Re: [PATCH] Don't fold away division by zero (PR middle-end/66127)

[Steve Ellcey <sellcey at imgtec dot com>]

Marek,

This may be unrelated to your patches for PR 66127 and 66066 but I am
having a new failure when building the latest glibc with the latest GCC.

I haven't yet tracked down exactly which patch caused the problem.  Included
is a cutdown test case from libio/memstream.c in glibc that results in a
strict-aliasing error.  Is this is something you already know about or have
seen?

In the mean time I will try to figure out exactly which patch caused this error
to trigger.

Steve Ellcey
sellcey@imgtec.com


typedef unsigned int size_t;
extern void *malloc (size_t __size) __attribute__ ((__nothrow__ )) __attribute__
 ((__malloc__)) ;

struct _IO_FILE_plus { void *vtable; };
void *_IO_mem_jumps;

struct _IO_streambuf { };

typedef struct _IO_strfile_
{
  struct _IO_streambuf _sbf;
} _IO_strfile;

struct _IO_FILE_memstream
{
  _IO_strfile _sf;
};

void open_memstream (int bufloc, int sizeloc)
{
  struct locked_FILE
  {
    struct _IO_FILE_memstream fp;
  } *new_f;
  new_f = (struct locked_FILE *) malloc (sizeof (struct locked_FILE));
  ((struct _IO_FILE_plus *) &new_f->fp._sf._sbf)->vtable = &_IO_mem_jumps;
}


x.c: In function 'open_memstream':
x.c:28:12: error: dereferencing type-punned pointer will break strict-aliasing rules [-Werror=strict-aliasing]
   ((struct _IO_FILE_plus *) &new_f->fp._sf._sbf)->vtable = &_IO_mem_jumps;
            ^
cc1: all warnings being treated as errors