Jonn
Posts: 35
Joined: Wed Aug 01, 2012 12:15 pm

Request tool/access to SD post mortem 'out-of-body'

Sun Aug 05, 2012 12:59 am

The minimum requirements for the RaspberryPi to boot are the SD card and power: the RaspberryPi may boot and work properly even if the other peripherals do not.

Problem:
My first session went smoothly, with keyboard and mouse on a hub and GUI via RCA composite to a TV, and with GUI via CygwinX on a PC through the network. However, it seems now that the keyboard stops working a few minutes after boot (even at typing the password or trying to 'startx'), and ssh times out. (I hope I did not break my Pi.)

Some forum posts recommend dmesg and similar commands for diagnosis, but when input is unavailable the only apparent option that remains is to photograph messages on the display. Some messages scroll off the screen very quickly...

Suggestion:
An option to copy (some or all) boot messages into a file on the SD card might be useful for diagnosing problems with the RaspberryPi or with its peripherals. I imagine one could set a flag in config.txt to enable this, but the default should be disabled, of course, to minimize writes to the SD card. (I do not see such an option in config.txt for 2012-07-15-wheezy-raspbian.img or at http://elinux.org/RPi_config.txt .)

I don't know how to control when to stop writing data to the file--there is no user control without input, and no control at all if the RaspberryPi freezes. Alternating output between two files and alternately closing them upon successful writes might provide an approximate solution for a freeze where data in one file might be lost.

Perhaps when the option is enabled, the boot process would also disable the same option in config.txt so that only one session's output is recorded at a time. Immediate analysis would be appropriate, and the user could then manually edit config.txt again to reactivate the option, if desired. (Time and date information would not distinguish the information from one session to the next, but sequential numberings of files as used in cameras could, pairing could be odd-even or with suffixes, e.g. btmsg001.txt & btmsg002.txt or btmsg001-1.txt, btmsg001-2.txt.

Is something like this already part of the Raspian kernel_emergency.img?
(Is it a good idea someone else already has had?)
Would this be useful for enough people to make it feasible?

Thanks. ; )

W. H. Heydt
Posts: 9533
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Request tool/access to SD post mortem 'out-of-body'

Sun Aug 05, 2012 9:49 pm

Assuming you have some other system (Linux, Mac, Windows...) you should be able to get into the Pi using an ssh connection. PuTTY is one of the easiest programs to do this with.

Once connected, you can look at messages. You can also use the "lsusb" and "usb-devices" commands to gather more data.

Jonn
Posts: 35
Joined: Wed Aug 01, 2012 12:15 pm

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:06 am

There's the rub--when what should work does not or has stopped working, such a tool or procedure might help with diagnosis.

alexchamberlain
Posts: 121
Joined: Thu Jun 14, 2012 11:20 am
Location: Leamington Spa, UK
Contact: Website

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 12:52 pm

Linux logs everything! Once your Raspberry Pi dies, you can plug the SD card into your Linux desktop, mount it and diagnose any problems.
Developer of piimg, a utility for working with RPi images.

User avatar
Jim JKla
Posts: 2219
Joined: Sun Jan 29, 2012 11:15 pm
Location: Newcastle upon Tyne UK

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 12:59 pm

So this should be worth a wiki page at least givingour Noob readers a clue on how to do this.

Considering we are starting from scratch and a lot of users are coming from the Windows/Microsoft/Apple school of no help at all we have to be able to do better.

I have started writing noob support for the wiki mainly in support of Wheezy and it has been getting some hits. We do need more of this support stuff in the wiki.
Noob is not derogatory the noob is just the lower end of the noob--geek spectrum being a noob is just your first step towards being an uber-geek ;)

If you find a solution please post it in the wiki the forum dies too quick

alexchamberlain
Posts: 121
Joined: Thu Jun 14, 2012 11:20 am
Location: Leamington Spa, UK
Contact: Website

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:07 pm

You can use `piimg` to mount an SD card properly - with /dev, /proc, /sys and /boot. Then you can chroot into sort out any problems if necessary.

I've written a bit a related guide over at Raspberry Pi.SE. It explains how to install QEMU etc.

Obviously, this is not necessary if you just want the log files, but I think it's the easiest way to install new software.

The only problem is Windows users! I think I might have to publish a Raspberry Pi rescue CD...
Developer of piimg, a utility for working with RPi images.

User avatar
Jim JKla
Posts: 2219
Joined: Sun Jan 29, 2012 11:15 pm
Location: Newcastle upon Tyne UK

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:12 pm

alexchamberlain wrote:You can use `piimg` to mount an SD card properly - with /dev, /proc, /sys and /boot. Then you can chroot into sort out any problems if necessary.

I've written a bit a related guide over at Raspberry Pi.SE. It explains how to install QEMU etc.

Obviously, this is not necessary if you just want the log files, but I think it's the easiest way to install new software.

The only problem is Windows users! I think I might have to publish a Raspberry Pi rescue CD...
Now how about a rewrite that the noob's can understand posted in the wiki the first line is pure uber-geek gibberish :D
Noob is not derogatory the noob is just the lower end of the noob--geek spectrum being a noob is just your first step towards being an uber-geek ;)

If you find a solution please post it in the wiki the forum dies too quick

PaulCheffus
Posts: 381
Joined: Tue May 29, 2012 2:51 pm
Location: Lincolnshire UK

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:15 pm

alexchamberlain wrote:Linux logs everything! Once your Raspberry Pi dies, you can plug the SD card into your Linux desktop, mount it and diagnose any problems.
Hi

Thats ok if you have a Linux desktop but what happens if you only have Windows available?

Cheers

Paul
Procrastination - The Thief of Time.

alexchamberlain
Posts: 121
Joined: Thu Jun 14, 2012 11:20 am
Location: Leamington Spa, UK
Contact: Website

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:18 pm

PaulCheffus wrote:Thats ok if you have a Linux desktop but what happens if you only have Windows available?
Would you prefer a Linux Live CD (designed for this situation) or a piece of Windows software that could read the Linux partitions?
Developer of piimg, a utility for working with RPi images.

PaulCheffus
Posts: 381
Joined: Tue May 29, 2012 2:51 pm
Location: Lincolnshire UK

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:39 pm

alexchamberlain wrote:
PaulCheffus wrote:Thats ok if you have a Linux desktop but what happens if you only have Windows available?
Would you prefer a Linux Live CD (designed for this situation) or a piece of Windows software that could read the Linux partitions?
Hi

Good question. Personally I don't have a preference.

Cheers

Paul
Procrastination - The Thief of Time.

User avatar
Jim JKla
Posts: 2219
Joined: Sun Jan 29, 2012 11:15 pm
Location: Newcastle upon Tyne UK

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:40 pm

Ask yourself would a true noob know the difference? ;)
Noob is not derogatory the noob is just the lower end of the noob--geek spectrum being a noob is just your first step towards being an uber-geek ;)

If you find a solution please post it in the wiki the forum dies too quick

alexchamberlain
Posts: 121
Joined: Thu Jun 14, 2012 11:20 am
Location: Leamington Spa, UK
Contact: Website

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:45 pm

Ok, then... What would you like to see in a rescue CD? The idea would be... You insert the CD into your x86 Windows desktop/laptop with the SD card in a reader attached. You can then use the Linux commands to manage the SD card.

Any thoughts?

My main concern is people destroying their hard drive...
Developer of piimg, a utility for working with RPi images.

User avatar
Jim JKla
Posts: 2219
Joined: Sun Jan 29, 2012 11:15 pm
Location: Newcastle upon Tyne UK

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:54 pm

No problem with your concerns.

I am just a moderate geek trying to hold a torch for the noobs to find their way.

But that idea sounds good. ;)
Noob is not derogatory the noob is just the lower end of the noob--geek spectrum being a noob is just your first step towards being an uber-geek ;)

If you find a solution please post it in the wiki the forum dies too quick

PaulCheffus
Posts: 381
Joined: Tue May 29, 2012 2:51 pm
Location: Lincolnshire UK

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 1:58 pm

alexchamberlain wrote: My main concern is people destroying their hard drive...
Hi

Having done this to a USB drive (luckily I had a backup) it would be one of my concerns too.

Cheers

Paul
Procrastination - The Thief of Time.

alexchamberlain
Posts: 121
Joined: Thu Jun 14, 2012 11:20 am
Location: Leamington Spa, UK
Contact: Website

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 2:00 pm

Would you want a GUI?
Developer of piimg, a utility for working with RPi images.

PaulCheffus
Posts: 381
Joined: Tue May 29, 2012 2:51 pm
Location: Lincolnshire UK

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 2:01 pm

alexchamberlain wrote:Would you want a GUI?
Hi

Again not too worried but if this is aimed at someone completely new who has only ever used Windows then it might be worthwhile.

Cheers

Paul
Procrastination - The Thief of Time.

Jonn
Posts: 35
Joined: Wed Aug 01, 2012 12:15 pm

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 2:52 pm

A utility for Windows would be fabulous: Windows users (such as I am) would be able to get the information without having to close applications for rebooting the PC. This would hold also for classrooms where Windows is the default OS. Read-only access should be sufficient (and safer) for diagnosis: if included, write access should be 'advanced'. Of course, a user could simply lock the SD card as a precaution.

I have a 'lightly-used' Linux live CD that I can boot on a second, older PC, so I'll see if I can download and set up piimg on that. (Probably I should save the installed-version piimg files to a USB stick to save having to re-install each time I start up fresh with the Linux live CD.)

I do have a question: I saved the SD card files (.img, .txt, etc.) to Windows directories a couple times (from re-imaging of the SD card different times), but only the config.txt files have differing timestamps because of raspi-config and/or manual editing. The SD card is writable, because I have reformatted it and re-imaged it ... in Windows. It did show empty of files and of full capacity, and then at 59 MB after re-imaging. File comparisons between two previous incarnations and the current version of kernel.img show the contents to be the same. Are the logs stored in the RAM and not written to the SD card (yet)?

alexchamberlain
Posts: 121
Joined: Thu Jun 14, 2012 11:20 am
Location: Leamington Spa, UK
Contact: Website

Re: Request tool/access to SD post mortem 'out-of-body'

Mon Aug 06, 2012 3:08 pm

Jonn wrote:A utility for Windows would be fabulous: Windows users (such as I am) would be able to get the information without having to close applications for rebooting the PC. This would hold also for classrooms where Windows is the default OS. Read-only access should be sufficient (and safer) for diagnosis: if included, write access should be 'advanced'. Of course, a user could simply lock the SD card as a precaution.
There are a couple of applications out there that claim to read ext partitions on Windows, but they aren't actively maintained from what I can tell. My fear would be that Windows upgrades would mean significant effort to make it work again. All this for a solution that limits it to reading.
Jonn wrote:I have a 'lightly-used' Linux live CD that I can boot on a second, older PC, so I'll see if I can download and set up piimg on that. (Probably I should save the installed-version piimg files to a USB stick to save having to re-install each time I start up fresh with the Linux live CD.)
This is the possibility I would vote for, except slightly more permanent. A specially designed Live CD for managing SD card images. I've just pushed to https://github.com/alexchamberlain/rpi_manage_live. Please raise an Issue to suggest any more features it should have.
Jonn wrote:I do have a question: I saved the SD card files (.img, .txt, etc.) to Windows directories a couple times (from re-imaging of the SD card different times), but only the config.txt files have differing timestamps because of raspi-config and/or manual editing. The SD card is writable, because I have reformatted it and re-imaged it ... in Windows. It did show empty of files and of full capacity, and then at 59 MB after re-imaging. File comparisons between two previous incarnations and the current version of kernel.img show the contents to be the same. Are the logs stored in the RAM and not written to the SD card (yet)?
No idea.
Developer of piimg, a utility for working with RPi images.

Jonn
Posts: 35
Joined: Wed Aug 01, 2012 12:15 pm

Re: Request tool/access to SD post mortem 'out-of-body'

Wed Aug 08, 2012 4:50 pm

Thank you alexchamberlain! (And thank you to the others too!) So far, I've found the logs on the raspian SD card's second partition, which is invisible to Windows XP. (One can search the forum and find posts that say it is invisible--if one realizes the need to do so and knows what search text to use, e.g. 'ext4'.) I still need to learn which log files to check and to find which messages point to problems that where my RaspberryPi stops responding to keyboard input, but that would be for the Troubleshooting section of the forum... I'm making progress. :)
...and I still have to try out piimg.


Following is how to get to those logs (with parts gleaned from various posts and other sources).

I downloaded the zip file from https://github.com/alexchamberlain/rpi_manage_live, unzipped it to a USB stick, and found the releng/root-image/root/install.txt file. It references https://wiki.archlinux.org/index.php/Installation_Guide for making an Arch Linux CD.

In Windows XP, I downloaded the Arch Linux ISO, confirmed the md5sum (with a utility for Windows) and wrote the .iso image to a CD-R using Infrarecorder (http://infrarecorder.org, v. 0.44.1.0) and at the speed rating of the CD-R.

I attached the USB reader and SD card and the USB stick to a second PC and booted it to the Arch Linux Live CD choosing 'Boot Arch Linux (i686)' from the top of the menu. It booted up and automatically logged in as root.

(It took a bit of looking things up and trial and error, but here is the summary of commands.) To log in from a Windows PC over the network, prepare on Arch Linux with these commands:
  • passwd
    ifconfig
    /etc/rc.d/sshd restart
to set the root password, get the IP address, and to generate keys for the 'new' PC where the Live CD was booted. Then make a user account (example account 'jonn'.) with
  • adduser jonn
and follow the prompts to create the user account.

I didn't know how to start a graphical interface in Arch Linux, so from the Windows PC, I connected with ssh (a Windows version) to the Arch Linux PC by IP address. This way I could easily copy and paste information into a file for documentation.

Once logged in as the user, enter
  • su
to become root.

The next step is to logically mount partitions from the SD card and USB stick. At the linux command line (on either the Arch Linux console or the other PC),
  • cd /dev
    ls sd*
This (Arch Linux) PC has sda, sdb, sdc, sdd, sde, and more followed by numbers starting from 1. My SD card with the 2012-07-15-wheezy-raspbian.img image has sdd1 and sdd2. The sdd1 partition is the one visible to Windows XP, and the sdd2 is the ext4 partition that is invisible to Windows. (A couple ways how to tell which sd* entry is the SD card: mount and check entries, or to boot without the USB card reader, check the list in /dev, and check again with the card attached to see what sd* entries have been added.)
  • cd /mnt
    ls
Make a new directory name that does not already appear in the /mnt directory, (which starts out empty anyway), and mount the device partition to the new directory (e.g. SD2 here):
  • mkdir /mnt/SD2
    mount /dev/sdd2 /mnt/SD2
This attaches the root of the RaspberryPi raspian file system to /mnt/SD2.
  • cd /mnt/SD2/var/log
There are several log files there. Several forum posts refer to dmesg, and there is a file with that name. Since the log files have timestamps starting each line, it looks like dmesg is the place to start and one could cross-reference into the other files by the timestamps.

The ssh (for Windows) program I used allowed me to see the directories and files, so I could drag and drop copies of the log files into my Windows PC, but I could also use scp in Cygwin (http://cygwin.org/):

Code: Select all

[email protected] ~
$ scp -p [email protected][i]IP_Address[/i]:/mnt/SD2/var/log/dmesg .
The authenticity of host '[i]IP_Address[/i]([i]IP_Address[/i])' can't be established.
RSA key fingerprint is 03:[i][edited][/i]:69.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added '[i]IP_Address[/i]' (RSA) to the list of known hosts.
[email protected][i]IP_Address[/i]'s password: 
dmesg                                         100%   19KB  18.9KB/s   00:00    

[email protected] ~
$ 

alexchamberlain
Posts: 121
Joined: Thu Jun 14, 2012 11:20 am
Location: Leamington Spa, UK
Contact: Website

Re: Request tool/access to SD post mortem 'out-of-body'

Wed Aug 08, 2012 5:30 pm

Jonn wrote: I downloaded the zip file from https://github.com/alexchamberlain/rpi_manage_live, unzipped it to a USB stick, and found the releng/root-image/root/install.txt file. It references https://wiki.archlinux.org/index.php/Installation_Guide for making an Arch Linux CD.
This is totally not ready, but at least you burned a raw Arch Linux disk and got stuff sorted. Be careful not to bork your own system with Live CDs though!
Developer of piimg, a utility for working with RPi images.

Jonn
Posts: 35
Joined: Wed Aug 01, 2012 12:15 pm

Re: Request tool/access to SD post mortem 'out-of-body'

Wed Aug 08, 2012 6:16 pm

I forgot to include the finish for the session after getting to the logs: For the remote access through ssh, use exit from su mode and logout. Then at the linux console use
  • shutdown -h now
(That way one does not have to manually enter commands to unmount the added devices: the Live CD will start up without them as it did before anyway.)
Lastly, when the PC is off, detach the USB stick and reader.
Be careful not to bork your own system with Live CDs though!
Happily I knew little enough that I had to proceed slowly. I think copying the files to view on another PC helped keep me from trouble too. I could have goofed up with an editor, but I didn't know or check for what editor is available in Arch Linux.

Jonn
Posts: 35
Joined: Wed Aug 01, 2012 12:15 pm

Re: Request tool/access to SD post mortem 'out-of-body'

Tue Aug 28, 2012 6:22 am

Well... now I can check for clues in the log files for keyboard or network problems. Since there were two good sessions out of dozens of times booting suggests a hardware problem--It would be nice to be able to point to something in the logs that would be more definitive.

After the RaspberryPi was off about a week, I had a second good session with working keyboard, mouse and network. I kept it on for a day or more. On subsequent boots it stopped and stayed at the login prompt with no keyboard or network response, so I let it rest another two or three days. It didn't improve the response. (Somewhere in the forum I read that the polyfuses should reset in hours, so I guess it isn't exclusively the polyfuses.) I put the new 2012-08-16-wheezy-raspbian.img onto a new SD card and booted with it. The result was that it still stopped, but on the raspi-config menu.

An optical mouse showed each USB port had power (the LED lit), and a TP1-TP2 measurement pushed the meter needle just over 5V. (Previous measurements with a borrowed digital meter showed around the range of 4.96 to 5.05V.)

During the good session, I got the lsusb list (so this keyboard [100mA] and mouse [60mA] 'work'):

[email protected] /var/log $ lsusb
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 001 Device 002: ID 0424:9512 Standard Microsystems Corp.
Bus 001 Device 003: ID 0424:ec00 Standard Microsystems Corp.
Bus 001 Device 004: ID 0430:0100 Sun Microsystems, Inc. 3-button Mouse
Bus 001 Device 005: ID 0430:0005 Sun Microsystems, Inc. Type 6 Keyboard


I copied the log files from the SD card (/var/log mounted as /mnt/SD2/var/log) to my PC where I could check them more easily (in Windows). I used different directories for sets of logs so I could see what they look like before and after a series of boot attempts.

If the problem is a non-oscillating 25 MHz crystal (like in the discussions http://www.raspberrypi.org/phpBB3/viewt ... =28&t=5766 and
http://elinux.org/R-Pi_Troubleshooting# ... card.2C_KB), what messages would show up (or be missing) in the log files?

NOHz appeared in dmesg several times, but it seemes to be related to the CPU:

usbcore: registered new device driver usb
Switching to clocksource stc
Switched to NOHz mode on CPU #0


NOHz shows up in files kern.log, messages, and syslog too.


I found that 'USB device found, idVendor=' shows up in logs for each device that shows up in the lsusb command. It is easy to get a count of how many times that string appears in a log simply by using grep or by searching for it using the Chrome browser, which also shows where it occurs in the file. So, I collected some info. The particular USB device message strings specifying a unique idVendor and idProduct pair , e.g. 'New USB device found, idVendor=1d6b' or 'New USB device found, idVendor=0424, idProduct=9512', appear the same number of times in kern.log as in the combined syslog and syslog.1 (if present) files.

On my 2012-07-15-wheezy-raspbian SD card (third of fourth time imaged) I saw
66 New USB device found, idVendor=
which breaks down to:
7 New USB device found, idVendor=1d6b {Linux Foundation 2.0 root hub}
40 New USB device found, idVendor=0424
which is a combination of:
___11 New USB device found, idVendor=0424, idProduct=9512 {Standard Microsystems Corp.}
___29 New USB device found, idVendor=0424, idProduct=ec00 {Standard Microsystems Corp.}
6 New USB device found, idVendor=0430
which is a combination of:
___ 5 New USB device found, idVendor=0430, idProduct=0005 {keyboard}
___ 1 New USB device found, idVendor=0430, idProduct=0100 {mouse}
7 New USB device found, idVendor=05e3 {idProduct=0605 iRocks IR-4300 four port USB 2.0 generic hub}
1 New USB device found, idVendor=1267
5 New USB device found, idVendor=1241


After a couple more boot attempts, the main count increased from 66 to 81, but all that was added was
15 New USB device found, idVendor=1d6b {Linux Foundation 2.0 root hub}
and none of the other USB devices.

On Friday, I put 2012-08-16-wheezy-raspbian on a new (second) SD card. The first (Friday) session's logs show:
9 New USB device found, idVendor=
9 New USB device found, idVendor=1d6b


There was no logging of the other USB devices. For the short second session there were 6 more entries(Sunday, 15 total) and no other devices were logged.

'Conclusion':
Presumably, the logs should show a one-to-one correspondence of the built-in devices (1d6b:0002, 0424:9512, and 0424:ec00) whether or not keyboard, mouse, or network cable is attached. The first logs show counts of (7,11,29) respectively, but the last logs show an additional (15,0,0).

I think my next step is to contact RS to ask about a return.

Would a non-oscillating crystal let the root hub be found but not the devices connected directly to it, corresponding to the (15,0,0) counts of devices found?

Jonn
Posts: 35
Joined: Wed Aug 01, 2012 12:15 pm

Re: Request tool/access to SD post mortem 'out-of-body'

Fri Aug 31, 2012 1:54 am

I found this:
Polyfuses can take a very long time to completely recover from a "trip", sometimes days, even weeks, in fact in a sense they never get completely the same, but if they get down to say 3 Ohm (for the USB polyfuses F1 & F2) or 0.2 Ohm for the input polyfise F3 on the back of the board you can say they are "cured". http://www.raspberrypi.org/phpBB3/viewt ... 799#p94555
... so maybe I'll leave it off for another week and try again...

Jonn
Posts: 35
Joined: Wed Aug 01, 2012 12:15 pm

Re: Request tool/access to SD post mortem 'out-of-body'

Wed Sep 12, 2012 2:39 pm

Yesterday I booted the RaspberryPi again a few times with the Monoprice CYgma4 5V 2.1A charger I bought for it. I was able to log in with username and password the first time, but only that time.

Having found some YouTube videos on power supplies and problems, I decided finally 'bite the bullet' and cut through a perfectly good USB-microUSB cable... (I did get two cables when I bought the charger, so I have another.) My old disused soldering iron finally got the solder to join the black (ground) wires together and the red (5V) wires together of the half-cable and floppy drive lines of an old 250W computer power supply--after all, the RaspbnerryPi *is* a computer! It didn't solve the problems, but it does have a power switch.

I used the July Raspbian SD card, then the August SD card, then the July SD card again since it seemed to yield more information. (Maybe when the next Raspbian distribution becomes available I’ll overwrite the August SD card with it.) The message count is at the end.

During some of the boot attempts I pressed Caps Lock and Num Lock a few times. I did see the keyboard LEDs come on during an attempt, but it only seems to work occasionally. The RaspberryPi boot messages displayed on the (RCA) monitor, "Num Lock on, Caps Lock on" before the login prompt, but the keyboard LEDs had gone out by then and the keyboard had stopped working—the Lock keys would not make the keyboard LEDs light. (This was on the July SD card only.) I also saw the network LEDs light up momentarily on the RaspberryPi during one of the boot attempts, but my router log did not show an IP address for it.

I saw that in response to a comment about penguinspuzzle locking up, Dom said:
  • “Sounds like power supply to me. The current drops because the chip crashes after the voltage drops. As an experiment, add:
    arm_freq=500
    gpu_freq=150
    If that fixes it, it is almost certainly power supply.” http://www.raspberrypi.org/archives/1967#comments
I added those two lines to config.txt in the Windows-readable partition of the July SD card. I did not notice a difference with keyboard response. I wonder if that was when the network LEDs lit up, but if the config.txt change was a fix for my RaspberryPi, the LEDs should have lit on the few subsequent boots too, but that did not happen.

I don’t know if this is the ‘elephant’ problem http://www.raspberrypi.org/phpBB3/viewt ... t=elephant or whether I excited it—an excited elephant? ;) The very first power source I had used was a Kindle power adapter (I had seen the approved list), but after the long working session and after the problems occurred, I saw that the Kindle adapter I had been using was marked 4.9V. I noted that I also used a powered USB hub (that has an adapter marked 5V). When I started having problems and found the different voltages, I worried that I had broken the RaspberryPi, by making power run backward through the USB hub and polyfuses. I also worried that I had one adapter plugged into a surge strip or UPS that the other was not plugged into, but both were on the same household circuit. After realizing that mistake I tried to remember always to keep the power supplies plugged into the same power, e.g. both on the same strip or extension cord. The second long working session suggested that the faux pas was not too bad or that the RaspberryPi and its polyfuses were recovering (if that was the problem issue).

The polyfuses measure ~5 ohm for F1 and F2 and ~0.5 for F3 with an analog meter calibrated for zero ohms when the probes touch. These measurements don't look bad when comparing with some working (and problem) boards on http://elinux.org/RaspberryPi_Boards or comparing to some of the discussion at http://www.raspberrypi.org/phpBB3/viewt ... 799#p94555. The 3 ohm quote in a previous post does suggest a problem, but the working boards link has counterexamples with higher resistance comparable to 5 ohms. Before ever being tripped, should polyfuses F1 and F2 have much lower resistances like F3 instead, or is there a designed relationship between the resistance and the tripping current?

I previously matched my board to the list with the RS board listed to ‘antijn’ that reports ‘USB/Keyboard/LAN issues’ and it appeared to also match one listed to ‘Mortimer’, that reports issues ‘None’. My board details are:
  • Model B board version BH1218, (RS Invoice date 29 May 2012, shipped from UK to US)
    Sticker: E2112RSV1.0B1.1
    RAM chip Hynix: H9TKNNN2GDMP, LRNDM, 217A, \VTKC9325A2
    USB chip SMSC: LAN9512 –JZX, B1215 –A1B17, 8R150188B, ASE-TW, (E3)
    C6: 2E-220-16H
    F1/F2/F3: “14”, “14”, “07” ; ~5, ~5, ~0.5 (As of 2012 Sep 11).
    RG2 SE8117T33, 1213-LF
Now in kern.log for the 2012-07-15-wheezy-raspbian SD card, my chrome browser 'find' feature yields search counts of the text as follows, with the following format:
(number of additional occurrences is in parentheses.), current count - previous count, search string.

(36) 117-81 New USB device found
which breaks down to:
(9) 31-(15+7) New USB device found, idVendor=1d6b {Linux Foundation 2.0 root hub}
(15) 55-40 New USB device found, idVendor=0424
which is a combination of:
__ (6) 17-11 New USB device found, idVendor=0424, idProduct=9512 {Standard Microsystems Corp.}
__ (9) 38-29 New USB device found, idVendor=0424, idProduct=ec00 {Standard Microsystems Corp.}
(12) 18-6 New USB device found, idVendor=0430
which is a combination of
__ (6) 11-5 New USB device found, idVendor=0430, idProduct=0005 {keyboard}
__ (6) 7-1 New USB device found, idVendor=0430, idProduct=0100 {mouse}
(0) 7-7 New USB device found, idVendor=05e3
(0) 1-1 New USB device found, idVendor=1267
(0) 5-5 New USB device found, idVendor=1241


That the 1d6b(root) and ec00 devices are recognized and listed 9 times more and the 9512 only 6 times more suggests a problem with recognizing or listing the 9512 device.
I may have left the mouse off sometimes, but that isn't reflected in the mouse and keyboard list count... the 9512 and keyboard and mouse all appear 6 times more. I suppose the keyboard and mouse depend upon the 9512 device.

For the August Raspian SD card, with which the RaspberryPi lands in raspi-config and the keyboard didn’t respond. I only tried a few times and these are the new total counts in the file:

17 'New USB device found'
17 New USB device found, idVendor=1d6b
0 idVendor=0424
0 idVendor=0430
17 New USB device strings

Perhaps the other devices are not logged simply because raspi-config has not yet been finished successfully?

So... I guess I'll wait for the September Raspbian and let the board rest some more.

User avatar
Licaon_Kter
Posts: 240
Joined: Wed Sep 05, 2012 10:12 am
Location: Between the keyboard and the chair.

Re: Request tool/access to SD post mortem 'out-of-body'

Wed Sep 12, 2012 2:59 pm

Jonn wrote:So... I guess I'll wait for the September Raspbian and let the board rest some more.
You know that running

Code: Select all

sudo apt-get update
sudo apt-get upgrade
sudo reboot
will update your system and there is no need for them to make a new image every month or for you to wait for a month, right?
BFQ+BFS or RT on a RPi? 4'real: https://github.com/licaon-kter/ (source and compiled!)

Return to “General discussion”