Comment # 13 on bug 1075876 from Benjamin Poirier

Created attachment 767941 [details]
e1000e: Log e1000e_get_base_timinca failures on spt macs (v2)

Thank you for the quick logs. Due to my mistake there was a problem in the
debugging patch which made the code always fail ("break" was moved inside a
for-loop and no longer exited the switch-case). I believe the cases with only
"Failure via adjfreq" would normally have been successes (ie. no "Failed to
restore TIMINCA clock rate delta") however the one case (comment 12) with "No
SYSCFI in TSYNCRXCTL: 0" is a genuine failure and shows the cause of the usual
problem.

To confirm that I've updated the kernel package in the
home:benjamin_poirier:branches:home:tiwai:bnc1075876 project with a
(hopefully) fixed debugging patch and removed Takashi's ~workaround. I've also
reduced E1000_SYSTIM_OVERFLOW_PERIOD from 4 hours to 10 mins so a crash should
occur sooner if the problem is not fixed. As instructed in comment 10, after a
boot when the "Failed to restore TIMINCA clock rate delta" message occurs,
please include dmesg. Also report if a similar oops occurs again.