Re: [Kernel-hackathon] [PATCH] urb_tx_buffer struct

the file tx_callback_h.patch has a bug, jus delete the las two lines
(signed off) and it will work fine ... (thanks Andres)

Ing. Javier Alejandro Jorge


On Sat, Sep 19, 2009 at 3:12 PM, Javier Jorge <javierbrk@???> wrote:
>
> Here is the tx_callback patch for current repository.
>
> are we going to use this functions?  do we need them ?
> ieee80211_wake_queues — wake all queues ieee80211_stop_queues — stop all queues maybe we could use them here ...
>
>     /*
>      * Take the oldest unused urb from the free anchor.
>      * This will unanchor it and get the ownership over the urb.
>      * If no free urb is available bail out
>      *
>      * TODO!! think if we want to wait instead(we'll need to
>      * signal the wait form the completion handler??)
>      */
>     urb = usb_get_from_anchor(&vt->tx_free_anchor);
>     if (!urb) {
>         error = -EAGAIN;
>         goto out;
>     }
>
> Ing. Javier Alejandro Jorge
>
>
> On Tue, Sep 8, 2009 at 5:23 PM, Kohn, Rodolfo <rodolfo.kohn@???> wrote:
>>
>> Probably this is the same thing as a structure I've created called vt6656_tx_ctx that allows vt6656_usb_tx_complete have a ptr to the skb and the vt, in my patch.
>>
>> +struct vt6656_tx_ctx {
>> +       struct vt6656_dev *vt;
>> +       struct sk_buff *skb;
>> +};
>>
>>
>>
>> Rodolfo Kohn
>> Software Engineer
>> Phone: 54-351-526-KOHN (5646)
>> Argentina Software Development Center
>> Intel
>>
>>
>> >-----Original Message-----
>> >From: kernel-hackathon-bounces@??? [mailto:kernel-hackathon-
>> >bounces@???] On Behalf Of Christoph Hellwig
>> >Sent: Tuesday, September 08, 2009 3:52 PM
>> >To: Gaston Ingaramo
>> >Cc: Mailing list for the VIA dongle driver hackathon
>> >Subject: Re: [Kernel-hackathon] [PATCH] urb_tx_buffer struct
>> >
>> >On Wed, Sep 02, 2009 at 02:44:57AM -0300, Gaston Ingaramo wrote:
>> >> Guys:
>> >>
>> >> What this patch here does is:
>> >>  * gives a new struct called urb_tx_buffer and
>> >>  * minor changes in usb_tx_complete, alloc_tx_urb and vt6656_tx to
>> >make all
>> >> work.
>> >>
>> >> It solves the problem of getting the vt6656_dev and skb  pointers
>> >inside
>> >> usb_tx_complete
>> >
>> >Good.  How does this interact with the TX bits from Rodolpho?  I'm a bit
>> >afraid to put any TX patches in while that work is going on..
>> >
>> >> BTW: I think I may've found a big mistake in
>> >*vt6656_usb_alloc_tx_urb*,
>> >> right after the "*error = 0;*" line, at the
>> >> bottom there's a "*usb_put_urb(urb)*" instruction!!!!
>> >
>> >It drops the reference we acquired, after this only the usb anchor
>> >owns a references
>> >
>> >_______________________________________________
>> >Kernel-hackathon mailing list
>> >Kernel-hackathon@???
>> >https://mx.grulic.org.ar/mailman/listinfo/kernel-hackathon
>> _______________________________________________
>> Kernel-hackathon mailing list
>> Kernel-hackathon@???
>> https://mx.grulic.org.ar/mailman/listinfo/kernel-hackathon
>