[PATCH] synchronize_irq needs a barrier

[Herbert Xu]

On Fri, Oct 19, 2007 at 02:26:54PM +1000, Benjamin Herrenschmidt wrote:
>
> I think a simple smp_mb(); here after foo = 1; is enough, which means
> basically just having an smp_mp(); inside napi_synchronize(), before
> the test_bit(). Or do I miss something ?

Yes I think you're right.  In this case we do have barriers
everywhere else so this should work.

Although if you want napi_synchronize to have the property that
when it returns all NAPI processing effects are visible then
you'd need another smp_mb() after the loop.

Cheers,
-- 
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert at gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt