| |||||||||||||
PCIe, power management, and problematic BIOSesPCIe, power management, and problematic BIOSesPosted Jun 29, 2011 16:53 UTC (Wed) by nix (subscriber, #2304)Parent article: PCIe, power management, and problematic BIOSes I wonder if this is a contributing factor to this bug I'm experiencing where the kernel's attempt to turn ASPM off on an e1000 device whose support is lethally buggy fails or is somehow reverted shortly after taking effect, leading to an interface freeze a few minutes to days later... So far the e1000e guys have declared defeat and I've been too lazy/busy to hand it off to the PCI guys for further diagnosis (in part because the sourceforge bugtracker is so bloody ugly that I'm mildly ashamed to point anyone at it. Oops.)
(Log in to post comments)
PCIe, power management, and problematic BIOSes Posted Jun 29, 2011 16:57 UTC (Wed) by cesarb (subscriber, #6266) [Link]
There is always the possibility that SMM is playing with you, especially if it is onboard...
PCIe, power management, and problematic BIOSes Posted Jun 30, 2011 12:35 UTC (Thu) by nix (subscriber, #2304) [Link]
This seems unlikely. Before 2.6.36 (I think it was) when the kernel started activating ASPM systemwide, ASPM was turned off on this card and it never failed (well, not due to *this* problem anyway). So this would have to be an SMM that kicks in only if ASPM is turned on... it looks like ASPM is being reset when the interface is brought up or something, and I know the kernel knows about this because it prints a message. So, no, this really does look more like a kernel bug to me.
|
|||||||||||||