[openib-general] [PATCH] mthca updates (2.6.8 dependent)

[Grant Grundler]

On Sat, Aug 14, 2004 at 09:36:53AM -0700, Roland Dreier wrote:
...
> Finally, I changed the logic to unconditionally reset the HCA during
> initialization, since I've found that the previous initialization
> logic may not work when handed an HCA in an unknown state.

This is good. Different firmware (BIOS, EFI, OpenBoot, etc) may
leave the adapter in a slightly different state. I've been bitten
by this twice in the past 18 monthes.

> MSI-X seems to improve performance somewhat, because it allows the
> driver to register three separate interrupt vectors and avoid having
> to do a slow MMIO read of the event cause register in the interrupt
> handler.  I don't know if there's much point to plain MSI, because
> Linux only allows a driver to have a single interrupt even in MSI mode.

Well, MSI was intended to allow devices to use multiple consective
vectors. I forgot Linux only allows one and why.

MSI is useful with one if the driver can avoid all MMIO read
in the normal interrupt case. MSI guarantees strong ordering
of DMA to host mem and delivery of the interrupt.

grant