Featured Articles
Announcing Pi
Technical Specifications
Raspberry Pi FAQs
Win a Pi
Raspberry Pi Wishlist
Actions
Forum Thread Details
  • Replies 6 replies
  • Subscribers 677 subscribers
  • Views 1536 views
  • Users 0 members are here
Related

Is my Pi toast?  It hangs (sort of) during boot.

Former Member

I have an issue with the first Pi I bought (I have 2 - a 256M version and the 512M version - both are model B).

 

It starts to boot, I get the color splash, then see console messages scrolling by.  About the time it is giving me messages about the ethernet, it seems to hang.  I do see USB events if I connect then disconnect a keyboard and/or mouse (through a powered hub).  The lines I can currently see on the screen are (typos here are mine, as I am typing these in.  Leading space in the timestamps is omitted.):

 

[2.873812] mount (476) used greatest stack depth: 5684 bytes left

[2.896007] vchiq (467) used greatest stack depth: 5580 bytes left

 

Greetings from standalone Raspberry Pi

 

 

Processing /etc/profile...Done

 

# [3.007599] usb 1-1: New USB device found, idVendor=0424, idProduct=9512

[3.018287] usb 1-1: New USB device strings: Mfr=0, Product=0, SerialNumber=0

[3.034940] hub 1-1:1.0: USB hub found

[3.041008] hub 1-1:1.0: 3 ports detected

[3.327492] usb 1-1.1: new high-speed USB device number 3 using dwc_otg

[3.448043] usb 1-1.1: New USB device found, idVendor=0424, isProduct=ec00

[3.458866] usb 1-1.1: New USB device strings: Mfr=0, Product=0, SerialNumber=0

[3.476024] smsc95xx v1.0.4

[3.549453] smsc95xx 1-1.1:1.0: eth0: register 'smsc95xx' at usb-bcm2708_usb-1.1, smsc95xx USB 2.0 Ethernet, b8:27:eb:xx:xx:xx (I left out the last part of the MAC address on purpose

 

Aside from the timestamps, I get the same thing every time.  This is happening with a known good, working, SD card (I can put it in the other Pi and it boots just fine).  I can reimage the card, same behavior.  Move the card to the other Pi and it boots successfully.

 

This is with only power, the SD card, and a monitor connected (via HDMI -> DVI cable).  I can use the same power and cabling on the other Pi with no problems.

 

To me, this rules out the SD card, the power supply, and peripherals.  I tried with various revisions of the image.  Currently, trying with 2013-05-25-wheezy-rasbian.  I have also tried the 2012-12-16, 2012-09-18, 2012-08-16, and the 2012-08-08-wheezy-armel images.  All fail in the same manner.

 

Once it hangs, if I have a keyboard attached anything I type shows up.  If I just hit enter, I get a hash prompt (#).  It appears that I am in a shell.  I went looking for logs (/var/log/), however I find none.  It seems to have booted most of the way, it just can't quite finish.

 

Any suggestions on how to ger this back up and running?  As I recall, I went to bring it up one time and it behaved like this.  It was a new image at the time.  Seraching, I find lots of almost-but-not-quite-close-enough issues others have reported.  I found referenced to a bad SD card (same one works in another Pi), bad images (see previous statement), bad power (power works with the newer Pi but not the old one, and it is the supply I bought from Adafruit when I got a starter kit).

  • Thank you for all the details.  I assume you have looked at similar issues at the RasPi Troubleshooting Wiki: http://elinux.org/R-Pi_Troubleshooting#Kernel_Panic_on_boot

     

    If you have a multimeter or voltmeter, it's easy to check for low voltage: http://elinux.org/R-Pi_Troubleshooting#Troubleshooting_power_problems

     

    Low voltage is a source for lots of problems, and as the boot process continues more of the Broadcom SoC switches on and takes more current.  Since your power supply and cable work fine in your other RasPi, a possible culprit could be polyfuse F3.  It acts as a resistor, and if its resistance is too high it can cause excessive voltage drop as the SoC requires more current.  The RasPi Troubleshooting Wiki talks about F3 in the the above link.

     

    Your 256MB RasPi has the old PC board layout which can cause some LAN9512 chips to get hot.  This can also lead to problems, but it's unlikely to affect boot.

     

    Hope this helps.

  • in reply to johnbeetem

    Thanks for the response.  I tried to include the details and information *I* would like to see up front when troubleshooting a problem for someone else.

     

    I did look at the wiki, but discounted the power.  It is a simple test and will not hurt to check.  I will grab my meter from the shop and check it tomorrow and report back if power is indeed the problem.  If it is the polyfuse, I wonder if I can sample one at work (my company makes those, among other things).

     

    The weird thing is that it is not a total hang.  I can still navigate the filesystem, and some commands work.  It is like it booted to a mini-shell, if there is such a thing.  Sort of like the recovery console on Windows (only this seems to be actually a bit more useful image).

  • in reply to Former Member

    coder27 wrote:

     

    Try adding avoid_safe_mode=1 to your config.txt

    Oh dear... just searched the other forum for that option since I'd not come across it before..   So something introduced almost a year ago here http://www.raspberrypi.org/phpBB3/viewtopic.php?p=129413 still seems to have problems today if http://www.raspberrypi.org/phpBB3/viewtopic.php?t=42134&p=367001 is anything to go by.  

     

    The sensible choice would have been to revert this 'feature' quickly when issues first appeared, and to redesign it to not rely on a pin that's changed function and a user-programmable piece of data that predictably won't be done correctly on every board. Especially when the board rev could be changed by the user, whether it's deliberate or accidental is a moot point if you end up with a board that doesn't boot properly.

    Interestingly, the issues started to appear before the 'feature' was announced : http://www.raspberrypi.org/phpBB3/viewtopic.php?f=28&t=11625

     

    At the very least, the distro should catch that it's in safe mode, tell the user why his board is broken, give a link to the appropriate posts on forum (or tell him what to google for)..

     

    Oh well, I'm off to ensure that kernel_emergency.img doesn't exist on any of my cards and to recompile all of my kernels so that they ignore any commandline passed to the kernel.

  • in reply to Former Member

    That seems to have done the trick!

     

    I will have to read through the posts linked in your post and in selsinork's post to see if I can more thoroughly understand the situation.  From my quick read, it looks like there is a minimal kernel that it is booting into instead of the regular kernel.  Makes sense that I was seeing a limited set of commands available.

  • in reply to Former Member

    If you do read through some of those threads you'll find out how to mount the partition with your data on it, but that's only really useful so that you can go and add avoid_safe_mode=1 to config.txt and then reboot.

     

    You can also remove /boot/kernel_emergency.img and it should boot to the normal system regardless, but it seems that it will replace anything in cmdlines.txt with a default version, so you could lose any settings you have in there.

     

    There's a couple of things that seem to be possibilities for the cause of your problem:

    • You have something connected to the GPIO header that's pulling the 'safe mode' pin low during boot.
    • The pullup resistor on that pin has been damaged/removed somehow.
    • The gpio has been damaged within the SOC, (wrong voltage applied externally, too high current etc)
    • The board revision on your problem board is programmed incorrectly

     

    One of the posts by 'dom' shows how to change the board revision, but be aware that there will be different values depending on exacty which version you have.