Bug #582
closedfailure to boot after using haswell nri
100%
Description
Good day,
I am experiencing a rather strange issue with my system after trying out the recently merged Haswell NRI.
It was working great, booted fine and ran memtest86+ for the whole day, 10 passes with 0 errors. I then turned off the system for the night by just holding the power button down.
The next morning, the system would not boot at all. No display output. Ever since then I haven't been able to get it to boot at all.
One thing I have noticed is that the system will restart after about 15 seconds if there is an SPI flash chip present.
If there is no present SPI chip this doesn't happen. So I assume the PCH is reading it or at least trying to.
Worth noting is it only restarts one time, then it will just remain powered on forever until I turn it off.
The restarting also doesn't seem to be consistent, it seems to only happen if I flip the switch on the power supply
on and off. If I don't do this it will just remain powered on forever.
My hardware:
Z97E-ITX a/c
16GB TridentX 2400mhz DDR3
i7 4790S
Things I have done:
- Reflashed the original stock firmware
- Tried a different known working kit of RAM
- Removed everything from the system that is nonessential to boot (disk, keyboard, USB drives, etc.)
- Removed the CMOS battery
- Tried a different known working PSU
- Tried different SPI flash chips with coreboot on them
- Tried the original SPI flash chip with coreboot on it
- Tried coreboot with flashconsole enabled, nothing is written to it
At this point I suspect the problem is either with the motherboard, CPU or SPI flash chip. They are the only things I haven't swapped out.
One reason I suspect the SPI flash chip is because I wrote to it a lot of times in my initial attempts to get coreboot working on this board. So, maybe that caused excessive wear, I'm not sure. The other reason is because one of the last flashes I did was with the internal motherboard programmer, which worked fine. At the time of the flash I was running the board with all the flash descriptor binaries and an external microcode binary included for my CPU since building it from tree was failing. The image that I flashed with the internal programmer was the same image with me_cleaner run on it. The system booted fine, I confirmed me_cleaner worked but then I tried to do another flash to include memtest86+ as a second payload and flashrom gave some errors about not being able to erase, telling me the flash chip ended up being unchanged. I then flashed the image using an external programmer (raspberry PI with a breadboard) and that worked fine. This makes me somewhat doubt it's the SPI flash chip, flashrom successfully verifies the flash with no errors, and I can read it back fine too and it's the same as what I flashed. I tried other flash chips but my understanding is they aren't all command-compatible so that could be why they didn't work. Unfortunately I haven't been able to try the stock bios on another flash chip because I don't have any motherboards that have chips of the same size.
I know this could maybe be a hardware failure unrelated to coreboot or haswell NRI but I figured I would submit this with the hope that the developer or someone else here might have some better ideas about what is going on.
Updated by Ron Dillenger 12 days ago ยท Edited
- Assignee deleted (
Angel Pons) - % Done changed from 0 to 100
Problem is almost certainly electrical. Got a new SPI flash chip and put it in after a few days of the board being unplugged. It booted the first two times, after unplugging the power again I am back to the same issue as before. Probably some bad capacitors on the board.
Updated by Martin Roth 6 days ago
- Category deleted (
coreboot common code) - Status changed from New to Closed
Thanks for the update. I like the electrical explanation, because I was at a complete loss as to where to look for a solution here.
I'm going to mark this as closed, feel free to update or reopen if you find any more information.