Fundamental defect of the concept of shared libraries

[Masataka Ohta]

In article <1991May22.063425.26144 at kithrup.COM>
	sef at kithrup.COM (Sean Eric Fagan) writes:

>>>Are there any architectures of interest in this discussion that can't
>>>support PC-relative references?
>>R3000.
>
>	_foo:
>		<entrycode>
>		jal	foo1$
>	foo1$:
>		addu	$at, $31, 0
>
>Guess what:  we can now do PC-relative references.  And this isn't even 
>the most efficient way to do it.

You poor boy, such an old trick is already known to me. I sometimes use
the trick if it is possible.

The problem here is that "jal" is not PC-relative.

You had better write PC-relative (not actually) reference as

<		la	$at,foo1$
<	foo1$:

It is as PC-relative as your trick, but is called immediate addressing.

Of course, both of above are unusable in PIC.

>Score one for our side.

Sigh.

						Masataka Ohta