Hi Tomas, On Sun, Jul 29, 2018 at 7:52 PM Tomas Janousek <t...@nomi.cz> wrote: > > Hi Julian, > > On Fri, Jul 27, 2018 at 09:31:42PM +1000, Julian Calaby wrote: > > > Does adding the kernel parameter "ahci.mobile_lpm_policy=1" avoid this? > > > > No, booting with that parameter makes no difference. > > Can you try "dm_mod.use_blk_mq=0 scsi_mod.use_blk_mq=0" as well?
Disabling multi-queue fixes it and allows it to boot. I have not tested this in combination with the AHCI LPM policy change Ben suggested as just disabling multi-queue Works Fine For Me. > Block multiqueue was enabled in debian's 4.17 > (https://salsa.debian.org/kernel-team/linux/commit/049487d8822c141bef503b024e73db55e2a695ff) > but there's no support for runtime power management yet: > https://github.com/torvalds/linux/blob/a26fb01c2879ed7026e6cbd78bb701912d249eef/block/blk-core.c#L3765 This seems unwize. I saw the patches implementing runtime PM for MQ go by and am firmly in the camp of not enabling the new way to do stuff until it's (mostly) feature-complete with the old way. > Unfortunately, the above code doesn't prevent laptop-mode-tools, tlp nor > custom udev rules from enabling runtime-pm for devices so if you use any of > that, you might be having runtime-pm enabled on devices that don't support it, > and therefore get hangs. > > See https://github.com/rickysarraf/laptop-mode-tools/issues/123 for more > details. Ah! I have laptop-mode-tools installed. Thanks, -- Julian Calaby Email: julian.cal...@gmail.com Profile: http://www.google.com/profiles/julian.calaby/
