Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Linux Hardware IT

Un-Bricking Linux Plug Computers 68

An anonymous reader writes "Accidentally 'bricking' a little Linux plug computer doesn't have to be forever. This is a good howto on repairing a non-booting Linux plug computer. For example if it uses the uBoot environment then it already has some good built-in recovery tools. The article also mentions ESIA, the Sheevaplug installer, openocd, and GuruPlug."
This discussion has been archived. No new comments can be posted.

Un-Bricking Linux Plug Computers

Comments Filter:
  • Oblig. pedantry (Score:5, Interesting)

    by butalearner ( 1235200 ) on Thursday February 10, 2011 @03:33PM (#35165268)

    Isn't the fact that you can "unbrick" it mean it's not really bricked?

    Also I think this requires JTAG, which comes with Sheevaplug but is sold separately from Guruplug, and AFAIK is not available on the PogoPlug, et al. So, not quite so useful for me (I just picked up a PogoPlug on the cheap with the intent of running Plugbox Linux [plugapps.com].

    • by Anonymous Coward

      So what you're saying is that you don't think that word means what he thinks it means?

    • Unless you've decided to mess with uboot(which is likely a bad idea, unless you are totally comfortable with JTAG...), you shouldn't need anything scarier than some way of speaking RS-232 at suitably low voltage. Re-flashing a device over a serial line is tedious; but not terribly challenging, and even the various Kirkwood-platform products that are "no user-serviceable parts inside" almost certainly have an accessible serial header somewhere, albeit likely unpopulated or even unmarked...

      If you've gone a
      • some way of speaking RS-232 at suitably low voltage.

        Ok, I'll bite. How do you speak RS-232 "at suitably low voltage", since RS-232 includes the voltage limits as part of the standard? What do you mean by this?

        • Well, -15v is pretty low compared to +15v. Both are officially the lower and upper bounds of the RS-232 signaling voltage range, after all. Certainly, the common -12v "mark" level is much lower (i.e., less positive) than the 3.3v or 5v typical for logic levels nowadays.

          Yes, GPP mistakenly conflated "serial" and "RS-232". I'm sure he meant "logic-voltage asynchronous serial". And I'm sure you understood, but that sure was good electronics pedantry.

        • Like this: http://www.instructables.com/id/Assembling-a-RS232-to-TTL-Serial-Adapter/ [instructables.com] . You can use it to speak at levels lower than TTL (5V)
        • Re:Oblig. pedantry (Score:5, Informative)

          by fuzzyfuzzyfungus ( 1223518 ) on Thursday February 10, 2011 @04:25PM (#35165862) Journal
          In principle, you are correct. In practice, an increasing number of devices(especially space or cost constrained ones) implement "RS-232" that behaves pretty much exactly the same way as would be expected by anything post-20mA loop, with the exception of voltage. For cost and board space reasons(and because they are not intending to address the "terminal across the electrically noisy building from the minicomputer" use case), they omit any voltage conversion or protection circuitry and simply depend on the attached hardware to do either 5v/0v or 3.3v/0v, or whatever their logic-level happens to be.

          Even an increasing number of supposedly-genuine RS-232 devices(especially laptops) don't generate anything near the +-12 swing of the old days. 12v/0v is more likely, or even 5v/0v, though such devices tend to, at least, have better tolerance for over-voltage than the little guys do.

          Because it is so close to RS-232(all you need is a dumb level converter, no logic/protocol translation required), I tend to fall into calling it "RS-232" colloquially, even though it technically isn't.
          • by sjames ( 1099 )

            It does all get a bit confusing and has been a bit of a frog boiling. The first deviations would fully inter-operate with fully compliant RS-232 devices provided the cable wasn't too long, so they called it RS-232 even though it wasn't quite. It's been pushing the limits ever since until now we get stuck with "rs-232" connections that will actually burn out if they are connected to a genuine rs-232 device. I was more or less OK with the first deviations, but the latter situation is past the limits IMHO. We

            • Considering how cheap it is to make the inputs real-RS-232-tolerant, you'd think they would do so. I mean, all that's needed are a couple of resistors, a transistor, and a diode (to prevent reverse breakdown of the base), as a part of the chip.
          • For cost and board space reasons(and because they are not intending to address the "terminal across the electrically noisy building from the minicomputer" use case), they omit any voltage conversion or protection circuitry and simply depend on the attached hardware to do either 5v/0v or 3.3v/0v, or whatever their logic-level happens to be.

            Actually, it's because they want to run on 5.0v and 3.3v and don't want to have to include an inverter to produce 12V to do RS232. All of that stuff is handled in the level converter chip so you don't need any of it onboard. And they will only use serial for the boot loader installation process, and the cable for connecting almost certainly has the serial interface IC built in.

        • He probably means TTL @ 0 - 5 VDC. http://en.wikipedia.org/wiki/RS-232#Related_standards [wikipedia.org]
      • by Nursie ( 632944 )

        Meh... a lot of folks changed uboot on their sheevaplug. Someone in the community released a version that was patched to be able to initialise and read from the SD slot so that you could boot from it.

        Very useful, no bricks I know of.

    • Re: (Score:2, Insightful)

      by Anonymous Coward

      You haven't done due diligence to pedantry at all; I'm disappointed.

      Isn't the fact that you can "unbrick" it mean it's not really bricked?

      How bricked is "really" bricked?

      I always understood brick to be a measure of subjective usefulness. So if my cellphone is bricked it doesn't necessarily mean it's completely unrecoverable, just that it's not practically recoverable in time to be useful to me, or it will be expensive to recover it, or I don't know how to recover it.

      I'm doubtful there's a good objective definition. For instance, you might say "having to replace hardware const

    • They were only literally bricked so they could be easily repaired. Now, if they had been figuratively bricked they'd be literal bricks.

    • by Anonymous Coward

      Why would you unbrick something that isn't bricked?

    • by KWTm ( 808824 ) on Thursday February 10, 2011 @04:14PM (#35165716) Journal

      Isn't the fact that you can "unbrick" it mean it's not really bricked?

      Hear hear! Let's recall that "brick" basically means "turning your equipment into something completely worthless, equivalent to a brick".

      I propose the following "USB cable" test:

      Has your device been rendered so unusable that you'd be willing to give it to me if I gave you a USB cable?

      If the answer is "yes", then you have bricked your device. Congratulations.

      If the answer is "no, let me work on this for a bit --I think I can restore partial functionality by pressing this reset button for 30 seconds, and then at least it will function as a glorified wall clock", then this is not "bricked".

      If you say, "This is the third time I've bricked my device --I had to SSH into it and do 'sudo reboot'" --then the brick is in your brain.

      Now, having said this, it's possible that the owner of the computer didn't know it was possible to undo the damage, in which case, yes, the device is bricked because he might as well have traded it in for a USB cable, prior to knowing how to salvage his device.

      You can substitute any marginally useful but cheap piece of equipment for "USB cable".

      Disclaimer: no, I haven't RTFA.

      • by sjames ( 1099 )

        Your definition of bricked is worthless since we already have words for burning out the hardware and anything else can be fixed given sufficiently heroic measures (including de-soldering the flash and soldering in a socket so you can re-flash extyernally)..

        Perhaps a better definition is that the device has been placed into a state where normal end-user procedures are no longer adequate to recover. Soldering pins onto the JTAG interface and re-flashing that way probably counts. Soldering in serial connection

        • by Rysc ( 136391 ) *

          A better definition is that the device has been placed in to a state where extraordinary, heroic and skilled actions are necessary to restore any kind of functionality. The average consumer can "brick" a PC to the point where he can't recover it just by sticking gum in the power port on the PSU, so let us not use them as the benchmark for "It's a brick."

      • Thank you. I have faith that Jesus sent you to speak these words of truth. Desperately needed to be said.

        While you're at it, would you mind visiting upon the bag of hammers running Phoronix? Somebody needs to explain to them what the word "regression" means (and doesn't mean).

      • I dub this the "KWT Test for Brickedness" and do name thee duke of East Looe for thy services.

    • by MattBD ( 1157291 )

      With the PogoPlug, all you're doing really on the device is stopping a shell script that's running, and installing a new bootloader. Everything else gets installed on whatever storage device you attach to it, so I think it's probably fairly difficult to properly brick it (although there are more obscure NAND installs that do have the potential to really screw it up).

      Good luck with your PogoPlug. I bought one in the new year sales, £20 off, and I ran Plugbox Linux on it for a couple of weeks, but I rea

      • With the PogoPlug, all you're doing really on the device is stopping a shell script that's running, and installing a new bootloader. Everything else gets installed on whatever storage device you attach to it, so I think it's probably fairly difficult to properly brick it (although there are more obscure NAND installs that do have the potential to really screw it up).

        I actually think Debian is a more obvious choice for these devices than an Arch-based distro as it has more packages than any other distro and has good support for ARM. In my case I really wanted a number of packages that were in Debian, such as byobu and procmail, and I use Ubuntu on the desktop and have always liked Debian-based distros, so it seemed the obvious choice.

        Funny story: I actually came across your blog yesterday as I browsed for info about setting up a mail server on the PogoPlug. I'm not sure I'll do it (getting a web server and website up and running is first), but just considering some options. I'm actually an Arch user normally, but I tend to agree that Debian might be better since I don't care about having bleeding edge stuff on there. Thanks for the info!

    • Miracle Max: It probably owes you data huh? I'll ask it.

      Inigo Montoya: It's bricked. It can't talk.

      Miracle Max: Whoo-hoo-hoo, look who knows so much. It just so happens that your friend here is only MOSTLY bricked. There's a big difference between mostly bricked and all bricked. Mostly bricked is slightly unbricked. With all bricked, well, with all bricked there's usually only one thing you can do.

      Inigo Montoya: What's that?

      Miracle Max: Go through its hard drive and look for loose data.

    • by arivanov ( 12034 )

      Yep. As long as you can still print all env values it is not dead yet.

      I made the mistake of not printing the values of a cheap and cheerful Chinese tablet before trying to smack a abrasive build on it. BAD mistake. Looks like it had a non-standard framebuffer offset so I will be guessing it till doomsday now...

    • Isn't the fact that you can "unbrick" it mean it's not really bricked?

      Eh... I've complained about this before, but for me the big difference is whether or not an easy fix was available at the time it happened. I mean if they went out of their way to create a tool to fix it later, then I don't personally have an issue with calling that 'de-bricking'.

      Then again, I don't like getting fussy about the term 'brick' until people use it to mean 'it froze'.

    • by mirix ( 1649853 )

      Nothing with JTAG or whatnot is really brickable at all, it's just a brick to folks that don't have the proper tools.

      Anyway, seems like a non-story to me. The sheevaplug having built in USBJTAG was one of the selling features.

      • by mirix ( 1649853 )

        grrr. forgot slashdot drops angle brackets. The sheevaplug has a JTAG interface that exists over USB. USB-"DOUBLE_ENDED_ARROW"-JTAG

  • This is just an OS reinstall. It's not like going in through the JTAG port and loading the firmware.

    These devices exist in the space between hard embedded systems and OSs with a user interface. With hard embedded systems, you do all development on another machine and download an entire image. With user-oriented OSs, you can interact with the machine in some reasonable way. These things live in a limbo between those two points - smart enough not to be total slaves, but not smart enough for standalone dev

    • Some of these plugs are powerful enough(or, more importantly, have enough RAM) that the only thing that really marks them as "embedded" is the fact that they have no video out and you have to deal with mtd devices rather than block devices...

      Compiling on a 1.2GHz ARM with 128-512MB of RAM(depending on variant), when every cheap Wintel/Lintel is some 2+GHz dual core beast with 3GB of RAM is, certainly, somewhat masochistic; but the system can run perfectly normal debian ARM, compile natively, and either r
  • Seems exploding power supplies are common and after opening mine up it had indeed exploded too. Ordered a replacement PS for 9.95, just waiting for it to arrive.
    • by akc ( 207721 )

      Mine did too - but it did quite a lot of damage to the electronics when doing so, so a replacement power supply did not revive it.

  • Anybody know if any of these little beasties come with two NIC ports?

  • by tunapez ( 1161697 ) on Thursday February 10, 2011 @04:34PM (#35166022)
    Have they fixed the power supply over-heating/failure issues or are we still talking about the same warts that have a shelf-life of 12 months or less?

    I'm serious, I would like an answer b/c I have been wanting one since the first of many, many, many slashvertisements appeared here two, three years ago. Problem is, everything I read on the outside says they fail prematurely and warranty fulfillment is spotty, at best. Am I wrong, Dude?
    • If it is the same device I once read about, yes, they fixed the over-heating problem.... if you like to have a siren where you plugged in the thing. AFAIK the installed a very loud fan.

    • by plover ( 150551 ) *

      I read on their site that they recommended running one of the network ports at 10/100 in order to manage the heat until they had a "solution". I was not comforted.

  • i realise locked down bootloaders are about control, but a brave hacker-friendly vendor would say 'here are the keys, no support offered but if you happen to brick your phone, here's how to factory restore'.

  • Remarkable timing for this story, as my Sheevaplug suddenly died last week after just over 330 days of flawless uptime. Turns out, there is a known issue with the original power supplies that were shipped, and they die pretty frequently after 8-10 months of use. This is so common that Globalscale now sells the PSUs separately [globalscal...logies.com] for ~$10 USD (plus another $15 for shipping, of course).

Beware of Programmers who carry screwdrivers. -- Leonard Brandwein

Working...