> Hi -
> fixing a -Wshadow warning I found that the code writing bytes/shorts
> to sparse address space looks strange.

Yes, it does indeed.

> Possibly, it has never worked correctly.

That's quite possible.  Another strangeness that your patch does not
address is that `off' and `shift' ends up with the same value:

                shift =3D off & 0x3;
                off &=3D 0x3;

as Chris replied to me when I asked him why the (former) 'v'
variable was computed and the result simply thrown away.  Chris
suggested that the last should possibly have been 'off &=3D ~0x3'.

> The appended patch makes the code work as appearently intended.
> Is anyone able to test this, or has some documentation to verify?

Sorry, I don't have a turbochannel alpha.  Chris wanted to know if the
code in this branch of tc_mem_write_1() and tc_mem_write_2() ever gets
executed.

Regards,

- H=E5vard