From: Mark Kettenis <mark.kettenis@xs4all.nl>
Subject: Re: qwz: enable suspend/resume
To: Marcus Glocker <marcus@nazgul.ch>
Cc: tech@openbsd.org
Date: Sun, 07 Jun 2026 10:34:58 +0200

> Date: Sun, 7 Jun 2026 08:39:07 +0200
> From: Marcus Glocker <marcus@nazgul.ch>
> 
> This makes suspend/resume with qwz(4) work on my Samsung Galaxy Book4
> Edge.
> 
> Ok?

So qwx(4) calls _core_deinit() from _stop(), whereas qwz(4) has:

        /*
         * Firmware stays running across ifconfig down/up; the chip is
         * only released on driver detach.  Do not clear pdevs_active
         * or call qwz_core_deinit() here.
         */

In that context, this diff makes sense.  But I do wonder whether
qwx(4) should do the same thing.  Things to consider are:

1. Are there any power savings to be had by stopping the firmware on
   ifconfig down?

2. Are there firmware differences that justify that qwz(4) diverges
   from qwx(4) here.

> Index: sys/dev/ic/qwz.c
> ===================================================================
> RCS file: /cvs/src/sys/dev/ic/qwz.c,v
> diff -u -p -u -p -r1.38 qwz.c
> --- sys/dev/ic/qwz.c	26 May 2026 14:55:16 -0000	1.38
> +++ sys/dev/ic/qwz.c	7 Jun 2026 06:01:46 -0000
> @@ -24857,8 +24857,14 @@ qwz_activate(struct device *self, int ac
>  			qwz_stop(ifp);
>  			rw_exit(&sc->ioctl_rwl);
>  		}
> +		if (sc->fw_initialized)
> +			qwz_core_deinit(sc);
>  		break;
>  	case DVACT_RESUME:
> +		err = qwz_hal_srng_init(sc);
> +		if (err)
> +			printf("%s: could not initialize hal\n",
> +			    sc->sc_dev.dv_xname);
>  		break;
>  	case DVACT_WAKEUP:
>  		if ((ifp->if_flags & (IFF_UP | IFF_RUNNING)) == IFF_UP) {
> 
>