Alignment problems

[David Thompson]

Yes, some compilers have a pack directive, but it doesn't always work.

For example, on SPARC, attempting to access a misaligned object causes
the program to crash. Here's a program that uses gcc's
__attribute__((packed)) extension: <snip>
If you refer to the misaligned member directly, so the compiler
knows that it's misaligned, it can access it correctly. If you've
saved a pointer to the misaligned member, there's no way for the
compiler to know that it's misaligned.

Or nonstandardly tag the pointers. That's what then-Tandem TNS/R did
for a (MIPS) 32bit reimplementation of a former custom 16bit ISA.
To (try to) futureproof they actually jumped to 64bit. If you indirect
through a 'refaligned 2' pointer the compiler uses 2 (or 4) 16bit
accesses; for 'refaligned 8' it does one 32bit (or 64bit) which faults
if misaligned (although IIRC they also offered a fixup handler, with
as noted elsethread performance degradation).

They used a pragma not a language qualifier, so it wasn't enforced or
verified across caller/callee, unfortunately.

Now-Tanpaqard changed further to Itanium, and I'm not up to date on
what they've done there.