[net PATCH 0/2] fix for cpsw interface up/down with high ethernet traffic

[net PATCH 0/2] fix for cpsw interface up/down with high ethernet traffic

[Mugunthan V N]

This patch series fixes the cpsw issue with interface up/dpwn with high
ethernet traffic.

Mugunthan V N (2):
  drivers: net: cpsw: discard all packets received when interface is
    down
  drivers: net: cpsw: enable interrupts after napi enable and clearing
    previous interrupts

 drivers/net/ethernet/ti/cpsw.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

-- 
1.9.2.459.g68773ac

[net PATCH 1/2] drivers: net: cpsw: discard all packets received when interface is down

[Mugunthan V N]

When the Ethernet interface is brought down during high Ethernet traffic,
then cpsw creates the following warn dump. When cpdma has already processed
the packet then the status will be greater than 0, so the cpsw_rx_handler
considers that the interface is up and try to resubmit one more rx buffer
to cpdma which fails as the DMA is in teardown process. This can be avoided
by checking the interface state and then process the received packet, if the
interface is down just discard and free the skb and return.

[ 2823.104591] WARNING: CPU: 0 PID: 1823 at drivers/net/ethernet/ti/cpsw.c:711 cpsw_rx_handler+0x148/0x164()
[ 2823.114654] Modules linked in:
[ 2823.117872] CPU: 0 PID: 1823 Comm: ifconfig Tainted: G        W     3.14.0-11992-gf34c4a3 #11
[ 2823.126860] [<c0014b5c>] (unwind_backtrace) from [<c00117e4>] (show_stack+0x10/0x14)
[ 2823.135030] [<c00117e4>] (show_stack) from [<c0533a9c>] (dump_stack+0x80/0x9c)
[ 2823.142619] [<c0533a9c>] (dump_stack) from [<c003f0e0>] (warn_slowpath_common+0x6c/0x90)
[ 2823.151141] [<c003f0e0>] (warn_slowpath_common) from [<c003f120>] (warn_slowpath_null+0x1c/0x24)
[ 2823.160336] [<c003f120>] (warn_slowpath_null) from [<c03caeb0>] (cpsw_rx_handler+0x148/0x164)
[ 2823.169314] [<c03caeb0>] (cpsw_rx_handler) from [<c03c730c>] (__cpdma_chan_free+0x90/0xa8)
[ 2823.178028] [<c03c730c>] (__cpdma_chan_free) from [<c03c7418>] (__cpdma_chan_process+0xf4/0x134)
[ 2823.187279] [<c03c7418>] (__cpdma_chan_process) from [<c03c7560>] (cpdma_chan_stop+0xb4/0x17c)
[ 2823.196349] [<c03c7560>] (cpdma_chan_stop) from [<c03c766c>] (cpdma_ctlr_stop+0x44/0x9c)
[ 2823.204872] [<c03c766c>] (cpdma_ctlr_stop) from [<c03cb708>] (cpsw_ndo_stop+0x154/0x188)
[ 2823.213321] [<c03cb708>] (cpsw_ndo_stop) from [<c046f0ec>] (__dev_close_many+0x84/0xc8)
[ 2823.221761] [<c046f0ec>] (__dev_close_many) from [<c046f158>] (__dev_close+0x28/0x3c)
[ 2823.230012] [<c046f158>] (__dev_close) from [<c0474ca8>] (__dev_change_flags+0x88/0x160)
[ 2823.238483] [<c0474ca8>] (__dev_change_flags) from [<c0474da0>] (dev_change_flags+0x18/0x48)
[ 2823.247316] [<c0474da0>] (dev_change_flags) from [<c04d12c4>] (devinet_ioctl+0x61c/0x6e0)
[ 2823.255884] [<c04d12c4>] (devinet_ioctl) from [<c045c660>] (sock_ioctl+0x68/0x2a4)
[ 2823.263789] [<c045c660>] (sock_ioctl) from [<c0125fe4>] (do_vfs_ioctl+0x78/0x61c)
[ 2823.271629] [<c0125fe4>] (do_vfs_ioctl) from [<c01265ec>] (SyS_ioctl+0x64/0x74)
[ 2823.279284] [<c01265ec>] (SyS_ioctl) from [<c000e580>] (ret_fast_syscall+0x0/0x48)

Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
---
 drivers/net/ethernet/ti/cpsw.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
index 6133e4a..4e3d197 100644
--- a/drivers/net/ethernet/ti/cpsw.c
+++ b/drivers/net/ethernet/ti/cpsw.c
@@ -687,7 +687,7 @@ static void cpsw_rx_handler(void *token, int len, int status)
 
 	cpsw_dual_emac_src_port_detect(status, priv, ndev, skb);
 
-	if (unlikely(status < 0)) {
+	if (unlikely(status < 0) || unlikely(!netif_running(ndev))) {
 		/* the interface is going down, skbs are purged */
 		dev_kfree_skb_any(skb);
 		return;
-- 
1.9.2.459.g68773ac

[net PATCH 2/2] drivers: net: cpsw: enable interrupts after napi enable and clearing previous interrupts

[Mugunthan V N]

When the Ethernet interface is put down and up with heavy Ethernet
traffic, then there is prossibility of an interrupt waiting in irq
controller to be processed, so when the interface is brought up again
just after enable interrupt, it goes to ISR due to the previous
unhandled interrutp and in ISR napi is not scheduled as the napi
is not enabled in ndo_open which results in disabled interrupt for
CPSW and no packets are received in cpsw. So this patch moves enabling
of interupts after napi_enable and clearing CPDMA interrupts.

Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
---
 drivers/net/ethernet/ti/cpsw.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
index 4e3d197..36aa109 100644
--- a/drivers/net/ethernet/ti/cpsw.c
+++ b/drivers/net/ethernet/ti/cpsw.c
@@ -1252,6 +1252,12 @@ static int cpsw_ndo_open(struct net_device *ndev)
 		cpsw_set_coalesce(ndev, &coal);
 	}
 
+	napi_enable(&priv->napi);
+	cpdma_ctlr_start(priv->dma);
+	cpsw_intr_enable(priv);
+	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_RX);
+	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_TX);
+
 	prim_cpsw = cpsw_get_slave_priv(priv, 0);
 	if (prim_cpsw->irq_enabled == false) {
 		if ((priv == prim_cpsw) || !netif_running(prim_cpsw->ndev)) {
@@ -1260,12 +1266,6 @@ static int cpsw_ndo_open(struct net_device *ndev)
 		}
 	}
 
-	napi_enable(&priv->napi);
-	cpdma_ctlr_start(priv->dma);
-	cpsw_intr_enable(priv);
-	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_RX);
-	cpdma_ctlr_eoi(priv->dma, CPDMA_EOI_TX);
-
 	if (priv->data.dual_emac)
 		priv->slaves[priv->emac_port].open_stat = true;
 	return 0;
-- 
1.9.2.459.g68773ac

Re: [net PATCH 0/2] fix for cpsw interface up/down with high ethernet traffic

[David Miller]

From: Mugunthan V N <mugunthanvnm@ti.com>
Date: Thu, 10 Apr 2014 14:23:22 +0530

> This patch series fixes the cpsw issue with interface up/dpwn with high
> ethernet traffic.

Series applied, thanks.