diff mbox

[v3,3/4] net: nb8800: Move HW init to ndo_open()

Message ID a047ecbd-c8a2-a8c0-7c9d-947b28aa8b9a@sigmadesigns.com (mailing list archive)
State New, archived
Headers show

Commit Message

Marc Gonzalez Nov. 15, 2017, 2:58 p.m. UTC 
On 14/11/2017 17:55, Måns Rullgård wrote:

> Marc Gonzalez wrote:
> 
>> I will run iperf3 tests with RX_DESC_COUNT lowered to 2.
>> Would that produce conclusive results?
>> Do you have other suggestions?
> 
> Leave RX_DESC_COUNT alone but add a delay in the nb8800_poll() loop.
> That should ensure that queue is drained slowly enough for the buffers
> to run out.

Using the following patch:



I saw both tango4 and tango5 wedged... :-(

tango5:

[  5] local 172.27.64.23 port 5201 connected to 172.27.64.1 port 57415
[ ID] Interval           Transfer     Bandwidth       Jitter    Lost/Total Datagrams
[  5]   0.00-1.00   sec  6.33 MBytes  53.1 Mbits/sec  0.030 ms  73897/78442 (94%)  
[  5]   1.00-2.00   sec  6.59 MBytes  55.3 Mbits/sec  0.040 ms  76944/81675 (94%)  
[  5]   2.00-3.00   sec  6.59 MBytes  55.3 Mbits/sec  0.036 ms  76944/81675 (94%)  
[  5]   3.00-4.00   sec  6.59 MBytes  55.3 Mbits/sec  0.038 ms  76976/81708 (94%)  
[  5]   4.00-5.00   sec  6.59 MBytes  55.3 Mbits/sec  0.038 ms  76976/81708 (94%)  
[  5]   5.00-6.00   sec  6.59 MBytes  55.3 Mbits/sec  0.039 ms  76944/81675 (94%)  
[  5]   6.00-7.00   sec  6.59 MBytes  55.3 Mbits/sec  0.042 ms  76941/81674 (94%)  
[  5]   7.00-8.00   sec  6.59 MBytes  55.3 Mbits/sec  0.048 ms  76976/81707 (94%)  
[  5]   8.00-9.00   sec  6.59 MBytes  55.3 Mbits/sec  0.041 ms  76928/81660 (94%)  
[  5]   9.00-10.00  sec  6.59 MBytes  55.3 Mbits/sec  0.039 ms  76960/81692 (94%)  
[  5]  10.00-10.04  sec   279 KBytes  55.1 Mbits/sec  0.094 ms  3088/3284 (94%)  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth       Jitter    Lost/Total Datagrams
[  5]   0.00-10.04  sec  0.00 Bytes  0.00 bits/sec  0.094 ms  769574/816900 (94%)  
^Ciperf3: interrupt - the server has terminated
# ping -c 10 172.27.64.1
PING 172.27.64.1 (172.27.64.1): 56 data bytes

--- 172.27.64.1 ping statistics ---
10 packets transmitted, 0 packets received, 100% packet loss


tango4:

[  5] local 172.27.64.23 port 5201 connected to 172.27.64.1 port 52983
[ ID] Interval           Transfer     Bandwidth       Jitter    Lost/Total Datagrams
[  5]   0.00-1.00   sec  6.24 MBytes  52.3 Mbits/sec  0.035 ms  74019/78498 (94%)  
[  5]   1.00-2.00   sec  6.51 MBytes  54.6 Mbits/sec  0.046 ms  77024/81702 (94%)  
[  5]   2.00-3.00   sec  6.51 MBytes  54.7 Mbits/sec  0.050 ms  77024/81703 (94%)  
[  5]   3.00-4.00   sec  6.51 MBytes  54.7 Mbits/sec  0.050 ms  77024/81703 (94%)  
[  5]   4.00-5.00   sec  6.52 MBytes  54.7 Mbits/sec  0.052 ms  77024/81705 (94%)  
[  5]   5.00-6.00   sec  6.52 MBytes  54.7 Mbits/sec  0.037 ms  76960/81640 (94%)  
[  5]   6.00-7.00   sec  6.52 MBytes  54.7 Mbits/sec  0.043 ms  77008/81689 (94%)  
[  5]   7.00-8.00   sec  6.52 MBytes  54.7 Mbits/sec  0.050 ms  77024/81704 (94%)  
[  5]   8.00-9.00   sec  6.52 MBytes  54.7 Mbits/sec  0.045 ms  77024/81705 (94%)  
[  5]   9.00-10.00  sec  6.52 MBytes  54.7 Mbits/sec  0.043 ms  77008/81688 (94%)  
[  5]  10.00-10.04  sec   275 KBytes  54.5 Mbits/sec  0.103 ms  3040/3233 (94%)  
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth       Jitter    Lost/Total Datagrams
[  5]   0.00-10.04  sec  0.00 Bytes  0.00 bits/sec  0.103 ms  770179/816970 (94%)  
^Ciperf3: interrupt - the server has terminated
# ping -c 10 172.27.64.1
PING 172.27.64.1 (172.27.64.1): 56 data bytes

--- 172.27.64.1 ping statistics ---
10 packets transmitted, 0 packets received, 100% packet loss


More tests needed...

Comments

Måns Rullgård Nov. 15, 2017, 3:11 p.m. UTC | #1 
Marc Gonzalez <marc_gonzalez@sigmadesigns.com> writes:

> On 14/11/2017 17:55, Måns Rullgård wrote:
>
>> Marc Gonzalez wrote:
>> 
>>> I will run iperf3 tests with RX_DESC_COUNT lowered to 2.
>>> Would that produce conclusive results?
>>> Do you have other suggestions?
>> 
>> Leave RX_DESC_COUNT alone but add a delay in the nb8800_poll() loop.
>> That should ensure that queue is drained slowly enough for the buffers
>> to run out.
>
> Using the following patch:
>
> diff --git a/drivers/net/ethernet/aurora/nb8800.c b/drivers/net/ethernet/aurora/nb8800.c
> index 31c3f0f10fbb..646300bb53b6 100644
> --- a/drivers/net/ethernet/aurora/nb8800.c
> +++ b/drivers/net/ethernet/aurora/nb8800.c
> @@ -317,6 +317,7 @@ static int nb8800_poll(struct napi_struct *napi, int budget)
>
>                 len = RX_BYTES_TRANSFERRED(rxd->report);
>
> +               udelay(200);
>                 if (IS_RX_ERROR(rxd->report))
>                         nb8800_rx_error(dev, rxd->report);
>                 else
> @@ -1416,9 +1417,9 @@ static int nb8800_probe(struct platform_device *pdev)
>         netdev_info(dev, "MAC address %pM\n", dev->dev_addr);
>
>         /* Auto-negotiate by default */
> -       priv->pause_aneg = true;
> -       priv->pause_rx = true;
> -       priv->pause_tx = true;
> +       priv->pause_aneg = false;
> +       priv->pause_rx = false;
> +       priv->pause_tx = false;
>
>         priv->ops = match->data;
>         priv->ops->power_down(dev);
> diff --git a/drivers/net/ethernet/aurora/nb8800.h b/drivers/net/ethernet/aurora/nb8800.h
> index 23fefca54804..9b59ea776e4a 100644
> --- a/drivers/net/ethernet/aurora/nb8800.h
> +++ b/drivers/net/ethernet/aurora/nb8800.h
> @@ -7,7 +7,7 @@
>  #include <linux/clk.h>
>  #include <linux/bitops.h>
>
> -#define RX_DESC_COUNT                  256
> +#define RX_DESC_COUNT                  16
>  #define TX_DESC_COUNT                  256
>
>  #define NB8800_DESC_LOW                        4
>
> I saw both tango4 and tango5 wedged... :-(

Well, that's not good.  I'll see if I can replicate it later this week.
diff mbox

Patch

diff --git a/drivers/net/ethernet/aurora/nb8800.c b/drivers/net/ethernet/aurora/nb8800.c
index 31c3f0f10fbb..646300bb53b6 100644
--- a/drivers/net/ethernet/aurora/nb8800.c
+++ b/drivers/net/ethernet/aurora/nb8800.c
@@ -317,6 +317,7 @@  static int nb8800_poll(struct napi_struct *napi, int budget)
 
                len = RX_BYTES_TRANSFERRED(rxd->report);
 
+               udelay(200);
                if (IS_RX_ERROR(rxd->report))
                        nb8800_rx_error(dev, rxd->report);
                else
@@ -1416,9 +1417,9 @@  static int nb8800_probe(struct platform_device *pdev)
        netdev_info(dev, "MAC address %pM\n", dev->dev_addr);
 
        /* Auto-negotiate by default */
-       priv->pause_aneg = true;
-       priv->pause_rx = true;
-       priv->pause_tx = true;
+       priv->pause_aneg = false;
+       priv->pause_rx = false;
+       priv->pause_tx = false;
 
        priv->ops = match->data;
        priv->ops->power_down(dev);
diff --git a/drivers/net/ethernet/aurora/nb8800.h b/drivers/net/ethernet/aurora/nb8800.h
index 23fefca54804..9b59ea776e4a 100644
--- a/drivers/net/ethernet/aurora/nb8800.h
+++ b/drivers/net/ethernet/aurora/nb8800.h
@@ -7,7 +7,7 @@ 
 #include <linux/clk.h>
 #include <linux/bitops.h>
 
-#define RX_DESC_COUNT                  256
+#define RX_DESC_COUNT                  16
 #define TX_DESC_COUNT                  256
 
 #define NB8800_DESC_LOW                        4