Re: [PATCH 0/3] misc fixes for eCos 3.0

[Andreas Oberritter <obi@saftware.de>]

Gary Thomas wrote:
> Not much use without some explanation of what they do.
> Accepted practice is to explain each change, so it can be
> properly understood and discussed.

I thought that the reason for these changes were pretty obvious. I'll
try to explain them now, though:

[PATCH 1/3] usbs_at91: call vendor_control_fn on USB_DEVREQ_TYPE_VENDOR:

For each type of USB control requests (USB_DEVREQ_TYPE_XXX), there is
a callback function called xxx_control_fn(). For vendor requests, the
wrong function was called. It looks like a copy and paste bug. The
function class_control_fn() should only be called for
USB_DEVREQ_TYPE_CLASS.

[PATCH 2/3] usbseth: compile fix: move misplaced #endif:

Compilation aborted, because HAL_DCACHE_LINE_SIZE was not defined. A
look at the code revealed, that the following code does not make sense:

#ifndef XXX
x = expr;
#else
#endif
x = expr using XXX;

[PATCH 3/3] tcpip: compile fix: remove unneeded cast to long in
TCPT_RANGESET:

Compilation aborted using the toolchain supplied by the eCos installer
(gcc 4.3.2), because of an invalid cast:

error: "lvalue required as left operand of assignment"

Therefore, TCPT_RANGESET was adjusted in two places to match the usage
pattern of other users of this macro, which simply omit the cast.

Regards,
Andreas