login
Search
Having several weeks to plan a careful and slow migration, I instead took a serious break from the web. Suddenly realizing that my original server is expiring on the 15'th (tomorrow), I have to make the jump quickly. Expect the site to keep changing (improving) for a while... But having people actively use the new site will be the encouragment I need to get everything working properly.
One feature I'm especially hopeful to see used is the new blogs. It'll allow users to keep their own journals, of which hopefully many will be appropriate to be posted on the front page.
The Broadcom Tigon3 boards arrived, but looking at their
driver made me want to scratch my eyeballs out. I think
I'll write the 2.4.x driver from scratch, thank you.
At least, I can print out their driver and use that when
we run out of toilet paper here at the apartment.
Jeff Garzik will help me on this thing and I let him loose on
the vger cvs tree today.
Someone told my wife about the multi-head experiment I
Winter is long gone, this snowboard powder bum is depressed.
I just int10 booted the x86 bios on a r128 PCI graphics
card on my Sparc SunBlade-100. Life is... ummm... good?
Will be
cleaning up these xf86-4.1.x mods and trying out tdfx
and other cards tomorrow. Maybe I'll even get r128 DRI
working on this box, which would be nice. Gareth Hughes
promised he would help.
What does this mean? Well basically I can now multi-head
Got multi-depth (8, 24), many visuals
(truecolor/directcolor/pseudocolor/staticgrey/etc), and hw
accelerated double-buffering working with
Creator/Creator3D. Just have to nail a few bugs then it's
off to DRI/DRM direct 3d rendering.
Paul Oakenfold is the ruler.
The PPPOE/PPPOX patches are in 2.3.x, should show up
in Linus's next pre-patch.
Starting looking into overlay support for FFB in xf86-4.0,
unfortunately the xf40 overlay layer lacks the concept of
window-id's for the 8-bit overlay (essentially, WIDs give
you
N 8-bit colormaps to use, where N is the number of window
IDs
supported by the ramdac) so we'll just use one WID for
xf86-4.0
until this is cleared up. I wonder if I can trick the xf86
Got the SysKonnect FDDI driver working and portable in 2.3.x
Need to cut a 2.2.x version of these changes and submit to
the
SysKonnect folks for review...
Checking out
axboe's
latest cut of his elevator hacks, let's
see
how this goes.
I believe AT&T Cable has a conspiracy going. Every time
I've gotten a new cable box, exactly one month later the
Spent a few days on my new paging stuff, killed most of the
bugs
but it still had a lot of problems. Suddenly I realized
that this
isn't what I should be working on and as such I have dropped
this
work for a while, but I sent off a snapshot of the work to
Stephen
Tweedie in case he wishes to play with it.
Playing with Jens's new elevator stuff, on my box the first
run
looks really bad... hope that's just a cold first run issue
because
he's sent this stuff off to Linus already.
Andrew broke 3c59x.c on non-little-endian machines and those
not supporting virt_to_bus/bus_to_bus in pre6-6, I've let
him
know about this. :-)
Starting to make more progress on my VM hacks, the anon
layer
was just the beginning. I hope after the 2nd or 3rd pot of
coffee
tonight I should have something that at least remotely
resembles
what I want it all to look like. If I arrive at something
people can
at least play with, I'll post patches.
Finally got to finishing half of the SKFP portability stuff.
Merged current tree to Linus.
Wrote an anon page layer for the Linux vm. It works, no
leaks, no bugs, etc.
Older UltraSparc firmware aparently writes out the
"assigned-address"
PCI device properties incorrectly for 64-bit MEM space
resources. I hate
having to write workarounds for crap like this. Nice
spotting on this
one by Matt Jacob.
