RE: [PATCH v9 5/7] ACPI: Translate the I/O range of non-MMIO devices before scanning

[Gabriele Paoloni <gabriele.paoloni@xxxxxxxxxx>]

Hi Mika

> -----Original Message-----
> From: Mika Westerberg [mailto:mika.westerberg@xxxxxxxxxxxxxxx]
> Sent: 13 June 2017 09:49
> To: Lorenzo Pieralisi
> Cc: Gabriele Paoloni; rafael@xxxxxxxxxx; Rafael J. Wysocki;
> catalin.marinas@xxxxxxx; will.deacon@xxxxxxx; robh+dt@xxxxxxxxxx;
> frowand.list@xxxxxxxxx; bhelgaas@xxxxxxxxxx; arnd@xxxxxxxx; linux-arm-
> kernel@xxxxxxxxxxxxxxxxxxx; mark.rutland@xxxxxxx;
> brian.starkey@xxxxxxx; olof@xxxxxxxxx; benh@xxxxxxxxxxxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx; linux-acpi@xxxxxxxxxxxxxxx; Linuxarm; linux-
> pci@xxxxxxxxxxxxxxx; minyard@xxxxxxx; John Garry; xuwei (O)
> Subject: Re: [PATCH v9 5/7] ACPI: Translate the I/O range of non-MMIO
> devices before scanning
> 
> On Mon, Jun 12, 2017 at 04:57:00PM +0100, Lorenzo Pieralisi wrote:
> > I had a more in-depth look at this series and from my understanding
> > the problem are the following to manage the LPC bindings in ACPI.
> >
> > (1) Child devices of an LPC controller require special handling when
> >     filling their resources (ie they need to be translated - in DT
> >     that's guaranteed by the "isa" binding, in ACPI it has to be
> >     done by new code)
> > (2) In DT systems, LPC child devices are created by the LPC bus
> >     controller driver through an of_platform_populate() call with
> >     the LPC controller node as the fwnode root. For ACPI to work
> >     the same way there must be a way to prevent LPC children to
> >     be enumerated in acpi_default_enumeration() something like
> >     I2C does (and then the LPC driver would enumerate its children as
> >     DT does)
> >
> > I am not sure how (1) and (2) can be managed with current ACPI
> bindings
> > and kernel code - I suspect it may be done by mirroring what's done
> > for I2C but I am not sure, that's why I CC'ed Mika (ie the LPC
> adapter
> > is matched as a platform device and it takes care of enumerating its
> > children - problem though is preventing enumeration from core ACPI
> code).
> 
> Is there an example ASL showing how these LPC devices are
> currently presented in ACPI?

Please find below the asl sketch for our LPC and IPMI

//
// LPC
//

Scope(_SB) {
  Device (LPC0) {
    Name (_HID, "HISI0191")  // HiSi LPC
    Name (_CRS, ResourceTemplate () {
      Memory32Fixed (ReadWrite, 0xa01b0000, 0x1000)
    })
  }

  Device (LPC0.IPMI) {
    Name (_HID, "IPI0001")
    Method (_IFT) {
      Return (0x03)
    }
    Name (LORS, ResourceTemplate() {
      QWordIO (
        ResourceConsumer,
	MinNotFixed,     // _MIF
	MaxNotFixed,     // _MAF
	PosDecode,
	EntireRange,
	0x0,             // _GRA
	0xe4,            // _MIN
	0x3fff,          // _MAX
	0x0,             // _TRA
	0x04,            // _LEN
	, ,
	BTIO
      )
    })
    CreateQWordField (LORS, BTIO._MIN, CMIN)
    CreateQWordField (LORS, BTIO._MAX, CMAX)
    CreateQWordField (LORS, BTIO._LEN, CLEN)

    Method (_PRS, 0) {
      Return (LORS)
    }

    Method (_CRS, 0) {
      Return (LORS)
    }
    Method (_SRS, 1) {
      CreateQWordField (Arg0, \_SB.LPC0.IPMI.BTIO._MIN, IMIN)
      Store (IMIN, CMIN)
      CreateQWordField (Arg0, \_SB.LPC0.IPMI.BTIO._MAX, IMAX)
      Store (IMAX, CMAX)
    }
  } 
[...]
}

Many thanks
Gab