Hello,

We have a custom board with ARM processor and FT2232H device, configured in 2 UARTs links. We used Linux OS and device is correctly found and attached:

[    0.031124] usbcore: registered new interface driver usbfs
[    0.031155] usbcore: registered new interface driver hub
[    0.031183] usbcore: registered new device driver usb
[    0.835577] dwc3 3100000.usb: Adding to iommu group 7
[    0.835896] dwc3 3110000.usb: Adding to iommu group 8
[    0.838035] usb usb2: We don't know the algorithms for LPM for this host, disabling LPM.
[    0.839505] usb usb4: We don't know the algorithms for LPM for this host, disabling LPM.
[    0.840173] usbcore: registered new interface driver usb-storage
[    0.840243] usbcore: registered new interface driver ftdi_sio
[    0.840263] usbserial: USB Serial support registered for FTDI USB Serial Device
[    1.212986] usb 1-1: new high-speed USB device number 2 using xhci-hcd
[    1.394520] usb 1-1: Detected FT2232H
[    1.394720] usb 1-1: FTDI USB Serial Device converter now attached to ttyUSB0
[    1.395004] usb 1-1: Detected FT2232H
[    1.395184] usb 1-1: FTDI USB Serial Device converter now attached to ttyUSB1

To perfom test, Tx is loopback externally to Rx.

Randomly, and most of time after several ten of hours, generally one of two channels got errors. On channel with error, data goes out from processor not with the good period and FT2232 Rx FIFO is never read for endpoint n ° 1. More strange, transfer mode is changed from bulk to "normal" transaction. Once error occurs, it never desapears.

We catch data exchange with USB analyser with and without errors.

has somebody have already seen this behaviour ?