[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <2025051549-flannels-lively-a46d@gregkh>
Date: Thu, 15 May 2025 09:52:29 +0200
From: Greg KH <gregkh@...uxfoundation.org>
To: Kuninori Morimoto <kuninori.morimoto.gx@...esas.com>
Cc: Linux-Kernel <linux-kernel@...r.kernel.org>, devicetree@...r.kernel.org,
shimoda <yoshihiro.shimoda.uh@...esas.com>,
goda <yusuke.goda.sx@...esas.com>,
Kurokawa <harunobu.kurokawa.dn@...esas.com>,
Kihara <takeshi.kihara.df@...esas.com>,
kazuya.mizuguchi.ks@...esas.com, takamitsu.honda.pv@...esas.com
Subject: Re: Question about UIO vs DT
On Thu, May 15, 2025 at 04:45:23AM +0000, Kuninori Morimoto wrote:
>
> Hi Greg / UIO / DT
>
> I would like to ask about UIO vs DT.
>
> If my understanding was correct, current UIO can use 1 IRQ / 1 reg per 1 UIO,
> but some device needs multi-IRQ/reg. In such case, we need to use
> multi-UIO. But it is not good much to DT rule. For example in case of
> the device which needs "2 regs 3 irqs". it will be
>
> (A) [1 reg, 1 IRQ] UIO
> (B) [1 reg, 1 IRQ] UIO
> (C) [0 reg, 1 IRQ] UIO
>
> and (C) will be DT error. Is this known issue ? Do we have better solution ?
Yes, write a real driver for the device as obviously it is a complex one
and UIO shouldn't be used for it :)
What type of device is this that requires this type of hardware control
and why do you feel that UIO is the proper solution?
thanks,
greg k-h
Powered by blists - more mailing lists