[PATCH] D18433: [IFUNC] Introduce GlobalIndirectSymbol as a base class for aliases and ifuncs

[Dmitry Polukhin via llvm-commits]

On Wed, Mar 30, 2016 at 8:33 PM, Duncan P. N. Exon Smith <
dexonsmith at apple.com> wrote:

> > +  /// These methods set and retrieve indirect symbol.
> > +  void setIndirectSymbol(Constant *Symbol) {
> > +    setOperand(0, Symbol);
> > +  }
>
> Should setIndirectSymbol() be protected?  I feel like the IFunc
> version might want to add some extra checks about the type of
> symbol passed in, or constrain the type somehow.
>
> Although maybe those checks would fit better in the verifier?
>

Both approaches have sense to me. Having setIndirectSymbol public gives an
ability to use GlobalIndirectSymbol polymorphic (for example, it is used in
BitcodeReader in ifunc patch). But GlobalAlias does have assert in
setAliasee for types, everything else is checked in verifier. I would
prefer to keep setIndirectSymbol public because setOperand is public so it
is possible to skip setAliasee check anyway. But if you think that it
should be private, I have no objections. please let me know what do you
prefer.

Duncan and Eric, thank you for review!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160331/86e72dbc/attachment.html>