crypto: qat - add support for 420xx devices

Add support for 420xx devices by including a new device driver that
supports such devices, updates to the firmware loader and capabilities.

Compared to 4xxx devices, 420xx devices have more acceleration engines
(16 service engines and 1 admin) and support the wireless cipher
algorithms ZUC and Snow 3G.
Signed-off-by: Jie Wang <jie.wang@intel.com>
Co-developed-by: Dong Xie <dong.xie@intel.com>
Signed-off-by: Dong Xie <dong.xie@intel.com>
Reviewed-by: Giovanni Cabiddu <giovanni.cabiddu@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

drivers/crypto/intel/qat/Kconfig

@@ -59,6 +59,17 @@ config CRYPTO_DEV_QAT_4XXX
 	  To compile this as a module, choose M here: the module
 	  will be called qat_4xxx.
 
+config CRYPTO_DEV_QAT_420XX
+	tristate "Support for Intel(R) QAT_420XX"
+	depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST)
+	select CRYPTO_DEV_QAT
+	help
+	  Support for Intel(R) QuickAssist Technology QAT_420xx
+	  for accelerating crypto and compression workloads.
+
+	  To compile this as a module, choose M here: the module
+	  will be called qat_420xx.
+
 config CRYPTO_DEV_QAT_DH895xCCVF
 	tristate "Support for Intel(R) DH895xCC Virtual Function"
 	depends on PCI && (!CPU_BIG_ENDIAN || COMPILE_TEST)

drivers/crypto/intel/qat/Makefile

@@ -4,6 +4,7 @@ obj-$(CONFIG_CRYPTO_DEV_QAT_DH895xCC) += qat_dh895xcc/
 obj-$(CONFIG_CRYPTO_DEV_QAT_C3XXX) += qat_c3xxx/
 obj-$(CONFIG_CRYPTO_DEV_QAT_C62X) += qat_c62x/
 obj-$(CONFIG_CRYPTO_DEV_QAT_4XXX) += qat_4xxx/
+obj-$(CONFIG_CRYPTO_DEV_QAT_420XX) += qat_420xx/
 obj-$(CONFIG_CRYPTO_DEV_QAT_DH895xCCVF) += qat_dh895xccvf/
 obj-$(CONFIG_CRYPTO_DEV_QAT_C3XXXVF) += qat_c3xxxvf/
 obj-$(CONFIG_CRYPTO_DEV_QAT_C62XVF) += qat_c62xvf/

drivers/crypto/intel/qat/qat_420xx/Makefile

+# SPDX-License-Identifier: GPL-2.0-only
+ccflags-y := -I $(srctree)/$(src)/../qat_common
+obj-$(CONFIG_CRYPTO_DEV_QAT_420XX) += qat_420xx.o
+qat_420xx-objs := adf_drv.o adf_420xx_hw_data.o

drivers/crypto/intel/qat/qat_420xx/adf_420xx_hw_data.h

+/* SPDX-License-Identifier: GPL-2.0-only */
+/* Copyright(c) 2023 Intel Corporation */
+#ifndef ADF_420XX_HW_DATA_H_
+#define ADF_420XX_HW_DATA_H_
+
+#include <adf_accel_devices.h>
+
+#define ADF_420XX_MAX_ACCELENGINES		17
+
+#define ADF_420XX_ACCELENGINES_MASK		0x1FFFF
+#define ADF_420XX_ADMIN_AE_MASK			0x10000
+
+#define ADF_420XX_HICPPAGENTCMDPARERRLOG_MASK	(0xFF)
+#define ADF_420XX_PARITYERRORMASK_ATH_CPH_MASK	(0xFF00FF)
+#define ADF_420XX_PARITYERRORMASK_CPR_XLT_MASK	(0x10001)
+#define ADF_420XX_PARITYERRORMASK_DCPR_UCS_MASK	(0xF0007)
+#define ADF_420XX_PARITYERRORMASK_PKE_MASK	(0xFFF)
+#define ADF_420XX_PARITYERRORMASK_WAT_WCP_MASK	(0x3FF03FF)
+
+/*
+ * SSMFEATREN bit mask
+ * BIT(4) - enables parity detection on CPP
+ * BIT(12) - enables the logging of push/pull data errors
+ *	     in pperr register
+ * BIT(16) - BIT(27) - enable parity detection on SPPs
+ */
+#define ADF_420XX_SSMFEATREN_MASK \
+	(BIT(4) | BIT(12) | BIT(16) | BIT(17) | BIT(18) | BIT(19) | BIT(20) | \
+	 BIT(21) | BIT(22) | BIT(23) | BIT(24) | BIT(25) | BIT(26) | BIT(27))
+
+/* Firmware Binaries */
+#define ADF_420XX_FW		"qat_420xx.bin"
+#define ADF_420XX_MMP		"qat_420xx_mmp.bin"
+#define ADF_420XX_SYM_OBJ	"qat_420xx_sym.bin"
+#define ADF_420XX_DC_OBJ	"qat_420xx_dc.bin"
+#define ADF_420XX_ASYM_OBJ	"qat_420xx_asym.bin"
+#define ADF_420XX_ADMIN_OBJ	"qat_420xx_admin.bin"
+
+/* RL constants */
+#define ADF_420XX_RL_PCIE_SCALE_FACTOR_DIV	100
+#define ADF_420XX_RL_PCIE_SCALE_FACTOR_MUL	102
+#define ADF_420XX_RL_DCPR_CORRECTION		1
+#define ADF_420XX_RL_SCANS_PER_SEC		954
+#define ADF_420XX_RL_MAX_TP_ASYM		173750UL
+#define ADF_420XX_RL_MAX_TP_SYM			95000UL
+#define ADF_420XX_RL_MAX_TP_DC			40000UL
+#define ADF_420XX_RL_SLICE_REF			1000UL
+
+/* Clocks frequency */
+#define ADF_420XX_AE_FREQ		(1000 * HZ_PER_MHZ)
+
+void adf_init_hw_data_420xx(struct adf_hw_device_data *hw_data, u32 dev_id);
+void adf_clean_hw_data_420xx(struct adf_hw_device_data *hw_data);
+
+#endif

drivers/crypto/intel/qat/qat_420xx/adf_drv.c

+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright(c) 2023 Intel Corporation */
+#include <linux/device.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include <adf_accel_devices.h>
+#include <adf_gen4_hw_data.h>
+#include <adf_gen4_config.h>
+#include <adf_cfg.h>
+#include <adf_common_drv.h>
+#include <adf_dbgfs.h>
+
+#include "adf_420xx_hw_data.h"
+
+static const struct pci_device_id adf_pci_tbl[] = {
+	{ PCI_VDEVICE(INTEL, ADF_420XX_PCI_DEVICE_ID), },
+	{ }
+};
+MODULE_DEVICE_TABLE(pci, adf_pci_tbl);
+
+static void adf_cleanup_accel(struct adf_accel_dev *accel_dev)
+{
+	if (accel_dev->hw_device) {
+		adf_clean_hw_data_420xx(accel_dev->hw_device);
+		accel_dev->hw_device = NULL;
+	}
+	adf_dbgfs_exit(accel_dev);
+	adf_cfg_dev_remove(accel_dev);
+	adf_devmgr_rm_dev(accel_dev, NULL);
+}
+
+static int adf_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
+{
+	struct adf_accel_dev *accel_dev;
+	struct adf_accel_pci *accel_pci_dev;
+	struct adf_hw_device_data *hw_data;
+	unsigned int i, bar_nr;
+	unsigned long bar_mask;
+	struct adf_bar *bar;
+	int ret;
+
+	if (num_possible_nodes() > 1 && dev_to_node(&pdev->dev) < 0) {
+		/*
+		 * If the accelerator is connected to a node with no memory
+		 * there is no point in using the accelerator since the remote
+		 * memory transaction will be very slow.
+		 */
+		dev_err(&pdev->dev, "Invalid NUMA configuration.\n");
+		return -EINVAL;
+	}
+
+	accel_dev = devm_kzalloc(&pdev->dev, sizeof(*accel_dev), GFP_KERNEL);
+	if (!accel_dev)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&accel_dev->crypto_list);
+	accel_pci_dev = &accel_dev->accel_pci_dev;
+	accel_pci_dev->pci_dev = pdev;
+
+	/*
+	 * Add accel device to accel table
+	 * This should be called before adf_cleanup_accel is called
+	 */
+	if (adf_devmgr_add_dev(accel_dev, NULL)) {
+		dev_err(&pdev->dev, "Failed to add new accelerator device.\n");
+		return -EFAULT;
+	}
+
+	accel_dev->owner = THIS_MODULE;
+	/* Allocate and initialise device hardware meta-data structure */
+	hw_data = devm_kzalloc(&pdev->dev, sizeof(*hw_data), GFP_KERNEL);
+	if (!hw_data) {
+		ret = -ENOMEM;
+		goto out_err;
+	}
+
+	accel_dev->hw_device = hw_data;
+	adf_init_hw_data_420xx(accel_dev->hw_device, ent->device);
+
+	pci_read_config_byte(pdev, PCI_REVISION_ID, &accel_pci_dev->revid);
+	pci_read_config_dword(pdev, ADF_GEN4_FUSECTL4_OFFSET, &hw_data->fuses);
+
+	/* Get Accelerators and Accelerators Engines masks */
+	hw_data->accel_mask = hw_data->get_accel_mask(hw_data);
+	hw_data->ae_mask = hw_data->get_ae_mask(hw_data);
+	accel_pci_dev->sku = hw_data->get_sku(hw_data);
+	/* If the device has no acceleration engines then ignore it */
+	if (!hw_data->accel_mask || !hw_data->ae_mask ||
+	    (~hw_data->ae_mask & 0x01)) {
+		dev_err(&pdev->dev, "No acceleration units found.\n");
+		ret = -EFAULT;
+		goto out_err;
+	}
+
+	/* Create device configuration table */
+	ret = adf_cfg_dev_add(accel_dev);
+	if (ret)
+		goto out_err;
+
+	/* Enable PCI device */
+	ret = pcim_enable_device(pdev);
+	if (ret) {
+		dev_err(&pdev->dev, "Can't enable PCI device.\n");
+		goto out_err;
+	}
+
+	/* Set DMA identifier */
+	ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
+	if (ret) {
+		dev_err(&pdev->dev, "No usable DMA configuration.\n");
+		goto out_err;
+	}
+
+	ret = adf_gen4_cfg_dev_init(accel_dev);
+	if (ret) {
+		dev_err(&pdev->dev, "Failed to initialize configuration.\n");
+		goto out_err;
+	}
+
+	/* Get accelerator capabilities mask */
+	hw_data->accel_capabilities_mask = hw_data->get_accel_cap(accel_dev);
+	if (!hw_data->accel_capabilities_mask) {
+		dev_err(&pdev->dev, "Failed to get capabilities mask.\n");
+		ret = -EINVAL;
+		goto out_err;
+	}
+
+	/* Find and map all the device's BARS */
+	bar_mask = pci_select_bars(pdev, IORESOURCE_MEM) & ADF_GEN4_BAR_MASK;
+
+	ret = pcim_iomap_regions_request_all(pdev, bar_mask, pci_name(pdev));
+	if (ret) {
+		dev_err(&pdev->dev, "Failed to map pci regions.\n");
+		goto out_err;
+	}
+
+	i = 0;
+	for_each_set_bit(bar_nr, &bar_mask, PCI_STD_NUM_BARS) {
+		bar = &accel_pci_dev->pci_bars[i++];
+		bar->virt_addr = pcim_iomap_table(pdev)[bar_nr];
+	}
+
+	pci_set_master(pdev);
+
+	if (pci_save_state(pdev)) {
+		dev_err(&pdev->dev, "Failed to save pci state.\n");
+		ret = -ENOMEM;
+		goto out_err;
+	}
+
+	accel_dev->ras_errors.enabled = true;
+	adf_dbgfs_init(accel_dev);
+
+	ret = adf_dev_up(accel_dev, true);
+	if (ret)
+		goto out_err_dev_stop;
+
+	ret = adf_sysfs_init(accel_dev);
+	if (ret)
+		goto out_err_dev_stop;
+
+	return ret;
+
+out_err_dev_stop:
+	adf_dev_down(accel_dev, false);
+out_err:
+	adf_cleanup_accel(accel_dev);
+	return ret;
+}
+
+static void adf_remove(struct pci_dev *pdev)
+{
+	struct adf_accel_dev *accel_dev = adf_devmgr_pci_to_accel_dev(pdev);
+
+	if (!accel_dev) {
+		pr_err("QAT: Driver removal failed\n");
+		return;
+	}
+	adf_dev_down(accel_dev, false);
+	adf_cleanup_accel(accel_dev);
+}
+
+static struct pci_driver adf_driver = {
+	.id_table = adf_pci_tbl,
+	.name = ADF_420XX_DEVICE_NAME,
+	.probe = adf_probe,
+	.remove = adf_remove,
+	.sriov_configure = adf_sriov_configure,
+	.err_handler = &adf_err_handler,
+};
+
+module_pci_driver(adf_driver);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Intel");
+MODULE_FIRMWARE(ADF_420XX_FW);
+MODULE_FIRMWARE(ADF_420XX_MMP);
+MODULE_DESCRIPTION("Intel(R) QuickAssist Technology");
+MODULE_VERSION(ADF_DRV_VERSION);
+MODULE_SOFTDEP("pre: crypto-intel_qat");
+MODULE_IMPORT_NS(CRYPTO_QAT);

drivers/crypto/intel/qat/qat_common/adf_accel_devices.h

@@ -19,12 +19,15 @@
 #define ADF_C3XXX_DEVICE_NAME "c3xxx"
 #define ADF_C3XXXVF_DEVICE_NAME "c3xxxvf"
 #define ADF_4XXX_DEVICE_NAME "4xxx"
+#define ADF_420XX_DEVICE_NAME "420xx"
 #define ADF_4XXX_PCI_DEVICE_ID 0x4940
 #define ADF_4XXXIOV_PCI_DEVICE_ID 0x4941
 #define ADF_401XX_PCI_DEVICE_ID 0x4942
 #define ADF_401XXIOV_PCI_DEVICE_ID 0x4943
 #define ADF_402XX_PCI_DEVICE_ID 0x4944
 #define ADF_402XXIOV_PCI_DEVICE_ID 0x4945
+#define ADF_420XX_PCI_DEVICE_ID 0x4946
+#define ADF_420XXIOV_PCI_DEVICE_ID 0x4947
 #define ADF_DEVICE_FUSECTL_OFFSET 0x40
 #define ADF_DEVICE_LEGFUSE_OFFSET 0x4C
 #define ADF_DEVICE_FUSECTL_MASK 0x80000000

drivers/crypto/intel/qat/qat_common/adf_cfg_common.h

@@ -47,6 +47,7 @@ enum adf_device_type {
 	DEV_C3XXX,
 	DEV_C3XXXVF,
 	DEV_4XXX,
+	DEV_420XX,
 };
 
 struct adf_dev_status_info {

drivers/crypto/intel/qat/qat_common/adf_gen4_hw_data.h

@@ -202,6 +202,8 @@ enum icp_qat_gen4_slice_mask {
 	ICP_ACCEL_GEN4_MASK_UCS_SLICE = BIT(4),
 	ICP_ACCEL_GEN4_MASK_EIA3_SLICE = BIT(5),
 	ICP_ACCEL_GEN4_MASK_SMX_SLICE = BIT(7),
+	ICP_ACCEL_GEN4_MASK_WCP_WAT_SLICE = BIT(8),
+	ICP_ACCEL_GEN4_MASK_ZUC_256_SLICE = BIT(9),
 };
 
 void adf_gen4_enable_error_correction(struct adf_accel_dev *accel_dev);

drivers/crypto/intel/qat/qat_common/icp_qat_hw.h

@@ -18,7 +18,12 @@ enum icp_qat_hw_ae_id {
 	ICP_QAT_HW_AE_9 = 9,
 	ICP_QAT_HW_AE_10 = 10,
 	ICP_QAT_HW_AE_11 = 11,
-	ICP_QAT_HW_AE_DELIMITER = 12
+	ICP_QAT_HW_AE_12 = 12,
+	ICP_QAT_HW_AE_13 = 13,
+	ICP_QAT_HW_AE_14 = 14,
+	ICP_QAT_HW_AE_15 = 15,
+	ICP_QAT_HW_AE_16 = 16,
+	ICP_QAT_HW_AE_DELIMITER = 17
 };
 
 enum icp_qat_hw_qat_id {
@@ -95,7 +100,7 @@ enum icp_qat_capabilities_mask {
 	/* Bits 10-11 are currently reserved */
 	ICP_ACCEL_CAPABILITIES_HKDF = BIT(12),
 	ICP_ACCEL_CAPABILITIES_ECEDMONT = BIT(13),
-	/* Bit 14 is currently reserved */
+	ICP_ACCEL_CAPABILITIES_EXT_ALGCHAIN = BIT(14),
 	ICP_ACCEL_CAPABILITIES_SHA3_EXT = BIT(15),
 	ICP_ACCEL_CAPABILITIES_AESGCM_SPC = BIT(16),
 	ICP_ACCEL_CAPABILITIES_CHACHA_POLY = BIT(17),
@@ -107,7 +112,10 @@ enum icp_qat_capabilities_mask {
 	ICP_ACCEL_CAPABILITIES_CNV_INTEGRITY64 = BIT(23),
 	ICP_ACCEL_CAPABILITIES_LZ4_COMPRESSION = BIT(24),
 	ICP_ACCEL_CAPABILITIES_LZ4S_COMPRESSION = BIT(25),
-	ICP_ACCEL_CAPABILITIES_AES_V2 = BIT(26)
+	ICP_ACCEL_CAPABILITIES_AES_V2 = BIT(26),
+	/* Bits 27-28 are currently reserved */
+	ICP_ACCEL_CAPABILITIES_ZUC_256 = BIT(29),
+	ICP_ACCEL_CAPABILITIES_WIRELESS_CRYPTO_EXT = BIT(30),
 };
 
 #define QAT_AUTH_MODE_BITPOS 4

drivers/crypto/intel/qat/qat_common/icp_qat_uclo.h

@@ -7,7 +7,7 @@
 #define ICP_QAT_AC_C62X_DEV_TYPE   0x01000000
 #define ICP_QAT_AC_C3XXX_DEV_TYPE  0x02000000
 #define ICP_QAT_AC_4XXX_A_DEV_TYPE 0x08000000
-#define ICP_QAT_UCLO_MAX_AE       12
+#define ICP_QAT_UCLO_MAX_AE       17
 #define ICP_QAT_UCLO_MAX_CTX      8
 #define ICP_QAT_UCLO_MAX_UIMAGE   (ICP_QAT_UCLO_MAX_AE * ICP_QAT_UCLO_MAX_CTX)
 #define ICP_QAT_UCLO_MAX_USTORE   0x4000

drivers/crypto/intel/qat/qat_common/qat_hal.c

@@ -697,12 +697,16 @@ static int qat_hal_chip_init(struct icp_qat_fw_loader_handle *handle,
 	case ADF_4XXX_PCI_DEVICE_ID:
 	case ADF_401XX_PCI_DEVICE_ID:
 	case ADF_402XX_PCI_DEVICE_ID:
+	case ADF_420XX_PCI_DEVICE_ID:
 		handle->chip_info->mmp_sram_size = 0;
 		handle->chip_info->nn = false;
 		handle->chip_info->lm2lm3 = true;
 		handle->chip_info->lm_size = ICP_QAT_UCLO_MAX_LMEM_REG_2X;
 		handle->chip_info->icp_rst_csr = ICP_RESET_CPP0;
-		handle->chip_info->icp_rst_mask = 0x100015;
+		if (handle->pci_dev->device == ADF_420XX_PCI_DEVICE_ID)
+			handle->chip_info->icp_rst_mask = 0x100155;
+		else
+			handle->chip_info->icp_rst_mask = 0x100015;
 		handle->chip_info->glb_clk_enable_csr = ICP_GLOBAL_CLK_ENABLE_CPP0;
 		handle->chip_info->misc_ctl_csr = MISC_CONTROL_C4XXX;
 		handle->chip_info->wakeup_event_val = 0x80000000;

drivers/crypto/intel/qat/qat_common/qat_uclo.c

@@ -733,6 +733,7 @@ qat_uclo_get_dev_type(struct icp_qat_fw_loader_handle *handle)
 	case ADF_4XXX_PCI_DEVICE_ID:
 	case ADF_401XX_PCI_DEVICE_ID:
 	case ADF_402XX_PCI_DEVICE_ID:
+	case ADF_420XX_PCI_DEVICE_ID:
 		return ICP_QAT_AC_4XXX_A_DEV_TYPE;
 	default:
 		pr_err("QAT: unsupported device 0x%x\n",