Re: [PATCH v4 05/18] nitro_enclaves: Handle PCI device command requests

[Alexander Graf <graf@amazon.de>]

On 22.06.20 22:03, Andra Paraschiv wrote:
> The Nitro Enclaves PCI device exposes a MMIO space that this driver
> uses to submit command requests and to receive command replies e.g. for
> enclave creation / termination or setting enclave resources.
> 
> Add logic for handling PCI device command requests based on the given
> command type.
> 
> Register an MSI-X interrupt vector for command reply notifications to
> handle this type of communication events.
> 
> Signed-off-by: Alexandru-Catalin Vasile <lexnv@amazon.com>
> Signed-off-by: Andra Paraschiv <andraprs@amazon.com>
> 
> Fix issue reported in:
> https://lore.kernel.org/lkml/202004231644.xTmN4Z1z%25lkp@intel.com/
> 
> Reported-by: kbuild test robot <lkp@intel.com>
> Signed-off-by: Andra Paraschiv <andraprs@amazon.com>
> ---
> Changelog
> 
> v3 -> v4
> 
> * Use dev_err instead of custom NE log pattern.
> * Return IRQ_NONE when interrupts are not handled.
> 
> v2 -> v3
> 
> * Remove the WARN_ON calls.
> * Update static calls sanity checks.
> * Remove "ratelimited" from the logs that are not in the ioctl call
>    paths.
> 
> v1 -> v2
> 
> * Add log pattern for NE.
> * Remove the BUG_ON calls.
> * Update goto labels to match their purpose.
> * Add fix for kbuild report.
> ---
>   drivers/virt/nitro_enclaves/ne_pci_dev.c | 232 +++++++++++++++++++++++
>   1 file changed, 232 insertions(+)
> 
> diff --git a/drivers/virt/nitro_enclaves/ne_pci_dev.c b/drivers/virt/nitro_enclaves/ne_pci_dev.c
> index 235fa3ecbee2..c24230cfe7c0 100644
> --- a/drivers/virt/nitro_enclaves/ne_pci_dev.c
> +++ b/drivers/virt/nitro_enclaves/ne_pci_dev.c
> @@ -27,6 +27,218 @@ static const struct pci_device_id ne_pci_ids[] = {
>   
>   MODULE_DEVICE_TABLE(pci, ne_pci_ids);
>   
> +/**
> + * ne_submit_request - Submit command request to the PCI device based on the
> + * command type.
> + *
> + * This function gets called with the ne_pci_dev mutex held.
> + *
> + * @pdev: PCI device to send the command to.
> + * @cmd_type: command type of the request sent to the PCI device.
> + * @cmd_request: command request payload.
> + * @cmd_request_size: size of the command request payload.
> + *
> + * @returns: 0 on success, negative return value on failure.
> + */
> +static int ne_submit_request(struct pci_dev *pdev,
> +			     enum ne_pci_dev_cmd_type cmd_type,
> +			     void *cmd_request, size_t cmd_request_size)
> +{
> +	struct ne_pci_dev *ne_pci_dev = pci_get_drvdata(pdev);
> +
> +	if (!ne_pci_dev || !ne_pci_dev->iomem_base)
> +		return -EINVAL;

How can this ever happen?

> +
> +	memcpy_toio(ne_pci_dev->iomem_base + NE_SEND_DATA, cmd_request,
> +		    cmd_request_size);
> +
> +	iowrite32(cmd_type, ne_pci_dev->iomem_base + NE_COMMAND);
> +
> +	return 0;
> +}
> +
> +/**
> + * ne_retrieve_reply - Retrieve reply from the PCI device.
> + *
> + * This function gets called with the ne_pci_dev mutex held.
> + *
> + * @pdev: PCI device to receive the reply from.
> + * @cmd_reply: command reply payload.
> + * @cmd_reply_size: size of the command reply payload.
> + *
> + * @returns: 0 on success, negative return value on failure.
> + */
> +static int ne_retrieve_reply(struct pci_dev *pdev,
> +			     struct ne_pci_dev_cmd_reply *cmd_reply,
> +			     size_t cmd_reply_size)
> +{
> +	struct ne_pci_dev *ne_pci_dev = pci_get_drvdata(pdev);
> +
> +	if (!ne_pci_dev || !ne_pci_dev->iomem_base)
> +		return -EINVAL;

Same.

> +
> +	memcpy_fromio(cmd_reply, ne_pci_dev->iomem_base + NE_RECV_DATA,
> +		      cmd_reply_size);
> +
> +	return 0;
> +}
> +
> +/**
> + * ne_wait_for_reply - Wait for a reply of a PCI command.
> + *
> + * This function gets called with the ne_pci_dev mutex held.
> + *
> + * @pdev: PCI device for which a reply is waited.
> + *
> + * @returns: 0 on success, negative return value on failure.
> + */
> +static int ne_wait_for_reply(struct pci_dev *pdev)
> +{
> +	struct ne_pci_dev *ne_pci_dev = pci_get_drvdata(pdev);
> +	int rc = -EINVAL;

Unused assignment?

> +
> +	if (!ne_pci_dev)
> +		return -EINVAL;

Same.

> +
> +	/*
> +	 * TODO: Update to _interruptible and handle interrupted wait event
> +	 * e.g. -ERESTARTSYS, incoming signals + add / update timeout.
> +	 */
> +	rc = wait_event_timeout(ne_pci_dev->cmd_reply_wait_q,
> +				atomic_read(&ne_pci_dev->cmd_reply_avail) != 0,
> +				msecs_to_jiffies(NE_DEFAULT_TIMEOUT_MSECS));
> +	if (!rc)
> +		return -ETIMEDOUT;
> +
> +	return 0;
> +}
> +
> +int ne_do_request(struct pci_dev *pdev, enum ne_pci_dev_cmd_type cmd_type,
> +		  void *cmd_request, size_t cmd_request_size,
> +		  struct ne_pci_dev_cmd_reply *cmd_reply, size_t cmd_reply_size)
> +{
> +	struct ne_pci_dev *ne_pci_dev = NULL;
> +	int rc = -EINVAL;
> +
> +	if (!pdev)
> +		return -ENODEV;

When can this happen?

> +
> +	ne_pci_dev = pci_get_drvdata(pdev);
> +	if (!ne_pci_dev || !ne_pci_dev->iomem_base)
> +		return -EINVAL;

Same

> +
> +	if (cmd_type <= INVALID_CMD || cmd_type >= MAX_CMD) {
> +		dev_err_ratelimited(&pdev->dev, "Invalid cmd type=%u\n",
> +				    cmd_type);
> +
> +		return -EINVAL;
> +	}
> +
> +	if (!cmd_request) {
> +		dev_err_ratelimited(&pdev->dev, "Null cmd request\n");
> +
> +		return -EINVAL;
> +	}
> +
> +	if (cmd_request_size > NE_SEND_DATA_SIZE) {
> +		dev_err_ratelimited(&pdev->dev,
> +				    "Invalid req size=%zu for cmd type=%u\n",
> +				    cmd_request_size, cmd_type);
> +
> +		return -EINVAL;
> +	}
> +
> +	if (!cmd_reply) {
> +		dev_err_ratelimited(&pdev->dev, "Null cmd reply\n");
> +
> +		return -EINVAL;
> +	}
> +
> +	if (cmd_reply_size > NE_RECV_DATA_SIZE) {
> +		dev_err_ratelimited(&pdev->dev, "Invalid reply size=%zu\n",
> +				    cmd_reply_size);
> +
> +		return -EINVAL;
> +	}
> +
> +	/*
> +	 * Use this mutex so that the PCI device handles one command request at
> +	 * a time.
> +	 */
> +	mutex_lock(&ne_pci_dev->pci_dev_mutex);
> +
> +	atomic_set(&ne_pci_dev->cmd_reply_avail, 0);
> +
> +	rc = ne_submit_request(pdev, cmd_type, cmd_request, cmd_request_size);
> +	if (rc < 0) {
> +		dev_err_ratelimited(&pdev->dev,
> +				    "Error in submit request [rc=%d]\n", rc);
> +
> +		goto unlock_mutex;
> +	}
> +
> +	rc = ne_wait_for_reply(pdev);
> +	if (rc < 0) {
> +		dev_err_ratelimited(&pdev->dev,
> +				    "Error in wait for reply [rc=%d]\n", rc);
> +
> +		goto unlock_mutex;
> +	}
> +
> +	rc = ne_retrieve_reply(pdev, cmd_reply, cmd_reply_size);
> +	if (rc < 0) {
> +		dev_err_ratelimited(&pdev->dev,
> +				    "Error in retrieve reply [rc=%d]\n", rc);
> +
> +		goto unlock_mutex;
> +	}
> +
> +	atomic_set(&ne_pci_dev->cmd_reply_avail, 0);
> +
> +	if (cmd_reply->rc < 0) {
> +		dev_err_ratelimited(&pdev->dev,
> +				    "Error in cmd process logic [rc=%d]\n",
> +				    cmd_reply->rc);
> +
> +		rc = cmd_reply->rc;
> +
> +		goto unlock_mutex;
> +	}
> +
> +	mutex_unlock(&ne_pci_dev->pci_dev_mutex);
> +
> +	return 0;

Can you just set rc to 0 and fall through?

> +
> +unlock_mutex:
> +	mutex_unlock(&ne_pci_dev->pci_dev_mutex);
> +
> +	return rc;
> +}
> +
> +/**
> + * ne_reply_handler - Interrupt handler for retrieving a reply matching
> + * a request sent to the PCI device for enclave lifetime management.
> + *
> + * @irq: received interrupt for a reply sent by the PCI device.
> + * @args: PCI device private data structure.
> + *
> + * @returns: IRQ_HANDLED on handled interrupt, IRQ_NONE otherwise.
> + */
> +static irqreturn_t ne_reply_handler(int irq, void *args)
> +{
> +	struct ne_pci_dev *ne_pci_dev = (struct ne_pci_dev *)args;
> +
> +	if (!ne_pci_dev)
> +		return IRQ_NONE;

How can this ever happen?


Alex

> +
> +	atomic_set(&ne_pci_dev->cmd_reply_avail, 1);
> +
> +	/* TODO: Update to _interruptible. */
> +	wake_up(&ne_pci_dev->cmd_reply_wait_q);
> +
> +	return IRQ_HANDLED;
> +}
> +
>   /**
>    * ne_setup_msix - Setup MSI-X vectors for the PCI device.
>    *
> @@ -59,7 +271,25 @@ static int ne_setup_msix(struct pci_dev *pdev)
>   		return rc;
>   	}
>   
> +	/*
> +	 * This IRQ gets triggered every time the PCI device responds to a
> +	 * command request. The reply is then retrieved, reading from the MMIO
> +	 * space of the PCI device.
> +	 */
> +	rc = request_irq(pci_irq_vector(pdev, NE_VEC_REPLY),
> +			 ne_reply_handler, 0, "enclave_cmd", ne_pci_dev);
> +	if (rc < 0) {
> +		dev_err(&pdev->dev, "Error in request irq reply [rc=%d]\n", rc);
> +
> +		goto free_irq_vectors;
> +	}
> +
>   	return 0;
> +
> +free_irq_vectors:
> +	pci_free_irq_vectors(pdev);
> +
> +	return rc;
>   }
>   
>   /**
> @@ -74,6 +304,8 @@ static void ne_teardown_msix(struct pci_dev *pdev)
>   	if (!ne_pci_dev)
>   		return;
>   
> +	free_irq(pci_irq_vector(pdev, NE_VEC_REPLY), ne_pci_dev);
> +
>   	pci_free_irq_vectors(pdev);
>   }
>   
> 



Amazon Development Center Germany GmbH
Krausenstr. 38
10117 Berlin
Geschaeftsfuehrung: Christian Schlaeger, Jonathan Weiss
Eingetragen am Amtsgericht Charlottenburg unter HRB 149173 B
Sitz: Berlin
Ust-ID: DE 289 237 879