w1 driver for omap error -- reversed byte sequence in netlink slave message

w1 driver for omap error -- reversed byte sequence in netlink slave message

[Paul Alfille]

OWFS (one-wire filesystem -- www.owfs.org) seems to be the only user
of the w1 netlink messages. We use them to extend the kernel 1-wire
system from a simple temperature sensor to a complete 1-wire bus
master. Thus we've uncovered a bug -- byte reversal -- in one of the
sub-modules.

1-wire devices have a unique 64-bit address (including an 8-bit CRC)
that is reported by the netlink W1_CMD_SEARCH message. For the omap
bus master, and only the omap bus master, the slave addresses are
byte-order reversed. If the slave byte order is reversed in that case,
the CRC8 is then correct, the slave address then matches the actual
device address lettered on the device, and all other commands work
correctly.

The driver is in kernel/drivers/w1/masters/omap_hdq.c

Paul Alfille

Re: w1 driver for omap error -- reversed byte sequence in netlink slave message

[Andrew Morton]

On Thu, 21 Apr 2011 21:53:39 -0400 Paul Alfille <paul.alfille@gmail.com> wrote:

> OWFS (one-wire filesystem -- www.owfs.org) seems to be the only user
> of the w1 netlink messages. We use them to extend the kernel 1-wire
> system from a simple temperature sensor to a complete 1-wire bus
> master. Thus we've uncovered a bug -- byte reversal -- in one of the
> sub-modules.
> 
> 1-wire devices have a unique 64-bit address (including an 8-bit CRC)
> that is reported by the netlink W1_CMD_SEARCH message. For the omap
> bus master, and only the omap bus master, the slave addresses are
> byte-order reversed. If the slave byte order is reversed in that case,
> the CRC8 is then correct, the slave address then matches the actual
> device address lettered on the device, and all other commands work
> correctly.
> 
> The driver is in kernel/drivers/w1/masters/omap_hdq.c
> 

Have you fixed it?  Can you send us a tested patch to look at?

Thanks.