Debian on Dell Wyse T10 / T50

@tmn505,

> @bodhi I have few patches You might be interested
> in and cleaned up the dts, these are not tested
> with T50, so validity of changes in PCIE and SD
> CARD node need to be confirmed.
> https://github.com/tmn505/openwrt/tree/dove/target/linux/dove

Thanks! I'll take a look.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Hi,

I recently bought some T10 devices and found this forum. Based on the information gathered here I was able to get results pretty quickly. Thank you all!
My long term plans include some hardware modifications, like adding sata, pcie and sd card connectors, and it would be nice to get the DVI output working.

I first tried openwrt and uboot as described by @tmn505. It started fine at first, but after a few minutes it froze and refused to start again. After a couple of hours and testing several devices, I found that some don't start at all, and some start to fail the memory test after 2-3 minutes. This worked pretty much stable with fan cooling.
My devices have samsung K4B2G0846Q-BYK0 memory chips, but this should be interchangeable with the H5TQ2G83CFRPBC-H9C chip mentioned by tmn505.
That's when I started comparing the Dram controller configuration part of the original firmware and the uboot.
I found two relevant differences: (I only looked at the CPU@1000MHZ DDR3@500MHZ part, I think the other modes are not used)
1: Trfc value is set to 118ns, but should be 160ns according to the chips datasheets.
2: The PHY Control Register 9 Phy_wck_ac_dly value is different, but it seems to be stable without fixing this, but I fixed this anyway.

At the beginning of the uboot binary I modified the boot script with hex editor in these places and now my 3 tested devices work fine at high temperatures.

So @tmn505 I would like to suggest you to modify the dram init script(dramregs_d2plug_hynix_1gb.txt) as below. I think this will also make the hynix chip version more stable.
line 90: 0xD0800060 0x74780504 #Trfc 118ns ->160
line105: 0xD08001F0 0x00003F33 #PHY Control Register 9 Phy_wck_ac_dly from original init script

I would like to ask you, what toolchain did you use to compile the uboot? Can you help me with this?

Thank for the report. I'll check later if the changes in DRAM configuration don't change stability on my unit. If they are good, I'll commit them to the repo by copying the dram config to a new file, with only 1000mhz@500mhz profile, since probably this is the only one probed (or You can create a PR, if yes, don't foget Your SoB). On side note, D2PLUG U-Boot source had neat RAM diagnostic built-in, which helped to find good DRAM file, patch in attachment.

For compilation I'm using Linaro arm-eabi toolchain version 4.9-2017.01. Check this workflow file which is used to publish binaries in my repository: https://github.com/tmn505/u-boot/blob/gh-pages/.github/workflows/publish.yml. Make sure that device tree compiler headers are not installed on Your system, otherwise compilation will fail.

DVI output might be difficult to achieve, since there's no IDT5V49EE503 driver in kernel. Maybe it's used to generate clocks only for VGA signal? So outputting digital signal could be possible?



Edited 1 time(s). Last edit at 04/16/2024 08:41AM by tmn505.