2020-11-09 12:11:12

by Srujana Challa

[permalink] [raw]
Subject: [PATCH v9,net-next,05/12] crypto: octeontx2: add mailbox communication with AF

In the resource virtualization unit (RVU) each of the PF and AF
(admin function) share a 64KB of reserved memory region for
communication. This patch initializes PF <=> AF mailbox IRQs,
registers handlers for processing these communication messages.

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 | 3 +-
.../marvell/octeontx2/otx2_cpt_common.h | 4 +
.../marvell/octeontx2/otx2_cpt_mbox_common.c | 37 +++++++
drivers/crypto/marvell/octeontx2/otx2_cptpf.h | 12 +++
.../marvell/octeontx2/otx2_cptpf_main.c | 102 +++++++++++++++++-
.../marvell/octeontx2/otx2_cptpf_mbox.c | 80 ++++++++++++++
6 files changed, 236 insertions(+), 2 deletions(-)
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cpt_mbox_common.c
create mode 100644 drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c

diff --git a/drivers/crypto/marvell/octeontx2/Makefile b/drivers/crypto/marvell/octeontx2/Makefile
index db763ad46a91..8c8262e94f78 100644
--- a/drivers/crypto/marvell/octeontx2/Makefile
+++ b/drivers/crypto/marvell/octeontx2/Makefile
@@ -1,6 +1,7 @@
# SPDX-License-Identifier: GPL-2.0-only
obj-$(CONFIG_CRYPTO_DEV_OCTEONTX2_CPT) += octeontx2-cpt.o

-octeontx2-cpt-objs := otx2_cptpf_main.o
+octeontx2-cpt-objs := otx2_cptpf_main.o otx2_cptpf_mbox.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 eff4ffa58dc4..b677f8c7e724 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_common.h
@@ -12,6 +12,7 @@
#include <linux/crypto.h>
#include "otx2_cpt_hw_types.h"
#include "rvu.h"
+#include "mbox.h"

#define OTX2_CPT_RVU_FUNC_ADDR_S(blk, slot, offs) \
(((blk) << 20) | ((slot) << 12) | (offs))
@@ -29,4 +30,7 @@ static inline u64 otx2_cpt_read64(void __iomem *reg_base, u64 blk, u64 slot,
return readq_relaxed(reg_base +
OTX2_CPT_RVU_FUNC_ADDR_S(blk, slot, offs));
}
+
+int otx2_cpt_send_ready_msg(struct otx2_mbox *mbox, struct pci_dev *pdev);
+int otx2_cpt_send_mbox_msg(struct otx2_mbox *mbox, struct pci_dev *pdev);
#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
new file mode 100644
index 000000000000..a122483b5976
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cpt_mbox_common.c
@@ -0,0 +1,37 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (C) 2020 Marvell. */
+
+#include "otx2_cpt_common.h"
+
+int otx2_cpt_send_mbox_msg(struct otx2_mbox *mbox, struct pci_dev *pdev)
+{
+ int ret;
+
+ otx2_mbox_msg_send(mbox, 0);
+ ret = otx2_mbox_wait_for_rsp(mbox, 0);
+ if (ret == -EIO) {
+ dev_err(&pdev->dev, "RVU MBOX timeout.\n");
+ return ret;
+ } else if (ret) {
+ dev_err(&pdev->dev, "RVU MBOX error: %d.\n", ret);
+ return -EFAULT;
+ }
+ return ret;
+}
+
+int otx2_cpt_send_ready_msg(struct otx2_mbox *mbox, struct pci_dev *pdev)
+{
+ struct mbox_msghdr *req;
+
+ req = otx2_mbox_alloc_msg_rsp(mbox, 0, sizeof(*req),
+ sizeof(struct ready_msg_rsp));
+ if (req == NULL) {
+ dev_err(&pdev->dev, "RVU MBOX failed to get message.\n");
+ return -EFAULT;
+ }
+ req->id = MBOX_MSG_READY;
+ req->sig = OTX2_MBOX_REQ_SIG;
+ req->pcifunc = 0;
+
+ return otx2_cpt_send_mbox_msg(mbox, pdev);
+}
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf.h b/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
index 84cdc8cc2c15..87fe4c6838e5 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf.h
@@ -5,9 +5,21 @@
#ifndef __OTX2_CPTPF_H
#define __OTX2_CPTPF_H

+#include "otx2_cpt_common.h"
+
struct otx2_cptpf_dev {
void __iomem *reg_base; /* CPT PF registers start address */
+ void __iomem *afpf_mbox_base; /* PF-AF mbox start address */
struct pci_dev *pdev; /* PCI device handle */
+ /* AF <=> PF mbox */
+ struct otx2_mbox afpf_mbox;
+ struct work_struct afpf_mbox_work;
+ struct workqueue_struct *afpf_mbox_wq;
+
+ u8 pf_id; /* RVU PF number */
};

+irqreturn_t otx2_cptpf_afpf_mbox_intr(int irq, void *arg);
+void otx2_cptpf_afpf_mbox_handler(struct work_struct *work);
+
#endif /* __OTX2_CPTPF_H */
diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
index 47781966de3e..2f5bf02436da 100644
--- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_main.c
@@ -10,6 +10,75 @@
#define OTX2_CPT_DRV_NAME "octeontx2-cpt"
#define OTX2_CPT_DRV_STRING "Marvell OcteonTX2 CPT Physical Function Driver"

+static void cptpf_disable_afpf_mbox_intr(struct otx2_cptpf_dev *cptpf)
+{
+ /* Disable AF-PF interrupt */
+ otx2_cpt_write64(cptpf->reg_base, BLKADDR_RVUM, 0, RVU_PF_INT_ENA_W1C,
+ 0x1ULL);
+ /* Clear interrupt if any */
+ otx2_cpt_write64(cptpf->reg_base, BLKADDR_RVUM, 0, RVU_PF_INT, 0x1ULL);
+}
+
+static int cptpf_register_afpf_mbox_intr(struct otx2_cptpf_dev *cptpf)
+{
+ struct pci_dev *pdev = cptpf->pdev;
+ struct device *dev = &pdev->dev;
+ int ret, irq;
+
+ irq = pci_irq_vector(pdev, RVU_PF_INT_VEC_AFPF_MBOX);
+ /* Register AF-PF mailbox interrupt handler */
+ ret = devm_request_irq(dev, irq, otx2_cptpf_afpf_mbox_intr, 0,
+ "CPTAFPF Mbox", cptpf);
+ if (ret) {
+ dev_err(dev,
+ "IRQ registration failed for PFAF mbox irq\n");
+ return ret;
+ }
+ /* Clear interrupt if any, to avoid spurious interrupts */
+ otx2_cpt_write64(cptpf->reg_base, BLKADDR_RVUM, 0, RVU_PF_INT, 0x1ULL);
+ /* Enable AF-PF interrupt */
+ otx2_cpt_write64(cptpf->reg_base, BLKADDR_RVUM, 0, RVU_PF_INT_ENA_W1S,
+ 0x1ULL);
+
+ ret = otx2_cpt_send_ready_msg(&cptpf->afpf_mbox, cptpf->pdev);
+ if (ret) {
+ dev_warn(dev,
+ "AF not responding to mailbox, deferring probe\n");
+ cptpf_disable_afpf_mbox_intr(cptpf);
+ return -EPROBE_DEFER;
+ }
+ return 0;
+}
+
+static int cptpf_afpf_mbox_init(struct otx2_cptpf_dev *cptpf)
+{
+ int err;
+
+ cptpf->afpf_mbox_wq = alloc_workqueue("cpt_afpf_mailbox",
+ WQ_UNBOUND | WQ_HIGHPRI |
+ WQ_MEM_RECLAIM, 1);
+ if (!cptpf->afpf_mbox_wq)
+ return -ENOMEM;
+
+ err = otx2_mbox_init(&cptpf->afpf_mbox, cptpf->afpf_mbox_base,
+ cptpf->pdev, cptpf->reg_base, MBOX_DIR_PFAF, 1);
+ if (err)
+ goto error;
+
+ INIT_WORK(&cptpf->afpf_mbox_work, otx2_cptpf_afpf_mbox_handler);
+ return 0;
+
+error:
+ destroy_workqueue(cptpf->afpf_mbox_wq);
+ return err;
+}
+
+static void cptpf_afpf_mbox_destroy(struct otx2_cptpf_dev *cptpf)
+{
+ destroy_workqueue(cptpf->afpf_mbox_wq);
+ otx2_mbox_destroy(&cptpf->afpf_mbox);
+}
+
static int cpt_is_pf_usable(struct otx2_cptpf_dev *cptpf)
{
u64 rev;
@@ -33,6 +102,7 @@ static int otx2_cptpf_probe(struct pci_dev *pdev,
const struct pci_device_id *ent)
{
struct device *dev = &pdev->dev;
+ resource_size_t offset, size;
struct otx2_cptpf_dev *cptpf;
int err;

@@ -69,8 +139,35 @@ static int otx2_cptpf_probe(struct pci_dev *pdev,
if (err)
goto clear_drvdata;

+ offset = pci_resource_start(pdev, PCI_MBOX_BAR_NUM);
+ size = pci_resource_len(pdev, PCI_MBOX_BAR_NUM);
+ /* Map AF-PF mailbox memory */
+ cptpf->afpf_mbox_base = devm_ioremap_wc(dev, offset, size);
+ if (!cptpf->afpf_mbox_base) {
+ dev_err(&pdev->dev, "Unable to map BAR4\n");
+ err = -ENODEV;
+ goto clear_drvdata;
+ }
+ err = pci_alloc_irq_vectors(pdev, RVU_PF_INT_VEC_CNT,
+ RVU_PF_INT_VEC_CNT, PCI_IRQ_MSIX);
+ if (err < 0) {
+ dev_err(dev, "Request for %d msix vectors failed\n",
+ RVU_PF_INT_VEC_CNT);
+ goto clear_drvdata;
+ }
+ /* Initialize AF-PF mailbox */
+ err = cptpf_afpf_mbox_init(cptpf);
+ if (err)
+ goto clear_drvdata;
+ /* Register mailbox interrupt */
+ err = cptpf_register_afpf_mbox_intr(cptpf);
+ if (err)
+ goto destroy_afpf_mbox;
+
return 0;

+destroy_afpf_mbox:
+ cptpf_afpf_mbox_destroy(cptpf);
clear_drvdata:
pci_set_drvdata(pdev, NULL);
return err;
@@ -82,7 +179,10 @@ static void otx2_cptpf_remove(struct pci_dev *pdev)

if (!cptpf)
return;
-
+ /* Disable AF-PF mailbox interrupt */
+ cptpf_disable_afpf_mbox_intr(cptpf);
+ /* Destroy AF-PF mbox */
+ cptpf_afpf_mbox_destroy(cptpf);
pci_set_drvdata(pdev, NULL);
}

diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
new file mode 100644
index 000000000000..0a8bd46b5686
--- /dev/null
+++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_mbox.c
@@ -0,0 +1,80 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Copyright (C) 2020 Marvell. */
+
+#include "otx2_cpt_common.h"
+#include "otx2_cptpf.h"
+#include "rvu_reg.h"
+
+irqreturn_t otx2_cptpf_afpf_mbox_intr(int __always_unused irq, void *arg)
+{
+ struct otx2_cptpf_dev *cptpf = arg;
+ u64 intr;
+
+ /* Read the interrupt bits */
+ intr = otx2_cpt_read64(cptpf->reg_base, BLKADDR_RVUM, 0, RVU_PF_INT);
+
+ if (intr & 0x1ULL) {
+ /* Schedule work queue function to process the MBOX request */
+ queue_work(cptpf->afpf_mbox_wq, &cptpf->afpf_mbox_work);
+ /* Clear and ack the interrupt */
+ otx2_cpt_write64(cptpf->reg_base, BLKADDR_RVUM, 0, RVU_PF_INT,
+ 0x1ULL);
+ }
+ return IRQ_HANDLED;
+}
+
+static void process_afpf_mbox_msg(struct otx2_cptpf_dev *cptpf,
+ struct mbox_msghdr *msg)
+{
+ struct device *dev = &cptpf->pdev->dev;
+
+ if (msg->id >= MBOX_MSG_MAX) {
+ dev_err(dev, "MBOX msg with unknown ID %d\n", msg->id);
+ return;
+ }
+ if (msg->sig != OTX2_MBOX_RSP_SIG) {
+ dev_err(dev, "MBOX msg with wrong signature %x, ID %d\n",
+ msg->sig, msg->id);
+ return;
+ }
+
+ switch (msg->id) {
+ case MBOX_MSG_READY:
+ cptpf->pf_id = (msg->pcifunc >> RVU_PFVF_PF_SHIFT) &
+ RVU_PFVF_PF_MASK;
+ break;
+ default:
+ dev_err(dev,
+ "Unsupported msg %d received.\n", msg->id);
+ break;
+ }
+}
+
+/* Handle mailbox messages received from AF */
+void otx2_cptpf_afpf_mbox_handler(struct work_struct *work)
+{
+ struct otx2_cptpf_dev *cptpf;
+ struct otx2_mbox *afpf_mbox;
+ struct otx2_mbox_dev *mdev;
+ struct mbox_hdr *rsp_hdr;
+ struct mbox_msghdr *msg;
+ int offset, i;
+
+ cptpf = container_of(work, struct otx2_cptpf_dev, afpf_mbox_work);
+ afpf_mbox = &cptpf->afpf_mbox;
+ mdev = &afpf_mbox->dev[0];
+ /* Sync mbox data into memory */
+ smp_wmb();
+
+ rsp_hdr = (struct mbox_hdr *)(mdev->mbase + afpf_mbox->rx_start);
+ offset = ALIGN(sizeof(*rsp_hdr), MBOX_MSG_ALIGN);
+
+ for (i = 0; i < rsp_hdr->num_msgs; i++) {
+ msg = (struct mbox_msghdr *)(mdev->mbase + afpf_mbox->rx_start +
+ offset);
+ process_afpf_mbox_msg(cptpf, msg);
+ offset = msg->next_msgoff;
+ mdev->msgs_acked++;
+ }
+ otx2_mbox_reset(afpf_mbox, 0);
+}
--
2.28.0


2020-11-12 01:35:22

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [PATCH v9,net-next,05/12] crypto: octeontx2: add mailbox communication with AF

On Mon, 9 Nov 2020 17:39:17 +0530 Srujana Challa wrote:
> + err = pci_alloc_irq_vectors(pdev, RVU_PF_INT_VEC_CNT,
> + RVU_PF_INT_VEC_CNT, PCI_IRQ_MSIX);

I don't see any pci_free_irq_vectors() in this patch

2020-11-12 07:12:04

by Srujana Challa

[permalink] [raw]
Subject: RE: [EXT] Re: [PATCH v9,net-next,05/12] crypto: octeontx2: add mailbox communication with AF

> On Mon, 9 Nov 2020 17:39:17 +0530 Srujana Challa wrote:
> > + err = pci_alloc_irq_vectors(pdev, RVU_PF_INT_VEC_CNT,
> > + RVU_PF_INT_VEC_CNT, PCI_IRQ_MSIX);
>
> I don't see any pci_free_irq_vectors() in this patch

This will be handled by the devres managed PCI interface.

Copied code for reference:

static void pcim_release(struct device *gendev, void *res)
{
struct pci_dev *dev = to_pci_dev(gendev);
struct pci_devres *this = res;
int i;

if (dev->msi_enabled)
pci_disable_msi(dev);
if (dev->msix_enabled)
pci_disable_msix(dev);

....
}

2020-11-12 15:51:53

by Jakub Kicinski

[permalink] [raw]
Subject: Re: [EXT] Re: [PATCH v9,net-next,05/12] crypto: octeontx2: add mailbox communication with AF

On Thu, 12 Nov 2020 07:07:34 +0000 Srujana Challa wrote:
> > On Mon, 9 Nov 2020 17:39:17 +0530 Srujana Challa wrote:
> > > + err = pci_alloc_irq_vectors(pdev, RVU_PF_INT_VEC_CNT,
> > > + RVU_PF_INT_VEC_CNT, PCI_IRQ_MSIX);
> >
> > I don't see any pci_free_irq_vectors() in this patch
>
> This will be handled by the devres managed PCI interface.

All good by me then, please resend once you get a ack from someone on
the crypto side.