On Thu, Nov 25, 2021 at 10:59:25PM +0100, Alexander Bluhm wrote:
> On Thu, Nov 25, 2021 at 05:13:16PM +0100, Tobias Heider wrote:
> > Now with the missing parts from pfkeyv2.c as noticed by Hrvoje.
>
> We have this code in pfkeyv2_send()
>
> if (headers[SADB_EXT_ADDRESS_SRC] ||
> headers[SADB_EXT_ADDRESS_PROXY]) {
> tdb_unlink(sa2);
> import_address((struct sockaddr
> *)&sa2->tdb_src,
> headers[SADB_EXT_ADDRESS_SRC]);
> import_address((struct sockaddr
> *)&sa2->tdb_dst,
> headers[SADB_EXT_ADDRESS_PROXY]);
> puttdb(sa2);
> }
> }
> NET_UNLOCK();
>
> Without the deleted flag, the pointers removed by tdb_unlink() and
> set by puttdb() guarantee that tdb_delete() is not called twice.
> In this piece of code they are missing for a short time.
>
> Net lock takes care of this. There should be a comment that describes
> this.
>
> /*
> * NET_LOCK prevents tdb_delete() between
> * tdb_unlink() and puttdb().
> */
> tdb_unlink(sa2);
> ...
> puttdb(sa2);
>
> Or we don't want to rely on net lock. Then we need a common mutex
> to pretect this.
>
> mtx_enter(&tdb_sadb_mtx);
> tdb_unlink_locked(tdbp);
> ...
> puttdb_locked(sa2);
> mtx_leave(&tdb_sadb_mtx);
>
Without regarding on this diff this could be right direction because
`tdb_sadb_mtx' mutex(9) protects TDB hash consistency.
