| Message ID | 20190529221523.22399-4-bigeasy@linutronix.de |
|---|---|
| State | Changes Requested |
| Delegated to: | David Miller |
| Headers | show |
| Series | Avoid local_irq_save() and use napi_alloc_frag() where possible | expand |
> -----Original Message----- > From: Sebastian Andrzej Siewior <bigeasy@linutronix.de> > Sent: Thursday, May 30, 2019 1:15 AM > To: netdev@vger.kernel.org > Cc: tglx@linutronix.de; David S. Miller <davem@davemloft.net>; Sebastian > Andrzej Siewior <bigeasy@linutronix.de>; Ioana Ciocoi Radulescu > <ruxandra.radulescu@nxp.com> > Subject: [PATCH net-next 3/7] dpaa2-eth: Remove preempt_disable() from > seed_pool() > > According to the comment, the preempt_disable() statement is required > due to synchronisation in napi_alloc_frag(). The awful truth is that > local_bh_disable() is required because otherwise the NAPI poll callback > can be invoked while the open function setup buffers. This isn't > unlikely since the dpaa2 provides multiple devices. > > The usage of napi_alloc_frag() has been removed in commit > > 27c874867c4e9 ("dpaa2-eth: Use a single page per Rx buffer") > > which means that the comment is not accurate and the preempt_disable() > statement is not required. > > Remove the outdated comment and the no longer required > preempt_disable(). > > Cc: Ioana Radulescu <ruxandra.radulescu@nxp.com> > Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Acked-by: Ioana Radulescu <ruxandra.radulescu@nxp.com> Thanks, Ioana
diff --git a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c index 63b1ecc18c26f..f9ae97ba63334 100644 --- a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c +++ b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c @@ -997,13 +997,6 @@ static int seed_pool(struct dpaa2_eth_priv *priv, u16 bpid) int i, j; int new_count; - /* This is the lazy seeding of Rx buffer pools. - * dpaa2_add_bufs() is also used on the Rx hotpath and calls - * napi_alloc_frag(). The trouble with that is that it in turn ends up - * calling this_cpu_ptr(), which mandates execution in atomic context. - * Rather than splitting up the code, do a one-off preempt disable. - */ - preempt_disable(); for (j = 0; j < priv->num_channels; j++) { for (i = 0; i < DPAA2_ETH_NUM_BUFS; i += DPAA2_ETH_BUFS_PER_CMD) { @@ -1011,12 +1004,10 @@ static int seed_pool(struct dpaa2_eth_priv *priv, u16 bpid) priv->channel[j]->buf_count += new_count; if (new_count < DPAA2_ETH_BUFS_PER_CMD) { - preempt_enable(); return -ENOMEM; } } } - preempt_enable(); return 0; }
According to the comment, the preempt_disable() statement is required due to synchronisation in napi_alloc_frag(). The awful truth is that local_bh_disable() is required because otherwise the NAPI poll callback can be invoked while the open function setup buffers. This isn't unlikely since the dpaa2 provides multiple devices. The usage of napi_alloc_frag() has been removed in commit 27c874867c4e9 ("dpaa2-eth: Use a single page per Rx buffer") which means that the comment is not accurate and the preempt_disable() statement is not required. Remove the outdated comment and the no longer required preempt_disable(). Cc: Ioana Radulescu <ruxandra.radulescu@nxp.com> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c | 9 --------- 1 file changed, 9 deletions(-)