On Wed, Jan 11, 2017 at 9:39 AM, Oleh Kravchenko <oleg@xxxxxxxxxx> wrote: > Hello! > > So, I tried to split i2c send messages to 5, 8, 16 bytes - that cause error: > cx231xx 3-1:1.1: cx231xx_send_usb_command: failed with status --32 > > This problem occurred only with Si2158 tuner, > demodulator Si2168 eats up to 15 bytes per i2c message. > > I think problem in cx231xx_send_vendor_cmd() function. > It's split reading, but not writing. > > PS: I use USB sniffer to capture traffic, please find it in attachment. Do you have a logic analyzer so you can inspect the I2C bus state? I'm speaking as someone who's done significant reverse engineering projects on many pieces of hardware, you'll get a much better understanding of platform specific I2C problems if you can see the bus directly. Any such issues tend to propagate upwards through micro-controller interfaces (cx231xx) and its much more obvious to deal with the controller problem when the state of the bus is absolutely known. I have no objections to a I2C bus controller implementation splitting reads and writes in a uniform and generic way, assuming windows is doing the same thing, providing that approach applies to all attached devices. We don't want board specific hacks infecting the various drivers......... unless we're absolutely clear as to why...... and we don't want to see I2C controller limitations infecting I2C demod/tuner drivers where at all possible. If its true that send_vendor_cmd() splits reads but not writes, fix the function to be uniform for all boards. Don't do boards specific overrides. -- Steven Toth - Kernel Labs http://www.kernellabs.com -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
