Re: [PATCH v10 4/8] PCI/sysfs: Allow userspace to query and set device reset mechanism

[Bjorn Helgaas <helgaas@kernel.org>]

On Fri, Jul 09, 2021 at 06:08:09PM +0530, Amey Narkhede wrote:
> Add reset_method sysfs attribute to enable user to query and set user
> preferred device reset methods and their ordering.
> 
> Co-developed-by: Alex Williamson <alex.williamson@redhat.com>
> Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
> Signed-off-by: Amey Narkhede <ameynarkhede03@gmail.com>
> ---
>  Documentation/ABI/testing/sysfs-bus-pci |  19 +++++
>  drivers/pci/pci-sysfs.c                 | 103 ++++++++++++++++++++++++
>  2 files changed, 122 insertions(+)
> 
> diff --git a/Documentation/ABI/testing/sysfs-bus-pci b/Documentation/ABI/testing/sysfs-bus-pci
> index ef00fada2..43f4e33c7 100644
> --- a/Documentation/ABI/testing/sysfs-bus-pci
> +++ b/Documentation/ABI/testing/sysfs-bus-pci
> @@ -121,6 +121,25 @@ Description:
>  		child buses, and re-discover devices removed earlier
>  		from this part of the device tree.
>  
> +What:		/sys/bus/pci/devices/.../reset_method
> +Date:		March 2021
> +Contact:	Amey Narkhede <ameynarkhede03@gmail.com>
> +Description:
> +		Some devices allow an individual function to be reset
> +		without affecting other functions in the same slot.
> +
> +		For devices that have this support, a file named
> +		reset_method will be present in sysfs. Initially reading
> +		this file will give names of the device supported reset
> +		methods and their ordering. After write, this file will
> +		give names and ordering of currently enabled reset methods.
> +		Writing the name or comma separated list of names of any of
> +		the device supported reset methods to this file will set
> +		the reset methods and their ordering to be used when
> +		resetting the device. Writing empty string to this file
> +		will disable ability to reset the device and writing
> +		"default" will return to the original value.
> +
>  What:		/sys/bus/pci/devices/.../reset
>  Date:		July 2009
>  Contact:	Michael S. Tsirkin <mst@redhat.com>
> diff --git a/drivers/pci/pci-sysfs.c b/drivers/pci/pci-sysfs.c
> index 316f70c3e..8a740e211 100644
> --- a/drivers/pci/pci-sysfs.c
> +++ b/drivers/pci/pci-sysfs.c
> @@ -1334,6 +1334,108 @@ static const struct attribute_group pci_dev_rom_attr_group = {
>  	.is_bin_visible = pci_dev_rom_attr_is_visible,
>  };
>  
> +static ssize_t reset_method_show(struct device *dev,
> +				 struct device_attribute *attr,
> +				 char *buf)
> +{
> +	struct pci_dev *pdev = to_pci_dev(dev);
> +	ssize_t len = 0;
> +	int i, idx;
> +
> +	for (i = 0; i < PCI_NUM_RESET_METHODS; i++) {
> +		idx = pdev->reset_methods[i];
> +		if (!idx)

I know it's totally OCD, but can you use the same "m" for indexing
pci_reset_fn_methods[] that you used in __pci_reset_function_locked()
and ... oops, I meant to say pci_init_reset_methods(), but there you
used "i".  Maybe pci_init_reset_methods() could use "m" as well, and
maybe it could use "i" instead of "n" to index dev->reset_methods[]?

> +			break;
> +
> +		len += sysfs_emit_at(buf, len, "%s%s", len ? "," : "",
> +				     pci_reset_fn_methods[idx].name);

Maybe separate with spaces instead of commas?  E.g.,

  device_specific flr pm bus

instead of

  device_specific,flr,pm,bus

I think the former is less error-prone when parsing.  Easier to split
in a shell script, for example.

> +	}
> +
> +	if (len)
> +		len += sysfs_emit_at(buf, len, "\n");
> +
> +	return len;
> +}
> +
> +static ssize_t reset_method_store(struct device *dev,
> +				  struct device_attribute *attr,
> +				  const char *buf, size_t count)
> +{
> +	struct pci_dev *pdev = to_pci_dev(dev);
> +	int n = 0;
> +	char *name, *options = NULL;
> +	u8 reset_methods[PCI_NUM_RESET_METHODS] = { 0 };
> +
> +	if (count >= (PAGE_SIZE - 1))
> +		return -EINVAL;

I'm not the sysfs expert, but surely the sysfs infrastructure already
guarantees this?

> +
> +	if (sysfs_streq(buf, "")) {
> +		pci_warn(pdev, "All device reset methods disabled by user");

Can we just do:

  pdev->reset_methods[0] = 0;

here and dispense with the memcpy()?

> +		goto set_reset_methods;
> +	}
> +
> +	if (sysfs_streq(buf, "default")) {
> +		pci_init_reset_methods(pdev);
> +		return count;
> +	}
> +
> +	options = kstrndup(buf, count, GFP_KERNEL);

I assume the kstrndup() is because strsep() writes into the buffer?
Aren't we allowed to write into the buffer we get from sysfs?  Does
the user ever see the buffer contents again?  I would think sysfs
would have already done a copy_from_user() or whatever.

> +	if (!options)
> +		return -ENOMEM;
> +
> +	while ((name = strsep(&options, ",")) != NULL) {

*If* you change the show to use spaces, obviously this would have to
change as well (as well as the doc).

> +		int i;
> +
> +		if (sysfs_streq(name, ""))
> +			continue;
> +
> +		name = strim(name);
> +
> +		for (i = 1; i < PCI_NUM_RESET_METHODS; i++) {
> +			if (sysfs_streq(name, pci_reset_fn_methods[i].name) &&
> +			    !pci_reset_fn_methods[i].reset_fn(pdev, 1)) {
> +				reset_methods[n++] = i;

Can we build this directly in pdev->reset_methods[] so we don't need
the memcpy() below?

> +				break;
> +			}
> +		}

Same "m" for indexing pci_reset_fn_methods[], "i" for
pdev->reset_methods[] comment here.

> +		if (i == PCI_NUM_RESET_METHODS) {
> +			kfree(options);
> +			return -EINVAL;
> +		}
> +	}
> +
> +	if (!pci_reset_fn_methods[1].reset_fn(pdev, 1) && reset_methods[0] != 1)
> +		pci_warn(pdev, "Device specific reset disabled/de-prioritized by user");

Hmmm.  I sort of see the point here, but I wish we didn't have the
implicit dependency on pci_reset_fn_methods[1] being
pci_dev_specific_reset().

I know we've talked about this before.  I'm still not 100% sure either
of these warnings is worthwhile, especially since we're not *using*
the reset here.  It might be useful at the point where we try to *do*
a reset.  I dunno.  Maybe this is the best place since this is where
the user potentially screwed up.

> +set_reset_methods:
> +	memcpy(pdev->reset_methods, reset_methods, sizeof(reset_methods));
> +	kfree(options);
> +	return count;
> +}
> +static DEVICE_ATTR_RW(reset_method);
> +
> +static struct attribute *pci_dev_reset_method_attrs[] = {
> +	&dev_attr_reset_method.attr,
> +	NULL,
> +};
> +
> +static umode_t pci_dev_reset_method_attr_is_visible(struct kobject *kobj,
> +						    struct attribute *a, int n)
> +{
> +	struct pci_dev *pdev = to_pci_dev(kobj_to_dev(kobj));
> +
> +	if (!pci_reset_supported(pdev))
> +		return 0;
> +
> +	return a->mode;
> +}
> +
> +static const struct attribute_group pci_dev_reset_method_attr_group = {
> +	.attrs = pci_dev_reset_method_attrs,
> +	.is_visible = pci_dev_reset_method_attr_is_visible,
> +};
> +
>  static ssize_t reset_store(struct device *dev, struct device_attribute *attr,
>  			   const char *buf, size_t count)
>  {
> @@ -1491,6 +1593,7 @@ const struct attribute_group *pci_dev_groups[] = {
>  	&pci_dev_config_attr_group,
>  	&pci_dev_rom_attr_group,
>  	&pci_dev_reset_attr_group,
> +	&pci_dev_reset_method_attr_group,
>  	&pci_dev_vpd_attr_group,
>  #ifdef CONFIG_DMI
>  	&pci_dev_smbios_attr_group,
> -- 
> 2.32.0
>