[ovs-dev] [PATCH] ovsdb-idl: Adjust indexes during transactions.

[Ben Pfaff]

On Thu, Aug 16, 2018 at 10:40:23AM -0700, Han Zhou wrote:
> On Thu, Aug 16, 2018 at 10:39 AM, Ben Pfaff <blp at ovn.org> wrote:
> >
> > On Wed, Aug 15, 2018 at 08:56:21PM -0700, Han Zhou wrote:
> > > On Wed, Aug 15, 2018 at 8:53 PM, Han Zhou <zhouhan at gmail.com> wrote:
> > > >
> > > >
> > > >
> > > > On Tue, Aug 14, 2018 at 12:39 PM, Han Zhou <zhouhan at gmail.com> wrote:
> > > > >
> > > > >
> > > > >
> > > > > On Tue, Aug 14, 2018 at 11:31 AM, Ben Pfaff <blp at ovn.org> wrote:
> > > > > >
> > > > > > When transactions modified tables with indexes, the indexes were
> not
> > > > > > properly updated to reflect the changes.  For deleted rows, in
> > > particular,
> > > > > > this could cause use-after-free errors.
> > > > > >
> > > > > > This commit fixes the problem and adds a very simple test case
> > > provided by
> > > > > > Han Zhou that, without the fix, causes a crash.
> > > > > >
> > > > > > Reported-by: Han Zhou <zhouhan at gmail.com>
> > > > > > Reported-at:
> > >
> https://mail.openvswitch.org/pipermail/ovs-discuss/2018-August/047185.html
> > > > > > Signed-off-by: Ben Pfaff <blp at ovn.org>
> > > > > > ---
> > > > > >  lib/ovsdb-idl.c    | 25 +++++++++++++++++++++++--
> > > > > >  tests/test-ovsdb.c |  1 +
> > > > > >  2 files changed, 24 insertions(+), 2 deletions(-)
> > > > > >
> > > > > > diff --git a/lib/ovsdb-idl.c b/lib/ovsdb-idl.c
> > > > > > index 8fdd18f4688e..a4d66113b18d 100644
> > > > > > --- a/lib/ovsdb-idl.c
> > > > > > +++ b/lib/ovsdb-idl.c
> > > > > > @@ -3508,9 +3508,18 @@ ovsdb_idl_txn_disassemble(struct
> ovsdb_idl_txn
> > > *txn)
> > > > > >      txn->db->txn = NULL;
> > > > > >
> > > > > >      HMAP_FOR_EACH_SAFE (row, next, txn_node, &txn->txn_rows) {
> > > > > > +        enum { INSERTED, MODIFIED, DELETED } op
> > > > > > +            = (!row->new_datum ? DELETED
> > > > > > +               : !row->old_datum ? INSERTED
> > > > > > +               : MODIFIED);
> > > > > > +
> > > > > > +        if (op != DELETED) {
> > > > > > +            ovsdb_idl_remove_from_indexes(row);
> > > > > > +        }
> > > > > > +
> > > > > >          ovsdb_idl_destroy_all_map_op_lists(row);
> > > > > >          ovsdb_idl_destroy_all_set_op_lists(row);
> > > > > > -        if (row->old_datum) {
> > > > > > +        if (op != INSERTED) {
> > > > > >              if (row->written) {
> > > > > >                  ovsdb_idl_row_unparse(row);
> > > > > >                  ovsdb_idl_row_clear_arcs(row, false);
> > > > > > @@ -3529,7 +3538,9 @@ ovsdb_idl_txn_disassemble(struct
> ovsdb_idl_txn
> > > *txn)
> > > > > >
> > > > > >          hmap_remove(&txn->txn_rows, &row->txn_node);
> > > > > >          hmap_node_nullify(&row->txn_node);
> > > > > > -        if (!row->old_datum) {
> > > > > > +        if (op != INSERTED) {
> > > > > > +            ovsdb_idl_add_to_indexes(row);
> > > > > > +        } else {
> > > > > >              hmap_remove(&row->table->rows, &row->hmap_node);
> > > > > >              free(row);
> > > > > >          }
> > > > > > @@ -4209,6 +4220,10 @@ ovsdb_idl_txn_write__(const struct
> > > ovsdb_idl_row *row_,
> > > > > >          goto discard_datum;
> > > > > >      }
> > > > > >
> > > > > > +    bool index_row = is_index_row(row);
> > > > > > +    if (!index_row) {
> > > > > > +        ovsdb_idl_remove_from_indexes(row);
> > > > > > +    }
> > > > > >      if (hmap_node_is_null(&row->txn_node)) {
> > > > > >          hmap_insert(&row->table->db->txn->txn_rows,
> &row->txn_node,
> > > > > >                      uuid_hash(&row->uuid));
> > > > > > @@ -4231,6 +4246,9 @@ ovsdb_idl_txn_write__(const struct
> > > ovsdb_idl_row *row_,
> > > > > >      }
> > > > > >      (column->unparse)(row);
> > > > > >      (column->parse)(row, &row->new_datum[column_idx]);
> > > > > > +    if (!index_row) {
> > > > > > +        ovsdb_idl_add_to_indexes(row);
> > > > > > +    }
> > > > > >      return;
> > > > > >
> > > > > >  discard_datum:
> > > > > > @@ -4358,6 +4376,8 @@ ovsdb_idl_txn_delete(const struct
> ovsdb_idl_row
> > > *row_)
> > > > > >      }
> > > > > >
> > > > > >      ovs_assert(row->new_datum != NULL);
> > > > > > +    ovs_assert(!is_index_row(row_));
> > > > > > +    ovsdb_idl_remove_from_indexes(row_);
> > > > > >      if (!row->old_datum) {
> > > > > >          ovsdb_idl_row_unparse(row);
> > > > > >          ovsdb_idl_row_clear_new(row);
> > > > > > @@ -4405,6 +4425,7 @@ ovsdb_idl_txn_insert(struct ovsdb_idl_txn
> *txn,
> > > > > >      row->new_datum = xmalloc(class->n_columns * sizeof
> > > *row->new_datum);
> > > > > >      hmap_insert(&row->table->rows, &row->hmap_node,
> > > uuid_hash(&row->uuid));
> > > > > >      hmap_insert(&txn->txn_rows, &row->txn_node,
> > > uuid_hash(&row->uuid));
> > > > > > +    ovsdb_idl_add_to_indexes(row);
> > > > > >      return row;
> > > > > >  }
> > > > > >
> > > > > > diff --git a/tests/test-ovsdb.c b/tests/test-ovsdb.c
> > > > > > index 793220400c3a..e981b588849d 100644
> > > > > > --- a/tests/test-ovsdb.c
> > > > > > +++ b/tests/test-ovsdb.c
> > > > > > @@ -2882,6 +2882,7 @@ test_idl_compound_index_single_column(struct
> > > ovsdb_idl *idl,
> > > > > >      ovs_assert(myRow->i == 4);
> > > > > >      txn = ovsdb_idl_txn_create(idl);
> > > > > >      idltest_simple_delete(myRow);
> > > > > > +    myRow = idltest_simple_index_find(i_index, toDelete);
> > > > > >      toInsert = idltest_simple_insert(txn);
> > > > > >      idltest_simple_set_i(toInsert, 54);
> > > > > >      idltest_simple_set_s(toInsert, "Lista054");
> > > > > > --
> > > > > > 2.16.1
> > > > > >
> > > > >
> > > > > Thanks Ben for the quick fix!! It may be better if we add a little
> more
> > > tests to ensure change is reflected in indexes before transaction
> commit.
> > > > >
> > > > > Acked-by: Han Zhou <hzhou8 at ebay.com>
> > > >
> > > > Hi Ben, I added below tests on top of your patch and it works as
> > > expected. So consider it as my Tested-by, too.
> >
> > Thanks a lot for the test improvements.
> >
> > I applied this, with the tests, to master and branch-2.10.
> 
> Thanks Ben. I think it is needed in 2.9 and 2.8, too.

The backport to 2.9 was pretty easy so I did it.

2.8 seems to lack a lot of related stuff.   I don't think we really used
indexes in that release, so I don't think the backport is worth it.