[Bug middle-end/102687] New: [12 Regression] bootstrap ICE in insert_access, at ipa-modref-tree.h:582

[Bug middle-end/102687] New: [12 Regression] bootstrap ICE in insert_access, at ipa-modref-tree.h:582

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102687

            Bug ID: 102687
           Summary: [12 Regression] bootstrap ICE in insert_access, at
                    ipa-modref-tree.h:582
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
          Assignee: unassigned at gcc dot gnu.org
          Reporter: msebor at gcc dot gnu.org
  Target Milestone: ---

The current GCC 12 trunk fails to bootstrap for me with the ICE below.  Based
on git log I wonder if g:008e7397dad971c03c08fc1b0a4a98fddccaaed8 might be to
blame.

during IPA pass: modref
In file included from /ssd/test/src/gcc/trunk/gcc/cp/module.cc:20075:
./gt-cp-module.h: In member function ‘bool trees_in::core_vals(tree)’:
./gt-cp-module.h:420:1: internal compiler error: in insert_access, at
ipa-modref-tree.h:582
  420 | }
      | ^
0x1452a63 modref_ref_node<int>::insert_access(modref_access_node, unsigned
long, bool)
        /ssd/test/src/gcc/trunk/gcc/ipa-modref-tree.h:582
0x1450ef5 modref_tree<int>::insert(int, int, modref_access_node, bool)
        /ssd/test/src/gcc/trunk/gcc/ipa-modref-tree.h:853
0x143f5aa record_access
        /ssd/test/src/gcc/trunk/gcc/ipa-modref.c:719
0x1441012 analyze_store
        /ssd/test/src/gcc/trunk/gcc/ipa-modref.c:1251
0x134baa4 walk_stmt_load_store_addr_ops(gimple*, void*, bool (*)(gimple*,
tree_node*, tree_node*, void*), bool (*)(gimple*, tree_node*, tree_node*,
void*), bool (*)(gimple*, tree_node*, tree_node*, void*))
        /ssd/test/src/gcc/trunk/gcc/gimple-walk.c:767
0x134cc0b walk_stmt_load_store_ops(gimple*, void*, bool (*)(gimple*,
tree_node*, tree_node*, void*), bool (*)(gimple*, tree_node*, tree_node*,
void*))
        /ssd/test/src/gcc/trunk/gcc/gimple-walk.c:966
0x14410cd analyze_stmt
        /ssd/test/src/gcc/trunk/gcc/ipa-modref.c:1274
0x1443b25 analyze_function
        /ssd/test/src/gcc/trunk/gcc/ipa-modref.c:2138
0x1443f0e modref_generate
        /ssd/test/src/gcc/trunk/gcc/ipa-modref.c:2216
0x1644856 execute_ipa_summary_passes(ipa_opt_pass_d*)
        /ssd/test/src/gcc/trunk/gcc/passes.c:2248
0x10d2ccb ipa_passes
        /ssd/test/src/gcc/trunk/gcc/cgraphunit.c:2179
0x10d312f symbol_table::compile()
        /ssd/test/src/gcc/trunk/gcc/cgraphunit.c:2289
0x10d3730 symbol_table::finalize_compilation_unit()
        /ssd/test/src/gcc/trunk/gcc/cgraphunit.c:2537
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.

[Bug middle-end/102687] [12 Regression] bootstrap ICE in insert_access, at ipa-modref-tree.h:582

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102687

--- Comment #1 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Sorry, I accidentally commited an unrelated change I had in my tree.
I am testing
diff --git a/gcc/ipa-modref-tree.h b/gcc/ipa-modref-tree.h
index 52f225b1aae..9795e2b8405 100644
--- a/gcc/ipa-modref-tree.h
+++ b/gcc/ipa-modref-tree.h
@@ -148,8 +148,7 @@ struct GTY(()) modref_access_node
               poly_int64 offset1, poly_int64 size1, poly_int64 max_size1,
               bool record_adjustments)
     {
-      if (known_eq (parm_offset, parm_offset1)
-         && known_eq (offset, offset1)
+      if (known_eq (offset, offset1)
          && known_eq (size, size1)
          && known_eq (max_size, max_size1))
        return;
@@ -578,10 +577,6 @@ struct GTY((user)) modref_ref_node
          }
        (*accesses)[best1].forced_merge (best2 < 0 ? a : (*accesses)[best2],
                                         record_adjustments);
-       /* CHeck that merging indeed merged ranges.  */
-       gcc_checking_assert ((*accesses)[best1].contains (best2 < 0 ? a :
(*accesses)[best2]));
-       /*if (best2 >= 0)
-         accesses->unordered_remove (best2);*/
        if (!(*accesses)[best1].useful_p ())
          {
            collapse ();

[Bug middle-end/102687] [12 Regression] bootstrap ICE in insert_access, at ipa-modref-tree.h:582

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102687

--- Comment #2 from Martin Sebor <msebor at gcc dot gnu.org> ---
The patch gets rid of the ICE and I don't see any unexpected failures in the
test suite, thanks!

[Bug ipa/102687] [12 Regression] bootstrap ICE in insert_access, at ipa-modref-tree.h:582

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102687

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |marxin at gcc dot gnu.org
   Target Milestone|---                         |12.0
           Keywords|                            |ice-on-valid-code
          Component|middle-end                  |ipa

[Bug ipa/102687] [12 Regression] bootstrap ICE in insert_access, at ipa-modref-tree.h:582

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102687

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
https://gcc.gnu.org/pipermail/gcc-cvs/2021-October/354847.html

r12-4330