On Wed, Feb 23, 2022 at 06:32:56AM +0000, Krishna Yarlagadda wrote: > > > + if (has_acpi_companion(tqspi->dev)) > > > + return 0; > > > /* flush all write which are in PPSB queue by reading back */ > > > tegra_qspi_readl(tqspi, QSPI_COMMAND1); > > As well as clock stuff this is also skipping flushing of pending writes > > - is that intentional? It's not called out in the changelog and seems like > > something that could cause issues if someone runs on a system where > > the firmware does implement runtime suspend. > Runtime suspend is not enabled with ACPI firmware. Converted compiler flag in v1 to runtime check. > We must add more changes like setting DPM flags for runtime pm support with ACPI. > Can take this as part of a different series. It at least needs to be clearer what's going on here, the changelog doesn't match the code and it's not obvious from the code that ACPI won't kick in and power manage the device as things stand.
Attachment:
signature.asc
Description: PGP signature
