Re: [lto][patch] Remove after_inlining flag

["Rafael Espindola" <espindola at google dot com>]

> In principle I wouldn't mind removing it, but I would like to
> understand why exactly.

With a inliner running in cc1 and one running lto1 it doesn't make
sense have a notion of "after_inline". The problem I was having was
with the verifier being confused about a node throwing an inner
exception in the middle of the block. This was caused by the use in
except.c.

The tests just finished. Beth C and C++ are OK. I have also updated
the comments.

> Jan, any insights?
>
>
> Thanks.  Diego.
>

Cheers,
-- 
Rafael Avila de Espindola

Google | Gordon House | Barrow Street | Dublin 4 | Ireland
Registered in Dublin, Ireland | Registration Number: 368047

diff --git a/gcc/except.c b/gcc/except.c
index 77a3049..18058e7 100644
--- a/gcc/except.c
+++ b/gcc/except.c
@@ -2510,13 +2510,9 @@ reachable_next_level (struct eh_region *region, tree type_thrown,
       /* Here we end our search, since no exceptions may propagate.
 	 If we've touched down at some landing pad previous, then the
 	 explicit function call we generated may be used.  Otherwise
-	 the call is made by the runtime.
+	 the call is made by the runtime. */
 
-         Before inlining, do not perform this optimization.  We may
-	 inline a subroutine that contains handlers, and that will
-	 change the value of saw_any_handlers.  */
-
-      if ((info && info->saw_any_handlers) || !cfun->after_inlining)
+      if (info && info->saw_any_handlers)
 	{
 	  add_reachable_handler (info, region, region);
 	  return RNL_CAUGHT;
diff --git a/gcc/function.h b/gcc/function.h
index 16f2324..8f17c59 100644
--- a/gcc/function.h
+++ b/gcc/function.h
@@ -577,8 +577,6 @@ struct function GTY(())
      not desire this behavior.  */
   unsigned int dont_save_pending_sizes_p : 1;
 
-  unsigned int after_inlining : 1;
-
   /* Fields below this point are not set for abstract functions; see
      allocate_struct_function.  */
 
diff --git a/gcc/tree-inline.c b/gcc/tree-inline.c
index 69f0c80..2899f87 100644
--- a/gcc/tree-inline.c
+++ b/gcc/tree-inline.c
@@ -1763,7 +1763,6 @@ initialize_cfun (tree new_fndecl, tree callee_fndecl, gcov_type count,
   cfun->has_nonlocal_label = src_cfun->has_nonlocal_label;
   cfun->stdarg = src_cfun->stdarg;
   cfun->dont_save_pending_sizes_p = src_cfun->dont_save_pending_sizes_p;
-  cfun->after_inlining = src_cfun->after_inlining;
   cfun->returns_struct = src_cfun->returns_struct;
   cfun->returns_pcc_struct = src_cfun->returns_pcc_struct;
   cfun->after_tree_profile = src_cfun->after_tree_profile;
@@ -3241,8 +3240,6 @@ expand_call_inline (basic_block bb, gimple stmt, copy_body_data *id)
   id->src_cfun = DECL_STRUCT_FUNCTION (fn);
   id->gimple_call = stmt;
 
-  gcc_assert (!id->src_cfun->after_inlining);
-
   id->entry_bb = bb;
   if (lookup_attribute ("cold", DECL_ATTRIBUTES (fn)))
     {
diff --git a/gcc/tree-optimize.c b/gcc/tree-optimize.c
index ba3d0fc..f2e1e3a 100644
--- a/gcc/tree-optimize.c
+++ b/gcc/tree-optimize.c
@@ -292,8 +292,6 @@ execute_fixup_cfg (void)
   gimple_stmt_iterator gsi;
   int todo = gimple_in_ssa_p (cfun) ? TODO_verify_ssa : 0;
 
-  cfun->after_inlining = true;
-
   if (cfun->eh)
     FOR_EACH_BB (bb)
       {
diff --git a/gcc/tree-ssa-live.c b/gcc/tree-ssa-live.c
index 4731518..927d737 100644
--- a/gcc/tree-ssa-live.c
+++ b/gcc/tree-ssa-live.c
@@ -508,11 +508,7 @@ remove_unused_scope_block_p (tree scope)
 	 only the used variables for cfgexpand's memory packing saving quite
 	 a lot of memory.  */
       else if (debug_info_level == DINFO_LEVEL_NORMAL
-	       || debug_info_level == DINFO_LEVEL_VERBOSE
-	       /* Removing declarations before inlining is going to affect
-		  DECL_UID that in turn is going to affect hashtables and
-		  code generation.  */
-	       || !cfun->after_inlining)
+	       || debug_info_level == DINFO_LEVEL_VERBOSE)
 	unused = false;
 
       else