No announcement yet.

Linux 5.5 Provides Knob To Toggle ASPM Link States Individually - Better Power-Savings

  • Filter
  • Time
  • Show
Clear All
new posts

  • Linux 5.5 Provides Knob To Toggle ASPM Link States Individually - Better Power-Savings

    Phoronix: Linux 5.5 Provides Knob To Toggle ASPM Link States Individually - Better Power-Savings

    ASPM can be a big boost to help power-savings on Linux laptops and desktops as shown by a prominent kernel regression a number of years ago. However, a number of Linux drivers are forced to disable Active State Power Management (ASPM) due to quirky/buggy hardware where it ends up not being sane to enable that power-saving feature by default. But with the Linux 5.5 kernel is support for toggling ASPM link states via sysfs as an easy-to-perform manner for achieving better power-savings with friendly devices...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Does it mean we can expect an improvement on this long-standing kernel bug? I'm tired of adding "pcie_aspm=off" to my kernel parameters and losing battery life for years.
    WORKAROUND: add pci=noaer to your kernel command line: 1) edit /etc/default/grub and and add pci=noaer to the line starting with GRUB_CMDLINE_LINUX_DEFAULT. It will look like this: GRUB_CMDLINE_LINUX_DEFAULT="quiet splash pci=noaer" 2) run "sudo update-grub" 3) reboot ---- My dmesg gets completely spammed with the following messages appearing over and over again. It stops after one s3 cycle; it only happens after reboot. [ 5315.986588] pcieport 0000:00:1c.0: AER: Corrected error received:...


    • #3
      It means now you can easily test-enable ASPM for your specific hardware if it is unnecessarily disabled because of other faulty hardware.

      If you need the flag to make your hardware work, then it is affected by an ASPM bug and enabling ASPM will probably cause issues (though I imagine the user enabling it after boot may bypasses a few issues), so likely not a direct solution to that...

      ...but at the very least makes debugging easier and shows kernel devs have their eyes on the subject now, so maybe a sign of future improvements


      • #4
        Thanks for mentioning PowerTOP in your article. That is exactly the sort of thing that should use these sysfs knobs.

        They may also be useful when debugging issues caused by defective hardware or software, but they should not be considered a solution to problems like that. The goal is that Linux should run correctly even if these knobs are never used.

        If we learn that using these knobs makes a device work better, e.g., if a device works reliably only when a certain ASPM state is disabled, we should change the driver or add a quirk to do that automatically.