Messages in this thread | | | Date | Wed, 27 Aug 2008 14:19:31 -0700 | From | Jeremy Fitzhardinge <> | Subject | Re: [patch] Add helper macros for little-endian bitfields |
| |
David Vrabel wrote: > Al Viro wrote: > >> On Sun, Aug 24, 2008 at 06:37:43PM -0700, Roland Dreier wrote: >> >>> > + * NOTE: When using multibyte bitfields, you need to convert the data >>> > + * from Little Endian to CPU before you can access the bitfield >>> > + * (to make it simpler): >>> >> NOTE: When tempted to use multibyte bitfields on fixed-layout data, you >> need to look in the mirror, ask yourself "what will they do to me during >> code review for that?", shudder and decide that some temptations are >> just not worth the pain. >> > > But why is this worthy of a crispy flaming? I've not seen anything > definite beyond a somewhat vague 'some compilers don't optimize > bitfields very well'. > > The structure definition and the DECL_BF_LEx() macros might be ugly but > the code using the structures is clearer. For example, > > get_random_bytes(&tiebreaker, sizeof(unsigned)); > drp_ie->tiebreaker = tiebreaker & 1; > > versus > > get_random_bytes(&tiebreaker, sizeof(unsigned)); > drp_ie->drp_control |= (tiebreaker & 1) > ? UWB_DRP_IE_DRP_CTRL_TIEBREAKER : 0; >
Why not just
drp_ie->drp_control |= tiebreaker & UWB_DRP_IE_DRP_CTRL_TIEBREAKER;
then? Doesn't matter which random bit you pick, does it?
J
| |