Re: Using binary attributes for configuration sysfs entries

From: Jay Aurabind <jay.aurabind@gmail.com>
To: Greg KH <greg@kroah.com>
Cc: kernelnewbies <kernelnewbies@kernelnewbies.org>
Subject: Re: Using binary attributes for configuration sysfs entries
Date: Thu, 14 Mar 2019 23:11:15 +0530	[thread overview]
Message-ID: <CAOsEZohtWCuCPUtedfDYdO5DG=qQbrUqXHRGsYNO-RPPnfoR7A@mail.gmail.com> (raw)
In-Reply-To: <20190314145749.GB12576@kroah.com>

On Thu, 14 Mar 2019 at 20:27, Greg KH <greg@kroah.com> wrote:
>
> On Thu, Mar 14, 2019 at 04:17:13PM +0530, Jay Aurabind wrote:
> > On Wed, 13 Mar 2019 at 19:55, Greg KH <greg@kroah.com> wrote:
> > >
> > > On Wed, Mar 13, 2019 at 12:02:02PM +0530, Jay Aurabind wrote:
> > > > Hi,
> > > >
> > > > Before I send this patch to actual mailing list, I'd appreciate if
> > > > someone could tell me if this is a bad idea!
> > > >
> > > > The driver in staging for pi433 (a wireless transceiver) uses IOCTLs
> > > > at the moment. I wish to add a sysfs interface to it that control the
> > > > various transmission and reception parameters. In the ioctl interface,
> > > > it uses two structs that have about 40 parameters in total.
> > > >
> > > > For the corresponding sysfs interface, since there are a lot of
> > > > parameters, would it be justified to use the same binary format though
> > > > sysfs_create_binary_file() ? The rationale is that it would be easier
> > > > to simply pack all the config options in the struct and send it in
> > > > once rather than individually write 40 files. This is what the
> > > > attached patch follows. Interface is added only for reception
> > > > parameters as of now.
> > >
> > > binary sysfs files are only allowed for "pass through" data, where the
> > > kernel does not touch the information at all and only passes it from the
> > > hardware, to userspace directly (or the other way around).  It can not
> > > be used for data that the kernel actually knows about and modifies /
> > > acts on.
> > >
> > > An example of valid binary sysfs files are USB and PCI device
> > > configuration information (read directly from the hardware), or firmware
> > > files that are send from userspace directly to the hardware without the
> > > kernel knowing what the data is.
> > >
> > > You can't use a binary sysfs file for ioctl-like data, that's not
> > > allowed, just use an ioctl for that.  Or better yet, use a common api
> > > interface for it, to match the other types of devices.
> >
> > Thank you for the explanation. What API would the most appropriate for
> > such a wireless transceiver ? If I could make it as a networking
> > driver, I could probably use setsockopt to pass in these
> > configuration. Is this recommended ?
>
> No, try looking at how v4l2 works, odds are they already have the
> correct api for this type of device there for you to use.

Thank you very much for the information!
>
> thanks,
>
> greg k-h


-- 

Thanks and Regards,
Jay

_______________________________________________
Kernelnewbies mailing list
Kernelnewbies@kernelnewbies.org
https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies