Hi,
Thank you for sending the patch over! A few small comments below.
[...]
> +static ssize_t pci_10bit_tag_store(struct device *dev,
> + struct device_attribute *attr,
> + const char *buf, size_t count)
> +{
> + struct pci_dev *pdev = to_pci_dev(dev);
> + bool enable;
Would you mind adding the following capabilities check here?
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
This is so we make sure that whatever user is going to use this sysfs
attribute actually has enough permissions to update this value safely.
> + if (kstrtobool(buf, &enable) < 0)
> + return -EINVAL;
> +
> + if (pdev->driver)
> + return -EBUSY;
> +
> + if (enable) {
> + if (!pcie_rp_10bit_tag_cmp_supported(pdev))
> + return -EPERM;
Would it make sense to also verify 10-Bit Tag Completer support on the
"disable" path too? We won't be able to set a value if there is no
support, but nothing will stop us from clearing it regardless - unless
this would be safe to do? What do you think?
> + pcie_capability_set_word(pdev, PCI_EXP_DEVCTL2,
> + PCI_EXP_DEVCTL2_10BIT_TAG_REQ_EN);
> + } else {
> + pcie_capability_clear_word(pdev, PCI_EXP_DEVCTL2,
> + PCI_EXP_DEVCTL2_10BIT_TAG_REQ_EN);
> + }
> +
> + return count;
> +}
[...]
> +> +static umode_t pcie_dev_10bit_tag_attrs_are_visible(struct kobject *kobj,
> + struct attribute *a, int n)
The preferred function name for the .is_visible() callback in a case when
there is only a single sysfs attribute being added would be:
pcie_dev_10bit_tag_attr_is_visible()
Albeit, I appreciate that you followed the existing naming pattern.
Krzysztof
