On Wed May 21, 2025 at 01:02:46PM +0200, Mark Kettenis wrote:
> > Date: Wed, 21 May 2025 11:36:47 +0200
> > From: Rafael Sadowski <raf...@sizeofvoid.org>
> >
> > On Sat May 10, 2025 at 10:03:21PM +0200, Mark Kettenis wrote:
> > > > Date: Sun, 4 May 2025 19:20:38 +0200
> > > > From: Rafael Sadowski <raf...@sizeofvoid.org>
> > > >
> > > > On Wed Apr 23, 2025 at 11:36:41PM +0200, Mark Kettenis wrote:
> > > > > > Date: Wed, 23 Apr 2025 14:16:14 +0200
> > > > > > From: Rafael Sadowski <raf...@sizeofvoid.org>
> > > > > >
> > > > > > Unfortunately, my new T14gen5 does not want to switch the suspend
> > > > > > state.
> > > > > > The red LED flashes quickly all the time but nothing happens.
> > > > >
> > > > > That may just mean the machine suspended but didn't reach the deepest
> > > > > sleep state.
> > > > >
> > > > > > Only a hard reset (hold the power button for 10 seconds) will help.
> > > > >
> > > > > That may just mean the methods you tried to wake it up didn't work.
> > > > >
> > > > > > If there is something (patch/flag..) I can use to build the kernel
> > > > > > or
> > > > > > contribute some information I would be happy to do so.
> > > > >
> > > > > Please try what happens if you press the power button when the machine
> > > > > is up. The machine should do a clean shutdown. But if that doesn't
> > > > > happen the power button handling is borked and pressing it won't wake
> > > > > up the laptop from suspend.
> > > >
> > > > It looks like the power button under OpenBSD is broken. Nothing happens
> > > > when I press the power button for 1,2,5 seconds. The fingerprint sensor
> > > > is disabled.
> > >
> > > That's the point where you make sure I have acpidump(8) output for
> > > this machine.
> > >
> >
> > When I run zzz(1) with ddb.suspend=1 and I can see:
> >
> > xhci1: command ring abort timeout
> > xhci2: command ring abort timeout
> >
> > Everything else printed with "detached". So i built an kernel with
> > XHCI_DEBUG, dmesg see below.
>
> These messages aren't necessarily a problem. At least I expect the machine
> would still resume (but maybe it would throw a few more errors.q
>
> What happens if you set the XHCI_NOCSS flag for these controllers in
> dev/pci/xhci_pci.c?
case PCI_VENDOR_AMD:
- if (PCI_PRODUCT(pa->pa_id) == PCI_PRODUCT_AMD_17_1X_XHCI_1 ||
- PCI_PRODUCT(pa->pa_id) == PCI_PRODUCT_AMD_17_1X_XHCI_2)
psc->sc.sc_flags |= XHCI_NOCSS;
break;
Unfortunately no luck.
