Re: [PATCH v3] PCI: rcar-gen4: Add missing PM ops

On 9/9/26 4:17 AM, Koichiro Den wrote:

Hello Den-san,

quoted

quoted

The s2idle test with pm_test=platform now passes in EP mode on S4 Spider even
with CONFIG_PCIE_DW_HOST=y. Thanks! Feel free to use the tag for it if it helps.

Tested-by: Koichiro Den <redacted>

Thank you for your test. I would like to ask -- is the EP side of PCIe link
behaving correctly during suspend/resume cycle on your S4 setup ? Does the
EP require any special handling during suspend/resume ?

To be clear, what I confirmed was the following:

- The panic I observed with v1 no longer occurs with this newer revision.
- I could create a vNTB EPF after running the test.
- With vNTB/ntb_netdev already connected before the test. ping resumed normally
   afterwards.

Note: I used the suspend test from the commit message on the EP:
   echo s2idle > /sys/power/mem_sleep && \
   echo platform > /sys/power/pm_test && \
   echo mem > /sys/power/state

Based on these, I believe this patch introduces no regression in EP mode on S4.
Any pre-existing EP suspend/resume issues could be addressed separately.

Understood.

Thank you for your help !

-- 
Best regards,
Marek Vasut