| Message ID | 159312906372.1850128.11611897078988158727.stgit@dwillia2-desk3.amr.corp.intel.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | ACPI/NVDIMM: Runtime Firmware Activation | expand |
On Thu, Jun 25, 2020 at 04:51:03PM -0700, Dan Williams wrote: > A common pattern for using plain DEVICE_ATTR() instead of > DEVICE_ATTR_RO() and DEVICE_ATTR_RW() is for attributes that want to > limit read to only root. I.e. many users of DEVICE_ATTR() are > specifying 0400 or 0600 for permissions. > > Given the expectation that CAP_SYS_ADMIN is needed to access these > sensitive attributes and an explicit helper with the _ADMIN_ identifier > for DEVICE_ATTR_ADMIN_{RO,RW}. > > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com> > Signed-off-by: Dan Williams <dan.j.williams@intel.com> > --- > include/linux/device.h | 4 ++++ > include/linux/sysfs.h | 7 +++++++ > 2 files changed, 11 insertions(+) > > diff --git a/include/linux/device.h b/include/linux/device.h > index 15460a5ac024..d7c2570368fa 100644 > --- a/include/linux/device.h > +++ b/include/linux/device.h > @@ -128,8 +128,12 @@ ssize_t device_store_bool(struct device *dev, struct device_attribute *attr, > __ATTR_PREALLOC(_name, _mode, _show, _store) > #define DEVICE_ATTR_RW(_name) \ > struct device_attribute dev_attr_##_name = __ATTR_RW(_name) > +#define DEVICE_ATTR_ADMIN_RW(_name) \ > + struct device_attribute dev_attr_##_name = __ATTR_RW_MODE(_name, 0600) > #define DEVICE_ATTR_RO(_name) \ > struct device_attribute dev_attr_##_name = __ATTR_RO(_name) > +#define DEVICE_ATTR_ADMIN_RO(_name) \ > + struct device_attribute dev_attr_##_name = __ATTR_RO_MODE(_name, 0400) > #define DEVICE_ATTR_WO(_name) \ > struct device_attribute dev_attr_##_name = __ATTR_WO(_name) > #define DEVICE_ULONG_ATTR(_name, _mode, _var) \ > diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h > index 86067dbe7745..34e84122f635 100644 > --- a/include/linux/sysfs.h > +++ b/include/linux/sysfs.h > @@ -123,6 +123,13 @@ struct attribute_group { > .show = _name##_show, \ > } > > +#define __ATTR_RW_MODE(_name, _mode) { \ > + .attr = { .name = __stringify(_name), \ > + .mode = VERIFY_OCTAL_PERMISSIONS(_mode) }, \ > + .show = _name##_show, \ > + .store = _name##_store, \ > +} > + > #define __ATTR_WO(_name) { \ > .attr = { .name = __stringify(_name), .mode = 0200 }, \ > .store = _name##_store, \ > Nice! Want me to take this now, I know of many other places that could be cleaned up to use this (like the raw device bug that I just fixed...) If not: Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On Thu, Jun 25, 2020 at 10:06 PM Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote: > > On Thu, Jun 25, 2020 at 04:51:03PM -0700, Dan Williams wrote: > > A common pattern for using plain DEVICE_ATTR() instead of > > DEVICE_ATTR_RO() and DEVICE_ATTR_RW() is for attributes that want to > > limit read to only root. I.e. many users of DEVICE_ATTR() are > > specifying 0400 or 0600 for permissions. > > > > Given the expectation that CAP_SYS_ADMIN is needed to access these > > sensitive attributes and an explicit helper with the _ADMIN_ identifier > > for DEVICE_ATTR_ADMIN_{RO,RW}. > > > > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > > Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com> > > Signed-off-by: Dan Williams <dan.j.williams@intel.com> > > --- > > include/linux/device.h | 4 ++++ > > include/linux/sysfs.h | 7 +++++++ > > 2 files changed, 11 insertions(+) > > > > diff --git a/include/linux/device.h b/include/linux/device.h > > index 15460a5ac024..d7c2570368fa 100644 > > --- a/include/linux/device.h > > +++ b/include/linux/device.h > > @@ -128,8 +128,12 @@ ssize_t device_store_bool(struct device *dev, struct device_attribute *attr, > > __ATTR_PREALLOC(_name, _mode, _show, _store) > > #define DEVICE_ATTR_RW(_name) \ > > struct device_attribute dev_attr_##_name = __ATTR_RW(_name) > > +#define DEVICE_ATTR_ADMIN_RW(_name) \ > > + struct device_attribute dev_attr_##_name = __ATTR_RW_MODE(_name, 0600) > > #define DEVICE_ATTR_RO(_name) \ > > struct device_attribute dev_attr_##_name = __ATTR_RO(_name) > > +#define DEVICE_ATTR_ADMIN_RO(_name) \ > > + struct device_attribute dev_attr_##_name = __ATTR_RO_MODE(_name, 0400) > > #define DEVICE_ATTR_WO(_name) \ > > struct device_attribute dev_attr_##_name = __ATTR_WO(_name) > > #define DEVICE_ULONG_ATTR(_name, _mode, _var) \ > > diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h > > index 86067dbe7745..34e84122f635 100644 > > --- a/include/linux/sysfs.h > > +++ b/include/linux/sysfs.h > > @@ -123,6 +123,13 @@ struct attribute_group { > > .show = _name##_show, \ > > } > > > > +#define __ATTR_RW_MODE(_name, _mode) { \ > > + .attr = { .name = __stringify(_name), \ > > + .mode = VERIFY_OCTAL_PERMISSIONS(_mode) }, \ > > + .show = _name##_show, \ > > + .store = _name##_store, \ > > +} > > + > > #define __ATTR_WO(_name) { \ > > .attr = { .name = __stringify(_name), .mode = 0200 }, \ > > .store = _name##_store, \ > > > > Nice! Want me to take this now, I know of many other places that could > be cleaned up to use this (like the raw device bug that I just fixed...) Sure, that'd be great. > If not: > > Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Thanks Greg.
diff --git a/include/linux/device.h b/include/linux/device.h index 15460a5ac024..d7c2570368fa 100644 --- a/include/linux/device.h +++ b/include/linux/device.h @@ -128,8 +128,12 @@ ssize_t device_store_bool(struct device *dev, struct device_attribute *attr, __ATTR_PREALLOC(_name, _mode, _show, _store) #define DEVICE_ATTR_RW(_name) \ struct device_attribute dev_attr_##_name = __ATTR_RW(_name) +#define DEVICE_ATTR_ADMIN_RW(_name) \ + struct device_attribute dev_attr_##_name = __ATTR_RW_MODE(_name, 0600) #define DEVICE_ATTR_RO(_name) \ struct device_attribute dev_attr_##_name = __ATTR_RO(_name) +#define DEVICE_ATTR_ADMIN_RO(_name) \ + struct device_attribute dev_attr_##_name = __ATTR_RO_MODE(_name, 0400) #define DEVICE_ATTR_WO(_name) \ struct device_attribute dev_attr_##_name = __ATTR_WO(_name) #define DEVICE_ULONG_ATTR(_name, _mode, _var) \ diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h index 86067dbe7745..34e84122f635 100644 --- a/include/linux/sysfs.h +++ b/include/linux/sysfs.h @@ -123,6 +123,13 @@ struct attribute_group { .show = _name##_show, \ } +#define __ATTR_RW_MODE(_name, _mode) { \ + .attr = { .name = __stringify(_name), \ + .mode = VERIFY_OCTAL_PERMISSIONS(_mode) }, \ + .show = _name##_show, \ + .store = _name##_store, \ +} + #define __ATTR_WO(_name) { \ .attr = { .name = __stringify(_name), .mode = 0200 }, \ .store = _name##_store, \
A common pattern for using plain DEVICE_ATTR() instead of DEVICE_ATTR_RO() and DEVICE_ATTR_RW() is for attributes that want to limit read to only root. I.e. many users of DEVICE_ATTR() are specifying 0400 or 0600 for permissions. Given the expectation that CAP_SYS_ADMIN is needed to access these sensitive attributes and an explicit helper with the _ADMIN_ identifier for DEVICE_ATTR_ADMIN_{RO,RW}. Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> --- include/linux/device.h | 4 ++++ include/linux/sysfs.h | 7 +++++++ 2 files changed, 11 insertions(+)