Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-5312

Conditional jump or move depends on uninitialised value(s) in ssl3_read_bytes

        Details

        • Type: Bug
        • Status: Closed
        • Priority: Major
        • Resolution: Cannot Reproduce
        • Affects Version/s: 10.0.5
        • Fix Version/s: 10.0.11
        • Component/s: None
        • Labels:
          None

          Description

          When MariaDB is built with the external openssl library

          cmake . -DCMAKE_INSTALL_PREFIX=/opt/mariadb-10.0 -DCMAKE_BUILD_TYPE=Debug -DWITH_VALGRIND=1

          some --valgrind test fail with the following output:

          ./mtr --valgrind ssl_cipher
...
==7154== Conditional jump or move depends on uninitialised value(s)
==7154==    at 0x5DC1B3: read_client_connect_attrs(char**, char*, charset_info_st const*) (sql_acl.cc:11144)
==7154==    by 0x5DD1F2: parse_client_handshake_packet(MPVIO_EXT*, unsigned char**, unsigned long) (sql_acl.cc:11527)
==7154==    by 0x5DD7D8: server_mpvio_read_packet(st_plugin_vio*, unsigned char**) (sql_acl.cc:11701)
==7154==    by 0x5DED61: native_password_authenticate(st_plugin_vio*, st_mysql_server_auth_info*) (sql_acl.cc:12272)
==7154==    by 0x5DDE31: do_auth_once(THD*, st_mysql_lex_string const*, MPVIO_EXT*) (sql_acl.cc:11869)
==7154==    by 0x5DE168: acl_authenticate(THD*, unsigned int, unsigned int) (sql_acl.cc:11971)
==7154==    by 0x7601E0: check_connection(THD*) (sql_connect.cc:1070)
==7154==    by 0x760353: login_connection(THD*) (sql_connect.cc:1138)
==7154==    by 0x7608D0: thd_prepare_connection(THD*) (sql_connect.cc:1301)
==7154==    by 0x760B5D: do_handle_one_connection(THD*) (sql_connect.cc:1370)
==7154==    by 0x7608A5: handle_one_connection (sql_connect.cc:1293)
==7154==    by 0xD3B682: pfs_spawn_thread (pfs.cc:1853)
==7154==    by 0x37C7807D14: start_thread (in /usr/lib64/libpthread-2.16.so)
==7154==    by 0x37C70F253C: clone (in /usr/lib64/libc-2.16.so)
==7154== Conditional jump or move depends on uninitialised value(s)
==7154==    at 0x37D1C296B2: ssl3_read_bytes (in /usr/lib64/libssl.so.1.0.1e)
==7154==    by 0x37D1C26D11: ??? (in /usr/lib64/libssl.so.1.0.1e)
==7154==    by 0xE6B296: vio_ssl_read (viossl.c:171)
==7154==    by 0x598E38: my_real_read(st_net*, unsigned long*) (net_serv.cc:870)
==7154==    by 0x59955C: my_net_read (net_serv.cc:1072)
==7154==    by 0x643936: do_command(THD*) (sql_parse.cc:925)
==7154==    by 0x760B7F: do_handle_one_connection(THD*) (sql_connect.cc:1379)
==7154==    by 0x7608A5: handle_one_connection (sql_connect.cc:1293)
==7154==    by 0xD3B682: pfs_spawn_thread (pfs.cc:1853)
==7154==    by 0x37C7807D14: start_thread (in /usr/lib64/libpthread-2.16.so)
==7154==    by 0x37C70F253C: clone (in /usr/lib64/libc-2.16.so)

          The same tests do not fail when built with the bundled ssl library:

          cmake . -DCMAKE_INSTALL_PREFIX=/opt/mariadb-10.0 -DCMAKE_BUILD_TYPE=Debug -DWITH_VALGRIND=1 -DWITH_SSL=bundled

          Note, this problem is not caught by BuildBot,
          because work-amd64-valgrind compiles with the bundled ssl library.

            Gliffy Diagrams

              Attachments

                Activity

                Ascending order - Click to sort in descending order
                Hide
                Permalink
                sanja Oleksandr Byelkin added a comment -

                It looks like valgrind thinks like a big part of read buffer is something random.

                Show
                sanja Oleksandr Byelkin added a comment - It looks like valgrind thinks like a big part of read buffer is something random.
                Hide
                Permalink
                sanja Oleksandr Byelkin added a comment - - edited

                Looks looks it is not related to client attributes but to the ssl library.

                Show
                sanja Oleksandr Byelkin added a comment - - edited Looks looks it is not related to client attributes but to the ssl library.
                Hide
                Permalink
                serg Sergei Golubchik added a comment -

                I cannot repeat that. I'm getting tons of valgrind warnings about openssl, but not this one.

                Show
                serg Sergei Golubchik added a comment - I cannot repeat that. I'm getting tons of valgrind warnings about openssl, but not this one.

                  People

                  • Assignee:
                    serg Sergei Golubchik
                    Reporter:
                    bar Alexander Barkov
                  • Votes:
                    0 Vote for this issue
                    Watchers:
                    3 Start watching this issue

                    Dates

                    • Due:
                      Created:
                      Updated:
                      Resolved: