Thanks Andrei, I have done some more testing and I have found that - with everything else being the same - I can only see the issue on kernel 4.4. On kernel 4.1 it is stable.
As an alternative workaround I have noticed that disabling USB 2.0 by booting with dwc_otg.speed=1 makes the system stable on any kernel version - albeit of course with much lower USB speeds. For my particular application either of these alternatives work but I would rather use the latest kernel and at full USB speed. I have tried changing the core frequency (i.e. core_freq=250) but that didn’t make any difference. The change seems to be mostly related to the serial ports on the RPi which is not the same as the problem I am experiencing. My problem is that the USB ports on the RPi stop working. /Martin > On 08 Jul 2016, at 15:42, Andrei Gherzan <and...@gherzan.ro> wrote: > > On Mon, Jun 13, 2016 at 11:24:08PM +0200, Martin Bergek wrote: >> Hi, >> >> I am testing an image on a system including a PL2303 (USB to serial bridge) >> and have noticed an issue on the Raspberry Pi 3. The system will run fine >> for an hour or two but then the kernel logs the following error: >> >> pl2303 ttyUSB0: usb_serial_generic_read_bulk_callback - urb stopped: -32 >> >> After this there is no more data coming from the USB device and I have to >> either reboot or disconnect/connect the USB device - neither of which is >> acceptable in my application. >> >> This is on kernel 4.4.9 and it happens only on the Raspberry Pi 3. On >> Raspberry Pi 2 the exact same firmware runs flawlessly. >> >> Has anyone else come across this or have some suggestions on how to solve >> it? Or how to pinpoint the cause? >> >> Thanks, >> Martin Bergek >> > > I'm usign exactly the same chipset and never saw this problem before. > There might be something related to > https://github.com/raspberrypi/firmware/issues/553 which you can test > with a core_freq=250 > > As well it might be connected to the USB power supply if you use the 5V pin > from PL2303 to power up your board. I do and I experience no issue but > it might be something there. > > -- > Andrei Gherzan -- _______________________________________________ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
