[patch] Document the restriction on removing a global from a COMDAT

[Rafael Espíndola]

The attached patch adds a note about why optimizations cannot drop
globlas from COMDATs.

Since this is the language ref I made the statement fairly generic, in
case there is some corner case where an optimization can figure out it
is actually safe to modify the symbols in a COMDAT.  Let me know if
you would prefer a more strict statement.

Cheers,
Rafael
-------------- next part --------------
A non-text attachment was scrubbed...
Name: t.patch
Type: text/x-patch
Size: 837 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141008/3d6914f1/attachment.bin>