Re: [ob/pushed] dwarf2read.c: fix latent buglet

[Pedro Alves <palves@redhat.com>]

On 08/03/2015 07:06 PM, Doug Evans wrote:
> On Mon, Aug 3, 2015 at 11:00 AM, Pedro Alves <palves@redhat.com> wrote:
>> cust->includes is:
>>
>> struct compunit_symtab
>> {
>> ...
>>   struct compunit_symtab **includes;
>>
>> gdb/ChangeLog:
>> 2015-08-03  Pedro Alves  <palves@redhat.com>
>>
>>         * dwarf2read.c (compute_compunit_symtab_includes): Use size of struct
>>         compunit_symtab pointer.
>> ---
>>  gdb/dwarf2read.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/gdb/dwarf2read.c b/gdb/dwarf2read.c
>> index 24a4022..b5ffd04 100644
>> --- a/gdb/dwarf2read.c
>> +++ b/gdb/dwarf2read.c
>> @@ -7983,7 +7983,7 @@ compute_compunit_symtab_includes (struct dwarf2_per_cu_data *per_cu)
>>        len = VEC_length (compunit_symtab_ptr, result_symtabs);
>>        cust->includes
>>         = obstack_alloc (&dwarf2_per_objfile->objfile->objfile_obstack,
>> -                        (len + 1) * sizeof (struct symtab *));
>> +                        (len + 1) * sizeof (struct compunit_symtab *));
>>        for (ix = 0;
>>            VEC_iterate (compunit_symtab_ptr, result_symtabs, ix,
>>                         compunit_symtab_iter);
> 
> Bleah.
> Since sizeof (struct symtab) < sizeof (struct compunit_symtab) (64 vs
> 112 for amd64)

Yes, but that's not the case here -- this is 'sizeof (foo *)' not
'sizeof (foo)'.  So it's actually pretty harmless.  Should have called
that out explicitly, sorry.

This was caught in the C++ conversion, where the "insert-casts"
script would generate:

       cust->includes
-         = obstack_alloc (&dwarf2_per_objfile->objfile->objfile_obstack,
+         = (struct symtab **) obstack_alloc (&dwarf2_per_objfile->objfile->objfile_obstack,
                         (len + 1) * sizeof (struct symtab *));

which would then fail to compile.

Thanks,
Pedro Alves