[Bug c++/91397] New: -Wstringop-overflow specified bound 18446744073709551615 exceeds maximum object size 9223372036854775807

[steinar+gcc at gunderson dot no]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91397

            Bug ID: 91397
           Summary: -Wstringop-overflow specified bound
                    18446744073709551615 exceeds maximum object size
                    9223372036854775807
           Product: gcc
           Version: 9.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: steinar+gcc at gunderson dot no
  Target Milestone: ---

Hi,

We saw this coming only with GCC 10 (probably different inlining decisions),
but the reduced test case also triggers with 9, so I'm going to file it against
9.1.0.

This test case is reduced from MySQL 8.0 (osfile0.cc):

#include <string.h>
#include <stdlib.h>

char *allocate(unsigned long g) {
        if (g == 0) return (char *)malloc(0);
        for (;;)        
                ;
} 

char *i;
char *j(long k) {
        char *l(allocate(k + 1));
        memcpy(l, i, k);
        return l;
} 

atum17:~> g++-9 -O2 -Wstringop-overflow -c os0file.cc
os0file.cc: In function ‘char* j(long int)’:
os0file.cc:12:9: warning: ‘void* memcpy(void*, const void*, size_t)’ specified
bound 18446744073709551615 exceeds maximum object size 9223372036854775807
[-Wstringop-overflow=]
   12 |   memcpy(l, i, k);
      |   ~~~~~~^~~~~~~~~

I don't even understand what the warning means (“specified bound” of what?),
but it looks spurious to me. Perhaps some signed/unsigned confusion?

Since the reduction is somewhat nonsensical with the empty infinite loop in
there, I'm also including an unreduced test case. With GCC 9.1.0 and -O2, it
gives the same warning.