soc: silabs: siwx91x: link pm state to power domain and adjust driver #100116
Conversation
This patch is needed to block the pm_state "PM_STATE_SUSPEND_TO_IDLE" when a device on the power domain (actually all the peripherals) is active. Without this patch, cpu can decide to go to deep sleep while a peripheral is active. Signed-off-by: Martin Hoff <[email protected]>
| } | ||
|
|
||
| /* Get the power management policy state lock */ | ||
| if (!data->zephyr_channel_info[channel].channel_active) { |
There was a problem hiding this comment.
Should we also call pm_device_runtime_put() and pm_device_runtime_get() in DMA drivers to ensure that the user doesn’t have to explicitly manage this for memory-to-memory transfers?
There was a problem hiding this comment.
I wonder to add that like intel has done here: dma_intel_adsp_hda.c
I will try but I fear that we miss some cases that introduce some bugs.
There was a problem hiding this comment.
that's done, seems to works
Added calls to pm_device_runtime_put() to ensure proper device runtime management when configuration or DMA transceive operations fail. Signed-off-by: Martin Hoff <[email protected]>
Martinhoff-maker
force-pushed
the
siwx91x_pm_correction
branch
from
3dfaad9 to
8523b0e
Compare
| /* Perform DMA transceive */ | ||
| ret = gspi_siwx91x_transceive_dma(dev, config); | ||
| if (ret < 0) { | ||
| pm_device_runtime_put(dev); |
There was a problem hiding this comment.
What happens if asynchronous==false && dma=true && ret==0 ?
There was a problem hiding this comment.
Since this PR is mainly about PM, would do it in another PR.
| if (ret < 0) { | ||
| pm_device_runtime_put(dev); | ||
| } | ||
| spi_context_release(&data->ctx, ret); |
There was a problem hiding this comment.
BTW, is the context released when asynchronous==true && dma=true && ret<0 ?
There was a problem hiding this comment.
same as above comment.
Martinhoff-maker
force-pushed
the
siwx91x_pm_correction
branch
from
8523b0e to
e60e513
Compare
Removed unnecessary power management state lock calls since the DMA is on a power domain that already handles state locking. It will now managed directly pm_device which will active power_domain and then block pm_state. Signed-off-by: Martin Hoff <[email protected]>
This patch fix a reg write that needs to be done before calling the callback. Since the user can start a new dma transfer in the callback, it can miss the interrupt from the new transfer if the new transfer is done before the interrupt bit is cleared. Signed-off-by: Martin Hoff <[email protected]>
Removed conditional checks for device runtime put in DMA RX and TX callbacks, ensuring that device runtime is always released asynchronously. Signed-off-by: Martin Hoff <[email protected]>
Martinhoff-maker
force-pushed
the
siwx91x_pm_correction
branch
from
e60e513 to
9f1da5c
Compare
|
zephyrbot
added
platform: Silabs
zephyrbot
requested review from
JordanYates,
anangl,
asmellby,
bjarki-andreasen,
ceolin,
decsny,
jhedberg,
nashif,
tbursztyka and
teburd
6 participants
The goal of this PR is to have correlation between pm state of the cpu and the soc power domain.
It also correct the driver behavior that are on the power domain.