Phil Endecott writes: > Andrew Haley wrote: > > gcc should always generate correct code for accessing unaligned data. > > but Paul Brook wrote: > > You can't reliably take the address of a member of a packed > > structure (gcc doesn't have unaligned pointers). > > Are you both right? At first your comments look contradictory, but > my slow brain is starting to understand now. > > I think you're saying that I can declare something that is > unaligned: > > char c; > struct foo f; // unaligned > > and then operate on it locally: > > f.a++; > > because gcc knows that f.a is unaligned and on a machine that > doesn't do unaligned accesses it can generate appropriate > byte-shuffling code. But if I take its address: > > func(&f.a); > > it won't work, because func() assumes that the pointer it is passed > is aligned. > > Is this correct? A quick test seems to indicate that this is indeed > what happens. Sounds right. A member of a packed struct must be accessed in there, where it is unaligned. > But I don't get any error or warning when I do this. Would you > agree that an error (or at least a "big fat warning") would be > appropriate at the point where I wrote &f.a? Not necessarily. It's perfectly appropriate to do something like: memcpy (&f.a, foo, sizeof f.a); And we don't want to generate bazillions of warnings for that. Andrew.
