After discussions on the discord a few days ago I'm submitting a draft/prototype hwsurvey python script. When run, this script gathers certain specific, limited, data about the system it is running on, before presenting a report to the user and giving them the option to save it as a text file. It also asks for the user to consider submitting the report back to us, along with feedback regarding their installation experience.

The script does NOT have any capability to actually submit the report, this is partly deliberate. I feel that asking the user to manually upload the text report or copy/paste the contents into a web form is most within the spirit of Funtoo; and in any case drobbins was especially insistent that nothing happen without explicit consent of the user. A manual upload process ensures this consent.

The script gathers the following information:

from /proc/cpuinfo:

vendor ID, family, model, stepping, model name, sibling count and core count

from /proc/meminfo:

total ram, free ram, total swap, free swap

from /sys/:

L1,2,3,4 cache sizes, cpuinfo_max_freq from cpufreq, nodes (populated cpu sockets) counted

from lsmod:

loaded module names

from uname:

-srv, and separately, -m

from epro:

subarch, mix-ins

CPU clock is determined either from cpu_max_freq or by looking for '@' in the model name (latter overrides former)

threads per core is calculated by dividing siblings by cores.

EDIT: Having seen discussions on the discord relating to running this script from the livecd, I'm just putting on record that if this script is run from livecd vs installed system, some of the info will be incorrect/unhelpful (specifically the Funtoo subarch and mix-ins) as they will be reporting the generic livecd environment rather than the user's actual system. This particular script ideally needs to be run from the installed system after the install is complete and the machine has been rebooted at least once.