Since bounce_user_kiobuf() is required anyway, I would prefer this
aproach:
it looks like a 2 line fix.
> [ or: ]
> The harder problem is to
> hide it inside ll_rw_block(), which we _do_ want to do if we move the
> page cache to high memory.
Is that planed for 2.[34]?
If not, then I'd wait: I don't like a large in-advance solution for a
not exactly defined problem [possible OOM problems]
Note that currently, kswapd automagically ignores pages from bigmem if
it cannot allocate the bounce buffer [mm/vmscan.c, line 159]. It will be
difficult to do that in ll_rw_block().
-- Manfred- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/