crypto: ccp - Change ISR handler method for a v3 CCP

The CCP has the ability to perform several operations simultaneously,
but only one interrupt.  When implemented as a PCI device and using
MSI-X/MSI interrupts, use a tasklet model to service interrupts. By
disabling and enabling interrupts from the CCP, coupled with the
queuing that tasklets provide, we can ensure that all events
(occurring on the device) are recognized and serviced.

This change fixes a problem wherein 2 or more busy queues can cause
notification bits to change state while a (CCP) interrupt is being
serviced, but after the queue state has been evaluated. This results
in the event being 'lost' and the queue hanging, waiting to be
serviced. Since the status bits are never fully de-asserted, the
CCP never generates another interrupt (all bits zero -> one or more
bits one), and no further CCP operations will be executed.

Cc: <stable@vger.kernel.org> # 4.9.x+
Signed-off-by: Gary R Hook <gary.hook@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

diff --git a/drivers/crypto/ccp/ccp-dev-v3.c b/drivers/crypto/ccp/ccp-dev-v3.c
index a3689a66cb1d7d32083ca2ba98870e94d5714732..367c2e30656fc3680020dbf50c881cfbf2f43614 100644 (file)
--- a/drivers/crypto/ccp/ccp-dev-v3.c
+++ b/drivers/crypto/ccp/ccp-dev-v3.c
@@ -315,17 +315,73 @@ static int ccp_perform_ecc(struct ccp_op *op)
        return ccp_do_cmd(op, cr, ARRAY_SIZE(cr));
 }
 
+static void ccp_disable_queue_interrupts(struct ccp_device *ccp)
+{
+       iowrite32(0x00, ccp->io_regs + IRQ_MASK_REG);
+}
+
+static void ccp_enable_queue_interrupts(struct ccp_device *ccp)
+{
+       iowrite32(ccp->qim, ccp->io_regs + IRQ_MASK_REG);
+}
+
+static void ccp_irq_bh(unsigned long data)
+{
+       struct ccp_device *ccp = (struct ccp_device *)data;
+       struct ccp_cmd_queue *cmd_q;
+       u32 q_int, status;
+       unsigned int i;
+
+       status = ioread32(ccp->io_regs + IRQ_STATUS_REG);
+
+       for (i = 0; i < ccp->cmd_q_count; i++) {
+               cmd_q = &ccp->cmd_q[i];
+
+               q_int = status & (cmd_q->int_ok | cmd_q->int_err);
+               if (q_int) {
+                       cmd_q->int_status = status;
+                       cmd_q->q_status = ioread32(cmd_q->reg_status);
+                       cmd_q->q_int_status = ioread32(cmd_q->reg_int_status);
+
+                       /* On error, only save the first error value */
+                       if ((q_int & cmd_q->int_err) && !cmd_q->cmd_error)
+                               cmd_q->cmd_error = CMD_Q_ERROR(cmd_q->q_status);
+
+                       cmd_q->int_rcvd = 1;
+
+                       /* Acknowledge the interrupt and wake the kthread */
+                       iowrite32(q_int, ccp->io_regs + IRQ_STATUS_REG);
+                       wake_up_interruptible(&cmd_q->int_queue);
+               }
+       }
+       ccp_enable_queue_interrupts(ccp);
+}
+
+static irqreturn_t ccp_irq_handler(int irq, void *data)
+{
+       struct device *dev = data;
+       struct ccp_device *ccp = dev_get_drvdata(dev);
+
+       ccp_disable_queue_interrupts(ccp);
+       if (ccp->use_tasklet)
+               tasklet_schedule(&ccp->irq_tasklet);
+       else
+               ccp_irq_bh((unsigned long)ccp);
+
+       return IRQ_HANDLED;
+}
+
 static int ccp_init(struct ccp_device *ccp)
 {
        struct device *dev = ccp->dev;
        struct ccp_cmd_queue *cmd_q;
        struct dma_pool *dma_pool;
        char dma_pool_name[MAX_DMAPOOL_NAME_LEN];
-       unsigned int qmr, qim, i;
+       unsigned int qmr, i;
        int ret;
 
        /* Find available queues */
-       qim = 0;
+       ccp->qim = 0;
        qmr = ioread32(ccp->io_regs + Q_MASK_REG);
        for (i = 0; i < MAX_HW_QUEUES; i++) {
                if (!(qmr & (1 << i)))
@@ -370,7 +426,7 @@ static int ccp_init(struct ccp_device *ccp)
                init_waitqueue_head(&cmd_q->int_queue);
 
                /* Build queue interrupt mask (two interrupts per queue) */
-               qim |= cmd_q->int_ok | cmd_q->int_err;
+               ccp->qim |= cmd_q->int_ok | cmd_q->int_err;
 
 #ifdef CONFIG_ARM64
                /* For arm64 set the recommended queue cache settings */
@@ -388,14 +444,14 @@ static int ccp_init(struct ccp_device *ccp)
        dev_notice(dev, "%u command queues available\n", ccp->cmd_q_count);
 
        /* Disable and clear interrupts until ready */
-       iowrite32(0x00, ccp->io_regs + IRQ_MASK_REG);
+       ccp_disable_queue_interrupts(ccp);
        for (i = 0; i < ccp->cmd_q_count; i++) {
                cmd_q = &ccp->cmd_q[i];
 
                ioread32(cmd_q->reg_int_status);
                ioread32(cmd_q->reg_status);
        }
-       iowrite32(qim, ccp->io_regs + IRQ_STATUS_REG);
+       iowrite32(ccp->qim, ccp->io_regs + IRQ_STATUS_REG);
 
        /* Request an irq */
        ret = ccp->get_irq(ccp);
@@ -404,6 +460,11 @@ static int ccp_init(struct ccp_device *ccp)
                goto e_pool;
        }
 
+       /* Initialize the ISR tasklet? */
+       if (ccp->use_tasklet)
+               tasklet_init(&ccp->irq_tasklet, ccp_irq_bh,
+                            (unsigned long)ccp);
+
        dev_dbg(dev, "Starting threads...\n");
        /* Create a kthread for each queue */
        for (i = 0; i < ccp->cmd_q_count; i++) {
@@ -426,7 +487,7 @@ static int ccp_init(struct ccp_device *ccp)
 
        dev_dbg(dev, "Enabling interrupts...\n");
        /* Enable interrupts */
-       iowrite32(qim, ccp->io_regs + IRQ_MASK_REG);
+       ccp_enable_queue_interrupts(ccp);
 
        dev_dbg(dev, "Registering device...\n");
        ccp_add_device(ccp);
@@ -463,7 +524,7 @@ static void ccp_destroy(struct ccp_device *ccp)
 {
        struct ccp_cmd_queue *cmd_q;
        struct ccp_cmd *cmd;
-       unsigned int qim, i;
+       unsigned int i;
 
        /* Unregister the DMA engine */
        ccp_dmaengine_unregister(ccp);
@@ -474,22 +535,15 @@ static void ccp_destroy(struct ccp_device *ccp)
        /* Remove this device from the list of available units */
        ccp_del_device(ccp);
 
-       /* Build queue interrupt mask (two interrupt masks per queue) */
-       qim = 0;
-       for (i = 0; i < ccp->cmd_q_count; i++) {
-               cmd_q = &ccp->cmd_q[i];
-               qim |= cmd_q->int_ok | cmd_q->int_err;
-       }
-
        /* Disable and clear interrupts */
-       iowrite32(0x00, ccp->io_regs + IRQ_MASK_REG);
+       ccp_disable_queue_interrupts(ccp);
        for (i = 0; i < ccp->cmd_q_count; i++) {
                cmd_q = &ccp->cmd_q[i];
 
                ioread32(cmd_q->reg_int_status);
                ioread32(cmd_q->reg_status);
        }
-       iowrite32(qim, ccp->io_regs + IRQ_STATUS_REG);
+       iowrite32(ccp->qim, ccp->io_regs + IRQ_STATUS_REG);
 
        /* Stop the queue kthreads */
        for (i = 0; i < ccp->cmd_q_count; i++)
@@ -516,40 +570,6 @@ static void ccp_destroy(struct ccp_device *ccp)
        }
 }
 
-static irqreturn_t ccp_irq_handler(int irq, void *data)
-{
-       struct device *dev = data;
-       struct ccp_device *ccp = dev_get_drvdata(dev);
-       struct ccp_cmd_queue *cmd_q;
-       u32 q_int, status;
-       unsigned int i;
-
-       status = ioread32(ccp->io_regs + IRQ_STATUS_REG);
-
-       for (i = 0; i < ccp->cmd_q_count; i++) {
-               cmd_q = &ccp->cmd_q[i];
-
-               q_int = status & (cmd_q->int_ok | cmd_q->int_err);
-               if (q_int) {
-                       cmd_q->int_status = status;
-                       cmd_q->q_status = ioread32(cmd_q->reg_status);
-                       cmd_q->q_int_status = ioread32(cmd_q->reg_int_status);
-
-                       /* On error, only save the first error value */
-                       if ((q_int & cmd_q->int_err) && !cmd_q->cmd_error)
-                               cmd_q->cmd_error = CMD_Q_ERROR(cmd_q->q_status);
-
-                       cmd_q->int_rcvd = 1;
-
-                       /* Acknowledge the interrupt and wake the kthread */
-                       iowrite32(q_int, ccp->io_regs + IRQ_STATUS_REG);
-                       wake_up_interruptible(&cmd_q->int_queue);
-               }
-       }
-
-       return IRQ_HANDLED;
-}
-
 static const struct ccp_actions ccp3_actions = {
        .aes = ccp_perform_aes,
        .xts_aes = ccp_perform_xts_aes,

diff --git a/drivers/crypto/ccp/ccp-dev.h b/drivers/crypto/ccp/ccp-dev.h
index 2dfec019a832e27684b5f548d5dc2c1cf780a28a..0cb09d0feeaf0bc748b87eebbb522bdf3e3d66b5 100644 (file)
--- a/drivers/crypto/ccp/ccp-dev.h
+++ b/drivers/crypto/ccp/ccp-dev.h
@@ -339,7 +339,10 @@ struct ccp_device {
        void *dev_specific;
        int (*get_irq)(struct ccp_device *ccp);
        void (*free_irq)(struct ccp_device *ccp);
+       unsigned int qim;
        unsigned int irq;
+       bool use_tasklet;
+       struct tasklet_struct irq_tasklet;
 
        /* I/O area used for device communication. The register mapping
         * starts at an offset into the mapped bar.

diff --git a/drivers/crypto/ccp/ccp-pci.c b/drivers/crypto/ccp/ccp-pci.c
index 28a9996c1085624cc325bbdec39c67d5000eb519..e880d4cf4ada867d96e01ad8784c71fd4b58aac7 100644 (file)
--- a/drivers/crypto/ccp/ccp-pci.c
+++ b/drivers/crypto/ccp/ccp-pci.c
@@ -69,6 +69,7 @@ static int ccp_get_msix_irqs(struct ccp_device *ccp)
                        goto e_irq;
                }
        }
+       ccp->use_tasklet = true;
 
        return 0;
 
@@ -100,6 +101,7 @@ static int ccp_get_msi_irq(struct ccp_device *ccp)
                dev_notice(dev, "unable to allocate MSI IRQ (%d)\n", ret);
                goto e_msi;
        }
+       ccp->use_tasklet = true;
 
        return 0;