Re: [PATCH v8 4/8] nvmem: core: Track the registered devices

[Miquel Raynal]

Hi Greg,

gregkh@xxxxxxxxxxxxxxxxxxx wrote on Mon, 7 Aug 2023 11:02:35 +0200:

> On Mon, Aug 07, 2023 at 10:24:15AM +0200, Miquel Raynal wrote:
> > Create a list with all the NVMEM devices registered in the
> > subsystem. This way we can iterate through them when needed (unused for
> > now).
> > 
> > Signed-off-by: Miquel Raynal <miquel.raynal@xxxxxxxxxxx>
> > ---
> >  drivers/nvmem/core.c | 12 ++++++++++++
> >  1 file changed, 12 insertions(+)
> > 
> > diff --git a/drivers/nvmem/core.c b/drivers/nvmem/core.c
> > index 257328887263..4e81e0aaf433 100644
> > --- a/drivers/nvmem/core.c
> > +++ b/drivers/nvmem/core.c
> > @@ -23,6 +23,7 @@
> >  struct nvmem_device {
> >  	struct module		*owner;
> >  	struct device		dev;
> > +	struct list_head	node;
> >  	int			stride;
> >  	int			word_size;
> >  	int			id;
> > @@ -76,6 +77,9 @@ static LIST_HEAD(nvmem_cell_tables);
> >  static DEFINE_MUTEX(nvmem_lookup_mutex);
> >  static LIST_HEAD(nvmem_lookup_list);
> >  
> > +static DEFINE_MUTEX(nvmem_devices_mutex);
> > +static LIST_HEAD(nvmem_devices_list);  
> 
> But this list should already be in the driver core, why create
> yet-another-list-and-lock?

I did not think about using it. I believe you mean using
bus_for_each_dev() here? Could definitely make the trick; I'll try.

> 
> Why is "when needed" not sufficient to use the list already present?
> 
> And now note, you have the same structure on 2 different lists, watch
> out for device lifetime rules :(
> 
> thanks,
> 
> greg k-h


Thanks,
Miquèl