Re: bwfm(4) PCI support

[Jason Thorpe <thorpej%me.com@localhost>]

I’ll also take a look at this later today (I’ve also been working on bwfm, so want to compare changes to provide additional feedback).

(Sorry for the top-post.)

-- thorpej
Sent from my iPhone.

> On May 4, 2018, at 3:15 AM, Jared McNeill <jmcneill%invisible.ca@localhost> wrote:
> 
>> On Fri, 4 May 2018, coypu%sdf.org@localhost wrote:
>> 
>> http://coypu.sdf.org/0001-sync-with-openbsd-bwfm.patch
> 
> Two issues:
> 
> @@ -307,6 +313,12 @@ bwfm_start(struct ifnet *ifp)
>            continue;
>        }
> 
> +        if (sc->sc_bus_ops->bs_txcheck(sc)) {
> +            IF_PREPEND(&ifp->if_snd, m);
> +            ifp->if_flags |= IFF_OACTIVE;
> +            break;
> +        }
> +
>        IFQ_DEQUEUE(&ifp->if_snd, m);
>        if (m == NULL)
>            break;
> 
> Remove the IF_PREPEND here. You only need to do that after dequeue. In this case, m will always be NULL.
> 
> -
> 
> +int
> +bwfm_usb_txcheck(struct bwfm_softc *bwfm)
> +{
> +    struct bwfm_usb_softc *sc = (void *)bwfm;
> +
> +    mutex_enter(&sc->sc_tx_lock);
> +
> +    if (TAILQ_EMPTY(&sc->sc_tx_free_list)) {
> +        mutex_exit(&sc->sc_tx_lock);
> +        return ENOBUFS;
> +    }
> +
> +    return 0;
> +}
> 
> Missing mutex_exit in the success path here.
> 
>> http://coypu.sdf.org/0004-Update-bwfm-4-to-note-PCI-support-and-mention-jmcnei.patch
> 
> I wouldn't say I wrote the driver, only ported it. Also note capitalization of my last name.
> 
> Cheers,
> Jared