Updated u-boot support?
Home › Forums › MultiConnect OCG › Updated u-boot support?
- This topic has 9 replies, 3 voices, and was last updated 10 years ago by Tim Scott.
-
AuthorPosts
-
September 10, 2014 at 10:36 am #6001Tim ScottParticipant
I’m running with the stock U-Boot that is installed on my MT100EOCG-SP: “U-Boot 1.3.4-mts (Nov 9 2011 – 08:45:23)”. This version is fairly old, and does not appear to include FAT FS operations. Ideally, I’d like to be able to load in kernel and filesystem images from a USB thumb drive. From what I can tell, this version of U-Boot does not support that. I believe that the “usbboot” command, which is included, only supports RAW partitions which is not exactly what I want.
Has there been any effort to patch a relatively recent version of U-Boot to work on the MT100EOCG-SP? I tried building vanilla U-Boot v2014.01 configured for the at91sam9g20ek. However, after loading in the resulting u-boot.bin file and trying to execute it, things just hang. I encounter the same thing with v2013.04.
Thanks,
–timSeptember 11, 2014 at 10:31 am #6003Mike FioreBlockedHi Tim,
We don’t have any work planned to update u-boot at the moment. You could try patching a newer u-boot; compare the board file in our u-boot recipe against the board file in the vanilla u-boot version you choose. I don’t think there should be too many changes/customizations.A couple things to keep in mind:
– make sure the nand configuration is correct (you should be building for at91sam9g20ek_nandflash_config)
– the vanilla u-boot may be using a different ethernet configuration (mii vs rmii), but that shouldn’t cause u-boot to hangIf you continue to have issues, pasting any console debug output will help us assist you in debugging the issue.
Cheers,
MikeOctober 3, 2014 at 2:24 pm #6039Jesse GillesBlockedHi Tim,
I was able to get u-boot 2012.10 for MT100EOCG/MTCDP working with the following patch:
http://www.multitech.net/downloads/u-boot-2012.10-mtcdp.patch
It was likely hanging due to the difference in CONFIG_SYS_TEXT_BASE. The older at91bootstrap on the EOCG is loading u-boot at a different address than the default in u-boot.
The rest of the patch is for boot defaults and ethernet functionality.
Hope that helps, the same changes may also apply to later versions.
The only issue I’ve noticed is that when I press the Tab key at the u-boot prompt, it hangs — I’m not sure why, but it works otherwise.
Jesse
November 13, 2014 at 11:26 am #6075Tim ScottParticipantHi Jesse,
I think I may be deploying my u-boot.bin file to the target incorrectly. I can’t load it at 0x23F00000 because that is where the existing u-boot resides, so it freezes if I try to do that. If I load things at 0x21400000 and follow that with a “go 0x2140000” I encounter the freeze as mentioned in my original post. I’m guessing this may be due to the CONFIG_SYS_TEXT_BASE setting.
How did you test your u-boot build?
Thanks,
–timNovember 13, 2014 at 2:54 pm #6078Tim ScottParticipantSweet. I recompiled U-boot v2012.10 with the patch, but changed then CONFIG_SYS_TEXT_BASE to 0x21400000 and was able to load / execute the .bin file just fine. I plan on investigating whether or not I can get the latest version of U-Boot running in a similar manner.
–tim
November 13, 2014 at 4:18 pm #6079Jesse GillesBlockedHow did you test your u-boot build?
I tested it by flashing it into the nand and then rebooting. I didn’t try executing it directly using ‘go’.
Jesse
November 13, 2014 at 4:23 pm #6080Jesse GillesBlockedYou’ll need to use the CONFIG_SYS_TEXT_BASE value in the patch once you are ready to flash it into the nand since that is where at91bootstrap is putting u-boot.
November 14, 2014 at 7:07 am #6081Tim ScottParticipantYep, that makes sense. For test purposes, I think loading in the .bin to RAM and running things that way will help prevent from bricking accidentally bricking the unit.
FWIW, I have the first pass of U-Boot v2014.10 working on the board. Ethernet doesn’t appear to function yet so I’ll be digging into that today.
–tim
November 14, 2014 at 9:33 am #6082Tim ScottParticipantDoes anyone know if it is possible to recover an incorrectly programmed flash using the MT100EOCG-DK? I found the link below, but it applies to a different board altogether:My board is working fine at the moment. I’d just like to know how to resolve a bricked unit in case something goes bad with my u-boot update.Found the answer in S000516C. Looks like I just need to short out R108 on the MT100 board and run the ATMEL software. Sweet!
Thanks,
–timNovember 17, 2014 at 10:43 am #6085Tim ScottParticipantJust wanted to post a quick update. I’m able to successfully run v2012.10 using the patch changes Jesse linked above. I can attach a USB thumb drive and download a file to that using USB / fatload commands. I was able to re-flash my MT100 using this and things are working well so far.
I seem to have taken a step back trying to run the latest version of U-boot, however. I applied the same patch changes to v2015.01-rc1 and the boot process appears to get hung up before it hits the prompt:
## Starting application at 0x21400000 …
U-Boot 2015.01-rc1-dirty (Nov 17 2014 – 10:30:52)
CPU: AT91SAM9G20
Crystal frequency: 18.432 MHz
CPU clock : 396.288 MHz
Master clock : 132.096 MHz
DRAM: 64 MiB
WARNING: Caches not enabledAt this point things hang indefinitely. Since I’ve got v2012.10 working, I can proceed with what I was originally working on. Still, it would be nice to know what the hang-up is for the latest version of U-boot. If I get anything figured out, I’ll be sure to post it here. If anyone has ideas for troubleshooting, I’d be interested to hear them.
Thanks!
–tim -
AuthorPosts
- You must be logged in to reply to this topic.