> > Such an interface can't really know what slot will be
> > picked by ide_register_hw() and can't "prepare" the HWIF with
> > special iops, so it won't be much harmed by the fact we are
> > calling init_hwif_data, but still, we should ultimately think
> > about splitting completely the fact of allocating an hwif slot,
> > setting it up, and triggering a probe on it. Those are 3 different
> > things that are currently mixed in bad ways. I don't beleive
> > fixing that fits in the 2.6 timeframe though.
>
> I just though about another possible crap, though I haven't looked
> enough to be sure, but PCI interfaces with no device will trigger
> a similar problem as "empty" ide-pmac interfaces in that sense that
> they will change dma ops, possibly mmio ops, etc.... If they hold
> no device, their hwif->present will not be set, and thus the hwif
> slot can possibly get re-used by thing like ide-cs (or anybody else
> that rely on ide_register_hw() to allocate a new slot) without
> those changes to hwif done by the PCI interface beeing cleared.
>
> So my patch may actually fix some cases there too.
No, look at ide_match_hwif() in setup-pci.c .
PCI grabs only ide_unknown interfaces.
btw, I think the only real long-term solution for all ordering issues
is customizable device mapper... 2.7?
-- Bartlomiej> Ben.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/