Skip to content

Conversation

@dchat-nordic
Copy link
Contributor

Change the default IPC backend of nRF5340 to icbmsg.

Signed-off-by: Dominik Chat [email protected]
(cherry picked from commit 70419bd)

@dchat-nordic
Copy link
Contributor Author

Rebased

@dchat-nordic dchat-nordic added this to the ncs-2.8.0 milestone Oct 9, 2024
@gmarull
Copy link
Member

gmarull commented Oct 10, 2024

handled here nrfconnect/sdk-nrf#17721

@gmarull gmarull closed this Oct 10, 2024
@dchat-nordic dchat-nordic reopened this Oct 17, 2024
@dchat-nordic
Copy link
Contributor Author

Reopening since nrfconnect/sdk-nrf#17721 doesn't handle it anymore.

@dchat-nordic dchat-nordic removed this from the ncs-2.8.0 milestone Oct 17, 2024
tx-region = <&cpuapp_cpunet_ipc_shm>;
rx-region = <&cpunet_cpuapp_ipc_shm>;
tx-blocks = <32>;
rx-blocks = <32>;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since we have 32K for each direction, we will get almost 1K per block. We can increase the number of blocks, because it has almost no cost and it will give better granularity. I'm suggesting 64 blocks.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Currently it's a cherry-pick from upstream, so I would suggest limited changes here, and changing in the upstream. The openamp worked with 32 blocks as I remember and there were no issues

@doki-nordic doki-nordic self-requested a review October 17, 2024 09:34
@dchat-nordic
Copy link
Contributor Author

Adding DNM since the sdk-nrf needs alignments in overlays etc. before this is merged.

adamkondraciuk and others added 25 commits November 6, 2024 11:37
This commit allows to switch the GRTC clock source to an RC oscillator.

Upstream PR #: 80622

Signed-off-by: Adam Kondraciuk <[email protected]>
… depend on !BT

The previous changes in
zephyrproject-rtos/zephyr#74304
assumed that because this is also handled in
`bt_hci_transport_setup` that it shouldn't be done on
initialisation too.

However, if someone wants to develop their own app which
uses BT and also wants to enable the CPUNET by default this
KConfig should be available to them.

Signed-off-by: Sean Madigan <[email protected]>
(cherry picked from commit e4fa386)
A bunch of copy-paste mistakes when adding the nRF54L15 DK added stale
references to the Preview DK (PDK).

Signed-off-by: Carles Cufi <[email protected]>
(cherry picked from commit c85e418)
…-managed PDs

Add DT binding for NXP's SCU-managed PDs. This binding describes
exactly _one_ power domain.

Signed-off-by: Laurentiu Mihalcea <[email protected]>
(cherry picked from commit fb84f4f)
Renamed soc from ace30_ptl to ace30.
We were previously using the wrong soc name.
The correct name is ace30.

There is only one ptl platform, but there can be several ace30 platforms.

Signed-off-by: Grzegorz Bernat <[email protected]>
(cherry picked from commit a654bfb)
Add power domain DT node for the irqsteer and a reference to
it in irqsteer's DT node.

Signed-off-by: Laurentiu Mihalcea <[email protected]>
(cherry picked from commit fdbf4d2)
Add bus clock for `sai1` node.

Signed-off-by: Laurentiu Mihalcea <[email protected]>
(cherry picked from commit 1174ea5)
Add build_all test for power_domain drivers.

Signed-off-by: TOKITA Hiroshi <[email protected]>
(cherry picked from commit 2c1274f)
…names

Some devices may belong to >1 power domain, so with the current design
this is something not possible to describe. It's worth to note that
Linux also uses the `power-domains` naming scheme, not `power-domain`.
This patch also introduces `power-domain-names` so that each entry in
`power-domains` can be given a name if needed. `#power-domain-cells`
is now required as well.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit f989711)
Add binding for Global Power Domain found in nRF54Hx SoCs.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit f3d29d6)
Add the global power domain entry. This domain is not memory-mapped
but controlled using NRFS services.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit a56a170)
So that it can be used to manually control certain power domains.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 01e285c)
…IVE1 peripherals

This patch introduces a new flag to indicate if a peripheral belongs
to FAST_ACTIVE1 domain. This way, pinctrl knows when to request the
SLOW_ACTIVE domain (where CTRLSEL multiplexer resides).

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 9925ec9)
…ase GPD

Add a new soc-level API that allows to manually request/release global
power domains.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 87a42a8)
It is required for some pinctrl changes.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 0b18f31)
… FAST_ACTIVE1

This patch allows to _safely_ configure GPIO ports that have their pad
on FAST_ACTIVE1 domain.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 56c5b1e)
So that we do not get false warnings about consistent spacing around
'*'.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 5e8905b)
…ention

This API needs to be called by FAST peripherals before/after
disabling/enabling them.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 77fc183)
When GPD is managed by pinctrl, pins retention needs to be controlled by
the driver to avoid glitches.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit b6d4542)
It is enabled by default if we enable device PM, but we do not want
this, otherwise we get linker errors (PM subsys, fun guaranteed!).

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 969326b)
Upstream not ready yet, revert once applied there.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
Busy-waiting for the result of the nrfs service calls can stall, so
let's use a callback that flags a semaphore instead. Since the API is
supposed to be callable in the context of pre-kernel, fallback to
busy-wait on that scenario.

Upstream PR #: 80819

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 714d82d232977309e0bdc8dadd2d7a71ee914a4b)
…FIG_DEBUG=y

Usage of K_SPINLOCK with CONFIG_DEBUG=y seems to trigger a compiler
warning about request not always being initialized. Fallback to
k_spin_lock/unlock calls to fix this issue.

Signed-off-by: Gerard Marull-Paretas <[email protected]>
(cherry picked from commit 5249619)
…bmsg

Change the default IPC backend of nRF5340 to icbmsg.

Signed-off-by: Dominik Chat <[email protected]>
(cherry picked from commit 70419bd)
… bsim

Add call to Z_SPIN_DELAY() to avoid issue in bsim where
time does not progress in loop causing other threads to
starve.

Signed-off-by: Johan Stridkvist <[email protected]>
(cherry picked from commit 50c47b0)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.