An uninformed reader (i.e. me) could believe that the current workaround against the e1000 bug included in .27 isn't efficient against this bug.
Is that true ?
Posted Oct 16, 2008 15:14 UTC (Thu) by corbet (editor, #1)
[Link]
The 2.6.27 workaround will prevent the bug from destroying your hardware. But the buggy code is still there and could potentially create other difficulties. So upgrading to 2.6.27.1 (or just configuring out ftrace) might be a prudent thing to do.
yes, 2.6.27 is safe wrt. e1000e
Posted Oct 16, 2008 15:24 UTC (Thu) by mingo (subscriber, #31122)
[Link]
Correct - while .27 is safe wrt. e1000e EEPROM corruption (write modifications to the EEPROM were disabled by the driver), updating to latest .27.1 will get you the memory corruption fix as well.
Note that it is CONFIG_DYNAMIC_FTRACE=y (a performance feature) that was causing the sporadic memory corruption - plain ftrace was still fine. Both options were default-off.
The development ftrace code does not have this bug.