On 6/20/23 3:39 PM, Denys Dmytriyenko wrote:
On Tue, Jun 20, 2023 at 03:06:01PM -0500, Andrew Davis wrote:
On 6/20/23 2:17 PM, Ryan Eatmon via lists.yoctoproject.org wrote:
On 6/20/2023 1:05 PM, Denys Dmytriyenko wrote:
On Tue, Jun 20, 2023 at 12:09:25PM -0500, Ryan Eatmon via
lists.yoctoproject.org wrote:
When the ti-eth-fw recipe was split out from the ti-rtos-firmware recipe
we forgot to add it as a dependency.
Signed-off-by: Ryan Eatmon <[email protected]>
---
meta-ti-bsp/recipes-kernel/linux/kernel-rdepends.inc | 3 +++
1 file changed, 3 insertions(+)
diff --git a/meta-ti-bsp/recipes-kernel/linux/kernel-rdepends.inc
b/meta-ti-bsp/recipes-kernel/linux/kernel-rdepends.inc
index 18984e80..1f0c1ee8 100644
--- a/meta-ti-bsp/recipes-kernel/linux/kernel-rdepends.inc
+++ b/meta-ti-bsp/recipes-kernel/linux/kernel-rdepends.inc
@@ -31,3 +31,6 @@ RDEPENDS:${KERNEL_PACKAGE_NAME}-base:append:j721e = "
vxd-dec-fw"
RDEPENDS:${KERNEL_PACKAGE_NAME}-base:append:j721s2 = " cnm-wave-fw"
RDEPENDS:${KERNEL_PACKAGE_NAME}-base:append:j784s4 = " cnm-wave-fw"
RDEPENDS:${KERNEL_PACKAGE_NAME}-base:append:am62axx = " cnm-wave-fw"
+
+# Add run-time dependency for TI ETH firmware to the rootfs
+RDEPENDS:${KERNEL_PACKAGE_NAME}-base:append:k3 = " ti-eth-fw"
Though, ETH FW is only enabled on j721e, j7200 and j784s4:
https://git.yoctoproject.org/meta-ti/commit/?id=91e5a1ca6541012e3d4285479ecf5a9f353307f9
ETH_FW_LIST = ""
ETH_FW_LIST:j721e = "${ETH_FW}"
ETH_FW_LIST:j7200 = "${ETH_FW}"
ETH_FW_LIST:j721s2 = ""
ETH_FW_LIST:j784s4 = "${ETH_FW}"
ETH_FW_LIST:am65xx = ""
ETH_FW_LIST:am64xx = ""
ETH_FW_LIST:am62xx = ""
ETH_FW_LIST:am62axx = ""
Other platforms generate an empty package w/o the binary in it. So I guess for
simpler dependency pulling it for all K3 platforms is an option? On one hand
it makes kernel dependencies simpler, but executes build/packaging process for
the firmware that ends up being no-op and produce an empty output...
Andrew and I talked about. It seems annoying to have to update two files to
add a platform, BUT doing it this way does mean we execute the recipe when the
recipe does nothing...
This is sort of a holdover from the ti-rtos-firmware way of doing things.
Let me think about it...
Another thing we may want to do is not have this recipe be device specific,
it is just a small firmware, no good reason to only deploy one per device.
I'd say we ship all unconditionally in this package, then it won't have to
be re-built for each device either.
Is the binary identical between the platforms?
No, but the binaries are named differently for each platform, so they do not
conflict.
Same as most other firmware packages, ship all the related firmware in the same
package,
and let the kernel pick the firmware it actually needs based on name for the
platform
currently running.
Andrew
Andrew
Similar question for DM FW? E.g.:
https://git.yoctoproject.org/meta-ti/commit/?id=b21d29d94694ac49b97a1f4ba428c7d8cd6fb64a
DM_FW_LIST = ""
DM_FW_LIST:j721e = "${DM_FIRMWARE}"
DM_FW_LIST:j7200 = "${DM_FIRMWARE}"
DM_FW_LIST:j721s2 = "${DM_FIRMWARE}"
DM_FW_LIST:j784s4 = "${DM_FIRMWARE}"
DM_FW_LIST:am65xx = ""
DM_FW_LIST:am64xx = ""
DM_FW_LIST:am62xx = "${DM_FIRMWARE}"
DM_FW_LIST:am62axx = "${DM_FIRMWARE}"
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#16750):
https://lists.yoctoproject.org/g/meta-ti/message/16750
Mute This Topic: https://lists.yoctoproject.org/mt/99648797/21656
Group Owner: [email protected]
Unsubscribe: https://lists.yoctoproject.org/g/meta-ti/unsub
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
