To be able to constify instances of struct attribute it has to be
possible to add them to struct attribute_group.
The current type of the attrs member however is not compatible with that.
Introduce a union that allows registration of both const and non-const
attributes to enable a piecewise transition.
As both union member types are compatible no logic needs to be adapted.
Technically it is now possible register a const struct
attribute and receive it as mutable pointer in the callbacks.
This is a soundness issue.
But this same soundness issue already exists today in
sysfs_create_file().
Also the struct definition and callback implementation are always
closely linked and are meant to be moved to const in lockstep.
Similar to commit 906c508afdca ("sysfs: attribute_group: allow registration of const bin_attribute")
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
---
include/linux/sysfs.h | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h
index f418aae4f1134f8126783d9e8eb575ba4278e927..a47092e837d9eb014894d1f7e49f0fd0f9a2e350 100644
--- a/include/linux/sysfs.h
+++ b/include/linux/sysfs.h
@@ -105,7 +105,10 @@ struct attribute_group {
size_t (*bin_size)(struct kobject *,
const struct bin_attribute *,
int);
- struct attribute **attrs;
+ union {
+ struct attribute **attrs;
+ const struct attribute *const *attrs_new;
+ };
union {
const struct bin_attribute *const *bin_attrs;
const struct bin_attribute *const *bin_attrs_new;
--
2.50.1
On Mon, Aug 11, 2025 at 11:14:27AM +0200, Thomas Weißschuh wrote:
> To be able to constify instances of struct attribute it has to be
> possible to add them to struct attribute_group.
> The current type of the attrs member however is not compatible with that.
> Introduce a union that allows registration of both const and non-const
> attributes to enable a piecewise transition.
> As both union member types are compatible no logic needs to be adapted.
>
> Technically it is now possible register a const struct
> attribute and receive it as mutable pointer in the callbacks.
> This is a soundness issue.
> But this same soundness issue already exists today in
> sysfs_create_file().
> Also the struct definition and callback implementation are always
> closely linked and are meant to be moved to const in lockstep.
>
> Similar to commit 906c508afdca ("sysfs: attribute_group: allow registration of const bin_attribute")
>
> Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
> ---
> include/linux/sysfs.h | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h
> index f418aae4f1134f8126783d9e8eb575ba4278e927..a47092e837d9eb014894d1f7e49f0fd0f9a2e350 100644
> --- a/include/linux/sysfs.h
> +++ b/include/linux/sysfs.h
> @@ -105,7 +105,10 @@ struct attribute_group {
> size_t (*bin_size)(struct kobject *,
> const struct bin_attribute *,
> int);
> - struct attribute **attrs;
> + union {
> + struct attribute **attrs;
> + const struct attribute *const *attrs_new;
I know you will drop the "_new" prefix after a while, but "new" is
relative, and not very descriptive. How about "_const"?
> + };
> union {
> const struct bin_attribute *const *bin_attrs;
> const struct bin_attribute *const *bin_attrs_new;
There is no bin_attrs_new anymore. Finally. sorry about that...
greg k-h
On 2025-08-19 13:22:55+0200, Greg Kroah-Hartman wrote:
> On Mon, Aug 11, 2025 at 11:14:27AM +0200, Thomas Weißschuh wrote:
> > To be able to constify instances of struct attribute it has to be
> > possible to add them to struct attribute_group.
> > The current type of the attrs member however is not compatible with that.
> > Introduce a union that allows registration of both const and non-const
> > attributes to enable a piecewise transition.
> > As both union member types are compatible no logic needs to be adapted.
> >
> > Technically it is now possible register a const struct
> > attribute and receive it as mutable pointer in the callbacks.
> > This is a soundness issue.
> > But this same soundness issue already exists today in
> > sysfs_create_file().
> > Also the struct definition and callback implementation are always
> > closely linked and are meant to be moved to const in lockstep.
> >
> > Similar to commit 906c508afdca ("sysfs: attribute_group: allow registration of const bin_attribute")
> >
> > Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
> > ---
> > include/linux/sysfs.h | 5 ++++-
> > 1 file changed, 4 insertions(+), 1 deletion(-)
> >
> > diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h
> > index f418aae4f1134f8126783d9e8eb575ba4278e927..a47092e837d9eb014894d1f7e49f0fd0f9a2e350 100644
> > --- a/include/linux/sysfs.h
> > +++ b/include/linux/sysfs.h
> > @@ -105,7 +105,10 @@ struct attribute_group {
> > size_t (*bin_size)(struct kobject *,
> > const struct bin_attribute *,
> > int);
> > - struct attribute **attrs;
> > + union {
> > + struct attribute **attrs;
> > + const struct attribute *const *attrs_new;
>
> I know you will drop the "_new" prefix after a while, but "new" is
> relative, and not very descriptive.
That is somewhat intentional to express that it is a transitional thing.
> How about "_const"?
At some point the regular variant will be const too, so "_const" would
be a bit weird.
> > + };
> > union {
> > const struct bin_attribute *const *bin_attrs;
> > const struct bin_attribute *const *bin_attrs_new;
>
> There is no bin_attrs_new anymore. Finally. sorry about that...
Thanks! No worries.
Thomas
On Tue, Aug 19, 2025 at 03:59:04PM +0200, Thomas Weißschuh wrote:
> On 2025-08-19 13:22:55+0200, Greg Kroah-Hartman wrote:
> > On Mon, Aug 11, 2025 at 11:14:27AM +0200, Thomas Weißschuh wrote:
> > > To be able to constify instances of struct attribute it has to be
> > > possible to add them to struct attribute_group.
> > > The current type of the attrs member however is not compatible with that.
> > > Introduce a union that allows registration of both const and non-const
> > > attributes to enable a piecewise transition.
> > > As both union member types are compatible no logic needs to be adapted.
> > >
> > > Technically it is now possible register a const struct
> > > attribute and receive it as mutable pointer in the callbacks.
> > > This is a soundness issue.
> > > But this same soundness issue already exists today in
> > > sysfs_create_file().
> > > Also the struct definition and callback implementation are always
> > > closely linked and are meant to be moved to const in lockstep.
> > >
> > > Similar to commit 906c508afdca ("sysfs: attribute_group: allow registration of const bin_attribute")
> > >
> > > Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
> > > ---
> > > include/linux/sysfs.h | 5 ++++-
> > > 1 file changed, 4 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h
> > > index f418aae4f1134f8126783d9e8eb575ba4278e927..a47092e837d9eb014894d1f7e49f0fd0f9a2e350 100644
> > > --- a/include/linux/sysfs.h
> > > +++ b/include/linux/sysfs.h
> > > @@ -105,7 +105,10 @@ struct attribute_group {
> > > size_t (*bin_size)(struct kobject *,
> > > const struct bin_attribute *,
> > > int);
> > > - struct attribute **attrs;
> > > + union {
> > > + struct attribute **attrs;
> > > + const struct attribute *const *attrs_new;
> >
> > I know you will drop the "_new" prefix after a while, but "new" is
> > relative, and not very descriptive.
>
> That is somewhat intentional to express that it is a transitional thing.
Fair, but given the huge quantity here, it's going to take a long time,
so "new" is going to be rough to push through for 6+ months.
> > How about "_const"?
>
> At some point the regular variant will be const too, so "_const" would
> be a bit weird.
Yes, that's when you "switch it back", right? You would have to do that
for _new as well.
thanks,
greg k-h
On 2025-08-19 16:10:42+0200, Greg Kroah-Hartman wrote:
> On Tue, Aug 19, 2025 at 03:59:04PM +0200, Thomas Weißschuh wrote:
> > On 2025-08-19 13:22:55+0200, Greg Kroah-Hartman wrote:
> > > On Mon, Aug 11, 2025 at 11:14:27AM +0200, Thomas Weißschuh wrote:
> > > > To be able to constify instances of struct attribute it has to be
> > > > possible to add them to struct attribute_group.
> > > > The current type of the attrs member however is not compatible with that.
> > > > Introduce a union that allows registration of both const and non-const
> > > > attributes to enable a piecewise transition.
> > > > As both union member types are compatible no logic needs to be adapted.
> > > >
> > > > Technically it is now possible register a const struct
> > > > attribute and receive it as mutable pointer in the callbacks.
> > > > This is a soundness issue.
> > > > But this same soundness issue already exists today in
> > > > sysfs_create_file().
> > > > Also the struct definition and callback implementation are always
> > > > closely linked and are meant to be moved to const in lockstep.
> > > >
> > > > Similar to commit 906c508afdca ("sysfs: attribute_group: allow registration of const bin_attribute")
> > > >
> > > > Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
> > > > ---
> > > > include/linux/sysfs.h | 5 ++++-
> > > > 1 file changed, 4 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h
> > > > index f418aae4f1134f8126783d9e8eb575ba4278e927..a47092e837d9eb014894d1f7e49f0fd0f9a2e350 100644
> > > > --- a/include/linux/sysfs.h
> > > > +++ b/include/linux/sysfs.h
> > > > @@ -105,7 +105,10 @@ struct attribute_group {
> > > > size_t (*bin_size)(struct kobject *,
> > > > const struct bin_attribute *,
> > > > int);
> > > > - struct attribute **attrs;
> > > > + union {
> > > > + struct attribute **attrs;
> > > > + const struct attribute *const *attrs_new;
> > >
> > > I know you will drop the "_new" prefix after a while, but "new" is
> > > relative, and not very descriptive.
> >
> > That is somewhat intentional to express that it is a transitional thing.
>
> Fair, but given the huge quantity here, it's going to take a long time,
> so "new" is going to be rough to push through for 6+ months.
Looking at how 'struct bin_attribute' went probably quite a bit longer.
> > > How about "_const"?
> >
> > At some point the regular variant will be const too, so "_const" would
> > be a bit weird.
>
> Yes, that's when you "switch it back", right? You would have to do that
> for _new as well.
There will probably be some overlap. But in the end it probably
doesn't matter. Let's go with "_const".
Thomas
© 2016 - 2026 Red Hat, Inc.