[BUG] Atheros QCA9377 Wi-Fi vanishes ONLY on battery boot

Forum › Forums › Official Releases › antiX-26 “Stephen Kapos” › [BUG] Atheros QCA9377 Wi-Fi vanishes ONLY on battery boot

  • This topic has 6 replies, 3 voices, and was last updated Apr 15-11:13 am by Brian Masinick.
Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
    Posts
  • #201739
    AntiSlop
    Member

      Hi everyone,

      I’m facing a very specific and frustrating power management issue with my Wi-Fi on antiX 26 (Kernel 6.6.119). I have spent hours debugging this and finally found the exact trigger. I would really appreciate some expert advice.

      The Hardware:
      * Laptop: Acer
      * Network Controller: Qualcomm Atheros QCA9377 802.11ac

      The Exact Trigger (The Catch):
      * If I boot, reboot, or shut down the laptop while the AC Adapter is PLUGGED IN, the Wi-Fi works flawlessly. wlan0 comes up, connects instantly, and is perfectly stable.
      * If I boot, reboot, or shut down the laptop while on BATTERY POWER, the Wi-Fi card completely “dies” and disappears. ip a only shows lo and eth0.

      The Symptoms & Logs (When booting on battery):
      During a battery-powered boot, the system hangs for about 15 seconds on Starting udev-settle.
      Checking sudo dmesg | grep -i ath10 reveals the driver is panicking because the card is stuck in a deep hardware sleep and won’t respond:

      [ ... ] ath10k_pci 0000:03:00.0: failed to wake target for write32 at 0x00034400: -110
      [ ... ] ath10k_pci 0000:03:00.0: failed to wake target for read32 at 0x00034410: -110
      [ ... ] ath10k_pci 0000:03:00.0: failed to read device register, device is gone
      [ ... ] ath10k_pci 0000:03:00.0: failed to reset chip: -5
      [ ... ] ath10k_pci: probe of 0000:03:00.0 failed with error -5

      (To temporarily revive the card, I have to do a hard power drain: unplug AC, hold the power button for 60 seconds, and boot with AC plugged in).

      What I have already tried (that FAILED to fix the battery boot issue):
      1. Disabled ASPM: Added pcie_aspm=off to GRUB_CMDLINE_LINUX_DEFAULT.
      2. Blacklisted Acer modules: Created a modprobe file with blacklist acer_wmi and blacklist acer_wireless.
      3. Disabled OTP memory check: Added options ath10k_core skip_otp=y to /etc/modprobe.d/ath10k.conf.
      4. Disabled “Fast Boot” and “Secure Boot” in the BIOS.

      My Question:
      It seems that Kernel 6.6.x applies an aggressive ACPI/power-saving profile to the PCIe bus during a battery boot that literally starves the Atheros card, causing it to lock up. (Note: This did NOT happen on antiX 23 with older kernels).

      Is there a specific kernel parameter, udev rule, or TLP/power-management configuration in antiX 26 that I can modify to force the system to treat the Wi-Fi PCIe slot as if it’s on AC power, even when booting on battery?

      Thank you in advance!

      #201740
      anticapitalista
      Forum Admin

        Have you tried using the 5.10 ‘legacy’ kernel that is also shipped on the iso?

        Philosophers have interpreted the world in many ways; the point is to change it.

        antiX with runit - leaner and meaner.

        #201741
        AntiSlop
        Member

          Hi anticapitalista,

          You were absolutely right! I just booted into the 5.10 ‘legacy’ kernel on battery power, and the Wi-Fi works flawlessly without any delays or hardware lockups.

          There is definitely a power management/ACPI regression in the 6.6 kernel series specifically for this Atheros card when running on battery.

          I will set the 5.10 kernel as my default and stick with it. Thank you so much for the incredibly fast help and for an amazing distro!

          #201742
          anticapitalista
          Forum Admin

            I agree with you that the 6.6 kernel has this issue and maybe later updates have fixed it. I hope so.
            At some point, I’ll build later versions of the 6.6 kernel.
            I notice the latest is now 6.6.134

            Philosophers have interpreted the world in many ways; the point is to change it.

            antiX with runit - leaner and meaner.

            #201834
            Brian Masinick
            Moderator

              I agree with you that the 6.6 kernel has this issue and maybe later updates have fixed it. I hope so.
              At some point, I’ll build later versions of the 6.6 kernel.
              I notice the latest is now 6.6.134

              I believe that the .134 kernel, at least the ones currently appearing, is 6.1.134; the most current 6.6 kernel showing up as of 11:30 AM EDT (GMT -4) is 6.6.119.

              --
              Brian Masinick
              Alternate "Search B":
              This search page

              #201836
              anticapitalista
              Forum Admin

                Brian – I meant latest from kernel.org is 6.6.134

                Philosophers have interpreted the world in many ways; the point is to change it.

                antiX with runit - leaner and meaner.

                #201842
                Brian Masinick
                Moderator

                  Brian – I meant latest from kernel.org is 6.6.134

                  Okay, then we’re good. As far as on my every day system, this Lenovo Thinkpad T14 Gen 3 has been pretty easy with kernels; while I do have to use a V6 kernel in order to get WiFi access, the V5 or V6 kernels boot to login and I’ve been able to run 6.1.134, 6.6.101, and 6.6.119. On my Dell, I’ve tested both the V5 and V6 antiX kernels and they work fine, as do various Debian and Liquorix kernels.

                  It’s been a long time since I’ve seen a problematic kernel. Ironically the most recent one I actually remember was the lone Liquorix kernel issue I’ve ever seen in the entire time I’ve used those kernels and it was fixed in a day or less and hasn’t been a problem since. Our kernels have been excellent; recently I’ve been sticking with antiX kernels on our releases and Liquorix kernels when I use MX Linux; they are all very solid. When or if the 6.6.134 kernel is built for antiX I anticipate similar positive results. One reason I have good kernel results is that I choose very ordinary mainstream systems that have been out a while and well tested; that helps, so quirks in the kernel are things that I rarely experience because of this.

                  --
                  Brian Masinick
                  Alternate "Search B":
                  This search page

                Viewing 7 posts - 1 through 7 (of 7 total)
                • You must be logged in to reply to this topic.