On 4/28/25 8:36 AM, Bryan Brattlof wrote:
On April 25, 2025 thus sayeth Andrew Davis:
Just like TF-A and OP-TEE, the documentation states a custom path for DM
can be provided at build time by setting TI_DM. This should then set
ti-dm-path which updates ti-dm node filenames in binman.

Two issues prevent this from functioning for most K3 boards. One is when
then DM firmware name is inside a blob-ext node instead of a ti-dm node.

The second is when the filename in the ti-dm node is a pointer to a
blob-ext node. In this case even though the filename is updated, the
filename in the blob-ext is not, so build can fail if the default
file in the blob-ext cannot be found, even if the updated ti-dm file
does exist.

Fix both of these for all K3 by removing any indirect ti-dm nodes and
making sure all DM nodes are labeled with "ti-dm".

Signed-off-by: Andrew Davis <a...@ti.com>

For the few sitars boards this works :)

Reviewed-by: Bryan Brattlof <b...@ti.com>

---
  arch/arm/dts/k3-am625-beagleplay-u-boot.dtsi  | 11 ++-----
  arch/arm/dts/k3-am625-phycore-som-binman.dtsi | 15 +++-------
  arch/arm/dts/k3-am625-sk-binman.dtsi          | 14 +++------
  .../dts/k3-am625-verdin-wifi-dev-binman.dtsi  | 11 ++-----
  arch/arm/dts/k3-am62a-phycore-som-binman.dtsi | 14 +++------
  arch/arm/dts/k3-am62a-sk-binman.dtsi          | 14 +++------
  arch/arm/dts/k3-am62p-sk-binman.dtsi          | 11 ++-----
  arch/arm/dts/k3-am67a-beagley-ai-u-boot.dtsi  | 12 ++------
  arch/arm/dts/k3-am69-sk-u-boot.dtsi           | 29 +++++++++++++------
  arch/arm/dts/k3-j7200-binman.dtsi             | 13 +++------
  .../dts/k3-j721e-beagleboneai64-u-boot.dtsi   | 11 ++-----
  arch/arm/dts/k3-j721e-binman.dtsi             | 13 +++------
  arch/arm/dts/k3-j721s2-binman.dtsi            | 13 +++------
  arch/arm/dts/k3-j722s-binman.dtsi             | 12 ++------
  arch/arm/dts/k3-j742s2-evm-u-boot.dtsi        | 29 +++++++++++++------
  arch/arm/dts/k3-j784s4-binman.dtsi            |  8 ++---
  arch/arm/dts/k3-j784s4-evm-u-boot.dtsi        | 29 +++++++++++++------
  17 files changed, 104 insertions(+), 155 deletions(-)


...

diff --git a/arch/arm/dts/k3-am625-sk-binman.dtsi b/arch/arm/dts/k3-am625-sk-binman.dtsi
index 534eb14795b..cc619f5920e 100644
--- a/arch/arm/dts/k3-am625-sk-binman.dtsi
+++ b/arch/arm/dts/k3-am625-sk-binman.dtsi
@@ -156,14 +156,6 @@
  #define AM625_SK_DTB "u-boot.dtb"
&binman {
-       ti-dm {
-               filename = "ti-dm.bin";
-               blob-ext {
-                       filename = 
"ti-dm/am62xx/ipc_echo_testb_mcu1_0_release_strip.xer5f";
-                       optional;
-               };
-       };
-
        tifsstub-hs {
                filename = "tifsstub.bin_hs";
                ti-secure-rom {
@@ -270,7 +262,8 @@
                                                keyfile = "custMpk.pem";
                                        };
                                        dm: ti-dm {
-                                               filename = "ti-dm.bin";
+                                               filename = 
"ti-dm/am62xx/ipc_echo_testb_mcu1_0_release_strip.xer5f";
+                                               optional;
                                        };

So I was curious of this optional flag. Is the issue here that we only
check for mandatory binaries via BINMAN_INDIRS and not through all the
custom arguments like TI_DM?


I would like to remove all these optional flags for DM here at some point.
DM is not optional for these platforms, without it you will not have a
functional U-Boot.

Looking at the history on the list, seems these were added to make automatic
testing with buildman happy. Marking these optional made the build pass for
test builds without needing to download our firmware repo. 
BINMAN_ALLOW_MISSING=1
is the solution but it was broken at one point. It seems to be working now
and these optional flags can go away.

I didn't do it as part of this series to keep things simple and as they were
before, but feel free to start removing them. Some platforms already have
this non-optional (AM62x Verdin, BeaglePlay).

Andrew

Reply via email to