Re: [RESEND PATCH v3 1/3] libfdt: Add function to find regions in an FDT

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]



Hi David,

On 5 June 2014 at 08:12, David Gibson <david@xxxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Mon, Jun 02, 2014 at 09:12:55PM -0600, Simon Glass wrote:
> > Hi David,
> >
> > On 1 May 2014 23:40, David Gibson <david@xxxxxxxxxxxxxxxxxxxxx> wrote:
> > > On Thu, May 01, 2014 at 09:13:51AM -0700, Simon Glass wrote:
> > >> Hi,
> > >>
> > >> On 22 March 2014 08:07, Simon Glass <sjg@xxxxxxxxxxxx> wrote:
> > >> > Given a set of nodes and properties, find the regions of the device tree
> > >> > which describe those parts.
> > >> >
> > >> > A test is provided which builds a tree while tracking where the regions
> > >> > should be, then calls fdt_first/next_region() to make sure that it agrees.
> > >> >
> > >> > Further tests will come as part of fdtgrep.
> > >> >
> > >> > Signed-off-by: Simon Glass <sjg@xxxxxxxxxxxx>
> > >>
> > >> Have I sent this to the right place? Any comments?
> > >
> > > Sorry.  I know you've resent this several times, and I've been
> > > procrastinating about it since forever.
> > >
> > > Basically, I'm just not convinced.  For all your efforts to explain
> > > the rationale, it just seems like a really ad-hoc set of flags and
> > > conditions that doesn't seem to form a coherent whole.
> >
> > Are you referring to fdtgrep or the new library function? I'd be happy
> > enough getting the library function to start with in if there isn't
> > much need for the grep utility.
>
> The library function primarily.
>
> > > From the lack of other responses, I'm assuming there's not really
> > > anyone else who sees it as a compelling feature either.
> >
> > Is there normally a lot of mailing list traffic for new libfdt
> > features? This is used for verified boot in U-Boot for example.
>
> Hm, that's true I guess.

I'd like to dredge this up again.

I'm currently working on adding device tree support in U-Boot SPL (the
tiny loader that loads U-Boot). U-Boot currently uses device tree for
normal operation, but since SPL can run in very small RAM sizes there
is often not enough space for a full (~40KB) device tree. In fact
often all that is needed is the UART and perhaps SPI/I2C. This is
<1KB.

There seem to be a several options here:

1. Create a separate device tree file for each board, just for SPL,
with just what is needed to boot SPL
2. Use #ifdef and the preprocessor to remove what is not needed, and
run dtc twice with different #defines
3. Use fdtgrep to chop down the device tree file by removing unwanted nodes

I think that option 3 is the best. It avoids copying source code, or
littering it with #ifdef.

If necessary I suppose this tool could just live in U-Boot but that
seems a shame. I understand that fdtgrep perhaps has too many options
for your liking. Is there some subset that you think would be worthy
of including as a new device-tree-compiler tool?

If so, then I would rather create fdtgrep (or fdtprune or whatever)
there than add a special tool to U-Boot.

Please let me know what you think.

Regards,
Simon
--
To unsubscribe from this list: send the line "unsubscribe devicetree-compiler" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [Device Tree]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]

  Powered by Linux