loaded. Added more comments. Fixed the perspective of a 24-bit
device from the original sun4 code to match the sun2. For now,
we don't double-map pages into DVMA space by hand, we're back
to using pmap_enter. Be more paranoid about saving and
restoring the MMU user context.
* run_rc_command():
- add support for "poll", which by default, uses wait_for_pids() on
the list of matching processes
- in the "restart" case, call "poll" between "stop" and "start"
this fixes the situation where certain services (e.g, amd, squid) would
take a bit longer to shutdown after receiving the kill signal to shutdown,
and "/etc/rc.d/foo restart" would fail in the `start' phase.
that became apparent when UBC was added: store a pointer to
the process itself, not a pointer to ci->ci_curproc.
This gets us back to where we were before UBC went in: MP
kernels get to single-user mode, and can run processes on
both CPUs, but things go south when we try to come into
multi-user mode.
reserving RAM in the bus_mem extent map. Problem pointed
out by Artur Grabowski.
- Work around a slightly annoying bit of behavior exhibited by
the UP1000 firmware. The UP1000 firmware reports the space
consumed by the "ISA hole" in the same MDDT entry as two
chunks of RAM (on either side of the hole) used by the PALcode,
all as one "reserved for PALcode" chunk. We must take this
into account when reserving RAM in the bus_mem extent map.
* turn ad-hoc `pmap resources allocated during sleep' into a panic condition
(this should all be prevented by higher level VM locking).
* arrange for ENOMEM to be returned if PMAP_CANFAIL is on, and malloc()
and/or pool_get() fail.
tweak esp glue to read xfer_len from dma map
add some debugging checks for dma driver invariants
read DD_LIMIT instead of DD_SAVED_LIMIT on scsi dma shutdowns
vfs_busy'ing just before the dounmount() call. This is to avoid
sleeping with the mountlist_slock held -- but we must acquire
syncer_lock before vfs_busy because the syncer itself uses
syncer_lock -> vfs_busy locking order.
fredette