[rt2x00-users] [PATCH] Fix RX and TX:

Vánek vanek.cz at email.cz
Wed Jul 15 06:44:09 EST 2009


jdete s tim do prdelel , SHITT
> ------------ Původní zpráva ------------
> Od: Ivo van Doorn <ivdoorn at gmail.com>
> Předmět: Re: [rt2x00-users] [PATCH] Fix RX and TX:
> Datum: 14.7.2009 22:42:40
> ----------------------------------------
> Hi,
> 
> Thanks for the patch, I can't comment on the entire patch right now,
> but I will look into the patch in a few days.
> 
> But with this patch, what is the exact status of the rt2800pci driver for
> your device?
> 
> > Move the register wpdma reset around to fix rx (and reverse the gpio value
> > sent to rfkill).
> 
> I believe there was some discussion earlier about this, and it turned out that
> the current intepretation of the bit was correct.
> 
> > For TX : make SD_LEN0 the length of the TXWI plus the full frame (header
> > +data).
> > Also for tx part the "kick tx" loop over queue entries to help survive the
> > remaining errors (something is wrong in the SD_LEN0 and/or MDP_TOTAL_BYTES
> > count for certain frames) longer.
> > 
> > Various small ghost glitches (desc_size behing TXWI_DESC_SIZE instead
> > of TXD_DESC_SIZE for beacon ... harmless as they are the same length
> > but for consistency lets fix thoses.
> > 
> > There are hacks in tx processing:
> > - in my old tests I had to add 1 to queue->index send via the TXWI to
> > the interrupt handler otherwise it was not passed (I remove the 1 from
> > the value in the interrupt handler). This has to be tested again adn removed
> > if not needed.
> > - well the queue loop in tx kick is good but the errors it helps survives
> > sometimes happens more than 23 times (the size of the queue) in a raw and in
> > this case one has to unload/reload rt2800pci driver to reinit the queue.
> > One should probably test a few padding and/or roundup if the skb->len of all
> > those missed queue entries is the same . Could be otherwise though I had
> > a lot of issues that revolved around this size matter. Could be txdesc->l2pad
> > related. I had not l2pad in the tests I made to get the tx working somewhat
> > so hard to tell.
> > - last is the queue entry_index that I don't get from the WirelessCLID (WCID)
> > due to well it was not working maybe for unrelated reasons (SD_LEN0 values I
> > passed back then where pretty in the wild). If it is correct the hack (remove
> 1 from Q_INDEX , bet it is no good for entry Q_INDEX == 0 ... well if one has
> > time to fix this one too either way (by using WCID or removing 1 properly to
> > Q_INDEX) it would be fixed.
> > 
> > Signed-off-by: Alban Browaeys <prahal at yahoo.com>
> > ---
> >  drivers/net/wireless/rt2x00/rt2800pci.c |  136
> +++++++++++++++++++++++++------
> >  1 files changed, 110 insertions(+), 26 deletions(-)
> 
> Thanks.
> 
> Ivo
> 
> 
> _______________________________________________
> users mailing list
> users at rt2x00.serialmonkey.com
> http://rt2x00.serialmonkey.com/mailman/listinfo/users_rt2x00.serialmonkey.com
> 
> 
> 



More information about the users mailing list