Date:   Fri, 23 Feb 2018 18:07:54 -0600
From:   Bjorn Helgaas <helgaas@...nel.org>
To:     Oza Pawandeep <poza@...eaurora.org>
Cc:     Bjorn Helgaas <bhelgaas@...gle.com>,
        Philippe Ombredanne <pombredanne@...b.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Kate Stewart <kstewart@...uxfoundation.org>,
        linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
        Dongdong Liu <liudongdong3@...wei.com>,
        Keith Busch <keith.busch@...el.com>, Wei Zhang <wzhang@...com>,
        Sinan Kaya <okaya@...eaurora.org>,
        Timur Tabi <timur@...eaurora.org>
Subject: Re: [PATCH v11 4/7] PCI/DPC: Unify and plumb error handling into DPC

On Fri, Feb 23, 2018 at 01:54:01PM +0530, Oza Pawandeep wrote:
> Current DPC driver does not do recovery, e.g. calling end-point's driver's
> callbacks, which sanitize the sw.
> 
> DPC driver implements link_reset callback, and calls pcie_do_recovery.

s/pcie_do_recovery/pcie_do_recovery()/

> Signed-off-by: Oza Pawandeep <poza@...eaurora.org>
> 
> diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
> index abc514e..f8575da 100644
> --- a/drivers/pci/pci.h
> +++ b/drivers/pci/pci.h
> @@ -343,6 +343,8 @@ static inline resource_size_t pci_resource_alignment(struct pci_dev *dev,
>  void pci_enable_acs(struct pci_dev *dev);
>  
>  /* PCI error reporting and recovery */
> +#define DPC_FATAL	4

This needs to go next to the AER_FATAL, etc., definitions because
DPC_FATAL shares the namespace and they all need to have distinct
values.  I can't tell from this patch whether they do or not.

>  void pcie_do_recovery(struct pci_dev *dev, int severity);
>  
>  #ifdef CONFIG_PCIEASPM
> diff --git a/drivers/pci/pcie/pcie-dpc.c b/drivers/pci/pcie/pcie-dpc.c
> index 38e40c6..5c01c63 100644
> --- a/drivers/pci/pcie/pcie-dpc.c
> +++ b/drivers/pci/pcie/pcie-dpc.c
> @@ -13,6 +13,7 @@
>  #include <linux/pcieport_if.h>
>  #include "../pci.h"
>  #include "aer/aerdrv.h"
> +#include "portdrv.h"
>  
>  struct dpc_dev {
>  	struct pcie_device	*dev;
> @@ -45,6 +46,58 @@ struct dpc_dev {
>  	"Memory Request Completion Timeout",		 /* Bit Position 18 */
>  };
>  
> +static int find_dpc_dev_iter(struct device *device, void *data)
> +{
> +	struct pcie_port_service_driver *service_driver;
> +	struct device **dev = (struct device **) data;;
> +
> +	if (device->bus == &pcie_port_bus_type && device->driver) {
> +		service_driver = to_service_driver(device->driver);
> +		if (service_driver->service == PCIE_PORT_SERVICE_DPC) {
> +			*dev = device;
> +			return 1;
> +		}
> +	}
> +
> +	return 0;
> +}
> +
> +static struct device *pci_find_dpc_dev(struct pci_dev *pdev)
> +{
> +	struct device *dev = NULL;
> +
> +	device_for_each_child(&pdev->dev, &dev, find_dpc_dev_iter);
> +
> +	return dev;
> +}

Ugh.  You're not responsible for this and you don't need to do
anything, but hanging the struct dpc_dev off the struct pcie_device
and then having to grub around like this to locate it from the pci_dev
is just ... clunky.  OK, rant over, sorry :)

> +static int find_dpc_service_iter(struct device *device, void *data)
> +{
> +	struct pcie_port_service_driver *service_driver;
> +	struct pcie_port_service_driver **drv =
> +		(struct pcie_port_service_driver **) data;
> +
> +	if (device->bus == &pcie_port_bus_type && device->driver) {
> +		service_driver = to_service_driver(device->driver);
> +		if (service_driver->service == PCIE_PORT_SERVICE_DPC) {
> +			*drv = service_driver;
> +			return 1;
> +		}
> +	}
> +
> +	return 0;
> +}
> +
> +struct pcie_port_service_driver *pci_find_dpc_service(struct pci_dev *dev)
> +{
> +	struct pcie_port_service_driver *drv = NULL;
> +
> +	device_for_each_child(&dev->dev, &drv, find_dpc_service_iter);
> +
> +	return drv;
> +}
> +EXPORT_SYMBOL_GPL(pci_find_dpc_service);

No module uses this, so it doesn't need to be exported.

This is a clone of find_aer_service().  Can you add a preliminary patch to
make a generic "find service" interface that accepts the service type
(PCIE_PORT_SERVICE_AER, PCIE_PORT_SERVICE_DPC) as a parameter?

This whole "find service" thing is ugly as sin.  You're not responsible for
cleaning it up, but maybe we can at least limit the proliferation of it.

>  static int dpc_wait_rp_inactive(struct dpc_dev *dpc)
>  {
>  	unsigned long timeout = jiffies + HZ;
> @@ -82,12 +135,25 @@ static void dpc_wait_link_inactive(struct dpc_dev *dpc)
>  		dev_warn(dev, "Link state not disabled for DPC event\n");
>  }
>  
> -static void dpc_work(struct work_struct *work)
> +/**
> + * dpc_reset_link - reset link DPC  routine

s/  / / (remove extra space)

> + * @dev: pointer to Root Port's pci_dev data structure
> + *
> + * Invoked by Port Bus driver when performing link reset at Root Port.
> + */
> +static pci_ers_result_t dpc_reset_link(struct pci_dev *pdev)
>  {
> -	struct dpc_dev *dpc = container_of(work, struct dpc_dev, work);
> -	struct pci_dev *dev, *temp, *pdev = dpc->dev->port;
>  	struct pci_bus *parent = pdev->subordinate;
> -	u16 cap = dpc->cap_pos, ctl;
> +	struct pci_dev *dev, *temp;
> +	struct dpc_dev *dpc;
> +	struct pcie_device *pciedev;
> +	struct device *devdpc;
> +	u16 cap, ctl;
> +
> +	devdpc = pci_find_dpc_dev(pdev);
> +	pciedev = to_pcie_device(devdpc);
> +	dpc = get_service_data(pciedev);
> +	cap = dpc->cap_pos;
>  
>  	pci_lock_rescan_remove();
>  	list_for_each_entry_safe_reverse(dev, temp, &parent->devices,
> @@ -104,21 +170,31 @@ static void dpc_work(struct work_struct *work)
>  
>  	dpc_wait_link_inactive(dpc);
>  	if (dpc->rp_extensions && dpc_wait_rp_inactive(dpc))
> -		return;
> +		return PCI_ERS_RESULT_DISCONNECT;
>  	if (dpc->rp_extensions && dpc->rp_pio_status) {
>  		pci_write_config_dword(pdev, cap + PCI_EXP_DPC_RP_PIO_STATUS,
>  				       dpc->rp_pio_status);
>  		dpc->rp_pio_status = 0;
>  	}
>  
> -	pci_write_config_word(pdev, cap + PCI_EXP_DPC_STATUS,
> +	pci_write_config_word(pdev, dpc->cap_pos + PCI_EXP_DPC_STATUS,
>  		PCI_EXP_DPC_STATUS_TRIGGER | PCI_EXP_DPC_STATUS_INTERRUPT);
>  
>  	pci_read_config_word(pdev, cap + PCI_EXP_DPC_CTL, &ctl);
>  	pci_write_config_word(pdev, cap + PCI_EXP_DPC_CTL,
> -			      ctl | PCI_EXP_DPC_CTL_INT_EN);
> +				ctl | PCI_EXP_DPC_CTL_INT_EN);

Align "ctl" with "pdev".

> +	return PCI_ERS_RESULT_RECOVERED;
>  }
>  
> +static void dpc_work(struct work_struct *work)
> +{
> +	struct dpc_dev *dpc = container_of(work, struct dpc_dev, work);
> +	struct pci_dev *pdev = dpc->dev->port;
> +
> +	/* From DPC point of view error is always FATAL. */
> +	pcie_do_recovery(pdev, DPC_FATAL);
> +}
>  static void dpc_process_rp_pio_error(struct dpc_dev *dpc)
>  {
>  	struct device *dev = &dpc->dev->device;
> @@ -297,6 +373,7 @@ static void dpc_remove(struct pcie_device *dev)
>  	.service	= PCIE_PORT_SERVICE_DPC,
>  	.probe		= dpc_probe,
>  	.remove		= dpc_remove,
> +	.reset_link     = dpc_reset_link,
>  };
>  
>  static int __init dpc_service_init(void)
> diff --git a/drivers/pci/pcie/pcie-err.c b/drivers/pci/pcie/pcie-err.c
> index f830975..1ea4b9a 100644
> --- a/drivers/pci/pcie/pcie-err.c
> +++ b/drivers/pci/pcie/pcie-err.c
> @@ -19,6 +19,7 @@
>  #include <linux/aer.h>
>  #include <linux/pcieport_if.h>
>  #include "portdrv.h"
> +#include "./../pci.h"
>  
>  static DEFINE_MUTEX(pci_err_recovery_lock);
>  
> @@ -181,7 +182,7 @@ static pci_ers_result_t default_reset_link(struct pci_dev *dev)
>  	return PCI_ERS_RESULT_RECOVERED;
>  }
>  
> -static pci_ers_result_t reset_link(struct pci_dev *dev)
> +static pci_ers_result_t reset_link(struct pci_dev *dev, int severity)
>  {
>  	struct pci_dev *udev;
>  	pci_ers_result_t status;
> @@ -195,9 +196,17 @@ static pci_ers_result_t reset_link(struct pci_dev *dev)
>  		udev = dev->bus->self;
>  	}
>  
> +
> +	/* Use the service driver of the component firstly */
> +#if IS_ENABLED(CONFIG_PCIE_DPC)

#ifdef CONFIG_PCIE_DPC

> +	if (severity == DPC_FATAL)
> +		driver = pci_find_dpc_service(udev);
> +#endif
>  #if IS_ENABLED(CONFIG_PCIEAER)
> -	/* Use the aer driver of the component firstly */
> -	driver = pci_find_aer_service(udev);
> +	if (severity == AER_FATAL ||
> +	    severity == AER_NONFATAL ||
> +	    severity == AER_CORRECTABLE)

This change (to check for AER_FATAL, etc) looks like it belongs in a
different patch.  This patch doesn't change any places that set the
severity.

> +		driver = pci_find_aer_service(udev);
>  #endif
>  
>  	if (driver && driver->reset_link) {
> @@ -287,7 +296,8 @@ void pcie_do_recovery(struct pci_dev *dev, int severity)
>  
>  	mutex_lock(&pci_err_recovery_lock);
>  
> -	if (severity == AER_FATAL)
> +	if (severity == AER_FATAL ||
> +	    severity == DPC_FATAL)
>  		state = pci_channel_io_frozen;
>  	else
>  		state = pci_channel_io_normal;
> @@ -297,10 +307,14 @@ void pcie_do_recovery(struct pci_dev *dev, int severity)
>  			"error_detected",
>  			report_error_detected);
>  
> -	if (severity == AER_FATAL) {
> -		result = reset_link(dev);
> +	if (severity == AER_FATAL ||
> +	    severity == DPC_FATAL) {
> +		result = reset_link(dev, severity);
>  		if (result != PCI_ERS_RESULT_RECOVERED)
>  			goto failed;
> +		else if (severity == DPC_FATAL)
> +			goto resume;
> +
>  	}
>  
>  	if (status == PCI_ERS_RESULT_CAN_RECOVER)
> @@ -324,6 +338,7 @@ void pcie_do_recovery(struct pci_dev *dev, int severity)
>  	if (status != PCI_ERS_RESULT_RECOVERED)
>  		goto failed;
>  
> +resume:
>  	broadcast_error_message(dev,
>  				state,
>  				"resume",
> diff --git a/drivers/pci/pcie/portdrv.h b/drivers/pci/pcie/portdrv.h
> index 4f1992d..b013e24 100644
> --- a/drivers/pci/pcie/portdrv.h
> +++ b/drivers/pci/pcie/portdrv.h
> @@ -80,4 +80,5 @@ static inline void pcie_port_platform_notify(struct pci_dev *port, int *mask){}
>  #endif /* !CONFIG_ACPI */
>  
>  struct pcie_port_service_driver *pci_find_aer_service(struct pci_dev *dev);
> +struct pcie_port_service_driver *pci_find_dpc_service(struct pci_dev *dev);
>  #endif /* _PORTDRV_H_ */
> -- 
> Qualcomm Datacenter Technologies, Inc. as an affiliate of Qualcomm Technologies, Inc.,
> a Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project.
> 

Powered by blists - more mailing lists