This series introduces crypto(CPT) drivers(PF & VF) for Marvell
OcteonTX2 CN96XX Soc.
OcteonTX2 SOC's resource virtualization unit (RVU) supports multiple
physical and virtual functions. Each of the PF/VF's functionality is
determined by what kind of resources are attached to it. When the CPT
block is attached to a VF, it can function as a security device.
The CPT PF driver is responsible for:
- Forwarding messages to/from VFs from/to admin function(AF),
- Enabling/disabling VFs,
- Loading/unloading microcode (creation/deletion of engine groups).
The CPT VF driver works as a crypto offload device.
This patch series includes:
- CPT PF driver patches that include AF<=>PF<=>VF mailbox communication,
sriov_configure, and firmware load to the acceleration engines.
- CPT VF driver patches that include VF<=>PF mailbox communication and
crypto offload support through the kernel cryptographic API.
This series is tested with CRYPTO_EXTRA_TESTS enabled and
CRYPTO_DISABLE_TESTS disabled.
Changes since v1:
* Resolved compilation warning.
Srujana Challa (9):
drivers: crypto: add Marvell OcteonTX2 CPT PF driver
crypto: octeontx2: add mailbox communication with AF
crypto: octeontx2: enable SR-IOV and mailbox communication with VF
crypto: octeontx2: load microcode and create engine groups
crypto: octeontx2: add LF framework
crypto: octeontx2: add support to get engine capabilities
crypto: octeontx2: add virtual function driver support
crypto: octeontx2: add support to process the crypto request
crypto: octeontx2: register with linux crypto framework
drivers/crypto/marvell/Kconfig | 14 +
drivers/crypto/marvell/Makefile | 1 +
drivers/crypto/marvell/octeontx2/Makefile | 10 +
.../marvell/octeontx2/otx2_cpt_common.h | 137 ++
.../marvell/octeontx2/otx2_cpt_hw_types.h | 464 +++++
.../marvell/octeontx2/otx2_cpt_mbox_common.c | 202 ++
.../marvell/octeontx2/otx2_cpt_reqmgr.h | 197 ++
drivers/crypto/marvell/octeontx2/otx2_cptlf.c | 429 ++++
drivers/crypto/marvell/octeontx2/otx2_cptlf.h | 353 ++++
drivers/crypto/marvell/octeontx2/otx2_cptpf.h | 61 +
.../marvell/octeontx2/otx2_cptpf_main.c | 713 +++++++
.../marvell/octeontx2/otx2_cptpf_mbox.c | 356 ++++
.../marvell/octeontx2/otx2_cptpf_ucode.c | 1415 +++++++++++++
.../marvell/octeontx2/otx2_cptpf_ucode.h | 162 ++
drivers/crypto/marvell/octeontx2/otx2_cptvf.h | 29 +
.../marvell/octeontx2/otx2_cptvf_algs.c | 1758 +++++++++++++++++
.../marvell/octeontx2/otx2_cptvf_algs.h | 178 ++
.../marvell/octeontx2/otx2_cptvf_main.c | 410 ++++
.../marvell/octeontx2/otx2_cptvf_mbox.c | 167 ++
.../marvell/octeontx2/otx2_cptvf_reqmgr.c | 541 +++++
20 files changed, 7597 insertions(+)
create mode 100644 drivers/crypto/marvell/octeontx2/Makefile
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_hw_types.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_mbox_common.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptlf.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptlf.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c
--
2.29.0
Add support for the Marvell OcteonTX2 CPT virtual function
driver. This patch includes probe, PCI specific initialization
and interrupt handling.
Signed-off-by: Suheil Chandran <[email protected]>
Signed-off-by: Lukasz Bartosik <[email protected]>
Signed-off-by: Srujana Challa <[email protected]>
---
drivers/crypto/marvell/octeontx2/Makefile | 4 +-
.../marvell/octeontx2/otx2_cpt_common.h | 1 +
.../marvell/octeontx2/otx2_cpt_mbox_common.c | 32 +++
drivers/crypto/marvell/octeontx2/otx2_cptvf.h | 28 +++
.../marvell/octeontx2/otx2_cptvf_main.c | 196 ++++++++++++++++++
.../marvell/octeontx2/otx2_cptvf_mbox.c | 113 ++++++++++
6 files changed, 373 insertions(+), 1 deletion(-)
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c
diff --git a/drivers/crypto/marvell/octeontx2/Makefile b/drivers/crypto/marvell/octeontx2/Makefile
index e47a55961bb8..ef6fb2ab3571 100644
--- a/drivers/crypto/marvell/octeontx2/Makefile
+++ b/drivers/crypto/marvell/octeontx2/Makefile
@@ -1,7 +1,9 @@
# SPDX-License-Identifier: GPL-2.0-only
-obj-$(CONFIG_CRYPTO_DEV_OCTEONTX2_CPT) += octeontx2-cpt.o
+obj-$(CONFIG_CRYPTO_DEV_OCTEONTX2_CPT) += octeontx2-cpt.o octeontx2-cptvf.o
octeontx2-cpt-objs := otx2_cptpf_main.o otx2_cptpf_mbox.o \
otx2_cpt_mbox_common.o otx2_cptpf_ucode.o otx2_cptlf.o
+octeontx2-cptvf-objs := otx2_cptvf_main.o otx2_cptvf_mbox.o otx2_cptlf.o \
+ otx2_cpt_mbox_common.o
ccflags-y += -I$(srctree)/drivers/net/ethernet/marvell/octeontx2/af
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h b/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
index 705a0503b962..ca220178e518 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
@@ -115,5 +115,6 @@ int otx2_cpt_write_af_reg(struct otx2_mbox *mbox, struct pci_dev *pdev,
struct otx2_cptlfs_info;
int otx2_cpt_attach_rscrs_msg(struct otx2_cptlfs_info *lfs);
int otx2_cpt_detach_rsrcs_msg(struct otx2_cptlfs_info *lfs);
+int otx2_cpt_msix_offset_msg(struct otx2_cptlfs_info *lfs);
#endif /* __OTX2_CPT_COMMON_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cpt_mbox_common.c b/drivers/crypto/marvell/octeontx2/otx2_cpt_mbox_common.c
index 0933031ac827..51cb6404ded7 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cpt_mbox_common.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_mbox_common.c
@@ -168,3 +168,35 @@ int otx2_cpt_detach_rsrcs_msg(struct otx2_cptlfs_info *lfs)
return ret;
}
+
+int otx2_cpt_msix_offset_msg(struct otx2_cptlfs_info *lfs)
+{
+ struct otx2_mbox *mbox = lfs->mbox;
+ struct pci_dev *pdev = lfs->pdev;
+ struct mbox_msghdr *req;
+ int ret, i;
+
+ req = otx2_mbox_alloc_msg_rsp(mbox, 0, sizeof(*req),
+ sizeof(struct msix_offset_rsp));
+ if (req == NULL) {
+ dev_err(&pdev->dev, "RVU MBOX failed to get message.\n");
+ return -EFAULT;
+ }
+
+ req->id = MBOX_MSG_MSIX_OFFSET;
+ req->sig = OTX2_MBOX_REQ_SIG;
+ req->pcifunc = 0;
+ ret = otx2_cpt_send_mbox_msg(mbox, pdev);
+ if (ret)
+ return ret;
+
+ for (i = 0; i < lfs->lfs_num; i++) {
+ if (lfs->lf[i].msix_offset == MSIX_VECTOR_INVALID) {
+ dev_err(&pdev->dev,
+ "Invalid msix offset %d for LF %d\n",
+ lfs->lf[i].msix_offset, i);
+ return -EINVAL;
+ }
+ }
+ return ret;
+}
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf.h b/drivers/crypto/marvell/octeontx2/otx2_cptvf.h
new file mode 100644
index 000000000000..4b01eb9d9f70
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf.h
@@ -0,0 +1,28 @@
+/* SPDX-License-Identifier: GPL-2.0-only
+ * Copyright (C) 2020 Marvell.
+ */
+
+#ifndef __OTX2_CPTVF_H
+#define __OTX2_CPTVF_H
+
+#include "mbox.h"
+#include "otx2_cptlf.h"
+
+struct otx2_cptvf_dev {
+ void __iomem *reg_base; /* Register start address */
+ void __iomem *pfvf_mbox_base; /* PF-VF mbox start address */
+ struct pci_dev *pdev; /* PCI device handle */
+ struct otx2_cptlfs_info lfs; /* CPT LFs attached to this VF */
+ u8 vf_id; /* Virtual function index */
+
+ /* PF <=> VF mbox */
+ struct otx2_mbox pfvf_mbox;
+ struct work_struct pfvf_mbox_work;
+ struct workqueue_struct *pfvf_mbox_wq;
+};
+
+irqreturn_t otx2_cptvf_pfvf_mbox_intr(int irq, void *arg);
+void otx2_cptvf_pfvf_mbox_handler(struct work_struct *work);
+int otx2_cptvf_send_eng_grp_num_msg(struct otx2_cptvf_dev *cptvf, int eng_type);
+
+#endif /* __OTX2_CPTVF_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c b/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
new file mode 100644
index 000000000000..6d84fafa2915
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
@@ -0,0 +1,196 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (C) 2020 Marvell. */
+
+#include "otx2_cpt_common.h"
+#include "otx2_cptvf.h"
+#include <rvu_reg.h>
+
+#define OTX2_CPTVF_DRV_NAME "octeontx2-cptvf"
+
+static void cptvf_enable_pfvf_mbox_intrs(struct otx2_cptvf_dev *cptvf)
+{
+ /* Clear interrupt if any */
+ otx2_cpt_write64(cptvf->reg_base, BLKADDR_RVUM, 0, OTX2_RVU_VF_INT,
+ 0x1ULL);
+
+ /* Enable PF-VF interrupt */
+ otx2_cpt_write64(cptvf->reg_base, BLKADDR_RVUM, 0,
+ OTX2_RVU_VF_INT_ENA_W1S, 0x1ULL);
+}
+
+static void cptvf_disable_pfvf_mbox_intrs(struct otx2_cptvf_dev *cptvf)
+{
+ /* Disable PF-VF interrupt */
+ otx2_cpt_write64(cptvf->reg_base, BLKADDR_RVUM, 0,
+ OTX2_RVU_VF_INT_ENA_W1C, 0x1ULL);
+
+ /* Clear interrupt if any */
+ otx2_cpt_write64(cptvf->reg_base, BLKADDR_RVUM, 0, OTX2_RVU_VF_INT,
+ 0x1ULL);
+}
+
+static int cptvf_register_interrupts(struct otx2_cptvf_dev *cptvf)
+{
+ int ret, irq;
+ u32 num_vec;
+
+ num_vec = pci_msix_vec_count(cptvf->pdev);
+ if (num_vec <= 0)
+ return -EINVAL;
+
+ /* Enable MSI-X */
+ ret = pci_alloc_irq_vectors(cptvf->pdev, num_vec, num_vec,
+ PCI_IRQ_MSIX);
+ if (ret < 0) {
+ dev_err(&cptvf->pdev->dev,
+ "Request for %d msix vectors failed\n", num_vec);
+ return ret;
+ }
+ irq = pci_irq_vector(cptvf->pdev, OTX2_CPT_VF_INT_VEC_E_MBOX);
+ /* Register VF<=>PF mailbox interrupt handler */
+ ret = devm_request_irq(&cptvf->pdev->dev, irq,
+ otx2_cptvf_pfvf_mbox_intr, 0,
+ "CPTPFVF Mbox", cptvf);
+ if (ret)
+ return ret;
+ /* Enable PF-VF mailbox interrupts */
+ cptvf_enable_pfvf_mbox_intrs(cptvf);
+
+ ret = otx2_cpt_send_ready_msg(&cptvf->pfvf_mbox, cptvf->pdev);
+ if (ret) {
+ dev_warn(&cptvf->pdev->dev,
+ "PF not responding to mailbox, deferring probe\n");
+ cptvf_disable_pfvf_mbox_intrs(cptvf);
+ return -EPROBE_DEFER;
+ }
+ return 0;
+}
+
+static int cptvf_pfvf_mbox_init(struct otx2_cptvf_dev *cptvf)
+{
+ int ret;
+
+ cptvf->pfvf_mbox_wq = alloc_workqueue("cpt_pfvf_mailbox",
+ WQ_UNBOUND | WQ_HIGHPRI |
+ WQ_MEM_RECLAIM, 1);
+ if (!cptvf->pfvf_mbox_wq)
+ return -ENOMEM;
+
+ ret = otx2_mbox_init(&cptvf->pfvf_mbox, cptvf->pfvf_mbox_base,
+ cptvf->pdev, cptvf->reg_base, MBOX_DIR_VFPF, 1);
+ if (ret)
+ goto free_wqe;
+
+ INIT_WORK(&cptvf->pfvf_mbox_work, otx2_cptvf_pfvf_mbox_handler);
+ return 0;
+
+free_wqe:
+ destroy_workqueue(cptvf->pfvf_mbox_wq);
+ return ret;
+}
+
+static void cptvf_pfvf_mbox_destroy(struct otx2_cptvf_dev *cptvf)
+{
+ destroy_workqueue(cptvf->pfvf_mbox_wq);
+ otx2_mbox_destroy(&cptvf->pfvf_mbox);
+}
+
+static int otx2_cptvf_probe(struct pci_dev *pdev,
+ const struct pci_device_id *ent)
+{
+ struct device *dev = &pdev->dev;
+ resource_size_t offset, size;
+ struct otx2_cptvf_dev *cptvf;
+ int ret;
+
+ cptvf = devm_kzalloc(dev, sizeof(*cptvf), GFP_KERNEL);
+ if (!cptvf)
+ return -ENOMEM;
+
+ ret = pcim_enable_device(pdev);
+ if (ret) {
+ dev_err(dev, "Failed to enable PCI device\n");
+ goto clear_drvdata;
+ }
+
+ ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(48));
+ if (ret) {
+ dev_err(dev, "Unable to get usable DMA configuration\n");
+ goto clear_drvdata;
+ }
+ /* Map VF's configuration registers */
+ ret = pcim_iomap_regions_request_all(pdev, 1 << PCI_PF_REG_BAR_NUM,
+ OTX2_CPTVF_DRV_NAME);
+ if (ret) {
+ dev_err(dev, "Couldn't get PCI resources 0x%x\n", ret);
+ goto clear_drvdata;
+ }
+ pci_set_master(pdev);
+ pci_set_drvdata(pdev, cptvf);
+ cptvf->pdev = pdev;
+
+ cptvf->reg_base = pcim_iomap_table(pdev)[PCI_PF_REG_BAR_NUM];
+
+ offset = pci_resource_start(pdev, PCI_MBOX_BAR_NUM);
+ size = pci_resource_len(pdev, PCI_MBOX_BAR_NUM);
+ /* Map PF-VF mailbox memory */
+ cptvf->pfvf_mbox_base = devm_ioremap_wc(dev, offset, size);
+ if (!cptvf->pfvf_mbox_base) {
+ dev_err(&pdev->dev, "Unable to map BAR4\n");
+ ret = -ENODEV;
+ goto clear_drvdata;
+ }
+ /* Initialize PF<=>VF mailbox */
+ ret = cptvf_pfvf_mbox_init(cptvf);
+ if (ret)
+ goto clear_drvdata;
+
+ /* Register interrupts */
+ ret = cptvf_register_interrupts(cptvf);
+ if (ret)
+ goto destroy_pfvf_mbox;
+
+ return 0;
+
+destroy_pfvf_mbox:
+ cptvf_pfvf_mbox_destroy(cptvf);
+clear_drvdata:
+ pci_set_drvdata(pdev, NULL);
+
+ return ret;
+}
+
+static void otx2_cptvf_remove(struct pci_dev *pdev)
+{
+ struct otx2_cptvf_dev *cptvf = pci_get_drvdata(pdev);
+
+ if (!cptvf) {
+ dev_err(&pdev->dev, "Invalid CPT VF device.\n");
+ return;
+ }
+ /* Disable PF-VF mailbox interrupt */
+ cptvf_disable_pfvf_mbox_intrs(cptvf);
+ /* Destroy PF-VF mbox */
+ cptvf_pfvf_mbox_destroy(cptvf);
+ pci_set_drvdata(pdev, NULL);
+}
+
+/* Supported devices */
+static const struct pci_device_id otx2_cptvf_id_table[] = {
+ {PCI_VDEVICE(CAVIUM, OTX2_CPT_PCI_VF_DEVICE_ID), 0},
+ { 0, } /* end of table */
+};
+
+static struct pci_driver otx2_cptvf_pci_driver = {
+ .name = OTX2_CPTVF_DRV_NAME,
+ .id_table = otx2_cptvf_id_table,
+ .probe = otx2_cptvf_probe,
+ .remove = otx2_cptvf_remove,
+};
+
+module_pci_driver(otx2_cptvf_pci_driver);
+
+MODULE_AUTHOR("Marvell");
+MODULE_DESCRIPTION("Marvell OcteonTX2 CPT Virtual Function Driver");
+MODULE_LICENSE("GPL v2");
+MODULE_DEVICE_TABLE(pci, otx2_cptvf_id_table);
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c b/drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c
new file mode 100644
index 000000000000..417099a86742
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c
@@ -0,0 +1,113 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (C) 2020 Marvell. */
+
+#include "otx2_cpt_common.h"
+#include "otx2_cptvf.h"
+#include <rvu_reg.h>
+
+irqreturn_t otx2_cptvf_pfvf_mbox_intr(int __always_unused irq, void *arg)
+{
+ struct otx2_cptvf_dev *cptvf = arg;
+ u64 intr;
+
+ /* Read the interrupt bits */
+ intr = otx2_cpt_read64(cptvf->reg_base, BLKADDR_RVUM, 0,
+ OTX2_RVU_VF_INT);
+
+ if (intr & 0x1ULL) {
+ /* Schedule work queue function to process the MBOX request */
+ queue_work(cptvf->pfvf_mbox_wq, &cptvf->pfvf_mbox_work);
+ /* Clear and ack the interrupt */
+ otx2_cpt_write64(cptvf->reg_base, BLKADDR_RVUM, 0,
+ OTX2_RVU_VF_INT, 0x1ULL);
+ }
+ return IRQ_HANDLED;
+}
+
+static void process_pfvf_mbox_mbox_msg(struct otx2_cptvf_dev *cptvf,
+ struct mbox_msghdr *msg)
+{
+ struct otx2_cptlfs_info *lfs = &cptvf->lfs;
+ struct cpt_rd_wr_reg_msg *rsp_reg;
+ struct msix_offset_rsp *rsp_msix;
+ int i;
+
+ if (msg->id >= MBOX_MSG_MAX) {
+ dev_err(&cptvf->pdev->dev,
+ "MBOX msg with unknown ID %d\n", msg->id);
+ return;
+ }
+ if (msg->sig != OTX2_MBOX_RSP_SIG) {
+ dev_err(&cptvf->pdev->dev,
+ "MBOX msg with wrong signature %x, ID %d\n",
+ msg->sig, msg->id);
+ return;
+ }
+ switch (msg->id) {
+ case MBOX_MSG_READY:
+ cptvf->vf_id = ((msg->pcifunc >> RVU_PFVF_FUNC_SHIFT)
+ & RVU_PFVF_FUNC_MASK) - 1;
+ break;
+ case MBOX_MSG_ATTACH_RESOURCES:
+ /* Check if resources were successfully attached */
+ if (!msg->rc)
+ lfs->are_lfs_attached = 1;
+ break;
+ case MBOX_MSG_DETACH_RESOURCES:
+ /* Check if resources were successfully detached */
+ if (!msg->rc)
+ lfs->are_lfs_attached = 0;
+ break;
+ case MBOX_MSG_MSIX_OFFSET:
+ rsp_msix = (struct msix_offset_rsp *) msg;
+ for (i = 0; i < rsp_msix->cptlfs; i++)
+ lfs->lf[i].msix_offset = rsp_msix->cptlf_msixoff[i];
+ break;
+ case MBOX_MSG_CPT_RD_WR_REGISTER:
+ rsp_reg = (struct cpt_rd_wr_reg_msg *) msg;
+ if (msg->rc) {
+ dev_err(&cptvf->pdev->dev,
+ "Reg %llx rd/wr(%d) failed %d\n",
+ rsp_reg->reg_offset, rsp_reg->is_write,
+ msg->rc);
+ return;
+ }
+ if (!rsp_reg->is_write)
+ *rsp_reg->ret_val = rsp_reg->val;
+ break;
+ default:
+ dev_err(&cptvf->pdev->dev, "Unsupported msg %d received.\n",
+ msg->id);
+ break;
+ }
+}
+
+void otx2_cptvf_pfvf_mbox_handler(struct work_struct *work)
+{
+ struct otx2_cptvf_dev *cptvf;
+ struct otx2_mbox *pfvf_mbox;
+ struct otx2_mbox_dev *mdev;
+ struct mbox_hdr *rsp_hdr;
+ struct mbox_msghdr *msg;
+ int offset, i;
+
+ /* sync with mbox memory region */
+ smp_rmb();
+
+ cptvf = container_of(work, struct otx2_cptvf_dev, pfvf_mbox_work);
+ pfvf_mbox = &cptvf->pfvf_mbox;
+ mdev = &pfvf_mbox->dev[0];
+ rsp_hdr = (struct mbox_hdr *)(mdev->mbase + pfvf_mbox->rx_start);
+ if (rsp_hdr->num_msgs == 0)
+ return;
+ offset = ALIGN(sizeof(struct mbox_hdr), MBOX_MSG_ALIGN);
+
+ for (i = 0; i < rsp_hdr->num_msgs; i++) {
+ msg = (struct mbox_msghdr *)(mdev->mbase + pfvf_mbox->rx_start +
+ offset);
+ process_pfvf_mbox_mbox_msg(cptvf, msg);
+ offset = msg->next_msgoff;
+ mdev->msgs_acked++;
+ }
+ otx2_mbox_reset(pfvf_mbox, 0);
+}
--
2.29.0
Adds skeleton for the Marvell OcteonTX2 CPT physical function
driver which includes probe, PCI specific initialization and
hardware register defines.
RVU defines are present in AF driver
(drivers/net/ethernet/marvell/octeontx2/af), header files from
AF driver are included here to avoid duplication.
Signed-off-by: Suheil Chandran <[email protected]>
Signed-off-by: Lukasz Bartosik <[email protected]>
Signed-off-by: Srujana Challa <[email protected]>
---
drivers/crypto/marvell/Kconfig | 10 +
drivers/crypto/marvell/Makefile | 1 +
drivers/crypto/marvell/octeontx2/Makefile | 6 +
.../marvell/octeontx2/otx2_cpt_common.h | 32 ++
.../marvell/octeontx2/otx2_cpt_hw_types.h | 464 ++++++++++++++++++
drivers/crypto/marvell/octeontx2/otx2_cptpf.h | 13 +
.../marvell/octeontx2/otx2_cptpf_main.c | 107 ++++
7 files changed, 633 insertions(+)
create mode 100644 drivers/crypto/marvell/octeontx2/Makefile
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_hw_types.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf.h
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
diff --git a/drivers/crypto/marvell/Kconfig b/drivers/crypto/marvell/Kconfig
index 13063384f958..1440ec9e1fb4 100644
--- a/drivers/crypto/marvell/Kconfig
+++ b/drivers/crypto/marvell/Kconfig
@@ -35,3 +35,13 @@ config CRYPTO_DEV_OCTEONTX_CPT
To compile this driver as module, choose M here:
the modules will be called octeontx-cpt and octeontx-cptvf
+
+config CRYPTO_DEV_OCTEONTX2_CPT
+ tristate "Marvell OcteonTX2 CPT driver"
+ depends on ARM64 || COMPILE_TEST
+ depends on PCI_MSI && 64BIT
+ select OCTEONTX2_MBOX
+ select CRYPTO_DEV_MARVELL
+ help
+ This driver allows you to utilize the Marvell Cryptographic
+ Accelerator Unit(CPT) found in OcteonTX2 series of processors.
diff --git a/drivers/crypto/marvell/Makefile b/drivers/crypto/marvell/Makefile
index 6c6a1519b0f1..39db6d9c0aaf 100644
--- a/drivers/crypto/marvell/Makefile
+++ b/drivers/crypto/marvell/Makefile
@@ -2,3 +2,4 @@
obj-$(CONFIG_CRYPTO_DEV_MARVELL_CESA) += cesa/
obj-$(CONFIG_CRYPTO_DEV_OCTEONTX_CPT) += octeontx/
+obj-$(CONFIG_CRYPTO_DEV_OCTEONTX2_CPT) += octeontx2/
diff --git a/drivers/crypto/marvell/octeontx2/Makefile b/drivers/crypto/marvell/octeontx2/Makefile
new file mode 100644
index 000000000000..db763ad46a91
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/Makefile
@@ -0,0 +1,6 @@
+# SPDX-License-Identifier: GPL-2.0-only
+obj-$(CONFIG_CRYPTO_DEV_OCTEONTX2_CPT) += octeontx2-cpt.o
+
+octeontx2-cpt-objs := otx2_cptpf_main.o
+
+ccflags-y += -I$(srctree)/drivers/net/ethernet/marvell/octeontx2/af
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h b/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
new file mode 100644
index 000000000000..eff4ffa58dc4
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: GPL-2.0-only
+ * Copyright (C) 2020 Marvell.
+ */
+
+#ifndef __OTX2_CPT_COMMON_H
+#define __OTX2_CPT_COMMON_H
+
+#include <linux/pci.h>
+#include <linux/types.h>
+#include <linux/module.h>
+#include <linux/delay.h>
+#include <linux/crypto.h>
+#include "otx2_cpt_hw_types.h"
+#include "rvu.h"
+
+#define OTX2_CPT_RVU_FUNC_ADDR_S(blk, slot, offs) \
+ (((blk) << 20) | ((slot) << 12) | (offs))
+
+static inline void otx2_cpt_write64(void __iomem *reg_base, u64 blk, u64 slot,
+ u64 offs, u64 val)
+{
+ writeq_relaxed(val, reg_base +
+ OTX2_CPT_RVU_FUNC_ADDR_S(blk, slot, offs));
+}
+
+static inline u64 otx2_cpt_read64(void __iomem *reg_base, u64 blk, u64 slot,
+ u64 offs)
+{
+ return readq_relaxed(reg_base +
+ OTX2_CPT_RVU_FUNC_ADDR_S(blk, slot, offs));
+}
+#endif /* __OTX2_CPT_COMMON_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cpt_hw_types.h b/drivers/crypto/marvell/octeontx2/otx2_cpt_hw_types.h
new file mode 100644
index 000000000000..ecafc42f37a2
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_hw_types.h
@@ -0,0 +1,464 @@
+/* SPDX-License-Identifier: GPL-2.0-only
+ * Copyright (C) 2020 Marvell.
+ */
+
+#ifndef __OTX2_CPT_HW_TYPES_H
+#define __OTX2_CPT_HW_TYPES_H
+
+#include <linux/types.h>
+
+/* Device IDs */
+#define OTX2_CPT_PCI_PF_DEVICE_ID 0xA0FD
+#define OTX2_CPT_PCI_VF_DEVICE_ID 0xA0FE
+
+/* Mailbox interrupts offset */
+#define OTX2_CPT_PF_MBOX_INT 6
+#define OTX2_CPT_PF_INT_VEC_E_MBOXX(x, a) ((x) + (a))
+
+/* Maximum supported microcode groups */
+#define OTX2_CPT_MAX_ENGINE_GROUPS 8
+
+/* CPT instruction size in bytes */
+#define OTX2_CPT_INST_SIZE 64
+/*
+ * CPT VF MSIX vectors and their offsets
+ */
+#define OTX2_CPT_VF_MSIX_VECTORS 1
+#define OTX2_CPT_VF_INTR_MBOX_MASK BIT(0)
+
+/* CPT LF MSIX vectors */
+#define OTX2_CPT_LF_MSIX_VECTORS 2
+
+/* OcteonTX2 CPT PF registers */
+#define OTX2_CPT_PF_CONSTANTS (0x0)
+#define OTX2_CPT_PF_RESET (0x100)
+#define OTX2_CPT_PF_DIAG (0x120)
+#define OTX2_CPT_PF_BIST_STATUS (0x160)
+#define OTX2_CPT_PF_ECC0_CTL (0x200)
+#define OTX2_CPT_PF_ECC0_FLIP (0x210)
+#define OTX2_CPT_PF_ECC0_INT (0x220)
+#define OTX2_CPT_PF_ECC0_INT_W1S (0x230)
+#define OTX2_CPT_PF_ECC0_ENA_W1S (0x240)
+#define OTX2_CPT_PF_ECC0_ENA_W1C (0x250)
+#define OTX2_CPT_PF_MBOX_INTX(b) (0x400 | (b) << 3)
+#define OTX2_CPT_PF_MBOX_INT_W1SX(b) (0x420 | (b) << 3)
+#define OTX2_CPT_PF_MBOX_ENA_W1CX(b) (0x440 | (b) << 3)
+#define OTX2_CPT_PF_MBOX_ENA_W1SX(b) (0x460 | (b) << 3)
+#define OTX2_CPT_PF_EXEC_INT (0x500)
+#define OTX2_CPT_PF_EXEC_INT_W1S (0x520)
+#define OTX2_CPT_PF_EXEC_ENA_W1C (0x540)
+#define OTX2_CPT_PF_EXEC_ENA_W1S (0x560)
+#define OTX2_CPT_PF_GX_EN(b) (0x600 | (b) << 3)
+#define OTX2_CPT_PF_EXEC_INFO (0x700)
+#define OTX2_CPT_PF_EXEC_BUSY (0x800)
+#define OTX2_CPT_PF_EXEC_INFO0 (0x900)
+#define OTX2_CPT_PF_EXEC_INFO1 (0x910)
+#define OTX2_CPT_PF_INST_REQ_PC (0x10000)
+#define OTX2_CPT_PF_INST_LATENCY_PC (0x10020)
+#define OTX2_CPT_PF_RD_REQ_PC (0x10040)
+#define OTX2_CPT_PF_RD_LATENCY_PC (0x10060)
+#define OTX2_CPT_PF_RD_UC_PC (0x10080)
+#define OTX2_CPT_PF_ACTIVE_CYCLES_PC (0x10100)
+#define OTX2_CPT_PF_EXE_CTL (0x4000000)
+#define OTX2_CPT_PF_EXE_STATUS (0x4000008)
+#define OTX2_CPT_PF_EXE_CLK (0x4000010)
+#define OTX2_CPT_PF_EXE_DBG_CTL (0x4000018)
+#define OTX2_CPT_PF_EXE_DBG_DATA (0x4000020)
+#define OTX2_CPT_PF_EXE_BIST_STATUS (0x4000028)
+#define OTX2_CPT_PF_EXE_REQ_TIMER (0x4000030)
+#define OTX2_CPT_PF_EXE_MEM_CTL (0x4000038)
+#define OTX2_CPT_PF_EXE_PERF_CTL (0x4001000)
+#define OTX2_CPT_PF_EXE_DBG_CNTX(b) (0x4001100 | (b) << 3)
+#define OTX2_CPT_PF_EXE_PERF_EVENT_CNT (0x4001180)
+#define OTX2_CPT_PF_EXE_EPCI_INBX_CNT(b) (0x4001200 | (b) << 3)
+#define OTX2_CPT_PF_EXE_EPCI_OUTBX_CNT(b) (0x4001240 | (b) << 3)
+#define OTX2_CPT_PF_ENGX_UCODE_BASE(b) (0x4002000 | (b) << 3)
+#define OTX2_CPT_PF_QX_CTL(b) (0x8000000 | (b) << 20)
+#define OTX2_CPT_PF_QX_GMCTL(b) (0x8000020 | (b) << 20)
+#define OTX2_CPT_PF_QX_CTL2(b) (0x8000100 | (b) << 20)
+#define OTX2_CPT_PF_VFX_MBOXX(b, c) (0x8001000 | (b) << 20 | \
+ (c) << 8)
+
+/* OcteonTX2 CPT LF registers */
+#define OTX2_CPT_LF_CTL (0x10)
+#define OTX2_CPT_LF_DONE_WAIT (0x30)
+#define OTX2_CPT_LF_INPROG (0x40)
+#define OTX2_CPT_LF_DONE (0x50)
+#define OTX2_CPT_LF_DONE_ACK (0x60)
+#define OTX2_CPT_LF_DONE_INT_ENA_W1S (0x90)
+#define OTX2_CPT_LF_DONE_INT_ENA_W1C (0xa0)
+#define OTX2_CPT_LF_MISC_INT (0xb0)
+#define OTX2_CPT_LF_MISC_INT_W1S (0xc0)
+#define OTX2_CPT_LF_MISC_INT_ENA_W1S (0xd0)
+#define OTX2_CPT_LF_MISC_INT_ENA_W1C (0xe0)
+#define OTX2_CPT_LF_Q_BASE (0xf0)
+#define OTX2_CPT_LF_Q_SIZE (0x100)
+#define OTX2_CPT_LF_Q_INST_PTR (0x110)
+#define OTX2_CPT_LF_Q_GRP_PTR (0x120)
+#define OTX2_CPT_LF_NQX(a) (0x400 | (a) << 3)
+#define OTX2_CPT_RVU_FUNC_BLKADDR_SHIFT 20
+/* LMT LF registers */
+#define OTX2_CPT_LMT_LFBASE BIT_ULL(OTX2_CPT_RVU_FUNC_BLKADDR_SHIFT)
+#define OTX2_CPT_LMT_LF_LMTLINEX(a) (OTX2_CPT_LMT_LFBASE | 0x000 | \
+ (a) << 12)
+/* RVU VF registers */
+#define OTX2_RVU_VF_INT (0x20)
+#define OTX2_RVU_VF_INT_W1S (0x28)
+#define OTX2_RVU_VF_INT_ENA_W1S (0x30)
+#define OTX2_RVU_VF_INT_ENA_W1C (0x38)
+
+/*
+ * Enumeration otx2_cpt_ucode_error_code_e
+ *
+ * Enumerates ucode errors
+ */
+enum otx2_cpt_ucode_comp_code_e {
+ OTX2_CPT_UCC_SUCCESS = 0x00,
+ OTX2_CPT_UCC_INVALID_OPCODE = 0x01,
+
+ /* Scatter gather */
+ OTX2_CPT_UCC_SG_WRITE_LENGTH = 0x02,
+ OTX2_CPT_UCC_SG_LIST = 0x03,
+ OTX2_CPT_UCC_SG_NOT_SUPPORTED = 0x04,
+
+};
+
+/*
+ * Enumeration otx2_cpt_comp_e
+ *
+ * OcteonTX2 CPT Completion Enumeration
+ * Enumerates the values of CPT_RES_S[COMPCODE].
+ */
+enum otx2_cpt_comp_e {
+ OTX2_CPT_COMP_E_NOTDONE = 0x00,
+ OTX2_CPT_COMP_E_GOOD = 0x01,
+ OTX2_CPT_COMP_E_FAULT = 0x02,
+ OTX2_CPT_COMP_E_HWERR = 0x04,
+ OTX2_CPT_COMP_E_INSTERR = 0x05,
+ OTX2_CPT_COMP_E_LAST_ENTRY = 0x06
+};
+
+/*
+ * Enumeration otx2_cpt_vf_int_vec_e
+ *
+ * OcteonTX2 CPT VF MSI-X Vector Enumeration
+ * Enumerates the MSI-X interrupt vectors.
+ */
+enum otx2_cpt_vf_int_vec_e {
+ OTX2_CPT_VF_INT_VEC_E_MBOX = 0x00
+};
+
+/*
+ * Enumeration otx2_cpt_lf_int_vec_e
+ *
+ * OcteonTX2 CPT LF MSI-X Vector Enumeration
+ * Enumerates the MSI-X interrupt vectors.
+ */
+enum otx2_cpt_lf_int_vec_e {
+ OTX2_CPT_LF_INT_VEC_E_MISC = 0x00,
+ OTX2_CPT_LF_INT_VEC_E_DONE = 0x01
+};
+
+/*
+ * Structure otx2_cpt_inst_s
+ *
+ * CPT Instruction Structure
+ * This structure specifies the instruction layout. Instructions are
+ * stored in memory as little-endian unless CPT()_PF_Q()_CTL[INST_BE] is set.
+ * cpt_inst_s_s
+ * Word 0
+ * doneint:1 Done interrupt.
+ * 0 = No interrupts related to this instruction.
+ * 1 = When the instruction completes, CPT()_VQ()_DONE[DONE] will be
+ * incremented,and based on the rules described there an interrupt may
+ * occur.
+ * Word 1
+ * res_addr [127: 64] Result IOVA.
+ * If nonzero, specifies where to write CPT_RES_S.
+ * If zero, no result structure will be written.
+ * Address must be 16-byte aligned.
+ * Bits <63:49> are ignored by hardware; software should use a
+ * sign-extended bit <48> for forward compatibility.
+ * Word 2
+ * grp:10 [171:162] If [WQ_PTR] is nonzero, the SSO guest-group to use when
+ * CPT submits work SSO.
+ * For the SSO to not discard the add-work request, FPA_PF_MAP() must map
+ * [GRP] and CPT()_PF_Q()_GMCTL[GMID] as valid.
+ * tt:2 [161:160] If [WQ_PTR] is nonzero, the SSO tag type to use when CPT
+ * submits work to SSO
+ * tag:32 [159:128] If [WQ_PTR] is nonzero, the SSO tag to use when CPT
+ * submits work to SSO.
+ * Word 3
+ * wq_ptr [255:192] If [WQ_PTR] is nonzero, it is a pointer to a
+ * work-queue entry that CPT submits work to SSO after all context,
+ * output data, and result write operations are visible to other
+ * CNXXXX units and the cores. Bits <2:0> must be zero.
+ * Bits <63:49> are ignored by hardware; software should
+ * use a sign-extended bit <48> for forward compatibility.
+ * Internal:
+ * Bits <63:49>, <2:0> are ignored by hardware, treated as always 0x0.
+ * Word 4
+ * ei0; [319:256] Engine instruction word 0. Passed to the AE/SE.
+ * Word 5
+ * ei1; [383:320] Engine instruction word 1. Passed to the AE/SE.
+ * Word 6
+ * ei2; [447:384] Engine instruction word 1. Passed to the AE/SE.
+ * Word 7
+ * ei3; [511:448] Engine instruction word 1. Passed to the AE/SE.
+ *
+ */
+union otx2_cpt_inst_s {
+ u64 u[8];
+
+ struct {
+ /* Word 0 */
+ u64 nixtxl:3;
+ u64 doneint:1;
+ u64 nixtx_addr:60;
+ /* Word 1 */
+ u64 res_addr;
+ /* Word 2 */
+ u64 tag:32;
+ u64 tt:2;
+ u64 grp:10;
+ u64 reserved_172_175:4;
+ u64 rvu_pf_func:16;
+ /* Word 3 */
+ u64 qord:1;
+ u64 reserved_194_193:2;
+ u64 wq_ptr:61;
+ /* Word 4 */
+ u64 ei0;
+ /* Word 5 */
+ u64 ei1;
+ /* Word 6 */
+ u64 ei2;
+ /* Word 7 */
+ u64 ei3;
+ } s;
+};
+
+/*
+ * Structure otx2_cpt_res_s
+ *
+ * CPT Result Structure
+ * The CPT coprocessor writes the result structure after it completes a
+ * CPT_INST_S instruction. The result structure is exactly 16 bytes, and
+ * each instruction completion produces exactly one result structure.
+ *
+ * This structure is stored in memory as little-endian unless
+ * CPT()_PF_Q()_CTL[INST_BE] is set.
+ * cpt_res_s_s
+ * Word 0
+ * doneint:1 [16:16] Done interrupt. This bit is copied from the
+ * corresponding instruction's CPT_INST_S[DONEINT].
+ * compcode:8 [7:0] Indicates completion/error status of the CPT coprocessor
+ * for the associated instruction, as enumerated by CPT_COMP_E.
+ * Core software may write the memory location containing [COMPCODE] to
+ * 0x0 before ringing the doorbell, and then poll for completion by
+ * checking for a nonzero value.
+ * Once the core observes a nonzero [COMPCODE] value in this case,the CPT
+ * coprocessor will have also completed L2/DRAM write operations.
+ * Word 1
+ * reserved
+ *
+ */
+union otx2_cpt_res_s {
+ u64 u[2];
+
+ struct {
+ u64 compcode:8;
+ u64 uc_compcode:8;
+ u64 doneint:1;
+ u64 reserved_17_63:47;
+ u64 reserved_64_127;
+ } s;
+};
+
+/*
+ * Register (RVU_PF_BAR0) cpt#_af_constants1
+ *
+ * CPT AF Constants Register
+ * This register contains implementation-related parameters of CPT.
+ */
+union otx2_cptx_af_constants1 {
+ u64 u;
+ struct otx2_cptx_af_constants1_s {
+ u64 se:16;
+ u64 ie:16;
+ u64 ae:16;
+ u64 reserved_48_63:16;
+ } s;
+};
+
+/*
+ * RVU_PFVF_BAR2 - cpt_lf_misc_int
+ *
+ * This register contain the per-queue miscellaneous interrupts.
+ *
+ */
+union otx2_cptx_lf_misc_int {
+ u64 u;
+ struct otx2_cptx_lf_misc_int_s {
+ u64 reserved_0:1;
+ u64 nqerr:1;
+ u64 irde:1;
+ u64 nwrp:1;
+ u64 reserved_4:1;
+ u64 hwerr:1;
+ u64 fault:1;
+ u64 reserved_7_63:57;
+ } s;
+};
+
+/*
+ * RVU_PFVF_BAR2 - cpt_lf_misc_int_ena_w1s
+ *
+ * This register sets interrupt enable bits.
+ *
+ */
+union otx2_cptx_lf_misc_int_ena_w1s {
+ u64 u;
+ struct otx2_cptx_lf_misc_int_ena_w1s_s {
+ u64 reserved_0:1;
+ u64 nqerr:1;
+ u64 irde:1;
+ u64 nwrp:1;
+ u64 reserved_4:1;
+ u64 hwerr:1;
+ u64 fault:1;
+ u64 reserved_7_63:57;
+ } s;
+};
+
+/*
+ * RVU_PFVF_BAR2 - cpt_lf_ctl
+ *
+ * This register configures the queue.
+ *
+ * When the queue is not execution-quiescent (see CPT_LF_INPROG[EENA,INFLIGHT]),
+ * software must only write this register with [ENA]=0.
+ */
+union otx2_cptx_lf_ctl {
+ u64 u;
+ struct otx2_cptx_lf_ctl_s {
+ u64 ena:1;
+ u64 fc_ena:1;
+ u64 fc_up_crossing:1;
+ u64 reserved_3:1;
+ u64 fc_hyst_bits:4;
+ u64 reserved_8_63:56;
+ } s;
+};
+
+/*
+ * RVU_PFVF_BAR2 - cpt_lf_done_wait
+ *
+ * This register specifies the per-queue interrupt coalescing settings.
+ */
+union otx2_cptx_lf_done_wait {
+ u64 u;
+ struct otx2_cptx_lf_done_wait_s {
+ u64 num_wait:20;
+ u64 reserved_20_31:12;
+ u64 time_wait:16;
+ u64 reserved_48_63:16;
+ } s;
+};
+
+/*
+ * RVU_PFVF_BAR2 - cpt_lf_done
+ *
+ * This register contain the per-queue instruction done count.
+ */
+union otx2_cptx_lf_done {
+ u64 u;
+ struct otx2_cptx_lf_done_s {
+ u64 done:20;
+ u64 reserved_20_63:44;
+ } s;
+};
+
+/*
+ * RVU_PFVF_BAR2 - cpt_lf_inprog
+ *
+ * These registers contain the per-queue instruction in flight registers.
+ *
+ */
+union otx2_cptx_lf_inprog {
+ u64 u;
+ struct otx2_cptx_lf_inprog_s {
+ u64 inflight:9;
+ u64 reserved_9_15:7;
+ u64 eena:1;
+ u64 grp_drp:1;
+ u64 reserved_18_30:13;
+ u64 grb_partial:1;
+ u64 grb_cnt:8;
+ u64 gwb_cnt:8;
+ u64 reserved_48_63:16;
+ } s;
+};
+
+/*
+ * RVU_PFVF_BAR2 - cpt_lf_q_base
+ *
+ * CPT initializes these CSR fields to these values on any CPT_LF_Q_BASE write:
+ * _ CPT_LF_Q_INST_PTR[XQ_XOR]=0.
+ * _ CPT_LF_Q_INST_PTR[NQ_PTR]=2.
+ * _ CPT_LF_Q_INST_PTR[DQ_PTR]=2.
+ * _ CPT_LF_Q_GRP_PTR[XQ_XOR]=0.
+ * _ CPT_LF_Q_GRP_PTR[NQ_PTR]=1.
+ * _ CPT_LF_Q_GRP_PTR[DQ_PTR]=1.
+ */
+union otx2_cptx_lf_q_base {
+ u64 u;
+ struct otx2_cptx_lf_q_base_s {
+ u64 fault:1;
+ u64 reserved_1_6:6;
+ u64 addr:46;
+ u64 reserved_53_63:11;
+ } s;
+};
+
+/*
+ * RVU_PFVF_BAR2 - cpt_lf_q_size
+ *
+ * CPT initializes these CSR fields to these values on any CPT_LF_Q_SIZE write:
+ * _ CPT_LF_Q_INST_PTR[XQ_XOR]=0.
+ * _ CPT_LF_Q_INST_PTR[NQ_PTR]=2.
+ * _ CPT_LF_Q_INST_PTR[DQ_PTR]=2.
+ * _ CPT_LF_Q_GRP_PTR[XQ_XOR]=0.
+ * _ CPT_LF_Q_GRP_PTR[NQ_PTR]=1.
+ * _ CPT_LF_Q_GRP_PTR[DQ_PTR]=1.
+ */
+union otx2_cptx_lf_q_size {
+ u64 u;
+ struct otx2_cptx_lf_q_size_s {
+ u64 size_div40:15;
+ u64 reserved_15_63:49;
+ } s;
+};
+
+/*
+ * RVU_PF_BAR0 - cpt_af_lf_ctl
+ *
+ * This register configures queues. This register should be written only
+ * when the queue is execution-quiescent (see CPT_LF_INPROG[INFLIGHT]).
+ */
+union otx2_cptx_af_lf_ctrl {
+ u64 u;
+ struct otx2_cptx_af_lf_ctrl_s {
+ u64 pri:1;
+ u64 reserved_1_8:8;
+ u64 pf_func_inst:1;
+ u64 cont_err:1;
+ u64 reserved_11_15:5;
+ u64 nixtx_en:1;
+ u64 reserved_17_47:31;
+ u64 grp:8;
+ u64 reserved_56_63:8;
+ } s;
+};
+
+#endif /* __OTX2_CPT_HW_TYPES_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf.h b/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
new file mode 100644
index 000000000000..84cdc8cc2c15
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
@@ -0,0 +1,13 @@
+/* SPDX-License-Identifier: GPL-2.0-only
+ * Copyright (C) 2020 Marvell.
+ */
+
+#ifndef __OTX2_CPTPF_H
+#define __OTX2_CPTPF_H
+
+struct otx2_cptpf_dev {
+ void __iomem *reg_base; /* CPT PF registers start address */
+ struct pci_dev *pdev; /* PCI device handle */
+};
+
+#endif /* __OTX2_CPTPF_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
new file mode 100644
index 000000000000..47781966de3e
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
@@ -0,0 +1,107 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (C) 2020 Marvell. */
+
+#include <linux/firmware.h>
+#include "otx2_cpt_hw_types.h"
+#include "otx2_cpt_common.h"
+#include "otx2_cptpf.h"
+#include "rvu_reg.h"
+
+#define OTX2_CPT_DRV_NAME "octeontx2-cpt"
+#define OTX2_CPT_DRV_STRING "Marvell OcteonTX2 CPT Physical Function Driver"
+
+static int cpt_is_pf_usable(struct otx2_cptpf_dev *cptpf)
+{
+ u64 rev;
+
+ rev = otx2_cpt_read64(cptpf->reg_base, BLKADDR_RVUM, 0,
+ RVU_PF_BLOCK_ADDRX_DISC(BLKADDR_RVUM));
+ rev = (rev >> 12) & 0xFF;
+ /*
+ * Check if AF has setup revision for RVUM block, otherwise
+ * driver probe should be deferred until AF driver comes up
+ */
+ if (!rev) {
+ dev_warn(&cptpf->pdev->dev,
+ "AF is not initialized, deferring probe\n");
+ return -EPROBE_DEFER;
+ }
+ return 0;
+}
+
+static int otx2_cptpf_probe(struct pci_dev *pdev,
+ const struct pci_device_id *ent)
+{
+ struct device *dev = &pdev->dev;
+ struct otx2_cptpf_dev *cptpf;
+ int err;
+
+ cptpf = devm_kzalloc(dev, sizeof(*cptpf), GFP_KERNEL);
+ if (!cptpf)
+ return -ENOMEM;
+
+ err = pcim_enable_device(pdev);
+ if (err) {
+ dev_err(dev, "Failed to enable PCI device\n");
+ goto clear_drvdata;
+ }
+
+ err = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(48));
+ if (err) {
+ dev_err(dev, "Unable to get usable DMA configuration\n");
+ goto clear_drvdata;
+ }
+ /* Map PF's configuration registers */
+ err = pcim_iomap_regions_request_all(pdev, 1 << PCI_PF_REG_BAR_NUM,
+ OTX2_CPT_DRV_NAME);
+ if (err) {
+ dev_err(dev, "Couldn't get PCI resources 0x%x\n", err);
+ goto clear_drvdata;
+ }
+ pci_set_master(pdev);
+ pci_set_drvdata(pdev, cptpf);
+ cptpf->pdev = pdev;
+
+ cptpf->reg_base = pcim_iomap_table(pdev)[PCI_PF_REG_BAR_NUM];
+
+ /* Check if AF driver is up, otherwise defer probe */
+ err = cpt_is_pf_usable(cptpf);
+ if (err)
+ goto clear_drvdata;
+
+ return 0;
+
+clear_drvdata:
+ pci_set_drvdata(pdev, NULL);
+ return err;
+}
+
+static void otx2_cptpf_remove(struct pci_dev *pdev)
+{
+ struct otx2_cptpf_dev *cptpf = pci_get_drvdata(pdev);
+
+ if (!cptpf)
+ return;
+
+ pci_set_drvdata(pdev, NULL);
+}
+
+/* Supported devices */
+static const struct pci_device_id otx2_cpt_id_table[] = {
+ { PCI_DEVICE(PCI_VENDOR_ID_CAVIUM, OTX2_CPT_PCI_PF_DEVICE_ID) },
+ { 0, } /* end of table */
+};
+
+static struct pci_driver otx2_cpt_pci_driver = {
+ .name = OTX2_CPT_DRV_NAME,
+ .id_table = otx2_cpt_id_table,
+ .probe = otx2_cptpf_probe,
+ .remove = otx2_cptpf_remove,
+};
+
+module_pci_driver(otx2_cpt_pci_driver);
+
+MODULE_AUTHOR("Marvell");
+MODULE_DESCRIPTION(OTX2_CPT_DRV_STRING);
+MODULE_LICENSE("GPL v2");
+MODULE_DEVICE_TABLE(pci, otx2_cpt_id_table);
--
2.29.0
Adds support to get engine capabilities and adds a new mailbox
to share capabilities with VF driver.
Signed-off-by: Suheil Chandran <[email protected]>
Signed-off-by: Srujana Challa <[email protected]>
---
.../marvell/octeontx2/otx2_cpt_common.h | 36 ++++
.../marvell/octeontx2/otx2_cpt_reqmgr.h | 51 ++++++
drivers/crypto/marvell/octeontx2/otx2_cptlf.h | 62 +++++++
drivers/crypto/marvell/octeontx2/otx2_cptpf.h | 3 +
.../marvell/octeontx2/otx2_cptpf_main.c | 5 +
.../marvell/octeontx2/otx2_cptpf_mbox.c | 31 ++++
.../marvell/octeontx2/otx2_cptpf_ucode.c | 161 ++++++++++++++++++
.../marvell/octeontx2/otx2_cptpf_ucode.h | 1 +
8 files changed, 350 insertions(+)
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h b/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
index d5576f5d3b90..705a0503b962 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
@@ -20,6 +20,7 @@
#define OTX2_CPT_INVALID_CRYPTO_ENG_GRP 0xFF
#define OTX2_CPT_NAME_LENGTH 64
+#define OTX2_CPT_DMA_MINALIGN 128
#define BAD_OTX2_CPT_ENG_TYPE OTX2_CPT_MAX_ENG_TYPES
@@ -32,6 +33,7 @@ enum otx2_cpt_eng_type {
/* Take mbox id from end of CPT mbox range in AF (range 0xA00 - 0xBFF) */
#define MBOX_MSG_GET_ENG_GRP_NUM 0xBFF
+#define MBOX_MSG_GET_CAPS 0xBFD
/*
* Message request and response to get engine group number
@@ -49,6 +51,40 @@ struct otx2_cpt_egrp_num_rsp {
u8 eng_grp_num;
};
+/* CPT HW capabilities */
+union otx2_cpt_eng_caps {
+ u64 u;
+ struct {
+ u64 reserved_0_4:5;
+ u64 mul:1;
+ u64 sha1_sha2:1;
+ u64 chacha20:1;
+ u64 zuc_snow3g:1;
+ u64 sha3:1;
+ u64 aes:1;
+ u64 kasumi:1;
+ u64 des:1;
+ u64 crc:1;
+ u64 reserved_14_63:50;
+ };
+};
+
+/*
+ * Message request and response to get HW capabilities for each
+ * engine type (SE, IE, AE).
+ * This messages are only used between CPT PF <=> CPT VF
+ */
+struct otx2_cpt_caps_msg {
+ struct mbox_msghdr hdr;
+};
+
+struct otx2_cpt_caps_rsp {
+ struct mbox_msghdr hdr;
+ u16 cpt_pf_drv_version;
+ u8 cpt_revision;
+ union otx2_cpt_eng_caps eng_caps[OTX2_CPT_MAX_ENG_TYPES];
+};
+
static inline void otx2_cpt_write64(void __iomem *reg_base, u64 blk, u64 slot,
u64 offs, u64 val)
{
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h b/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
new file mode 100644
index 000000000000..9184f91c68c1
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
@@ -0,0 +1,51 @@
+/* SPDX-License-Identifier: GPL-2.0-only
+ * Copyright (C) 2020 Marvell.
+ */
+
+#ifndef __OTX2_CPT_REQMGR_H
+#define __OTX2_CPT_REQMGR_H
+
+#include "otx2_cpt_common.h"
+
+/* Completion code size and initial value */
+#define OTX2_CPT_COMPLETION_CODE_SIZE 8
+#define OTX2_CPT_COMPLETION_CODE_INIT OTX2_CPT_COMP_E_NOTDONE
+
+union otx2_cpt_opcode {
+ u16 flags;
+ struct {
+ u8 major;
+ u8 minor;
+ } s;
+};
+
+/*
+ * CPT_INST_S software command definitions
+ * Words EI (0-3)
+ */
+union otx2_cpt_iq_cmd_word0 {
+ u64 u;
+ struct {
+ __be16 opcode;
+ __be16 param1;
+ __be16 param2;
+ __be16 dlen;
+ } s;
+};
+
+union otx2_cpt_iq_cmd_word3 {
+ u64 u;
+ struct {
+ u64 cptr:61;
+ u64 grp:3;
+ } s;
+};
+
+struct otx2_cpt_iq_command {
+ union otx2_cpt_iq_cmd_word0 cmd;
+ u64 dptr;
+ u64 rptr;
+ union otx2_cpt_iq_cmd_word3 cptr;
+};
+
+#endif /* __OTX2_CPT_REQMGR_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptlf.h b/drivers/crypto/marvell/octeontx2/otx2_cptlf.h
index 37ae0d2dcf55..efb4f395a54a 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptlf.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptlf.h
@@ -4,9 +4,11 @@
#ifndef __OTX2_CPTLF_H
#define __OTX2_CPTLF_H
+#include <linux/soc/marvell/octeontx2/asm.h>
#include <mbox.h>
#include <rvu.h>
#include "otx2_cpt_common.h"
+#include "otx2_cpt_reqmgr.h"
/*
* CPT instruction and pending queues user requested length in CPT_INST_S msgs
@@ -272,6 +274,66 @@ static inline void otx2_cptlf_enable_iqueues(struct otx2_cptlfs_info *lfs)
}
}
+static inline void otx2_cpt_fill_inst(union otx2_cpt_inst_s *cptinst,
+ struct otx2_cpt_iq_command *iq_cmd,
+ u64 comp_baddr)
+{
+ cptinst->u[0] = 0x0;
+ cptinst->s.doneint = true;
+ cptinst->s.res_addr = comp_baddr;
+ cptinst->u[2] = 0x0;
+ cptinst->u[3] = 0x0;
+ cptinst->s.ei0 = iq_cmd->cmd.u;
+ cptinst->s.ei1 = iq_cmd->dptr;
+ cptinst->s.ei2 = iq_cmd->rptr;
+ cptinst->s.ei3 = iq_cmd->cptr.u;
+}
+
+/*
+ * On OcteonTX2 platform the parameter insts_num is used as a count of
+ * instructions to be enqueued. The valid values for insts_num are:
+ * 1 - 1 CPT instruction will be enqueued during LMTST operation
+ * 2 - 2 CPT instructions will be enqueued during LMTST operation
+ */
+static inline void otx2_cpt_send_cmd(union otx2_cpt_inst_s *cptinst,
+ u32 insts_num, struct otx2_cptlf_info *lf)
+{
+ void __iomem *lmtline = lf->lmtline;
+ long ret;
+
+ /*
+ * Make sure memory areas pointed in CPT_INST_S
+ * are flushed before the instruction is sent to CPT
+ */
+ dma_wmb();
+
+ do {
+ /* Copy CPT command to LMTLINE */
+ memcpy_toio(lmtline, cptinst, insts_num * OTX2_CPT_INST_SIZE);
+
+ /*
+ * LDEOR initiates atomic transfer to I/O device
+ * The following will cause the LMTST to fail (the LDEOR
+ * returns zero):
+ * - No stores have been performed to the LMTLINE since it was
+ * last invalidated.
+ * - The bytes which have been stored to LMTLINE since it was
+ * last invalidated form a pattern that is non-contiguous, does
+ * not start at byte 0, or does not end on a 8-byte boundary.
+ * (i.e.comprises a formation of other than 1–16 8-byte
+ * words.)
+ *
+ * These rules are designed such that an operating system
+ * context switch or hypervisor guest switch need have no
+ * knowledge of the LMTST operations; the switch code does not
+ * need to store to LMTCANCEL. Also note as LMTLINE data cannot
+ * be read, there is no information leakage between processes.
+ */
+ ret = otx2_lmt_flush(lf->ioreg);
+
+ } while (!ret);
+}
+
int otx2_cptlf_init(struct otx2_cptlfs_info *lfs, u8 eng_grp_msk, int pri,
int lfs_num);
void otx2_cptlf_shutdown(struct otx2_cptlfs_info *lfs);
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf.h b/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
index f441c2dbe7bf..ede230dc205f 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
@@ -31,6 +31,9 @@ struct otx2_cptpf_dev {
struct otx2_cptvf_info vf[OTX2_CPT_MAX_VFS_NUM];
struct otx2_cpt_eng_grps eng_grps;/* Engine groups information */
struct otx2_cptlfs_info lfs; /* CPT LFs attached to this PF */
+ /* HW capabilities for each engine type */
+ union otx2_cpt_eng_caps eng_caps[OTX2_CPT_MAX_ENG_TYPES];
+ bool is_eng_caps_discovered;
/* AF <=> PF mbox */
struct otx2_mbox afpf_mbox;
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
index f0ad45c04df7..502bcf21ea89 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
@@ -500,6 +500,11 @@ static int cptpf_sriov_enable(struct pci_dev *pdev, int num_vfs)
if (ret)
goto destroy_flr;
+ /* Get CPT HW capabilities using LOAD_FVC operation. */
+ ret = otx2_cpt_discover_eng_capabilities(cptpf);
+ if (ret)
+ goto disable_intr;
+
ret = otx2_cpt_create_eng_grps(cptpf->pdev, &cptpf->eng_grps);
if (ret)
goto disable_intr;
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
index 2270967eaf65..0834dc3b3e59 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
@@ -5,6 +5,12 @@
#include "otx2_cptpf.h"
#include "rvu_reg.h"
+/*
+ * CPT PF driver version, It will be incremented by 1 for every feature
+ * addition in CPT mailbox messages.
+ */
+#define OTX2_CPT_PF_DRV_VERSION 0x1
+
static int forward_to_af(struct otx2_cptpf_dev *cptpf,
struct otx2_cptvf_info *vf,
struct mbox_msghdr *req, int size)
@@ -35,6 +41,28 @@ static int forward_to_af(struct otx2_cptpf_dev *cptpf,
return 0;
}
+static int handle_msg_get_caps(struct otx2_cptpf_dev *cptpf,
+ struct otx2_cptvf_info *vf,
+ struct mbox_msghdr *req)
+{
+ struct otx2_cpt_caps_rsp *rsp;
+
+ rsp = (struct otx2_cpt_caps_rsp *)
+ otx2_mbox_alloc_msg(&cptpf->vfpf_mbox, vf->vf_id,
+ sizeof(*rsp));
+ if (!rsp)
+ return -ENOMEM;
+
+ rsp->hdr.id = MBOX_MSG_GET_CAPS;
+ rsp->hdr.sig = OTX2_MBOX_RSP_SIG;
+ rsp->hdr.pcifunc = req->pcifunc;
+ rsp->cpt_pf_drv_version = OTX2_CPT_PF_DRV_VERSION;
+ rsp->cpt_revision = cptpf->pdev->revision;
+ memcpy(&rsp->eng_caps, &cptpf->eng_caps, sizeof(rsp->eng_caps));
+
+ return 0;
+}
+
static int handle_msg_get_eng_grp_num(struct otx2_cptpf_dev *cptpf,
struct otx2_cptvf_info *vf,
struct mbox_msghdr *req)
@@ -72,6 +100,9 @@ static int cptpf_handle_vf_req(struct otx2_cptpf_dev *cptpf,
case MBOX_MSG_GET_ENG_GRP_NUM:
err = handle_msg_get_eng_grp_num(cptpf, vf, req);
break;
+ case MBOX_MSG_GET_CAPS:
+ err = handle_msg_get_caps(cptpf, vf, req);
+ break;
default:
err = forward_to_af(cptpf, vf, req, size);
break;
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
index 64f0e7e2652a..1dc3ba298139 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
@@ -6,6 +6,8 @@
#include "otx2_cptpf_ucode.h"
#include "otx2_cpt_common.h"
#include "otx2_cptpf.h"
+#include "otx2_cptlf.h"
+#include "otx2_cpt_reqmgr.h"
#include "rvu_reg.h"
#define CSR_DELAY 30
@@ -1252,3 +1254,162 @@ int otx2_cpt_init_eng_grps(struct pci_dev *pdev,
otx2_cpt_cleanup_eng_grps(pdev, eng_grps);
return ret;
}
+
+static int create_eng_caps_discovery_grps(struct pci_dev *pdev,
+ struct otx2_cpt_eng_grps *eng_grps)
+{
+ struct otx2_cpt_uc_info_t *uc_info[OTX2_CPT_MAX_ETYPES_PER_GRP] = { };
+ struct otx2_cpt_engines engs[OTX2_CPT_MAX_ETYPES_PER_GRP] = { {0} };
+ struct fw_info_t fw_info;
+ int ret;
+
+ ret = cpt_ucode_load_fw(pdev, &fw_info);
+ if (ret)
+ return ret;
+
+ uc_info[0] = get_ucode(&fw_info, OTX2_CPT_SE_TYPES);
+ if (uc_info[0] == NULL) {
+ dev_err(&pdev->dev, "Unable to find firmware for AE\n");
+ ret = -EINVAL;
+ goto release_fw;
+ }
+ engs[0].type = OTX2_CPT_AE_TYPES;
+ engs[0].count = 2;
+
+ ret = create_engine_group(&pdev->dev, eng_grps, engs, 1,
+ (void **) uc_info, 0);
+ if (ret)
+ goto release_fw;
+
+ uc_info[0] = get_ucode(&fw_info, OTX2_CPT_SE_TYPES);
+ if (uc_info[0] == NULL) {
+ dev_err(&pdev->dev, "Unable to find firmware for SE\n");
+ ret = -EINVAL;
+ goto delete_eng_grp;
+ }
+ engs[0].type = OTX2_CPT_SE_TYPES;
+ engs[0].count = 2;
+
+ ret = create_engine_group(&pdev->dev, eng_grps, engs, 1,
+ (void **) uc_info, 0);
+ if (ret)
+ goto delete_eng_grp;
+
+ uc_info[0] = get_ucode(&fw_info, OTX2_CPT_IE_TYPES);
+ if (uc_info[0] == NULL) {
+ dev_err(&pdev->dev, "Unable to find firmware for IE\n");
+ ret = -EINVAL;
+ goto delete_eng_grp;
+ }
+ engs[0].type = OTX2_CPT_IE_TYPES;
+ engs[0].count = 2;
+
+ ret = create_engine_group(&pdev->dev, eng_grps, engs, 1,
+ (void **) uc_info, 0);
+ if (ret)
+ goto delete_eng_grp;
+
+ cpt_ucode_release_fw(&fw_info);
+ return 0;
+
+delete_eng_grp:
+ delete_engine_grps(pdev, eng_grps);
+release_fw:
+ cpt_ucode_release_fw(&fw_info);
+ return ret;
+}
+
+/*
+ * Get CPT HW capabilities using LOAD_FVC operation.
+ */
+int otx2_cpt_discover_eng_capabilities(struct otx2_cptpf_dev *cptpf)
+{
+ struct otx2_cptlfs_info *lfs = &cptpf->lfs;
+ struct otx2_cpt_iq_command iq_cmd;
+ union otx2_cpt_opcode opcode;
+ union otx2_cpt_res_s *result;
+ union otx2_cpt_inst_s inst;
+ dma_addr_t rptr_baddr;
+ struct pci_dev *pdev;
+ u32 len, compl_rlen;
+ int ret, etype;
+ void *rptr;
+
+ /*
+ * We don't get capabilities if it was already done
+ * (when user enabled VFs for the first time)
+ */
+ if (cptpf->is_eng_caps_discovered)
+ return 0;
+
+ pdev = cptpf->pdev;
+ /*
+ * Create engine groups for each type to submit LOAD_FVC op and
+ * get engine's capabilities.
+ */
+ ret = create_eng_caps_discovery_grps(pdev, &cptpf->eng_grps);
+ if (ret)
+ goto delete_grps;
+
+ lfs->pdev = pdev;
+ lfs->reg_base = cptpf->reg_base;
+ lfs->mbox = &cptpf->afpf_mbox;
+ ret = otx2_cptlf_init(&cptpf->lfs, OTX2_CPT_ALL_ENG_GRPS_MASK,
+ OTX2_CPT_QUEUE_HI_PRIO, 1);
+ if (ret)
+ goto delete_grps;
+
+ compl_rlen = ALIGN(sizeof(union otx2_cpt_res_s), OTX2_CPT_DMA_MINALIGN);
+ len = compl_rlen + LOADFVC_RLEN;
+
+ result = kzalloc(len, GFP_KERNEL);
+ if (!result) {
+ ret = -ENOMEM;
+ goto lf_cleanup;
+ }
+ rptr_baddr = dma_map_single(&pdev->dev, (void *)result, len,
+ DMA_BIDIRECTIONAL);
+ if (dma_mapping_error(&pdev->dev, rptr_baddr)) {
+ dev_err(&pdev->dev, "DMA mapping failed\n");
+ ret = -EFAULT;
+ goto free_result;
+ }
+ rptr = (u8 *)result + compl_rlen;
+
+ /* Fill in the command */
+ opcode.s.major = LOADFVC_MAJOR_OP;
+ opcode.s.minor = LOADFVC_MINOR_OP;
+
+ iq_cmd.cmd.u = 0;
+ iq_cmd.cmd.s.opcode = cpu_to_be16(opcode.flags);
+
+ /* 64-bit swap for microcode data reads, not needed for addresses */
+ cpu_to_be64s(&iq_cmd.cmd.u);
+ iq_cmd.dptr = 0;
+ iq_cmd.rptr = rptr_baddr + compl_rlen;
+ iq_cmd.cptr.u = 0;
+
+ for (etype = 1; etype < OTX2_CPT_MAX_ENG_TYPES; etype++) {
+ result->s.compcode = OTX2_CPT_COMPLETION_CODE_INIT;
+ iq_cmd.cptr.s.grp = otx2_cpt_get_eng_grp(&cptpf->eng_grps,
+ etype);
+ otx2_cpt_fill_inst(&inst, &iq_cmd, rptr_baddr);
+ otx2_cpt_send_cmd(&inst, 1, &cptpf->lfs.lf[0]);
+
+ while (result->s.compcode == OTX2_CPT_COMPLETION_CODE_INIT)
+ cpu_relax();
+
+ cptpf->eng_caps[etype].u = be64_to_cpup(rptr);
+ }
+ dma_unmap_single(&pdev->dev, rptr_baddr, len, DMA_BIDIRECTIONAL);
+ cptpf->is_eng_caps_discovered = true;
+
+free_result:
+ kfree(result);
+lf_cleanup:
+ otx2_cptlf_shutdown(&cptpf->lfs);
+delete_grps:
+ delete_engine_grps(pdev, &cptpf->eng_grps);
+
+ return ret;
+}
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.h b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.h
index 96556399a58c..6b0d432de0af 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.h
@@ -157,5 +157,6 @@ int otx2_cpt_create_eng_grps(struct pci_dev *pdev,
struct otx2_cpt_eng_grps *eng_grps);
int otx2_cpt_disable_all_cores(struct otx2_cptpf_dev *cptpf);
int otx2_cpt_get_eng_grp(struct otx2_cpt_eng_grps *eng_grps, int eng_type);
+int otx2_cpt_discover_eng_capabilities(struct otx2_cptpf_dev *cptpf);
#endif /* __OTX2_CPTPF_UCODE_H */
--
2.29.0
CPT offload module utilises the linux crypto framework to offload
crypto processing. This patch registers supported algorithms by
calling registration functions provided by the kernel crypto API.
The module currently supports:
- AES block cipher in CBC,ECB and XTS mode.
- 3DES block cipher in CBC and ECB mode.
- AEAD algorithms.
authenc(hmac(sha1),cbc(aes)),
authenc(hmac(sha256),cbc(aes)),
authenc(hmac(sha384),cbc(aes)),
authenc(hmac(sha512),cbc(aes)),
authenc(hmac(sha1),ecb(cipher_null)),
authenc(hmac(sha256),ecb(cipher_null)),
authenc(hmac(sha384),ecb(cipher_null)),
authenc(hmac(sha512),ecb(cipher_null)),
rfc4106(gcm(aes)).
Signed-off-by: Suheil Chandran <[email protected]>
Signed-off-by: Lukasz Bartosik <[email protected]>
Signed-off-by: Srujana Challa <[email protected]>
---
drivers/crypto/marvell/Kconfig | 4 +
drivers/crypto/marvell/octeontx2/Makefile | 3 +-
.../marvell/octeontx2/otx2_cpt_reqmgr.h | 1 +
.../marvell/octeontx2/otx2_cptvf_algs.c | 1758 +++++++++++++++++
.../marvell/octeontx2/otx2_cptvf_algs.h | 178 ++
.../marvell/octeontx2/otx2_cptvf_main.c | 12 +-
.../marvell/octeontx2/otx2_cptvf_reqmgr.c | 7 +
7 files changed, 1961 insertions(+), 2 deletions(-)
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.h
diff --git a/drivers/crypto/marvell/Kconfig b/drivers/crypto/marvell/Kconfig
index 1440ec9e1fb4..2efbd79180ce 100644
--- a/drivers/crypto/marvell/Kconfig
+++ b/drivers/crypto/marvell/Kconfig
@@ -40,8 +40,12 @@ config CRYPTO_DEV_OCTEONTX2_CPT
tristate "Marvell OcteonTX2 CPT driver"
depends on ARM64 || COMPILE_TEST
depends on PCI_MSI && 64BIT
+ depends on CRYPTO_LIB_AES
select OCTEONTX2_MBOX
select CRYPTO_DEV_MARVELL
+ select CRYPTO_SKCIPHER
+ select CRYPTO_HASH
+ select CRYPTO_AEAD
help
This driver allows you to utilize the Marvell Cryptographic
Accelerator Unit(CPT) found in OcteonTX2 series of processors.
diff --git a/drivers/crypto/marvell/octeontx2/Makefile b/drivers/crypto/marvell/octeontx2/Makefile
index 41c0a5832b3f..b9c6201019e0 100644
--- a/drivers/crypto/marvell/octeontx2/Makefile
+++ b/drivers/crypto/marvell/octeontx2/Makefile
@@ -4,6 +4,7 @@ obj-$(CONFIG_CRYPTO_DEV_OCTEONTX2_CPT) += octeontx2-cpt.o octeontx2-cptvf.o
octeontx2-cpt-objs := otx2_cptpf_main.o otx2_cptpf_mbox.o \
otx2_cpt_mbox_common.o otx2_cptpf_ucode.o otx2_cptlf.o
octeontx2-cptvf-objs := otx2_cptvf_main.o otx2_cptvf_mbox.o otx2_cptlf.o \
- otx2_cpt_mbox_common.o otx2_cptvf_reqmgr.o
+ otx2_cpt_mbox_common.o otx2_cptvf_reqmgr.o \
+ otx2_cptvf_algs.o
ccflags-y += -I$(srctree)/drivers/net/ethernet/marvell/octeontx2/af
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h b/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
index 597a998c6df6..dbb1ee746f4c 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
@@ -192,5 +192,6 @@ struct otx2_cptlf_wqe;
int otx2_cpt_do_request(struct pci_dev *pdev, struct otx2_cpt_req_info *req,
int cpu_num);
void otx2_cpt_post_process(struct otx2_cptlf_wqe *wqe);
+int otx2_cpt_get_kcrypto_eng_grp_num(struct pci_dev *pdev);
#endif /* __OTX2_CPT_REQMGR_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.c b/drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.c
new file mode 100644
index 000000000000..a72723455df7
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.c
@@ -0,0 +1,1758 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (C) 2020 Marvell. */
+
+#include <crypto/aes.h>
+#include <crypto/authenc.h>
+#include <crypto/cryptd.h>
+#include <crypto/des.h>
+#include <crypto/internal/aead.h>
+#include <crypto/sha1.h>
+#include <crypto/sha2.h>
+#include <crypto/xts.h>
+#include <crypto/gcm.h>
+#include <crypto/scatterwalk.h>
+#include <linux/rtnetlink.h>
+#include <linux/sort.h>
+#include <linux/module.h>
+#include "otx2_cptvf.h"
+#include "otx2_cptvf_algs.h"
+#include "otx2_cpt_reqmgr.h"
+
+/* Size of salt in AES GCM mode */
+#define AES_GCM_SALT_SIZE 4
+/* Size of IV in AES GCM mode */
+#define AES_GCM_IV_SIZE 8
+/* Size of ICV (Integrity Check Value) in AES GCM mode */
+#define AES_GCM_ICV_SIZE 16
+/* Offset of IV in AES GCM mode */
+#define AES_GCM_IV_OFFSET 8
+#define CONTROL_WORD_LEN 8
+#define KEY2_OFFSET 48
+#define DMA_MODE_FLAG(dma_mode) \
+ (((dma_mode) == OTX2_CPT_DMA_MODE_SG) ? (1 << 7) : 0)
+
+/* Truncated SHA digest size */
+#define SHA1_TRUNC_DIGEST_SIZE 12
+#define SHA256_TRUNC_DIGEST_SIZE 16
+#define SHA384_TRUNC_DIGEST_SIZE 24
+#define SHA512_TRUNC_DIGEST_SIZE 32
+
+static DEFINE_MUTEX(mutex);
+static int is_crypto_registered;
+
+struct cpt_device_desc {
+ struct pci_dev *dev;
+ int num_queues;
+};
+
+struct cpt_device_table {
+ atomic_t count;
+ struct cpt_device_desc desc[OTX2_CPT_MAX_LFS_NUM];
+};
+
+static struct cpt_device_table se_devices = {
+ .count = ATOMIC_INIT(0)
+};
+
+static inline int get_se_device(struct pci_dev **pdev, int *cpu_num)
+{
+ int count;
+
+ count = atomic_read(&se_devices.count);
+ if (count < 1)
+ return -ENODEV;
+
+ *cpu_num = get_cpu();
+ /*
+ * On OcteonTX2 platform CPT instruction queue is bound to each
+ * local function LF, in turn LFs can be attached to PF
+ * or VF therefore we always use first device. We get maximum
+ * performance if one CPT queue is available for each cpu
+ * otherwise CPT queues need to be shared between cpus.
+ */
+ if (*cpu_num >= se_devices.desc[0].num_queues)
+ *cpu_num %= se_devices.desc[0].num_queues;
+ *pdev = se_devices.desc[0].dev;
+
+ put_cpu();
+
+ return 0;
+}
+
+static inline int validate_hmac_cipher_null(struct otx2_cpt_req_info *cpt_req)
+{
+ struct otx2_cpt_req_ctx *rctx;
+ struct aead_request *req;
+ struct crypto_aead *tfm;
+
+ req = container_of(cpt_req->areq, struct aead_request, base);
+ tfm = crypto_aead_reqtfm(req);
+ rctx = aead_request_ctx(req);
+ if (memcmp(rctx->fctx.hmac.s.hmac_calc,
+ rctx->fctx.hmac.s.hmac_recv,
+ crypto_aead_authsize(tfm)) != 0)
+ return -EBADMSG;
+
+ return 0;
+}
+
+static void otx2_cpt_aead_callback(int status, void *arg1, void *arg2)
+{
+ struct otx2_cpt_inst_info *inst_info = arg2;
+ struct crypto_async_request *areq = arg1;
+ struct otx2_cpt_req_info *cpt_req;
+ struct pci_dev *pdev;
+
+ if (inst_info) {
+ cpt_req = inst_info->req;
+ if (!status) {
+ /*
+ * When selected cipher is NULL we need to manually
+ * verify whether calculated hmac value matches
+ * received hmac value
+ */
+ if (cpt_req->req_type ==
+ OTX2_CPT_AEAD_ENC_DEC_NULL_REQ &&
+ !cpt_req->is_enc)
+ status = validate_hmac_cipher_null(cpt_req);
+ }
+ pdev = inst_info->pdev;
+ otx2_cpt_info_destroy(pdev, inst_info);
+ }
+ if (areq)
+ areq->complete(areq, status);
+}
+
+static void output_iv_copyback(struct crypto_async_request *areq)
+{
+ struct otx2_cpt_req_info *req_info;
+ struct otx2_cpt_req_ctx *rctx;
+ struct skcipher_request *sreq;
+ struct crypto_skcipher *stfm;
+ struct otx2_cpt_enc_ctx *ctx;
+ u32 start, ivsize;
+
+ sreq = container_of(areq, struct skcipher_request, base);
+ stfm = crypto_skcipher_reqtfm(sreq);
+ ctx = crypto_skcipher_ctx(stfm);
+ if (ctx->cipher_type == OTX2_CPT_AES_CBC ||
+ ctx->cipher_type == OTX2_CPT_DES3_CBC) {
+ rctx = skcipher_request_ctx(sreq);
+ req_info = &rctx->cpt_req;
+ ivsize = crypto_skcipher_ivsize(stfm);
+ start = sreq->cryptlen - ivsize;
+
+ if (req_info->is_enc) {
+ scatterwalk_map_and_copy(sreq->iv, sreq->dst, start,
+ ivsize, 0);
+ } else {
+ if (sreq->src != sreq->dst) {
+ scatterwalk_map_and_copy(sreq->iv, sreq->src,
+ start, ivsize, 0);
+ } else {
+ memcpy(sreq->iv, req_info->iv_out, ivsize);
+ kfree(req_info->iv_out);
+ }
+ }
+ }
+}
+
+static void otx2_cpt_skcipher_callback(int status, void *arg1, void *arg2)
+{
+ struct otx2_cpt_inst_info *inst_info = arg2;
+ struct crypto_async_request *areq = arg1;
+ struct pci_dev *pdev;
+
+ if (areq) {
+ if (!status)
+ output_iv_copyback(areq);
+ if (inst_info) {
+ pdev = inst_info->pdev;
+ otx2_cpt_info_destroy(pdev, inst_info);
+ }
+ areq->complete(areq, status);
+ }
+}
+
+static inline void update_input_data(struct otx2_cpt_req_info *req_info,
+ struct scatterlist *inp_sg,
+ u32 nbytes, u32 *argcnt)
+{
+ req_info->req.dlen += nbytes;
+
+ while (nbytes) {
+ u32 len = (nbytes < inp_sg->length) ? nbytes : inp_sg->length;
+ u8 *ptr = sg_virt(inp_sg);
+
+ req_info->in[*argcnt].vptr = (void *)ptr;
+ req_info->in[*argcnt].size = len;
+ nbytes -= len;
+ ++(*argcnt);
+ inp_sg = sg_next(inp_sg);
+ }
+}
+
+static inline void update_output_data(struct otx2_cpt_req_info *req_info,
+ struct scatterlist *outp_sg,
+ u32 offset, u32 nbytes, u32 *argcnt)
+{
+ u32 len, sg_len;
+ u8 *ptr;
+
+ req_info->rlen += nbytes;
+
+ while (nbytes) {
+ sg_len = outp_sg->length - offset;
+ len = (nbytes < sg_len) ? nbytes : sg_len;
+ ptr = sg_virt(outp_sg);
+
+ req_info->out[*argcnt].vptr = (void *) (ptr + offset);
+ req_info->out[*argcnt].size = len;
+ nbytes -= len;
+ ++(*argcnt);
+ offset = 0;
+ outp_sg = sg_next(outp_sg);
+ }
+}
+
+static inline int create_ctx_hdr(struct skcipher_request *req, u32 enc,
+ u32 *argcnt)
+{
+ struct crypto_skcipher *stfm = crypto_skcipher_reqtfm(req);
+ struct otx2_cpt_req_ctx *rctx = skcipher_request_ctx(req);
+ struct otx2_cpt_enc_ctx *ctx = crypto_skcipher_ctx(stfm);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ struct otx2_cpt_fc_ctx *fctx = &rctx->fctx;
+ int ivsize = crypto_skcipher_ivsize(stfm);
+ u32 start = req->cryptlen - ivsize;
+ gfp_t flags;
+
+ flags = (req->base.flags & CRYPTO_TFM_REQ_MAY_SLEEP) ?
+ GFP_KERNEL : GFP_ATOMIC;
+ req_info->ctrl.s.dma_mode = OTX2_CPT_DMA_MODE_SG;
+ req_info->ctrl.s.se_req = 1;
+
+ req_info->req.opcode.s.major = OTX2_CPT_MAJOR_OP_FC |
+ DMA_MODE_FLAG(OTX2_CPT_DMA_MODE_SG);
+ if (enc) {
+ req_info->req.opcode.s.minor = 2;
+ } else {
+ req_info->req.opcode.s.minor = 3;
+ if ((ctx->cipher_type == OTX2_CPT_AES_CBC ||
+ ctx->cipher_type == OTX2_CPT_DES3_CBC) &&
+ req->src == req->dst) {
+ req_info->iv_out = kmalloc(ivsize, flags);
+ if (!req_info->iv_out)
+ return -ENOMEM;
+
+ scatterwalk_map_and_copy(req_info->iv_out, req->src,
+ start, ivsize, 0);
+ }
+ }
+ /* Encryption data length */
+ req_info->req.param1 = req->cryptlen;
+ /* Authentication data length */
+ req_info->req.param2 = 0;
+
+ fctx->enc.enc_ctrl.e.enc_cipher = ctx->cipher_type;
+ fctx->enc.enc_ctrl.e.aes_key = ctx->key_type;
+ fctx->enc.enc_ctrl.e.iv_source = OTX2_CPT_FROM_CPTR;
+
+ if (ctx->cipher_type == OTX2_CPT_AES_XTS)
+ memcpy(fctx->enc.encr_key, ctx->enc_key, ctx->key_len * 2);
+ else
+ memcpy(fctx->enc.encr_key, ctx->enc_key, ctx->key_len);
+
+ memcpy(fctx->enc.encr_iv, req->iv, crypto_skcipher_ivsize(stfm));
+
+ cpu_to_be64s(&fctx->enc.enc_ctrl.u);
+
+ /*
+ * Storing Packet Data Information in offset
+ * Control Word First 8 bytes
+ */
+ req_info->in[*argcnt].vptr = (u8 *)&rctx->ctrl_word;
+ req_info->in[*argcnt].size = CONTROL_WORD_LEN;
+ req_info->req.dlen += CONTROL_WORD_LEN;
+ ++(*argcnt);
+
+ req_info->in[*argcnt].vptr = (u8 *)fctx;
+ req_info->in[*argcnt].size = sizeof(struct otx2_cpt_fc_ctx);
+ req_info->req.dlen += sizeof(struct otx2_cpt_fc_ctx);
+
+ ++(*argcnt);
+
+ return 0;
+}
+
+static inline int create_input_list(struct skcipher_request *req, u32 enc,
+ u32 enc_iv_len)
+{
+ struct otx2_cpt_req_ctx *rctx = skcipher_request_ctx(req);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ u32 argcnt = 0;
+ int ret;
+
+ ret = create_ctx_hdr(req, enc, &argcnt);
+ if (ret)
+ return ret;
+
+ update_input_data(req_info, req->src, req->cryptlen, &argcnt);
+ req_info->in_cnt = argcnt;
+
+ return 0;
+}
+
+static inline void create_output_list(struct skcipher_request *req,
+ u32 enc_iv_len)
+{
+ struct otx2_cpt_req_ctx *rctx = skcipher_request_ctx(req);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ u32 argcnt = 0;
+
+ /*
+ * OUTPUT Buffer Processing
+ * AES encryption/decryption output would be
+ * received in the following format
+ *
+ * ------IV--------|------ENCRYPTED/DECRYPTED DATA-----|
+ * [ 16 Bytes/ [ Request Enc/Dec/ DATA Len AES CBC ]
+ */
+ update_output_data(req_info, req->dst, 0, req->cryptlen, &argcnt);
+ req_info->out_cnt = argcnt;
+}
+
+static int skcipher_do_fallback(struct skcipher_request *req, bool is_enc)
+{
+ struct crypto_skcipher *stfm = crypto_skcipher_reqtfm(req);
+ struct otx2_cpt_req_ctx *rctx = skcipher_request_ctx(req);
+ struct otx2_cpt_enc_ctx *ctx = crypto_skcipher_ctx(stfm);
+ int ret;
+
+ if (ctx->fbk_cipher) {
+ skcipher_request_set_tfm(&rctx->sk_fbk_req, ctx->fbk_cipher);
+ skcipher_request_set_callback(&rctx->sk_fbk_req,
+ req->base.flags,
+ req->base.complete,
+ req->base.data);
+ skcipher_request_set_crypt(&rctx->sk_fbk_req, req->src,
+ req->dst, req->cryptlen, req->iv);
+ ret = is_enc ? crypto_skcipher_encrypt(&rctx->sk_fbk_req) :
+ crypto_skcipher_decrypt(&rctx->sk_fbk_req);
+ } else {
+ ret = -EINVAL;
+ }
+ return ret;
+}
+
+static inline int cpt_enc_dec(struct skcipher_request *req, u32 enc)
+{
+ struct crypto_skcipher *stfm = crypto_skcipher_reqtfm(req);
+ struct otx2_cpt_req_ctx *rctx = skcipher_request_ctx(req);
+ struct otx2_cpt_enc_ctx *ctx = crypto_skcipher_ctx(stfm);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ u32 enc_iv_len = crypto_skcipher_ivsize(stfm);
+ struct pci_dev *pdev;
+ int status, cpu_num;
+
+ if (req->cryptlen == 0)
+ return 0;
+
+ if (!IS_ALIGNED(req->cryptlen, ctx->enc_align_len))
+ return -EINVAL;
+
+ if (req->cryptlen > OTX2_CPT_MAX_REQ_SIZE)
+ return skcipher_do_fallback(req, enc);
+
+ /* Clear control words */
+ rctx->ctrl_word.flags = 0;
+ rctx->fctx.enc.enc_ctrl.u = 0;
+
+ status = create_input_list(req, enc, enc_iv_len);
+ if (status)
+ return status;
+ create_output_list(req, enc_iv_len);
+
+ status = get_se_device(&pdev, &cpu_num);
+ if (status)
+ return status;
+
+ req_info->callback = otx2_cpt_skcipher_callback;
+ req_info->areq = &req->base;
+ req_info->req_type = OTX2_CPT_ENC_DEC_REQ;
+ req_info->is_enc = enc;
+ req_info->is_trunc_hmac = false;
+ req_info->ctrl.s.grp = otx2_cpt_get_kcrypto_eng_grp_num(pdev);
+
+ /*
+ * We perform an asynchronous send and once
+ * the request is completed the driver would
+ * intimate through registered call back functions
+ */
+ status = otx2_cpt_do_request(pdev, req_info, cpu_num);
+
+ return status;
+}
+
+static int otx2_cpt_skcipher_encrypt(struct skcipher_request *req)
+{
+ return cpt_enc_dec(req, true);
+}
+
+static int otx2_cpt_skcipher_decrypt(struct skcipher_request *req)
+{
+ return cpt_enc_dec(req, false);
+}
+
+static int otx2_cpt_skcipher_xts_setkey(struct crypto_skcipher *tfm,
+ const u8 *key, u32 keylen)
+{
+ struct otx2_cpt_enc_ctx *ctx = crypto_skcipher_ctx(tfm);
+ const u8 *key2 = key + (keylen / 2);
+ const u8 *key1 = key;
+ int ret;
+
+ ret = xts_check_key(crypto_skcipher_tfm(tfm), key, keylen);
+ if (ret)
+ return ret;
+ ctx->key_len = keylen;
+ ctx->enc_align_len = 1;
+ memcpy(ctx->enc_key, key1, keylen / 2);
+ memcpy(ctx->enc_key + KEY2_OFFSET, key2, keylen / 2);
+ ctx->cipher_type = OTX2_CPT_AES_XTS;
+ switch (ctx->key_len) {
+ case 2 * AES_KEYSIZE_128:
+ ctx->key_type = OTX2_CPT_AES_128_BIT;
+ break;
+ case 2 * AES_KEYSIZE_192:
+ ctx->key_type = OTX2_CPT_AES_192_BIT;
+ break;
+ case 2 * AES_KEYSIZE_256:
+ ctx->key_type = OTX2_CPT_AES_256_BIT;
+ break;
+ default:
+ return -EINVAL;
+ }
+ return crypto_skcipher_setkey(ctx->fbk_cipher, key, keylen);
+}
+
+static int cpt_des_setkey(struct crypto_skcipher *tfm, const u8 *key,
+ u32 keylen, u8 cipher_type)
+{
+ struct otx2_cpt_enc_ctx *ctx = crypto_skcipher_ctx(tfm);
+
+ if (keylen != DES3_EDE_KEY_SIZE)
+ return -EINVAL;
+
+ ctx->key_len = keylen;
+ ctx->cipher_type = cipher_type;
+ ctx->enc_align_len = 8;
+
+ memcpy(ctx->enc_key, key, keylen);
+
+ return crypto_skcipher_setkey(ctx->fbk_cipher, key, keylen);
+}
+
+static int cpt_aes_setkey(struct crypto_skcipher *tfm, const u8 *key,
+ u32 keylen, u8 cipher_type)
+{
+ struct otx2_cpt_enc_ctx *ctx = crypto_skcipher_ctx(tfm);
+
+ switch (keylen) {
+ case AES_KEYSIZE_128:
+ ctx->key_type = OTX2_CPT_AES_128_BIT;
+ break;
+ case AES_KEYSIZE_192:
+ ctx->key_type = OTX2_CPT_AES_192_BIT;
+ break;
+ case AES_KEYSIZE_256:
+ ctx->key_type = OTX2_CPT_AES_256_BIT;
+ break;
+ default:
+ return -EINVAL;
+ }
+ if (cipher_type == OTX2_CPT_AES_CBC || cipher_type == OTX2_CPT_AES_ECB)
+ ctx->enc_align_len = 16;
+ else
+ ctx->enc_align_len = 1;
+
+ ctx->key_len = keylen;
+ ctx->cipher_type = cipher_type;
+
+ memcpy(ctx->enc_key, key, keylen);
+
+ return crypto_skcipher_setkey(ctx->fbk_cipher, key, keylen);
+}
+
+static int otx2_cpt_skcipher_cbc_aes_setkey(struct crypto_skcipher *tfm,
+ const u8 *key, u32 keylen)
+{
+ return cpt_aes_setkey(tfm, key, keylen, OTX2_CPT_AES_CBC);
+}
+
+static int otx2_cpt_skcipher_ecb_aes_setkey(struct crypto_skcipher *tfm,
+ const u8 *key, u32 keylen)
+{
+ return cpt_aes_setkey(tfm, key, keylen, OTX2_CPT_AES_ECB);
+}
+
+static int otx2_cpt_skcipher_cbc_des3_setkey(struct crypto_skcipher *tfm,
+ const u8 *key, u32 keylen)
+{
+ return cpt_des_setkey(tfm, key, keylen, OTX2_CPT_DES3_CBC);
+}
+
+static int otx2_cpt_skcipher_ecb_des3_setkey(struct crypto_skcipher *tfm,
+ const u8 *key, u32 keylen)
+{
+ return cpt_des_setkey(tfm, key, keylen, OTX2_CPT_DES3_ECB);
+}
+
+static int cpt_skcipher_fallback_init(struct otx2_cpt_enc_ctx *ctx,
+ struct crypto_alg *alg)
+{
+ if (alg->cra_flags & CRYPTO_ALG_NEED_FALLBACK) {
+ ctx->fbk_cipher =
+ crypto_alloc_skcipher(alg->cra_name, 0,
+ CRYPTO_ALG_ASYNC |
+ CRYPTO_ALG_NEED_FALLBACK);
+ if (IS_ERR(ctx->fbk_cipher)) {
+ pr_err("%s() failed to allocate fallback for %s\n",
+ __func__, alg->cra_name);
+ return PTR_ERR(ctx->fbk_cipher);
+ }
+ }
+ return 0;
+}
+
+static int otx2_cpt_enc_dec_init(struct crypto_skcipher *stfm)
+{
+ struct otx2_cpt_enc_ctx *ctx = crypto_skcipher_ctx(stfm);
+ struct crypto_tfm *tfm = crypto_skcipher_tfm(stfm);
+ struct crypto_alg *alg = tfm->__crt_alg;
+
+ memset(ctx, 0, sizeof(*ctx));
+ /*
+ * Additional memory for skcipher_request is
+ * allocated since the cryptd daemon uses
+ * this memory for request_ctx information
+ */
+ crypto_skcipher_set_reqsize(stfm, sizeof(struct otx2_cpt_req_ctx) +
+ sizeof(struct skcipher_request));
+
+ return cpt_skcipher_fallback_init(ctx, alg);
+}
+
+static void otx2_cpt_skcipher_exit(struct crypto_skcipher *tfm)
+{
+ struct otx2_cpt_enc_ctx *ctx = crypto_skcipher_ctx(tfm);
+
+ if (ctx->fbk_cipher) {
+ crypto_free_skcipher(ctx->fbk_cipher);
+ ctx->fbk_cipher = NULL;
+ }
+}
+
+static int cpt_aead_fallback_init(struct otx2_cpt_aead_ctx *ctx,
+ struct crypto_alg *alg)
+{
+ if (alg->cra_flags & CRYPTO_ALG_NEED_FALLBACK) {
+ ctx->fbk_cipher =
+ crypto_alloc_aead(alg->cra_name, 0,
+ CRYPTO_ALG_ASYNC |
+ CRYPTO_ALG_NEED_FALLBACK);
+ if (IS_ERR(ctx->fbk_cipher)) {
+ pr_err("%s() failed to allocate fallback for %s\n",
+ __func__, alg->cra_name);
+ return PTR_ERR(ctx->fbk_cipher);
+ }
+ }
+ return 0;
+}
+
+static int cpt_aead_init(struct crypto_aead *atfm, u8 cipher_type, u8 mac_type)
+{
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(atfm);
+ struct crypto_tfm *tfm = crypto_aead_tfm(atfm);
+ struct crypto_alg *alg = tfm->__crt_alg;
+
+ ctx->cipher_type = cipher_type;
+ ctx->mac_type = mac_type;
+
+ /*
+ * When selected cipher is NULL we use HMAC opcode instead of
+ * FLEXICRYPTO opcode therefore we don't need to use HASH algorithms
+ * for calculating ipad and opad
+ */
+ if (ctx->cipher_type != OTX2_CPT_CIPHER_NULL) {
+ switch (ctx->mac_type) {
+ case OTX2_CPT_SHA1:
+ ctx->hashalg = crypto_alloc_shash("sha1", 0,
+ CRYPTO_ALG_ASYNC);
+ if (IS_ERR(ctx->hashalg))
+ return PTR_ERR(ctx->hashalg);
+ break;
+
+ case OTX2_CPT_SHA256:
+ ctx->hashalg = crypto_alloc_shash("sha256", 0,
+ CRYPTO_ALG_ASYNC);
+ if (IS_ERR(ctx->hashalg))
+ return PTR_ERR(ctx->hashalg);
+ break;
+
+ case OTX2_CPT_SHA384:
+ ctx->hashalg = crypto_alloc_shash("sha384", 0,
+ CRYPTO_ALG_ASYNC);
+ if (IS_ERR(ctx->hashalg))
+ return PTR_ERR(ctx->hashalg);
+ break;
+
+ case OTX2_CPT_SHA512:
+ ctx->hashalg = crypto_alloc_shash("sha512", 0,
+ CRYPTO_ALG_ASYNC);
+ if (IS_ERR(ctx->hashalg))
+ return PTR_ERR(ctx->hashalg);
+ break;
+ }
+ }
+ switch (ctx->cipher_type) {
+ case OTX2_CPT_AES_CBC:
+ case OTX2_CPT_AES_ECB:
+ ctx->enc_align_len = 16;
+ break;
+ case OTX2_CPT_DES3_CBC:
+ case OTX2_CPT_DES3_ECB:
+ ctx->enc_align_len = 8;
+ break;
+ case OTX2_CPT_AES_GCM:
+ case OTX2_CPT_CIPHER_NULL:
+ ctx->enc_align_len = 1;
+ break;
+ }
+ crypto_aead_set_reqsize(atfm, sizeof(struct otx2_cpt_req_ctx));
+
+ return cpt_aead_fallback_init(ctx, alg);
+}
+
+static int otx2_cpt_aead_cbc_aes_sha1_init(struct crypto_aead *tfm)
+{
+ return cpt_aead_init(tfm, OTX2_CPT_AES_CBC, OTX2_CPT_SHA1);
+}
+
+static int otx2_cpt_aead_cbc_aes_sha256_init(struct crypto_aead *tfm)
+{
+ return cpt_aead_init(tfm, OTX2_CPT_AES_CBC, OTX2_CPT_SHA256);
+}
+
+static int otx2_cpt_aead_cbc_aes_sha384_init(struct crypto_aead *tfm)
+{
+ return cpt_aead_init(tfm, OTX2_CPT_AES_CBC, OTX2_CPT_SHA384);
+}
+
+static int otx2_cpt_aead_cbc_aes_sha512_init(struct crypto_aead *tfm)
+{
+ return cpt_aead_init(tfm, OTX2_CPT_AES_CBC, OTX2_CPT_SHA512);
+}
+
+static int otx2_cpt_aead_ecb_null_sha1_init(struct crypto_aead *tfm)
+{
+ return cpt_aead_init(tfm, OTX2_CPT_CIPHER_NULL, OTX2_CPT_SHA1);
+}
+
+static int otx2_cpt_aead_ecb_null_sha256_init(struct crypto_aead *tfm)
+{
+ return cpt_aead_init(tfm, OTX2_CPT_CIPHER_NULL, OTX2_CPT_SHA256);
+}
+
+static int otx2_cpt_aead_ecb_null_sha384_init(struct crypto_aead *tfm)
+{
+ return cpt_aead_init(tfm, OTX2_CPT_CIPHER_NULL, OTX2_CPT_SHA384);
+}
+
+static int otx2_cpt_aead_ecb_null_sha512_init(struct crypto_aead *tfm)
+{
+ return cpt_aead_init(tfm, OTX2_CPT_CIPHER_NULL, OTX2_CPT_SHA512);
+}
+
+static int otx2_cpt_aead_gcm_aes_init(struct crypto_aead *tfm)
+{
+ return cpt_aead_init(tfm, OTX2_CPT_AES_GCM, OTX2_CPT_MAC_NULL);
+}
+
+static void otx2_cpt_aead_exit(struct crypto_aead *tfm)
+{
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(tfm);
+
+ kfree(ctx->ipad);
+ kfree(ctx->opad);
+ if (ctx->hashalg)
+ crypto_free_shash(ctx->hashalg);
+ kfree(ctx->sdesc);
+
+ if (ctx->fbk_cipher) {
+ crypto_free_aead(ctx->fbk_cipher);
+ ctx->fbk_cipher = NULL;
+ }
+}
+
+static int otx2_cpt_aead_gcm_set_authsize(struct crypto_aead *tfm,
+ unsigned int authsize)
+{
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(tfm);
+
+ if (crypto_rfc4106_check_authsize(authsize))
+ return -EINVAL;
+
+ tfm->authsize = authsize;
+ /* Set authsize for fallback case */
+ if (ctx->fbk_cipher)
+ ctx->fbk_cipher->authsize = authsize;
+
+ return 0;
+}
+
+static int otx2_cpt_aead_set_authsize(struct crypto_aead *tfm,
+ unsigned int authsize)
+{
+ tfm->authsize = authsize;
+
+ return 0;
+}
+
+static int otx2_cpt_aead_null_set_authsize(struct crypto_aead *tfm,
+ unsigned int authsize)
+{
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(tfm);
+
+ ctx->is_trunc_hmac = true;
+ tfm->authsize = authsize;
+
+ return 0;
+}
+
+static struct otx2_cpt_sdesc *alloc_sdesc(struct crypto_shash *alg)
+{
+ struct otx2_cpt_sdesc *sdesc;
+ int size;
+
+ size = sizeof(struct shash_desc) + crypto_shash_descsize(alg);
+ sdesc = kmalloc(size, GFP_KERNEL);
+ if (!sdesc)
+ return NULL;
+
+ sdesc->shash.tfm = alg;
+
+ return sdesc;
+}
+
+static inline void swap_data32(void *buf, u32 len)
+{
+ cpu_to_be32_array(buf, buf, len / 4);
+}
+
+static inline void swap_data64(void *buf, u32 len)
+{
+ u64 *src = buf;
+ int i = 0;
+
+ for (i = 0 ; i < len / 8; i++, src++)
+ cpu_to_be64s(src);
+}
+
+static int copy_pad(u8 mac_type, u8 *out_pad, u8 *in_pad)
+{
+ struct sha512_state *sha512;
+ struct sha256_state *sha256;
+ struct sha1_state *sha1;
+
+ switch (mac_type) {
+ case OTX2_CPT_SHA1:
+ sha1 = (struct sha1_state *) in_pad;
+ swap_data32(sha1->state, SHA1_DIGEST_SIZE);
+ memcpy(out_pad, &sha1->state, SHA1_DIGEST_SIZE);
+ break;
+
+ case OTX2_CPT_SHA256:
+ sha256 = (struct sha256_state *) in_pad;
+ swap_data32(sha256->state, SHA256_DIGEST_SIZE);
+ memcpy(out_pad, &sha256->state, SHA256_DIGEST_SIZE);
+ break;
+
+ case OTX2_CPT_SHA384:
+ case OTX2_CPT_SHA512:
+ sha512 = (struct sha512_state *) in_pad;
+ swap_data64(sha512->state, SHA512_DIGEST_SIZE);
+ memcpy(out_pad, &sha512->state, SHA512_DIGEST_SIZE);
+ break;
+
+ default:
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static int aead_hmac_init(struct crypto_aead *cipher)
+{
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(cipher);
+ int state_size = crypto_shash_statesize(ctx->hashalg);
+ int ds = crypto_shash_digestsize(ctx->hashalg);
+ int bs = crypto_shash_blocksize(ctx->hashalg);
+ int authkeylen = ctx->auth_key_len;
+ u8 *ipad = NULL, *opad = NULL;
+ int ret = 0, icount = 0;
+
+ ctx->sdesc = alloc_sdesc(ctx->hashalg);
+ if (!ctx->sdesc)
+ return -ENOMEM;
+
+ ctx->ipad = kzalloc(bs, GFP_KERNEL);
+ if (!ctx->ipad) {
+ ret = -ENOMEM;
+ goto calc_fail;
+ }
+
+ ctx->opad = kzalloc(bs, GFP_KERNEL);
+ if (!ctx->opad) {
+ ret = -ENOMEM;
+ goto calc_fail;
+ }
+
+ ipad = kzalloc(state_size, GFP_KERNEL);
+ if (!ipad) {
+ ret = -ENOMEM;
+ goto calc_fail;
+ }
+
+ opad = kzalloc(state_size, GFP_KERNEL);
+ if (!opad) {
+ ret = -ENOMEM;
+ goto calc_fail;
+ }
+
+ if (authkeylen > bs) {
+ ret = crypto_shash_digest(&ctx->sdesc->shash, ctx->key,
+ authkeylen, ipad);
+ if (ret)
+ goto calc_fail;
+
+ authkeylen = ds;
+ } else {
+ memcpy(ipad, ctx->key, authkeylen);
+ }
+
+ memset(ipad + authkeylen, 0, bs - authkeylen);
+ memcpy(opad, ipad, bs);
+
+ for (icount = 0; icount < bs; icount++) {
+ ipad[icount] ^= 0x36;
+ opad[icount] ^= 0x5c;
+ }
+
+ /*
+ * Partial Hash calculated from the software
+ * algorithm is retrieved for IPAD & OPAD
+ */
+
+ /* IPAD Calculation */
+ crypto_shash_init(&ctx->sdesc->shash);
+ crypto_shash_update(&ctx->sdesc->shash, ipad, bs);
+ crypto_shash_export(&ctx->sdesc->shash, ipad);
+ ret = copy_pad(ctx->mac_type, ctx->ipad, ipad);
+ if (ret)
+ goto calc_fail;
+
+ /* OPAD Calculation */
+ crypto_shash_init(&ctx->sdesc->shash);
+ crypto_shash_update(&ctx->sdesc->shash, opad, bs);
+ crypto_shash_export(&ctx->sdesc->shash, opad);
+ ret = copy_pad(ctx->mac_type, ctx->opad, opad);
+ if (ret)
+ goto calc_fail;
+
+ kfree(ipad);
+ kfree(opad);
+
+ return 0;
+
+calc_fail:
+ kfree(ctx->ipad);
+ ctx->ipad = NULL;
+ kfree(ctx->opad);
+ ctx->opad = NULL;
+ kfree(ipad);
+ kfree(opad);
+ kfree(ctx->sdesc);
+ ctx->sdesc = NULL;
+
+ return ret;
+}
+
+static int otx2_cpt_aead_cbc_aes_sha_setkey(struct crypto_aead *cipher,
+ const unsigned char *key,
+ unsigned int keylen)
+{
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(cipher);
+ struct crypto_authenc_key_param *param;
+ int enckeylen = 0, authkeylen = 0;
+ struct rtattr *rta = (void *)key;
+ int status;
+
+ if (!RTA_OK(rta, keylen))
+ return -EINVAL;
+
+ if (rta->rta_type != CRYPTO_AUTHENC_KEYA_PARAM)
+ return -EINVAL;
+
+ if (RTA_PAYLOAD(rta) < sizeof(*param))
+ return -EINVAL;
+
+ param = RTA_DATA(rta);
+ enckeylen = be32_to_cpu(param->enckeylen);
+ key += RTA_ALIGN(rta->rta_len);
+ keylen -= RTA_ALIGN(rta->rta_len);
+ if (keylen < enckeylen)
+ return -EINVAL;
+
+ if (keylen > OTX2_CPT_MAX_KEY_SIZE)
+ return -EINVAL;
+
+ authkeylen = keylen - enckeylen;
+ memcpy(ctx->key, key, keylen);
+
+ switch (enckeylen) {
+ case AES_KEYSIZE_128:
+ ctx->key_type = OTX2_CPT_AES_128_BIT;
+ break;
+ case AES_KEYSIZE_192:
+ ctx->key_type = OTX2_CPT_AES_192_BIT;
+ break;
+ case AES_KEYSIZE_256:
+ ctx->key_type = OTX2_CPT_AES_256_BIT;
+ break;
+ default:
+ /* Invalid key length */
+ return -EINVAL;
+ }
+
+ ctx->enc_key_len = enckeylen;
+ ctx->auth_key_len = authkeylen;
+
+ status = aead_hmac_init(cipher);
+ if (status)
+ return status;
+
+ return 0;
+}
+
+static int otx2_cpt_aead_ecb_null_sha_setkey(struct crypto_aead *cipher,
+ const unsigned char *key,
+ unsigned int keylen)
+{
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(cipher);
+ struct crypto_authenc_key_param *param;
+ struct rtattr *rta = (void *)key;
+ int enckeylen = 0;
+
+ if (!RTA_OK(rta, keylen))
+ return -EINVAL;
+
+ if (rta->rta_type != CRYPTO_AUTHENC_KEYA_PARAM)
+ return -EINVAL;
+
+ if (RTA_PAYLOAD(rta) < sizeof(*param))
+ return -EINVAL;
+
+ param = RTA_DATA(rta);
+ enckeylen = be32_to_cpu(param->enckeylen);
+ key += RTA_ALIGN(rta->rta_len);
+ keylen -= RTA_ALIGN(rta->rta_len);
+ if (enckeylen != 0)
+ return -EINVAL;
+
+ if (keylen > OTX2_CPT_MAX_KEY_SIZE)
+ return -EINVAL;
+
+ memcpy(ctx->key, key, keylen);
+ ctx->enc_key_len = enckeylen;
+ ctx->auth_key_len = keylen;
+
+ return 0;
+}
+
+static int otx2_cpt_aead_gcm_aes_setkey(struct crypto_aead *cipher,
+ const unsigned char *key,
+ unsigned int keylen)
+{
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(cipher);
+
+ /*
+ * For aes gcm we expect to get encryption key (16, 24, 32 bytes)
+ * and salt (4 bytes)
+ */
+ switch (keylen) {
+ case AES_KEYSIZE_128 + AES_GCM_SALT_SIZE:
+ ctx->key_type = OTX2_CPT_AES_128_BIT;
+ ctx->enc_key_len = AES_KEYSIZE_128;
+ break;
+ case AES_KEYSIZE_192 + AES_GCM_SALT_SIZE:
+ ctx->key_type = OTX2_CPT_AES_192_BIT;
+ ctx->enc_key_len = AES_KEYSIZE_192;
+ break;
+ case AES_KEYSIZE_256 + AES_GCM_SALT_SIZE:
+ ctx->key_type = OTX2_CPT_AES_256_BIT;
+ ctx->enc_key_len = AES_KEYSIZE_256;
+ break;
+ default:
+ /* Invalid key and salt length */
+ return -EINVAL;
+ }
+
+ /* Store encryption key and salt */
+ memcpy(ctx->key, key, keylen);
+
+ return crypto_aead_setkey(ctx->fbk_cipher, key, keylen);
+}
+
+static inline int create_aead_ctx_hdr(struct aead_request *req, u32 enc,
+ u32 *argcnt)
+{
+ struct otx2_cpt_req_ctx *rctx = aead_request_ctx(req);
+ struct crypto_aead *tfm = crypto_aead_reqtfm(req);
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(tfm);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ struct otx2_cpt_fc_ctx *fctx = &rctx->fctx;
+ int mac_len = crypto_aead_authsize(tfm);
+ int ds;
+
+ rctx->ctrl_word.e.enc_data_offset = req->assoclen;
+
+ switch (ctx->cipher_type) {
+ case OTX2_CPT_AES_CBC:
+ if (req->assoclen > 248 || !IS_ALIGNED(req->assoclen, 8))
+ return -EINVAL;
+
+ fctx->enc.enc_ctrl.e.iv_source = OTX2_CPT_FROM_CPTR;
+ /* Copy encryption key to context */
+ memcpy(fctx->enc.encr_key, ctx->key + ctx->auth_key_len,
+ ctx->enc_key_len);
+ /* Copy IV to context */
+ memcpy(fctx->enc.encr_iv, req->iv, crypto_aead_ivsize(tfm));
+
+ ds = crypto_shash_digestsize(ctx->hashalg);
+ if (ctx->mac_type == OTX2_CPT_SHA384)
+ ds = SHA512_DIGEST_SIZE;
+ if (ctx->ipad)
+ memcpy(fctx->hmac.e.ipad, ctx->ipad, ds);
+ if (ctx->opad)
+ memcpy(fctx->hmac.e.opad, ctx->opad, ds);
+ break;
+
+ case OTX2_CPT_AES_GCM:
+ if (crypto_ipsec_check_assoclen(req->assoclen))
+ return -EINVAL;
+
+ fctx->enc.enc_ctrl.e.iv_source = OTX2_CPT_FROM_DPTR;
+ /* Copy encryption key to context */
+ memcpy(fctx->enc.encr_key, ctx->key, ctx->enc_key_len);
+ /* Copy salt to context */
+ memcpy(fctx->enc.encr_iv, ctx->key + ctx->enc_key_len,
+ AES_GCM_SALT_SIZE);
+
+ rctx->ctrl_word.e.iv_offset = req->assoclen - AES_GCM_IV_OFFSET;
+ break;
+
+ default:
+ /* Unknown cipher type */
+ return -EINVAL;
+ }
+ cpu_to_be64s(&rctx->ctrl_word.flags);
+
+ req_info->ctrl.s.dma_mode = OTX2_CPT_DMA_MODE_SG;
+ req_info->ctrl.s.se_req = 1;
+ req_info->req.opcode.s.major = OTX2_CPT_MAJOR_OP_FC |
+ DMA_MODE_FLAG(OTX2_CPT_DMA_MODE_SG);
+ if (enc) {
+ req_info->req.opcode.s.minor = 2;
+ req_info->req.param1 = req->cryptlen;
+ req_info->req.param2 = req->cryptlen + req->assoclen;
+ } else {
+ req_info->req.opcode.s.minor = 3;
+ req_info->req.param1 = req->cryptlen - mac_len;
+ req_info->req.param2 = req->cryptlen + req->assoclen - mac_len;
+ }
+
+ fctx->enc.enc_ctrl.e.enc_cipher = ctx->cipher_type;
+ fctx->enc.enc_ctrl.e.aes_key = ctx->key_type;
+ fctx->enc.enc_ctrl.e.mac_type = ctx->mac_type;
+ fctx->enc.enc_ctrl.e.mac_len = mac_len;
+ cpu_to_be64s(&fctx->enc.enc_ctrl.u);
+
+ /*
+ * Storing Packet Data Information in offset
+ * Control Word First 8 bytes
+ */
+ req_info->in[*argcnt].vptr = (u8 *)&rctx->ctrl_word;
+ req_info->in[*argcnt].size = CONTROL_WORD_LEN;
+ req_info->req.dlen += CONTROL_WORD_LEN;
+ ++(*argcnt);
+
+ req_info->in[*argcnt].vptr = (u8 *)fctx;
+ req_info->in[*argcnt].size = sizeof(struct otx2_cpt_fc_ctx);
+ req_info->req.dlen += sizeof(struct otx2_cpt_fc_ctx);
+ ++(*argcnt);
+
+ return 0;
+}
+
+static inline void create_hmac_ctx_hdr(struct aead_request *req, u32 *argcnt,
+ u32 enc)
+{
+ struct otx2_cpt_req_ctx *rctx = aead_request_ctx(req);
+ struct crypto_aead *tfm = crypto_aead_reqtfm(req);
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(tfm);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+
+ req_info->ctrl.s.dma_mode = OTX2_CPT_DMA_MODE_SG;
+ req_info->ctrl.s.se_req = 1;
+ req_info->req.opcode.s.major = OTX2_CPT_MAJOR_OP_HMAC |
+ DMA_MODE_FLAG(OTX2_CPT_DMA_MODE_SG);
+ req_info->is_trunc_hmac = ctx->is_trunc_hmac;
+
+ req_info->req.opcode.s.minor = 0;
+ req_info->req.param1 = ctx->auth_key_len;
+ req_info->req.param2 = ctx->mac_type << 8;
+
+ /* Add authentication key */
+ req_info->in[*argcnt].vptr = ctx->key;
+ req_info->in[*argcnt].size = round_up(ctx->auth_key_len, 8);
+ req_info->req.dlen += round_up(ctx->auth_key_len, 8);
+ ++(*argcnt);
+}
+
+static inline int create_aead_input_list(struct aead_request *req, u32 enc)
+{
+ struct otx2_cpt_req_ctx *rctx = aead_request_ctx(req);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ u32 inputlen = req->cryptlen + req->assoclen;
+ u32 status, argcnt = 0;
+
+ status = create_aead_ctx_hdr(req, enc, &argcnt);
+ if (status)
+ return status;
+ update_input_data(req_info, req->src, inputlen, &argcnt);
+ req_info->in_cnt = argcnt;
+
+ return 0;
+}
+
+static inline void create_aead_output_list(struct aead_request *req, u32 enc,
+ u32 mac_len)
+{
+ struct otx2_cpt_req_ctx *rctx = aead_request_ctx(req);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ u32 argcnt = 0, outputlen = 0;
+
+ if (enc)
+ outputlen = req->cryptlen + req->assoclen + mac_len;
+ else
+ outputlen = req->cryptlen + req->assoclen - mac_len;
+
+ update_output_data(req_info, req->dst, 0, outputlen, &argcnt);
+ req_info->out_cnt = argcnt;
+}
+
+static inline void create_aead_null_input_list(struct aead_request *req,
+ u32 enc, u32 mac_len)
+{
+ struct otx2_cpt_req_ctx *rctx = aead_request_ctx(req);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ u32 inputlen, argcnt = 0;
+
+ if (enc)
+ inputlen = req->cryptlen + req->assoclen;
+ else
+ inputlen = req->cryptlen + req->assoclen - mac_len;
+
+ create_hmac_ctx_hdr(req, &argcnt, enc);
+ update_input_data(req_info, req->src, inputlen, &argcnt);
+ req_info->in_cnt = argcnt;
+}
+
+static inline int create_aead_null_output_list(struct aead_request *req,
+ u32 enc, u32 mac_len)
+{
+ struct otx2_cpt_req_ctx *rctx = aead_request_ctx(req);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ struct scatterlist *dst;
+ u8 *ptr = NULL;
+ int argcnt = 0, status, offset;
+ u32 inputlen;
+
+ if (enc)
+ inputlen = req->cryptlen + req->assoclen;
+ else
+ inputlen = req->cryptlen + req->assoclen - mac_len;
+
+ /*
+ * If source and destination are different
+ * then copy payload to destination
+ */
+ if (req->src != req->dst) {
+
+ ptr = kmalloc(inputlen, (req_info->areq->flags &
+ CRYPTO_TFM_REQ_MAY_SLEEP) ?
+ GFP_KERNEL : GFP_ATOMIC);
+ if (!ptr)
+ return -ENOMEM;
+
+ status = sg_copy_to_buffer(req->src, sg_nents(req->src), ptr,
+ inputlen);
+ if (status != inputlen) {
+ status = -EINVAL;
+ goto error_free;
+ }
+ status = sg_copy_from_buffer(req->dst, sg_nents(req->dst), ptr,
+ inputlen);
+ if (status != inputlen) {
+ status = -EINVAL;
+ goto error_free;
+ }
+ kfree(ptr);
+ }
+
+ if (enc) {
+ /*
+ * In an encryption scenario hmac needs
+ * to be appended after payload
+ */
+ dst = req->dst;
+ offset = inputlen;
+ while (offset >= dst->length) {
+ offset -= dst->length;
+ dst = sg_next(dst);
+ if (!dst)
+ return -ENOENT;
+ }
+
+ update_output_data(req_info, dst, offset, mac_len, &argcnt);
+ } else {
+ /*
+ * In a decryption scenario calculated hmac for received
+ * payload needs to be compare with hmac received
+ */
+ status = sg_copy_buffer(req->src, sg_nents(req->src),
+ rctx->fctx.hmac.s.hmac_recv, mac_len,
+ inputlen, true);
+ if (status != mac_len)
+ return -EINVAL;
+
+ req_info->out[argcnt].vptr = rctx->fctx.hmac.s.hmac_calc;
+ req_info->out[argcnt].size = mac_len;
+ argcnt++;
+ }
+
+ req_info->out_cnt = argcnt;
+ return 0;
+
+error_free:
+ kfree(ptr);
+ return status;
+}
+
+static int aead_do_fallback(struct aead_request *req, bool is_enc)
+{
+ struct otx2_cpt_req_ctx *rctx = aead_request_ctx(req);
+ struct crypto_aead *aead = crypto_aead_reqtfm(req);
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(aead);
+ int ret;
+
+ if (ctx->fbk_cipher) {
+ /* Store the cipher tfm and then use the fallback tfm */
+ aead_request_set_tfm(&rctx->fbk_req, ctx->fbk_cipher);
+ aead_request_set_callback(&rctx->fbk_req, req->base.flags,
+ req->base.complete, req->base.data);
+ aead_request_set_crypt(&rctx->fbk_req, req->src,
+ req->dst, req->cryptlen, req->iv);
+ ret = is_enc ? crypto_aead_encrypt(&rctx->fbk_req) :
+ crypto_aead_decrypt(&rctx->fbk_req);
+ } else {
+ ret = -EINVAL;
+ }
+
+ return ret;
+}
+
+static int cpt_aead_enc_dec(struct aead_request *req, u8 reg_type, u8 enc)
+{
+ struct otx2_cpt_req_ctx *rctx = aead_request_ctx(req);
+ struct otx2_cpt_req_info *req_info = &rctx->cpt_req;
+ struct crypto_aead *tfm = crypto_aead_reqtfm(req);
+ struct otx2_cpt_aead_ctx *ctx = crypto_aead_ctx(tfm);
+ struct pci_dev *pdev;
+ int status, cpu_num;
+
+ /* Clear control words */
+ rctx->ctrl_word.flags = 0;
+ rctx->fctx.enc.enc_ctrl.u = 0;
+
+ req_info->callback = otx2_cpt_aead_callback;
+ req_info->areq = &req->base;
+ req_info->req_type = reg_type;
+ req_info->is_enc = enc;
+ req_info->is_trunc_hmac = false;
+
+ switch (reg_type) {
+ case OTX2_CPT_AEAD_ENC_DEC_REQ:
+ status = create_aead_input_list(req, enc);
+ if (status)
+ return status;
+ create_aead_output_list(req, enc, crypto_aead_authsize(tfm));
+ break;
+
+ case OTX2_CPT_AEAD_ENC_DEC_NULL_REQ:
+ create_aead_null_input_list(req, enc,
+ crypto_aead_authsize(tfm));
+ status = create_aead_null_output_list(req, enc,
+ crypto_aead_authsize(tfm));
+ if (status)
+ return status;
+ break;
+
+ default:
+ return -EINVAL;
+ }
+ if (!IS_ALIGNED(req_info->req.param1, ctx->enc_align_len))
+ return -EINVAL;
+
+ if (!req_info->req.param2 ||
+ (req_info->req.param1 > OTX2_CPT_MAX_REQ_SIZE) ||
+ (req_info->req.param2 > OTX2_CPT_MAX_REQ_SIZE))
+ return aead_do_fallback(req, enc);
+
+ status = get_se_device(&pdev, &cpu_num);
+ if (status)
+ return status;
+
+ req_info->ctrl.s.grp = otx2_cpt_get_kcrypto_eng_grp_num(pdev);
+
+ /*
+ * We perform an asynchronous send and once
+ * the request is completed the driver would
+ * intimate through registered call back functions
+ */
+ return otx2_cpt_do_request(pdev, req_info, cpu_num);
+}
+
+static int otx2_cpt_aead_encrypt(struct aead_request *req)
+{
+ return cpt_aead_enc_dec(req, OTX2_CPT_AEAD_ENC_DEC_REQ, true);
+}
+
+static int otx2_cpt_aead_decrypt(struct aead_request *req)
+{
+ return cpt_aead_enc_dec(req, OTX2_CPT_AEAD_ENC_DEC_REQ, false);
+}
+
+static int otx2_cpt_aead_null_encrypt(struct aead_request *req)
+{
+ return cpt_aead_enc_dec(req, OTX2_CPT_AEAD_ENC_DEC_NULL_REQ, true);
+}
+
+static int otx2_cpt_aead_null_decrypt(struct aead_request *req)
+{
+ return cpt_aead_enc_dec(req, OTX2_CPT_AEAD_ENC_DEC_NULL_REQ, false);
+}
+
+static struct skcipher_alg otx2_cpt_skciphers[] = { {
+ .base.cra_name = "xts(aes)",
+ .base.cra_driver_name = "cpt_xts_aes",
+ .base.cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .base.cra_blocksize = AES_BLOCK_SIZE,
+ .base.cra_ctxsize = sizeof(struct otx2_cpt_enc_ctx),
+ .base.cra_alignmask = 7,
+ .base.cra_priority = 4001,
+ .base.cra_module = THIS_MODULE,
+
+ .init = otx2_cpt_enc_dec_init,
+ .exit = otx2_cpt_skcipher_exit,
+ .ivsize = AES_BLOCK_SIZE,
+ .min_keysize = 2 * AES_MIN_KEY_SIZE,
+ .max_keysize = 2 * AES_MAX_KEY_SIZE,
+ .setkey = otx2_cpt_skcipher_xts_setkey,
+ .encrypt = otx2_cpt_skcipher_encrypt,
+ .decrypt = otx2_cpt_skcipher_decrypt,
+}, {
+ .base.cra_name = "cbc(aes)",
+ .base.cra_driver_name = "cpt_cbc_aes",
+ .base.cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .base.cra_blocksize = AES_BLOCK_SIZE,
+ .base.cra_ctxsize = sizeof(struct otx2_cpt_enc_ctx),
+ .base.cra_alignmask = 7,
+ .base.cra_priority = 4001,
+ .base.cra_module = THIS_MODULE,
+
+ .init = otx2_cpt_enc_dec_init,
+ .exit = otx2_cpt_skcipher_exit,
+ .ivsize = AES_BLOCK_SIZE,
+ .min_keysize = AES_MIN_KEY_SIZE,
+ .max_keysize = AES_MAX_KEY_SIZE,
+ .setkey = otx2_cpt_skcipher_cbc_aes_setkey,
+ .encrypt = otx2_cpt_skcipher_encrypt,
+ .decrypt = otx2_cpt_skcipher_decrypt,
+}, {
+ .base.cra_name = "ecb(aes)",
+ .base.cra_driver_name = "cpt_ecb_aes",
+ .base.cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .base.cra_blocksize = AES_BLOCK_SIZE,
+ .base.cra_ctxsize = sizeof(struct otx2_cpt_enc_ctx),
+ .base.cra_alignmask = 7,
+ .base.cra_priority = 4001,
+ .base.cra_module = THIS_MODULE,
+
+ .init = otx2_cpt_enc_dec_init,
+ .exit = otx2_cpt_skcipher_exit,
+ .ivsize = 0,
+ .min_keysize = AES_MIN_KEY_SIZE,
+ .max_keysize = AES_MAX_KEY_SIZE,
+ .setkey = otx2_cpt_skcipher_ecb_aes_setkey,
+ .encrypt = otx2_cpt_skcipher_encrypt,
+ .decrypt = otx2_cpt_skcipher_decrypt,
+}, {
+ .base.cra_name = "cbc(des3_ede)",
+ .base.cra_driver_name = "cpt_cbc_des3_ede",
+ .base.cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .base.cra_blocksize = DES3_EDE_BLOCK_SIZE,
+ .base.cra_ctxsize = sizeof(struct otx2_cpt_enc_ctx),
+ .base.cra_alignmask = 7,
+ .base.cra_priority = 4001,
+ .base.cra_module = THIS_MODULE,
+
+ .init = otx2_cpt_enc_dec_init,
+ .exit = otx2_cpt_skcipher_exit,
+ .min_keysize = DES3_EDE_KEY_SIZE,
+ .max_keysize = DES3_EDE_KEY_SIZE,
+ .ivsize = DES_BLOCK_SIZE,
+ .setkey = otx2_cpt_skcipher_cbc_des3_setkey,
+ .encrypt = otx2_cpt_skcipher_encrypt,
+ .decrypt = otx2_cpt_skcipher_decrypt,
+}, {
+ .base.cra_name = "ecb(des3_ede)",
+ .base.cra_driver_name = "cpt_ecb_des3_ede",
+ .base.cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .base.cra_blocksize = DES3_EDE_BLOCK_SIZE,
+ .base.cra_ctxsize = sizeof(struct otx2_cpt_enc_ctx),
+ .base.cra_alignmask = 7,
+ .base.cra_priority = 4001,
+ .base.cra_module = THIS_MODULE,
+
+ .init = otx2_cpt_enc_dec_init,
+ .exit = otx2_cpt_skcipher_exit,
+ .min_keysize = DES3_EDE_KEY_SIZE,
+ .max_keysize = DES3_EDE_KEY_SIZE,
+ .ivsize = 0,
+ .setkey = otx2_cpt_skcipher_ecb_des3_setkey,
+ .encrypt = otx2_cpt_skcipher_encrypt,
+ .decrypt = otx2_cpt_skcipher_decrypt,
+} };
+
+static struct aead_alg otx2_cpt_aeads[] = { {
+ .base = {
+ .cra_name = "authenc(hmac(sha1),cbc(aes))",
+ .cra_driver_name = "cpt_hmac_sha1_cbc_aes",
+ .cra_blocksize = AES_BLOCK_SIZE,
+ .cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .cra_ctxsize = sizeof(struct otx2_cpt_aead_ctx),
+ .cra_priority = 4001,
+ .cra_alignmask = 0,
+ .cra_module = THIS_MODULE,
+ },
+ .init = otx2_cpt_aead_cbc_aes_sha1_init,
+ .exit = otx2_cpt_aead_exit,
+ .setkey = otx2_cpt_aead_cbc_aes_sha_setkey,
+ .setauthsize = otx2_cpt_aead_set_authsize,
+ .encrypt = otx2_cpt_aead_encrypt,
+ .decrypt = otx2_cpt_aead_decrypt,
+ .ivsize = AES_BLOCK_SIZE,
+ .maxauthsize = SHA1_DIGEST_SIZE,
+}, {
+ .base = {
+ .cra_name = "authenc(hmac(sha256),cbc(aes))",
+ .cra_driver_name = "cpt_hmac_sha256_cbc_aes",
+ .cra_blocksize = AES_BLOCK_SIZE,
+ .cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .cra_ctxsize = sizeof(struct otx2_cpt_aead_ctx),
+ .cra_priority = 4001,
+ .cra_alignmask = 0,
+ .cra_module = THIS_MODULE,
+ },
+ .init = otx2_cpt_aead_cbc_aes_sha256_init,
+ .exit = otx2_cpt_aead_exit,
+ .setkey = otx2_cpt_aead_cbc_aes_sha_setkey,
+ .setauthsize = otx2_cpt_aead_set_authsize,
+ .encrypt = otx2_cpt_aead_encrypt,
+ .decrypt = otx2_cpt_aead_decrypt,
+ .ivsize = AES_BLOCK_SIZE,
+ .maxauthsize = SHA256_DIGEST_SIZE,
+}, {
+ .base = {
+ .cra_name = "authenc(hmac(sha384),cbc(aes))",
+ .cra_driver_name = "cpt_hmac_sha384_cbc_aes",
+ .cra_blocksize = AES_BLOCK_SIZE,
+ .cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .cra_ctxsize = sizeof(struct otx2_cpt_aead_ctx),
+ .cra_priority = 4001,
+ .cra_alignmask = 0,
+ .cra_module = THIS_MODULE,
+ },
+ .init = otx2_cpt_aead_cbc_aes_sha384_init,
+ .exit = otx2_cpt_aead_exit,
+ .setkey = otx2_cpt_aead_cbc_aes_sha_setkey,
+ .setauthsize = otx2_cpt_aead_set_authsize,
+ .encrypt = otx2_cpt_aead_encrypt,
+ .decrypt = otx2_cpt_aead_decrypt,
+ .ivsize = AES_BLOCK_SIZE,
+ .maxauthsize = SHA384_DIGEST_SIZE,
+}, {
+ .base = {
+ .cra_name = "authenc(hmac(sha512),cbc(aes))",
+ .cra_driver_name = "cpt_hmac_sha512_cbc_aes",
+ .cra_blocksize = AES_BLOCK_SIZE,
+ .cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .cra_ctxsize = sizeof(struct otx2_cpt_aead_ctx),
+ .cra_priority = 4001,
+ .cra_alignmask = 0,
+ .cra_module = THIS_MODULE,
+ },
+ .init = otx2_cpt_aead_cbc_aes_sha512_init,
+ .exit = otx2_cpt_aead_exit,
+ .setkey = otx2_cpt_aead_cbc_aes_sha_setkey,
+ .setauthsize = otx2_cpt_aead_set_authsize,
+ .encrypt = otx2_cpt_aead_encrypt,
+ .decrypt = otx2_cpt_aead_decrypt,
+ .ivsize = AES_BLOCK_SIZE,
+ .maxauthsize = SHA512_DIGEST_SIZE,
+}, {
+ .base = {
+ .cra_name = "authenc(hmac(sha1),ecb(cipher_null))",
+ .cra_driver_name = "cpt_hmac_sha1_ecb_null",
+ .cra_blocksize = 1,
+ .cra_flags = CRYPTO_ALG_ASYNC,
+ .cra_ctxsize = sizeof(struct otx2_cpt_aead_ctx),
+ .cra_priority = 4001,
+ .cra_alignmask = 0,
+ .cra_module = THIS_MODULE,
+ },
+ .init = otx2_cpt_aead_ecb_null_sha1_init,
+ .exit = otx2_cpt_aead_exit,
+ .setkey = otx2_cpt_aead_ecb_null_sha_setkey,
+ .setauthsize = otx2_cpt_aead_null_set_authsize,
+ .encrypt = otx2_cpt_aead_null_encrypt,
+ .decrypt = otx2_cpt_aead_null_decrypt,
+ .ivsize = 0,
+ .maxauthsize = SHA1_DIGEST_SIZE,
+}, {
+ .base = {
+ .cra_name = "authenc(hmac(sha256),ecb(cipher_null))",
+ .cra_driver_name = "cpt_hmac_sha256_ecb_null",
+ .cra_blocksize = 1,
+ .cra_flags = CRYPTO_ALG_ASYNC,
+ .cra_ctxsize = sizeof(struct otx2_cpt_aead_ctx),
+ .cra_priority = 4001,
+ .cra_alignmask = 0,
+ .cra_module = THIS_MODULE,
+ },
+ .init = otx2_cpt_aead_ecb_null_sha256_init,
+ .exit = otx2_cpt_aead_exit,
+ .setkey = otx2_cpt_aead_ecb_null_sha_setkey,
+ .setauthsize = otx2_cpt_aead_null_set_authsize,
+ .encrypt = otx2_cpt_aead_null_encrypt,
+ .decrypt = otx2_cpt_aead_null_decrypt,
+ .ivsize = 0,
+ .maxauthsize = SHA256_DIGEST_SIZE,
+}, {
+ .base = {
+ .cra_name = "authenc(hmac(sha384),ecb(cipher_null))",
+ .cra_driver_name = "cpt_hmac_sha384_ecb_null",
+ .cra_blocksize = 1,
+ .cra_flags = CRYPTO_ALG_ASYNC,
+ .cra_ctxsize = sizeof(struct otx2_cpt_aead_ctx),
+ .cra_priority = 4001,
+ .cra_alignmask = 0,
+ .cra_module = THIS_MODULE,
+ },
+ .init = otx2_cpt_aead_ecb_null_sha384_init,
+ .exit = otx2_cpt_aead_exit,
+ .setkey = otx2_cpt_aead_ecb_null_sha_setkey,
+ .setauthsize = otx2_cpt_aead_null_set_authsize,
+ .encrypt = otx2_cpt_aead_null_encrypt,
+ .decrypt = otx2_cpt_aead_null_decrypt,
+ .ivsize = 0,
+ .maxauthsize = SHA384_DIGEST_SIZE,
+}, {
+ .base = {
+ .cra_name = "authenc(hmac(sha512),ecb(cipher_null))",
+ .cra_driver_name = "cpt_hmac_sha512_ecb_null",
+ .cra_blocksize = 1,
+ .cra_flags = CRYPTO_ALG_ASYNC,
+ .cra_ctxsize = sizeof(struct otx2_cpt_aead_ctx),
+ .cra_priority = 4001,
+ .cra_alignmask = 0,
+ .cra_module = THIS_MODULE,
+ },
+ .init = otx2_cpt_aead_ecb_null_sha512_init,
+ .exit = otx2_cpt_aead_exit,
+ .setkey = otx2_cpt_aead_ecb_null_sha_setkey,
+ .setauthsize = otx2_cpt_aead_null_set_authsize,
+ .encrypt = otx2_cpt_aead_null_encrypt,
+ .decrypt = otx2_cpt_aead_null_decrypt,
+ .ivsize = 0,
+ .maxauthsize = SHA512_DIGEST_SIZE,
+}, {
+ .base = {
+ .cra_name = "rfc4106(gcm(aes))",
+ .cra_driver_name = "cpt_rfc4106_gcm_aes",
+ .cra_blocksize = 1,
+ .cra_flags = CRYPTO_ALG_ASYNC | CRYPTO_ALG_NEED_FALLBACK,
+ .cra_ctxsize = sizeof(struct otx2_cpt_aead_ctx),
+ .cra_priority = 4001,
+ .cra_alignmask = 0,
+ .cra_module = THIS_MODULE,
+ },
+ .init = otx2_cpt_aead_gcm_aes_init,
+ .exit = otx2_cpt_aead_exit,
+ .setkey = otx2_cpt_aead_gcm_aes_setkey,
+ .setauthsize = otx2_cpt_aead_gcm_set_authsize,
+ .encrypt = otx2_cpt_aead_encrypt,
+ .decrypt = otx2_cpt_aead_decrypt,
+ .ivsize = AES_GCM_IV_SIZE,
+ .maxauthsize = AES_GCM_ICV_SIZE,
+} };
+
+static inline int cpt_register_algs(void)
+{
+ int i, err = 0;
+
+ if (!IS_ENABLED(CONFIG_DM_CRYPT)) {
+ for (i = 0; i < ARRAY_SIZE(otx2_cpt_skciphers); i++)
+ otx2_cpt_skciphers[i].base.cra_flags &=
+ ~CRYPTO_ALG_DEAD;
+
+ err = crypto_register_skciphers(otx2_cpt_skciphers,
+ ARRAY_SIZE(otx2_cpt_skciphers));
+ if (err)
+ return err;
+ }
+
+ for (i = 0; i < ARRAY_SIZE(otx2_cpt_aeads); i++)
+ otx2_cpt_aeads[i].base.cra_flags &= ~CRYPTO_ALG_DEAD;
+
+ err = crypto_register_aeads(otx2_cpt_aeads,
+ ARRAY_SIZE(otx2_cpt_aeads));
+ if (err) {
+ crypto_unregister_skciphers(otx2_cpt_skciphers,
+ ARRAY_SIZE(otx2_cpt_skciphers));
+ return err;
+ }
+
+ return 0;
+}
+
+static inline void cpt_unregister_algs(void)
+{
+ crypto_unregister_skciphers(otx2_cpt_skciphers,
+ ARRAY_SIZE(otx2_cpt_skciphers));
+ crypto_unregister_aeads(otx2_cpt_aeads, ARRAY_SIZE(otx2_cpt_aeads));
+}
+
+static int compare_func(const void *lptr, const void *rptr)
+{
+ const struct cpt_device_desc *ldesc = (struct cpt_device_desc *) lptr;
+ const struct cpt_device_desc *rdesc = (struct cpt_device_desc *) rptr;
+
+ if (ldesc->dev->devfn < rdesc->dev->devfn)
+ return -1;
+ if (ldesc->dev->devfn > rdesc->dev->devfn)
+ return 1;
+ return 0;
+}
+
+static void swap_func(void *lptr, void *rptr, int size)
+{
+ struct cpt_device_desc *ldesc = lptr;
+ struct cpt_device_desc *rdesc = rptr;
+ struct cpt_device_desc desc;
+
+ desc = *ldesc;
+ *ldesc = *rdesc;
+ *rdesc = desc;
+}
+
+int otx2_cpt_crypto_init(struct pci_dev *pdev, struct module *mod,
+ int num_queues, int num_devices)
+{
+ int ret = 0;
+ int count;
+
+ mutex_lock(&mutex);
+ count = atomic_read(&se_devices.count);
+ if (count >= OTX2_CPT_MAX_LFS_NUM) {
+ dev_err(&pdev->dev, "No space to add a new device\n");
+ ret = -ENOSPC;
+ goto unlock;
+ }
+ se_devices.desc[count].num_queues = num_queues;
+ se_devices.desc[count++].dev = pdev;
+ atomic_inc(&se_devices.count);
+
+ if (atomic_read(&se_devices.count) == num_devices &&
+ is_crypto_registered == false) {
+ if (cpt_register_algs()) {
+ dev_err(&pdev->dev,
+ "Error in registering crypto algorithms\n");
+ ret = -EINVAL;
+ goto unlock;
+ }
+ try_module_get(mod);
+ is_crypto_registered = true;
+ }
+ sort(se_devices.desc, count, sizeof(struct cpt_device_desc),
+ compare_func, swap_func);
+
+unlock:
+ mutex_unlock(&mutex);
+ return ret;
+}
+
+void otx2_cpt_crypto_exit(struct pci_dev *pdev, struct module *mod)
+{
+ struct cpt_device_table *dev_tbl;
+ bool dev_found = false;
+ int i, j, count;
+
+ mutex_lock(&mutex);
+
+ dev_tbl = &se_devices;
+ count = atomic_read(&dev_tbl->count);
+ for (i = 0; i < count; i++) {
+ if (pdev == dev_tbl->desc[i].dev) {
+ for (j = i; j < count-1; j++)
+ dev_tbl->desc[j] = dev_tbl->desc[j+1];
+ dev_found = true;
+ break;
+ }
+ }
+
+ if (!dev_found) {
+ dev_err(&pdev->dev, "%s device not found\n", __func__);
+ goto unlock;
+ }
+ if (atomic_dec_and_test(&se_devices.count)) {
+ cpt_unregister_algs();
+ module_put(mod);
+ is_crypto_registered = false;
+ }
+
+unlock:
+ mutex_unlock(&mutex);
+}
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.h b/drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.h
new file mode 100644
index 000000000000..f04184bd1744
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.h
@@ -0,0 +1,178 @@
+/* SPDX-License-Identifier: GPL-2.0-only
+ * Copyright (C) 2020 Marvell.
+ */
+
+#ifndef __OTX2_CPT_ALGS_H
+#define __OTX2_CPT_ALGS_H
+
+#include <crypto/hash.h>
+#include <crypto/skcipher.h>
+#include <crypto/aead.h>
+#include "otx2_cpt_common.h"
+
+#define OTX2_CPT_MAX_ENC_KEY_SIZE 32
+#define OTX2_CPT_MAX_HASH_KEY_SIZE 64
+#define OTX2_CPT_MAX_KEY_SIZE (OTX2_CPT_MAX_ENC_KEY_SIZE + \
+ OTX2_CPT_MAX_HASH_KEY_SIZE)
+enum otx2_cpt_request_type {
+ OTX2_CPT_ENC_DEC_REQ = 0x1,
+ OTX2_CPT_AEAD_ENC_DEC_REQ = 0x2,
+ OTX2_CPT_AEAD_ENC_DEC_NULL_REQ = 0x3,
+ OTX2_CPT_PASSTHROUGH_REQ = 0x4
+};
+
+enum otx2_cpt_major_opcodes {
+ OTX2_CPT_MAJOR_OP_MISC = 0x01,
+ OTX2_CPT_MAJOR_OP_FC = 0x33,
+ OTX2_CPT_MAJOR_OP_HMAC = 0x35,
+};
+
+enum otx2_cpt_cipher_type {
+ OTX2_CPT_CIPHER_NULL = 0x0,
+ OTX2_CPT_DES3_CBC = 0x1,
+ OTX2_CPT_DES3_ECB = 0x2,
+ OTX2_CPT_AES_CBC = 0x3,
+ OTX2_CPT_AES_ECB = 0x4,
+ OTX2_CPT_AES_CFB = 0x5,
+ OTX2_CPT_AES_CTR = 0x6,
+ OTX2_CPT_AES_GCM = 0x7,
+ OTX2_CPT_AES_XTS = 0x8
+};
+
+enum otx2_cpt_mac_type {
+ OTX2_CPT_MAC_NULL = 0x0,
+ OTX2_CPT_MD5 = 0x1,
+ OTX2_CPT_SHA1 = 0x2,
+ OTX2_CPT_SHA224 = 0x3,
+ OTX2_CPT_SHA256 = 0x4,
+ OTX2_CPT_SHA384 = 0x5,
+ OTX2_CPT_SHA512 = 0x6,
+ OTX2_CPT_GMAC = 0x7
+};
+
+enum otx2_cpt_aes_key_len {
+ OTX2_CPT_AES_128_BIT = 0x1,
+ OTX2_CPT_AES_192_BIT = 0x2,
+ OTX2_CPT_AES_256_BIT = 0x3
+};
+
+union otx2_cpt_encr_ctrl {
+ u64 u;
+ struct {
+#if defined(__BIG_ENDIAN_BITFIELD)
+ u64 enc_cipher:4;
+ u64 reserved_59:1;
+ u64 aes_key:2;
+ u64 iv_source:1;
+ u64 mac_type:4;
+ u64 reserved_49_51:3;
+ u64 auth_input_type:1;
+ u64 mac_len:8;
+ u64 reserved_32_39:8;
+ u64 encr_offset:16;
+ u64 iv_offset:8;
+ u64 auth_offset:8;
+#else
+ u64 auth_offset:8;
+ u64 iv_offset:8;
+ u64 encr_offset:16;
+ u64 reserved_32_39:8;
+ u64 mac_len:8;
+ u64 auth_input_type:1;
+ u64 reserved_49_51:3;
+ u64 mac_type:4;
+ u64 iv_source:1;
+ u64 aes_key:2;
+ u64 reserved_59:1;
+ u64 enc_cipher:4;
+#endif
+ } e;
+};
+
+struct otx2_cpt_cipher {
+ const char *name;
+ u8 value;
+};
+
+struct otx2_cpt_fc_enc_ctx {
+ union otx2_cpt_encr_ctrl enc_ctrl;
+ u8 encr_key[32];
+ u8 encr_iv[16];
+};
+
+union otx2_cpt_fc_hmac_ctx {
+ struct {
+ u8 ipad[64];
+ u8 opad[64];
+ } e;
+ struct {
+ u8 hmac_calc[64]; /* HMAC calculated */
+ u8 hmac_recv[64]; /* HMAC received */
+ } s;
+};
+
+struct otx2_cpt_fc_ctx {
+ struct otx2_cpt_fc_enc_ctx enc;
+ union otx2_cpt_fc_hmac_ctx hmac;
+};
+
+struct otx2_cpt_enc_ctx {
+ u32 key_len;
+ u8 enc_key[OTX2_CPT_MAX_KEY_SIZE];
+ u8 cipher_type;
+ u8 key_type;
+ u8 enc_align_len;
+ struct crypto_skcipher *fbk_cipher;
+};
+
+union otx2_cpt_offset_ctrl {
+ u64 flags;
+ struct {
+#if defined(__BIG_ENDIAN_BITFIELD)
+ u64 reserved:32;
+ u64 enc_data_offset:16;
+ u64 iv_offset:8;
+ u64 auth_offset:8;
+#else
+ u64 auth_offset:8;
+ u64 iv_offset:8;
+ u64 enc_data_offset:16;
+ u64 reserved:32;
+#endif
+ } e;
+};
+
+struct otx2_cpt_req_ctx {
+ struct otx2_cpt_req_info cpt_req;
+ union otx2_cpt_offset_ctrl ctrl_word;
+ struct otx2_cpt_fc_ctx fctx;
+ union {
+ struct skcipher_request sk_fbk_req;
+ struct aead_request fbk_req;
+ };
+};
+
+struct otx2_cpt_sdesc {
+ struct shash_desc shash;
+};
+
+struct otx2_cpt_aead_ctx {
+ u8 key[OTX2_CPT_MAX_KEY_SIZE];
+ struct crypto_shash *hashalg;
+ struct otx2_cpt_sdesc *sdesc;
+ struct crypto_aead *fbk_cipher;
+ u8 *ipad;
+ u8 *opad;
+ u32 enc_key_len;
+ u32 auth_key_len;
+ u8 cipher_type;
+ u8 mac_type;
+ u8 key_type;
+ u8 is_trunc_hmac;
+ u8 enc_align_len;
+};
+int otx2_cpt_crypto_init(struct pci_dev *pdev, struct module *mod,
+ int num_queues, int num_devices);
+void otx2_cpt_crypto_exit(struct pci_dev *pdev, struct module *mod);
+
+#endif /* __OTX2_CPT_ALGS_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c b/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
index 97d57bcb2b94..9663be38ee40 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
@@ -4,6 +4,7 @@
#include "otx2_cpt_common.h"
#include "otx2_cptvf.h"
#include "otx2_cptlf.h"
+#include "otx2_cptvf_algs.h"
#include <rvu_reg.h>
#define OTX2_CPTVF_DRV_NAME "octeontx2-cptvf"
@@ -214,6 +215,8 @@ static void cptvf_lf_shutdown(struct otx2_cptlfs_info *lfs)
otx2_cptlf_free_irqs_affinity(lfs);
/* Disable instruction queue */
otx2_cptlf_disable_iqueues(lfs);
+ /* Unregister crypto algorithms */
+ otx2_cpt_crypto_exit(lfs->pdev, THIS_MODULE);
/* Unregister LFs interrupts */
otx2_cptlf_unregister_interrupts(lfs);
/* Cleanup LFs software side */
@@ -278,9 +281,16 @@ static int cptvf_lf_init(struct otx2_cptvf_dev *cptvf)
goto unregister_intr;
atomic_set(&lfs->state, OTX2_CPTLF_STARTED);
-
+ /* Register crypto algorithms */
+ ret = otx2_cpt_crypto_init(lfs->pdev, THIS_MODULE, lfs_num, 1);
+ if (ret) {
+ dev_err(&lfs->pdev->dev, "algorithms registration failed\n");
+ goto disable_irqs;
+ }
return 0;
+disable_irqs:
+ otx2_cptlf_free_irqs_affinity(lfs);
unregister_intr:
otx2_cptlf_unregister_interrupts(lfs);
cleanup_lf_sw:
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c b/drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c
index b3b371ce06f9..d5c1c1b7c7e4 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c
@@ -532,3 +532,10 @@ void otx2_cpt_post_process(struct otx2_cptlf_wqe *wqe)
process_pending_queue(wqe->lfs->pdev,
&wqe->lfs->lf[wqe->lf_num].pqueue);
}
+
+int otx2_cpt_get_kcrypto_eng_grp_num(struct pci_dev *pdev)
+{
+ struct otx2_cptvf_dev *cptvf = pci_get_drvdata(pdev);
+
+ return cptvf->lfs.kcrypto_eng_grp_num;
+}
--
2.29.0
Attach LFs to CPT VF to process the crypto requests and register
LF interrupts.
Signed-off-by: Suheil Chandran <[email protected]>
Signed-off-by: Lukasz Bartosik <[email protected]>
Signed-off-by: Srujana Challa <[email protected]>
---
drivers/crypto/marvell/octeontx2/Makefile | 2 +-
.../marvell/octeontx2/otx2_cpt_common.h | 17 +
.../marvell/octeontx2/otx2_cpt_reqmgr.h | 145 +++++
drivers/crypto/marvell/octeontx2/otx2_cptlf.h | 8 +
drivers/crypto/marvell/octeontx2/otx2_cptpf.h | 1 +
.../marvell/octeontx2/otx2_cptpf_main.c | 47 ++
.../marvell/octeontx2/otx2_cptpf_mbox.c | 22 +
drivers/crypto/marvell/octeontx2/otx2_cptvf.h | 1 +
.../marvell/octeontx2/otx2_cptvf_main.c | 204 +++++++
.../marvell/octeontx2/otx2_cptvf_mbox.c | 54 ++
.../marvell/octeontx2/otx2_cptvf_reqmgr.c | 534 ++++++++++++++++++
11 files changed, 1034 insertions(+), 1 deletion(-)
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c
diff --git a/drivers/crypto/marvell/octeontx2/Makefile b/drivers/crypto/marvell/octeontx2/Makefile
index ef6fb2ab3571..41c0a5832b3f 100644
--- a/drivers/crypto/marvell/octeontx2/Makefile
+++ b/drivers/crypto/marvell/octeontx2/Makefile
@@ -4,6 +4,6 @@ obj-$(CONFIG_CRYPTO_DEV_OCTEONTX2_CPT) += octeontx2-cpt.o octeontx2-cptvf.o
octeontx2-cpt-objs := otx2_cptpf_main.o otx2_cptpf_mbox.o \
otx2_cpt_mbox_common.o otx2_cptpf_ucode.o otx2_cptlf.o
octeontx2-cptvf-objs := otx2_cptvf_main.o otx2_cptvf_mbox.o otx2_cptlf.o \
- otx2_cpt_mbox_common.o
+ otx2_cpt_mbox_common.o otx2_cptvf_reqmgr.o
ccflags-y += -I$(srctree)/drivers/net/ethernet/marvell/octeontx2/af
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h b/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
index ca220178e518..3518fac29834 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
@@ -17,6 +17,9 @@
#define OTX2_CPT_MAX_VFS_NUM 128
#define OTX2_CPT_RVU_FUNC_ADDR_S(blk, slot, offs) \
(((blk) << 20) | ((slot) << 12) | (offs))
+#define OTX2_CPT_RVU_PFFUNC(pf, func) \
+ ((((pf) & RVU_PFVF_PF_MASK) << RVU_PFVF_PF_SHIFT) | \
+ (((func) & RVU_PFVF_FUNC_MASK) << RVU_PFVF_FUNC_SHIFT))
#define OTX2_CPT_INVALID_CRYPTO_ENG_GRP 0xFF
#define OTX2_CPT_NAME_LENGTH 64
@@ -34,6 +37,7 @@ enum otx2_cpt_eng_type {
/* Take mbox id from end of CPT mbox range in AF (range 0xA00 - 0xBFF) */
#define MBOX_MSG_GET_ENG_GRP_NUM 0xBFF
#define MBOX_MSG_GET_CAPS 0xBFD
+#define MBOX_MSG_GET_KVF_LIMITS 0xBFC
/*
* Message request and response to get engine group number
@@ -51,6 +55,19 @@ struct otx2_cpt_egrp_num_rsp {
u8 eng_grp_num;
};
+/*
+ * Message request and response to get kernel crypto limits
+ * This messages are only used between CPT PF <-> CPT VF
+ */
+struct otx2_cpt_kvf_limits_msg {
+ struct mbox_msghdr hdr;
+};
+
+struct otx2_cpt_kvf_limits_rsp {
+ struct mbox_msghdr hdr;
+ u8 kvf_limits;
+};
+
/* CPT HW capabilities */
union otx2_cpt_eng_caps {
u64 u;
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h b/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
index 9184f91c68c1..597a998c6df6 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
@@ -10,6 +10,22 @@
/* Completion code size and initial value */
#define OTX2_CPT_COMPLETION_CODE_SIZE 8
#define OTX2_CPT_COMPLETION_CODE_INIT OTX2_CPT_COMP_E_NOTDONE
+/*
+ * Maximum total number of SG buffers is 100, we divide it equally
+ * between input and output
+ */
+#define OTX2_CPT_MAX_SG_IN_CNT 50
+#define OTX2_CPT_MAX_SG_OUT_CNT 50
+
+/* DMA mode direct or SG */
+#define OTX2_CPT_DMA_MODE_DIRECT 0
+#define OTX2_CPT_DMA_MODE_SG 1
+
+/* Context source CPTR or DPTR */
+#define OTX2_CPT_FROM_CPTR 0
+#define OTX2_CPT_FROM_DPTR 1
+
+#define OTX2_CPT_MAX_REQ_SIZE 65535
union otx2_cpt_opcode {
u16 flags;
@@ -19,6 +35,13 @@ union otx2_cpt_opcode {
} s;
};
+struct otx2_cptvf_request {
+ u32 param1;
+ u32 param2;
+ u16 dlen;
+ union otx2_cpt_opcode opcode;
+};
+
/*
* CPT_INST_S software command definitions
* Words EI (0-3)
@@ -48,4 +71,126 @@ struct otx2_cpt_iq_command {
union otx2_cpt_iq_cmd_word3 cptr;
};
+struct otx2_cpt_pending_entry {
+ void *completion_addr; /* Completion address */
+ void *info;
+ /* Kernel async request callback */
+ void (*callback)(int status, void *arg1, void *arg2);
+ struct crypto_async_request *areq; /* Async request callback arg */
+ u8 resume_sender; /* Notify sender to resume sending requests */
+ u8 busy; /* Entry status (free/busy) */
+};
+
+struct otx2_cpt_pending_queue {
+ struct otx2_cpt_pending_entry *head; /* Head of the queue */
+ u32 front; /* Process work from here */
+ u32 rear; /* Append new work here */
+ u32 pending_count; /* Pending requests count */
+ u32 qlen; /* Queue length */
+ spinlock_t lock; /* Queue lock */
+};
+
+struct otx2_cpt_buf_ptr {
+ u8 *vptr;
+ dma_addr_t dma_addr;
+ u16 size;
+};
+
+union otx2_cpt_ctrl_info {
+ u32 flags;
+ struct {
+#if defined(__BIG_ENDIAN_BITFIELD)
+ u32 reserved_6_31:26;
+ u32 grp:3; /* Group bits */
+ u32 dma_mode:2; /* DMA mode */
+ u32 se_req:1; /* To SE core */
+#else
+ u32 se_req:1; /* To SE core */
+ u32 dma_mode:2; /* DMA mode */
+ u32 grp:3; /* Group bits */
+ u32 reserved_6_31:26;
+#endif
+ } s;
+};
+
+struct otx2_cpt_req_info {
+ /* Kernel async request callback */
+ void (*callback)(int status, void *arg1, void *arg2);
+ struct crypto_async_request *areq; /* Async request callback arg */
+ struct otx2_cptvf_request req;/* Request information (core specific) */
+ union otx2_cpt_ctrl_info ctrl;/* User control information */
+ struct otx2_cpt_buf_ptr in[OTX2_CPT_MAX_SG_IN_CNT];
+ struct otx2_cpt_buf_ptr out[OTX2_CPT_MAX_SG_OUT_CNT];
+ u8 *iv_out; /* IV to send back */
+ u16 rlen; /* Output length */
+ u8 in_cnt; /* Number of input buffers */
+ u8 out_cnt; /* Number of output buffers */
+ u8 req_type; /* Type of request */
+ u8 is_enc; /* Is a request an encryption request */
+ u8 is_trunc_hmac;/* Is truncated hmac used */
+};
+
+struct otx2_cpt_inst_info {
+ struct otx2_cpt_pending_entry *pentry;
+ struct otx2_cpt_req_info *req;
+ struct pci_dev *pdev;
+ void *completion_addr;
+ u8 *out_buffer;
+ u8 *in_buffer;
+ dma_addr_t dptr_baddr;
+ dma_addr_t rptr_baddr;
+ dma_addr_t comp_baddr;
+ unsigned long time_in;
+ u32 dlen;
+ u32 dma_len;
+ u8 extra_time;
+};
+
+struct otx2_cpt_sglist_component {
+ __be16 len0;
+ __be16 len1;
+ __be16 len2;
+ __be16 len3;
+ __be64 ptr0;
+ __be64 ptr1;
+ __be64 ptr2;
+ __be64 ptr3;
+};
+
+static inline void otx2_cpt_info_destroy(struct pci_dev *pdev,
+ struct otx2_cpt_inst_info *info)
+{
+ struct otx2_cpt_req_info *req;
+ int i;
+
+ if (info->dptr_baddr)
+ dma_unmap_single(&pdev->dev, info->dptr_baddr,
+ info->dma_len, DMA_BIDIRECTIONAL);
+
+ if (info->req) {
+ req = info->req;
+ for (i = 0; i < req->out_cnt; i++) {
+ if (req->out[i].dma_addr)
+ dma_unmap_single(&pdev->dev,
+ req->out[i].dma_addr,
+ req->out[i].size,
+ DMA_BIDIRECTIONAL);
+ }
+
+ for (i = 0; i < req->in_cnt; i++) {
+ if (req->in[i].dma_addr)
+ dma_unmap_single(&pdev->dev,
+ req->in[i].dma_addr,
+ req->in[i].size,
+ DMA_BIDIRECTIONAL);
+ }
+ }
+ kfree(info);
+}
+
+struct otx2_cptlf_wqe;
+int otx2_cpt_do_request(struct pci_dev *pdev, struct otx2_cpt_req_info *req,
+ int cpu_num);
+void otx2_cpt_post_process(struct otx2_cptlf_wqe *wqe);
+
#endif /* __OTX2_CPT_REQMGR_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptlf.h b/drivers/crypto/marvell/octeontx2/otx2_cptlf.h
index efb4f395a54a..314e97354100 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptlf.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptlf.h
@@ -80,6 +80,7 @@ struct otx2_cptlf_info {
u8 slot; /* Slot number of this LF */
struct otx2_cpt_inst_queue iqueue;/* Instruction queue */
+ struct otx2_cpt_pending_queue pqueue; /* Pending queue */
struct otx2_cptlf_wqe *wqe; /* Tasklet work info */
};
@@ -91,6 +92,8 @@ struct otx2_cptlfs_info {
struct otx2_mbox *mbox;
u8 are_lfs_attached; /* Whether CPT LFs are attached */
u8 lfs_num; /* Number of CPT LFs */
+ u8 kcrypto_eng_grp_num; /* Kernel crypto engine group number */
+ u8 kvf_limits; /* Kernel crypto limits */
atomic_t state; /* LF's state. started/reset */
};
@@ -334,6 +337,11 @@ static inline void otx2_cpt_send_cmd(union otx2_cpt_inst_s *cptinst,
} while (!ret);
}
+static inline bool otx2_cptlf_started(struct otx2_cptlfs_info *lfs)
+{
+ return atomic_read(&lfs->state) == OTX2_CPTLF_STARTED;
+}
+
int otx2_cptlf_init(struct otx2_cptlfs_info *lfs, u8 eng_grp_msk, int pri,
int lfs_num);
void otx2_cptlf_shutdown(struct otx2_cptlfs_info *lfs);
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf.h b/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
index ede230dc205f..8c899ad531a5 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
@@ -50,6 +50,7 @@ struct otx2_cptpf_dev {
u8 pf_id; /* RVU PF number */
u8 max_vfs; /* Maximum number of VFs supported by CPT */
u8 enabled_vfs; /* Number of enabled VFs */
+ u8 kvf_limits; /* Kernel crypto limits */
};
irqreturn_t otx2_cptpf_afpf_mbox_intr(int irq, void *arg);
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
index 502bcf21ea89..5277e04badd9 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
@@ -392,6 +392,46 @@ static void cptpf_afpf_mbox_destroy(struct otx2_cptpf_dev *cptpf)
otx2_mbox_destroy(&cptpf->afpf_mbox);
}
+static ssize_t kvf_limits_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ struct otx2_cptpf_dev *cptpf = dev_get_drvdata(dev);
+
+ return sprintf(buf, "%d\n", cptpf->kvf_limits);
+}
+
+static ssize_t kvf_limits_store(struct device *dev,
+ struct device_attribute *attr,
+ const char *buf, size_t count)
+{
+ struct otx2_cptpf_dev *cptpf = dev_get_drvdata(dev);
+ int lfs_num;
+
+ if (kstrtoint(buf, 0, &lfs_num)) {
+ dev_err(dev, "lfs count %d must be in range [1 - %d]\n",
+ lfs_num, num_online_cpus());
+ return -EINVAL;
+ }
+ if (lfs_num < 1 || lfs_num > num_online_cpus()) {
+ dev_err(dev, "lfs count %d must be in range [1 - %d]\n",
+ lfs_num, num_online_cpus());
+ return -EINVAL;
+ }
+ cptpf->kvf_limits = lfs_num;
+
+ return count;
+}
+
+static DEVICE_ATTR_RW(kvf_limits);
+static struct attribute *cptpf_attrs[] = {
+ &dev_attr_kvf_limits.attr,
+ NULL
+};
+
+static const struct attribute_group cptpf_sysfs_group = {
+ .attrs = cptpf_attrs,
+};
+
static int cpt_is_pf_usable(struct otx2_cptpf_dev *cptpf)
{
u64 rev;
@@ -616,8 +656,13 @@ static int otx2_cptpf_probe(struct pci_dev *pdev,
if (err)
goto unregister_intr;
+ err = sysfs_create_group(&dev->kobj, &cptpf_sysfs_group);
+ if (err)
+ goto cleanup_eng_grps;
return 0;
+cleanup_eng_grps:
+ otx2_cpt_cleanup_eng_grps(pdev, &cptpf->eng_grps);
unregister_intr:
cptpf_disable_afpf_mbox_intr(cptpf);
destroy_afpf_mbox:
@@ -635,6 +680,8 @@ static void otx2_cptpf_remove(struct pci_dev *pdev)
return;
cptpf_sriov_disable(pdev);
+ /* Delete sysfs entry created for kernel VF limits */
+ sysfs_remove_group(&pdev->dev.kobj, &cptpf_sysfs_group);
/* Cleanup engine groups */
otx2_cpt_cleanup_eng_grps(pdev, &cptpf->eng_grps);
/* Disable AF-PF mailbox interrupt */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
index 0834dc3b3e59..186f1c1190c1 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
@@ -86,6 +86,25 @@ static int handle_msg_get_eng_grp_num(struct otx2_cptpf_dev *cptpf,
return 0;
}
+static int handle_msg_kvf_limits(struct otx2_cptpf_dev *cptpf,
+ struct otx2_cptvf_info *vf,
+ struct mbox_msghdr *req)
+{
+ struct otx2_cpt_kvf_limits_rsp *rsp;
+
+ rsp = (struct otx2_cpt_kvf_limits_rsp *)
+ otx2_mbox_alloc_msg(&cptpf->vfpf_mbox, vf->vf_id, sizeof(*rsp));
+ if (!rsp)
+ return -ENOMEM;
+
+ rsp->hdr.id = MBOX_MSG_GET_KVF_LIMITS;
+ rsp->hdr.sig = OTX2_MBOX_RSP_SIG;
+ rsp->hdr.pcifunc = req->pcifunc;
+ rsp->kvf_limits = cptpf->kvf_limits;
+
+ return 0;
+}
+
static int cptpf_handle_vf_req(struct otx2_cptpf_dev *cptpf,
struct otx2_cptvf_info *vf,
struct mbox_msghdr *req, int size)
@@ -103,6 +122,9 @@ static int cptpf_handle_vf_req(struct otx2_cptpf_dev *cptpf,
case MBOX_MSG_GET_CAPS:
err = handle_msg_get_caps(cptpf, vf, req);
break;
+ case MBOX_MSG_GET_KVF_LIMITS:
+ err = handle_msg_kvf_limits(cptpf, vf, req);
+ break;
default:
err = forward_to_af(cptpf, vf, req, size);
break;
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf.h b/drivers/crypto/marvell/octeontx2/otx2_cptvf.h
index 4b01eb9d9f70..4f0a169fddbd 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptvf.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf.h
@@ -24,5 +24,6 @@ struct otx2_cptvf_dev {
irqreturn_t otx2_cptvf_pfvf_mbox_intr(int irq, void *arg);
void otx2_cptvf_pfvf_mbox_handler(struct work_struct *work);
int otx2_cptvf_send_eng_grp_num_msg(struct otx2_cptvf_dev *cptvf, int eng_type);
+int otx2_cptvf_send_kvf_limits_msg(struct otx2_cptvf_dev *cptvf);
#endif /* __OTX2_CPTVF_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c b/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
index 6d84fafa2915..97d57bcb2b94 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
@@ -3,6 +3,7 @@
#include "otx2_cpt_common.h"
#include "otx2_cptvf.h"
+#include "otx2_cptlf.h"
#include <rvu_reg.h>
#define OTX2_CPTVF_DRV_NAME "octeontx2-cptvf"
@@ -95,6 +96,201 @@ static void cptvf_pfvf_mbox_destroy(struct otx2_cptvf_dev *cptvf)
otx2_mbox_destroy(&cptvf->pfvf_mbox);
}
+static void cptlf_work_handler(unsigned long data)
+{
+ otx2_cpt_post_process((struct otx2_cptlf_wqe *) data);
+}
+
+static void cleanup_tasklet_work(struct otx2_cptlfs_info *lfs)
+{
+ int i;
+
+ for (i = 0; i < lfs->lfs_num; i++) {
+ if (!lfs->lf[i].wqe)
+ continue;
+
+ tasklet_kill(&lfs->lf[i].wqe->work);
+ kfree(lfs->lf[i].wqe);
+ lfs->lf[i].wqe = NULL;
+ }
+}
+
+static int init_tasklet_work(struct otx2_cptlfs_info *lfs)
+{
+ struct otx2_cptlf_wqe *wqe;
+ int i, ret = 0;
+
+ for (i = 0; i < lfs->lfs_num; i++) {
+ wqe = kzalloc(sizeof(struct otx2_cptlf_wqe), GFP_KERNEL);
+ if (!wqe) {
+ ret = -ENOMEM;
+ goto cleanup_tasklet;
+ }
+
+ tasklet_init(&wqe->work, cptlf_work_handler, (u64) wqe);
+ wqe->lfs = lfs;
+ wqe->lf_num = i;
+ lfs->lf[i].wqe = wqe;
+ }
+ return 0;
+
+cleanup_tasklet:
+ cleanup_tasklet_work(lfs);
+ return ret;
+}
+
+static void free_pending_queues(struct otx2_cptlfs_info *lfs)
+{
+ int i;
+
+ for (i = 0; i < lfs->lfs_num; i++) {
+ kfree(lfs->lf[i].pqueue.head);
+ lfs->lf[i].pqueue.head = NULL;
+ }
+}
+
+static int alloc_pending_queues(struct otx2_cptlfs_info *lfs)
+{
+ int size, ret, i;
+
+ if (!lfs->lfs_num)
+ return -EINVAL;
+
+ for (i = 0; i < lfs->lfs_num; i++) {
+ lfs->lf[i].pqueue.qlen = OTX2_CPT_INST_QLEN_MSGS;
+ size = lfs->lf[i].pqueue.qlen *
+ sizeof(struct otx2_cpt_pending_entry);
+
+ lfs->lf[i].pqueue.head = kzalloc(size, GFP_KERNEL);
+ if (!lfs->lf[i].pqueue.head) {
+ ret = -ENOMEM;
+ goto error;
+ }
+
+ /* Initialize spin lock */
+ spin_lock_init(&lfs->lf[i].pqueue.lock);
+ }
+ return 0;
+
+error:
+ free_pending_queues(lfs);
+ return ret;
+}
+
+static void lf_sw_cleanup(struct otx2_cptlfs_info *lfs)
+{
+ cleanup_tasklet_work(lfs);
+ free_pending_queues(lfs);
+}
+
+static int lf_sw_init(struct otx2_cptlfs_info *lfs)
+{
+ int ret;
+
+ ret = alloc_pending_queues(lfs);
+ if (ret) {
+ dev_err(&lfs->pdev->dev,
+ "Allocating pending queues failed\n");
+ return ret;
+ }
+ ret = init_tasklet_work(lfs);
+ if (ret) {
+ dev_err(&lfs->pdev->dev,
+ "Tasklet work init failed\n");
+ goto pending_queues_free;
+ }
+ return 0;
+
+pending_queues_free:
+ free_pending_queues(lfs);
+ return ret;
+}
+
+static void cptvf_lf_shutdown(struct otx2_cptlfs_info *lfs)
+{
+ atomic_set(&lfs->state, OTX2_CPTLF_IN_RESET);
+
+ /* Remove interrupts affinity */
+ otx2_cptlf_free_irqs_affinity(lfs);
+ /* Disable instruction queue */
+ otx2_cptlf_disable_iqueues(lfs);
+ /* Unregister LFs interrupts */
+ otx2_cptlf_unregister_interrupts(lfs);
+ /* Cleanup LFs software side */
+ lf_sw_cleanup(lfs);
+ /* Send request to detach LFs */
+ otx2_cpt_detach_rsrcs_msg(lfs);
+}
+
+static int cptvf_lf_init(struct otx2_cptvf_dev *cptvf)
+{
+ struct otx2_cptlfs_info *lfs = &cptvf->lfs;
+ struct device *dev = &cptvf->pdev->dev;
+ int ret, lfs_num;
+ u8 eng_grp_msk;
+
+ /* Get engine group number for symmetric crypto */
+ cptvf->lfs.kcrypto_eng_grp_num = OTX2_CPT_INVALID_CRYPTO_ENG_GRP;
+ ret = otx2_cptvf_send_eng_grp_num_msg(cptvf, OTX2_CPT_SE_TYPES);
+ if (ret)
+ return ret;
+
+ if (cptvf->lfs.kcrypto_eng_grp_num == OTX2_CPT_INVALID_CRYPTO_ENG_GRP) {
+ dev_err(dev, "Engine group for kernel crypto not available\n");
+ ret = -ENOENT;
+ return ret;
+ }
+ eng_grp_msk = 1 << cptvf->lfs.kcrypto_eng_grp_num;
+
+ ret = otx2_cptvf_send_kvf_limits_msg(cptvf);
+ if (ret)
+ return ret;
+
+ lfs->reg_base = cptvf->reg_base;
+ lfs->pdev = cptvf->pdev;
+ lfs->mbox = &cptvf->pfvf_mbox;
+
+ lfs_num = cptvf->lfs.kvf_limits ? cptvf->lfs.kvf_limits :
+ num_online_cpus();
+ ret = otx2_cptlf_init(lfs, eng_grp_msk, OTX2_CPT_QUEUE_HI_PRIO,
+ lfs_num);
+ if (ret)
+ return ret;
+
+ /* Get msix offsets for attached LFs */
+ ret = otx2_cpt_msix_offset_msg(lfs);
+ if (ret)
+ goto cleanup_lf;
+
+ /* Initialize LFs software side */
+ ret = lf_sw_init(lfs);
+ if (ret)
+ goto cleanup_lf;
+
+ /* Register LFs interrupts */
+ ret = otx2_cptlf_register_interrupts(lfs);
+ if (ret)
+ goto cleanup_lf_sw;
+
+ /* Set interrupts affinity */
+ ret = otx2_cptlf_set_irqs_affinity(lfs);
+ if (ret)
+ goto unregister_intr;
+
+ atomic_set(&lfs->state, OTX2_CPTLF_STARTED);
+
+ return 0;
+
+unregister_intr:
+ otx2_cptlf_unregister_interrupts(lfs);
+cleanup_lf_sw:
+ lf_sw_cleanup(lfs);
+cleanup_lf:
+ otx2_cptlf_shutdown(lfs);
+
+ return ret;
+}
+
static int otx2_cptvf_probe(struct pci_dev *pdev,
const struct pci_device_id *ent)
{
@@ -150,8 +346,15 @@ static int otx2_cptvf_probe(struct pci_dev *pdev,
if (ret)
goto destroy_pfvf_mbox;
+ /* Initialize CPT LFs */
+ ret = cptvf_lf_init(cptvf);
+ if (ret)
+ goto unregister_interrupts;
+
return 0;
+unregister_interrupts:
+ cptvf_disable_pfvf_mbox_intrs(cptvf);
destroy_pfvf_mbox:
cptvf_pfvf_mbox_destroy(cptvf);
clear_drvdata:
@@ -168,6 +371,7 @@ static void otx2_cptvf_remove(struct pci_dev *pdev)
dev_err(&pdev->dev, "Invalid CPT VF device.\n");
return;
}
+ cptvf_lf_shutdown(&cptvf->lfs);
/* Disable PF-VF mailbox interrupt */
cptvf_disable_pfvf_mbox_intrs(cptvf);
/* Destroy PF-VF mbox */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c b/drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c
index 417099a86742..5d73b711cba6 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c
@@ -28,6 +28,8 @@ static void process_pfvf_mbox_mbox_msg(struct otx2_cptvf_dev *cptvf,
struct mbox_msghdr *msg)
{
struct otx2_cptlfs_info *lfs = &cptvf->lfs;
+ struct otx2_cpt_kvf_limits_rsp *rsp_limits;
+ struct otx2_cpt_egrp_num_rsp *rsp_grp;
struct cpt_rd_wr_reg_msg *rsp_reg;
struct msix_offset_rsp *rsp_msix;
int i;
@@ -75,6 +77,14 @@ static void process_pfvf_mbox_mbox_msg(struct otx2_cptvf_dev *cptvf,
if (!rsp_reg->is_write)
*rsp_reg->ret_val = rsp_reg->val;
break;
+ case MBOX_MSG_GET_ENG_GRP_NUM:
+ rsp_grp = (struct otx2_cpt_egrp_num_rsp *) msg;
+ cptvf->lfs.kcrypto_eng_grp_num = rsp_grp->eng_grp_num;
+ break;
+ case MBOX_MSG_GET_KVF_LIMITS:
+ rsp_limits = (struct otx2_cpt_kvf_limits_rsp *) msg;
+ cptvf->lfs.kvf_limits = rsp_limits->kvf_limits;
+ break;
default:
dev_err(&cptvf->pdev->dev, "Unsupported msg %d received.\n",
msg->id);
@@ -111,3 +121,47 @@ void otx2_cptvf_pfvf_mbox_handler(struct work_struct *work)
}
otx2_mbox_reset(pfvf_mbox, 0);
}
+
+int otx2_cptvf_send_eng_grp_num_msg(struct otx2_cptvf_dev *cptvf, int eng_type)
+{
+ struct otx2_mbox *mbox = &cptvf->pfvf_mbox;
+ struct pci_dev *pdev = cptvf->pdev;
+ struct otx2_cpt_egrp_num_msg *req;
+
+ req = (struct otx2_cpt_egrp_num_msg *)
+ otx2_mbox_alloc_msg_rsp(mbox, 0, sizeof(*req),
+ sizeof(struct otx2_cpt_egrp_num_rsp));
+ if (req == NULL) {
+ dev_err(&pdev->dev, "RVU MBOX failed to get message.\n");
+ return -EFAULT;
+ }
+ req->hdr.id = MBOX_MSG_GET_ENG_GRP_NUM;
+ req->hdr.sig = OTX2_MBOX_REQ_SIG;
+ req->hdr.pcifunc = OTX2_CPT_RVU_PFFUNC(cptvf->vf_id, 0);
+ req->eng_type = eng_type;
+
+ return otx2_cpt_send_mbox_msg(mbox, pdev);
+}
+
+int otx2_cptvf_send_kvf_limits_msg(struct otx2_cptvf_dev *cptvf)
+{
+ struct otx2_mbox *mbox = &cptvf->pfvf_mbox;
+ struct pci_dev *pdev = cptvf->pdev;
+ struct mbox_msghdr *req;
+ int ret;
+
+ req = (struct mbox_msghdr *)
+ otx2_mbox_alloc_msg_rsp(mbox, 0, sizeof(*req),
+ sizeof(struct otx2_cpt_kvf_limits_rsp));
+ if (req == NULL) {
+ dev_err(&pdev->dev, "RVU MBOX failed to get message.\n");
+ return -EFAULT;
+ }
+ req->id = MBOX_MSG_GET_KVF_LIMITS;
+ req->sig = OTX2_MBOX_REQ_SIG;
+ req->pcifunc = OTX2_CPT_RVU_PFFUNC(cptvf->vf_id, 0);
+
+ ret = otx2_cpt_send_mbox_msg(mbox, pdev);
+
+ return ret;
+}
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c b/drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c
new file mode 100644
index 000000000000..b3b371ce06f9
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c
@@ -0,0 +1,534 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (C) 2020 Marvell. */
+
+#include "otx2_cptvf.h"
+#include "otx2_cpt_common.h"
+
+/* SG list header size in bytes */
+#define SG_LIST_HDR_SIZE 8
+
+/* Default timeout when waiting for free pending entry in us */
+#define CPT_PENTRY_TIMEOUT 1000
+#define CPT_PENTRY_STEP 50
+
+/* Default threshold for stopping and resuming sender requests */
+#define CPT_IQ_STOP_MARGIN 128
+#define CPT_IQ_RESUME_MARGIN 512
+
+/* Default command timeout in seconds */
+#define CPT_COMMAND_TIMEOUT 4
+#define CPT_TIME_IN_RESET_COUNT 5
+
+static void otx2_cpt_dump_sg_list(struct pci_dev *pdev,
+ struct otx2_cpt_req_info *req)
+{
+ int i;
+
+ pr_debug("Gather list size %d\n", req->in_cnt);
+ for (i = 0; i < req->in_cnt; i++) {
+ pr_debug("Buffer %d size %d, vptr 0x%p, dmaptr 0x%p\n", i,
+ req->in[i].size, req->in[i].vptr,
+ (void *) req->in[i].dma_addr);
+ pr_debug("Buffer hexdump (%d bytes)\n",
+ req->in[i].size);
+ print_hex_dump_debug("", DUMP_PREFIX_NONE, 16, 1,
+ req->in[i].vptr, req->in[i].size, false);
+ }
+ pr_debug("Scatter list size %d\n", req->out_cnt);
+ for (i = 0; i < req->out_cnt; i++) {
+ pr_debug("Buffer %d size %d, vptr 0x%p, dmaptr 0x%p\n", i,
+ req->out[i].size, req->out[i].vptr,
+ (void *) req->out[i].dma_addr);
+ pr_debug("Buffer hexdump (%d bytes)\n", req->out[i].size);
+ print_hex_dump_debug("", DUMP_PREFIX_NONE, 16, 1,
+ req->out[i].vptr, req->out[i].size, false);
+ }
+}
+
+static inline struct otx2_cpt_pending_entry *get_free_pending_entry(
+ struct otx2_cpt_pending_queue *q,
+ int qlen)
+{
+ struct otx2_cpt_pending_entry *ent = NULL;
+
+ ent = &q->head[q->rear];
+ if (unlikely(ent->busy))
+ return NULL;
+
+ q->rear++;
+ if (unlikely(q->rear == qlen))
+ q->rear = 0;
+
+ return ent;
+}
+
+static inline u32 modulo_inc(u32 index, u32 length, u32 inc)
+{
+ if (WARN_ON(inc > length))
+ inc = length;
+
+ index += inc;
+ if (unlikely(index >= length))
+ index -= length;
+
+ return index;
+}
+
+static inline void free_pentry(struct otx2_cpt_pending_entry *pentry)
+{
+ pentry->completion_addr = NULL;
+ pentry->info = NULL;
+ pentry->callback = NULL;
+ pentry->areq = NULL;
+ pentry->resume_sender = false;
+ pentry->busy = false;
+}
+
+static inline int setup_sgio_components(struct pci_dev *pdev,
+ struct otx2_cpt_buf_ptr *list,
+ int buf_count, u8 *buffer)
+{
+ struct otx2_cpt_sglist_component *sg_ptr = NULL;
+ int ret = 0, i, j;
+ int components;
+
+ if (unlikely(!list)) {
+ dev_err(&pdev->dev, "Input list pointer is NULL\n");
+ return -EFAULT;
+ }
+
+ for (i = 0; i < buf_count; i++) {
+ if (unlikely(!list[i].vptr))
+ continue;
+ list[i].dma_addr = dma_map_single(&pdev->dev, list[i].vptr,
+ list[i].size,
+ DMA_BIDIRECTIONAL);
+ if (unlikely(dma_mapping_error(&pdev->dev, list[i].dma_addr))) {
+ dev_err(&pdev->dev, "Dma mapping failed\n");
+ ret = -EIO;
+ goto sg_cleanup;
+ }
+ }
+ components = buf_count / 4;
+ sg_ptr = (struct otx2_cpt_sglist_component *)buffer;
+ for (i = 0; i < components; i++) {
+ sg_ptr->len0 = cpu_to_be16(list[i * 4 + 0].size);
+ sg_ptr->len1 = cpu_to_be16(list[i * 4 + 1].size);
+ sg_ptr->len2 = cpu_to_be16(list[i * 4 + 2].size);
+ sg_ptr->len3 = cpu_to_be16(list[i * 4 + 3].size);
+ sg_ptr->ptr0 = cpu_to_be64(list[i * 4 + 0].dma_addr);
+ sg_ptr->ptr1 = cpu_to_be64(list[i * 4 + 1].dma_addr);
+ sg_ptr->ptr2 = cpu_to_be64(list[i * 4 + 2].dma_addr);
+ sg_ptr->ptr3 = cpu_to_be64(list[i * 4 + 3].dma_addr);
+ sg_ptr++;
+ }
+ components = buf_count % 4;
+
+ switch (components) {
+ case 3:
+ sg_ptr->len2 = cpu_to_be16(list[i * 4 + 2].size);
+ sg_ptr->ptr2 = cpu_to_be64(list[i * 4 + 2].dma_addr);
+ fallthrough;
+ case 2:
+ sg_ptr->len1 = cpu_to_be16(list[i * 4 + 1].size);
+ sg_ptr->ptr1 = cpu_to_be64(list[i * 4 + 1].dma_addr);
+ fallthrough;
+ case 1:
+ sg_ptr->len0 = cpu_to_be16(list[i * 4 + 0].size);
+ sg_ptr->ptr0 = cpu_to_be64(list[i * 4 + 0].dma_addr);
+ break;
+ default:
+ break;
+ }
+ return ret;
+
+sg_cleanup:
+ for (j = 0; j < i; j++) {
+ if (list[j].dma_addr) {
+ dma_unmap_single(&pdev->dev, list[j].dma_addr,
+ list[j].size, DMA_BIDIRECTIONAL);
+ }
+
+ list[j].dma_addr = 0;
+ }
+ return ret;
+}
+
+static inline struct otx2_cpt_inst_info *info_create(struct pci_dev *pdev,
+ struct otx2_cpt_req_info *req,
+ gfp_t gfp)
+{
+ int align = OTX2_CPT_DMA_MINALIGN;
+ struct otx2_cpt_inst_info *info;
+ u32 dlen, align_dlen, info_len;
+ u16 g_sz_bytes, s_sz_bytes;
+ u32 total_mem_len;
+
+ if (unlikely(req->in_cnt > OTX2_CPT_MAX_SG_IN_CNT ||
+ req->out_cnt > OTX2_CPT_MAX_SG_OUT_CNT)) {
+ dev_err(&pdev->dev, "Error too many sg components\n");
+ return NULL;
+ }
+
+ g_sz_bytes = ((req->in_cnt + 3) / 4) *
+ sizeof(struct otx2_cpt_sglist_component);
+ s_sz_bytes = ((req->out_cnt + 3) / 4) *
+ sizeof(struct otx2_cpt_sglist_component);
+
+ dlen = g_sz_bytes + s_sz_bytes + SG_LIST_HDR_SIZE;
+ align_dlen = ALIGN(dlen, align);
+ info_len = ALIGN(sizeof(*info), align);
+ total_mem_len = align_dlen + info_len + sizeof(union otx2_cpt_res_s);
+
+ info = kzalloc(total_mem_len, gfp);
+ if (unlikely(!info))
+ return NULL;
+
+ info->dlen = dlen;
+ info->in_buffer = (u8 *)info + info_len;
+
+ ((u16 *)info->in_buffer)[0] = req->out_cnt;
+ ((u16 *)info->in_buffer)[1] = req->in_cnt;
+ ((u16 *)info->in_buffer)[2] = 0;
+ ((u16 *)info->in_buffer)[3] = 0;
+ cpu_to_be64s((u64 *)info->in_buffer);
+
+ /* Setup gather (input) components */
+ if (setup_sgio_components(pdev, req->in, req->in_cnt,
+ &info->in_buffer[8])) {
+ dev_err(&pdev->dev, "Failed to setup gather list\n");
+ goto destroy_info;
+ }
+
+ if (setup_sgio_components(pdev, req->out, req->out_cnt,
+ &info->in_buffer[8 + g_sz_bytes])) {
+ dev_err(&pdev->dev, "Failed to setup scatter list\n");
+ goto destroy_info;
+ }
+
+ info->dma_len = total_mem_len - info_len;
+ info->dptr_baddr = dma_map_single(&pdev->dev, info->in_buffer,
+ info->dma_len, DMA_BIDIRECTIONAL);
+ if (unlikely(dma_mapping_error(&pdev->dev, info->dptr_baddr))) {
+ dev_err(&pdev->dev, "DMA Mapping failed for cpt req\n");
+ goto destroy_info;
+ }
+ /*
+ * Get buffer for union otx2_cpt_res_s response
+ * structure and its physical address
+ */
+ info->completion_addr = info->in_buffer + align_dlen;
+ info->comp_baddr = info->dptr_baddr + align_dlen;
+
+ return info;
+
+destroy_info:
+ otx2_cpt_info_destroy(pdev, info);
+ return NULL;
+}
+
+static int process_request(struct pci_dev *pdev, struct otx2_cpt_req_info *req,
+ struct otx2_cpt_pending_queue *pqueue,
+ struct otx2_cptlf_info *lf)
+{
+ struct otx2_cptvf_request *cpt_req = &req->req;
+ struct otx2_cpt_pending_entry *pentry = NULL;
+ union otx2_cpt_ctrl_info *ctrl = &req->ctrl;
+ struct otx2_cpt_inst_info *info = NULL;
+ union otx2_cpt_res_s *result = NULL;
+ struct otx2_cpt_iq_command iq_cmd;
+ union otx2_cpt_inst_s cptinst;
+ int retry, ret = 0;
+ u8 resume_sender;
+ gfp_t gfp;
+
+ gfp = (req->areq->flags & CRYPTO_TFM_REQ_MAY_SLEEP) ? GFP_KERNEL :
+ GFP_ATOMIC;
+ if (unlikely(!otx2_cptlf_started(lf->lfs)))
+ return -ENODEV;
+
+ info = info_create(pdev, req, gfp);
+ if (unlikely(!info)) {
+ dev_err(&pdev->dev, "Setting up cpt inst info failed");
+ return -ENOMEM;
+ }
+ cpt_req->dlen = info->dlen;
+
+ result = info->completion_addr;
+ result->s.compcode = OTX2_CPT_COMPLETION_CODE_INIT;
+
+ spin_lock_bh(&pqueue->lock);
+ pentry = get_free_pending_entry(pqueue, pqueue->qlen);
+ retry = CPT_PENTRY_TIMEOUT / CPT_PENTRY_STEP;
+ while (unlikely(!pentry) && retry--) {
+ spin_unlock_bh(&pqueue->lock);
+ udelay(CPT_PENTRY_STEP);
+ spin_lock_bh(&pqueue->lock);
+ pentry = get_free_pending_entry(pqueue, pqueue->qlen);
+ }
+
+ if (unlikely(!pentry)) {
+ ret = -ENOSPC;
+ goto destroy_info;
+ }
+
+ /*
+ * Check if we are close to filling in entire pending queue,
+ * if so then tell the sender to stop/sleep by returning -EBUSY
+ * We do it only for context which can sleep (GFP_KERNEL)
+ */
+ if (gfp == GFP_KERNEL &&
+ pqueue->pending_count > (pqueue->qlen - CPT_IQ_STOP_MARGIN)) {
+ pentry->resume_sender = true;
+ } else
+ pentry->resume_sender = false;
+ resume_sender = pentry->resume_sender;
+ pqueue->pending_count++;
+
+ pentry->completion_addr = info->completion_addr;
+ pentry->info = info;
+ pentry->callback = req->callback;
+ pentry->areq = req->areq;
+ pentry->busy = true;
+ info->pentry = pentry;
+ info->time_in = jiffies;
+ info->req = req;
+
+ /* Fill in the command */
+ iq_cmd.cmd.u = 0;
+ iq_cmd.cmd.s.opcode = cpu_to_be16(cpt_req->opcode.flags);
+ iq_cmd.cmd.s.param1 = cpu_to_be16(cpt_req->param1);
+ iq_cmd.cmd.s.param2 = cpu_to_be16(cpt_req->param2);
+ iq_cmd.cmd.s.dlen = cpu_to_be16(cpt_req->dlen);
+
+ /* 64-bit swap for microcode data reads, not needed for addresses*/
+ cpu_to_be64s(&iq_cmd.cmd.u);
+ iq_cmd.dptr = info->dptr_baddr;
+ iq_cmd.rptr = 0;
+ iq_cmd.cptr.u = 0;
+ iq_cmd.cptr.s.grp = ctrl->s.grp;
+
+ /* Fill in the CPT_INST_S type command for HW interpretation */
+ otx2_cpt_fill_inst(&cptinst, &iq_cmd, info->comp_baddr);
+
+ /* Print debug info if enabled */
+ otx2_cpt_dump_sg_list(pdev, req);
+ pr_debug("Cpt_inst_s hexdump (%d bytes)\n", OTX2_CPT_INST_SIZE);
+ print_hex_dump_debug("", 0, 16, 1, &cptinst, OTX2_CPT_INST_SIZE, false);
+ pr_debug("Dptr hexdump (%d bytes)\n", cpt_req->dlen);
+ print_hex_dump_debug("", 0, 16, 1, info->in_buffer,
+ cpt_req->dlen, false);
+
+ /* Send CPT command */
+ otx2_cpt_send_cmd(&cptinst, 1, lf);
+
+ /*
+ * We allocate and prepare pending queue entry in critical section
+ * together with submitting CPT instruction to CPT instruction queue
+ * to make sure that order of CPT requests is the same in both
+ * pending and instruction queues
+ */
+ spin_unlock_bh(&pqueue->lock);
+
+ ret = resume_sender ? -EBUSY : -EINPROGRESS;
+ return ret;
+
+destroy_info:
+ spin_unlock_bh(&pqueue->lock);
+ otx2_cpt_info_destroy(pdev, info);
+ return ret;
+}
+
+int otx2_cpt_do_request(struct pci_dev *pdev, struct otx2_cpt_req_info *req,
+ int cpu_num)
+{
+ struct otx2_cptvf_dev *cptvf = pci_get_drvdata(pdev);
+ struct otx2_cptlfs_info *lfs = &cptvf->lfs;
+
+ return process_request(lfs->pdev, req, &lfs->lf[cpu_num].pqueue,
+ &lfs->lf[cpu_num]);
+}
+
+static int cpt_process_ccode(struct pci_dev *pdev,
+ union otx2_cpt_res_s *cpt_status,
+ struct otx2_cpt_inst_info *info,
+ u32 *res_code)
+{
+ u8 uc_ccode = cpt_status->s.uc_compcode;
+ u8 ccode = cpt_status->s.compcode;
+
+ switch (ccode) {
+ case OTX2_CPT_COMP_E_FAULT:
+ dev_err(&pdev->dev,
+ "Request failed with DMA fault\n");
+ otx2_cpt_dump_sg_list(pdev, info->req);
+ break;
+
+ case OTX2_CPT_COMP_E_HWERR:
+ dev_err(&pdev->dev,
+ "Request failed with hardware error\n");
+ otx2_cpt_dump_sg_list(pdev, info->req);
+ break;
+
+ case OTX2_CPT_COMP_E_INSTERR:
+ dev_err(&pdev->dev,
+ "Request failed with instruction error\n");
+ otx2_cpt_dump_sg_list(pdev, info->req);
+ break;
+
+ case OTX2_CPT_COMP_E_NOTDONE:
+ /* check for timeout */
+ if (time_after_eq(jiffies, info->time_in +
+ CPT_COMMAND_TIMEOUT * HZ))
+ dev_warn(&pdev->dev,
+ "Request timed out 0x%p", info->req);
+ else if (info->extra_time < CPT_TIME_IN_RESET_COUNT) {
+ info->time_in = jiffies;
+ info->extra_time++;
+ }
+ return 1;
+
+ case OTX2_CPT_COMP_E_GOOD:
+ /*
+ * Check microcode completion code, it is only valid
+ * when completion code is CPT_COMP_E::GOOD
+ */
+ if (uc_ccode != OTX2_CPT_UCC_SUCCESS) {
+ /*
+ * If requested hmac is truncated and ucode returns
+ * s/g write length error then we report success
+ * because ucode writes as many bytes of calculated
+ * hmac as available in gather buffer and reports
+ * s/g write length error if number of bytes in gather
+ * buffer is less than full hmac size.
+ */
+ if (info->req->is_trunc_hmac &&
+ uc_ccode == OTX2_CPT_UCC_SG_WRITE_LENGTH) {
+ *res_code = 0;
+ break;
+ }
+
+ dev_err(&pdev->dev,
+ "Request failed with software error code 0x%x\n",
+ cpt_status->s.uc_compcode);
+ otx2_cpt_dump_sg_list(pdev, info->req);
+ break;
+ }
+ /* Request has been processed with success */
+ *res_code = 0;
+ break;
+
+ default:
+ dev_err(&pdev->dev,
+ "Request returned invalid status %d\n", ccode);
+ break;
+ }
+ return 0;
+}
+
+static inline void process_pending_queue(struct pci_dev *pdev,
+ struct otx2_cpt_pending_queue *pqueue)
+{
+ struct otx2_cpt_pending_entry *resume_pentry = NULL;
+ void (*callback)(int status, void *arg, void *req);
+ struct otx2_cpt_pending_entry *pentry = NULL;
+ union otx2_cpt_res_s *cpt_status = NULL;
+ struct otx2_cpt_inst_info *info = NULL;
+ struct otx2_cpt_req_info *req = NULL;
+ struct crypto_async_request *areq;
+ u32 res_code, resume_index;
+
+ while (1) {
+ spin_lock_bh(&pqueue->lock);
+ pentry = &pqueue->head[pqueue->front];
+
+ if (WARN_ON(!pentry)) {
+ spin_unlock_bh(&pqueue->lock);
+ break;
+ }
+
+ res_code = -EINVAL;
+ if (unlikely(!pentry->busy)) {
+ spin_unlock_bh(&pqueue->lock);
+ break;
+ }
+
+ if (unlikely(!pentry->callback)) {
+ dev_err(&pdev->dev, "Callback NULL\n");
+ goto process_pentry;
+ }
+
+ info = pentry->info;
+ if (unlikely(!info)) {
+ dev_err(&pdev->dev, "Pending entry post arg NULL\n");
+ goto process_pentry;
+ }
+
+ req = info->req;
+ if (unlikely(!req)) {
+ dev_err(&pdev->dev, "Request NULL\n");
+ goto process_pentry;
+ }
+
+ cpt_status = pentry->completion_addr;
+ if (unlikely(!cpt_status)) {
+ dev_err(&pdev->dev, "Completion address NULL\n");
+ goto process_pentry;
+ }
+
+ if (cpt_process_ccode(pdev, cpt_status, info, &res_code)) {
+ spin_unlock_bh(&pqueue->lock);
+ return;
+ }
+ info->pdev = pdev;
+
+process_pentry:
+ /*
+ * Check if we should inform sending side to resume
+ * We do it CPT_IQ_RESUME_MARGIN elements in advance before
+ * pending queue becomes empty
+ */
+ resume_index = modulo_inc(pqueue->front, pqueue->qlen,
+ CPT_IQ_RESUME_MARGIN);
+ resume_pentry = &pqueue->head[resume_index];
+ if (resume_pentry &&
+ resume_pentry->resume_sender) {
+ resume_pentry->resume_sender = false;
+ callback = resume_pentry->callback;
+ areq = resume_pentry->areq;
+
+ if (callback) {
+ spin_unlock_bh(&pqueue->lock);
+
+ /*
+ * EINPROGRESS is an indication for sending
+ * side that it can resume sending requests
+ */
+ callback(-EINPROGRESS, areq, info);
+ spin_lock_bh(&pqueue->lock);
+ }
+ }
+
+ callback = pentry->callback;
+ areq = pentry->areq;
+ free_pentry(pentry);
+
+ pqueue->pending_count--;
+ pqueue->front = modulo_inc(pqueue->front, pqueue->qlen, 1);
+ spin_unlock_bh(&pqueue->lock);
+
+ /*
+ * Call callback after current pending entry has been
+ * processed, we don't do it if the callback pointer is
+ * invalid.
+ */
+ if (callback)
+ callback(res_code, areq, info);
+ }
+}
+
+void otx2_cpt_post_process(struct otx2_cptlf_wqe *wqe)
+{
+ process_pending_queue(wqe->lfs->pdev,
+ &wqe->lfs->lf[wqe->lf_num].pqueue);
+}
--
2.29.0
On Fri, Jan 15, 2021 at 07:22:18PM +0530, Srujana Challa wrote:
> This series introduces crypto(CPT) drivers(PF & VF) for Marvell
> OcteonTX2 CN96XX Soc.
>
> OcteonTX2 SOC's resource virtualization unit (RVU) supports multiple
> physical and virtual functions. Each of the PF/VF's functionality is
> determined by what kind of resources are attached to it. When the CPT
> block is attached to a VF, it can function as a security device.
>
> The CPT PF driver is responsible for:
> - Forwarding messages to/from VFs from/to admin function(AF),
> - Enabling/disabling VFs,
> - Loading/unloading microcode (creation/deletion of engine groups).
>
> The CPT VF driver works as a crypto offload device.
>
> This patch series includes:
> - CPT PF driver patches that include AF<=>PF<=>VF mailbox communication,
> sriov_configure, and firmware load to the acceleration engines.
> - CPT VF driver patches that include VF<=>PF mailbox communication and
> crypto offload support through the kernel cryptographic API.
>
> This series is tested with CRYPTO_EXTRA_TESTS enabled and
> CRYPTO_DISABLE_TESTS disabled.
>
> Changes since v1:
> * Resolved compilation warning.
>
>
> Srujana Challa (9):
> drivers: crypto: add Marvell OcteonTX2 CPT PF driver
> crypto: octeontx2: add mailbox communication with AF
> crypto: octeontx2: enable SR-IOV and mailbox communication with VF
> crypto: octeontx2: load microcode and create engine groups
> crypto: octeontx2: add LF framework
> crypto: octeontx2: add support to get engine capabilities
> crypto: octeontx2: add virtual function driver support
> crypto: octeontx2: add support to process the crypto request
> crypto: octeontx2: register with linux crypto framework
>
> drivers/crypto/marvell/Kconfig | 14 +
> drivers/crypto/marvell/Makefile | 1 +
> drivers/crypto/marvell/octeontx2/Makefile | 10 +
> .../marvell/octeontx2/otx2_cpt_common.h | 137 ++
> .../marvell/octeontx2/otx2_cpt_hw_types.h | 464 +++++
> .../marvell/octeontx2/otx2_cpt_mbox_common.c | 202 ++
> .../marvell/octeontx2/otx2_cpt_reqmgr.h | 197 ++
> drivers/crypto/marvell/octeontx2/otx2_cptlf.c | 429 ++++
> drivers/crypto/marvell/octeontx2/otx2_cptlf.h | 353 ++++
> drivers/crypto/marvell/octeontx2/otx2_cptpf.h | 61 +
> .../marvell/octeontx2/otx2_cptpf_main.c | 713 +++++++
> .../marvell/octeontx2/otx2_cptpf_mbox.c | 356 ++++
> .../marvell/octeontx2/otx2_cptpf_ucode.c | 1415 +++++++++++++
> .../marvell/octeontx2/otx2_cptpf_ucode.h | 162 ++
> drivers/crypto/marvell/octeontx2/otx2_cptvf.h | 29 +
> .../marvell/octeontx2/otx2_cptvf_algs.c | 1758 +++++++++++++++++
> .../marvell/octeontx2/otx2_cptvf_algs.h | 178 ++
> .../marvell/octeontx2/otx2_cptvf_main.c | 410 ++++
> .../marvell/octeontx2/otx2_cptvf_mbox.c | 167 ++
> .../marvell/octeontx2/otx2_cptvf_reqmgr.c | 541 +++++
> 20 files changed, 7597 insertions(+)
> create mode 100644 drivers/crypto/marvell/octeontx2/Makefile
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_hw_types.h
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_mbox_common.c
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_reqmgr.h
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptlf.c
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptlf.h
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf.h
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.h
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf.h
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.c
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_algs.h
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_main.c
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_mbox.c
> create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptvf_reqmgr.c
All applied. Thanks.
--
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt