Hi Rafal,
thanks for the review.
Rafal Somla, 21.05.2009 11:36:
...
> Ingo Struewing wrote:
...
>> + if (m_errno == ER_DB_DROP_RMDIR)
>> + {
>> + size_t len= get_name()->length();
>> + char oldname[FN_REFLEN];
>> + char newname[FN_REFLEN];
>
> [2] I think newname buffer should have size FN_REFLEN +
> MY_UUID_STRING_LENGTH + 1 to satisfy my_uuid2str() requirements. This is
> because oldname can take up to FN_REFLEN bytes and then there should
> still be MY_UUID_STRING_LENGTH + 1 bytes left for the uuid.
...
>> + if (len + 3 + MY_UUID_STRING_LENGTH >= FN_REFLEN)
>> + goto err; /* purecov: inspected */
And then this "if" can go away, right?
...
>> + /* Build new name as old + uuid. */
>> + memcpy(newname, oldname, len);
>> + newname[len++]= '-';
>> + my_uuid(uuid);
>> + my_uuid2str(uuid, newname + len);
>
> [3] Here, there should be at least MY_UUID_STRING_LENGTH + 1 bytes left
> at newname + len.
The "if" above ensured that anyway. But yes, without the extended
buffer, there could be cases, where DROP would fail because the new name
could not be created. A longer buffer can avoid that.
Regards
Ingo
--
Ingo Strüwing, Database Group
Sun Microsystems GmbH, Sonnenallee 1, D-85551 Kirchheim-Heimstetten
Geschäftsführer: Thomas Schröder, Wolfgang Engels, Wolf Frenkel
Vorsitzender des Aufsichtsrates: Martin Häring HRB München 161028
