[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <2024111843-olive-handball-c703@gregkh>
Date: Mon, 18 Nov 2024 15:17:58 +0100
From: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To: Thomas Weißschuh <linux@...ssschuh.net>
Cc: linux-kernel@...r.kernel.org, "Rafael J. Wysocki" <rafael@...nel.org>
Subject: Re: [PATCH 1/2] sysfs: attribute_group: allow registration of const
bin_attribute
On Mon, Nov 18, 2024 at 12:36:06PM +0100, Thomas Weißschuh wrote:
> Hi Greg,
>
> On 2024-11-15 17:42:48+0100, Thomas Weißschuh wrote:
>
> > [..]
>
> > diff --git a/include/linux/sysfs.h b/include/linux/sysfs.h
> > index d713a6445a6267145a7014f308df3bb25b8c3287..0f2fcd244523f050c5286f19d4fe1846506f9214 100644
> > --- a/include/linux/sysfs.h
> > +++ b/include/linux/sysfs.h
> > @@ -106,7 +106,10 @@ struct attribute_group {
> > const struct bin_attribute *,
> > int);
> > struct attribute **attrs;
> > - struct bin_attribute **bin_attrs;
> > + union {
> > + struct bin_attribute **bin_attrs;
> > + const struct bin_attribute *const *bin_attrs_new;
> > + };
>
> Unfortunately this triggers warnings in two drivers.
> These incorrectly have a trailing NULL literal in their struct attribute
> definition (full list at the end of the mail):
>
> >> drivers/perf/arm-ni.c:248:63: warning: missing braces around initializer [-Wmissing-braces]
> 248 | static const struct attribute_group arm_ni_other_attr_group = {
> | ^
>
>
> vim +248 drivers/perf/arm-ni.c
>
> 4d5a7680f2b4d0 Robin Murphy 2024-09-04 247
> 4d5a7680f2b4d0 Robin Murphy 2024-09-04 @248 static const struct attribute_group arm_ni_other_attr_group = {
> 4d5a7680f2b4d0 Robin Murphy 2024-09-04 249 .attrs = arm_ni_other_attrs,
> 4d5a7680f2b4d0 Robin Murphy 2024-09-04 250 NULL
> 4d5a7680f2b4d0 Robin Murphy 2024-09-04 251 };
> 4d5a7680f2b4d0 Robin Murphy 2024-09-04 252
>
> These trailing NULLs should first be removed.
> How do you want to proceed?
Odd, it passed 0-day testing.
Just send me a patch to fix up these obvious problems, strange it built
in the first place (it's a mix of named and not named identifiers, I
thought the compiler would complain about that...)
> Cocci script and results, only the first two results are relevant at
> this moment.
>
> virtual patch
>
> @@
> identifier ag, pattrs;
> @@
>
> struct attribute_group ag = {
> .attrs = pattrs,
> - NULL
> };
>
> diff -u -p a/drivers/s390/char/con3215.c b/drivers/s390/char/con3215.c
> --- a/drivers/s390/char/con3215.c
> +++ b/drivers/s390/char/con3215.c
> @@ -803,7 +803,6 @@ static struct attribute *con3215_drv_att
>
> static struct attribute_group con3215_drv_attr_group = {
> .attrs = con3215_drv_attrs,
> - NULL,
> };
>
> static const struct attribute_group *con3215_drv_attr_groups[] = {
> diff -u -p a/drivers/perf/arm-ni.c b/drivers/perf/arm-ni.c
> --- a/drivers/perf/arm-ni.c
> +++ b/drivers/perf/arm-ni.c
> @@ -247,7 +247,6 @@ static struct attribute *arm_ni_other_at
>
> static const struct attribute_group arm_ni_other_attr_group = {
> .attrs = arm_ni_other_attrs,
> - NULL
> };
>
> static const struct attribute_group *arm_ni_attr_groups[] = {
> diff -u -p a/kernel/cpu.c b/kernel/cpu.c
> --- a/kernel/cpu.c
> +++ b/kernel/cpu.c
> @@ -2866,7 +2866,6 @@ static struct attribute *cpuhp_cpu_attrs
> static const struct attribute_group cpuhp_cpu_attr_group = {
> .attrs = cpuhp_cpu_attrs,
> .name = "hotplug",
> - NULL
> };
>
> static ssize_t states_show(struct device *dev,
> @@ -2898,7 +2897,6 @@ static struct attribute *cpuhp_cpu_root_
> static const struct attribute_group cpuhp_cpu_root_attr_group = {
> .attrs = cpuhp_cpu_root_attrs,
> .name = "hotplug",
> - NULL
> };
>
> #ifdef CONFIG_HOTPLUG_SMT
> @@ -3020,7 +3018,6 @@ static struct attribute *cpuhp_smt_attrs
> static const struct attribute_group cpuhp_smt_attr_group = {
> .attrs = cpuhp_smt_attrs,
> .name = "smt",
> - NULL
> };
>
> static int __init cpu_smt_sysfs_init(void)
Looks sane, send me a patch?
thanks,
greg k-h
Powered by blists - more mailing lists