pgcl-2.5.64-2 fixes up the initial attempt at swapping so it actually
passes touch testing. Whether shmem.c handles this properly is
uncertain, but nothing obvious stands out as being in need of repair.
I bumped down PAGE_SIZE to 32K, 64K appeared to trigger issues with
qlogicisp.c on NUMA-Q. That may or may not eventually get fixed. It'd
certainly be interesting to go beyond the limits of the 2.4 patch.
Dealing with q->max_sectors*512 < PAGE_SIZE in a generic fashion looks
painful, and the driver is crusty, so swapping hardware sound good.
If this really holds up and there aren't enough fs or driver issues
to keep me occupied I might actually start dealing with performance
issues wrt. fragmentation of anonymous memory and L3 pagetables.
(Well, they're L2 pagetables in the nomenclature starting at L0.)
As usual, available from:
ftp://ftp.kernel.org/pub/linux/kernel/people/wli/vm/pgcl/
-- wli
-
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/