Re: [PATCHv2 4/4] pci: make reset_subordinate hotplug safe

[<dan.j.williams@intel.com>]

Keith Busch wrote:
> From: Keith Busch <kbusch@kernel.org>
> 
> Use the slot reset method when resetting the bridge if the bus contains

What about some renames to make this easier to read, because it is
ironic that this says "Use the slot reset method" when the functional
change is to now start using the __pci_reset_slot() method. If I am
reading this "slot reset vs reset slot" conversion correctly.

Something like:

s/__pci_reset_slot/pci_try_masked_slot_reset/

Since it is a conditional reset, and it is masking slot events.

...but pulling that string also means:

s/__pci_reset_bus/pci_try_reset_bus/

> hot plug slots. This fixes spurious hot plug events that are triggered
> by the secondary bus reset that bypasses the slot's detection disabling.
> 
> Resetting a bridge's subordinate bus can be done like this:
> 
>   # echo 1 > /sys/bus/pci/devices/0000:50:01.0/reset_subordinate
> 
> Prior to this patch, an example kernel message may show something like:
> 
>   pcieport 0000:50:01.0: pciehp: Slot(40): Link Down
> 
> With this change, the pciehp driver ignores the link event during the
> reset, so may show this message instead:
> 
>   pcieport 0000:50:01.0: pciehp: Slot(40): Link Down/Up ignored
> 
> Signed-off-by: Keith Busch <kbusch@kernel.org>
> ---
>  drivers/pci/pci-sysfs.c |  3 +-
>  drivers/pci/pci.c       | 70 +++++++++++++++++++++++++++--------------
>  drivers/pci/pci.h       |  2 +-
>  3 files changed, 48 insertions(+), 27 deletions(-)
> 
> diff --git a/drivers/pci/pci-sysfs.c b/drivers/pci/pci-sysfs.c
> index c2df915ad2d29..0e1cef1c1c73b 100644
> --- a/drivers/pci/pci-sysfs.c
> +++ b/drivers/pci/pci-sysfs.c
> @@ -553,7 +553,6 @@ static ssize_t reset_subordinate_store(struct device *dev,
>  				const char *buf, size_t count)
>  {
>  	struct pci_dev *pdev = to_pci_dev(dev);
> -	struct pci_bus *bus = pdev->subordinate;
>  	unsigned long val;
>  
>  	if (!capable(CAP_SYS_ADMIN))
> @@ -563,7 +562,7 @@ static ssize_t reset_subordinate_store(struct device *dev,
>  		return -EINVAL;
>  
>  	if (val) {
> -		int ret = __pci_reset_bus(bus);
> +		int ret = pci_reset_bridge(pdev);
>  
>  		if (ret)
>  			return ret;
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 36427fbf7a747..12cf61a9bbeb1 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -50,6 +50,7 @@ EXPORT_SYMBOL(pci_pci_problems);
>  unsigned int pci_pm_d3hot_delay;
>  
>  static void pci_pme_list_scan(struct work_struct *work);
> +static int __pci_reset_bridge(struct pci_dev *bridge, bool save);
>  
>  static LIST_HEAD(pci_pme_list);
>  static DEFINE_MUTEX(pci_pme_list_mutex);
> @@ -5419,29 +5420,7 @@ static int pci_bus_reset(struct pci_bus *bus, bool probe)
>   */
>  int pci_bus_error_reset(struct pci_dev *bridge)
>  {
> -	struct pci_bus *bus = bridge->subordinate;
> -	struct pci_slot *slot;
> -
> -	if (!bus)
> -		return -ENOTTY;
> -
> -	mutex_lock(&pci_slot_mutex);
> -	if (list_empty(&bus->slots))
> -		goto bus_reset;
> -
> -	list_for_each_entry(slot, &bus->slots, list)
> -		if (pci_probe_reset_slot(slot))
> -			goto bus_reset;
> -
> -	list_for_each_entry(slot, &bus->slots, list)
> -		if (pci_slot_reset(slot, PCI_RESET_DO_RESET))
> -			goto bus_reset;
> -
> -	mutex_unlock(&pci_slot_mutex);
> -	return 0;
> -bus_reset:
> -	mutex_unlock(&pci_slot_mutex);
> -	return pci_bus_reset(bridge->subordinate, PCI_RESET_DO_RESET);
> +	return __pci_reset_bridge(bridge, false);

Similar to how PCI_RESET_{DO_RESET,PROBE} makes the ambiguous
"true/false" readable, how about adding something like:

#define PCI_SLOT_RESET_MASKED true
#define PCI_SLOT_RESET_UNMASKED false

>  }
>  
>  /**
> @@ -5462,7 +5441,7 @@ EXPORT_SYMBOL_GPL(pci_probe_reset_bus);
>   *
>   * Same as above except return -EAGAIN if the bus cannot be locked
>   */
> -int __pci_reset_bus(struct pci_bus *bus)
> +static int __pci_reset_bus(struct pci_bus *bus)
>  {
>  	int rc;
>  
> @@ -5482,6 +5461,49 @@ int __pci_reset_bus(struct pci_bus *bus)
>  	return rc;
>  }
>  
> +static int __pci_reset_bridge(struct pci_dev *bridge, bool save)
> +{
> +	struct pci_bus *bus = bridge->subordinate;
> +	struct pci_slot *slot;
> +
> +	if (!bus)
> +		return -ENOTTY;
> +
> +	mutex_lock(&pci_slot_mutex);
> +	if (list_empty(&bus->slots))
> +		goto bus_reset;
> +
> +	list_for_each_entry(slot, &bus->slots, list)
> +		if (pci_probe_reset_slot(slot))
> +			goto bus_reset;
> +
> +	list_for_each_entry(slot, &bus->slots, list) {
> +		int ret;
> +
> +		if (save)
> +			ret = __pci_reset_slot(slot);
> +		else
> +			ret = pci_slot_reset(slot, PCI_RESET_DO_RESET);
> +
> +		if (ret)
> +			goto bus_reset;
> +	}
> +
> +	mutex_unlock(&pci_slot_mutex);
> +	return 0;
> +bus_reset:
> +	mutex_unlock(&pci_slot_mutex);
> +
> +	if (save)
> +		return __pci_reset_bus(bus);
> +	return pci_bus_reset(bridge->subordinate, PCI_RESET_DO_RESET);

Logic looks equivalent in the PCI_SLOT_RESET_UNMASKED case, I think it
wants a comment that it turns into a "try_reset" when
PCI_SLOT_RESET_MASKED...

> +}
> +
> +int pci_reset_bridge(struct pci_dev *bridge)
> +{
> +	return __pci_reset_bridge(bridge, true);
> +}

...and maybe that comment is helped by naming this helper
pci_try_reset_bridge()?

I would want a second opinion on the renames before saying they are
absolutely required.

Reviewed-by: Dan Williams <dan.j.williams@intel.com>

...note I did not look at anything beyond the claims in the changelog,
like potential unwanted knock-on effect from the new masking.