues in the
APIs, it's particularly nice to see the transition from digital_mute()
to mute_stream() finally completed. There's also been a lot of work on
the x86 code again, this time a big focus has been on cleaning up some
issues identified by various static tests, and on the Freescale systems.
Otherwise the biggest thing has been a lot of driver additions:

 - Convert users of digital_mute() to mute_stream().
 - Simplify I/O helper functions.
 - Add a helper for getting the RTD from a substream.
 - Many, many fixes and cleanups to the x86 code.
 - New drivers for Freescale MQS and i.MX6sx, Intel KeemBay I2S, Maxim
   MAX98360A and MAX98373 Soundwire, several Mediatek boards, nVidia
   Tegra 186 and 210, RealTek RL6231, Samsung Midas and Aries boards (some
   of the first phones I worked on!) and TI J721e EVM.

19 files changed, 309 insertions, 134 deletions

diff --git a/drivers/misc/atmel-ssc.c b/drivers/misc/atmel-ssc.c
index ab4144ea1f11..d6cd5537126c 100644
--- a/drivers/misc/atmel-ssc.c
+++ b/drivers/misc/atmel-ssc.c
@@ -10,7 +10,7 @@
 #include <linux/clk.h>
 #include <linux/err.h>
 #include <linux/io.h>
-#include <linux/spinlock.h>
+#include <linux/mutex.h>
 #include <linux/atmel-ssc.h>
 #include <linux/slab.h>
 #include <linux/module.h>
@@ -20,7 +20,7 @@
 #include "../../sound/soc/atmel/atmel_ssc_dai.h"
 
 /* Serialize access to ssc_list and user count */
-static DEFINE_SPINLOCK(user_lock);
+static DEFINE_MUTEX(user_lock);
 static LIST_HEAD(ssc_list);
 
 struct ssc_device *ssc_request(unsigned int ssc_num)
@@ -28,7 +28,7 @@ struct ssc_device *ssc_request(unsigned int ssc_num)
 	int ssc_valid = 0;
 	struct ssc_device *ssc;
 
-	spin_lock(&user_lock);
+	mutex_lock(&user_lock);
 	list_for_each_entry(ssc, &ssc_list, list) {
 		if (ssc->pdev->dev.of_node) {
 			if (of_alias_get_id(ssc->pdev->dev.of_node, "ssc")
@@ -44,18 +44,18 @@ struct ssc_device *ssc_request(unsigned int ssc_num)
 	}
 
 	if (!ssc_valid) {
-		spin_unlock(&user_lock);
+		mutex_unlock(&user_lock);
 		pr_err("ssc: ssc%d platform device is missing\n", ssc_num);
 		return ERR_PTR(-ENODEV);
 	}
 
 	if (ssc->user) {
-		spin_unlock(&user_lock);
+		mutex_unlock(&user_lock);
 		dev_dbg(&ssc->pdev->dev, "module busy\n");
 		return ERR_PTR(-EBUSY);
 	}
 	ssc->user++;
-	spin_unlock(&user_lock);
+	mutex_unlock(&user_lock);
 
 	clk_prepare(ssc->clk);
 
@@ -67,14 +67,14 @@ void ssc_free(struct ssc_device *ssc)
 {
 	bool disable_clk = true;
 
-	spin_lock(&user_lock);
+	mutex_lock(&user_lock);
 	if (ssc->user)
 		ssc->user--;
 	else {
 		disable_clk = false;
 		dev_dbg(&ssc->pdev->dev, "device already free\n");
 	}
-	spin_unlock(&user_lock);
+	mutex_unlock(&user_lock);
 
 	if (disable_clk)
 		clk_unprepare(ssc->clk);
@@ -237,9 +237,9 @@ static int ssc_probe(struct platform_device *pdev)
 		return -ENXIO;
 	}
 
-	spin_lock(&user_lock);
+	mutex_lock(&user_lock);
 	list_add_tail(&ssc->list, &ssc_list);
-	spin_unlock(&user_lock);
+	mutex_unlock(&user_lock);
 
 	platform_set_drvdata(pdev, ssc);
 
@@ -258,9 +258,9 @@ static int ssc_remove(struct platform_device *pdev)
 
 	ssc_sound_dai_remove(ssc);
 
-	spin_lock(&user_lock);
+	mutex_lock(&user_lock);
 	list_del(&ssc->list);
-	spin_unlock(&user_lock);
+	mutex_unlock(&user_lock);
 
 	return 0;
 }
diff --git a/drivers/misc/habanalabs/command_submission.c b/drivers/misc/habanalabs/command_submission.c
index f82974a916c3..f3a8f113865d 100644
--- a/drivers/misc/habanalabs/command_submission.c
+++ b/drivers/misc/habanalabs/command_submission.c
@@ -62,6 +62,12 @@ static void hl_fence_release(struct dma_fence *fence)
 		container_of(fence, struct hl_cs_compl, base_fence);
 	struct hl_device *hdev = hl_cs_cmpl->hdev;
 
+	/* EBUSY means the CS was never submitted and hence we don't have
+	 * an attached hw_sob object that we should handle here
+	 */
+	if (fence->error == -EBUSY)
+		goto free;
+
 	if ((hl_cs_cmpl->type == CS_TYPE_SIGNAL) ||
 			(hl_cs_cmpl->type == CS_TYPE_WAIT)) {
 
@@ -92,6 +98,7 @@ static void hl_fence_release(struct dma_fence *fence)
 		kref_put(&hl_cs_cmpl->hw_sob->kref, hl_sob_reset);
 	}
 
+free:
 	kfree_rcu(hl_cs_cmpl, base_fence.rcu);
 }
 
@@ -328,10 +335,16 @@ static void cs_do_release(struct kref *ref)
 
 	hl_ctx_put(cs->ctx);
 
+	/* We need to mark an error for not submitted because in that case
+	 * the dma fence release flow is different. Mainly, we don't need
+	 * to handle hw_sob for signal/wait
+	 */
 	if (cs->timedout)
 		dma_fence_set_error(cs->fence, -ETIMEDOUT);
 	else if (cs->aborted)
 		dma_fence_set_error(cs->fence, -EIO);
+	else if (!cs->submitted)
+		dma_fence_set_error(cs->fence, -EBUSY);
 
 	dma_fence_signal(cs->fence);
 	dma_fence_put(cs->fence);
@@ -486,11 +499,19 @@ static int validate_queue_index(struct hl_device *hdev,
 	struct asic_fixed_properties *asic = &hdev->asic_prop;
 	struct hw_queue_properties *hw_queue_prop;
 
+	/* This must be checked here to prevent out-of-bounds access to
+	 * hw_queues_props array
+	 */
+	if (chunk->queue_index >= HL_MAX_QUEUES) {
+		dev_err(hdev->dev, "Queue index %d is invalid\n",
+			chunk->queue_index);
+		return -EINVAL;
+	}
+
 	hw_queue_prop = &asic->hw_queues_props[chunk->queue_index];
 
-	if ((chunk->queue_index >= HL_MAX_QUEUES) ||
-			(hw_queue_prop->type == QUEUE_TYPE_NA)) {
-		dev_err(hdev->dev, "Queue index %d is invalid\n",
+	if (hw_queue_prop->type == QUEUE_TYPE_NA) {
+		dev_err(hdev->dev, "Queue index %d is not applicable\n",
 			chunk->queue_index);
 		return -EINVAL;
 	}
diff --git a/drivers/misc/habanalabs/debugfs.c b/drivers/misc/habanalabs/debugfs.c
index 3c8dcdfba20c..0bc036e01ee8 100644
--- a/drivers/misc/habanalabs/debugfs.c
+++ b/drivers/misc/habanalabs/debugfs.c
@@ -36,7 +36,7 @@ static int hl_debugfs_i2c_read(struct hl_device *hdev, u8 i2c_bus, u8 i2c_addr,
 	pkt.i2c_reg = i2c_reg;
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-					HL_DEVICE_TIMEOUT_USEC, (long *) val);
+						0, (long *) val);
 
 	if (rc)
 		dev_err(hdev->dev, "Failed to read from I2C, error %d\n", rc);
@@ -63,7 +63,7 @@ static int hl_debugfs_i2c_write(struct hl_device *hdev, u8 i2c_bus, u8 i2c_addr,
 	pkt.value = cpu_to_le64(val);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-					HL_DEVICE_TIMEOUT_USEC, NULL);
+						0, NULL);
 
 	if (rc)
 		dev_err(hdev->dev, "Failed to write to I2C, error %d\n", rc);
@@ -87,7 +87,7 @@ static void hl_debugfs_led_set(struct hl_device *hdev, u8 led, u8 state)
 	pkt.value = cpu_to_le64(state);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-						HL_DEVICE_TIMEOUT_USEC, NULL);
+						0, NULL);
 
 	if (rc)
 		dev_err(hdev->dev, "Failed to set LED %d, error %d\n", led, rc);
@@ -480,7 +480,7 @@ out:
 	return 0;
 }
 
-static ssize_t mmu_write(struct file *file, const char __user *buf,
+static ssize_t mmu_asid_va_write(struct file *file, const char __user *buf,
 		size_t count, loff_t *f_pos)
 {
 	struct seq_file *s = file->private_data;
@@ -981,7 +981,7 @@ static ssize_t hl_clk_gate_read(struct file *f, char __user *buf,
 	if (*ppos)
 		return 0;
 
-	sprintf(tmp_buf, "%d\n", hdev->clock_gating);
+	sprintf(tmp_buf, "0x%llx\n", hdev->clock_gating_mask);
 	rc = simple_read_from_buffer(buf, strlen(tmp_buf) + 1, ppos, tmp_buf,
 			strlen(tmp_buf) + 1);
 
@@ -993,7 +993,7 @@ static ssize_t hl_clk_gate_write(struct file *f, const char __user *buf,
 {
 	struct hl_dbg_device_entry *entry = file_inode(f)->i_private;
 	struct hl_device *hdev = entry->hdev;
-	u32 value;
+	u64 value;
 	ssize_t rc;
 
 	if (atomic_read(&hdev->in_reset)) {
@@ -1002,19 +1002,12 @@ static ssize_t hl_clk_gate_write(struct file *f, const char __user *buf,
 		return 0;
 	}
 
-	rc = kstrtouint_from_user(buf, count, 10, &value);
+	rc = kstrtoull_from_user(buf, count, 16, &value);
 	if (rc)
 		return rc;
 
-	if (value) {
-		hdev->clock_gating = 1;
-		if (hdev->asic_funcs->enable_clock_gating)
-			hdev->asic_funcs->enable_clock_gating(hdev);
-	} else {
-		if (hdev->asic_funcs->disable_clock_gating)
-			hdev->asic_funcs->disable_clock_gating(hdev);
-		hdev->clock_gating = 0;
-	}
+	hdev->clock_gating_mask = value;
+	hdev->asic_funcs->set_clock_gating(hdev);
 
 	return count;
 }
@@ -1125,7 +1118,7 @@ static const struct hl_info_list hl_debugfs_list[] = {
 	{"command_submission_jobs", command_submission_jobs_show, NULL},
 	{"userptr", userptr_show, NULL},
 	{"vm", vm_show, NULL},
-	{"mmu", mmu_show, mmu_write},
+	{"mmu", mmu_show, mmu_asid_va_write},
 	{"engines", engines_show, NULL}
 };
 
diff --git a/drivers/misc/habanalabs/device.c b/drivers/misc/habanalabs/device.c
index 2b38a119704c..59608d1bac88 100644
--- a/drivers/misc/habanalabs/device.c
+++ b/drivers/misc/habanalabs/device.c
@@ -608,7 +608,7 @@ int hl_device_set_debug_mode(struct hl_device *hdev, bool enable)
 		hdev->in_debug = 0;
 
 		if (!hdev->hard_reset_pending)
-			hdev->asic_funcs->enable_clock_gating(hdev);
+			hdev->asic_funcs->set_clock_gating(hdev);
 
 		goto out;
 	}
diff --git a/drivers/misc/habanalabs/firmware_if.c b/drivers/misc/habanalabs/firmware_if.c
index baf790cf4b78..d27841cb5bcb 100644
--- a/drivers/misc/habanalabs/firmware_if.c
+++ b/drivers/misc/habanalabs/firmware_if.c
@@ -61,7 +61,7 @@ int hl_fw_send_pci_access_msg(struct hl_device *hdev, u32 opcode)
 	pkt.ctl = cpu_to_le32(opcode << ARMCP_PKT_CTL_OPCODE_SHIFT);
 
 	return hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt,
-				sizeof(pkt), HL_DEVICE_TIMEOUT_USEC, NULL);
+						sizeof(pkt), 0, NULL);
 }
 
 int hl_fw_send_cpu_message(struct hl_device *hdev, u32 hw_queue_id, u32 *msg,
@@ -144,7 +144,7 @@ int hl_fw_unmask_irq(struct hl_device *hdev, u16 event_type)
 	pkt.value = cpu_to_le64(event_type);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-			HL_DEVICE_TIMEOUT_USEC, &result);
+						0, &result);
 
 	if (rc)
 		dev_err(hdev->dev, "failed to unmask RAZWI IRQ %d", event_type);
@@ -183,7 +183,7 @@ int hl_fw_unmask_irq_arr(struct hl_device *hdev, const u32 *irq_arr,
 						ARMCP_PKT_CTL_OPCODE_SHIFT);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) pkt,
-			total_pkt_size, HL_DEVICE_TIMEOUT_USEC, &result);
+						total_pkt_size, 0, &result);
 
 	if (rc)
 		dev_err(hdev->dev, "failed to unmask IRQ array\n");
@@ -204,7 +204,7 @@ int hl_fw_test_cpu_queue(struct hl_device *hdev)
 	test_pkt.value = cpu_to_le64(ARMCP_PACKET_FENCE_VAL);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &test_pkt,
-			sizeof(test_pkt), HL_DEVICE_TIMEOUT_USEC, &result);
+						sizeof(test_pkt), 0, &result);
 
 	if (!rc) {
 		if (result != ARMCP_PACKET_FENCE_VAL)
@@ -248,7 +248,7 @@ int hl_fw_send_heartbeat(struct hl_device *hdev)
 	hb_pkt.value = cpu_to_le64(ARMCP_PACKET_FENCE_VAL);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &hb_pkt,
-			sizeof(hb_pkt), HL_DEVICE_TIMEOUT_USEC, &result);
+						sizeof(hb_pkt), 0, &result);
 
 	if ((rc) || (result != ARMCP_PACKET_FENCE_VAL))
 		rc = -EIO;
diff --git a/drivers/misc/habanalabs/gaudi/gaudi.c b/drivers/misc/habanalabs/gaudi/gaudi.c
index 61f88e9884ce..637a9d608707 100644
--- a/drivers/misc/habanalabs/gaudi/gaudi.c
+++ b/drivers/misc/habanalabs/gaudi/gaudi.c
@@ -80,6 +80,7 @@
 #define GAUDI_PLDM_QMAN0_TIMEOUT_USEC	(HL_DEVICE_TIMEOUT_USEC * 30)
 #define GAUDI_PLDM_TPC_KERNEL_WAIT_USEC	(HL_DEVICE_TIMEOUT_USEC * 30)
 #define GAUDI_BOOT_FIT_REQ_TIMEOUT_USEC	1000000		/* 1s */
+#define GAUDI_MSG_TO_CPU_TIMEOUT_USEC	4000000		/* 4s */
 
 #define GAUDI_QMAN0_FENCE_VAL		0x72E91AB9
 
@@ -96,7 +97,12 @@
 
 #define GAUDI_NUM_OF_QM_ARB_ERR_CAUSE	3
 
-#define GAUDI_ARB_WDT_TIMEOUT		0x400000
+#define GAUDI_ARB_WDT_TIMEOUT		0x1000000
+
+#define GAUDI_CLK_GATE_DEBUGFS_MASK	(\
+		BIT(GAUDI_ENGINE_ID_MME_0) |\
+		BIT(GAUDI_ENGINE_ID_MME_2) |\
+		GENMASK_ULL(GAUDI_ENGINE_ID_TPC_7, GAUDI_ENGINE_ID_TPC_0))
 
 static const char gaudi_irq_name[GAUDI_MSI_ENTRIES][GAUDI_MAX_STRING_LEN] = {
 		"gaudi cq 0_0", "gaudi cq 0_1", "gaudi cq 0_2", "gaudi cq 0_3",
@@ -106,14 +112,14 @@ static const char gaudi_irq_name[GAUDI_MSI_ENTRIES][GAUDI_MAX_STRING_LEN] = {
 };
 
 static const u8 gaudi_dma_assignment[GAUDI_DMA_MAX] = {
-	[GAUDI_PCI_DMA_1] = 0,
-	[GAUDI_PCI_DMA_2] = 1,
-	[GAUDI_PCI_DMA_3] = 5,
-	[GAUDI_HBM_DMA_1] = 2,
-	[GAUDI_HBM_DMA_2] = 3,
-	[GAUDI_HBM_DMA_3] = 4,
-	[GAUDI_HBM_DMA_4] = 6,
-	[GAUDI_HBM_DMA_5] = 7
+	[GAUDI_PCI_DMA_1] = GAUDI_ENGINE_ID_DMA_0,
+	[GAUDI_PCI_DMA_2] = GAUDI_ENGINE_ID_DMA_1,
+	[GAUDI_PCI_DMA_3] = GAUDI_ENGINE_ID_DMA_5,
+	[GAUDI_HBM_DMA_1] = GAUDI_ENGINE_ID_DMA_2,
+	[GAUDI_HBM_DMA_2] = GAUDI_ENGINE_ID_DMA_3,
+	[GAUDI_HBM_DMA_3] = GAUDI_ENGINE_ID_DMA_4,
+	[GAUDI_HBM_DMA_4] = GAUDI_ENGINE_ID_DMA_6,
+	[GAUDI_HBM_DMA_5] = GAUDI_ENGINE_ID_DMA_7
 };
 
 static const u8 gaudi_cq_assignment[NUMBER_OF_CMPLT_QUEUES] = {
@@ -1819,7 +1825,7 @@ static void gaudi_init_golden_registers(struct hl_device *hdev)
 
 	gaudi_init_rate_limiter(hdev);
 
-	gaudi_disable_clock_gating(hdev);
+	hdev->asic_funcs->disable_clock_gating(hdev);
 
 	for (tpc_id = 0, tpc_offset = 0;
 				tpc_id < TPC_NUMBER_OF_ENGINES;
@@ -1893,6 +1899,8 @@ static void gaudi_init_pci_dma_qman(struct hl_device *hdev, int dma_id,
 	WREG32(mmDMA0_QM_CP_MSG_BASE3_ADDR_LO_0 + q_off, so_base_ws_lo);
 	WREG32(mmDMA0_QM_CP_MSG_BASE3_ADDR_HI_0 + q_off, so_base_ws_hi);
 
+	WREG32(mmDMA0_QM_CP_BARRIER_CFG_0 + q_off, 0x100);
+
 	/* The following configuration is needed only once per QMAN */
 	if (qman_id == 0) {
 		/* Configure RAZWI IRQ */
@@ -2529,46 +2537,55 @@ static void gaudi_tpc_stall(struct hl_device *hdev)
 	WREG32(mmTPC7_CFG_TPC_STALL, 1 << TPC0_CFG_TPC_STALL_V_SHIFT);
 }
 
-static void gaudi_enable_clock_gating(struct hl_device *hdev)
+static void gaudi_set_clock_gating(struct hl_device *hdev)
 {
 	struct gaudi_device *gaudi = hdev->asic_specific;
 	u32 qman_offset;
 	int i;
 
-	if (!hdev->clock_gating)
-		return;
-
-	if (gaudi->hw_cap_initialized & HW_CAP_CLK_GATE)
-		return;
-
 	/* In case we are during debug session, don't enable the clock gate
 	 * as it may interfere
 	 */
 	if (hdev->in_debug)
 		return;
 
-	for (i = 0, qman_offset = 0 ; i < PCI_DMA_NUMBER_OF_CHNLS ; i++) {
+	for (i = GAUDI_PCI_DMA_1, qman_offset = 0 ; i < GAUDI_HBM_DMA_1 ; i++) {
+		if (!(hdev->clock_gating_mask &
+					(BIT_ULL(gaudi_dma_assignment[i]))))
+			continue;
+
 		qman_offset = gaudi_dma_assignment[i] * DMA_QMAN_OFFSET;
 		WREG32(mmDMA0_QM_CGM_CFG1 + qman_offset, QMAN_CGM1_PWR_GATE_EN);
 		WREG32(mmDMA0_QM_CGM_CFG + qman_offset,
 				QMAN_UPPER_CP_CGM_PWR_GATE_EN);
 	}
 
-	for (; i < HBM_DMA_NUMBER_OF_CHNLS ; i++) {
+	for (i = GAUDI_HBM_DMA_1 ; i < GAUDI_DMA_MAX ; i++) {
+		if (!(hdev->clock_gating_mask &
+					(BIT_ULL(gaudi_dma_assignment[i]))))
+			continue;
+
 		qman_offset = gaudi_dma_assignment[i] * DMA_QMAN_OFFSET;
 		WREG32(mmDMA0_QM_CGM_CFG1 + qman_offset, QMAN_CGM1_PWR_GATE_EN);
 		WREG32(mmDMA0_QM_CGM_CFG + qman_offset,
 				QMAN_COMMON_CP_CGM_PWR_GATE_EN);
 	}
 
-	WREG32(mmMME0_QM_CGM_CFG1, QMAN_CGM1_PWR_GATE_EN);
-	WREG32(mmMME0_QM_CGM_CFG,
-			QMAN_COMMON_CP_CGM_PWR_GATE_EN);
-	WREG32(mmMME2_QM_CGM_CFG1, QMAN_CGM1_PWR_GATE_EN);
-	WREG32(mmMME2_QM_CGM_CFG,
-			QMAN_COMMON_CP_CGM_PWR_GATE_EN);
+	if (hdev->clock_gating_mask & (BIT_ULL(GAUDI_ENGINE_ID_MME_0))) {
+		WREG32(mmMME0_QM_CGM_CFG1, QMAN_CGM1_PWR_GATE_EN);
+		WREG32(mmMME0_QM_CGM_CFG, QMAN_COMMON_CP_CGM_PWR_GATE_EN);
+	}
+
+	if (hdev->clock_gating_mask & (BIT_ULL(GAUDI_ENGINE_ID_MME_2))) {
+		WREG32(mmMME2_QM_CGM_CFG1, QMAN_CGM1_PWR_GATE_EN);
+		WREG32(mmMME2_QM_CGM_CFG, QMAN_COMMON_CP_CGM_PWR_GATE_EN);
+	}
 
 	for (i = 0, qman_offset = 0 ; i < TPC_NUMBER_OF_ENGINES ; i++) {
+		if (!(hdev->clock_gating_mask &
+					(BIT_ULL(GAUDI_ENGINE_ID_TPC_0 + i))))
+			continue;
+
 		WREG32(mmTPC0_QM_CGM_CFG1 + qman_offset,
 				QMAN_CGM1_PWR_GATE_EN);
 		WREG32(mmTPC0_QM_CGM_CFG + qman_offset,
@@ -2661,7 +2678,7 @@ static void gaudi_halt_engines(struct hl_device *hdev, bool hard_reset)
 	gaudi_stop_hbm_dma_qmans(hdev);
 	gaudi_stop_pci_dma_qmans(hdev);
 
-	gaudi_disable_clock_gating(hdev);
+	hdev->asic_funcs->disable_clock_gating(hdev);
 
 	msleep(wait_timeout_ms);
 
@@ -2725,6 +2742,12 @@ static int gaudi_mmu_init(struct hl_device *hdev)
 	WREG32(mmSTLB_HOP_CONFIGURATION,
 			hdev->mmu_huge_page_opt ? 0x30440 : 0x40440);
 
+	/*
+	 * The H/W expects the first PI after init to be 1. After wraparound
+	 * we'll write 0.
+	 */
+	gaudi->mmu_cache_inv_pi = 1;
+
 	gaudi->hw_cap_initialized |= HW_CAP_MMU;
 
 	return 0;
@@ -2995,7 +3018,7 @@ static int gaudi_hw_init(struct hl_device *hdev)
 
 	gaudi_init_tpc_qmans(hdev);
 
-	gaudi_enable_clock_gating(hdev);
+	hdev->asic_funcs->set_clock_gating(hdev);
 
 	gaudi_enable_timestamp(hdev);
 
@@ -3104,7 +3127,9 @@ static void gaudi_hw_fini(struct hl_device *hdev, bool hard_reset)
 					HW_CAP_HBM_DMA | HW_CAP_PLL |
 					HW_CAP_MMU |
 					HW_CAP_SRAM_SCRAMBLER |
-					HW_CAP_HBM_SCRAMBLER);
+					HW_CAP_HBM_SCRAMBLER |
+					HW_CAP_CLK_GATE);
+
 	memset(gaudi->events_stat, 0, sizeof(gaudi->events_stat));
 }
 
@@ -3455,6 +3480,9 @@ static int gaudi_send_cpu_message(struct hl_device *hdev, u32 *msg,
 		return 0;
 	}
 
+	if (!timeout)
+		timeout = GAUDI_MSG_TO_CPU_TIMEOUT_USEC;
+
 	return hl_fw_send_cpu_message(hdev, GAUDI_QUEUE_ID_CPU_PQ, msg, len,
 						timeout, result);
 }
@@ -3790,6 +3818,25 @@ static int gaudi_validate_dma_pkt_no_mmu(struct hl_device *hdev,
 						src_in_host);
 }
 
+static int gaudi_validate_load_and_exe_pkt(struct hl_device *hdev,
+					struct hl_cs_parser *parser,
+					struct packet_load_and_exe *user_pkt)
+{
+	u32 cfg;
+
+	cfg = le32_to_cpu(user_pkt->cfg);
+
+	if (cfg & GAUDI_PKT_LOAD_AND_EXE_CFG_DST_MASK) {
+		dev_err(hdev->dev,
+			"User not allowed to use Load and Execute\n");
+		return -EPERM;
+	}
+
+	parser->patched_cb_size += sizeof(struct packet_load_and_exe);
+
+	return 0;
+}
+
 static int gaudi_validate_cb(struct hl_device *hdev,
 			struct hl_cs_parser *parser, bool is_mmu)
 {
@@ -3838,6 +3885,17 @@ static int gaudi_validate_cb(struct hl_device *hdev,
 			rc = -EPERM;
 			break;
 
+		case PACKET_WREG_BULK:
+			dev_err(hdev->dev,
+				"User not allowed to use WREG_BULK\n");
+			rc = -EPERM;
+			break;
+
+		case PACKET_LOAD_AND_EXE:
+			rc = gaudi_validate_load_and_exe_pkt(hdev, parser,
+				(struct packet_load_and_exe *) user_pkt);
+			break;
+
 		case PACKET_LIN_DMA:
 			parser->contains_dma_pkt = true;
 			if (is_mmu)
@@ -3848,14 +3906,12 @@ static int gaudi_validate_cb(struct hl_device *hdev,
 			break;
 
 		case PACKET_WREG_32:
-		case PACKET_WREG_BULK:
 		case PACKET_MSG_LONG:
 		case PACKET_MSG_SHORT:
 		case PACKET_REPEAT:
 		case PACKET_FENCE:
 		case PACKET_NOP:
 		case PACKET_ARB_POINT:
-		case PACKET_LOAD_AND_EXE:
 			parser->patched_cb_size += pkt_size;
 			break;
 
@@ -4490,13 +4546,18 @@ static int gaudi_debugfs_read32(struct hl_device *hdev, u64 addr, u32 *val)
 	int rc = 0;
 
 	if ((addr >= CFG_BASE) && (addr < CFG_BASE + CFG_SIZE)) {
-		if (gaudi->hw_cap_initialized & HW_CAP_CLK_GATE) {
+
+		if ((gaudi->hw_cap_initialized & HW_CAP_CLK_GATE) &&
+				(hdev->clock_gating_mask &
+						GAUDI_CLK_GATE_DEBUGFS_MASK)) {
+
 			dev_err_ratelimited(hdev->dev,
 				"Can't read register - clock gating is enabled!\n");
 			rc = -EFAULT;
 		} else {
 			*val = RREG32(addr - CFG_BASE);
 		}
+
 	} else if ((addr >= SRAM_BASE_ADDR) &&
 			(addr < SRAM_BASE_ADDR + SRAM_BAR_SIZE)) {
 		*val = readl(hdev->pcie_bar[SRAM_BAR_ID] +
@@ -4532,13 +4593,18 @@ static int gaudi_debugfs_write32(struct hl_device *hdev, u64 addr, u32 val)
 	int rc = 0;
 
 	if ((addr >= CFG_BASE) && (addr < CFG_BASE + CFG_SIZE)) {
-		if (gaudi->hw_cap_initialized & HW_CAP_CLK_GATE) {
+
+		if ((gaudi->hw_cap_initialized & HW_CAP_CLK_GATE) &&
+				(hdev->clock_gating_mask &
+						GAUDI_CLK_GATE_DEBUGFS_MASK)) {
+
 			dev_err_ratelimited(hdev->dev,
 				"Can't write register - clock gating is enabled!\n");
 			rc = -EFAULT;
 		} else {
 			WREG32(addr - CFG_BASE, val);
 		}
+
 	} else if ((addr >= SRAM_BASE_ADDR) &&
 			(addr < SRAM_BASE_ADDR + SRAM_BAR_SIZE)) {
 		writel(val, hdev->pcie_bar[SRAM_BAR_ID] +
@@ -4574,7 +4640,11 @@ static int gaudi_debugfs_read64(struct hl_device *hdev, u64 addr, u64 *val)
 	int rc = 0;
 
 	if ((addr >= CFG_BASE) && (addr <= CFG_BASE + CFG_SIZE - sizeof(u64))) {
-		if (gaudi->hw_cap_initialized & HW_CAP_CLK_GATE) {
+
+		if ((gaudi->hw_cap_initialized & HW_CAP_CLK_GATE) &&
+				(hdev->clock_gating_mask &
+						GAUDI_CLK_GATE_DEBUGFS_MASK)) {
+
 			dev_err_ratelimited(hdev->dev,
 				"Can't read register - clock gating is enabled!\n");
 			rc = -EFAULT;
@@ -4584,6 +4654,7 @@ static int gaudi_debugfs_read64(struct hl_device *hdev, u64 addr, u64 *val)
 
 			*val = (((u64) val_h) << 32) | val_l;
 		}
+
 	} else if ((addr >= SRAM_BASE_ADDR) &&
 		   (addr <= SRAM_BASE_ADDR + SRAM_BAR_SIZE - sizeof(u64))) {
 		*val = readq(hdev->pcie_bar[SRAM_BAR_ID] +
@@ -4620,7 +4691,11 @@ static int gaudi_debugfs_write64(struct hl_device *hdev, u64 addr, u64 val)
 	int rc = 0;
 
 	if ((addr >= CFG_BASE) && (addr <= CFG_BASE + CFG_SIZE - sizeof(u64))) {
-		if (gaudi->hw_cap_initialized & HW_CAP_CLK_GATE) {
+
+		if ((gaudi->hw_cap_initialized & HW_CAP_CLK_GATE) &&
+				(hdev->clock_gating_mask &
+						GAUDI_CLK_GATE_DEBUGFS_MASK)) {
+
 			dev_err_ratelimited(hdev->dev,
 				"Can't write register - clock gating is enabled!\n");
 			rc = -EFAULT;
@@ -4629,6 +4704,7 @@ static int gaudi_debugfs_write64(struct hl_device *hdev, u64 addr, u64 val)
 			WREG32(addr + sizeof(u32) - CFG_BASE,
 				upper_32_bits(val));
 		}
+
 	} else if ((addr >= SRAM_BASE_ADDR) &&
 		   (addr <= SRAM_BASE_ADDR + SRAM_BAR_SIZE - sizeof(u64))) {
 		writeq(val, hdev->pcie_bar[SRAM_BAR_ID] +
@@ -4850,7 +4926,7 @@ static void gaudi_mmu_prepare(struct hl_device *hdev, u32 asid)
 	gaudi_mmu_prepare_reg(hdev, mmPSOC_GLOBAL_CONF_TRACE_ARUSER, asid);
 	gaudi_mmu_prepare_reg(hdev, mmPSOC_GLOBAL_CONF_TRACE_AWUSER, asid);
 
-	hdev->asic_funcs->enable_clock_gating(hdev);
+	hdev->asic_funcs->set_clock_gating(hdev);
 
 	mutex_unlock(&gaudi->clk_gate_mutex);
 }
@@ -5231,7 +5307,7 @@ static void gaudi_print_ecc_info_generic(struct hl_device *hdev,
 	}
 
 	if (disable_clock_gating) {
-		hdev->asic_funcs->enable_clock_gating(hdev);
+		hdev->asic_funcs->set_clock_gating(hdev);
 		mutex_unlock(&gaudi->clk_gate_mutex);
 	}
 }
@@ -5718,7 +5794,7 @@ static bool gaudi_tpc_read_interrupts(struct hl_device *hdev, u8 tpc_id,
 	/* Clear interrupts */
 	WREG32(mmTPC0_CFG_TPC_INTR_CAUSE + tpc_offset, 0);
 
-	hdev->asic_funcs->enable_clock_gating(hdev);
+	hdev->asic_funcs->set_clock_gating(hdev);
 
 	mutex_unlock(&gaudi->clk_gate_mutex);
 
@@ -5994,6 +6070,8 @@ static int gaudi_mmu_invalidate_cache(struct hl_device *hdev, bool is_hard,
 	mutex_lock(&hdev->mmu_cache_lock);
 
 	/* L0 & L1 invalidation */
+	WREG32(mmSTLB_INV_PS, 3);
+	WREG32(mmSTLB_CACHE_INV, gaudi->mmu_cache_inv_pi++);
 	WREG32(mmSTLB_INV_PS, 2);
 
 	rc = hl_poll_timeout(
@@ -6232,7 +6310,7 @@ static bool gaudi_is_device_idle(struct hl_device *hdev, u32 *mask,
 	if (s)
 		seq_puts(s, "\n");
 
-	hdev->asic_funcs->enable_clock_gating(hdev);
+	hdev->asic_funcs->set_clock_gating(hdev);
 
 	mutex_unlock(&gaudi->clk_gate_mutex);
 
@@ -6333,7 +6411,7 @@ static int gaudi_run_tpc_kernel(struct hl_device *hdev, u64 tpc_kernel,
 		dev_err(hdev->dev,
 			"Timeout while waiting for TPC%d icache prefetch\n",
 			tpc_id);
-		hdev->asic_funcs->enable_clock_gating(hdev);
+		hdev->asic_funcs->set_clock_gating(hdev);
 		mutex_unlock(&gaudi->clk_gate_mutex);
 		return -EIO;
 	}
@@ -6362,7 +6440,7 @@ static int gaudi_run_tpc_kernel(struct hl_device *hdev, u64 tpc_kernel,
 		1000,
 		kernel_timeout);
 
-	hdev->asic_funcs->enable_clock_gating(hdev);
+	hdev->asic_funcs->set_clock_gating(hdev);
 	mutex_unlock(&gaudi->clk_gate_mutex);
 
 	if (rc) {
@@ -6703,7 +6781,7 @@ static const struct hl_asic_funcs gaudi_funcs = {
 	.mmu_invalidate_cache = gaudi_mmu_invalidate_cache,
 	.mmu_invalidate_cache_range = gaudi_mmu_invalidate_cache_range,
 	.send_heartbeat = gaudi_send_heartbeat,
-	.enable_clock_gating = gaudi_enable_clock_gating,
+	.set_clock_gating = gaudi_set_clock_gating,
 	.disable_clock_gating = gaudi_disable_clock_gating,
 	.debug_coresight = gaudi_debug_coresight,
 	.is_device_idle = gaudi_is_device_idle,
diff --git a/drivers/misc/habanalabs/gaudi/gaudiP.h b/drivers/misc/habanalabs/gaudi/gaudiP.h
index a46530d375fa..41a8d9bff6bf 100644
--- a/drivers/misc/habanalabs/gaudi/gaudiP.h
+++ b/drivers/misc/habanalabs/gaudi/gaudiP.h
@@ -229,6 +229,8 @@ struct gaudi_internal_qman_info {
  * @multi_msi_mode: whether we are working in multi MSI single MSI mode.
  *                  Multi MSI is possible only with IOMMU enabled.
  * @ext_queue_idx: helper index for external queues initialization.
+ * @mmu_cache_inv_pi: PI for MMU cache invalidation flow. The H/W expects an
+ *                    8-bit value so use u8.
  */
 struct gaudi_device {
 	int (*armcp_info_get)(struct hl_device *hdev);
@@ -248,6 +250,7 @@ struct gaudi_device {
 	u32				hw_cap_initialized;
 	u8				multi_msi_mode;
 	u8				ext_queue_idx;
+	u8				mmu_cache_inv_pi;
 };
 
 void gaudi_init_security(struct hl_device *hdev);
diff --git a/drivers/misc/habanalabs/goya/goya.c b/drivers/misc/habanalabs/goya/goya.c
index 0d2952bb58df..88460b2138d8 100644
--- a/drivers/misc/habanalabs/goya/goya.c
+++ b/drivers/misc/habanalabs/goya/goya.c
@@ -88,6 +88,7 @@
 #define GOYA_PLDM_MMU_TIMEOUT_USEC	(MMU_CONFIG_TIMEOUT_USEC * 100)
 #define GOYA_PLDM_QMAN0_TIMEOUT_USEC	(HL_DEVICE_TIMEOUT_USEC * 30)
 #define GOYA_BOOT_FIT_REQ_TIMEOUT_USEC	1000000		/* 1s */
+#define GOYA_MSG_TO_CPU_TIMEOUT_USEC	4000000		/* 4s */
 
 #define GOYA_QMAN0_FENCE_VAL		0xD169B243
 
@@ -2830,6 +2831,9 @@ int goya_send_cpu_message(struct hl_device *hdev, u32 *msg, u16 len,
 		return 0;
 	}
 
+	if (!timeout)
+		timeout = GOYA_MSG_TO_CPU_TIMEOUT_USEC;
+
 	return hl_fw_send_cpu_message(hdev, GOYA_QUEUE_ID_CPU_PQ, msg, len,
 					timeout, result);
 }
@@ -4431,8 +4435,8 @@ static int goya_unmask_irq_arr(struct hl_device *hdev, u32 *irq_arr,
 	pkt->armcp_pkt.ctl = cpu_to_le32(ARMCP_PACKET_UNMASK_RAZWI_IRQ_ARRAY <<
 						ARMCP_PKT_CTL_OPCODE_SHIFT);
 
-	rc = goya_send_cpu_message(hdev, (u32 *) pkt, total_pkt_size,
-			HL_DEVICE_TIMEOUT_USEC, &result);
+	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) pkt,
+						total_pkt_size,	0, &result);
 
 	if (rc)
 		dev_err(hdev->dev, "failed to unmask IRQ array\n");
@@ -4464,8 +4468,8 @@ static int goya_unmask_irq(struct hl_device *hdev, u16 event_type)
 				ARMCP_PKT_CTL_OPCODE_SHIFT);
 	pkt.value = cpu_to_le64(event_type);
 
-	rc = goya_send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-			HL_DEVICE_TIMEOUT_USEC, &result);
+	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
+						0, &result);
 
 	if (rc)
 		dev_err(hdev->dev, "failed to unmask RAZWI IRQ %d", event_type);
@@ -5028,14 +5032,14 @@ int goya_armcp_info_get(struct hl_device *hdev)
 	return 0;
 }
 
-static void goya_enable_clock_gating(struct hl_device *hdev)
+static void goya_set_clock_gating(struct hl_device *hdev)
 {
-
+	/* clock gating not supported in Goya */
 }
 
 static void goya_disable_clock_gating(struct hl_device *hdev)
 {
-
+	/* clock gating not supported in Goya */
 }
 
 static bool goya_is_device_idle(struct hl_device *hdev, u32 *mask,
@@ -5259,7 +5263,7 @@ static const struct hl_asic_funcs goya_funcs = {
 	.mmu_invalidate_cache = goya_mmu_invalidate_cache,
 	.mmu_invalidate_cache_range = goya_mmu_invalidate_cache_range,
 	.send_heartbeat = goya_send_heartbeat,
-	.enable_clock_gating = goya_enable_clock_gating,
+	.set_clock_gating = goya_set_clock_gating,
 	.disable_clock_gating = goya_disable_clock_gating,
 	.debug_coresight = goya_debug_coresight,
 	.is_device_idle = goya_is_device_idle,
diff --git a/drivers/misc/habanalabs/habanalabs.h b/drivers/misc/habanalabs/habanalabs.h
index 1ecdcf8b763a..194d83352696 100644
--- a/drivers/misc/habanalabs/habanalabs.h
+++ b/drivers/misc/habanalabs/habanalabs.h
@@ -578,8 +578,9 @@ enum hl_pll_frequency {
  * @mmu_invalidate_cache_range: flush specific MMU STLB cache lines with
  *                              ASID-VA-size mask.
  * @send_heartbeat: send is-alive packet to ArmCP and verify response.
- * @enable_clock_gating: enable clock gating for reducing power consumption.
- * @disable_clock_gating: disable clock for accessing registers on HBW.
+ * @set_clock_gating: enable/disable clock gating per engine according to
+ *                    clock gating mask in hdev
+ * @disable_clock_gating: disable clock gating completely
  * @debug_coresight: perform certain actions on Coresight for debugging.
  * @is_device_idle: return true if device is idle, false otherwise.
  * @soft_reset_late_init: perform certain actions needed after soft reset.
@@ -587,7 +588,11 @@ enum hl_pll_frequency {
  * @hw_queues_unlock: release H/W queues lock.
  * @get_pci_id: retrieve PCI ID.
  * @get_eeprom_data: retrieve EEPROM data from F/W.
- * @send_cpu_message: send buffer to ArmCP.
+ * @send_cpu_message: send message to F/W. If the message is timedout, the
+ *                    driver will eventually reset the device. The timeout can
+ *                    be determined by the calling function or it can be 0 and
+ *                    then the timeout is the default timeout for the specific
+ *                    ASIC
  * @get_hw_state: retrieve the H/W state
  * @pci_bars_map: Map PCI BARs.
  * @set_dram_bar_base: Set DRAM BAR to map specific device address. Returns
@@ -680,7 +685,7 @@ struct hl_asic_funcs {
 	int (*mmu_invalidate_cache_range)(struct hl_device *hdev, bool is_hard,
 			u32 asid, u64 va, u64 size);
 	int (*send_heartbeat)(struct hl_device *hdev);
-	void (*enable_clock_gating)(struct hl_device *hdev);
+	void (*set_clock_gating)(struct hl_device *hdev);
 	void (*disable_clock_gating)(struct hl_device *hdev);
 	int (*debug_coresight)(struct hl_device *hdev, void *data);
 	bool (*is_device_idle)(struct hl_device *hdev, u32 *mask,
@@ -1398,6 +1403,9 @@ struct hl_device_idle_busy_ts {
  * @max_power: the max power of the device, as configured by the sysadmin. This
  *             value is saved so in case of hard-reset, the driver will restore
  *             this value and update the F/W after the re-initialization
+ * @clock_gating_mask: is clock gating enabled. bitmask that represents the
+ *                     different engines. See debugfs-driver-habanalabs for
+ *                     details.
  * @in_reset: is device in reset flow.
  * @curr_pll_profile: current PLL profile.
  * @cs_active_cnt: number of active command submissions on this device (active
@@ -1425,7 +1433,6 @@ struct hl_device_idle_busy_ts {
  * @init_done: is the initialization of the device done.
  * @mmu_enable: is MMU enabled.
  * @mmu_huge_page_opt: is MMU huge pages optimization enabled.
- * @clock_gating: is clock gating enabled.
  * @device_cpu_disabled: is the device CPU disabled (due to timeouts)
  * @dma_mask: the dma mask that was set for this device
  * @in_debug: is device under debug. This, together with fpriv_list, enforces
@@ -1493,6 +1500,7 @@ struct hl_device {
 	atomic64_t			dram_used_mem;
 	u64				timeout_jiffies;
 	u64				max_power;
+	u64				clock_gating_mask;
 	atomic_t			in_reset;
 	enum hl_pll_frequency		curr_pll_profile;
 	int				cs_active_cnt;
@@ -1514,7 +1522,6 @@ struct hl_device {
 	u8				dram_default_page_mapping;
 	u8				pmmu_huge_range;
 	u8				init_done;
-	u8				clock_gating;
 	u8				device_cpu_disabled;
 	u8				dma_mask;
 	u8				in_debug;
diff --git a/drivers/misc/habanalabs/habanalabs_drv.c b/drivers/misc/habanalabs/habanalabs_drv.c
index 8652c7e5d7f1..22716da9f85f 100644
--- a/drivers/misc/habanalabs/habanalabs_drv.c
+++ b/drivers/misc/habanalabs/habanalabs_drv.c
@@ -232,7 +232,7 @@ static void set_driver_behavior_per_device(struct hl_device *hdev)
 	hdev->fw_loading = 1;
 	hdev->cpu_queues_enable = 1;
 	hdev->heartbeat = 1;
-	hdev->clock_gating = 1;
+	hdev->clock_gating_mask = ULONG_MAX;
 
 	hdev->reset_pcilink = 0;
 	hdev->axi_drain = 0;
diff --git a/drivers/misc/habanalabs/hwmon.c b/drivers/misc/habanalabs/hwmon.c
index 8c6cd77e6af6..b997336fa75f 100644
--- a/drivers/misc/habanalabs/hwmon.c
+++ b/drivers/misc/habanalabs/hwmon.c
@@ -10,7 +10,6 @@
 #include <linux/pci.h>
 #include <linux/hwmon.h>
 
-#define SENSORS_PKT_TIMEOUT		1000000	/* 1s */
 #define HWMON_NR_SENSOR_TYPES		(hwmon_pwm + 1)
 
 int hl_build_hwmon_channel_info(struct hl_device *hdev,
@@ -323,7 +322,7 @@ int hl_get_temperature(struct hl_device *hdev,
 	pkt.type = __cpu_to_le16(attr);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-			SENSORS_PKT_TIMEOUT, value);
+						0, value);
 
 	if (rc) {
 		dev_err(hdev->dev,
@@ -350,7 +349,7 @@ int hl_set_temperature(struct hl_device *hdev,
 	pkt.value = __cpu_to_le64(value);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-						SENSORS_PKT_TIMEOUT, NULL);
+						0, NULL);
 
 	if (rc)
 		dev_err(hdev->dev,
@@ -374,7 +373,7 @@ int hl_get_voltage(struct hl_device *hdev,
 	pkt.type = __cpu_to_le16(attr);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-					SENSORS_PKT_TIMEOUT, value);
+						0, value);
 
 	if (rc) {
 		dev_err(hdev->dev,
@@ -400,7 +399,7 @@ int hl_get_current(struct hl_device *hdev,
 	pkt.type = __cpu_to_le16(attr);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-					SENSORS_PKT_TIMEOUT, value);
+						0, value);
 
 	if (rc) {
 		dev_err(hdev->dev,
@@ -426,7 +425,7 @@ int hl_get_fan_speed(struct hl_device *hdev,
 	pkt.type = __cpu_to_le16(attr);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-					SENSORS_PKT_TIMEOUT, value);
+						0, value);
 
 	if (rc) {
 		dev_err(hdev->dev,
@@ -452,7 +451,7 @@ int hl_get_pwm_info(struct hl_device *hdev,
 	pkt.type = __cpu_to_le16(attr);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-					SENSORS_PKT_TIMEOUT, value);
+						0, value);
 
 	if (rc) {
 		dev_err(hdev->dev,
@@ -479,7 +478,7 @@ void hl_set_pwm_info(struct hl_device *hdev, int sensor_index, u32 attr,
 	pkt.value = cpu_to_le64(value);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-					SENSORS_PKT_TIMEOUT, NULL);
+						0, NULL);
 
 	if (rc)
 		dev_err(hdev->dev,
@@ -502,7 +501,7 @@ int hl_set_voltage(struct hl_device *hdev,
 	pkt.value = __cpu_to_le64(value);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-						SENSORS_PKT_TIMEOUT, NULL);
+						0, NULL);
 
 	if (rc)
 		dev_err(hdev->dev,
@@ -527,7 +526,7 @@ int hl_set_current(struct hl_device *hdev,
 	pkt.value = __cpu_to_le64(value);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-						SENSORS_PKT_TIMEOUT, NULL);
+						0, NULL);
 
 	if (rc)
 		dev_err(hdev->dev,
diff --git a/drivers/misc/habanalabs/include/gaudi/gaudi_packets.h b/drivers/misc/habanalabs/include/gaudi/gaudi_packets.h
index 9a5800b0086b..0f0cd067bb43 100644
--- a/drivers/misc/habanalabs/include/gaudi/gaudi_packets.h
+++ b/drivers/misc/habanalabs/include/gaudi/gaudi_packets.h
@@ -197,6 +197,9 @@ struct packet_wait {
 	__le32 ctl;
 };
 
+#define GAUDI_PKT_LOAD_AND_EXE_CFG_DST_SHIFT	0
+#define GAUDI_PKT_LOAD_AND_EXE_CFG_DST_MASK	0x00000001
+
 struct packet_load_and_exe {
 	__le32 cfg;
 	__le32 ctl;
diff --git a/drivers/misc/habanalabs/sysfs.c b/drivers/misc/habanalabs/sysfs.c
index 5d78d5e1c782..70b6b1863c2e 100644
--- a/drivers/misc/habanalabs/sysfs.c
+++ b/drivers/misc/habanalabs/sysfs.c
@@ -9,9 +9,6 @@
 
 #include <linux/pci.h>
 
-#define SET_CLK_PKT_TIMEOUT	1000000	/* 1s */
-#define SET_PWR_PKT_TIMEOUT	1000000	/* 1s */
-
 long hl_get_frequency(struct hl_device *hdev, u32 pll_index, bool curr)
 {
 	struct armcp_packet pkt;
@@ -29,7 +26,7 @@ long hl_get_frequency(struct hl_device *hdev, u32 pll_index, bool curr)
 	pkt.pll_index = cpu_to_le32(pll_index);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-						SET_CLK_PKT_TIMEOUT, &result);
+						0, &result);
 
 	if (rc) {
 		dev_err(hdev->dev,
@@ -54,7 +51,7 @@ void hl_set_frequency(struct hl_device *hdev, u32 pll_index, u64 freq)
 	pkt.value = cpu_to_le64(freq);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-					SET_CLK_PKT_TIMEOUT, NULL);
+						0, NULL);
 
 	if (rc)
 		dev_err(hdev->dev,
@@ -74,7 +71,7 @@ u64 hl_get_max_power(struct hl_device *hdev)
 				ARMCP_PKT_CTL_OPCODE_SHIFT);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-						SET_PWR_PKT_TIMEOUT, &result);
+						0, &result);
 
 	if (rc) {
 		dev_err(hdev->dev, "Failed to get max power, error %d\n", rc);
@@ -96,7 +93,7 @@ void hl_set_max_power(struct hl_device *hdev, u64 value)
 	pkt.value = cpu_to_le64(value);
 
 	rc = hdev->asic_funcs->send_cpu_message(hdev, (u32 *) &pkt, sizeof(pkt),
-					SET_PWR_PKT_TIMEOUT, NULL);
+						0, NULL);
 
 	if (rc)
 		dev_err(hdev->dev, "Failed to set max power, error %d\n", rc);
diff --git a/drivers/misc/kgdbts.c b/drivers/misc/kgdbts.c
index bccd341e9ae1..d5d2af4d10e6 100644
--- a/drivers/misc/kgdbts.c
+++ b/drivers/misc/kgdbts.c
@@ -828,7 +828,7 @@ static void run_plant_and_detach_test(int is_early)
 	char before[BREAK_INSTR_SIZE];
 	char after[BREAK_INSTR_SIZE];
 
-	probe_kernel_read(before, (char *)kgdbts_break_test,
+	copy_from_kernel_nofault(before, (char *)kgdbts_break_test,
 	  BREAK_INSTR_SIZE);
 	init_simple_test();
 	ts.tst = plant_and_detach_test;
@@ -836,8 +836,8 @@ static void run_plant_and_detach_test(int is_early)
 	/* Activate test with initial breakpoint */
 	if (!is_early)
 		kgdb_breakpoint();
-	probe_kernel_read(after, (char *)kgdbts_break_test,
-	  BREAK_INSTR_SIZE);
+	copy_from_kernel_nofault(after, (char *)kgdbts_break_test,
+			BREAK_INSTR_SIZE);
 	if (memcmp(before, after, BREAK_INSTR_SIZE)) {
 		printk(KERN_CRIT "kgdbts: ERROR kgdb corrupted memory\n");
 		panic("kgdb memory corruption");
diff --git a/drivers/misc/mei/bus.c b/drivers/misc/mei/bus.c
index 8d468e0a950a..f476dbc7252b 100644
--- a/drivers/misc/mei/bus.c
+++ b/drivers/misc/mei/bus.c
@@ -745,9 +745,8 @@ static int mei_cl_device_remove(struct device *dev)
 
 	mei_cl_bus_module_put(cldev);
 	module_put(THIS_MODULE);
-	dev->driver = NULL;
-	return ret;
 
+	return ret;
 }
 
 static ssize_t name_show(struct device *dev, struct device_attribute *a,
diff --git a/drivers/misc/mei/hw-me-regs.h b/drivers/misc/mei/hw-me-regs.h
index 9392934e3a06..7becfc768bbc 100644
--- a/drivers/misc/mei/hw-me-regs.h
+++ b/drivers/misc/mei/hw-me-regs.h
@@ -94,6 +94,7 @@
 #define MEI_DEV_ID_JSP_N      0x4DE0  /* Jasper Lake Point N */
 
 #define MEI_DEV_ID_TGP_LP     0xA0E0  /* Tiger Lake Point LP */
+#define MEI_DEV_ID_TGP_H      0x43E0  /* Tiger Lake Point H */
 
 #define MEI_DEV_ID_MCC        0x4B70  /* Mule Creek Canyon (EHL) */
 #define MEI_DEV_ID_MCC_4      0x4B75  /* Mule Creek Canyon 4 (EHL) */
@@ -107,6 +108,8 @@
 #  define PCI_CFG_HFS_1_D0I3_MSK     0x80000000
 #define PCI_CFG_HFS_2         0x48
 #define PCI_CFG_HFS_3         0x60
+#  define PCI_CFG_HFS_3_FW_SKU_MSK   0x00000070
+#  define PCI_CFG_HFS_3_FW_SKU_SPS   0x00000060
 #define PCI_CFG_HFS_4         0x64
 #define PCI_CFG_HFS_5         0x68
 #define PCI_CFG_HFS_6         0x6C
diff --git a/drivers/misc/mei/hw-me.c b/drivers/misc/mei/hw-me.c
index f620442addf5..7649710a2ab9 100644
--- a/drivers/misc/mei/hw-me.c
+++ b/drivers/misc/mei/hw-me.c
@@ -1366,7 +1366,7 @@ static bool mei_me_fw_type_nm(struct pci_dev *pdev)
 #define MEI_CFG_FW_NM                           \
 	.quirk_probe = mei_me_fw_type_nm
 
-static bool mei_me_fw_type_sps(struct pci_dev *pdev)
+static bool mei_me_fw_type_sps_4(struct pci_dev *pdev)
 {
 	u32 reg;
 	unsigned int devfn;
@@ -1382,7 +1382,36 @@ static bool mei_me_fw_type_sps(struct pci_dev *pdev)
 	return (reg & 0xf0000) == 0xf0000;
 }
 
-#define MEI_CFG_FW_SPS                           \
+#define MEI_CFG_FW_SPS_4                          \
+	.quirk_probe = mei_me_fw_type_sps_4
+
+/**
+ * mei_me_fw_sku_sps() - check for sps sku
+ *
+ * Read ME FW Status register to check for SPS Firmware.
+ * The SPS FW is only signaled in pci function 0
+ *
+ * @pdev: pci device
+ *
+ * Return: true in case of SPS firmware
+ */
+static bool mei_me_fw_type_sps(struct pci_dev *pdev)
+{
+	u32 reg;
+	u32 fw_type;
+	unsigned int devfn;
+
+	devfn = PCI_DEVFN(PCI_SLOT(pdev->devfn), 0);
+	pci_bus_read_config_dword(pdev->bus, devfn, PCI_CFG_HFS_3, &reg);
+	trace_mei_pci_cfg_read(&pdev->dev, "PCI_CFG_HFS_3", PCI_CFG_HFS_3, reg);
+	fw_type = (reg & PCI_CFG_HFS_3_FW_SKU_MSK);
+
+	dev_dbg(&pdev->dev, "fw type is %d\n", fw_type);
+
+	return fw_type == PCI_CFG_HFS_3_FW_SKU_SPS;
+}
+
+#define MEI_CFG_FW_SPS                          \
 	.quirk_probe = mei_me_fw_type_sps
 
 #define MEI_CFG_FW_VER_SUPP                     \
@@ -1452,10 +1481,17 @@ static const struct mei_cfg mei_me_pch8_cfg = {
 };
 
 /* PCH8 Lynx Point with quirk for SPS Firmware exclusion */
-static const struct mei_cfg mei_me_pch8_sps_cfg = {
+static const struct mei_cfg mei_me_pch8_sps_4_cfg = {
 	MEI_CFG_PCH8_HFS,
 	MEI_CFG_FW_VER_SUPP,
-	MEI_CFG_FW_SPS,
+	MEI_CFG_FW_SPS_4,
+};
+
+/* LBG with quirk for SPS (4.0) Firmware exclusion */
+static const struct mei_cfg mei_me_pch12_sps_4_cfg = {
+	MEI_CFG_PCH8_HFS,
+	MEI_CFG_FW_VER_SUPP,
+	MEI_CFG_FW_SPS_4,
 };
 
 /* Cannon Lake and newer devices */
@@ -1465,10 +1501,20 @@ static const struct mei_cfg mei_me_pch12_cfg = {
 	MEI_CFG_DMA_128,
 };
 
-/* LBG with quirk for SPS Firmware exclusion */
+/* Cannon Lake with quirk for SPS 5.0 and newer Firmware exclusion */
 static const struct mei_cfg mei_me_pch12_sps_cfg = {
 	MEI_CFG_PCH8_HFS,
 	MEI_CFG_FW_VER_SUPP,
+	MEI_CFG_DMA_128,
+	MEI_CFG_FW_SPS,
+};
+
+/* Cannon Lake with quirk for SPS 5.0 and newer Firmware exclusion
+ * w/o DMA support
+ */
+static const struct mei_cfg mei_me_pch12_nodma_sps_cfg = {
+	MEI_CFG_PCH8_HFS,
+	MEI_CFG_FW_VER_SUPP,
 	MEI_CFG_FW_SPS,
 };
 
@@ -1480,6 +1526,15 @@ static const struct mei_cfg mei_me_pch15_cfg = {
 	MEI_CFG_TRC,
 };
 
+/* Tiger Lake with quirk for SPS 5.0 and newer Firmware exclusion */
+static const struct mei_cfg mei_me_pch15_sps_cfg = {
+	MEI_CFG_PCH8_HFS,
+	MEI_CFG_FW_VER_SUPP,
+	MEI_CFG_DMA_128,
+	MEI_CFG_TRC,
+	MEI_CFG_FW_SPS,
+};
+
 /*
  * mei_cfg_list - A list of platform platform specific configurations.
  * Note: has to be synchronized with  enum mei_cfg_idx.
@@ -1492,10 +1547,13 @@ static const struct mei_cfg *const mei_cfg_list[] = {
 	[MEI_ME_PCH7_CFG] = &mei_me_pch7_cfg,
 	[MEI_ME_PCH_CPT_PBG_CFG] = &mei_me_pch_cpt_pbg_cfg,
 	[MEI_ME_PCH8_CFG] = &mei_me_pch8_cfg,
-	[MEI_ME_PCH8_SPS_CFG] = &mei_me_pch8_sps_cfg,
+	[MEI_ME_PCH8_SPS_4_CFG] = &mei_me_pch8_sps_4_cfg,
 	[MEI_ME_PCH12_CFG] = &mei_me_pch12_cfg,
+	[MEI_ME_PCH12_SPS_4_CFG] = &mei_me_pch12_sps_4_cfg,
 	[MEI_ME_PCH12_SPS_CFG] = &mei_me_pch12_sps_cfg,
+	[MEI_ME_PCH12_SPS_NODMA_CFG] = &mei_me_pch12_nodma_sps_cfg,
 	[MEI_ME_PCH15_CFG] = &mei_me_pch15_cfg,
+	[MEI_ME_PCH15_SPS_CFG] = &mei_me_pch15_sps_cfg,
 };
 
 const struct mei_cfg *mei_me_get_cfg(kernel_ulong_t idx)
diff --git a/drivers/misc/mei/hw-me.h b/drivers/misc/mei/hw-me.h
index b6b94e211464..6a8973649c49 100644
--- a/drivers/misc/mei/hw-me.h
+++ b/drivers/misc/mei/hw-me.h
@@ -1,6 +1,6 @@
 /* SPDX-License-Identifier: GPL-2.0 */
 /*
- * Copyright (c) 2012-2019, Intel Corporation. All rights reserved.
+ * Copyright (c) 2012-2020, Intel Corporation. All rights reserved.
  * Intel Management Engine Interface (Intel MEI) Linux driver
  */
 
@@ -76,14 +76,20 @@ struct mei_me_hw {
  *                         with quirk for Node Manager exclusion.
  * @MEI_ME_PCH8_CFG:       Platform Controller Hub Gen8 and newer
  *                         client platforms.
- * @MEI_ME_PCH8_SPS_CFG:   Platform Controller Hub Gen8 and newer
+ * @MEI_ME_PCH8_SPS_4_CFG: Platform Controller Hub Gen8 and newer
  *                         servers platforms with quirk for
  *                         SPS firmware exclusion.
  * @MEI_ME_PCH12_CFG:      Platform Controller Hub Gen12 and newer
- * @MEI_ME_PCH12_SPS_CFG:  Platform Controller Hub Gen12 and newer
+ * @MEI_ME_PCH12_SPS_4_CFG:Platform Controller Hub Gen12 up to 4.0
+ *                         servers platforms with quirk for
+ *                         SPS firmware exclusion.
+ * @MEI_ME_PCH12_SPS_CFG:  Platform Controller Hub Gen12 5.0 and newer
  *                         servers platforms with quirk for
  *                         SPS firmware exclusion.
  * @MEI_ME_PCH15_CFG:      Platform Controller Hub Gen15 and newer
+ * @MEI_ME_PCH15_SPS_CFG:  Platform Controller Hub Gen15 and newer
+ *                         servers platforms with quirk for
+ *                         SPS firmware exclusion.
  * @MEI_ME_NUM_CFG:        Upper Sentinel.
  */
 enum mei_cfg_idx {
@@ -94,10 +100,13 @@ enum mei_cfg_idx {
 	MEI_ME_PCH7_CFG,
 	MEI_ME_PCH_CPT_PBG_CFG,
 	MEI_ME_PCH8_CFG,
-	MEI_ME_PCH8_SPS_CFG,
+	MEI_ME_PCH8_SPS_4_CFG,
 	MEI_ME_PCH12_CFG,
+	MEI_ME_PCH12_SPS_4_CFG,
 	MEI_ME_PCH12_SPS_CFG,
+	MEI_ME_PCH12_SPS_NODMA_CFG,
 	MEI_ME_PCH15_CFG,
+	MEI_ME_PCH15_SPS_CFG,
 	MEI_ME_NUM_CFG,
 };
 
diff --git a/drivers/misc/mei/pci-me.c b/drivers/misc/mei/pci-me.c
index 71f795b510ce..2a3f2fd5df50 100644
--- a/drivers/misc/mei/pci-me.c
+++ b/drivers/misc/mei/pci-me.c
@@ -59,18 +59,18 @@ static const struct pci_device_id mei_me_pci_tbl[] = {
 	{MEI_PCI_DEVICE(MEI_DEV_ID_PPT_1, MEI_ME_PCH7_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_PPT_2, MEI_ME_PCH7_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_PPT_3, MEI_ME_PCH7_CFG)},
-	{MEI_PCI_DEVICE(MEI_DEV_ID_LPT_H, MEI_ME_PCH8_SPS_CFG)},
-	{MEI_PCI_DEVICE(MEI_DEV_ID_LPT_W, MEI_ME_PCH8_SPS_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_LPT_H, MEI_ME_PCH8_SPS_4_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_LPT_W, MEI_ME_PCH8_SPS_4_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_LPT_LP, MEI_ME_PCH8_CFG)},
-	{MEI_PCI_DEVICE(MEI_DEV_ID_LPT_HR, MEI_ME_PCH8_SPS_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_LPT_HR, MEI_ME_PCH8_SPS_4_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_WPT_LP, MEI_ME_PCH8_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_WPT_LP_2, MEI_ME_PCH8_CFG)},
 
 	{MEI_PCI_DEVICE(MEI_DEV_ID_SPT, MEI_ME_PCH8_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_SPT_2, MEI_ME_PCH8_CFG)},
-	{MEI_PCI_DEVICE(MEI_DEV_ID_SPT_H, MEI_ME_PCH8_SPS_CFG)},
-	{MEI_PCI_DEVICE(MEI_DEV_ID_SPT_H_2, MEI_ME_PCH8_SPS_CFG)},
-	{MEI_PCI_DEVICE(MEI_DEV_ID_LBG, MEI_ME_PCH12_SPS_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_SPT_H, MEI_ME_PCH8_SPS_4_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_SPT_H_2, MEI_ME_PCH8_SPS_4_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_LBG, MEI_ME_PCH12_SPS_4_CFG)},
 
 	{MEI_PCI_DEVICE(MEI_DEV_ID_BXT_M, MEI_ME_PCH8_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_APL_I, MEI_ME_PCH8_CFG)},
@@ -84,8 +84,8 @@ static const struct pci_device_id mei_me_pci_tbl[] = {
 
 	{MEI_PCI_DEVICE(MEI_DEV_ID_CNP_LP, MEI_ME_PCH12_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_CNP_LP_3, MEI_ME_PCH8_CFG)},
-	{MEI_PCI_DEVICE(MEI_DEV_ID_CNP_H, MEI_ME_PCH12_CFG)},
-	{MEI_PCI_DEVICE(MEI_DEV_ID_CNP_H_3, MEI_ME_PCH8_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_CNP_H, MEI_ME_PCH12_SPS_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_CNP_H_3, MEI_ME_PCH12_SPS_NODMA_CFG)},
 
 	{MEI_PCI_DEVICE(MEI_DEV_ID_CMP_LP, MEI_ME_PCH12_CFG)},
 	{MEI_PCI_DEVICE(MEI_DEV_ID_CMP_LP_3, MEI_ME_PCH8_CFG)},
@@ -96,6 +96,7 @@ static const struct pci_device_id mei_me_pci_tbl[] = {
 	{MEI_PCI_DEVICE(MEI_DEV_ID_ICP_LP, MEI_ME_PCH12_CFG)},
 
 	{MEI_PCI_DEVICE(MEI_DEV_ID_TGP_LP, MEI_ME_PCH15_CFG)},
+	{MEI_PCI_DEVICE(MEI_DEV_ID_TGP_H, MEI_ME_PCH15_SPS_CFG)},
 
 	{MEI_PCI_DEVICE(MEI_DEV_ID_JSP_N, MEI_ME_PCH15_CFG)},