Thrilleratplay
[m920q] No display output
I just compiled both the SeaBIOS version as well as the edk2 version. The compilation process went off file. I had dumped the original BIOS from the flash chips earlier.
However, there was no display output with either of these coreboot binaries. I would love to help debug / fix this issue.
I'm attaching a screenshot of the coreboot compilation output.
If you have both DIMM sockets filled, I had a similar issue. You need to take out the DIMM1 and leave in DIMM0. I tracked it down to a specific change in coreboot. I am not sure what is meant by "catch potential configuration errors early on" or how this is resolved.
~SIDE NOTE: The edk2 isn't completing initialization, the bootsplash appears but then hangs. I will investigate this later today.~ This seems to be working now.
If you have both DIMM sockets filled, I had a similar issue. You need to take out the DIMM1 and leave in DIMM0.
Thank you very much for this tip! It helped me boot the board finally! :)
~SIDE NOTE: The edk2 isn't completing initialization, the bootsplash appears but then hangs. I will investigate this later today.~ This seems to be working now.
Let me recompile, re-flash and test out.
I'll add this to the readme as well but if you have any more issues, you can't resolve you read the debug output using a serial com port (search for the part number FRU 04x2733) and a USB to male serial adapter. Then on another computer use a serial TTY program like minicom or tio. When all connected all I had to do run run sudo tio /dev/ttyUSB0 then realized it was too much information and did it again but outputted to a log file. Make sure you start the program before turning on the M920q.
Can confirm that the issue with DIMM2 RAM slot persists. I've tried multiple combinations: different commits (497298708c330deb479fbf3ee6a3acc8bff2a102, 640a41f3ee938b794b140218921e0fd63b1d9235 and 7945a31e9172ec2939cc1abc6036962a6efb601b, which is the latest commit in the master branch at the moment), placing RAM in DIMM1, DIMM1+DIMM2 and only in DIMM2 slots. None of them worked except DIMM1 had a stick of RAM installed and DIMM2 was empty. Both DDR4 RAM sticks in DIMM1 were working fine:
- Kingston 32GB (2666MHz, CL 19-19-19-43)
- Samsung 8GB (2400MHz, CL 17-17-17-39)
The motherboard at some point managed to successfully boot once with the second RAM stick inserted in DIMM2 slot, but the Memtest86 showed only one RAM stick present during the tests (apparently the DIMM1).
FYI, I believe this is now fixed, as suggested by https://ticket.coreboot.org/issues/592#change-2093
@wrobelda I have not confirmed this yet. Have you had success with more recent versions of coreboot?
@wrobelda I have not confirmed this yet. Have you had success with more recent versions of coreboot?
I am corebootng my first device as I speak, I'll report back later coming week.
Well, turns out I am having trouble with the memory sticks I just ordered. Even though the system supports 64GB, somehow it refuses to boot with even a single DIMM module out of the two (Crucial CT2K32G4SFD832A).
That's quite surprising, I would assume it would just work with at least one of the modules.
EDIT: reported here: https://ticket.coreboot.org/issues/598
That RAM may be too fast. 2666MHz I believe is the max for the M920q and while it says that ram is downclockable to that speed, I'm not sure if it is something you need specifically set as a BIOS setting (whether corebort or stock)
That's fortunately not how this normally works, the RAM eeprom comes with SPD info, which is then used to set up the timings accordingly. Sometimes the SPD might be off, but this would pose a problem when trying to run RAM at faster speed than supported, not slower.
I reflashed the stock rom for now and the RAM worked just fine off the bat. It's a matter of SPD info missing in coreboot, I updated the upstream ticket with some extra findings on that.