Modules G+F interoperability

This is gonna be a continuation of my previous post from general discussion. I went ahead and ordered an F module to my existing AG (G etched on the PCB, just to be certain that I didn’t get B by accident).
After a whole day of debugging many problems I come to a state where I have my NAS working or I have Wi-Fi (the G one). This is a log from when I tried to disabled the Wi-Fi, rebooted, checked my drives are mounted and working, and then enabled Wi-Fi again. The moment I tried to run a speed check on my phone, my drives disconnected and basically the whole F disconnected as evidenced by the log. Is there something obvious that I’m missing or is AGF not really viable?

Dec 15 17:45:46 turris hostapd: Configuration file: /var/run/hostapd-phy0.conf
Dec 15 18:45:47 turris kernel: [ 116.418490] ath10k_pci 0000:02:00.0: 10.1 wmi init: vdevs: 16 peers: 127 tid: 256
Dec 15 18:45:47 turris kernel: [ 116.436394] ath10k_pci 0000:02:00.0: wmi print ‘P 128 V 8 T 410’
Dec 15 18:45:47 turris kernel: [ 116.442706] ath10k_pci 0000:02:00.0: wmi print ‘msdu-desc: 1424 sw-crypt: 0’
Dec 15 18:45:47 turris kernel: [ 116.449818] ath10k_pci 0000:02:00.0: wmi print ‘alloc rem: 26384 iram: 26852’
Dec 15 18:45:47 turris kernel: [ 116.531563] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
Dec 15 18:45:47 turris kernel: [ 116.541983] br-lan: port 3(wlan0) entered blocking state
Dec 15 18:45:47 turris kernel: [ 116.547244] br-lan: port 3(wlan0) entered disabled state
Dec 15 18:45:47 turris kernel: [ 116.553397] device wlan0 entered promiscuous mode
Dec 15 17:45:47 turris hostapd: wlan0: interface state UNINITIALIZED->COUNTRY_UPDATE
Dec 15 17:45:47 turris hostapd: wlan0: interface state COUNTRY_UPDATE->HT_SCAN
Dec 15 17:45:47 turris hostapd: Using interface wlan0 with hwaddr 04:f0:21:45:cd:5d and ssid “Turris”
Dec 15 18:45:47 turris kernel: [ 116.792546] IPv6: ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready
Dec 15 18:45:47 turris kernel: [ 116.798943] br-lan: port 3(wlan0) entered blocking state
Dec 15 18:45:47 turris kernel: [ 116.804492] br-lan: port 3(wlan0) entered forwarding state
Dec 15 17:45:47 turris hostapd: wlan0: interface state HT_SCAN->ENABLED
Dec 15 17:45:47 turris hostapd: wlan0: AP-ENABLED
Dec 15 17:45:47 turris netifd: Network device ‘wlan0’ link is up
Dec 15 17:45:47 turris netifd: lan (3905): udhcpc: performing DHCP renew
Dec 15 17:45:47 turris netifd: lan (3905): udhcpc: sending renew to 0.0.0.0
Dec 15 17:45:47 turris netifd: lan (3905): udhcpc: lease of 192.168.0.10 obtained, lease time 3600
Dec 15 18:45:48 turris kernel: [ 117.416550] ath10k_pci 0000:02:00.0: NOTE: Firmware DBGLOG output disabled in debug_mask: 0x10000000
Dec 15 17:46:01 turris hostapd: wlan0: STA 08:c5:e1:a9:5f:13 IEEE 802.11: authenticated
Dec 15 17:46:01 turris hostapd: wlan0: STA 08:c5:e1:a9:5f:13 IEEE 802.11: associated (aid 1)
Dec 15 17:46:01 turris hostapd: wlan0: AP-STA-CONNECTED 08:c5:e1:a9:5f:13
Dec 15 17:46:01 turris hostapd: wlan0: STA 08:c5:e1:a9:5f:13 RADIUS: starting accounting session F6FA699A10DF938F
Dec 15 17:46:01 turris hostapd: wlan0: STA 08:c5:e1:a9:5f:13 WPA: pairwise key handshake completed (RSN)
Dec 15 18:46:54 turris kernel: [ 183.290528] xhci_hcd 0000:03:00.0: xHCI host not responding to stop endpoint command.
Dec 15 18:46:54 turris kernel: [ 183.298975] xhci_hcd 0000:03:00.0: xHCI host controller not responding, assume dead
Dec 15 18:46:54 turris kernel: [ 183.307165] xhci_hcd 0000:03:00.0: HC died; cleaning up
Dec 15 18:46:54 turris kernel: [ 183.307288] usb 2-1-port4: cannot reset (err = -22)
Dec 15 18:46:54 turris kernel: [ 183.315695] usb 2-1: USB disconnect, device number 2
Dec 15 18:46:54 turris kernel: [ 183.318165] usb 2-1.4: USB disconnect, device number 3
Dec 15 18:46:54 turris kernel: [ 183.360418] sd 0:0:0:0: [sda] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x01 driverbyte=0x00
Dec 15 18:46:54 turris kernel: [ 183.368559] sd 0:0:0:0: [sda] tag#0 CDB: opcode=0x28 28 00 00 00 09 40 00 00 08 00
Dec 15 18:46:54 turris kernel: [ 183.376521] print_req_error: I/O error, dev sda, sector 2368
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read index block: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read of MFT, mft=17 count=1 br=-1: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Could not decode the type of inode 17
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read of MFT, mft=3826 count=1 br=-1: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Could not decode the type of inode 3826
Dec 15 18:46:54 turris kernel: [ 183.391976] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.397359] Buffer I/O error on dev sda1, logical block 786436, async page read
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read of MFT, mft=17 count=1 br=-1: I/O error
Dec 15 18:46:54 turris kernel: [ 183.409624] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.414892] Buffer I/O error on dev sda1, logical block 788526, async page read
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read of MFT, mft=8376 count=1 br=-1: I/O error
Dec 15 18:46:54 turris kernel: [ 183.427400] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.432404] Buffer I/O error on dev sda1, logical block 786441, async page read
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read of MFT, mft=36 count=1 br=-1: I/O error
Dec 15 18:46:54 turris kernel: [ 183.442959] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.448038] Buffer I/O error on dev sda1, logical block 787388, async page read
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read of MFT, mft=3826 count=1 br=-1: I/O error
Dec 15 17:46:54 turris block: hotplug-call call failed
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read of MFT, mft=3827 count=1 br=-1: I/O error
Dec 15 18:46:54 turris kernel: [ 183.462236] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.467139] Buffer I/O error on dev sda1, logical block 787388, async page read
Dec 15 18:46:54 turris kernel: [ 183.478339] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.483581] Buffer I/O error on dev sda1, logical block 4258931, async page read
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read vcn 0x28: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read vcn 0x28: I/O error
Dec 15 18:46:54 turris kernel: [ 183.491967] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.497046] Buffer I/O error on dev sda1, logical block 4258931, async page read
Dec 15 18:46:54 turris kernel: [ 183.505613] blk_partition_remap: fail for partition 1
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read index block: I/O error
Dec 15 18:46:54 turris kernel: [ 183.510676] Buffer I/O error on dev sda1, logical block 4258926, async page read
Dec 15 18:46:54 turris kernel: [ 183.519393] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.524670] Buffer I/O error on dev sda1, logical block 4258931, async page read
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read vcn 0x28: I/O error
Dec 15 18:46:54 turris kernel: [ 183.583274] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.588482] Buffer I/O error on dev sda1, logical block 4258931, async page read
Dec 15 18:46:54 turris kernel: [ 183.597354] blk_partition_remap: fail for partition 1
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read vcn 0x28: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read vcn 0x28: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read index block: I/O error
Dec 15 18:46:54 turris kernel: [ 183.603599] blk_partition_remap: fail for partition 1
Dec 15 18:46:54 turris kernel: [ 183.609498] blk_partition_remap: fail for partition 1
Dec 15 17:46:54 turris ntfs-3g[4501]: ntfs_attr_pread_i: ntfs_pread failed: I/O error
Dec 15 17:46:54 turris ntfs-3g[4501]: Failed to read vcn 0x28: I/O error
Dec 15 17:47:09 turris ntfs-3g[4501]: Unmounting /dev/sda1 (RZR-MBK)

I’d like to bump this thread to at least get some discussion going on, because I’m not considering my MOX to be working as intended.

Current state:

The device itself is stable, no weird reboots, HDDs are working fine, Qualcomm Wi-Fi still unusable.

What I’ve tried so far:

  1. Rescue mode 6 and installing minimum packages apart from LXC and SMB
  2. Making sure that I have all the necessary kmod drivers (that alone is a separate issue to a G module) - How to force USB 3 device to use mass-storage driver using quirks?
  3. Using additional power supply to make sure MOX doesn’t have enough juice
  4. Using only 1 known good USB drive that does not require USB power
  5. Checking the Qualcomm’s mPCIe card is properly seated - which is known to be working flawlessly before
  6. Checking G module is properly seated
  7. Trying HBT branch with OS 5.0

Without knowing the specifics of MOX I suspect that both G and F are sharing PCI lanes and when I hit the Wi-Fi with a load (e.g. speedtest), I effectively cut the communication to the USB hub as the G module gets priority over that. That would render the F module completely useless.

I can dig up any info necessary for debugging but everything seems normal during boot and I already posted the log of what happens when I try to use Qualcomm Wi-Fi with HDDs connected.

Could anyone please share any information or reasoning of what I’m experiencing with my MOX?

Hi,
I have a similar configuration AGFC and I have experienced the same problems.
Everything we connect to USB ports (even in case of USB3.0 port on A module) is disconnected in while. The noticeable difference between A and F USB 3.0 ports is that USB 3.0 port on A module is sometimes able to recovery from failure state (tested with USB 3.0 hub). The F USB 3.0 ports are unable to detect the connection of anything until reboot.

The devices which I tried are, external USB Sandisk SSD, FTDI serial, Behringer USB2.0 sound card, USBI2C01B converter.

Could you confirm that if you disable the Wi-Fi card in the G module (Qualcomm), your F module gets stable? I’m still fairly sure that the culprit is somewhere along the PCI bus.

Yes, I could confirm that. I do not disable the Qualcomm Atheros QCA9880 802.11bgnac module itself. I had only disabled all wifi connections in the LuCi interface and USB FTDI serial device connected to the F module become stable without noticeable failures.

Is there some software option to disable USB HUB in the F module to do a reciprocal test with wifi?

I’ve been trying unsuccessfully to get any help from the Turris team for over a month now. There’s no official response to my thread nor to my support ticket. I have a sneaking suspicion that it’s a hardware flaw that cannot be fixed by software change.
Now I’m stuck with a network device that actually does no real networking and only provides access to my drives. And that itself could be substituted by a simpler (and cheaper) device.

Still hoping I’m wrong but until somebody actually shows up, I declare my project as a expensive failure.

I wrote at Turris tech support yesterday, sending the “diagnostics data”, but no response until now.

I suppose there is something wrong with PCIe data transmission. Therefore I had try to test possible wifi interference on PCIe bus. The bus is probably 2.5Gbps wchich is pretty close to 2.4GHz and 5 GHz wifi carrier frequency. Unfortunatelly I noticed the F module malfunction even in case the wifi power was set to 1 mW (no matter if was 2.4GHz or 5 GHz), but I do not know if power settings is applied on every transmitted wifi packet.