[ARM/RISC-V] Maybe check /sys/devices/soc0

[ThomasKaiser]

Another sbc-bench try to differentiate SoCs / SoC revisions is parsing /sys/devices/soc0. But if doing so checking for /sys/devices/soc* might be an even better idea.

Most of the time this sysfs entry doesn't exist, sometimes it contains valuable hardware information, sometimes stuff we don't understand and sometimes rather confusing info (examples in this order):

• SpacemiT K1/M1: M1-8571 rev C on MilkV Jupier vs. K1-6370 rev C on SiPEED LPi3A – unlike device-tree this is hardware since SpacemiT's BSP kernel fills in the information based on the SoC's serials
• Ampere Altra: jep106:0a16 jep106:0a16:0001 rev 0x000000a1
• Exynos 5422 reads Exynos EXYNOS5800 rev 1 on Hardkernel's ODROID-XU4

Hmm... having checked sbc-bench's results collection right now with only SpacemiT K1/M1 providing relevant info via this sysfs node it's most probably not worth the efforts.

[ThomasKaiser]

Well, on other devices /sys/devices/soc0 also seems to contain useful information: https://github.com/search?q=repo%3ADr-Noob%2Fcpufetch+%2Fsys%2Fdevices%2Fsoc0&type=issues

[ThomasKaiser]

Texas Instruments SoCs running with their BSP kernel also seem to expose SoC info via /sys/devices/soc0. For example the BeagleY-AI board equipped with a TI AM67A SoC from the J722S family exposes this:

• /sys/devices/soc0/family = J722S
• /sys/devices/soc0/revision = SR1.0