On Thu, Oct 05, 2017 at 02:41:33PM +0200, Wolfram Sang wrote:
> On Mon, Oct 02, 2017 at 02:45:19PM +0200, Pontus Andersson wrote:
> > Commit b6c159a9cb69 ("i2c: ismt: Don't duplicate the receive length for
> > block reads") broke I2C block reads. It aimed to fix normal SMBus block
> > read, but changed the correct behavior of I2C block read in the process.
> >
> > According to Documentation/i2c/smbus-protocol, one vital difference
> > between normal SMBus block read and I2C block read is that there is no
> > byte count prefixed in the data sent on the wire:
> >
> > SMBus Block Read: i2c_smbus_read_block_data()
> > S Addr Wr [A] Comm [A]
> > S Addr Rd [A] [Count] A [Data] A [Data] A ... A [Data] NA P
> >
> > I2C Block Read: i2c_smbus_read_i2c_block_data()
> > S Addr Wr [A] Comm [A]
> > S Addr Rd [A] [Data] A [Data] A ... A [Data] NA P
> >
> > Therefore the two transaction types need to be processed differently in
> > the driver by copying of the dma_buffer as done previously for the
> > I2C_SMBUS_I2C_BLOCK_DATA case.
> >
> > Fixes: b6c159a9cb69 ("i2c: ismt: Don't duplicate the receive length for block reads")
> > Signed-off-by: Pontus Andersson <epontan@xxxxxxxxx>
> > Cc: stable@xxxxxxxxxxxxxxx
>
> Applied to for-current, thanks!
Thank you! Is it safe to asume it will be backported to 4.4 stable
branch and upwards as the problematic commit did (once in mainline)?
> Might have been good to CC the patch author of the problematic commit,
> too.
Yes, of course! Stephen is CC now at least (kept the commit message
quoted at his convenience).
