On Tue, 20 Mar 2012, Tristan Gingold wrote:
> On Mar 20, 2012, at 1:21 PM, Richard Guenther wrote:
> > On Tue, 20 Mar 2012, Tristan Gingold wrote:
> >> On Mar 15, 2012, at 10:37 AM, Richard Guenther wrote:
> >>> On Wed, 14 Mar 2012, Tristan Gingold wrote:
> >> […]
> >>> Well. To make this work in LTO the "main" function (thus, the program
> >>> entry point) should be marked at cgraph level and all users of
> >>> MAIN_NAME_P should instead check a flag on the cgraph node.
> >>>> Will write a predicate in tree.[ch].
> >>> Please instead transition "main-ness" to the graph.
> >> Hi,
> >> here is the patch I wrote. Does it match what you had in mind ?
> > Basically yes. Comments below.
> >> main_identifier_node is now set in tree.c
> > Looks good, hopefully my review-grep was as good as yours ;)
> >> diff --git a/gcc/cfgexpand.c b/gcc/cfgexpand.c
> >> index bd21169..7a7a774 100644
> >> --- a/gcc/cfgexpand.c
> >> +++ b/gcc/cfgexpand.c
> >> @@ -4513,9 +4513,8 @@ gimple_expand_cfg (void)
> >> /* If this function is `main', emit a call to `__main'
> >> to run global initializers, etc. */
> >> - if (DECL_NAME (current_function_decl)
> >> - && MAIN_NAME_P (DECL_NAME (current_function_decl))
> >> - && DECL_FILE_SCOPE_P (current_function_decl))
> >> + if (DECL_FILE_SCOPE_P (current_function_decl)
> >> + && cgraph_main_function_p (cgraph_get_node (current_function_decl)))
> >> expand_main_function ();
> > The DECL_FILE_SCOPE_P check is redundant, please remove them everywhere
> > you call cgraph_main_function_p. I suppose returning false if the
> > cgraph node is NULL in cgraph_main_function_p would be good.
> Ok. (I added the DECL_FILE_SCOPE_P check to avoid the cgraph lookup for
> speed reason)
> >> diff --git a/gcc/cgraphunit.c b/gcc/cgraphunit.c
> >> index 516f187..4a59f63 100644
> >> --- a/gcc/cgraphunit.c
> >> +++ b/gcc/cgraphunit.c
> >> @@ -346,6 +346,10 @@ cgraph_finalize_function (tree decl, bool nested)
> >> notice_global_symbol (decl);
> >> node->local.finalized = true;
> >> node->lowered = DECL_STRUCT_FUNCTION (decl)->cfg != NULL;
> >> + node->local.main_function =
> >> + DECL_FILE_SCOPE_P (decl)
> >> + && ((!DECL_ASSEMBLER_NAME_SET_P (decl) && MAIN_NAME_P (DECL_NAME
> >> (decl)))
> >> + || decl_assembler_name_equal (decl, main_identifier_node));
> > If we finalize a function we should always create an assembler name,
> > thus I'd change the above to
> > node->local.main_function = decl_assembler_name_equal (decl,
> > main_identifier_node);
> Indeed. At worst, the assembler name is created during the call to
> > btw, decl_assembler_name_equal doesn't seem to remove target-specific
> > mangling - do some OSes "mangle" main differently (I'm thinking of
> > leading underscores or complete renames)? Thus, I guess the
> > targets might want to be able to provide the main_identifier_assember_name
> > you use here.
> I think this is currently OK because decl_assembler_name_equal deals
> with leading underscore correctly. I have checked that on Darwin,
> which has a leading underscore.
> The only target that mangle names is i386 cygwin/mingw, which 'annotates'
> stdcall and fastcall function, but main() is regular.
> But I agree this mechanism is fragile.
> In order to make this mechanism stronger, we could add main_function_node,
> designates the FUNCTION_DECL that is the main function (if not NULL_TREE),
> a fallback on main_identifier_node for regular languages such as C or C++.
I'd rather get away from using a global main_identifier_node, instead
make that frontend specific, and introduce targetm.main_assembler_name
which the assembler-name creating langhook would make sure to use
when mangling what the FE thinks main is. main_identifier_node should
not serve any purpose outside of Frontends.
But I see both as a possible cleanup opportunity, not a necessary change.