- Set mem->am_nseg, particularly before trying to use it.
- Avoid error-prone local copies in agp_i810_bind_memory_main.
- Omit missing DMA unload/free in agp_i810_unbind_memory.
Previously this read the MSAC register from the wrong PCI device.
Experiments reveal that even on the right PCI device, per the
datasheet, the GMADR BAR size is correct.
Not clear that the driver has any business reading any of these GMCH
registers to discern the aperture size -- seems to me they are meant
to be written by the BIOS to determine the GMADR BAR size. Lacking a
variety of hardware to test, though, I'll leave this as is for now...
no route to it, assume there are no quotas. While this might sound like
an impossible scenario, it actually happens inside rump tests when we have
a virtual shmif network but are querying quotas for / which happens to
be on NFS (but of course outside of the shmif setup).
This fixes tests/fs/nfs/t_rquotad on diskless clients.
Principal reason is that the GTT size is not necessarily the same as
the aperture size: the GPU may have a bigger virtual address space
than the CPU can see through the aperture.
While here, factor the code a little more legibly and name some magic
constants.
(dom_alive), create a state enumeration (domainstates) and use it
instead.
Instead of three states (new, alive, and, effectively, 'troubled') go
to five: new, alive, pinging, lost, and dead.
Domains start in the NEW state. When we get a reply from a server, the
state goes to ALIVE. The state is set to PINGING when we ping the
server (once a minute normally) and if the ping times out, it goes to
LOST. If we stay lost for a minute, go to DEAD, and in DEAD, do
exponential backoff of nag_servers calls.
Getting rid of the broken logic attached to the 'troubled' state fixes
PR 15355 (ypbind defeats disk idle spindown) -- it will now only
rewrite the binding file when the binding changes.
Also, fix the HEURISTIC code so it doesn't trigger except in ALIVE
state. I think this was the source of a lot of the spamming behavior
seen in PR 32519, which is now fixed.
Might also fix PR 23135 (broadcast ypbind sometimes fails to find
servers).
Processing arguments just sets flags -- may as well do it first, and
this way detection of silly errors isn't contingent on having things
fully configured and operating.
"evil", this isn't one of them. Since the logic involved is to wait
until the default domain binds before backgrounding, call the variable
"started" instead.
- Make struct agp_i810_softc::gatt specific to i810 chipsets; use other
members of struct agp_i810_softc for non-i810 chipsets.
- agp_i810_init detects and sets isc->gtt_size.
- Map GTT based on the GTT size detected by agp_i810_init.
- Sprinkle some comments particularly about questionable calculations.
- net.sdl for the active link-layer adddress (the MAC)
- net.ether.multicast for the Ethernet multicast addresses
- net.inet6.multicast for the IPv6 multicast groups
- net.inet6.multicast_kludge for temporarily removed multicast groups
Use this sysctls for replacing the kmem grovelling in ifmcstat(8).
