Re: [lto][patch] Small fixes for DECL_ASSEMBLE_NAME computation

[Diego Novillo <dnovillo at google dot com>]

On Wed, Jan 14, 2009 at 14:09, Rafael Espindola <espindola@google.com> wrote:

> This patches avoids most of the problem by replacing
> lang_hooks.set_decl_assembler_name with lhd_set_decl_assembler_name.
> That way we can be sure that any computation of the assembler name
> from this point on will not use language specific information. The
> patch also has two small fixes

This part is fine but it needs a test case.


> *) DECL_IGNORED_P is a debug info thing. We still need to compute
> assembly names of DECL_IGNORE_P decls.

Why?  What is this fixing?

> *) compute the name of every decl in the call graph. We write them to
> disk, so we need the names for the section names (at least).

But these should be caught by the traversal of decl_for_uid_map,
why aren't they?

It's not clear to me what you are fixing with these two.  I would
rather have a test case for all these changes.  If not, it's hard
to see the rationale behind them.


Diego.