Comment 2 for bug 2020656

Digging the Internet I reached some posts/bugs that talked about a problem with some amd cpu that hard time getting out of C6 power state.

So I applied one of the suggested workaround that is: prevent cpu to enter C6 state so it won't have problem to exit ^^

For this purpose is used cpupower, more precisely I used this command: cpupower idle-set --disable-by-latency 350

before I ran this command, cpupower moniter résulted in:
    | Mperf || Idle_Stats
 CPU| C0 | Cx | Freq || POLL | C1 | C2 | C3
   0| 0,07| 99,93| 1429|| 0,00| 0,00| 0,00| 99,86
nearly the same for other cores

I guess it means that all cores can sleep down as much as possible…

after the command, cpupower moniter résulted in:
    | Mperf || Idle_Stats
 CPU| C0 | Cx | Freq || POLL | C1 | C2 | C3
   0| 0,07| 99,93| 1429|| 0,00| 0,00| 99,91| 0,00
nearly the same for other cores

It seems cores can't sleep as much as before.

I've not really understood what the C6 state is and what the mapping with those Cx shown by cpupower is (max for cpupower is C4).
So far, cpupower idle-set --disable-by-latency 350 seems to do a good job preventing the cpu to go to far in sleeping state.

Nearly one full day without a freeze makes my computer usable.

Some questions please:
- any help to understand the C6/Cx mapping?
- is it a software problem (bug in the kernel like I'm guessing and I can hope for a fix) or a hardware problem and I should send the computer back to the manufacturer?
- I'm on laptop, how to evaluate the impact on the battery usage?

Many thanks