svn commit: r298612 - head/sys/dev/iwm

Andriy Voskoboinyk avos at freebsd.org
Tue Apr 26 06:27:29 UTC 2016


Tue, 26 Apr 2016 07:40:59 +0300 було написано Adrian Chadd  
<adrian at freebsd.org>:

> Author: adrian
> Date: Tue Apr 26 04:40:59 2016
> New Revision: 298612
> URL: https://svnweb.freebsd.org/changeset/base/298612
>
> Log:
>   [iwm] implement suspend/resume through ieee80211_{suspend,resume}_all
>  This allows wifi to associate correctly after a suspend/resume cycle.
>  Yes, I'm using this now day to day.
>  Tested:
>  * Intel 7260AC, STA mode
>
> Modified:
>   head/sys/dev/iwm/if_iwm.c
>   head/sys/dev/iwm/if_iwmvar.h
>
> Modified: head/sys/dev/iwm/if_iwm.c
> ==============================================================================
> --- head/sys/dev/iwm/if_iwm.c	Tue Apr 26 03:24:28 2016	(r298611)
> +++ head/sys/dev/iwm/if_iwm.c	Tue Apr 26 04:40:59 2016	(r298612)
> @@ -4934,6 +4934,8 @@ iwm_init_task(void *arg1)
>  static int
>  iwm_resume(device_t dev)
>  {
> +	struct iwm_softc *sc = device_get_softc(dev);
> +	int do_reinit = 0;
>  	uint16_t reg;
> 	/* Clear device-specific "PCI retry timeout" register (41h). */
> @@ -4941,17 +4943,33 @@ iwm_resume(device_t dev)
>  	pci_write_config(dev, 0x40, reg & ~0xff00, sizeof(reg));
>  	iwm_init_task(device_get_softc(dev));
> +	IWM_LOCK(sc);
> +	if (sc->sc_flags & IWM_FLAG_DORESUME) {
> +		sc->sc_flags &= ~IWM_FLAG_DORESUME;
> +		do_reinit = 1;

If no vap was running, then nothing will be started
(because of IEEE80211_FEXT_RESUME flag).

> +	}
> +	IWM_UNLOCK(sc);
> +
> +	if (do_reinit)
> +		ieee80211_resume_all(&sc->sc_ic);

AFAIK, suspend_all() / resume_all() should work without any
additional logic
(except (?) device-specific PCI registers (41h) pre-setup on resume).

> +
>  	return 0;
>  }
> static int
>  iwm_suspend(device_t dev)
>  {
> +	int do_stop = 0;
>  	struct iwm_softc *sc = device_get_softc(dev);
> -	if (sc->sc_ic.ic_nrunning > 0) {
> +	do_stop = !! (sc->sc_ic.ic_nrunning > 0);
> +
> +	ieee80211_suspend_all(&sc->sc_ic);
> +
> +	if (do_stop) {
>  		IWM_LOCK(sc);
>  		iwm_stop(sc);

Is it really needed? (iwm_stop() will be already executed when the last vap
will be stopped + suspend_resume() waits for ic_parent method).

> +		sc->sc_flags |= IWM_FLAG_DORESUME;
>  		IWM_UNLOCK(sc);
>  	}
>
> Modified: head/sys/dev/iwm/if_iwmvar.h
> ==============================================================================
> --- head/sys/dev/iwm/if_iwmvar.h	Tue Apr 26 03:24:28 2016	(r298611)
> +++ head/sys/dev/iwm/if_iwmvar.h	Tue Apr 26 04:40:59 2016	(r298612)
> @@ -405,6 +405,7 @@ struct iwm_softc {
>  #define IWM_FLAG_STOPPED	(1 << 2)
>  #define IWM_FLAG_RFKILL		(1 << 3)
>  #define IWM_FLAG_BUSY		(1 << 4)
> +#define	IWM_FLAG_DORESUME	(1 << 5)
> 	struct intr_config_hook sc_preinit_hook;
>  	struct callout		sc_watchdog_to;


More information about the svn-src-head mailing list