On Thu, May 27, 2021 at 7:40 PM Christoph Hellwig <hch@xxxxxx> wrote: > > On Thu, May 27, 2021 at 12:24:06AM +0800, Kai-Heng Feng wrote: > > Yes, that's exactly what they said. Because Windows Modern Standby > > always keep the NVMe at D0, so D3hot is untested by the vendors. > > So all the problems we've deal with latetly are that platforms cut the > power off. So it is not kept always at D0 which would be the right thing, > but moved into D3cold. It's still okay to use D3cold under s2idle if nvme_acpi_storage_d3() check passes. Otherwise D0 should be used. I am unsure how to proceed with HMB NVMes if we can't setup DMA after resume... > > Which makes this patch to disable D3cold rather counterintuitive. The intention of this patch is indeed very unclear. Kai-Heng