On 9/1/22 11:33 AM, Ryan Eatmon wrote:


On 9/1/2022 11:20, Andrew Davis wrote:
On 9/1/22 11:10 AM, Denys Dmytriyenko wrote:
On Thu, Sep 01, 2022 at 10:31:42AM -0500, Ryan Eatmon via 
lists.yoctoproject.org wrote:


On 8/31/2022 12:53, Andrew Davis wrote:
In many places we use "j7" (the family name) when we should be using a
specific SoC or EVM and vice-versa. For instance when including graphics
drivers we add for all the "j7" family, but what we really want is only
for some specific j7 SoCs.

First step in untangling this is to make it clear that there is no SoC
called "j7". There is a SoC called j721e with a j721e-evm.

Fix up this naming here.

Signed-off-by: Andrew Davis <[email protected]>
---
  .../{j7-evm-k3r5.conf => j721e-evm-k3r5.conf}  |  4 ++--
  conf/machine/{j7-evm.conf => j721e-evm.conf}   |  4 ++--
  ...sr1-1.conf => j721e-hs-evm-k3r5-sr1-1.conf} |  6 +++---
  ...hs-evm-k3r5.conf => j721e-hs-evm-k3r5.conf} |  4 ++--
  .../{j7-hs-evm.conf => j721e-hs-evm.conf}      |  6 +++---
  .../cadence-mhdp-fw/cadence-mhdp-fw_git.bb     |  2 +-
  .../ti-img-rogue-driver_1.15.6133109.bb        | 10 +++++-----
  recipes-bsp/sciclient/sciclient-rtos_git.bb    |  6 +++---
  .../viddec-test-app_1.0.0.bb                   |  2 +-
  .../videnc-test-app_1.0.0.bb                   |  2 +-
  recipes-bsp/ti-sci-fw/ti-sci-fw_git.bb         | 18 +++++++++---------
  .../trusted-firmware-a_%.bbappend              |  2 +-
  .../ti-img-rogue-umlibs_1.15.6133109.bb        |  6 +++---
  recipes-kernel/linux/kernel-rdepends.inc       |  4 ++--
  recipes-security/optee/optee-os_%.bbappend     |  2 +-
  recipes-ti/ti-rtos-bin/ti-rtos-firmware.bb     |  2 +-
  16 files changed, 40 insertions(+), 40 deletions(-)
  rename conf/machine/{j7-evm-k3r5.conf => j721e-evm-k3r5.conf} (81%)
  rename conf/machine/{j7-evm.conf => j721e-evm.conf} (87%)
  rename conf/machine/{j7-hs-evm-k3r5-sr1-1.conf => 
j721e-hs-evm-k3r5-sr1-1.conf} (53%)
  rename conf/machine/{j7-hs-evm-k3r5.conf => j721e-hs-evm-k3r5.conf} (81%)
  rename conf/machine/{j7-hs-evm.conf => j721e-hs-evm.conf} (85%)

diff --git a/conf/machine/j7-evm-k3r5.conf b/conf/machine/j721e-evm-k3r5.conf
similarity index 81%
rename from conf/machine/j7-evm-k3r5.conf
rename to conf/machine/j721e-evm-k3r5.conf


So this changes the name of the yocto build machine label...  This
will be impactful to users used to building j7-evm and for our build
and testing scripts.  I will need to test this and work with Minas
to make changes on the testing side before I can merge it in.

Right. This was a good and timely change for kirkstone, but is it worth the
effort for dunfell? Supposed to be a stable release branch and avoid breakages
like this...


I did consider that, but it seems we will be on dunfell for a little bit
longer than I expected. If we have plans to move to kirkstone in the
near future I'd have been okay with this fix only on kirkstone/master.

Holding off on valid fixes because we are waiting for a next version we
don't plan to switch to for a month+ is not a good strategy.


Is it possible to change the name of MACHINE after the code starts?  So that 
from the outside we call it j7-evm, but internally in the recipes we drop the 
j7 and go to j721e?  For dunfell, but for kirkstone we can change the conf file 
name too?



Do we gain much from that outside of not needing to change the command in the 
build/test farm?

We already made this change in kirkstone, this is basically a backport (I didn't
call it such as it is a bit different due to how much has changed between 
dunfell
and kirkstone).

We will need to update the processor docs in a couple spots but I can handle 
that
before the next Jacinto targeting release.
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#14988): 
https://lists.yoctoproject.org/g/meta-ti/message/14988
Mute This Topic: https://lists.yoctoproject.org/mt/93376998/21656
Group Owner: [email protected]
Unsubscribe: https://lists.yoctoproject.org/g/meta-ti/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to