[openib-general] Re: [PATCH][3/26] IB/mthca: improve CQ locking part 1

Roland Dreier roland at topspin.com
Thu Mar 3 16:40:14 PST 2005


> @@ -783,6 +777,11 @@
>  			  cq->cqn & (dev->limits.num_cqs - 1));
>  	spin_unlock_irq(&dev->cq_table.lock);
>  +	if (dev->mthca_flags & MTHCA_FLAG_MSI_X)
> +		synchronize_irq(dev->eq_table.eq[MTHCA_EQ_COMP].msi_x_vector);
> +	else
> +		synchronize_irq(dev->pdev->irq);
> +

    Jeff> Tangent: I think we need a pci_irq_sync() rather than
    Jeff> putting the above code into each driver.

The problem with trying to make it generic is that mthca has multiple
MSI-X vectors, and only the driver author could know that we only need
to synchronize with the completion event vector.

 - R.



More information about the general mailing list