From: Elliott Mitchell <ehem+undef@m5p.com>
To: Rob Herring <robh@kernel.org>
Cc: Stefano Stabellini <sstabellini@kernel.org>,
Julien Grall <julien.grall.oss@gmail.com>,
Elliott Mitchell <ehem+xen@m5p.com>,
xen-devel <xen-devel@lists.xenproject.org>,
Volodymyr Babchuk <Volodymyr_Babchuk@epam.com>,
julien@xen.org
Subject: Re: Question on PCIe Device Tree bindings, Was: [PATCH] xen/arm: domain_build: Ignore device nodes with invalid addresses
Date: Thu, 4 Feb 2021 19:32:13 -0800 [thread overview]
Message-ID: <YBy8PbYeLEHjcELY@mattapan.m5p.com> (raw)
In-Reply-To: <CAL_Jsq+cedzG5NBfLRub=msZEK6umBrk-O7FYB=Dk34=k9fuCA@mail.gmail.com>
On Thu, Feb 04, 2021 at 03:52:26PM -0600, Rob Herring wrote:
> On Thu, Feb 4, 2021 at 3:33 PM Stefano Stabellini
> <sstabellini@kernel.org> wrote:
> >
> > On Thu, 4 Feb 2021, Rob Herring wrote:
> > > On Thu, Feb 4, 2021 at 2:36 PM Stefano Stabellini
> > > <sstabellini@kernel.org> wrote:
> > > >
> > > > On Thu, 4 Feb 2021, Rob Herring wrote:
> > > > > On Thu, Feb 4, 2021 at 11:56 AM Stefano Stabellini
> > > > > <sstabellini@kernel.org> wrote:
> > > > > >
> > > > > > Hi Rob,
> > > > > >
> > > > > > We have a question on the PCIe device tree bindings. In summary, we have
> > > > > > come across the Raspberry Pi 4 PCIe description below:
> > > > > >
> > > > > >
> > > > > > pcie0: pcie@7d500000 {
> > > > > > compatible = "brcm,bcm2711-pcie";
> > > > > > reg = <0x0 0x7d500000 0x0 0x9310>;
> > > > > > device_type = "pci";
> > > > > > #address-cells = <3>;
> > > > > > #interrupt-cells = <1>;
> > > > > > #size-cells = <2>;
> > > > > > interrupts = <GIC_SPI 148 IRQ_TYPE_LEVEL_HIGH>,
> > > > > > <GIC_SPI 148 IRQ_TYPE_LEVEL_HIGH>;
> > > > > > interrupt-names = "pcie", "msi";
> > > > > > interrupt-map-mask = <0x0 0x0 0x0 0x7>;
> > > > > > interrupt-map = <0 0 0 1 &gicv2 GIC_SPI 143
> > > > > > IRQ_TYPE_LEVEL_HIGH>;
> > > > > > msi-controller;
> > > > > > msi-parent = <&pcie0>;
> > > > > >
> > > > > > ranges = <0x02000000 0x0 0xc0000000 0x6 0x00000000
> > > > > > 0x0 0x40000000>;
> > > > > > /*
> > > > > > * The wrapper around the PCIe block has a bug
> > > > > > * preventing it from accessing beyond the first 3GB of
> > > > > > * memory.
> > > > > > */
> > > > > > dma-ranges = <0x02000000 0x0 0x00000000 0x0 0x00000000
> > > > > > 0x0 0xc0000000>;
> > > > > > brcm,enable-ssc;
> > > > > >
> > > > > > pci@1,0 {
> > > > > > #address-cells = <3>;
> > > > > > #size-cells = <2>;
> > > > > > ranges;
> > > > > >
> > > > > > reg = <0 0 0 0 0>;
> > > > > >
> > > > > > usb@1,0 {
> > > > > > reg = <0x10000 0 0 0 0>;
> > > > > > resets = <&reset RASPBERRYPI_FIRMWARE_RESET_ID_USB>;
> > > > > > };
> > > > > > };
> > > > > > };
> > > > > >
> > > > > >
> > > > > > Xen fails to parse it with an error because it tries to remap reg =
> > > > > > <0x10000 0 0 0 0> as if it was a CPU address and of course it fails.
> > > > > >
> > > > > > Reading the device tree description in details, I cannot tell if Xen has
> > > > > > a bug: the ranges property under pci@1,0 means that pci@1,0 is treated
> > > > > > like a default bus (not a PCI bus), hence, the children regs are
> > > > > > translated using the ranges property of the parent (pcie@7d500000).
> > > > > >
> > > > > > Is it possible that the device tree is missing device_type =
> > > > > > "pci" under pci@1,0? Or is it just implied because pci@1,0 is a child of
> > > > > > pcie@7d500000?
> > > > >
> > > > > Indeed, it should have device_type. Linux (only recently due to
> > > > > another missing device_type case) will also look at node name, but
> > > > > only 'pcie'.
> > > > >
> > > > > We should be able to create (or extend pci-bus.yaml) a schema to catch
> > > > > this case.
> > > >
> > > > Ah, that is what I needed to know, thank you! Is Linux considering a
> > > > node named "pcie" as if it has device_type = "pci"?
> > >
> > > Yes, it was added for Rockchip RK3399 to avoid a DT update and regression.
> > >
> > > > In Xen, also to cover the RPi4 case, maybe I could add a check for the
> > > > node name to be "pci" or "pcie" and if so Xen could assume device_type =
> > > > "pci".
> > >
> > > I assume this never worked for RPi4 (and Linux will have the same
> > > issue), so can't we just update the DT in this case?
> >
> > I am not sure where the DT is coming from, probably from the RPi4 kernel
> > trees or firmware. I think it would be good if somebody got in touch to
> > tell them they have an issue.
>
> So you just take whatever downstream RPi invents? Good luck.
>
> > Elliot, where was that device tree coming from originally?
Please excuse my very weak device-tree fu...
I'm unsure which section is the problem, but looking at `git blame` what
shows is commt d5c8dc0d4c880fbde5293cc186b1ab23466254c4.
This commit is present in the Linux master branch and the linux-5.10.y
branch.
You were saying?
--
(\___(\___(\______ --=> 8-) EHM <=-- ______/)___/)___/)
\BS ( | ehem+sigmsg@m5p.com PGP 87145445 | ) /
\_CS\ | _____ -O #include <stddisclaimer.h> O- _____ | / _/
8A19\___\_|_/58D2 7E3D DDF4 7BA6 <-PGP-> 41D1 B375 37D0 8714\_|_/___/5445
next prev parent reply other threads:[~2021-02-05 3:33 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-02 17:47 [PATCH] xen/arm: domain_build: Ignore device nodes with invalid addresses Elliott Mitchell
2021-02-02 18:12 ` Julien Grall
2021-02-02 19:21 ` Julien Grall
2021-02-03 0:18 ` Stefano Stabellini
2021-02-03 17:44 ` Julien Grall
2021-02-03 22:18 ` Stefano Stabellini
2021-02-03 22:52 ` Julien Grall
2021-02-04 0:13 ` Stefano Stabellini
2021-02-04 16:29 ` Julien Grall
2021-02-04 17:56 ` Question on PCIe Device Tree bindings, Was: " Stefano Stabellini
2021-02-04 18:31 ` Rob Herring
2021-02-04 20:36 ` Stefano Stabellini
2021-02-04 21:08 ` Rob Herring
2021-02-04 21:33 ` Stefano Stabellini
2021-02-04 21:52 ` Rob Herring
2021-02-04 22:02 ` Stefano Stabellini
2021-02-05 3:32 ` Elliott Mitchell [this message]
2021-02-05 13:56 ` Rob Herring
2021-02-04 22:39 ` Stefano Stabellini
2021-02-06 8:47 ` Julien Grall
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=YBy8PbYeLEHjcELY@mattapan.m5p.com \
--to=ehem+undef@m5p.com \
--cc=Volodymyr_Babchuk@epam.com \
--cc=ehem+xen@m5p.com \
--cc=julien.grall.oss@gmail.com \
--cc=julien@xen.org \
--cc=robh@kernel.org \
--cc=sstabellini@kernel.org \
--cc=xen-devel@lists.xenproject.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).