On Fri, Aug 04, 2017 at 10:18:26AM +0800, Dongdong Liu wrote:
> From: Gabriele Paoloni <gabriele.paoloni@xxxxxxxxxx>
>
> Currently if an uncorrectable error is reported by an EP the AER
> driver walks over all the devices connected to the upstream port
> bus and in turns call the report_error_detected() callback.
> If any of the devices connected to the bus does not implement
> dev->driver->err_handler->error_detected() do_recovery() will fail.
>
> However for non fatal errors the PCIe link should not be considered
> compromised, therefore it makes sense to report the error only to
> all the functions of a multifunction device.
> This patch implements this new behaviour for non fatal errors.
Why do we bother even with other functions in the same multifunction
device? PCIe r3.1, sec 6.2.2.2.2, says non-fatal errors only affect a
particular transaction, and "devices not associated with the
transaction in error are not impacted."
A transaction is only associated with one function, so other functions
in the same device shouldn't be affected, should they?
> Signed-off-by: Gabriele Paoloni <gabriele.paoloni@xxxxxxxxxx>
> Signed-off-by: Dongdong Liu <liudongdong3@xxxxxxxxxx>
> ---
> drivers/pci/bus.c | 38 ++++++++++++++++++++++++++++++++++++++
> drivers/pci/pcie/aer/aerdrv_core.c | 13 ++++++++++++-
> include/linux/pci.h | 3 ++-
> 3 files changed, 52 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/pci/bus.c b/drivers/pci/bus.c
> index bc56cf1..bc8f8b2 100644
> --- a/drivers/pci/bus.c
> +++ b/drivers/pci/bus.c
> @@ -364,6 +364,44 @@ void pci_bus_add_devices(const struct pci_bus *bus)
> }
> EXPORT_SYMBOL(pci_bus_add_devices);
>
> +/** pci_walk_mf_dev - walk all functions of a multi-function
> + * device calling callback.
> + * @dev a function in a multi-function device
> + * @cb callback to be called for each device found
> + * @userdata arbitrary pointer to be passed to callback.
> + *
> + * Walk, on a given bus, only the adjacent functions of a
> + * multi-function device. Call the provided callback on each
> + * device found.
> + *
> + * We check the return of @cb each time. If it returns anything
> + * other than 0, we break out.
> + *
> + */
> +void pci_walk_mf_dev(struct pci_dev *dev, int (*cb)(struct pci_dev *, void *),
> + void *userdata)
> +{
> + int retval;
> + struct pci_bus *bus;
> + struct pci_dev *pdev;
> + int ndev;
> +
> + bus = dev->bus;
> + ndev = PCI_SLOT(dev->devfn);
> +
> + down_read(&pci_bus_sem);
> + /* call cb for all the functions of the mf device */
> + list_for_each_entry(pdev, &bus->devices, bus_list) {
> + if (PCI_SLOT(pdev->devfn) == ndev) {
> + retval = cb(pdev, userdata);
> + if (retval)
> + break;
> + }
> + }
> + up_read(&pci_bus_sem);
> +}
> +EXPORT_SYMBOL_GPL(pci_walk_mf_dev);
> +
> /** pci_walk_bus - walk devices on/under bus, calling callback.
> * @top bus whose devices should be walked
> * @cb callback to be called for each device found
> diff --git a/drivers/pci/pcie/aer/aerdrv_core.c b/drivers/pci/pcie/aer/aerdrv_core.c
> index b1303b3..67c3dc0 100644
> --- a/drivers/pci/pcie/aer/aerdrv_core.c
> +++ b/drivers/pci/pcie/aer/aerdrv_core.c
> @@ -390,7 +390,18 @@ static pci_ers_result_t broadcast_error_message(struct pci_dev *dev,
> * If the error is reported by an end point, we think this
> * error is related to the upstream link of the end point.
> */
> - pci_walk_bus(dev->bus, cb, &result_data);
> + if ((state == pci_channel_io_normal) &&
> + (!pci_ari_enabled(dev->bus)))
> + /*
> + * the error is non fatal so the bus is ok, just walk
> + * through all the functions in a multifunction device.
> + * if ARI is enabled on the bus then there can be only
> + * one device under that bus (so walk all the functions
> + * under the bus).
> + */
> + pci_walk_mf_dev(dev, cb, &result_data);
> + else
> + pci_walk_bus(dev->bus, cb, &result_data);
> }
>
> return result_data.result;
> diff --git a/include/linux/pci.h b/include/linux/pci.h
> index 4869e66..69e77bb 100644
> --- a/include/linux/pci.h
> +++ b/include/linux/pci.h
> @@ -1269,7 +1269,8 @@ const struct pci_device_id *pci_match_id(const struct pci_device_id *ids,
> struct pci_dev *dev);
> int pci_scan_bridge(struct pci_bus *bus, struct pci_dev *dev, int max,
> int pass);
> -
> +void pci_walk_mf_dev(struct pci_dev *dev, int (*cb)(struct pci_dev *, void *),
> + void *userdata);
> void pci_walk_bus(struct pci_bus *top, int (*cb)(struct pci_dev *, void *),
> void *userdata);
> int pci_cfg_space_size(struct pci_dev *dev);
> --
> 1.9.1
>
