https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104560

            Bug ID: 104560
           Summary: False positive from -Wanalyzer-free-of-non-heap seen
                    with rdma-core
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: analyzer
          Assignee: dmalcolm at gcc dot gnu.org
          Reporter: dmalcolm at gcc dot gnu.org
  Target Milestone: ---

Created attachment 52450
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52450&action=edit
Reduced reproducer

I'm seeing various -Wanalyzer-free-of-non-heap when compiling rdma-core with
-fanalyzer, e.g.:

rdma-core-33.0/providers/mlx5/mlx5.c: scope_hint: In function
'mlx5_uninit_device'
rdma-core-33.0/providers/mlx5/mlx5.c:2101:9:
warning[-Wanalyzer-free-of-non-heap]: 'free' of '&*verbs_device.device' which
points to memory not on the heap
# 2099|         struct mlx5_device *dev = to_mdev(&verbs_device->device);
# 2100|   
# 2101|->       free(dev);
# 2102|   }
# 2103|   

Am attaching a reproducer, which triggers the current false positive with trunk
when compiled without optimization:

<source>: In function 'mlx5_uninit_device':
<source>:42:9: warning: 'free' of 'dev' which points to memory not on the heap
[CWE-590] [-Wanalyzer-free-of-non-heap]
   42 |         __builtin_free(dev);
      |         ^~~~~~~~~~~~~~~~~~~
  'mlx5_uninit_device': events 1-3
    |
    |   40 |         struct mlx5_device *dev = to_mdev(&verbs_device->device);
    |      |                                   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    |      |                                   |
    |      |                                   (1) pointer is from here
    |      |                                   (2) pointer is from here
    |   41 | 
    |   42 |         __builtin_free(dev);
    |      |         ~~~~~~~~~~~~~~~~~~~        
    |      |         |
    |      |         (3) call to 'free' here
    |

Uploaded to Compiler Explorer as:
  https://godbolt.org/z/nazc15z1h

Reply via email to