Comments on: DROID Gets Amish Autofocus Fix [UPDATED]

By: YankeeDudeL

YankeeDudeL — Thu, 19 Nov 2009 06:52:24 +0000

I had an issue w/ my phone yesterday and brought it in to the VZW store. While there, the tech guy said that there were a “few patches” that were available for it that he would install. Alas, we’re green lights across the board. So maybe it’s the bug, maybe it’s the patch, but regardless, it’s working now.

By: MarvinAndroid

MarvinAndroid — Wed, 18 Nov 2009 20:13:00 +0000

I’m I the only one who read this explanation and thought it was a complete joke? Now that I’ve read some believable reasons why I guess it makes sense.. but when I first read that post I was thinking “yeah right.. this guy is having some fun with us”. lol

By: a

Wed, 18 Nov 2009 15:32:40 +0000

550mhz arm cortex a8 cpu
Android 2.0
Google Nav
16gb sd card
3.7 inch 854×480 capacitive lcd touch screen
physical qwerty keyboard
5mp autoblur camera

By: DroidRiffic

DroidRiffic — Wed, 18 Nov 2009 13:19:54 +0000

Perhaps their QA process lasts all of 24.5 days so they didn’t notice the bug.

By: Bradbury

Bradbury — Wed, 18 Nov 2009 13:19:05 +0000

You wouldn’t see this bug if testing was in the 24.5day window, duh.

By: BearsSuperfan

BearsSuperfan — Wed, 18 Nov 2009 10:00:29 +0000

Doesn’t Motorola bother to test their garbage before they put it on the market?

By: KBlack

KBlack — Wed, 18 Nov 2009 06:01:49 +0000

That’s one of the most interesting bugs I’ve ever seen.

By: kevinh52

kevinh52 — Wed, 18 Nov 2009 05:52:54 +0000

i just got green corners 15 times in a row, and didn’t do anything to the lens. I guess its in the day cycle. use to get about 2 green out out of 10.

By: Wing

Wing — Wed, 18 Nov 2009 01:55:54 +0000

From:

http://www.engadget.com/2009/11/17/motorola-droid-camera-autofocus-fixed-in-secrecy/3#comments

Tam @ Nov 17th 2009 3:21PM
I think I know the nature of the bug. The number of milliseconds that can be counted by a 32-bit number is approximately 49.7 days (i.e. about twice “24.5″).

Now, that’s for an “unsigned” (non-negative) 32-bit integer. For a signed 32-bit quantity, once the number of milliseconds exceeds ~24.8 days, the millisecond count becomes the lowest negative number (and starts counting back up towards zero).

The bug is that code is using a signed 32-bit integer math instead of unsigned so for half of the span of 32 bits (24.8 days) the timer math used for the autofocus is choking on negative millisecond values.

By: Thanassi44

Thanassi44 — Wed, 18 Nov 2009 01:26:50 +0000

so let me get this straight. lol. during the “red” cycle, if i want to take good pictures, i have to trick my phone into thinking its Doc Brown. updates please, like now.

By: Anton S

Anton S — Wed, 18 Nov 2009 00:17:08 +0000

The 24.5 days is a rounded value for 24.85513481481…, which is 2^31 milliseconds (*signed* 32 bit integer used for milliseconds). The focusing could use timestamps for calculating how far to move the focus-point during a given time-span, by measuring how fast the focus-engine is.

Maybe during each alternating 24.5 day period the timestamps obtained switch from positive in one period to negative in another. The ones with the negative timestamp values are unexpected and could be exhibiting the problem.

But then again… the cause could be entirely different… just guessing here… :-)

By: Ross

Ross — Wed, 18 Nov 2009 00:11:54 +0000

How many people went ahead and buffed off the scratch resistant coating on their camera lens because of this prematurely posted “fix”? Some small degree of fact checking might have been nice before spreading this rumor. For that matter, the update doesn’t even bother to refute the “clean the lens” theory.

As for how a timestamp could cause this, if someone used the clock as a cheap RNG (counting minutes), and stored the result in a signed long, you’d flip from negatives to positives about evert 24 days and change. I have no idea if this is what is actually happening, but the math works.

By: Brad

Brad — Tue, 17 Nov 2009 21:16:53 +0000

Okay, that’s weird. When I posted my earlier comment, it showed me as the second poster. When I refreshed the page now, it shows my comment way down the page, posted after comments that made my comment irrelevant. Go figure. Sorry about that guys.
While we’re on the subject though, how the heck can a timestamp cause the camera to not focus? That has gotta be hands-down one of the weirdest bugs I’ve ever heard of, and I work with computers and other technological gadgetry for a living, meaning I’ve seen some pretty weird stuff already.

By: Verges

Verges — Tue, 17 Nov 2009 20:23:58 +0000

Actually there was an OTA fix, but the Droid is so powerful that when you change the date you are actually going back in time to before the fix was issued.

By: kidphat

kidphat — Tue, 17 Nov 2009 19:46:39 +0000

@Brad, it could be because no one actually knows whether it was in fact a silent OTA. There is rumor that this is a cyclical bug that auto-corrected itself.