[PATCH] libertas: fix spinlock recursion bug

[Holger Schurig]

Hi Dan !

I like most of your proposition. A few things:


> struct lbs_event {
> 	u32 event;  /* MACREG_INT_CODE_xxxxx */
> 	u32 len;
> 	u8 buf[LBS_UPLD_SIZE];
> };

Ugh, as LBS_UPLD_SIZE equals to to 2312, this looks ugly. I'm not 
sure if accessing the cmd_ctrl_node list directly would be a 
good idea, thought.


> Seems better than what's there.

Definitely. Maybe David Woodhouse can give his opinion as well.


> I might take a stab at this if people think it seems sane
> enough.

Do you have time for such a thing?  I see that you don't ack some 
of my patches, probably because you don't even have time to 
check them. What time frame do you have in mind?


> The memory hit over what's there now would be ~15K of allocated
> memory at driver load per card.

The memory hit won't be nice for embedded devices. But maybe it's 
better to try this solution first, to see if it's working. Then 
we have all the time for optimizations, e.g. with two lists.