[Prism54-devel] [patch 0/6] [prism54 svn trunk] various fixes for prism54 svn

[jean-baptiste.note at wanadoo.fr]

Dear list, Joerg, Bob,

This is a set of patches to try and somehow improve the current SVN
state. Please tell me if the format is incorrect, i'm still trying to
learn the ropes. They should apply cleanly on prism54 svn trunk, one
after another.

* First patch fixes the problem on rmmoding the driver. this allows
testing to be done properly. The problem was fairly trivial (but
important), confusing skb for skb->data. You should try it, Joerg !

* Second patch make the rx in skbs, and frees the skb when done. This
brings nothing to the table _now_, but it can be used to get rid of the
needless memcpy in the mgmt rx path. To be done soon if this patch is
accepted.

* Third and forth patch implement Rainer's shortening of the irq path.

* Fifth puts the cleanup work in a tasklet. I'm not sure this is a good
idea, but we do a big chunk of work in this codepath, and my
understanding was that we shouldn't do all of this in an interrupt
handler. However, i've not taken great care to use tasklet_disable but
when removing the driver. This may need some more work.

* Sixth brings a fixup from the softmac driver. I don't know if the case
can happen with fullmac cards, but it costs practically nothing to do
the check.

I have hope that those patches will fix the mgmt timeout issue that a
number of people are experimenting. I don't know the precise bugzilla
numbers associated with them, but i'd really but thankful if people
affected could try out these patches.

In particular, could you try these before resorting to the windows
driver fix, Bob ?

Cheers,
JB

--
Jean-Baptiste Note
+33 (0)6 83 03 42 38
jean-baptiste.note at wanadoo.fr