On 11/23/2015 08:53 AM, Pierre-Marie de Rodat wrote:
Do you think the other patches could make it before the branch? (if they
could, I will rebase+retest them as quick as possible).
Probably, yes. I can't find the DW_AT_static_link patch, though; it
doesn't seem to have been attached to your initial mail.
+ /* If we already met this node, there is nothing to compute anymore. */
+ if (visited.contains (l))
+ {
+#if ENABLE_CHECKING
+ /* Make sure that the stack size is consistent wherever the execution
+ flow comes from. */
+ gcc_assert ((unsigned) l->dw_loc_frame_offset == frame_offset_);
+#endif
+ break;
+ }
+ visited.add (l);
The 'add' function returns whether or not the set already contained the
entry, so you don't need to also call 'contains'.
+ /* The called DWARF procedure consumes one stack slot per argument
+ and returns one stack slot. */
+ tree func
+ = lookup_dwarf_proc_decl (l->dw_loc_oprnd1.v.val_die_ref.die);
+
+ frame_offset += 1;
+ for (tree args = DECL_ARGUMENTS (func);
+ args != NULL;
+ args = DECL_CHAIN (args))
+ frame_offset_--;
Can you avoid the new hash table by counting the
DW_TAG_formal_parameters instead of the DECL_ARGUMENTS?
Jason
