| Age | Commit message (Collapse) | Author |
|---|
|
struct acpidock_softc and stuff devices into it during attach.
This list is not yet used, but I have code ready to handle the ACPI side
of this. However, it still doesn't handle that there may in fact be
multiple dock devices in a machine (e.g. on pre-60 series ThinkPads), but
I need to figure out how to do so properly first. In the mean time I want
this in the tree so I don't lose the code.
|
|
fornt panel of the xserve (the button with the triangle on it). This driver
will dump you into ddb at the press of a button if the ddb.console sysctl
is set. The bug still has one unresolved issue if ddb.console is not set
and you press and hold the button you will cause a interrupt storm that will
slow the system down. So if you keep your xserve near a toddler or you
yourself are a toddler you might want to consider disabling this device
remember buttons arn't toys.
ok deraadt.
|
|
for sparc64 pmap. They take care of all the magic of checking for
initialized pages, zeroing them, etc.
- merge pmap_create and pmap_pinit.
- don't do the page allocation dance in pmap_kenter, we have growkernel.
- Clean up if I was close to something dirty.
kettenis@ ok
|
|
and while there, fix the interrupt name for vmstat -iz; ok gwk
|
|
dma sync the rx mbuf before we push it to the hardware.
technically this isnt necessary cos bus_dmamap_sync is a nop on i386 and
amd64. but i like to be correct, and someone might read this as an example
of how to write a driver.
|
|
been completed. it's the simplest fifo of the lot, im sorry i didnt start
with it.
this is the last of the fifos we have to deal with. now we need an
interrupt handler, but i wont be doing that till i get my cable.
|
|
Add chunks to debug prints.
|
|
|
|
pv_entry head of lists are no longer preallocated at boot, and will only be
allocated for managed pages (instead of all physical memory pages, including
those containing the kernel).
pmap and pv_entry will now be allocated from a pool, instead of malloc for the
former and a homegrown allocator which never relinquishes unused elements to
the VM system for the latter.
The net result is a slight decrease in memory usage, and better behaviour in
low-memory conditions.
|
|
interface send queue and puts them on the hardwares tx task fifo. the tx
completion path hasnt been written yet.
the mbuf dmamap loading has been stolen from vic.
there's obviously a lot of code that is shared between the rxf fifo loading
and the txt fifo loading. i also have to think about how to better
calculate and keep track of space in the fifos.
this is getting harder and harder. it sucks writing tons of code without
being able to verify that it works. i might take a break for a bit.
|
|
|
|
it. returning an error to the scsi midlayer is all we need to do, which in
turn will do the right thing.
|
|
had a bracket though, because it almost falls out of the pci slot...
|
|
prevents dmesg spam when we return this ccb to the free list.
|
|
with it. prevents free panics on sparc64.
found by deraadt on a v215.
|
|
tx now.
|
|
that returns rx buffers to us that we previously posted in the rx free.
theoretically we should now be receiving packets.
practically, im still waiting on a cable so i can plug this card into
something and test it, and then that's going to be difficult cos i only
have two tht cards to test it. i need tx and rx working before i can test
if tx and rx work. fun fun fun.
this code doesnt process rx data change descriptors yet.
|
|
properly these days.
when draining the rxf fifo we want to POSTREAD the mbuf, not PREREAD it.
|
|
|
|
Spacing as well.
|
|
|
|
Pompted by deraadt
|
|
|
|
filesystem related.
pointed out by deraadt
|
|
and its not helping us on those that arent.
|
|
|
|
Align metadata to 64 bit; from deraadt.
|
|
Error report by Johan Mson Lindman <tybollt@solace.miun.se>
|
|
|
|
soft interrupts
from NetBSD, 'reads good art@', also tested by robert@
|
|
|
|
cores by the same amount, i.e. if you do hw.setperf=50 both cores will
be scaled to the opearting state corresponing to 50%. Tested by many with
est (mainly on core2duo machines like X60 thinkpads). Only enable est
during GENERIC.MP build no one tested powernow.
ok art@
|
|
|
|
|
|
|
|
|
|
because turtles are slow but reliable and trustworthy, packets stays
on the net for a long period of time. bigger turtles can stay much longer.
that is the hidden secret reason for the name of KAME project (i'm lying).
j> some IETFers need to be sent to bondage/SM club and spanked/whipped
j> by thousands of dominas and then chopped into million peaces by samurai
j> swords.
t> maybe that is what they actually want, and that is why they
t> fucked RFC1883 and put rosemary's baby into RFC2460.
j> I am king of IETF now, and tomorrow i may become beggar on the IETF venue
j> hotel corridor.
http://www.secdev.org/conf/IPv6_RH_security-csw07.pdf
ok by myself, deraadt@, <samurais at kame.net>
|
|
ok deraadt@
|
|
|
|
|
|
|
|
panel of the Xserve G4 and G5. The leds will provide an indication of the
system load slowing as the load climbs, and system health e.g. no
lights means its dead or in DDB.
Split some of the shared registers and clock divisors from i2s into
i2sreg.h while there nuke some evil C++ style comments.
Man page to follow shortly.
ok deraadt
|
|
|
|
for some reason, so i had some code in there that tried to guarantee that.
however, newly allocated mbufs and clusters are physically contiguous and
bigger than 128 bytes. therefore we dont need to do that check.
while here check if there is enough space in the fifo before trying to
fill any of it. this means we can skip dma syncs if we're not going to end
up writing anything in the fifo.
|
|
keep a list of used pkts in tht_pkt_list as well as free ones. add
tht_pkt_used which can be used to see if a tht_pkt_list is in use by
returning the head of the used list inside itself.
(ab)use this to drain the rx free fifo when the interface is brought down.
tht_rxf_drain pulls used pkts off, unmaps their mbufs, frees them, and then
returns the pkt to the free list. the whole tht_pkt_list can then be freed
safely.
|
|
more work coming soon.
kettenis@ ok
|
|
possible. it loops till it runs out of tht_pkts to use on the rx descriptor
free list, until it runs out of space in the fifo, or until it cant
allocate or map any more mbufs.
tht has a weird requirement that the first physical buffer in an sg list is
at least 128 bytes long. i have code that tries to guarantee that, but id
like someone else to look at it and tell me if its necessary or dumb.
|
|
the hardware. oops.
|
|
|
|
|
