| Message ID | 20210609103326.278782-16-toke@redhat.com |
|---|---|
| State | Superseded |
| Headers | show |
| Series | Clean up and document RCU-based object protection for XDP_REDIRECT | expand |
On Wed, 9 Jun 2021 at 13:33, Toke Høiland-Jørgensen <toke@redhat.com> wrote: > > The netsec driver has a rcu_read_lock()/rcu_read_unlock() pair around the > full RX loop, covering everything up to and including xdp_do_flush(). This > is actually the correct behaviour, but because it all happens in a single > NAPI poll cycle (and thus under local_bh_disable()), it is also technically > redundant. > > With the addition of RCU annotations to the XDP_REDIRECT map types that > take bh execution into account, lockdep even understands this to be safe, > so there's really no reason to keep the rcu_read_lock() around anymore, so > let's just remove it. > > Cc: Jassi Brar <jaswinder.singh@linaro.org> > Cc: Ilias Apalodimas <ilias.apalodimas@linaro.org> > Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> > --- > drivers/net/ethernet/socionext/netsec.c | 3 --- > 1 file changed, 3 deletions(-) > > diff --git a/drivers/net/ethernet/socionext/netsec.c b/drivers/net/ethernet/socionext/netsec.c > index dfc85cc68173..20d148c019d8 100644 > --- a/drivers/net/ethernet/socionext/netsec.c > +++ b/drivers/net/ethernet/socionext/netsec.c > @@ -958,7 +958,6 @@ static int netsec_process_rx(struct netsec_priv *priv, int budget) > > xdp_init_buff(&xdp, PAGE_SIZE, &dring->xdp_rxq); > > - rcu_read_lock(); > xdp_prog = READ_ONCE(priv->xdp_prog); > dma_dir = page_pool_get_dma_dir(dring->page_pool); > > @@ -1069,8 +1068,6 @@ static int netsec_process_rx(struct netsec_priv *priv, int budget) > } > netsec_finalize_xdp_rx(priv, xdp_act, xdp_xmit); > > - rcu_read_unlock(); > - > return done; > } > > -- > 2.31.1 > Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
diff --git a/drivers/net/ethernet/socionext/netsec.c b/drivers/net/ethernet/socionext/netsec.c index dfc85cc68173..20d148c019d8 100644 --- a/drivers/net/ethernet/socionext/netsec.c +++ b/drivers/net/ethernet/socionext/netsec.c @@ -958,7 +958,6 @@ static int netsec_process_rx(struct netsec_priv *priv, int budget) xdp_init_buff(&xdp, PAGE_SIZE, &dring->xdp_rxq); - rcu_read_lock(); xdp_prog = READ_ONCE(priv->xdp_prog); dma_dir = page_pool_get_dma_dir(dring->page_pool); @@ -1069,8 +1068,6 @@ static int netsec_process_rx(struct netsec_priv *priv, int budget) } netsec_finalize_xdp_rx(priv, xdp_act, xdp_xmit); - rcu_read_unlock(); - return done; }
The netsec driver has a rcu_read_lock()/rcu_read_unlock() pair around the full RX loop, covering everything up to and including xdp_do_flush(). This is actually the correct behaviour, but because it all happens in a single NAPI poll cycle (and thus under local_bh_disable()), it is also technically redundant. With the addition of RCU annotations to the XDP_REDIRECT map types that take bh execution into account, lockdep even understands this to be safe, so there's really no reason to keep the rcu_read_lock() around anymore, so let's just remove it. Cc: Jassi Brar <jaswinder.singh@linaro.org> Cc: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> --- drivers/net/ethernet/socionext/netsec.c | 3 --- 1 file changed, 3 deletions(-)