Silicon Valley Redneck

Whew!

Sometimes things come out right....

Off at the client site again this afternoon, checking that I'm looking at the right polarity for all the signals that come into my controller from their box.  Comes time to test the excessive-pulse-width signal (a rough check in hardware, as my precision pulse-width checking depends on the pulse actually having a trailing edge).

Oops.  It doesn't latch.  An overly-wide input pulse generates about a 300 microsecond error pulse, which my polling doesn't often catch.

But wait!  I routed that signal in on P0.15 for a reason!  P0.15 can also be configured as EINT2, for prompt response and edge-triggeredness!

Now, how do I go about that... and (eep!) is the polarity programmable?

Aha!  Yes, the pin configuration register is used to turn it into EINT2, and the EXTMODE and EXTPOLAR registers of the LPC2134 (as this happens to be) are used to configure edge vs. level, and the polarity.  All's well!

Curious beliefs

Browsing Slashdot on my lunch break, I came across this item, which contains the assertion:

(without toes you cannot remain standing or walk, btw)

Now, I've often heard this one.  Oddly, Snopes doesn't seem to have addressed it.  A couple of the Slashdot commenters note that they've known toeless people who could walk.  But, really, now...

While I'm personally disinclined to do this, I'm aware that there are people out there who routinely encase their feet in rigid structures which completely immobilize their toes.  (I think these contraptions might be called "shoes" or "boots" or something.)  Despite having their toes effectively nonfunctional, these people still somehow manage to walk.

Given the large number of people who wear shoes, surely almost everyone must know someone who does.  So... how the heck did the "can't walk without toes" belief get started?  How does it persist?

I'll admit that walking without the use of toes can sometimes be challenging; the first few times I put on ice skates, I got toe cramps from trying to grab the ice when (a) my toes were immobilized within the boots, and (b) the ice was 3 inches below toe level anyway.

Perhaps the statement should have been:

(without toes you cannot pick up dropped popcorn without bending down, btw)

Update: I just thought of another test.  I'm going to set a 2x4 on my patio, and fearlessly attempt to stand on it with my toes danging off the edge, into free space!

...Back now.  Not only could I stand just fine with only the back halves of my feet touching the board, but I could step sideways without trouble.  Walking forward seemed inadvisable.  This leads to the new, improved reformulation:

(without having solid ground beneath your toes, you probably shouldn't step forward, btw)

Uh-oh.

Home from errands.  Check Slashdot.  Find link to USA Today article announcing that the largest object in the known universe turns out to be a giant space amoeba.

Should I just go hide under my bed now?

Lunacy

If lunacy happens around the time of the new moon, should it be called novolunacy instead?  Things have sure seemed full-moonish lately, despite being two weeks out of phase.

I suppose the recent heat wave might have been a factor, but... how much does that really contribute to crazy driving?  And why has the crazy driving persisted?  News reports on the radio this morning sounded like crazy season was still very much on, with (assuming I heard it correctly) multiple fatal crazy incidents in the area.

Tuesday evening was time for the nominally weekly trip to the mice rink.  Not only was it crowded, but it was crowded with crazies, zigzagging madly through the mob.  This didn't do wonders for the condition of the ice, which rapidly became excessively anisotropic, not to mention rough.  The scene reminded me of the time, about 15 years ago, that I went to an outdoor shooting range on the first dry weekend of spring, and it seemed like all the nuts had come out of hibernation (e.g., yuppies with shiny new wondernines and only the vaguest notion of how the sights worked, and none of range protocol).

A few days ago, I heard on the radio a fine example of economic lunacy: a spokesman for some initiative to tax oil (and other energy?) companies that do business in California, and to use the new tax to fund "alternative energy" projects, saying that (paraphrased from memory) "This will make the oil companies pay, not the consumers."  Er... and the oil companies will get the money to pay this tax from where, exactly?  If not from consumers, then from... er... robbing banks?  Raiding the pension funds that have invested in the oil company stocks?  Some folks are real good at stringin' words together, and not so good at mindin' what they mean.

Progress!

A couple of days ago, I tracked down the problem with AGROS that was making it excessively toolchain-sensitive.  Turned out it had a couple of goofs in the assembly/C interface in the semaphore-handling code.

Aside from that, my implementation of priority queues was FUBAR - the data structure was fine, but I was handling two of the cases entirely wrong.  That being fixed, task scheduling is no longer sensitive to the order of task creation.

And so it came to pass that I spent yesterday afternoon at the client site, making lights blink, tweaking the application code (now that I can compile the code on my laptop and have it work), and seeing how the client's hardware works.

I think this will make for a happy client... several people were playing spectator at various times, and they were greatly relieved to see their new baby actually doing something.

There are still some fuzzy areas to be resolved (there always are, when there's a user interface, even when the user interface is just lights and a toggle switch), but it's aliiiiive!

Stupid phrase of the day

"Scalar weapon."

Really, I have no idea what the heck it's supposed to mean, but apparently some crackpots have been claiming the Rooskies got 'em.

But... but... aren't vector weapons much more useful?  After all, a vector has a magnitude and a direction; a scalar has only a magnitude.  A directionless weapon doesn't sound like a notably good idea.  Though I suppose hand grenades might qualify.

Well, that's one way to cool off...

After three nights in a row of not getting much sleep because it was so darn hot, I've been kinda dozing off at my desk.

This made it seem like time for a nap.  Maybe out on the front porch, with the fog spritzer going?

Attach spritzer to hose.  Turn on water.  Fog commences.  Spritzer falls over.  Hm?  Oh, the hose seems to have gotten a bit torqued.  Give it a couple of turns in the direction the hose is trying to twist, set it back up, and GOOOSH!  Aieeee!

I'd been meaning to replace the rather decrepit old garden hoses in the back yard.  Apparently the one in the front yard needed replacing too.

Good thing I hadn't taken my laptop out there to try to get some work done.

Well, I'm cooled off now.

And I've got it better than some people... outside temperature is in the upper 80s (F), down from 100ish over the weekend, yesterday's electrical outage was an exception, not the rule, and nobody's threatened to kill me for wearing shorts in public.  Kinda puts my little inconveniences in perspective.

Poodle with a megaphone

My parents' poodle, in true poodle fashion, recently mouthed off to a dog twice her size... which, during the bark-off, managed to slip his leash, cross the street, and get physical.  See the result:

"Look - it's Radley Barko!"

"Xmphlaca bi flucalaka un Bijnana y Aquaholder."  "Gne!"

"I've named him Hoover."

Prequels

It's just come to my attention that Dave Barry (with the assistance of Ridley Pearson, whoever he is) has gotten into the prequel business.

These guys were at Kepler's yesterday afternoon, yakking about their new prequel to Peter Pan, reading bits of it, and so forth.  Personally, I think they were just looking for an air-conditioned place to hang out, and I blame Barry for bringing Florida weather to the Bay Area.

Anyway, today's rumor is that they'll be branching out from Peter Pan prequels to other things.  Prequels to the Bible, The Art of Computer Programming, Mein Kampf, and On Beyond Zebra are rumored to be in the queue.

Isn't that right, Mr. Highly-Placed Anonymous Source In the Publishing Industry?

<sockpuppet>Oh, yes, SVR.  In fact, their publisher is also rumored to be seriously considering prequels to The Wealth of Nations, Gulliver's Travels, and FM 23-65.</sockpuppet>

And there you have it - confirmed by a highly-placed source!

More unsolicited reboots

I guess I shouldn't complain about Windows rebooting after I told it not to....

Yesterday afternoon, my FreeBSD server rebooted itself, after shutting down a measly 3 hours into a 5-hour blackout.  Sheesh.

Maybe I should have left it the car keys, so it could go to the store for more batteries when the power was out for so long.

(Perhaps I should also try to figure out why clamd didn't start up automatically when power came back on.  The system was alive and serving up web pages when I got home, but bouncing all incoming mail with "try again later" status because the virus checker wasn't running.)

Goofy idea!  Plug the existing UPS into the spare biggish UPS that used to be at Humball.  When power fails, the downstream UPS doesn't see it for a while, because it's running off the upstream UPS.  When the upstream UPS's battery runs down, the computer continues running off the downstream UPS....

Some project names just shouldn't be reused

So I get home from a meeting, check my mail, check Slashdot... what's this headline???? "Project Orion to Bring U.S. Back to the Moon" - what????

Well, the linked story is about the new generation of manned space hardware - the Crew Exploration Vehicle.  Apparently this has gotten named "Orion."

Can any space enthusiast hear (or see) that name without thinking of Old Boom-Boom?

Missed an anniversary...

... but I assume these three guys have better things to do than be mad at me for forgetting.

Laurence at IMAO, unfortunately, didn't forget. As you value your sanity, don't click his link!

As a sane alternative to Laurence's link, consider clicking here and scrolling down to "Hope Eyrie."  Or scare up a copy of "A Toast for Unknown Heroes."  (Sorry, no MP3 of the latter; I have one, but no permission to put it up on the Interweb thingy.)

Attention Microsoft: No means No!

So far, I've been working on this project in the familiar environment of my Linux workstation for editing and compiling, and the Windows machine in the lab for talking to the debug pod.

Real Soon Now, I'll be wanting to do development on the go... namely, at the client site, where (a) the client will be able to see the progress and provide feedback on details, and (b) it's air-conditioned.

So, being as how the debug-pod software is Windows-only, this calls for booting up the laptop in Windows and installing Cygwin and an ARM toolchain.

Turn on laptop, select Windows, browse to cygwin.com, start the installer...

Windows Update pops up a box asking permission to reboot.  I tell it no.

...Select the software to install under Cygwin, tell it to go do it...

Off to the kitchen.  Milk, cream, sugar, Donvier, peel & dissect a peach, carry Donvier back to where the laptop is...

...Is running Linux.  Somebody, I say somebody, went and rebooted my laptop.  Right in the middle of the Cygwin install. That just ain't right, folks!

Fortunately, the Cygwin installer, on being restarted, picked up more or less where it left off.  For now, I've left the laptop pondering Windows Update again.  Just in case it decides it needs another spontaneous reboot, and won't take no for an answer.

Yeesh, the things people in the Windows world put up with...!  I guess I've gotten spoiled with my FreeBSD server (reboot once a year, whether it needs it or not) and Linux workstation (reboot every couple of months, or when there's some bizarre wedgie, typically involving OpenGL, or when I've built a new kernel configuration).

Snazzle fratz!

This isn't a Soviet washing machine in the strictest sense; it appears to be more a case of a de facto standard (i.e., a grand game of follow the lemming) having preserved half of a useful feature - presumably because it had always been there - but lost the other half - perhaps because nobody remembered what it was for - thereby becoming more of a defective standard.

Yes, today's gripe is the UART in the LPC21xx microcontrollers, and in particular the way it was copied from the ubiquitous 16C550, which was copied... well, eventually it goes back to the 8250, and before that to the old 40-pin UARTs (AY-n-nnnn, whatever the numbers were).

Y'see, there are basically two transmitter status bits.  One says it's time to feed the transmitter (originally THRE, Transmitter Holding Register Empty); the other says the transmitter has actually run dry (TEMT, Transmitter EMpTy).

Now, on a full-duplex line, all you need is the THRE bit, which tells you that you can stuff more bytes into the transmitter.  All the current UARTs provide an interrupt on this condition.  This works nicely.

For half-duplex, though, it's really useful to know when the last bit has been transmitted, so that you can turn the channel around.  Doing this when the transmitter has just started shifting out the last character (the THRE interrupt) is simply not acceptable.

And so it comes to pass that I, attempting to implement a half-duplex line discipline, wish to receive an interrupt on TEMT.

Oops.

It appears that a TEMT status interrupt isn't available.

Somewhere along the line - and I'm looking at you, Mr. National Semiconductor, and your INS82C50A, and your Series 32000 Databook for June 1985 - the canonical UART with microprocessor bus interface was designed with a TEMT status bit that can be polled, but no ability to get an interrupt when it goes active.

Grrrr.

This means that, for the half-duplex scenario, I need to set an alarm after loading the buffer for the last time (which may be up to 16 characters), and poll for the freakin' TEMT bit, at system-timer intervals.

Life just got a whole bunch more complicated.

Update: not as complicated as I'd feared.  Add a static variable containing the line status register contents for each UART as of the last TX or line-status interrupt, and a routine (should be an optionally-linked-in module, really) called from the 1ms polling task to check for TEMT having changed for the higher, and in any case to update the old-LSR variable.

I'm not crazy about the up-to-1ms turnaround time after transmission, but, since the RS-485 interface in the application at hand will only be running at 9600 Baud, that's only one character time, and it so happens that the other side is expected to transmit infrequently.

More fun with gcc

Somewhere along the line, I really need to figure out why AGROS compiled with -O2 works, but with -Os it fails miserably and mysteriously.

Given that the intended target is little-bitty microcontrollers, being able to optimize for size is kinda important.

Right at the moment, AGROS (with device drivers for UART, I²C master, SPI master, system timer, ADC, and pulse capture) plus a small application takes up a bit over 16K of flash (compiled with -O2), which is fine when the app at hand runs on an LPC2134 (with 128K of flash), but I'd like to have it work usefully, including a full set of device drivers and room for some actual application code, on an LPC2102 (16K flash, 4K RAM, $2.50 in hundreds).  Well, the LPC2103 (32K flash, 8K RAM) is only $3 in hundreds, but still....

I suppose I could start from -O2 and selectively turn off things like loop unrolling and whatever it is that makes switch statements generate largish tables....

Hey, some of the old OSes fit easily in little teeny address spaces, back in the 16-bit minicomputer days.  Of course, they were dealing with simpler peripherals, only worked with one vendor's hardware, were written in assembly language, and must have been nightmares to maintain.  (Dusts off old RDOS manual.)  Oh, and each instruction took up 2 bytes in those days, not 4 (I'm not using thumb instructions for system code).

Update: actually, the size difference between -O2 and -Os isn't much... around 800 bytes out of 16K (though that 16K does include various data which shouldn't be affected by the optimization options).  When I run the -Os version for a while, then stop it and run a panic dump, it seems that the ready queue is empty... as is the task list.  So, somehow my data structures aren't getting built up.

Oh, well... for a few hundred fiddly bytes, I'm not in the mood to worry overmuch about what the compiler is doing differently.  For now, I'll just stick with -O2, get the current project done and delivered, and then contemplate some code restructuring to save space.

Bar code attack!

My mailbox has sprouted a mysterious USPS sticker with a bar code, my address, and the words "MANAGED SERVICE POINT."

Does this mean my mailbox is now SNMP-enabled?  It looks just the same as ever, except for the sticker: a simple metal box with a hinged top.

Did my mail service get managed for some reason?  It's not like I get a large volume of mail.  Maybe it's the nature of the mail I get?  Did I do something to get the USPS's attention?

Perhaps, when setting up a hydroponic plutonium farm in the garage, it's best not to buy the heavy water by mail order?

Two trains of thought just collided

Taking a break between milling some connector shells down to fit in the space available and returning to that software project, I checked Slashdot...

...which linked to a Businessweek article about oddball iPod accessories...

...which included this dingus: a locking iPod anchor, like the locking laptop anchors wherewith we're all familiar.

Apparently it's meant to anchor a small, readily-stolen iPod to a small, readily-stolen backpack or purse.  Clever, huh?

Now, it happens that just yesterday there was a story on the radio about the things government agencies had been buying lately, like a giant plasma TV for, IIRC, FEMA.  (Wait a minute... wasn't the big complaint during Katrina that FEMA wasn't watching TV, and therefore wasn't up to speed on all the sensational tales of cannibalism in the Superdome?)  Anyway, some agency or other had bought a bunch of iPods; this was presented as scandalous, though the official justification for them wasn't presented, and I had a vague thought that they might be used for listening to, e.g., recorded presentations while commuting or some such.

Combining these two...

If you're using a government-issue iPod to listen to classified briefings, shouldn't it be locked to your wrist, government attaché case style?  This calls for an iCuff.  New product idea!

...With a pair of iCuffs, you'd have stereo, but might need some rather, er, specialized content.

Thought for a hot day

One thing I haven't seen yet - though I admit I haven't been looking very hard:

Energy pops!

Frozen Red Bull onna stick!

Just the thing for caffeinating on a hot day!

Yeah, I know, in the freezer case they'd look like kid stuff:

"And if you went to gobble a snack at Baskin,
You could see the greatest horror man has known:
For little kids, for nothing more than askin',
Could get their coffee in an ice cream cone!"
            - Gary McGath,
            "The Alternate World Temperance Union"

A thought for today

Mosquitoes are nature's little Al Gores - flying around fighting global warming by attacking sources of CO₂.

(Why, yes, there were rather a lot of mosquitoes about at Big Basin yesterday.)

Still bashing on that project...

Looks like AGROS, running on a 60 MHz ARM7, has a typical interrupt latency of 5 microseconds (not too shabby), but occasional excursions of up to 55 microseconds... which suggests that I need to take a careful look at the system code that turns off interrupts, and of course the length of interrupt routines (I'm not currently allowing nested interrupts).

It also means that my use of two timer capture registers to measure pulse width and interval (by capturing a rising edge and a falling edge, and remembering the previous falling edge) will get flakey as the pulse frequency gets up around 20 kHz, so a certain amount of trickery may be called for.  It'll capture the width reliably (or else come up with a negative width, which is detectable), but detecting dropped intervals is problematic.  Adding an external flip-flop and a third timer channel might solve everything, but that's not in the current hardware.

Maybe I'll move pulse capture to an FIQ routine, leave FIQ enabled in most of the system code, and be very careful about what the FIQ routine does.

There's also some weirdness with the timer capture interrupt if I have an ADC polling task running.  Another mystery to sort out...!