On 22/08/21 13:06, Stijn Tintel wrote:
The T1042 CPU in the M200 is e5500, as opposed to the e6500 in the M300.
As Rui already explained, the T1042 does not support AltiVec, which musl
expects for all PPC64 CPUs. I was able to boot Linux on it after
patching the vector instructions out of the musl PPC64 assembly code
[1], but I suspect I also got bitten by the THREAD_SHIFT value being set
to the default for PPC32. I have just verified that with THREAD_SHIFT
correctly set and that musl hack, the M200 also boots fine, to be able
to give you the data about different RAM modules and ECC. The m200
branch in my staging tree [2] should be a good starting point for people
who want to continue on it. Once the qoriq target made it in the tree,
I'll have another go at the M200, and I'll move it to the qoriq target
then (it's still in the mpc85xx target atm), and do the DTS like I did
for the M300.


Thanks for taking the time to test on both devices. ECC support is a very rare feature in OpenWrt devices outside of x86 target. Sadly I'm not able to deal with compile issues in Musl so I'll not be working on M200.

-Alberto

_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel

Reply via email to