On Tue, 10 Sept 2024 at 11:26, Grumbach, Emmanuel <emmanuel.grumbach@xxxxxxxxx> wrote: > > On Tue, 2024-09-10 at 10:23 +0100, Chris Bainbridge wrote: > > On Sat, 7 Sept 2024 at 22:21, Chris Bainbridge > > <chris.bainbridge@xxxxxxxxx> wrote: > > > > > > It's possible that the > > > HP BIOS turns off the wifi when suspended on battery power > > > > This is not the case. Suspending while on battery power and then > > resuming (still on battery power) works without error. > > > > If the laptop is docked, and powered via both the dock USB-C *and* a > > mains PSU on the separate PSU connector, then suspended, then either > > the dock or the PSU are disconnected, then the error happens on > > resume, even though the laptop is continuously mains powered via the > > other power supply. It appears there's something about a powered > > dock/PSU unplug event while suspended that causes the error, as > > opposed to just being suspended while running on battery. > > Ok.. so that's really pointing in the direction of the BIOS or the system's firmware. Sure we made a > change in the driver that is now sensitive to that problem so it pops up. > Since wifi is available for you after resume, I think it's better not to fix that for the coming > release which is almost out. I'll funnel the fix patch I sent through the regular channel but you'll > see errors even with that patch. Was this ever fixed? I see there was some kind of fix that was picked up by regzbot but it looks like it was for a different issue caused by the same original commit. I'm now on a different laptop, and hitting the "Timeout waiting for hardware access (CSR_GP_CNTRL 0x08040008) ... iwl_mvm_check_rt_status failed, device is gone during suspend" error on every suspend/resume cycle (even if laptop remains plugged in while suspended). I see it's also been reported elsewhere too, e.g. https://www.reddit.com/r/debian/comments/1g7yook/iwlwifi_crashes_after_resume_from_s3_sleep_on/ I know the wifi is still functional after the error, but spamming the kernel log with a stack trace and register dump seems excessive given that the BIOS turning off power to the wifi when suspended seems reasonably common.
