Did you know stuff happened today while you weren't watching?

A completely unproductive bunch of nonsense. [That is probably a better way to end this than it is to start it, but at present, I'm not one to care.]

Friday. I have no idea. About all remember is Brock and David showing up rather late. Finally figured out the truth to the mural on my wall (and some interestingly nasty errors that the superimposer made). David left shortly after that revelation. Brock never did. Well, he did. Just not then. He disappeared around 1400 today.

Somewhere in between, I fell asleep on the coach. What was more amazing than that however was how I ended up on the side of the house with a couch on it in the first place. If you figure it out, let me know.

And now for some gratuitous, and inconsequentially amusing, links...

The prime example of the Windows user. Perhaps I'm merely overly tired, but I've spent the last ten minutes laughing at that. And yes, I can tie my shoes. I just don't like to.

And don't miss the (twisted) history of the official mnemonic for the lack thereof.

Short week, short month, short summer.

Nothing terribly exciting has been going on, except for a few things. One of them concerns AOL, which I'm not going to mention anything about this early for fear of causing a ruckous. Another involves something that I'm not even going to mention any clues of, because it may hinder the effectiveness of everything else thats going on. The third I will mention.

The third was today. And involved me getting a really fast inet connection. Via SpeedChoice. Its fantastic. Averages anywhere between 60kb/sec and 150kb/sec down, and "t1-like" speeds upstream (as you probably noticed that this loads a bit faster). Of course, it will be tough for a while, until DNS gets all resynched again. (Which should be anywhere from eight to twelve hours from now.) In any case, I am satisfied. The techs were mediocre (not as bright as the PLUG people led on, but not completely clueless). Took about two hours to set up. The coax drop didn't actually get done (its strung across the ceiling instead -- long story). Aparently line-of-sight was no problem. The antenna (surprisingly square) is on the far east end of the house, on a five foot pole at that. Not terribly obtrusive, but noticable.

Back to previous events. The week didn't really start til Wed. I threw together a very rough model driver for Russel Kroll's Smart UPS Tools, for my Best UPSes. Unfortunatly, I ran into the same stupid problems I had way back when I tried to write code for the things two years ago. Two major difficulties: 1) echo and 2) that damn prompt. The echo is a problem because you have to suck your own data back off the line whenever you give it a command. Annoying. The prompt is the most trouble. Prompts are great for human interaction, but for computers, its tough to parse. Gets in the way. The most difficult part is trying to figure out why its such a big problem. Let me know if you know. I've yet to figure it out.

Sara dropped by and picked something out of my collection as a birthday present. She decided on an AOL4.0 CD. She also (amusingly) found out that my keyboard has no keycaps.

Then I went to an event which I'm not going to talk about. Came home and took a nap. Brock and David came by. Went to see Deep Blue Sea Now that's a laughable movie. You should go see this movie for one scene: where that business guy starts off on his lovely speech and gets kindly (and unexpectetly) terminated by a large (badly CGI'd) shark. Other than that, the movie's utter nonsense. (And the helicopter scenes look like screenshots from a video game.)

Sleepitime after that. Woken up at 0900 by an Army recruiter. I thought about going into a long, drawn-out political reason why I would never come close to that brand of public service, but I decided to be nice and keep my opinions to myself. (Did I just say that?) The speedchoice guys came about noon. Stayed til 1400 or so. After convincing them that the garage was not in fact going to harm them in any way, they started installing.

Sorry about not writing here much. There's been a combination of extremes (very little and way-too-much-to-think-about) going about around here. Its just the beginning as well. As I said before, this was my last week of unset hours. Next week is seven hours/day of band stuff. Then school. And after nine months of that hell, then I can do whatever I want again. To an extent. Perhaps an even lesser extent for a time. In fact, it's been irratically getting worse for quite a while now... just over seventeen years now, methinks....

Wow...its Sat already...

Not much has particularly been going on around here. I'll point out the minor bits...

I put the simulator on freshmeat on thursday. Of course, since freshmeat was down for two days, it got pushed to the very hidden bottom of the list. It did pick up a few hits however, including a couple from Analog. Of course, no one has yet to comment. Hah. Since that release, I've put up three more. The latest of them has the beginnings of Shifter and MAC execution, as well as one (count it) one ALU op. Its all mostly untested, of course.

If you're a libfaim user, and for some reason can't connect to AIM anymore (starting yesterday), please go bug your client's maintainer about upgrading to the latest (as of yesterday) libfaim. AOL added a particular five bytes to the authorization frame that broke libfaim in a particularly fatal way. If you want to hear the anti-Microsoft angle, go to slashdot. Of course, you probably shouldn't believe that AOL did this intentionally. Since that's fairly unlikely. This is definitly the side effect of something quite unrelated.

I think the third and final interesting tidbit is that Dirk-Willem van Gulik (the one who dug up that old Novell RPL protocol document) has successfully booted an 8227 using his own code. Unfortunatly, it uses the BPF, and I don't have BSD. I will have to move it over to libpcap or something before I can use it (or install a BSD box, which is certainly something I've been planning on for a while now -- waiting for speedchoice so I can do big downloads -- less than a week to go).

And that means I only have a week of summer left too. Sob.

Monday must not have been memorable. I think I slept a lot. Too much. Damn.

Now tuesday involved simulator stuff. Stupid problems. However, it can now do DO UNTIL loops as well as properly handle the 14bit stuff (which fairly automagically brought circular buffers into functionality, though I did have to create myself some internal B registers to handle that, I really don't know how the 2181 does that). Anyway, those were the last few tough non-ALU features. Which leaves ALU/MAC/Shifter stuff. And it really won't be that hard to implement. Again, the most difficult part of these ops is the decoding of them.

And I decided that I'd waited long enough. See here for where to get the source. I'd like to get the math stuff done soon. I'm soooo very close to having something useful. After all that stuff is done, I'll write a quick emulated 16550 for io() space, and then a half-assed codec act-a-like for sport0. Oh, my current idea for emulating the codec is through the use of a sound card. RS232 doesn't have enough bandwidth to send raw codec data, but a sound card certainly does. There's still a lot of problems with it, but its interesting to think about. Opinions?

Still need to either write a gcc backend or port Analog's up to a decently new gcc (its for 2.3 i believe). Will also need binutils (gas) backend. Eventually I also would like to start on the SHARC chip. It's a bunch more complicated, but I'm not sure if I could live with myself if I didn't attempt a kernel port to it (its a _very_ worthy target). Besides, Linux/RT might be very useful running on a DSP. Not sure if anyone has actually done a port to DSPs or not. Have to use uCLinux (no MMU).

Also, someone mailed me today and said they have 500 8227s to get rid of. The 8227 and a single card for $80. Email moores@tdi.net for details.

Oh well.

Somehow I doubt I went to bed when I said I was going to. Of course, since I can't particularly account for that time, I really couldn't say that for sure. (It is in fact a common tradition for me to make a point of not going to bed when I say I am. And in a corrolary to that, it is common tradition for me to make a point of not doing anything I say I am going to do. Just remember that the next time I say I'm not going to do something....)

The time I can account for however was spent sleeping (although I also remember somewhat of an underflow in that account). Tested a splendid DSP idea that aparently only half-works.

About 0500, I tried booting up the Apollo. Nothing. The monitor syncs. But it just sits there. I have no idea what all the little lights on the front mean, and therefore have no real way of debugging it. I thought it was expecting a keyboard, so I pulled the graphics card, but still no serial console. Also pulled the token ring card out (its ISA, interestingly enough) to take a look at it. I thought I knew everything useless there was to know about TR. But aparently "Domain Ring" is completely different physically from IBM TR. The connector is a DB9 with only four pins, the others replaced with two largish shielded connectors embeded into the db9 (reminiscent of a shielded video connector). Oh well. Gave up on that after a while.

Then started working with the Sun3s a bit more. Couldn't get the /2xx CPU board to boot at first (diag mode worked, norm didnt). So I tried the Sun4/330 (yes, VME SPARC) in the /260 chassis. Got serial output while in diag mode (this is quite remarkable to me since this is the board that I would have bet had a busted uart, go figure). But it wouldn't boot in norm mode either.

Put the 3/260 board back in. This time, however, I followed the cardcage configuration charts in the Sun Hardware FAQ. (Contrary to what the backplane appears to be, it is not just a simple backplane. Order Does Matter. This was new to me.) CPU in slot one (leftmost). First RAM card in slot six. Voila. Normal mode boots with serial console. So I got adventurous after reading about the RAM cards and decided to try installing the other 8mb card I had. Since it would be installed between the CPU and an existing RAM card, the termination block had to be removed, and the external jumpers need to be set to 1 instead of 0. That worked amazingly too (the second RAM card must go in slot four, aparently there is some heat dissipation theory going on with that). Lastly, I put in the video card (must be in slot three) and plugged in my type4 keyboard via my homemade epoxy type4->type3 connector. Worked. Graphical console, complete with Sun logo. Played around with the console commands a bit (nothing new, same set that is on my 3/60). Did a RAM test (one of the warning lights on the secondary RAM card came on during bootup, I thought I should check it -- nothing wrong aparently). I much enjoy the "CPU" LED, which lights up if the CPU is currently accessing memory off of that board. During the memory test, it flips back and forth between the two boards. Okay, so I thought it was neat... In any case, when you turn on the 3/260 chassis, not only do all the UPSes in the house come on, but with ten fans, it also conjurs up quite a dust cloud. [Not to mention noise. The first time I powered it up, there was a terrible racket coming from the lower rack of fans. Closer inspection found a leaf stuck in one of them. Much quieter after that was removed. Surprisingly, the leaf was unbroken and in quite good shape. And I think its been in there for quite a while. And it was not a leaf from a tree native to Arizona. More of an aspen motif, which would fit. The system came from CO.]

I really picked a bad (read: hot) part of the day to start the second operation (it is in fact an operation that I said I was never going to do -- see paragraph one). That is, power up the Eagle drives and the 3/280 cardcage (though not the rack, I still dont have a plug for that). This was also a test of the rating on the power strip I had laying around. It was rated at 17amps. It survived (and so did I), so it must not be lying too badly (5a/eagle, 4a for the chassis, plus a monitor). This was all quite a trick to connect mostly because the /280 rack is quarantined to the netherregions (not air conditioned) part of the garage (and past that, its behind the Orangish Door, as opposed to the main Whitish Door -- why our garage doors are color coded is something you probably don't want to ask about -- so that puts it even farther out of reach). I did get it connected, however. And after putting each VME card into the right slots, I turned the cardcage on. Graphical console. Goody. Next step: power up and connect an Eagle. Which one to choose turned out to be an easy choice. Only one of them has the right connectors on it (the bottom one, of course). Plugged it in. Shut the cardcage off (just to make sure I didn't blow anything). Figured out that the Eagle has a breaker on the back that must be flipped first. Then it gets really noisy. The two largest switches on the front should be started with protection on (up) and start off (down). When everything sounds good, you push the start switch to on and wait a few minutes. When it sounds like its ready to take off, its ready. Turn protection off. I managed to (partially) boot SunOS (4.2 Release 3.4) off of it. The other eagle is listed in its fstab, so it of course fails to boot all the way. And aparently I have yet another problem because when it drops me to the shell prompt, I can't type anything. I'm thinking I'll be going back to a full serial console here soon. [My theory is that although the PROM supports the type4 keyboard, that old SunOS probably does not. [The '4.2 Release 3.4' thing I believe is interpretted as 4.2BSD (Tahoe -- or is it Reno? always confusing those) and SunOS 3.4. In fact, I don't remember if its 3.4. I think I am not remembering that right.]

I decided that was pretty damned successful for now, and closed the Orangish Door and came inside. Put up some way-too-large pictures of the 3/280 rack (I also added comments to the 3/260 pictures from yesterday). That should be enough of those for now.

While all that was going on, the [insert a multihour pause here where Dan came and shot my spledidly elegent DSP idea to hell], I forgot where I was going with this sentence; its been many hours after all. Perhaps it will come to me a bit later.

I think I'll probably sit here a while. Thinking about DSP stuff no less. [Of course, in case you've forgotten its content, refer to paragraph one.] I don't really have any idea what time it is in the Real World. Which is probably why I laughed when I asked myself what time it was just a few minutes(?) ago.

Hrm. David is not responding. Eaten by that... oh what did he call it... "lawn mower" i suppose.

Btw, the two things holding up the 2181 simulator are ALU/MAC/Shifter opcode decoding (decoding is far more troublesom than actually implementing the operations) and the fixing of the Major 14bit integer problems (ie, DAGs).

Hrm.

Slept through a lot of Thurs. I wasn't feeling terribly great.

Friday was better. Spent the early half of the day wandering around the inet. Left about 1130 for Tempe. Got there. Realized that the /280 rack was on the high end of the size I was expecting, though the /260 deskside was quite a bit larger than I expected. Removed the Eagle drives from the rack and loaded them first. Then the main rack. Heavy indeed. Stuffed the /260, the Eagles, the Apollo and the NCD around the rack. It is rather amusing to see people avoiding you on the freeway for fear of something flying out the window at them. Came home and unloaded it all. Yanked the Sun VME cards. Got a few more than I expected (I was anticipating the /260 being empty). I was not able to obtain the 6250bpi tape drive (which fills the top half of the rack) because we ran out of room in the vehicle. Maybe someday when I'm in tempe for some reason, I'll pick it up.

The /260...

• 501111612866 -- 8bit video
• 501115503777 -- tape controller
• 501116623000 -- SMD controller
• 501115310854 -- Ethernet (intel)
• 501114909674 -- "Sun-2" internal SCSI controller
• 501110265930 -- 8mb DRAM
• An unknown tape drive of some sort, SCSI i believe, though I haven't looked

And the /280...

• 501120606388 -- 3/2x0 CPU
• 501110215489 -- 8mb DRAM
• An SMD and a tape controller

Also included in the /280 rack was a strange board made by Clearpoint Research (no, I haven't heard of them either). The guy it came from was told it was an FPU. That seems unlikely. It's labelled 'SNX2RAM/8'. Serial number 639, copyright 1988. The major chips are a CSI5134 and 5133. No idea.

A few (largish) pictures of the /260 are here. Don't mind the surrounding mess. I'll have pictures of the /280 rack and Eagles as soon as I find a key to the other half of the garage. All in all, those two machines set the record as the first and second largest machines I've ever seen, let alone own. I'd like to put together an informative site much like Techrat's (for his 3/160).

I haven't taken a look at the NCD and Apollo yet. I'm still rather sore from moving the Suns around.

DSP debugging stuff from 0030 to 0730 (while the camera battery was charging). Took above pictures of the /260. Came back here. Probably go to bed here in a bit.

AP scores came yesterday. The oh-so-much time I spent preparing for that APUSH test got me a four (I was (rather optimistically) expecting a three). And of course that APCS test that I only ended up taking the A part of because a screw-up in administration, earned a five. Blah. Who cares. So that proves that I can operate on arrays in that AP pseudo-C++ syntax. Big deal. [Oh, for those of you who for the past ten months have been wondering what APUSH stands for, its AP US History. A very difficult test. Fun class though.]

Hrm. Think thats it. Good enough for now. Going to sleep now.

I'm about out of energy. Been up way too long.

Last night was interesting. If you'll remember a while back, I mentioned that I lost an auction for a local Apollo. The seller mailed me and informed me that since the high bidder was not responding, I get it. He then explained that he also had a Sun3/280 up for grabs (free for pickup). And 17in NCD Xterms for $25. Neat, I say. He's over in the northern part of Tempe. Not a long distance.

I was all excited about the 3/280, however, this picture of a /180 (the big one -- full rack) and a /160 (the small(er) one -- 'deskside') has scared me a bit. I would hate to refuse such an interesting offer because of lack of space, but if its THAT huge, I think I might. It's not like that is something I could just sneak in. I may convince my parents that I wont keep it for long. But I see no way of getting rid of it once I have it. No one else really wants an 8ftx3ftx3ft rack chassis either. I mean, I'd like it. But I just don't want to cause too much of a rift around here. Chassis' like that can cause rifts. Large ones. Yeah, so I don't know what I'll do. Definitly get the Apollo though, and maybe an NCD (I've always wanted one of those). After further research, I do believe that the /180 rack in the above picture is indeed what I'd be getting. Arg. Well, I've always wanted a big VAX. A big-iron Sun can't be too bad, right.... I've always wanted 150lbs SMD Eagle hard drives and a 19in rack anyway.... heh. (That two-foot tall thing at the very bottom of the big rack in the picture is an SMD Eagle. Yes. Real Winchester drive, five 11in platters.) Sheesh. I'm still trying to think about how to get that one by the owners of this house. Hmmm...I hope its not three-phase...

I was actually planning on driving over today and picking everything up. However, since most of Phoenix has about two feet of water standing, and that I'm really tired, I would wait until tomorrow. Maybe take someone with me to help move stuff. Hey brock, your truck functional? Maybe I could offload the big rack onto him....hmmm.....

Well, anyway. Enough about that. I slept a good chunk of yesterday. Not enough. Did DSP work til 0930 this morning (sans a few minutes in the middle at which time I went outside to get thoroughly showered by what one would surmise would be natural rain water, however, although the process of the rain falling seemed natural enough, the water itself was disturbingly slimy).

Analog mailed me back. Firstly, they won't release docs on the ICE protocol (no big surprise there). Secondly, despite the fact that the tutorial and all the rest of the VisualDSP docs document it, the EZICE target for the debugger is quite nonexistant. I've never seen such well-documented vapourware.

I can say with confidence on behalf of the linux-dvd list and myself, that slashdot has reached a peak in its trend of terribly bad timing. That CSS decryption stuff has got several weeks left of technical work left to go before its even reliable at what it was written for, let alone for use in a real software DVD decoder (and not mention weeks and maybe months worth of (il?)legal work to get through). Am I the only one who often wishes that slashdork would just go away?

Can anyone tell me why the hell I am still awake?

Onsale? Egghead? Merger? Why? I frankly still can't believe someone (VIA) actually picked up Cyrix...THAT was a Bad Idea...

Hmm...its Wed now. I think I missed Tue. Oh, wait, no I didn't. Slept through most of it, I remember. I also somehow remember ending up with a Sound Blaster Ensoniq (sp?) (the ES1373 one). Put it in mediocris and pulled the PAS. Loaded the ALSA driver for the ES1371. Mixer works. Opened /dev/dsp. Death. Null pointer dereference. Reboot; fsck. Load OSS driver for the ES1371. Mixer works. Opened /dev/dsp. Works. And still works. So now I have a PCI card to play mp3s with -- ie, no more ISA DMA rubbish. Yay. Also recollect the writing of an rs232 line monitor (sits between a DTE and DCE, logging all the data). Decided that the EZ-ICE board for the 218x basically does straight JTAG. Decided to ask Analog anyway. No response yet. Also asked them about where exactly to find the EZ-ICE target that supposedly exists that lets the debugger use the real DSP instead of the simulator, getting rid of the simulator's problems (inability to simulate codecs, etc) while keeping the benefits of the debugger (profiling, full-blown memory watching, symbolic trapping, etc). No response there either, of course. I suppose I should just remember to ask Dan about such things, since I could imagine it helping us a bit....

Well, that paragraph was nearly strenuous to avoid understanding; must not be trying hard enough.

Hell if I know what else happened. Basically woke up around 0100 today when I found myself wandering around with Dan looking for mtndew (at far more locations than logically (and rationally) necessary).

I'll sit here looking amazed at a container of water for a bit, then find my way to sleep...or is the other way around...

Hmm...would you really stick this thing in your ear? I always pictured the babelfish as someone a bit more...perky?

Aparently I AM a stark raving mad looney after all.

Lessee...what day is it....

Slept most of Sat, and in fact slept most of Sun too. Was awake long enough to watch MP, work more, and go to bed. And then on Sun, I did basically the same thing, sans replacing MP with the normal Sun TV. Went to bed around 0400 this morning. And got up around 1400.

After getting the word from the PLUG list that SpeedChoice does indeed include a fixed IP with all of their services, I called SpeedChoice and verified that. And then ordered (two-way for $54.95/mnth). I'll talk to them about the TV service when they do the install...which has to wait until the TWENTY-NINTH OF JULY (and that _early_ date is because of a CANCELLATION!). I am thinking that they only have one two-way installation crew or something silly like that. So, less than two and a half weeks left for me in 56k dialup hell...

That's all I've gotten accomplished today so far. And that's probably the extent of it too.

Sex Education School "only $39.95/ month", and remember: "Never date sluts"..."Dont commit gay acts unless you are gay"..."Do not have sex with animals"... Don't you wish American telephone c ompanies took an ethical stand every once in a while?

Use as many nifty imaging tricks as possible, using the most hardware you can requisition, all to find 'strange lights and UFOs' in an automated way. And look! They found the nasty ALIEN BUGS!!! AHHHHHH!!!!!

Hmmm...I'm fairly spent. This may be a bit short.

Dragged myself out of bed around 1300 (I nearly typed that in base-offset hex notation). Wrote some simulator regression test routines (just a few, mostly testing ALU, autoincrementing data write, circular buffer, and zero-overhead looping operations). Simpy stuff (mostly less than ten lines each). And not only that, since hardly any of the above mentioned operations are implemented in the simulator, all the tests fail splendidly. We'll work on that today.

From about 2130 to midnight, I hacked a rather handy UI into my 2181 simulator. All I can say 'I love GNU readline'. That simplified everything. I just hope it doesn't limit my portability too much. Does readline compile on djgpp? The only Big Feature I can think of to add to the UI is a microassembler (to execute instructions at the simulator prompt). But that requires implementing an assembler, of course. Which is yet another pita. I think I'm definitly ready for an initial release, ALU/MAC or not. I'll get stuff implemented to pass my current set of regressions, then release. I promise. I know my week is up...

Dan eventually arrived at midnight or so. [I was sitting in the hypothetical center line of the road, however, I decided after being out there for three quarters of an hour that the hypothetical mosquitos had bitten me enough. I say hypothetically mainly because firstly I never saw them, and secondly, all the bite marks magically disappeared a few hours later (I had a half-dozen of them on my right hand alone). Were they real mosquitos? Or is this yet another manifestation of my inate insanity? You decide.] Shoved "HI" out of the serial port just after 0700 (I was going to spring for an all-out "Hello World", but since the bytes have to crammed out one at a time (very manually), that was a bit much for a one-time test.)

I'm going to eat some breakfast and go to bed. I'm tired. It is friday, so you're probably expecting the normal friday-flood of links from NTK. I'll pass them along tomorrow.

Oh, and don't forget... A&E starts airing Monty Python's Flying Circus today at 2000. Also, And Now for Something Completely Different will be on at midnight local. It would be better if PBS still had the syndication rights for this stuff, but this is better for me, since KAET (our PBS affiliate) never aired it anyway. Circus will air weekly at that starting tonight. A&E is also rumored to be planning a DVD collection representing all the Circus episodes. Release date unknown. However, And Now for Something Completly Different will be released on DVD sometime this month. (If you've never seen that one, I'll warn you that its Circus-style, not Grail style -- many people only like the latter. But it does have that absolutly hilarious Mountainier sketch, and not to mention the classic Dead-parrot sketch.)

Anyway, enough about MP. I'm going to bed.

I have yet to reset my clock since the power outage. The nice part is that I really don't have to. It managed to get turned at just the right time to be a mere 97minutes fast. ('Bah', you say, 'that's still outrageously inaccurate!'. And to that, I say one of 'It's just 97minutes, subtract!' or 'Clock? What is this clock you speak of?') [Not to mention the fact that you don't actually have to subtract. You could add. Why you only have to add a meager 563minutes. You can do that without a calculator, cant you?] (append 'abovelistofreplies ("And what's wrong with Deepthought Standard Time?"))

Hmmm...I think that's enough (pseudo-)LISP for one day. I'll try to be more linear next time, so such things don't end up coming out.

I must admit that I really didn't get as much done today as I hoped. Slight simulator work (I really need to stop thinking about that and just start writing what needs to be done -- list slightly updated -- and getting shorter -- and you'll notice mediocris is now for some reason keeping UTC, it did that all by itself when I rebooted yesterday). Brock came over and pointed out the severe need for some sort of debugging interactivity interface with it. Funny. I never really realized the need for one until someone pointed it out. Now I can't imagine what I was thinking by not putting one in from the beginning (stepping, start, stop, etc). Considering how much I now find the need for one, that will probably get done next. I actually really thought about releasing a 0.01 of it today, just to have a clean copy for when I heavily break it by putting in UI crap. I will probably do that right before I start on it. Hopefully tomorrow (err...today -- that sun is creeping back up again). Yes. ALU/MAC operations will get done soon. I promise. (Heh.)

Around 0200 this morning, I about had a 2.3.10-mid1 (I'm going to try to prove Josh's statement that 'anyone with initials can release their own kernel tree'). I say 'had' because it was all going well until I actually tried to boot it. It died (Oopsed out on their page freeing). My main change (and the cause of the Oops) is my attempt at a partially reorganized page allocator that takes into account the 16mb+ ISA DMA problem. (Not that this is a new problem or that fixing it is something new either, I just decided I try it for myself, since none of the other patches work with 2.3.10.) I'm guessing I was trying to be too ambitious again. I will probably just end up porting Jaroslav Kysela's patch for 2.2.9. Anyway, now I'm running flat 2.3.10 (the first 2.3 I've tried since the deadly 2.3.7). Seems stable enough. (Upgrading a kernel on mediocris also means recompiling alsa, bttv, and the couple of my modules (who's code hasn't gotten put into anyone's standard tree yet (except for the AVEC tuner code, but that was in the mainstream kernel bttv, which I don't use -- its got Bad Vibes, and it doesn't work with my nifty TEA6300 mixer driver, which I've yet to submit to Gerd)).) (Weee that was fun closing all those -- thank the gods for emacs! Well, RMS anyway. Is he a god? I think so. Not sure about the official word on that though...)

So you may actually see a 2.3.10-mid1 patch floating around in the next few days (if I can figure out where its oopsing at). Mahler symphonies are rather difficult to listen to in a house with other people in it...too many volume changes.

So I'm sitting here listening to as many mp3s as I can before my ISA DMA window fills up (which there's nothing stopping it from doing -- that's why I need it fixed). Deepthought has been quite untroublesome since it moved in with me. The drive starts clicking when my room its 90degF though. We'll just have to keep that from happening.

The weather was rather disappointing today. The news had us getting all excited about yet another huge storm, and yet all we ended up with were some (somewhat unsubstantial) cloudworks. Nothing more than a slight breeze. Perhaps tomorrow will be nicer. (Aparently my house was the only one that lost power yesterday...)

Speaking of power...I finally got into my UPS. Straight-through DB25 female-male cable is what it wanted. And it wanted 1200bps. (That last part took a bit to figure out.) And then increasing the baud requires a password. Had to find that. It's 18473. (Don't bother trying to do anything with that -- my UPSes are not internet-connected (I failed at that project nearly two years ago, deciding that I hated programming serial as much as I did trying to get serial to work right).) Which brings up yet another strange bit of near-fact... Our AC is running at 60.01Hz. Which is not overly odd, except for the fact that for the past two years, its always run at precisly 59.95Hz. Of course, I'm not too sure how accurate these numbers are. It also says its only 47% loaded (which is possible -- its a relatively large UPS (500va)) and that runtime is 75min and thats fully charged (which means it fully charged itself from zeropoint to full charge in a mere couple of hours), which it definitly didn't get to last night -- 45min tops. Oh well.

Hi.

For having been such a rough evening, its really quite a nice morning. Partly (or is it mostly?) cloudy, waxing moon halway to the zenith, and that blaring too-bright light on the neighbors garage. Bit humid. And a bit warm (~80deg). Oh, and I forgot the annoying whir of the freeway. I'm not real sure about the time (I can't get the official Deepthought Standard Time when I'm outside). Very still. Too still. And Mahler's Eighth (part two) as background music, with insect noises and water dripping courtesy of Nature.

So now that I gave away that it was a rough evening, I might as well explain the rest of the day. Got up around noon. More minor simulator work (all the tiny details take forever to get anything accomplished). Actually, I spent most of the day doing that, on and off. I stopped at exactly 1830 (you know why -- it starts with a capital 's'). And just like last night, everything started going down from there. Though nothing died (well, except for the cricket in the bathroom (that was not my idea, I was perfectly willing to pretend it wasn't there and spare its humble life, but no, All Things Must Die is the motto around here), and the spiders in refrigerator (okay, I admit, I killed those ones -- I tried to coax them out, but they refused)). Anyway, it was real windy. Even more windy than yesterday. Windy to the point that at about 2000, the power got knocked out (that wasn't my fault). Not simply went out, of course. It had to sit there and flicker on and off a few dozen times first. But then it just stayed out. For nearly three hours. The UPS in my room died about 30min into it (I had gotten them all halted (mediocris and dt) approximatly three minutes before they died, which was a rather lucky guess, since I had no information on how its load status or anything (in that 30minutes, I stumbled around in the dark trying to find the appropriate serial cables to find that information, but I eventually gave up...just in time)). That UPS was probably over capacity anyway, considering it had my two largest machines on it (one advantage to having dt in the garage -- its own UPS). After the wind died down, it started to rain a bit (okay, a lot). Sat in that. After all, it was about 100deg in the house (no air conditioning) -- the rain was a relief even if it was warm. Power came back on just before 2300 (I think -- I didn't really have any clocks for reference). Got everything powered up just after 2300. That went relatively smoothly (mediocris hadn't been rebooted for 27days or something). dt didn't have any trouble of course because its probably been rebooted three or four times today already (oh, I forgot to mention that).

I sure wish I knew what these bugs were that are crawling all over me. They seem to be attempting to bite or something. Mosqs perhaps.

A good chunk of the afternoon was spent fiddling with deepthought. The primary drive in the raid (hda) is dying. Not sure how much longer its got. I've got to do something about it eventually. (Its impending failure is related to that 'lost interrupt' message I got a few weeks ago, when it killed its 90+day uptime.) The secondary drive seems to be doing just fine. The problem is that with striped RAID0, you can't just replace one drive.

Maybe these bugs will go away when the sun comes up (which appears to be going to happen in a couple minutes). If not, at least I'll be able to see them.

I really came to no conclusion other than that its dying. I replaced its IDE cable (which appeared to be stressed a bit and a possible tear in pin one), and have had no problems since. Somehow I doubt that was the problem, however. Cabling doesn't usually have the tendency to grow failures like media does.

Post-2300, I spent looking for a cheap PC110 (those neat, but rather old 486sx-33's that are the size of a Coke can (including keyboard, LCD, etc)). Since they stopped production on them, they're even harder to find than they were new (they've never been sold by IBM directly in the US, only Japan). Decided the modern equivelent is the Toshiba Libretto. Decided I didn't want one bad enough to pay for one of those. (The PC110 is still smaller anyway.) I don't need, nor can I afford either one of them. That was the end of that. (Oh, and I ALMOST won an HP Apollo on Ebay...and it was local too ($20 and no shipping!). I had forgotten that there's no foolproof way of winning on Ebay these days. Online auctions are a lot like online gambling, except with more gambling involved.)

Then spent a few hours (yes, too long) retrofitting a usable power connector onto my Proteon hermaphroditic token ring MAU (which I have still yet to boot up -- it has funky power connectors, but it is managed, and I'd like to at least poke around in the management PROM a bit (has a serial port on the back that resembles the DEC VAX MMJ pinout)). No luck. It gets the 12v line, but aparently there's still a loose connection on the 5v. I'll take a look some other time, when I'm even more bored. I did manage to accidently spew super glue on a few of my fingers (urf). And I can't find any pumice in this house. I'm open to suggestions on how else to take off skin painlessly.

Neighbors do the strangest things at five am.

As dawn arrives, I see that storm didn't do as much tree pruning as expected. Small branch grounded from our pecan tree. A moderatly sized chunk on the ground from the neighbor's cottonwood (which has never been the same after that 'microburst' a while back (has it been two years now?)).

I didn't get any nifty pictures off TV this time, because I didn't have a TV, let alone a Bt848. I've been thinking for a long time about finding a place (inside, pointing out a window) to mount a camera to automatically webify those respectibly attractive sunrises and sunsets we have here. (With a side feature of showing off the gnarly cloud structures that appear during 'the rainy season' (ie, now).) Still haven't found a good location yet, however. Our house has a really bad view of the horizon (I just smashed a bug on the word 'horizon' -- looks like an ordinary knat-like creature, useless buggers).

Just as I thought. They go away at sunrise. The only problem is that bigger ones come in their place....flies.

Well, I think I will sit here for a bit longer. Really have nothing else to write. I'll go in when the flies get too bad. I still need to copy this up to deepthought too.

Okay, I came in, dated it and copied it up. I think one of our dogs is going through some sort of mid-life dilema. She seems to be staring a lot.

Sheesh...it was just one of those days that I would just like to remove from history....

Got up around noon. Did some simulator work I think. Nothing major I'm sure. You know, I really can't remember what I spent the afternoon doing. Memory lapse.

A biggo storm rolled in around 1700. The day started to go downhill from there. About 1730, the air conditioner in the garage got shut off. The temperature shot up about 15deg in a couple of minutes. That seems to be a bit too fast for WD hard drives to do their thermal recal. Deepthought died. Turned it off. Wouldn't boot up. After getting rather concerned, I just drug it into my room, waited a minute, plugged it in, and booted it up. Its in fact still sitting in my room, a couple inches behind me. Might as well leave it there, i suppose. Its really a pita to move (I require myself to find some reason to complain about that case at minimum once a month). Anyway, it did boot up. fsck took probably 20 minutes. Went out to the living room. Threw away all my junk mail. Went outside. Saw our last sheep died. Watched it rain. Sat in the rain. Buried the sheep. Came in. Found the bathrooms heavily infested with giant ants (okay, so one or two _may_ have reached 2mm). Watched it rain a bit more. A load of other worthless crap happened after that (not to mention me getting thoroughly frustrated by dgjpp (I actually gave up an used Microsoft product)). And then KASW had the nerve to mess up the Simpsons tape. AHH!

Oh, this little utility (writing for DOS) may be useful in extracting the Service Tag number from some newer Dell machines. I don't have one to try. And, btw, it crashes your machine. (You really don't want to know how long it took to find the _MK_FP() macro -- thanks anyway Orb.)

Oh, and I still really don't know why, but Brock, David, and Booth for some reason appeared at my house. Brock ate Taco Bell. David informed me that the TI89 is in fact a dark bluish color, not the black it appears. And Booth helped my sister with her math (which I would have ended up doing if he didn't). And then they all left. Boggle.

The newspeople got all freaked out about that storm today. I guess they always do that with the first of the season. Oh well. (If you'd like to take a look, heres a few random shots from TV: here and here are courtesy of the local FOX affiliate (ch10), and this cheery one is from the local NBC affiliate (ch12). Oh, and that last one brings up the fact that we made it below 85degF for the first time in a month.)

Oh, and today is Gustav Mahler's birthday. (Anyone know where to get what was written of his 10th symphony? I've heard rumors that the BBC orchestra did an arrangement and recording, but I've yet to find it. That, and The Song of the Earth, are all I need to complete my collection...)

Damnit, the sun came back.

Okay...let's attempt to decoct this lovely yet puny day into something more substantial than it was...

I probably didn't get up til 1400 or so (though I believe I started thinking about the whole process around 1100).

Probably the most eventful part of the day involved an itch. In my ear no less. I was listening to random mp3s (its a public directory and I would link to it except I dont want people grabbing 192kb encoded mp3s across my 56k...not that anyone but me really wants pre-20th century music in mp3s anyway...I seem to be menacing up some terribly high tangetal acceleration this morning) and my hear was itching. I was using those little speakers that you stuff inside your ear. I really don't remember what they call those. But they do work nicely. Anyway, I took it out of my ear and soon found out why my ear was itching. The sheeth on the wire had gotten pushed away somehow, leaving the wire touching my skin. I find it amusing that it was merely an itch and not something more like a tingle or somesuch. Oh, and while we're on mp3s. Never play 192kb mp3s across linux nfs (unfsd...probably should try knfsd) while also running an XEmacs window across the same ether connection. Too much data. Too little width.

Anyway. Got a message from Dan denoting how badly my simulation figures were off. In reality, I need to be able to simulate an instruction in 27ns. (Never let me do math.) Which, of course is absolutly impossible (particularly on PCs). So, the goal is no longer real-time simulation, but just mere simulation. Which is a much less lofty goal. And one I can actually have a hope at reaching.

I really did attempt to concentrate at various times today. But something usually ended up distracting me.

And something especially distracting was the weather. First real monsoonish day of the year. Nice big winds and all that. It was so pleasant, that I went out and sat in it for an hour or so.

Oh, and for those wanting to know why I didn't see any fireworks yesterday, the Republic will inform you (it can also inform you about the death involved). [I think its also amusing how the Diamondbacks game is more important than Clinton wandering around in South Phoenix. See what happens when I actually read the news? Scary...]

Learn all about Tellurium.

I did spend some time fiddling with the simulator today. Implemented the PC and Status stacks. So now I have both Proper IRQ handling and returnable call support (ie, rti and rts actual do stuff now). Also did register flipping (I like the way I did that -- I won't explain it because its too simple to make sense out of context). I think I did some other things too. The actual TODO is really quite short considering this project is only a couple of days old. Everything left revolves around decoding instructions. There's really very little architecture stuff left to do. Mainly ALU and shifter ops. And conditionals. I've yet to do any conditions (rts/call/etc all ignore them for now).

Oh, the other thing I did was impelement the itty bitty feature of "software rebooting". This fixed my problem of the random jumping in the sample code. And now I'm into a really messy portion of it that uses mostly instructions that I don't have decodes for yet.

I doubt anyone is reading this live. Because it seems Primenet has decided to die again. Oh, too bad. Oh, wait. That is bad. I need Speedchoice. Now (because I also want to grab a Debian CD to install on zaphod, but I really don't want to do that over 56k). Looks like their Radius server is screwed. Oh, wait, there it is...

Bah. Dangerous day.

Found no reason to get up before noontime, so I didn't. That sounds right logical enough now doesn't it.

Somehow or another I forgot that I woke up and blantently attempted to wake up twice. But in doing so, I just got bored and started fiddling with various things laying around the garage (such as removing a presumidly faulty 16mb of ram from zaphod and taking a look at the ICE board uses across that serial connection (no, I'm not going to write my own ICE software...yet ;)).

There was then some lovely brotworst laying about, and I decided it would be a nice time to eat that. So in my usual single-minded attitute, I did. And it was good. I then proceded to the crowd outdoors, at which point, we sat and sat until about 2300, when we decided there were going to be no fireballs in the sky, as is tradition. After talking to David some time later, I found out that that was because of some hilarious stuff that happened. Only further proving my ealier statements about that place being full of idiots.

So, no fireworks. Big loss.

Conversing with Orb about various things (you're right, that hasnt happened in quite some time). For some reason, he was looking at this picture and said "You know, that picture of you looks like anakin skywalker threw a temper-tantrum and got arrested." Yes, that amused me.

Oh well. Back to work on the simulator tomorrow. Remember...release in a week. I'm sure everyone is quite unanxious to see my code.

Okay, yet another entry in this Volume of Blatently Pessimistic Diatribe...

Slept in yet again (til nearly noon this time). Managed to do nothing for quite some time.

Fixed some major atrocities in my adsp-2181 emulator. I had neglected to remember the difference between 24bit, 16bit, and 8bit memory. Aparently I was accessing 24bit PM buffer bytewise, and had completely forgotten about only accessing 16bit DM buffers wordwise. Oops. So that cleared up all my stupid DMA problems. And now I have both 24bit PM DMA (BTYPE=0x00) and 16bit DM DMA (BYTPE=0x01) implemented. I will only implement 8bit DM and 8bit DM byte-swapped if I really need to. The sample code now gets through all its boot loaders, except what happens after the last one (the one at 0x009e), which jumps to somewhere quite randomly. Also, the IRQ routines now actually modify the PC to jump to the interrupt vector, however, it still doesn't push the status and PC stacks because I have yet to implement stacks. Oh, and I did some profiling. NOPs have an overhead of 3usec. Register assignments take between 3 and 4usec. DM accesses take between 3 and 5usec (depending on page flipping, etc). All in all, that should be low enough overhead to be Real Time. Hopefully. [Oh, I implemented DM page flipping as well.] Next step is stacks, I suppose.

That didn't take a long while (though it did require messing with the ICE to figure out just what DMA parameters are in what units). Anyway, haven't done much since then. Spent a large portion of the later evening and early morning mocking the sheer stupidity of television (except the simpsons).

Oh, aparently SpeedChoice has a residential two-way microwave service now. I think I will call out a consultant guy next week. Its a mere $55/mnth. The plan is to split the cost with my parents. Yay. Hopefully we're not too far out. It will be nice to have multimegabit downstream (and nice for you to have 35kbit/sec guarenteed upstream). Hopefully they give fixed IPs. This way I could get rid of Primenet and a USwest line.

Anyway. Aparently today is somesort of pyromaniacal holiday. However, as an improvement to last year, we are not going up to that wretched place where we went last year ("The Place Full of Idiots Who Think Shining Bright Multi-Segmented Stadium Lights in My Bedroom Window Is Somehow Amusing", as I put it a year ago) to watch said holiday activities. We will be watching things blow up from the damp, humid, overheated comfort of our backyard during some sort of party-like-structured event. Odd indeed.

Well so this was a rather overly worthless meaningless group of hours. Woke up around 1100 (I had very little urge to get up even then -- bad sign from the start). Took a shower. Read mail. Found a very interesting bit of mail.

Much thanks goes to Dirk-Willem van Gulik, who managed to sucessfully dig up a copy of the Novell Boot ROM Developer's Guide. Which happens to contain a short section with some very interesting RPL info in it. So, I updated my RPL information. Of course, I then spent the whole rest of the day sulking about how I still couldn't get it to work. And that was AFTER I found the two stupid bugs that I was tripping on before. Firstly, the protocol type field of the ethernet headers coming from the server must be 0x003d for FOUND frames and 0x05dc for Data frames (not the 0x0056 that the client uses). Secondly, I had the source and destination arguments to my raw ethernet writer mixed up and was sending things to myself. That was a stupid mistake. Fixing both of those got me to getting the 8227 to send me a Send File Request, but then it wont listen to my file data frames. Of which there are 270 of. With NO handshake. Stupid dumb protocols. Still no working. And its after midnight.

So that leaves me here, eating my Dreyer's, ignoring Tara's insistance that I am going insane. And goddamnit, my ice cream melted. Tara, stop that. Ha-HA! I finished it before it all melted! Take that, evil undoer!

In any case, I should just generally go to bed now, before I cast out my antiproductiveness on the rest of you. But not without the gratuitous random links...

The Ultimate in Waste-Everyone's-Time-Especially-Your-Own-And-Your-Legos'. I'd rather use standard 8balls. At least they don't have race conditions that reorder my queue. And if that Furby just isn't making you happy anymore, you can always rebuild it. You know, you just gotta wonder about people who watch ASCII porn (of course, the people who code it are perfectly normal). Okay, thats enough of that. Time to grab a bedtime mtndew.

Figured I'd pop in and do a rotate. Maybe I'll write a bit while I'm here...

Last month is here; though you could of course have just gone to the index to get that.

So anyhow. Yesterday was spent basically hacking together the rudiments of an ADSP-21xx (specifically, 2181) disassembler. Then started working with the ICE. Determined that using Win95 to get serious work done was an atrocity, not to mention not very productive. So I started working on a 2181 simulator. Yes, a simulator, from scratch. [Note that I really didn't think this was a big deal...its just a dinky 16bit DSP afterall...I still don't think its a really big deal. Oh well.] I believe I had a basic framework written before I went to bed (first around 0100, again at 0200). However, still no instruction decode and no interrupts. But I did have the Big Ass Struct laid out. And the basics of reading/writing program and data memories (the 2181 has no cache and no on-board RAM...this was easy). Its in fact mostly easy. (Oh, I got up after going to bed the first time because I remembered I hadn't talked to Tara about SATs...which I needed to talk about before I forgot again I guess...(a measly duple of 660verbal+620math=1280, btw).)

I slept in a bit this morning...0900 or so (I've somehow managed to contrive a system of seminormalcy with my sleeping habits...truely odd indeed). Got up and stared at my simulation hackery. Decided it was utterly nifty and needed to continue. I was interrupted, however, by the sudden and unexpected arrival of random CPUs and RAM falling from the doorway. A iP200 classic and 2x32mb EDO sticks. The latter went into zaphod (now at 80mb). The former was the cause of an unexpectetly long downtime for deepthought. Yes, deepthought is no longer SMP. Its now merely UP. But it is two-tenths of a bogomip faster! Now just 1x200mhz instead of 2x100mhz. I think it will suit my needs better. Besides, we all know Pentium SMP is dreadfully contentious. I can't really notice anything. It streams mp3s (icecast) about the same, which is really the only performance-conscious thing it ever does (besides opening up my 35000+ message mailboxes...but nothing does that very well...in fact, I keep indexes of that size around merely to remind myself that (1) I don't have infinite cycles and (2) pine flails somewhat miserably on most everything (yes, pine sucks, but its the only thing I can remember how to use)). Anyway, so mostly all of the morning and a chunk of the afternoon were spent slowly fiddling with deepthought [note to any random self, especially mine: fractional multiplier ratios are the same as fixed-point ones, just not reduced -- this DSP work is making me use funny words...fixed-point?!]. Anyway, I'm sure deepthought would be happy to accept the donation of another classic (non-mmx) iP200 if anyone feels inclined to pleasure it with one...

After calming down from the heart attack I was nearing from removing and then replacing the three dozen screws that hold deepthought together, I started simulator work again. Three hours later, I had a mostly functional IRQ firing framework (minus the actual vector jump -- I still don't have jumps, or any of the stacks for that matter, implemented) working along with the beginnings of the disassembler hack further hacked to support actual operations on the decoded opcodes. I also had Pete and Dan standing nearby at that point (who each aparently had an overly stimulating drive over here). Displayed my rather useless simulator (I was in fact hoping to have the sample code getting past at least the boot loaders and into jumping and calling and real interrupts, however, the many problems I had when implementing the three-cycle deadspot between an IRQ fire and the actual vector response, along with getting the tremendously inane idle instruction to work with all that, delayed my proceedings a few more hours than planned (IRQs were suppose to be a one-hour project, I still haven't finished them), and still the only IRQ I have tested AT ALL is the BDMA completion interrupt -- I'm happy to say that BDMA emulation should be complete now, though thats a minor function, at best). [Sheesh I'm rambling...my logs are starting to look like LISP...] I really forgot where I was headed with this paragraph. I think I'll just start a new one.

They agreed with the decision of doing a simulator, I think (little do they know I did it more to learn the DSP more than because of Win95's mad deaths, but I guess they know now that I just said that...). Simulator good. Working software good. [nod] I seem to have forgotten what else we talked about. It will come back in a few days, I'm sure (about five point four two nanofortnights after I need it). Oh, also... SHARC simulator nearly impossible (has cache, on-board ram, a four-instruction prefetcher (like the 2181 IRQ delay, but bigger, longer, and stranger), one of hell of a rubbishpile called a memory map, and other things). Using Analog's software good for SHARC (at least until I run out of better things to do). Oh, and unless there's any objections, the GPL'ed source for the ADSP-2181 core simulator will be popping up on your local internet in a week or so, when I have a clean, useful package (actually, I've managed to keep it rather clean, I'm actually rather proud of myself on that front!). We'll see where the SHARC stuff goes when I get to it. I'm not even CLOSE to done with the 2181, so its going to be awhile before I should even be really thinking about the drool-worthy SHARCs...

Good laugh at a random human, with the added bonus of involving overly large amounts of electricity: here.

Oh, and two days, nearly 60kb of code, and eleven mtndew's later, I wrote a log entry. Erm...make that twelve mtndews.