[Qemu-devel] [PATCH 0/2] Fix tests on recent gcc

[Juan Quintela <quintela@redhat.com>]

Hi

On Fedora 26 (gcc-7.1.1 and glib2 2.52.3) tests don't compile.

In file included from /usr/include/string.h:639:0,
                 from /mnt/kvm/qemu/cleanup/include/qemu/osdep.h:69,
                 from /mnt/kvm/qemu/cleanup/tests/test-iov.c:1:
In function ‘memcpy’,
    inlined from ‘iov_from_buf’ at /mnt/kvm/qemu/cleanup/include/qemu/iov.h:51:9,
    inlined from ‘test_to_from_buf_1’ at /mnt/kvm/qemu/cleanup/tests/test-iov.c:88:14,
    inlined from ‘test_to_from_buf’ at /mnt/kvm/qemu/cleanup/tests/test-iov.c:144:9:
/usr/include/bits/string3.h:53:10: error: ‘__builtin_memcpy’: specified size 18446744073709551615 exceeds maximum object size 9223372036854775807 [-Werror=stringop-overflow=]
   return __builtin___memcpy_chk (__dest, __src, __len, __bos0 (__dest));
          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In function ‘memcpy’,
    inlined from ‘iov_to_buf’ at /mnt/kvm/qemu/cleanup/include/qemu/iov.h:64:9,
    inlined from ‘test_to_from_buf_1’ at /mnt/kvm/qemu/cleanup/tests/test-iov.c:94:14,
    inlined from ‘test_to_from_buf’ at /mnt/kvm/qemu/cleanup/tests/test-iov.c:144:9:
/usr/include/bits/string3.h:53:10: error: ‘__builtin_memcpy’: specified size 18446744073709551615 exceeds maximum object size 9223372036854775807 [-Werror=stringop-overflow=]
   return __builtin___memcpy_chk (__dest, __src, __len, __bos0 (__dest));
          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors

In this case, we are abusing the functions and call a size_t argument
with -1, which gives us a very big number.  gcc gets overzealous and
gets confused about it.  Notice that this was introduced in 2015 by Paolo:

commit ad523bca56a7202d2498c550a41be5c986c4d33c
Author: Paolo Bonzini <pbonzini@redhat.com>
Date:   Tue Dec 22 12:03:33 2015 +0100

    iov: avoid memcpy for "simple" iov_from_buf/iov_to_buf

I fixed it by using the real sizes in the tests insntead of -1.  It is already calculated.




In file included from /usr/include/glib-2.0/glib/glist.h:32:0,
                 from /usr/include/glib-2.0/glib/ghash.h:33,
                 from /usr/include/glib-2.0/glib.h:50,
                 from /mnt/kvm/qemu/cleanup/tests/vmgenid-test.c:11:
In function ‘acpi_find_vgia’,
    inlined from ‘read_guid_from_memory’ at /mnt/kvm/qemu/cleanup/tests/vmgenid-test.c:103:18:
/usr/include/glib-2.0/glib/gmem.h:216:10: error: argument 1 range [18446744071562067968, 18446744073709551615] exceeds maximum object size 9223372036854775807 [-Werror=alloc-size-larger-than=]
      __p = g_##func##_n (__n, __s);   \
      ~~~~^~~~~~~~~~~~~~~~~~~~~~~~~
/usr/include/glib-2.0/glib/gmem.h:278:42: note: in expansion of macro ‘_G_NEW’
 #define g_new0(struct_type, n_structs)   _G_NEW (struct_type, n_structs, malloc0)
                                          ^~~~~~
/mnt/kvm/qemu/cleanup/tests/vmgenid-test.c:70:14: note: in expansion of macro ‘g_new0’
     tables = g_new0(uint32_t, tables_nr);
              ^~~~~~
/mnt/kvm/qemu/cleanup/tests/vmgenid-test.c: In function ‘read_guid_from_memory’:
/usr/include/glib-2.0/glib/gmem.h:96:10: note: in a call to allocation function ‘g_malloc0_n’ declared here
 gpointer g_malloc0_n      (gsize  n_blocks,
          ^~~~~~~~~~~
cc1: all warnings being treated as errors
make: *** [/mnt/kvm/qemu/cleanup/rules.mak:66: tests/vmgenid-test.o] Error 1


this cames form line:

    tables = g_new0(uint32_t, tables_nr);

glib/gcc gets completely confused about this and think that 1st
argument can be 2^64-1.  Documentation says that you should use g_new
for struct types, so ...  I moved to g_malloc0() and call it a day.

What do you think?

Later, Juan.






Juan Quintela (2):
  tests: Use real size for iov tests
  tests: Make acpid test compile

 tests/test-iov.c     | 8 ++++----
 tests/vmgenid-test.c | 2 +-
 2 files changed, 5 insertions(+), 5 deletions(-)

-- 
2.13.5