Re: [Qemu-devel] [PATCHv2 3/3] Split ISA and sysbus versions of m48t59 device

[Paolo Bonzini <pbonzini@redhat.com>]

----- Original Message -----
> From: "Mark Cave-Ayland" <mark.cave-ayland@ilande.co.uk>
> To: "Paolo Bonzini" <pbonzini@redhat.com>, "David Gibson" <david@gibson.dropbear.id.au>, "edgar iglesias"
> <edgar.iglesias@gmail.com>, michael@walle.cc, proljc@gmail.com, borntraeger@de.ibm.com, "cornelia huck"
> <cornelia.huck@de.ibm.com>, kbastian@mail.uni-paderborn.de, jcmvbkbc@gmail.com
> Cc: veroniabahaa@gmail.com, "peter maydell" <peter.maydell@linaro.org>, mst@redhat.com, armbru@redhat.com,
> qemu-devel@nongnu.org, agraf@suse.de
> Sent: Thursday, November 10, 2016 4:16:36 PM
> Subject: Re: [Qemu-devel] [PATCHv2 3/3] Split ISA and sysbus versions of m48t59 device
> 
> On 10/11/16 14:57, Paolo Bonzini wrote:
> > 
> > 
> > On 09/11/2016 13:22, David Gibson wrote:
> >> The m48t59 device supports both ISA and direct sysbus attached versions of
> >> the device in the one .c file.  This can be awkward for some embedded
> >> machine types which need the sysbus M48T59, but don't want to pull in the
> >> ISA bus code and its other dependencies.
> >>
> >> Therefore, this patch splits out the code for the ISA attached M48T59 into
> >> its own C file.  It will be built when both CONFIG_M48T59 and
> >> CONFIG_ISA_BUS are enabled.
> >>
> >> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> > 
> > Who needs the ISA M48T59?  Perhaps it should be a separate symbol
> > altogether.  Let's document that SPARC will stop providing it in 2.9,
> > for example, if it's only a PReP thing.
> > 
> > Paolo
> 
> Hi Paolo,
> 
> The ISA M48T59 is still actively used by the sun4u machine on
> qemu-system-sparc64. In real terms it's actually connected to the ebus,
> but for all intents and purposes it's the same as an ISA bus connected
> via a PCI bridge.

sun4u is actually using the sysbus M48T59, and mapping it into the ebus
space:

    nvram = m48t59_init(NULL, 0, 0, NVRAM_SIZE, 1968, 59);
    s = SYS_BUS_DEVICE(nvram);
    memory_region_add_subregion(get_system_io(), 0x2000,
                                sysbus_mmio_get_region(s, 0));

Paolo