[U-Boot] IP_t should be a "packed" struct
Ben Warren
biggerbadderben at gmail.com
Wed Jan 28 22:52:15 CET 2009
Wolfgang Denk wrote:
> Dear Ben Warren,
>
> In message <4980CC59.1070302 at gmail.com> you wrote:
>
>>> My idea should be to declare a define like this
>>>
>>> #define PKT_HEADER __attribute__((__packed__))
>>>
>>> my 2EuroCents.
>>>
>>> best regards,
>>>
>>> luigi
>>>
>>>
>>>
>> OK, sounds good. Send a patch please.
>>
>
> Hm... and what does this give us?
>
> How long until sombebody detects that all the data structures
> describing device register layout or so many processors and chips
> don't use any __packed__ either? Except for some (IMHO broken, but I
> know that I'm more attached to PowerPC anyway) ARM systems it is
> sufficient to use a "sane" selection of data types.
>
>
>
Well, yeah, I alluded to that in my other e-mail. I'm not crazy about
this either, but don't see any significant downside. Obviously if this
was a common problem, it would have surfaced a long time ago...
> I vote against changing this code. Just for a quick cross-check: do
> the corresponding data structs in the Linux kernel use any
> __packed__?
>
> Here is for example a copy of /usr/include/netinet/ip.h :
>
> ...
> struct iphdr
> {
> #if __BYTE_ORDER == __LITTLE_ENDIAN
> unsigned int ihl:4;
> unsigned int version:4;
> #elif __BYTE_ORDER == __BIG_ENDIAN
> unsigned int version:4;
> unsigned int ihl:4;
> #else
> # error "Please fix <bits/endian.h>"
> #endif
> u_int8_t tos;
> u_int16_t tot_len;
> u_int16_t id;
> u_int16_t frag_off;
> u_int8_t ttl;
> u_int8_t protocol;
> u_int16_t check;
> u_int32_t saddr;
> u_int32_t daddr;
> /*The options start here. */
> };
> ...
>
> Do you see any __packed__?
>
>
Yeah, I made the same observation, but am not fluent enough in the black
magic of Linux header files to know if packing was being enforced
somewhere else or in some other way that my little brain can't comprehend.
> Best regards,
>
> Wolfgang Denk
>
>
regards,
Ben
More information about the U-Boot
mailing list