Re: [PATCH ARM]: PR67745: Fix function alignment after __attribute__ 2/2

[Christian Bruel <christian dot bruel at st dot com>]

On 10/07/2015 07:50 PM, Bernd Schmidt wrote:
> On 10/07/2015 07:37 PM, Bernd Schmidt wrote:
> > On 10/07/2015 12:45 PM, Christian Bruel wrote:
>
> > > On 10/07/2015 12:18 PM, Bernd Schmidt wrote:
> > > > On 10/07/2015 09:04 AM, Christian Bruel wrote:
> > > > > +      /* Similarly, relayout function's alignment if not forced.  */
> > > > > +      if (!DECL_USER_ALIGN (fndecl)
> > > > > +          && (TREE_CODE (fntype) != METHOD_TYPE
> > > > > +          || TARGET_PTRMEMFUNC_VBIT_LOCATION !=
> > > > > ptrmemfunc_vbit_in_pfn))
> > > > > +        DECL_ALIGN (fndecl) = FUNCTION_BOUNDARY;
> > > > >         }
>
> Hmm, more questions are coming to mind - are other places that set
> DECL_ALIGN for functions redundant after this change? It looks like the
> main place where it's set is make_node_stat where it's also set to
> FUNCTION_BOUNDARY. That's small enough not to bother changing it,

Yes it is. I thought about poisoning it but make_node_stat is used for 
synthesized functions as well and we need to keep the default correct.

> but it
> raises another issue: we don't allocate a struct function for mere
> declarations, so does that mean they can have an incorrect DECL_ALIGN?

Probably at the time of start_decl, because DECL_ALIGN will have the 
boundary given by the global target_flags at that time. But this 
shouldn't be a problem since what matters is the DECL_ALIGN recomputed 
with the definition when there is something to layout.