Matthew Dillon wrote:
quoted text
> :That's the perfect solution!
> :
> :Patch appended. _lwkt_trytokref needs the same change, as it might 
> :acquire the token as well. I think setting lastref = NULL in 
> :lwkt_reltoken is also not absolutely neccessary, but it's not wrong
> :either, so I did it.
> :
> :Regards,
> :
> :   Michael
 >
quoted text
> [...]
 >
quoted text
>     This will require some surgery, would you like to have a go at it?  
>     Despite the work needed it won't actually complicate the code any more,
>     it's more on the order of a rearrangement.

A patch is appended.

lwkt_token_is_stale(lwkt_tokref_t) detects if the token was acquired by
another thread while the thread was sleeping. I don't care about
tokrefs, instead the granularity is a thread, so that recursive calls to
acquire a token within the same thread DO NOT stale out a token.
Furthermore the code for UP and SMP is now basically the same, except
for spin-locks or preemption issues. That is, I keep a per-thread count
instead of the globalcount on UP and maintain thread ownership as well.
I noticed that _lwkt_trytokref and _lwkt_gettokref were basically using
the same code (it was copy & pasted), so I unified that as well.

Regards,

   Michael