Re: [PATCH 1/1] vmalloc: purge_fragmented_blocks: Acquire spinlockbefore reading vmap_block

[David Rientjes]

On Thu, 8 Dec 2011, Kautuk Consul wrote:

> > In the original code, if the if-clause fails, the lock is only then taken
> > and the exact same test occurs again while protected. ÂIf the test now
> > fails, the lock is immediately dropped. ÂA branch here is faster than a
> > contented spinlock.
> 
> But, if there is some concurrent change happening to vb->free and
> vb->dirty, dont you think
> that it will continue and then go to the next vmap_block ?
> 
> If yes, then it will not be put into the purge list.
> 

That's intentional as an optimization, we don't care if 
vb->free + vb->dirty == VMAP_BBMAP_BITS && vb->dirty != VMAP_BBMAP_BITS 
would speculatively be true after we grab vb->lock, we'll have to purge it 
next time instead.  We certainly don't want to grab vb->lock for blocks 
that aren't candidates, so this optimization is a singificant speedup.