Re: [PATCH 0/3] Semi-Falcon Boot Support

Tom Rini <[email protected]>
Newsgroups org.u-boot-project.lists.u-boot
Message-ID <20260817212208.GW3297518@bill-the-cat>
On Mon, Aug 17, 2026 at 01:46:10PM +0530, Gokul Praveen wrote:
> Hi Tom,
> 
> On 14/08/26 20:26, Tom Rini wrote:
> > On Fri, Aug 14, 2026 at 03:51:50PM +0530, Gokul Praveen wrote:
> > 
> > > Semi-Falcon Mode is a hybrid boot approach that bypasses
> > > the A72 SPL while retaining U-Boot in the boot flow. This mode provides faster
> > > boot times compared to normal boot while maintaining U-Boot's flexibility for
> > > kernel loading and system configuration..
> > > 
> > > Boot logs link :
> > > 
> > > https://gist.github.com/GokulPraveen2001/fef407a482e2dad30272006de557627c
> > > 
> > > Gokul Praveen (3):
> > >    configs: add semi-falcon mode fragment for k3 devices
> > >    arm: k3-binman: add tisemifalcon.bin for semi-falcon mode
> > >    doc: ti: document semi-falcon mode for J784S4 EVM
> > This is just doing what many other aarch64 platforms do, of skipping the
> > SPL phase in Cortex-A because something else has already done any
> > required initialization, unless I'm reading all of this wrong. So my
> > feedback is:
> > - It's not "semi-falcon", it's just "no SPL". Naming differences, if
> >    any, should be done with that in mind.
> 
> Actually, there is an SPL in Cortex-R5 which is still present in the
> semi-falcon bootflow.

Yes. But it's not "semi-falcon", at all, as "falcon" refers to booting
directly to Linux.

> Its only the Cortex-A SPL which is avoided, which is why I did not name it
> as "no SPL"
> 
> *Normal K3 Bootflow*: R5 SPL + TFA + OPTEE + DM + Cortex-A SPL + UBOOT +
> Kernel
> 
> *Semi-Falcon* *boot flow* is : R5 SPL + TFA + OPTEE + DM + UBOOT + Kernel

Yes, and this is similar to other platforms, where the Cortex-R or
similar isn't exposed in the same fashion as it is for TI.

> > - You should make sure to structure this so any K3 platform *could* take
> >    advantage. I seem to recall part of the rationale for the current
> >    approach is that some variants might only have a little SRAM (512k?
> >    256k?) while others might have 1-2M and so easily fit U-Boot in SRAM
> >    prior to relocation. It sounds like this specific family has a large
> >    enough lower memory bound.
> > 
> Yes, this is structured in a way that all K3 platforms can take advantage of
> this approach and changes.
> 
> Additionally, as there is an SPL running in cortex-R5 even with this
> bootflow(semi-falcon).
> 
> this bootflow can be used on all variants irrespective of whether they have
> less or more memory.

OK good. Just come up with a different way of describing this that
doesn't say "falcon", thanks.

-- 
Tom
signature.asc (application/pgp-signature, 228 B)
-----BEGIN PGP SIGNATURE-----

iHUEABYKAB0WIQTzzqh0PWDgGS+bTHor4qD1Cr/kCgUCaoN7fwAKCRAr4qD1Cr/k
CgOyAQCBrXlklFYrHX2TsDgzS11Kda1zlEFvCnqSNUGLoBVSWQD+L1Q9KWFVhY38
m24/RYb7KCPqWf/ccKTjWcQd3FdscAg=
=i3XT
-----END PGP SIGNATURE-----
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.