Re: [EXT] Re: [PATCH v5 2/2] misc: nxp-sr1xx: UWB driver support for sr1xx series chip

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Tue, Dec 20, 2022 at 08:03:08PM +0530, Manjunatha Venkatesh wrote:
> 
> On 11/30/2022 12:53 PM, Greg Kroah-Hartman wrote:
> > Caution: EXT Email
> > 
> > On Wed, Nov 30, 2022 at 09:39:59AM +0530, Manjunatha Venkatesh wrote:
> > > On 10/7/2022 8:27 PM, Greg Kroah-Hartman wrote:
> > > > Caution: EXT Email
> > > > 
> > > > On Fri, Oct 07, 2022 at 07:34:25PM +0530, Manjunatha Venkatesh wrote:
> > > > > On 9/14/2022 8:39 PM, Arnd Bergmann wrote:
> > > > > > Caution: EXT Email
> > > > > > 
> > > > > > On Wed, Sep 14, 2022, at 4:29 PM, Manjunatha Venkatesh wrote:
> > > > > > 
> > > > > > > NXP has SR1XX family of UWB Subsystems (UWBS) devices. SR1XX SOCs
> > > > > > > are FiRa Compliant. SR1XX SOCs are flash less devices and they need
> > > > > > > Firmware Download on every device boot. More details on the SR1XX Family
> > > > > > > can be found athttps://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.nxp.com%2Fproducts%2F%3AUWB-TRIMENSION&data=05%7C01%7Cmanjunatha.venkatesh%40nxp.com%7C46c5718c03ee429cf57208dad2a3cad7%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C638053898170779252%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=S2BswHaF22edAfiZXEKUwGfUTNi1nuQzQSdGDb26peI%3D&reserved=0
> > > > > > > 
> > > > > > > The sr1xx driver work the SR1XX Family of UWBS, and uses UWB Controller
> > > > > > > Interface (UCI).  The corresponding details are available in the FiRa
> > > > > > > Consortium Website (https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.firaconsortium.org%2F&data=05%7C01%7Cmanjunatha.venkatesh%40nxp.com%7C46c5718c03ee429cf57208dad2a3cad7%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C638053898170779252%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=0fFcimUd6gOxTV0EKS%2BfxRZfrMDg0fytq1eSDmkMZ9E%3D&reserved=0).
> > > > > > I know nothing about UWB, so I have no idea if the user interface
> > > > > > you propose here makes sense. My guess is that there is a good chance
> > > > > > that there are other implementations of UWB that would not work
> > > > > > with this specific driver interface, so you probably need a
> > > > > > slightly higher-level abstraction.
> > > > > > 
> > > > > > We had an older subsystem that was called UWB and that got removed
> > > > > > a while ago:
> > > > > > 
> > > > > > https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgit.kernel.org%2Fpub%2Fscm%2Flinux%2Fkernel%2Fgit%2Ftorvalds%2Flinux.git%2Fcommit%2Fdrivers%2Fstaging%2Fuwb%3Fid%3Dcaa6772db4c1deb5d9add48e95d6eab50699ee5e&data=05%7C01%7Cmanjunatha.venkatesh%40nxp.com%7C46c5718c03ee429cf57208dad2a3cad7%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C638053898170779252%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=XXYgofE9LlBCPGW1aKxKUOBEIGF0aQv%2Bh6x6iNATkLQ%3D&reserved=0
> > > > > > 
> > > > > > Is that the same UWB or something completely different?
> > > > > Basically, it is SPI device driver which supports UCI(Ultra-wide band
> > > > > Command Interface) packet structure. It is not same as in mentioned link.
> > > > Why isn't this just a normal SPI driver and you do the "UCI" commands
> > > > from userspace through the device node there?
> > > > 
> > > > I know I asked this before, but I can't remember the answer, sorry, so
> > > > please include that in the changelog information when you resubmit.
> > > > 
> > > > thanks,
> > > > 
> > > > greg k-h
> > > The IO Handshake needed with SR1XX Family of SOCs cannot use the RAW SPI
> > > Module's APIs and hence custom APIs are added for communication with the
> > > UWBS,
> > I do not understand, what "IO handshake"?  What is missing from the
> > userspace spi api that is needed here?
> > 
> > > With this will get required throughput for UWBS use cases to avoid multiple
> > > round trip between user  and kernel mode.
> > Based on the speed of the SPI bus, this should not be an issue at all.
> > If it is, please provide us real performance numbers showing the
> > problem, as there are ways of speeding that up.
> 
> Not only throughput and also this driver customized ioctls to be controlled
> 
> from the user space for different scenarios.

Then you need to strongly document this.

> Current driver have UCI (UWB Command Interface) specific header parsing
> logic.

What does this mean?

> There is a specific GPIOs hand shake mechanism required between Host Driver
> and UWBS
> 
> at driver level which is tightly coupled with our UWBS chip.

Why can't you do this in userspace?

> Basically UWBS expecting acknowledgement from Host driver after first
> interrupt request
> 
> triggered then Host driver acknowledge to UWBS through dedicated GPIOs.

Again, why can't you do this in userspace?

> After this one more interrupt request will be triggered from UWBS for read
> operation.

Again, userspace?

thanks,

greg k-h