[ Splitting my replies up ] On Wed, Oct 13, 2021 at 04:45:50PM +0200, François Ozog wrote:
Le mer. 13 oct. 2021 à 15:30, Tom Rini trini@konsulko.com a écrit :
On Wed, Oct 13, 2021 at 03:12:02PM +0200, François Ozog wrote:
Le mer. 13 oct. 2021 à 14:55, Tom Rini trini@konsulko.com a écrit :
On Wed, Oct 13, 2021 at 10:15:02AM +0200, François Ozog wrote:
Le sam. 18 sept. 2021 à 15:18, Tom Rini trini@konsulko.com a
écrit :
On Sat, Sep 18, 2021 at 03:38:45AM -0600, Simon Glass wrote: > Hi, > > On Fri, 10 Sept 2021 at 16:44, Tom Rini trini@konsulko.com
wrote:
> > > > On Sat, Sep 11, 2021 at 12:09:40AM +0200, Mark Kettenis wrote: > > > > Date: Fri, 10 Sep 2021 17:17:37 -0400 > > > > From: Tom Rini trini@konsulko.com > > > > > > > > On Fri, Sep 10, 2021 at 11:12:20PM +0200, Mark Kettenis
wrote:
> > > > > > Date: Fri, 10 Sep 2021 08:34:20 -0400 > > > > > > From: Tom Rini trini@konsulko.com > > > > > > > > > > > > On Fri, Sep 10, 2021 at 10:38:17AM +0200, Heinrich
Schuchardt
wrote: > > > > > > > > > > > > > > > > > > > > > On 9/9/21 10:10 PM, Simon Glass wrote: > > > > > > > > At present some of the ideas and techniques behind devicetree in U-Boot > > > > > > > > are assumed, implied or unsaid. Add some
documentation
to
cover how > > > > > > > > devicetree is build, how it can be modified and the
rules
about using > > > > > > > > the various CONFIG_OF_... options. > > > > > > > > > > > > > > > > Signed-off-by: Simon Glass sjg@chromium.org > > > > > > > > Reviewed-by: Marcel Ziswiler <
marcel.ziswiler@toradex.com>
> > > > > > > > --- > > > > > > > > > > > > > > > > Changes in v3: > > > > > > > > - Fix typos linst suppled receive EFL > > > > > > > > - Drop 'and' before 'self-defeating' > > > > > > > > - Reword mention of control of QEMU's devicetree
generation
> > > > > > > > - Add mention of dropping CONFIG_OF_BOARD > > > > > > > > - Clarify the 'Once this bug is fixed' paragraph a
bit
> > > > > > > > - Expand ways that CONFIG_OF_PRIOR_STAGE can
support
the
U-Boot devicetree > > > > > > > > - Add a note at the top explaining that his patch
covers
'now', not 'future' > > > > > > > > - Add note 'Note: Some boards use a devicetree in
U-Boot
which does not match' > > > > > > > > > > > > > > > > Changes in v2: > > > > > > > > - Fix typos per Sean (thank you!) and a few others > > > > > > > > - Add a 'Use of U-Boot /config node' section > > > > > > > > - Drop mention of dm-verity since that actually
uses
the
kernel cmdline > > > > > > > > - Explain that OF_BOARD will still work after these changes (in > > > > > > > > 'Once this bug is fixed...' paragraph) > > > > > > > > - Expand a bit on the reason why the 'Current
situation'
is bad > > > > > > > > - Clarify in a second place that Linux and U-Boot
use
the
same devicetree > > > > > > > > in 'To be clear, while U-Boot...' > > > > > > > > - Expand on why we should have rules for other
projects in
> > > > > > > > 'Devicetree in another project' > > > > > > > > - Add a comment as to why devicetree in U-Boot is
not
'bad
design' > > > > > > > > - Reword 'in-tree U-Boot devicetree' to 'devicetree
source
in U-Boot' > > > > > > > > - Rewrite 'Devicetree generated on-the-fly in
another
project' to cover > > > > > > > > points raised on v1 > > > > > > > > - Add 'Why does U-Boot have its nodes and
properties?'
> > > > > > > > - Add 'Why not have two devicetrees?' > > > > > > > > > > > > > > > > doc/develop/index.rst | 1 + > > > > > > > > doc/develop/package/devicetree.rst | 583 +++++++++++++++++++++++++++++ > > > > > > > > doc/develop/package/index.rst | 1 + > > > > > > > > 3 files changed, 585 insertions(+) > > > > > > > > create mode 100644
doc/develop/package/devicetree.rst
> > > > > > > > > > > > > > > > diff --git a/doc/develop/index.rst
b/doc/develop/index.rst
> > > > > > > > index 83c929babda..d5ad8f9fe53 100644 > > > > > > > > --- a/doc/develop/index.rst > > > > > > > > +++ b/doc/develop/index.rst > > > > > > > > @@ -36,6 +36,7 @@ Packaging > > > > > > > > :maxdepth: 1 > > > > > > > > > > > > > > > > package/index > > > > > > > > + package/devicetree > > > > > > > > > > > > > > > > Testing > > > > > > > > ------- > > > > > > > > diff --git a/doc/develop/package/devicetree.rst b/doc/develop/package/devicetree.rst > > > > > > > > new file mode 100644 > > > > > > > > index 00000000000..b1bd310d906 > > > > > > > > --- /dev/null > > > > > > > > +++ b/doc/develop/package/devicetree.rst > > > > > > > > @@ -0,0 +1,583 @@ > > > > > > > > +.. SPDX-License-Identifier: GPL-2.0+ > > > > > > > > + > > > > > > > > +Updating the devicetree > > > > > > > > +======================= > > > > > > > > + > > > > > > > > +Note: This documentation describes how things are
today,
mostly, with some > > > > > > > > +mention of things that need to be fixed. It is not intended to point the way to > > > > > > > > +what might be done in the future. That should be
the
subject of discussions on > > > > > > > > +the mailing list. > > > > > > > > + > > > > > > > > +U-Boot uses devicetree for runtime configuration
and
storing required blobs or > > > > > > > > +any other information it needs to operate. It is
possible
to update the > > > > > > > > +devicetree separately from actually building
U-Boot.
This
provides a good degree > > > > > > > > +of control and flexibility for firmware that uses
U-Boot
in conjunction with > > > > > > > > +other project. > > > > > > > > + > > > > > > > > +There are many reasons why it is useful to modify
the
devicetree after building > > > > > > > > +it: > > > > > > > > + > > > > > > > > +- Configuration can be changed, e.g. which UART
to use
> > > > > > > > +- A serial number can be added > > > > > > > > +- Public keys can be added to allow image
verification
> > > > > > > > +- Console output can be changed (e.g. to select
serial or
vidconsole) > > > > > > > > + > > > > > > > > +This section describes how to work with
devicetree to
accomplish your goals. > > > > > > > > + > > > > > > > > +See also :doc:`../devicetree/control` for a basic
summary
of the available > > > > > > > > +features. > > > > > > > > + > > > > > > > > + > > > > > > > > +Devicetree source > > > > > > > > +----------------- > > > > > > > > + > > > > > > > > +Every board in U-Boot must include a devicetree sufficient to build and boot > > > > > > > > +that board on suitable hardware (or emulation).
This
is
specified using the > > > > > > > > +`CONFIG DEFAULT_DEVICE_TREE` option. > > > > > > > > + > > > > > > > > + > > > > > > > > +Current situation (August 2021) > > > > > > > > +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > > > > > > > > + > > > > > > > > +As an aside, at present U-Boot allows `CONFIG_DEFAULT_DEVICE_TREE` to be empty, > > > > > > > > +e.g. if `CONFIG_OF_BOARD` or
`CONFIG_OF_PRIOR_STAGE`
are
used. This has > > > > > > > > +unfortunately created an enormous amount of
confusion
and
some wasted effort. > > > > > > > > +This was not intended and this bug will be fixed
soon.
> > > > > > > > + > > > > > > > > +Some of the problems created are: > > > > > > > > + > > > > > > > > +- It is not obvious that the devicetree is coming
from
another project > > > > > > > > + > > > > > > > > +- There is no way to see even a sample devicetree
for
these platform in U-Boot, > > > > > > > > + so it is hard to know what is going on, e.g.
which
devices are typically > > > > > > > > + present > > > > > > > > + > > > > > > > > +- The other project may not provide a way to
support
U-Boot's requirements for > > > > > > > > + devicetree, such as the /config node. Note: On
the
U-Boot mailing list, this > > > > > > > > + was only discovered after weeks of discussion
and
confusion > > > > > > > > + > > > > > > > > +- For QEMU specifically, consulting two QEMU
source
files
is required, for which > > > > > > > > + there are no references in U-Boot
documentation. The
code is generating a > > > > > > > > + devicetree, with some control from command-line
args,
but it is not clear > > > > > > > > + how to add properties required by U-Boot. > > > > > > > > + > > > > > > > > +Specifically on the changes in U-Boot: > > > > > > > > + > > > > > > > > +- `CONFIG_OF_BOARD` was added in rpi_patch_ for
Raspberry
Pi, which does have > > > > > > > > + an in-tree devicetree, but this feature has
since
been
used for boards that > > > > > > > > + don't > > > > > > > > +- `CONFIG_OF_PRIOR_STAGE` was added in bcm_patch_
as
part
of a larger Broadcom > > > > > > > > + change with a tag indicating it only affected
one
board, so the change in > > > > > > > > + behaviour was not noticed at the time. It has
since
been used by RISC-V qemu > > > > > > > > + boards. > > > > > > > > + > > > > > > > > +Note: It is not clear that we actually need both
of
these. Possibly > > > > > > > > +`CONFIG_OF_BOARD` can be dropped. > > > > > > > > + > > > > > > > > +Once this bug is fixed, CONFIG_OF_BOARD and CONFIG_OF_PRIOR_STAGE will override > > > > > > > > > > > > > > What does "bug" refer to? Above you describe the
current
design not a bug. > > > > > > > > > > > > The bug is that we have two options to provide
seemingly
the
same > > > > > > functionality. Is there a functional difference
between
CONFIG_OF_BOARD > > > > > > and CONFIG_OF_PRIOR_STAGE ? > > > > > > > > > > With CONFIG_OF_BOARD there is a function that returns the pointer to > > > > > the DTB, so you can do all sort of things with it. > > > > > > > > > > With CONFIG_OF_PRIOR_STAGE there is a variable that you
need
to
set in > > > > > low-level code to point at the DTB and there is a
pre-defined
function > > > > > that returns that pointer. > > > > > > > > > > CONFIG_OF_BOARD is more flexible than
CONFIG_OF_PRIOR_STAGE,
but
if > > > > > the only thing you want to do is to pass on a DTB that is
passed
in a > > > > > CPU register to U-Boot then CONFIG_OF_PRIOR_STAGE is
probably
easier > > > > > to use. > > > > > > > > > > I'm not convinced there is a bug here. > > > > > > > > Thanks for explaining. Couldn't CONFIG_OF_PRIOR_STAGE be rewritten as > > > > an implementation of CONFIG_OF_BOARD, possibly at the same
or
less
> > > > overall code size? That I think is the potential bug. > > > > > > Probably a little bit more code: > > > > > > void * > > > board_fdt_blob_setup(void) > > > { > > > return (void *)(uintptr_t)prior_stage_fdt_address; > > > } > > > > Tiny bit more. Probably worth doing to make the choices
clearer on
> > which to select when? Bin, Rick, thoughts on this since riscv
is
the
> > main user of CONFIG_OF_PRIOR_STAGE at this point? > > Bin, Rick? > > What is the prior stage in the RISC-V stage? Could we get it to
set
up
> a bloblist? Then we can add a devicetree in there, with the
option to
> add more things in future.
I'm suggesting we don't need to do anything upstream of us, just
rework
things to use the other hook for "provided a DTB by caller, use
it", so
that we have a single hook for that.
What was the rationale in posting in kernel.org (
https://lore.kernel.org/all/20211003125134.2.I7733f5a849476e908cc51f0c71b8a5...
) and not in U-Boot knowing there is still no consensus on the big
picture ?
Well, because we need to get our bindings reviewed and made official, and that looked like a reasonable place and choice to start with? v2 cc'd a different set of lists, at Rob's suggestion.
making it “ Official” does not necessarily mean in the Linux DT. What
would
you think with a hardware description file that comes with U-Boot,
No? A project-specific binding is only valid within the project. So unless it's a linux,foo binding it's not "Linux DT", it's generic. Part of the point in pushing this up for real review is that IF we want to be passed this information then it needs to be a real official and reviewed binding. And that process in turn meanings convincing other (generally skeptical) people it's a good idea. And maybe that it should take a more generic approach.
LinuxBoot and EDK2 config options ? No need for Linux to know about those U-Boot internal config? Why would those U-Boot elements be passed down to any OS ?
I don't see the harm?
U-Boot per say does not define a standardized contract handover to OS. Multiboot, AVB, Linux and UEFi do.
I disagree. Linux, in general / looking forward, is trying to not have N ad-hoc per-architecture entry points. Even as I understand it (and I might have some of the exact terminology wrong, sorry) on x86_64, grub/systemd-boot/etc on UEFI systems (the modern, and even not so modern these days, norm) can set things up such that the kernel is launched by UEFI. Just like it would do for Free/Open/NetBSD or Windows. That standard defined in parts by UEFI and ACPI where who shall provide what. And sometimes that means "here's the standard way to add some info to the ACPI tables".
I'll admit I don't know how exactly on the x86_64 world information gets passed around between the secure/non-secure world and all of those other fun challenge that we're talking about here.
Looking at https://www.gnu.org/software/grub/manual/multiboot2/multiboot.html it's certainly written as a specification, but I don't know off-hand how widely it's used right now.
Assuming AVB in this context means Android Verified Boot, well, I guess at this point I'm now going to fail to not make this joke: https://xkcd.com/927/
It makes sense to expose some stuff related to the contract in the DT, added by the relevant “contract manager” in U-Boot. Should one want to define further VBE for instance with information that is useful to OS, then it make sense to have VBE information exposes. Exposing information by U-Boot for its own consumption is like an information leak: something that is part of a non existing ABI that can create bad legacy.
I think Rob's feedback on Simon's thread is helpful here, and the information in question possibly more belongs strictly under the /chosen node where this kind of information is already present. I would encourage you to chime in on the thread there.