Re: [PATCH] gold/arm: define $a/$d markers in .plt

[David Miller <davem at davemloft dot net>]

From: Ian Lance Taylor <iant@google.com>
Date: Wed, 25 Apr 2012 12:44:12 -0700

> This seems to make sense in general, in that at runtime one shared
> library can not refer to a hidden or internal symbol defined in a
> different shared library.  I can see how it matters for SPARC, but it
> still seems more or less like the right thing to do.

Ok, thanks for all the info.

I will say in passing that this made things confusing for me from a
gold target's perspective.  It seemd to me that the whole point of
having seperate global and local Scan methods is that relocations are
processed differently for STB_LOCAL vs. STB_GLOBAL symbols.

So, from this perspective, this forced local case adds at least a
small wrinkle.  Imagine my surprise while debugging that a gsym seen
in global relocation scanning ended up STB_LOCAL in the final symbol
table :-)

Anyways, thanks again Ian.