Re: [PATCH] [REPOST] enforce host byte order on

[Guy Harris <guy () alum mit edu>]

Abeni Paolo wrote:

> The patch contained in the previous post does conatain an offset
> error  while accessing the linux usb header struct. The one attached
> here should fix the issue. Please give it a review, since this sort
> of thing is a bit tricky.

Checked in.

> I was tempted by implementing and using an OFFSETOF() macro like:
>
> #define OFFSETOF(TYPE, MEMBER) (((size_t) &((TYPE *)0)->MEMBER) + sizeof(((TYPE *)0)->MEMBER))
>
> It will help to avoid the error I incurred in (substituting
> hard-coded  offset values, with compiler generated ones), but I feared some
> portabilty issue across different compiler. What do you think?

I suspect most if not all platforms on which libpcap/WinPcap is built 
support ANSI C, so offsetof() is available.

The structure itself (the part that precedes the pcap_usb_setup 
structure) appears to put all values on their natural boundaries, so it 
should be the same on 32-bit and 64-bit platforms (and its length is a 
multiple of 8, so compilers such as, I think, some ARM GCC versions that 
pad structures to a multiple of 4 bytes won't pad it), so it should be 
OK on all the compilers we use.
-
This is the tcpdump-workers list.
Visit https://cod.sandelman.ca/ to unsubscribe.