On Fri, Jan 16, 2009 at 4:19 PM, Ian Lance Taylor <i...@google.com> wrote:
> Adam Nemet <ane...@caviumnetworks.com> writes:
>
>> struct s
>> {
>> char a:4;
>> char b:8;
>> char c:4;
>> } __attribute__ ((packed))
>>
>> is 3 bytes long because b gets pushed to the next byte boundary.
>
> Sounds like a bug.
>
>
>> The reason for this behavior is that finish_struct does not propagate packed
>> to fields whose type has alignment <= BITS_PER_UNIT:
>
> This was introduced for PR 21166. I think you're right that the
> alignment check should be ignored if the bitfield is a bitfield.
We should make sure to not introduce ABI changes (if the PR21166 change
didn't change it already) and properly document even this "fixing" of the
ABI, possibly emitting a default-on warning.
Richard.
