[Intel Euclid] CMOS reset -- Not update, actually reset. Euclid will not boot due to settings in BIOS.

[TJoyn]

Hi all. I saw that someone asked how to clear nvram/cmos before and was just given the BIOS update file... ( )

That won't work here as I enabled some boot settings in the BIOS and now there's no output on the UART nor on video/HDMI; therefore I *cannot* alter it's filesystem.

My USB keyboard doesn't even turn on, but I do get a green light, it gets really hot, and my USB hub turns on. Standard BIOS halt.

I don't see any buttons, and I don't even see any in scanlime's teardown. I'm at a loss here.

Can someone from Intel tell me how I can reset it or point me in the right direction? @amitmoran maybe?

I really, really don't want to take this thing apart just to clear the CMOS.

I tried to file a customer support ticket, but your website sends broken 302 redirects that include `null` at the end of the domain instead of the URI component (`https://www-ssl.intel.comnull/ https://www-ssl.intel.comnull/ `), so I'm here.

Also, anyone that's used UART on the Euclid -- What output do you see during boot?

(Trying to figure out if it's at all overridable from serial ala uboot somehow and I'm just missing something here.)

Thanks,

Trevor

Message was edited by: Trevor Joynson

[TJoyn]

Also -- Is there really no Euclid specific forum here? Strange as it's so much different than just a realsense camera. It contains one, sure, but most of it is not realsense, you feel me?

[TJoyn]

Okay, it *looks* like from the teardown that it may be the pair of pads above the debug port, ie next to "20". Pic is upside down:

Is this true? If I short that does it void my warranty? To be honest I have absolutely no idea here so I'm not touching it until I get a response. I don't see any battery here either, which makes me wonder how long I may have to leave it unpowered before it loses it's memory (if it even does, what if it doesn't use a CMOS circuit and instead writes to flash? that would be rather terrible.)

It scares me that the pads are right next to the SDIO expander though. hmm.

[TJoyn]

From what I've gathered the CMOS logic is actually contained *inside* the PCH (https://www.intel.com/content/dam/www/public/us/en/documents/datasheets/7-series-chipset-pch-datasheet.pdf https://www.intel.com/content/dam/www/public/us/en/documents/datasheets/7-series-chipset-pch-datasheet.pdf), which is interesting. There's apparently a pin, RTCRST, meant just for that jumper we're used to seeing. (It warns vendors in the datasheet about pulling VCCRTC down with the jumper, I'm guessing that's to save the battery over time?). Not that any of that really helps me here as that chip is on the other side of the board (judging again from scanlime's teardown here), underneath a large shield, which I highly doubt it would be put there.

For anyone else curious, this is the relevant part of the datasheet for the PCH: https://www.intel.com/content/dam/www/public/us/en/documents/datasheets/7-series-chipset-pch-datasheet.pdf# G3.1384352 https://www.intel.com/content/dam/www/public/us/en/documents/datasheets/7-series-chipset-pch-datasheet.pdf# G3.1384352

I'm starting to think it may not be 20 after seeing that schematic.

Intel, help?

[TJoyn]

Should I just RMA it at this point? It seems a bit extreme to have to do that for just changing BIOS boot-time settings, doesn't it? (Plus what happens when I do it again?)

I've been up and down the documentation, and, I'm surprised there isn't anything in here related to this.

The Joule has/had excellent docs with schematics to boot

There _has_ to be a nice way to reset the BIOS without voiding my warranty, and I've got a feeling it's accessible from either the debug port or those pads around it (please correct me if I'm wrong).

I figured at first it would of course just be a button combination during power on, but I've tried all common button combinations I can think of to try to get it into some kind of bios recovery mode/alternate boot rom..

- I took the `.fv` bios flash file for the Euclid and placed it on a ESP/vfat formatted sdcard hoping it would look there first or something, but it seems not to notice.

- I left it unplugged for 6 hours with the battery out, hoping that since I don't see a battery on scanlime's teardown it would reset within that time. (I'm guessing those caps are pretty good, because it just laughed at me with it's green power light.)

I suppose I'll give it a bit more before RMA just in case (I'm not sure how often Intel frequents these forums).

Trevor

[idata]

Hello Tervorj,

We have escalated your question to the Euclid engineering. We will post their response here as soon as it is available.Regards,Jesus G.Intel Customer Support

[TJoyn]

Thanks. Appreciate that. Do you know "about" how long it may be by chance just so I can plan? I'd really like to get to work on my project over the weekend!

[idata]

Hello TrevorJ,

The feedback from the Euclid team is that it sounds like you have a severe hardware problem. Best to RMA at this point. I'm sorry about this, I know this isn't what you wanted to hear.Regards,Jesus G.Intel Customer Support

[TJoyn]

It's not a hardware problem. It's a BIOS problem. There are options in there related to bootup and init that soft-brick it like this (as I found).

An RMA will work around this, but it's not going to help me when I or anyone else does this again trying to troubleshoot another problem like I was.

Can you please request that they tell me how to reset the CMOS? That will, I'm 99% certain, make this unit work again. (That's also my original question )

(also thank you!)

[idata]

Hello trevorj,

We are working with engineering to determine what can be done in this case. Please stay tuned. Thank you for your patience.Regards,Jesus G.Intel Customer Support

[Daohua_W_Intel]

Hi,

It is very hard for us to reproduce the failure that user encountered. We need to know what actions user did or what boot settings user changed that caused the system brick, and then try to reproduce the symptom. Otherwise, the BIOS setting is stored in SPINOR flash, not in CMOS.

If user just want to know how to reset BIOS setting to recovery the device, here is the method:

1. Copy the attached bios image(with sudo) to /boot/efi/IFWI.bin, then reboot the device and wait about 5 minutes for the BIOS updating.

or

2. Flash the attached bios image using debug card and dediprog tool.

[TJoyn]

Ah, I didn't see that you had edited your reply with more info @HowardWu ! Ty!

I would love to tell you exactly what settings caused this, but without seeing the BIOS in front of me I won't be able to remember the exact text.

Once this is back up and functioning however, I can tell you once I go in there and see it.

SPI flash -- makes sense. Also may make this harder to fix, but hopefully the init sequence looks for this file way before it crashes and burns.

Would this work from microsd? Because I can't place anything onto the euclid itself since it won't boot.

I'll try placing it onto a microsd (gpt formatted with just a sole EFI partition; IFWI.bin placed in the efi folder as you mentioned) and will report back.

If not, can I control or interrupt the boot process over serial ala uboot to either force flashing the firmware from microsd, or even reset perhaps?

It's possible that the serial rig I made for this simply isn't correct, I don't have any other devices that utilize UART over microsd connector to test it against.

Edit:

By "debug card", do you mean something like what's used here for Cherry Trail? (eg vs a microsd)

https://software.intel.com/en-us/articles/how-to-run-intel-system-debugger-aka-xdb-for-target-cht-ffd How to run Intel® System Debugger (aka XDB) for target CHT-FFD | Intel® Software

The ITP-XDP3 is $3k from the looks of it, which is definitely on the expensive side for me; If so, are there any cheaper alternatives to the model `ITP-XDP3` by chance?

By `dediprog` do you mean ala the `SF-100` programmer that's used to flash other Intel boards (eg Gallieo, MinnowBoard, others)?

If so, which cable(s) should I expect to be compatible here?

Ty, will reply shortly.

Trevor

[TJoyn]

Welp, it did not do anything with it placed on the micro-sd (gpt partitioned with a single EFI-labelled vfat partition with IFWI.bin placed in the said place).

SPI flash -- I assume that means it's on the same winbond chip that the firmware itself is on.

Is it read in one go upon init? Or does it start the firmware which then loads it's configuration from SPI during init (two separate reads)?

Is it stored at the beginning or the end of the flash?

Does the config being stored on SPI mean that RTCRST does not clear the configuration at all on this device?

[Daohua_W_Intel]

Debug card doesnt mean the XDP. it is a dedicated debug board only used on Euclid.

Dediprog means SF-100 programmer,

[Daohua_W_Intel]

the SPI flash(winbond chip) stores the UEFI BIOS firmware , the bios configuration data is part of the firmware. The RTCRST can not clear it, we does not reserve the solution to clear data in SPI flash.

[TJoyn]

Awesome, thank you for the info!

Not sure if you saw this by the way, but just for my curiosity for when this happens again (it will I'm sure, I just started RMA a few days ago after giving a flash a shot from microsd and usb3):

> Is it read in one go upon init? Or does it start the firmware which then loads it's configuration from SPI during init (two separate reads)?

> Is it stored at the beginning or the end of the flash?

I'll keep track of which settings I've changed this time and do it in stages. This way I'll find our approximately which settings cause this soft brick so you can fix it.

I'll respond back here. Keep in mind this will be a while, because I haven't yet gotten a response, I'm sure just due to the holidays, then I have to wait for a shipment and such.

[TJoyn]

Oh, and I read in another post that a debug board is not available? Is this true?

If the card is not available:

- Can we make it available so I can fix the unit myself and in addition get access to i2c peripherals?

I somewhat expected that this unit would be treated like the Joule, eg with everything out in the open, since it's development kit hardware, and having the ability to debug would be extremely beneficial!

For instance, I wouldn't have to do an RMA right now, plus I'd get a cornucopia of features I wish I had right now, such as i2c!

Thanks again.

[TJoyn]

This is the debug board, correct?

/servlet/JiveServlet/downloadImage/2-481828-378003/debugcard.jpg https://communities.intel.com/servlet/JiveServlet/downloadImage/2-481828-378003/debugcard.jpg

[Daohua_W_Intel]

During BIOS boot up, the boot up code may read and write the configuration data many times by items.

the SPI Flash size is 8M bytes, BIOS firmware with nvram area is the upper 3M bytes.

And please do not change the OS boot options in BIOS, the Euclid device doesnt support Windows, only support ubuntu.

[Daohua_W_Intel]

yes, that is the debug card for Euclid.

[idata]

Hello TrevorJ,

You are correct, Intel does not provide the debug card.Regards,JesusIntel Customer Support