Re: Help with Atomic_Components and whole array assignment

[kenner@lab.ultra.nyu.edu (Richard Kenner)]

In article <9570vu$fv0$1@nnrp1.deja.com> Robert Dewar <robert_dewar@my-deja.com> writes:
>The GNU/Linux kernel had endless problems from assumptions
>of this kind, and finally they have all been replaced by
>machine language insertions, to avoid the problems. What
>happened in this case is that new improved and correct gcc
>compilers blew the kernel out of the water by breaking these
>undocumented assumptions.

That's an oversimplification, unfortunately.  Most of the assumptions
were indeed in machine language insertions, but the assumption were
things that GCC would do *around* those insertions (such as register
choice).  GCC has powerful facilities for saying where operands are
found and such, but they weren't fully used since "it worked without them".
The change was to use the machine-code insertions properly.

One other thing that would sometimes be seen was going off the end of
an array purposely in order to reference a variable in the calling
program.