On Sat, Sep 30, 2017 at 07:56:33PM +0000, Mario.Limonciello@xxxxxxxx wrote: > > -----Original Message----- > > From: Darren Hart [mailto:dvhart@xxxxxxxxxxxxx] > > Sent: Friday, September 29, 2017 9:17 PM > > To: Limonciello, Mario <Mario_Limonciello@xxxxxxxx> > > Cc: Andy Shevchenko <andy.shevchenko@xxxxxxxxx>; LKML <linux- > > kernel@xxxxxxxxxxxxxxx>; platform-driver-x86@xxxxxxxxxxxxxxx; Andy Lutomirski > > <luto@xxxxxxxxxx>; quasisec@xxxxxxxxxx; pali.rohar@xxxxxxxxx > > Subject: Re: [PATCH v3 0/8] Introduce support for Dell SMBIOS over WMI > > > > On Wed, Sep 27, 2017 at 11:02:12PM -0500, Mario Limonciello wrote: ... > > My other concern is the freeform structure around creating the file > > operations in each driver for the chardev IOCTL. It seems like we need > > some kind of defined mapping from METHOD index to IOCTL number, or else > > some way to advertise what it is? > > > > I was originally thinking it would be a good way to do this cleanly too, but my > main concern is this one character device may handle multiple methods problem. > If you map method instance (index) to ioctl number you will most likely run into > clashes. So maybe it's worth not allowing that? > > So I've got another idea. How about instead of a variety of freeform ioctl per driver, > provide three ioctl functions for all the character devices that come in through > the WMI bus to use (say IOC 'W') with either read, write or write/read. The WMI bus > should be able to know which driver to pass it on to by the character device used. > This is not something I have any experience with. I see you added it to v4, so let's see what others have to say there. -- Darren Hart VMware Open Source Technology Center
