MACCHIATObin

[Marcin Wojtas]

Hi,

pon., 29 paź 2018 o 21:05 Greg V <greg at unrelenting.technology> napisał(a):
>
>
>
> On Mon, Oct 29, 2018 at 10:46 PM, Emmanuel Vadot
> <manu at bidouilliste.com> wrote:
> > On Mon, 29 Oct 2018 22:37:59 +0300
> > Greg V <greg at unrelenting.technology> wrote:
> >
> >>
> >>
> >>  On Sat, Aug 18, 2018 at 10:07 AM, Marcin Wojtas <mw at semihalf.com>
> >> wrote:
> >>  > sob., 18 sie 2018 o 08:43 Marcin Wojtas <mw at semihalf.com>
> >> napisa?(a):
> >>  >>
> >>  >>  sob., 18 sie 2018 o 01:10 Jim Thompson <jim at netgate.com>
> >>  >> napisa?(a):
> >>  >>  >
> >>  >>  > On Aug 17, 2018, at 5:42 PM, Mark Millard via freebsd-arm
> >>  >> <freebsd-arm at freebsd.org> wrote:
> >>  >>  >
> >>  >>  > > Marcin Wojtas mw at semihalf.com wrote on
> >>  >>  > > Fri May 11 07:55:40 UTC 2018 :
> >>  >>  > >
> >>  >>  > >> Short status of the support - last year we enabled most of
> >> the
> >>  >>  > >> platform functionalities (core support, USB, AHCI, RTC).
> >> Three
> >>  >> big
> >>  >>  > >> items remained left:
> >>  >>  > >> - PCIE root complex (this should work soon with the work
> >> done
> >>  >> for
> >>  >>  > >> another SoC, not merged yet)
> >>  >>  > >> - Network PPv2
> >>  >>  > >> - Xenon SD/MMC controller
> >>  >>  > >
> >>  >>  > >
> >>  >>  > > I noticed a check in that deals with that "Xenon"
> >>  >>  > > SD/MMC controller:
> >>  >>  > >
> >>  >>  > > Author: loos
> >>  >>  > > Date: Tue Aug 14 16:33:30 2018
> >>  >>  > > New Revision: 337772
> >>  >>  > > URL:
> >>  >>  > > https://svnweb.freebsd.org/changeset/base/337772
> >>  >>
> >>  [?]
> >>  >>  > >
> >>  >>  > > But I've not noticed check-ins for the other of the "big
> >>  >>  > > items" going by. (Though they may have.)
> >>  >>  > >
> >>  >>  > > Care to comment-on/update-the actual status for the
> >>  >>  > > Macchiato.bin(s)?
> >>  >>  >
> >>  >>  >
> >>  >>  > tl;dr: It?s not ?there? yet.
> >>  >>  >
> >>  >>  > Specifics:
> >>  >>  >
> >>  >>  > - loos@ has done a lot of work to get the espresso.bin
> >> working,
> >>  >> and some of this carries over to the 8k/7k.
> >>  >>  > - We have another internal developer working on an EIP-97
> >> driver
> >>  >> for crypto offload.  This should be a foundation for the EIP-197
> >> in
> >>  >> the 8K.
> >>  >>  > - I know manu at is working on pin controllers and clocks
> >> specific
> >>  >> to Machiatto.bin (80x0/70x0).
> >>  >>  > - Getting the NICs, PCIe, etc working still remains to be
> >> done.
> >>  >>  >
> >>  >>
> >>  >>  About the latter - the NIC is pretty complex, however we
> >> (Semihalf)
> >>  >>  have really huge experience with all its support
> >> implementations and
> >>  >>  the platform itself. I'll put it straightforward - it's only a
> >>  >> matter
> >>  >>  of development funding, if it's guaranteed, we will do it with
> >>  >>  pleasure :) As well as the NIC support (DW Synopsys driver for
> >> DT
> >>  >> and
> >>  >>  verify/improve on pcie-host-generic with ACPI).
> >>  >
> >>  > "As well as PCIE support..." of course.
> >>  >
> >>  > Marcin
> >>
> >>  Hi,
> >>
> >>  I've noticed that people successfully use ACPI-PCIe on the
> >> MACCHIATObin
> >>  with Linux,
> >>  so? has anyone tested FreeBSD pcie-host-generic there?
> >>
> >>  I'm considering buying a MACCHIATObin board to test my drm-next
> >> build
> >>  for aarch64
> >>  ( https://github.com/FreeBSDDesktop/kms-drm/pull/89 ), but I'd like
> >> to
> >>  first know if PCIe actually works :)
> >
> >  I'll will push support for mcbin soon in the tree (with PCI support).
> >  I don't know what ACPI-PCIe is exactly but I don't see how it can
> > work
> > without a dedicated driver for the armada 8K.
>
> dev/pci/pci_host_generic_acpi.c — generic ECAM PCIe driver.
>
> I think ThunderX2 uses it (2018 Cavium copyright in the file), probably
> Ampere eMAG (haven't seen anything about that here, but the vendor
> lists FreeBSD in their OS compatibility list), should be possible to
> use on the SoftIron Overdrives (NetBSD just added ACPI/aarch64 and they
> tested on the Overdrive:
> https://dmesgd.nycbug.org/index.cgi?do=view&id=4526 — but all
> FreeBSD+SoftIron dmesgs on dmesgd.nycbug so far are in FDT mode, not
> ACPI)…
>
> And the EDK2 firmware for the Armada8k does have an ACPI mode…
>

I confirm - thanks to the initial configuration in UEFI, ACPI tables
support PCI, which OS can use with the pure ECAM pci-host-generic
driver. Here's also the latest Centos7 bootlog, where you can see it:
https://pastebin.com/Aa2NvDnM

Marcin