On Thu 03 Oct 14:11 PDT 2019, mnalajal@xxxxxxxxxxxxxx wrote:
> On 2019-10-03 11:33, Greg KH wrote:
> > On Thu, Oct 03, 2019 at 11:23:45AM -0700, mnalajal@xxxxxxxxxxxxxx wrote:
> > > On 2019-10-03 00:06, Greg KH wrote:
> > > > On Wed, Oct 02, 2019 at 05:06:14PM -0700, Murali Nalajala wrote:
> > > > > Soc framework exposed sysfs entries are not sufficient for some
> > > > > of the h/w platforms. Currently there is no interface where soc
> > > > > drivers can expose further information about their SoCs via soc
> > > > > framework. This change address this limitation where clients can
> > > > > pass their custom entries as attribute group and soc framework
> > > > > would expose them as sysfs properties.
> > > > >
> > > > > Signed-off-by: Murali Nalajala <mnalajal@xxxxxxxxxxxxxx>
> > > > > ---
> > > > > drivers/base/soc.c | 26 ++++++++++++++++++--------
> > > > > include/linux/sys_soc.h | 1 +
> > > > > 2 files changed, 19 insertions(+), 8 deletions(-)
> > > > >
> > > > > diff --git a/drivers/base/soc.c b/drivers/base/soc.c
> > > > > index 7c0c5ca..ec70a58 100644
> > > > > --- a/drivers/base/soc.c
> > > > > +++ b/drivers/base/soc.c
> > > > > @@ -15,6 +15,8 @@
> > > > > #include <linux/err.h>
> > > > > #include <linux/glob.h>
> > > > >
> > > > > +#define NUM_ATTR_GROUPS 3
> > > > > +
> > > > > static DEFINE_IDA(soc_ida);
> > > > >
> > > > > static ssize_t soc_info_get(struct device *dev,
> > > > > @@ -104,11 +106,6 @@ static ssize_t soc_info_get(struct device *dev,
> > > > > .is_visible = soc_attribute_mode,
> > > > > };
> > > > >
> > > > > -static const struct attribute_group *soc_attr_groups[] = {
> > > > > - &soc_attr_group,
> > > > > - NULL,
> > > > > -};
> > > > > -
> > > > > static void soc_release(struct device *dev)
> > > > > {
> > > > > struct soc_device *soc_dev = container_of(dev, struct soc_device,
> > > > > dev);
> > > > > @@ -121,6 +118,7 @@ static void soc_release(struct device *dev)
> > > > > struct soc_device *soc_device_register(struct soc_device_attribute
> > > > > *soc_dev_attr)
> > > > > {
> > > > > struct soc_device *soc_dev;
> > > > > + const struct attribute_group **soc_attr_groups = NULL;
> > > > > int ret;
> > > > >
> > > > > if (!soc_bus_type.p) {
> > > > > @@ -136,10 +134,20 @@ struct soc_device *soc_device_register(struct
> > > > > soc_device_attribute *soc_dev_attr
> > > > > goto out1;
> > > > > }
> > > > >
> > > > > + soc_attr_groups = kzalloc(sizeof(*soc_attr_groups) *
> > > > > + NUM_ATTR_GROUPS, GFP_KERNEL);
> > > > > + if (!soc_attr_groups) {
> > > > > + ret = -ENOMEM;
> > > > > + goto out2;
> > > > > + }
> > > > > + soc_attr_groups[0] = &soc_attr_group;
> > > > > + soc_attr_groups[1] = soc_dev_attr->custom_attr_group;
> > > > > + soc_attr_groups[2] = NULL;
> > > > > +
> > > > > /* Fetch a unique (reclaimable) SOC ID. */
> > > > > ret = ida_simple_get(&soc_ida, 0, 0, GFP_KERNEL);
> > > > > if (ret < 0)
> > > > > - goto out2;
> > > > > + goto out3;
> > > > > soc_dev->soc_dev_num = ret;
> > > > >
> > > > > soc_dev->attr = soc_dev_attr;
> > > > > @@ -151,14 +159,16 @@ struct soc_device *soc_device_register(struct
> > > > > soc_device_attribute *soc_dev_attr
> > > > >
> > > > > ret = device_register(&soc_dev->dev);
> > > > > if (ret)
> > > > > - goto out3;
> > > > > + goto out4;
> > > > >
> > > > > return soc_dev;
> > > > >
> > > > > -out3:
> > > > > +out4:
> > > > > ida_simple_remove(&soc_ida, soc_dev->soc_dev_num);
> > > > > put_device(&soc_dev->dev);
> > > > > soc_dev = NULL;
> > > > > +out3:
> > > > > + kfree(soc_attr_groups);
> > > > > out2:
> > > > > kfree(soc_dev);
> > > > > out1:
> > > > > diff --git a/include/linux/sys_soc.h b/include/linux/sys_soc.h
> > > > > index 48ceea8..d9b3cf0 100644
> > > > > --- a/include/linux/sys_soc.h
> > > > > +++ b/include/linux/sys_soc.h
> > > > > @@ -15,6 +15,7 @@ struct soc_device_attribute {
> > > > > const char *serial_number;
> > > > > const char *soc_id;
> > > > > const void *data;
> > > > > + const struct attribute_group *custom_attr_group;
> > > >
> > > > Shouldn't you make this:
> > > > const struct attribute_group **soc_groups;
> > > >
> > > > to match up with the rest of the way the driver core works?
> > > Assumption is, soc drivers send their custom attribute group and soc
> > > framework has already soc_attr_group" (basic info exposed).
> > > With my changes i am combining these two groups and passing to
> > > "device_register()".
> > > I do not think soc drivers have a requirement where they can pass
> > > various
> > > groups rather one single group attribute.
> >
> > Ok, I guess this is "good enough" such that no individual SOC driver
> > will want to create subdirs and lots of fun like that. If they do, then
> > we can change the api at that point in time :)
> >
> > thanks,
> >
> > greg k-h
>
> I trying to fix an issue in the existing "soc_device_register()" code. This
> looks to me a memory leak.
>
> ret = device_register(&soc_dev->dev);
> if (ret)
> goto out3;
> return soc_dev;
> out3:
> ida_simple_remove(&soc_ida, soc_dev->soc_dev_num);
> put_device(&soc_dev->dev);
This put_device() will invoke soc_release() which will free soc_dev.
> soc_dev = NULL;
So setting soc_dev to NULL here turns below kfree() into a nop.
> out2:
> kfree(soc_dev);
>
> Here we are assigning "soc_dev=NULL" before freeing. I see this assignment
> is unnecessary here.
The code works as intended and the assignment prevents a double free.
But it's perhaps slightly too clever.
Swapping the allocation order of the ida and soc_dev would make this
clearer.
Regards,
Bjorn
