Interesting blog

[danfoshizzle]

http://wuffcode.wolfpuppy.org.uk/

The guy on the blog Torne Wuff, mainly posts relating to the joggler

[hawsey]

And I thought all the joggler work was going on in this forum Dan.
I had a look but as usual it is not in my range of knowledge lol.
I see he is using refit .

[danfoshizzle]

Not my range either, I talked to him on google plus, he may pop in one day I hope, another ultra smart linux guy!

[torne]

I am here already, I just haven't posted much. There's not a lot of end-user useful stuff to report yet; the blog is for people who like reading about stuff being developed.

I'm using rEFIt temporarily because it gives me easy access to the EFI shell and choosing what bootloader to start, but it's rather suboptimal on the joggler (or in general, really) and the ultimate goal is to have debian-installer able to just install grub2 and add it to the EFI boot manager, so that the joggler starts directly into debian.

[cbredfred]

Hello Torne,

With your knowledge of how the Joggler boots, do you have any thoughts about the large number of Jogglers that lose the ability to boot from the internal 1GB flash memory?
As far as I know every test that people have run on the memory hasn't revealed any errors. EFI chips have been swapped in from working Jogglers to no avail.
There is a theory that gets bandied around the forums that the 1GB flash is very fragile, although I haven't heard of anyone that has proved that their Jogglers flash memory is broken other than Mevi, and I'm not sure what tests he ran. Anyway, I for one would appreciate your thoughts on this.

Clive

[torne]

Is there some description somewhere in more detail?

I assume the devices still boot from USB: can you tell whether they are booting via startup.nsh or boot.nsh? Normally the joggler will not read a startup.nsh from a USB stick (as it searches the eMMC first), so knowing which script it's executing would give you a clue as to which point the boot process is not able to read the eMMC.

Can you boot stuff on fs0: manually from the efi shell, if you start one from USB?

Edit: If someone who actually has a joggler with this problem wants to chat and see if we can work out what's going on, feel free to hit me up on IRC. I'm always on Freenode as Torne even if I'm not currently in #joggler.

[cbredfred]

Here's the main threads where it has been discussed:

Reflashing thread:
http://www.jogglerwiki.com/forum/viewto ... 0&start=60

Job lots:
http://www.jogglerwiki.com/forum/viewto ... ?f=2&t=365

Dead?? thread:
http://www.jogglerwiki.com/forum/viewto ... 6280#p6280

Yes they still boot from USB.
As far as the rest of your questions...I'm not sure. If they haven't been answered by the above threads, then can you lead me through the testing to find out?

[acesabe]

I am here already, I just haven't posted much. There's not a lot of end-user useful stuff to report yet; the blog is for people who like reading about stuff being developed.

I'm using rEFIt temporarily because it gives me easy access to the EFI shell and choosing what bootloader to start, but it's rather suboptimal on the joggler (or in general, really) and the ultimate goal is to have debian-installer able to just install grub2 and add it to the EFI boot manager, so that the joggler starts directly into debian.

I'd be interested to find out more about extending the 'extensible firmware interface' boot manager to utilize it's pxe boot capabillity so I can attempt to boot the Joggler off an LTSP server..not got too far yet due to the vagiaries of EFI, so if you can point me to any useful (read: usable!) info on getting more out of it, that'd be great! Or just a simple rEFIt howto would suffice!

[torne]

Here's the main threads where it has been discussed:

Reflashing thread:
http://www.jogglerwiki.com/forum/viewto ... 0&start=60

Job lots:
http://www.jogglerwiki.com/forum/viewto ... ?f=2&t=365

Dead?? thread:
http://www.jogglerwiki.com/forum/viewto ... 6280#p6280

Yes they still boot from USB.
As far as the rest of your questions...I'm not sure. If they haven't been answered by the above threads, then can you lead me through the testing to find out?

There's not a lot of immediately obvious information in those threads, and I have limited patience to trawl through

To test what stage in the boot it's getting to, you can try the following:

1) Make a USB stick with *something* on it that can boot (any old linux), with some bootloader efi binary (grub).
2) Try creating *just* a boot.nsh script which invokes the bootloader. Don't put a startup.nsh on the stick. Does that boot? Try both fs0: and fs1: in the script, check which boots. If boot.nsh doesn't work, try calling it boot2.nsh instead.
3) Try *just* a startup.nsh script which invokes the bootloader (directly, not by running boot.nsh). Don't put a boot.nsh on the stick. Does *that* boot? Try both fs0: and fs1: in the script, check which boots.
4) Try renaming the bootloader to boot.efi or boot2.efi and just putting it on the stick with no script file at all. Does *that* boot?

That's a lot of combinations, but the main thing is "can it start from something called boot". If it can, then the eMMC must be being accessed, because nothing other than the startup.nsh file on the eMMC can tell it to do that in the absence of a startup.nsh on the USB device.

Then, additionally to that, boot an EFI shell and see what files actually exist in fs0: and what they contain. *Is* there a startup.nsh? A boot.nsh? a boot.efi? what do they contain?

[torne]

I'd be interested to find out more about extending the 'extensible firmware interface' boot manager to utilize it's pxe boot capabillity so I can attempt to boot the Joggler off an LTSP server..not got too far yet due to the vagiaries of EFI, so if you can point me to any useful (read: usable!) info on getting more out of it, that'd be great! Or just a simple rEFIt howto would suffice!

EFI doesn't automatically imply the ability to boot via PXE. There would have to actually be an EFI network driver for the wired interface in the image, as well as the EFI networking/TFTP modules. Is there any evidence that this is the case on the Joggler? I haven't looked, but my guess would be no

EFI is just a platform; one EFI machine having the ability to boot via PXE, or even the existence of a standard mechanism for doing PXE on EFI, doesn't mean that any EFI machine can do it.

[Mevi]

Good to have you on the forum, Torne.

Re: my Joggler with the dead memory.......

Without a USB stick it just sits on the Openpeak logo. It doesn't boot into the factory OS.

It won't boot Linux, strangely. I don't know if grub needs to see the internal partitions t correctly identify the USB stick partitions. I should probably look in the grub config file to see if that's the case. It just sits on the logo screen again.

It does boot XP, however. This is where I identified the faulty internal flash as faulty as it showed up in device manager with an exclamation mark and an "error 10". I'd not seen this before on my other 4 Jogglers. This Joggler is my XP BitTorrent box, anyway, so not a biggie.

[torne]

It won't boot Linux, strangely. I don't know if grub needs to see the internal partitions t correctly identify the USB stick partitions. I should probably look in the grub config file to see if that's the case. It just sits on the logo screen again.

It does boot XP, however. This is where I identified the faulty internal flash as faulty as it showed up in device manager with an exclamation mark and an "error 10". I'd not seen this before on my other 4 Jogglers. This Joggler is my XP BitTorrent box, anyway, so not a biggie.

grub is probably attempting to probe all of the storage devices EFI tells it about, which presumably includes the nonresponsive eMMC. If it also doesn't appear to respond at all in a booted OS then I'd assume the eMMC has failed; some MMC/SD cards have really dumb flash translation layers which can get into a state where not only can you not write to them (as happens normally if the card has far too many bad blocks to do anything) but they don't even respond to IDENTIFY queries, let alone reads, because the card firmware has basically crashed.

The problem the previous poster was talking about implied that having booted from USB the eMMC was accessible and usable, which is more interesting