[dpdk-dev] [PATCH] examples/flow_filtering: add delay during updating link status

[dpdk-dev] [PATCH] examples/flow_filtering: add delay during updating link status

[Beilei Xing]

Add up to 9s delay for getting link status to make sure NIC updates
link status successfully, just like other applications such as
testpmd and l2fwd.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
---
 examples/flow_filtering/main.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/examples/flow_filtering/main.c b/examples/flow_filtering/main.c
index 4a07b63..1788f24 100644
--- a/examples/flow_filtering/main.c
+++ b/examples/flow_filtering/main.c
@@ -55,6 +55,7 @@
 #include <rte_mbuf.h>
 #include <rte_net.h>
 #include <rte_flow.h>
+#include <rte_cycles.h>
 
 static volatile bool force_quit;
 
@@ -119,13 +120,22 @@ main_loop(void)
 	rte_eth_dev_close(port_id);
 }
 
+#define CHECK_INTERVAL 1000  /* 100ms */
+#define MAX_REPEAT_TIME 90   /* 9s (90 * 100ms) in total */
+
 static void
 assert_link_status(void)
 {
 	struct rte_eth_link link;
+	uint8_t rep_cnt = MAX_REPEAT_TIME;
 
 	memset(&link, 0, sizeof(link));
-	rte_eth_link_get(port_id, &link);
+	do {
+		rte_eth_link_get(port_id, &link);
+		if (link.link_status == ETH_LINK_UP)
+			break;
+	} while (--rep_cnt);
+
 	if (link.link_status == ETH_LINK_DOWN)
 		rte_exit(EXIT_FAILURE, ":: error: link is still down\n");
 }
-- 
2.5.5

Re: [dpdk-dev] [PATCH] examples/flow_filtering: add delay during updating link status

[Ori Kam]

Hi Beilei,

PSB

> -----Original Message-----
> From: Beilei Xing [mailto:beilei.xing@intel.com]
> Sent: Wednesday, January 24, 2018 9:53 AM
> To: Ori Kam <orika@mellanox.com>
> Cc: dev@dpdk.org
> Subject: [PATCH] examples/flow_filtering: add delay during updating link
> status
> 
> Add up to 9s delay for getting link status to make sure NIC updates link status
> successfully, just like other applications such as testpmd and l2fwd.
> 
> Signed-off-by: Beilei Xing <beilei.xing@intel.com>
> ---
>  examples/flow_filtering/main.c | 12 +++++++++++-
>  1 file changed, 11 insertions(+), 1 deletion(-)
> 
> diff --git a/examples/flow_filtering/main.c b/examples/flow_filtering/main.c
> index 4a07b63..1788f24 100644
> --- a/examples/flow_filtering/main.c
> +++ b/examples/flow_filtering/main.c
> @@ -55,6 +55,7 @@
>  #include <rte_mbuf.h>
>  #include <rte_net.h>
>  #include <rte_flow.h>
> +#include <rte_cycles.h>
> 
>  static volatile bool force_quit;
> 
> @@ -119,13 +120,22 @@ main_loop(void)
>  	rte_eth_dev_close(port_id);
>  }
> 
> +#define CHECK_INTERVAL 1000  /* 100ms */

This define is not used.

> +#define MAX_REPEAT_TIME 90   /* 9s (90 * 100ms) in total */
> +
>  static void
>  assert_link_status(void)
>  {
>  	struct rte_eth_link link;
> +	uint8_t rep_cnt = MAX_REPEAT_TIME;
> 
>  	memset(&link, 0, sizeof(link));
> -	rte_eth_link_get(port_id, &link);
> +	do {
> +		rte_eth_link_get(port_id, &link);
> +		if (link.link_status == ETH_LINK_UP)
> +			break;
> +	} while (--rep_cnt);

I think you are missing the rte_delay_ms(CHECK_INTERVAL);
Currently the code will work for 90 iterations  but we can't grantee the duration.

> +
>  	if (link.link_status == ETH_LINK_DOWN)
>  		rte_exit(EXIT_FAILURE, ":: error: link is still down\n");  }
> --
> 2.5.5

Regards,
Ori

Re: [dpdk-dev] [PATCH] examples/flow_filtering: add delay during updating link status

[Xing, Beilei]

> -----Original Message-----
> From: Ori Kam [mailto:orika@mellanox.com]
> Sent: Wednesday, January 24, 2018 4:17 PM
> To: Xing, Beilei <beilei.xing@intel.com>
> Cc: dev@dpdk.org
> Subject: RE: [PATCH] examples/flow_filtering: add delay during updating link
> status
> 
> Hi Beilei,
> 
> PSB
> 
> > -----Original Message-----
> > From: Beilei Xing [mailto:beilei.xing@intel.com]
> > Sent: Wednesday, January 24, 2018 9:53 AM
> > To: Ori Kam <orika@mellanox.com>
> > Cc: dev@dpdk.org
> > Subject: [PATCH] examples/flow_filtering: add delay during updating
> > link status
> >
> > Add up to 9s delay for getting link status to make sure NIC updates
> > link status successfully, just like other applications such as testpmd and
> l2fwd.
> >
> > Signed-off-by: Beilei Xing <beilei.xing@intel.com>
> > ---
> >  examples/flow_filtering/main.c | 12 +++++++++++-
> >  1 file changed, 11 insertions(+), 1 deletion(-)
> >
> > diff --git a/examples/flow_filtering/main.c
> > b/examples/flow_filtering/main.c index 4a07b63..1788f24 100644
> > --- a/examples/flow_filtering/main.c
> > +++ b/examples/flow_filtering/main.c
> > @@ -55,6 +55,7 @@
> >  #include <rte_mbuf.h>
> >  #include <rte_net.h>
> >  #include <rte_flow.h>
> > +#include <rte_cycles.h>
> >
> >  static volatile bool force_quit;
> >
> > @@ -119,13 +120,22 @@ main_loop(void)
> >  	rte_eth_dev_close(port_id);
> >  }
> >
> > +#define CHECK_INTERVAL 1000  /* 100ms */
> 
> This define is not used.
> 
> > +#define MAX_REPEAT_TIME 90   /* 9s (90 * 100ms) in total */
> > +
> >  static void
> >  assert_link_status(void)
> >  {
> >  	struct rte_eth_link link;
> > +	uint8_t rep_cnt = MAX_REPEAT_TIME;
> >
> >  	memset(&link, 0, sizeof(link));
> > -	rte_eth_link_get(port_id, &link);
> > +	do {
> > +		rte_eth_link_get(port_id, &link);
> > +		if (link.link_status == ETH_LINK_UP)
> > +			break;
> > +	} while (--rep_cnt);
> 
> I think you are missing the rte_delay_ms(CHECK_INTERVAL); Currently the
> code will work for 90 iterations  but we can't grantee the duration.

Sorry for missing rte_delay_ms in this patch, will send v2 later, thanks.

> 
> > +
> >  	if (link.link_status == ETH_LINK_DOWN)
> >  		rte_exit(EXIT_FAILURE, ":: error: link is still down\n");  }
> > --
> > 2.5.5
> 
> Regards,
> Ori

[dpdk-dev] [PATCH v2] examples/flow_filtering: add delay during updating link status

[Beilei Xing]

Add up to 9s delay for getting link status to make sure NIC updates
link status successfully, just like other applications such as
testpmd and l2fwd.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
---

v2 changes:
 - Add rte_delay_ms(CHECK_INTERVAL) which is missed in v1.

 examples/flow_filtering/main.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/examples/flow_filtering/main.c b/examples/flow_filtering/main.c
index 4a07b63..85d5727 100644
--- a/examples/flow_filtering/main.c
+++ b/examples/flow_filtering/main.c
@@ -55,6 +55,7 @@
 #include <rte_mbuf.h>
 #include <rte_net.h>
 #include <rte_flow.h>
+#include <rte_cycles.h>
 
 static volatile bool force_quit;
 
@@ -119,13 +120,23 @@ main_loop(void)
 	rte_eth_dev_close(port_id);
 }
 
+#define CHECK_INTERVAL 1000  /* 100ms */
+#define MAX_REPEAT_TIME 90   /* 9s (90 * 100ms) in total */
+
 static void
 assert_link_status(void)
 {
 	struct rte_eth_link link;
+	uint8_t rep_cnt = MAX_REPEAT_TIME;
 
 	memset(&link, 0, sizeof(link));
-	rte_eth_link_get(port_id, &link);
+	do {
+		rte_eth_link_get(port_id, &link);
+		if (link.link_status == ETH_LINK_UP)
+			break;
+		rte_delay_ms(CHECK_INTERVAL);
+	} while (--rep_cnt);
+
 	if (link.link_status == ETH_LINK_DOWN)
 		rte_exit(EXIT_FAILURE, ":: error: link is still down\n");
 }
-- 
2.5.5

Re: [dpdk-dev] [PATCH v2] examples/flow_filtering: add delay during updating link status

[Ori Kam]

Hi


> -----Original Message-----
> From: Beilei Xing [mailto:beilei.xing@intel.com]
> Sent: Wednesday, January 24, 2018 10:35 AM
> To: Ori Kam <orika@mellanox.com>
> Cc: dev@dpdk.org
> Subject: [PATCH v2] examples/flow_filtering: add delay during updating link
> status
> 
> Add up to 9s delay for getting link status to make sure NIC updates link status
> successfully, just like other applications such as testpmd and l2fwd.
> 
> Signed-off-by: Beilei Xing <beilei.xing@intel.com>
> ---
> 
> v2 changes:
>  - Add rte_delay_ms(CHECK_INTERVAL) which is missed in v1.
> 
>  examples/flow_filtering/main.c | 13 ++++++++++++-
>  1 file changed, 12 insertions(+), 1 deletion(-)
> 
> diff --git a/examples/flow_filtering/main.c b/examples/flow_filtering/main.c
> index 4a07b63..85d5727 100644
> --- a/examples/flow_filtering/main.c
> +++ b/examples/flow_filtering/main.c
> @@ -55,6 +55,7 @@
>  #include <rte_mbuf.h>
>  #include <rte_net.h>
>  #include <rte_flow.h>
> +#include <rte_cycles.h>
> 
>  static volatile bool force_quit;
> 
> @@ -119,13 +120,23 @@ main_loop(void)
>  	rte_eth_dev_close(port_id);
>  }
> 
> +#define CHECK_INTERVAL 1000  /* 100ms */
> +#define MAX_REPEAT_TIME 90   /* 9s (90 * 100ms) in total */

I know that in other examples there is use of 
MAX_REPEAT_TIME but don't you think the name is incorrect,
It should be called:
MAX_REPEAT_TIMES or MAX_REPEAT_COUNT?
Since it doesn't represent time but iterations.
What do you think?

> +
>  static void
>  assert_link_status(void)
>  {
>  	struct rte_eth_link link;
> +	uint8_t rep_cnt = MAX_REPEAT_TIME;
> 
>  	memset(&link, 0, sizeof(link));
> -	rte_eth_link_get(port_id, &link);
> +	do {
> +		rte_eth_link_get(port_id, &link);
> +		if (link.link_status == ETH_LINK_UP)
> +			break;
> +		rte_delay_ms(CHECK_INTERVAL);
> +	} while (--rep_cnt);
> +
>  	if (link.link_status == ETH_LINK_DOWN)
>  		rte_exit(EXIT_FAILURE, ":: error: link is still down\n");  }
> --
> 2.5.5

Re: [dpdk-dev] [PATCH v2] examples/flow_filtering: add delay during updating link status

[Xing, Beilei]

> -----Original Message-----
> From: Ori Kam [mailto:orika@mellanox.com]
> Sent: Wednesday, January 24, 2018 4:56 PM
> To: Xing, Beilei <beilei.xing@intel.com>
> Cc: dev@dpdk.org
> Subject: RE: [PATCH v2] examples/flow_filtering: add delay during updating
> link status
> 
> Hi
> 
> 
> > -----Original Message-----
> > From: Beilei Xing [mailto:beilei.xing@intel.com]
> > Sent: Wednesday, January 24, 2018 10:35 AM
> > To: Ori Kam <orika@mellanox.com>
> > Cc: dev@dpdk.org
> > Subject: [PATCH v2] examples/flow_filtering: add delay during updating
> > link status
> >
> > Add up to 9s delay for getting link status to make sure NIC updates
> > link status successfully, just like other applications such as testpmd and
> l2fwd.
> >
> > Signed-off-by: Beilei Xing <beilei.xing@intel.com>
> > ---
> >
> > v2 changes:
> >  - Add rte_delay_ms(CHECK_INTERVAL) which is missed in v1.
> >
> >  examples/flow_filtering/main.c | 13 ++++++++++++-
> >  1 file changed, 12 insertions(+), 1 deletion(-)
> >
> > diff --git a/examples/flow_filtering/main.c
> > b/examples/flow_filtering/main.c index 4a07b63..85d5727 100644
> > --- a/examples/flow_filtering/main.c
> > +++ b/examples/flow_filtering/main.c
> > @@ -55,6 +55,7 @@
> >  #include <rte_mbuf.h>
> >  #include <rte_net.h>
> >  #include <rte_flow.h>
> > +#include <rte_cycles.h>
> >
> >  static volatile bool force_quit;
> >
> > @@ -119,13 +120,23 @@ main_loop(void)
> >  	rte_eth_dev_close(port_id);
> >  }
> >
> > +#define CHECK_INTERVAL 1000  /* 100ms */
> > +#define MAX_REPEAT_TIME 90   /* 9s (90 * 100ms) in total */
> 
> I know that in other examples there is use of MAX_REPEAT_TIME but don't
> you think the name is incorrect, It should be called:
> MAX_REPEAT_TIMES or MAX_REPEAT_COUNT?
> Since it doesn't represent time but iterations.
> What do you think?
> 

Make sense, looks like MAX_REPEAT_TIMES should be more accurate than MAX_REPEAT_TIME.
Will update in next version.

> > +
> >  static void
> >  assert_link_status(void)
> >  {
> >  	struct rte_eth_link link;
> > +	uint8_t rep_cnt = MAX_REPEAT_TIME;
> >
> >  	memset(&link, 0, sizeof(link));
> > -	rte_eth_link_get(port_id, &link);
> > +	do {
> > +		rte_eth_link_get(port_id, &link);
> > +		if (link.link_status == ETH_LINK_UP)
> > +			break;
> > +		rte_delay_ms(CHECK_INTERVAL);
> > +	} while (--rep_cnt);
> > +
> >  	if (link.link_status == ETH_LINK_DOWN)
> >  		rte_exit(EXIT_FAILURE, ":: error: link is still down\n");  }
> > --
> > 2.5.5

[dpdk-dev] [PATCH v3] examples/flow_filtering: add delay during updating link status

[Beilei Xing]

Add up to 9s delay for getting link status to make sure NIC updates
link status successfully, just like other applications such as
testpmd and l2fwd.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
---
v3 changes:
 - Modify MAX_REPEAT_TIME with MAX_REPEAT_TIMES
v2 changes:
 - Add rte_delay_ms(CHECK_INTERVAL) which is missed in v1.

 examples/flow_filtering/main.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/examples/flow_filtering/main.c b/examples/flow_filtering/main.c
index 4a07b63..d16a0a5 100644
--- a/examples/flow_filtering/main.c
+++ b/examples/flow_filtering/main.c
@@ -55,6 +55,7 @@
 #include <rte_mbuf.h>
 #include <rte_net.h>
 #include <rte_flow.h>
+#include <rte_cycles.h>
 
 static volatile bool force_quit;
 
@@ -119,13 +120,23 @@ main_loop(void)
 	rte_eth_dev_close(port_id);
 }
 
+#define CHECK_INTERVAL 1000  /* 100ms */
+#define MAX_REPEAT_TIMES 90  /* 9s (90 * 100ms) in total */
+
 static void
 assert_link_status(void)
 {
 	struct rte_eth_link link;
+	uint8_t rep_cnt = MAX_REPEAT_TIMES;
 
 	memset(&link, 0, sizeof(link));
-	rte_eth_link_get(port_id, &link);
+	do {
+		rte_eth_link_get(port_id, &link);
+		if (link.link_status == ETH_LINK_UP)
+			break;
+		rte_delay_ms(CHECK_INTERVAL);
+	} while (--rep_cnt);
+
 	if (link.link_status == ETH_LINK_DOWN)
 		rte_exit(EXIT_FAILURE, ":: error: link is still down\n");
 }
-- 
2.5.5

Re: [dpdk-dev] [PATCH v3] examples/flow_filtering: add delay during updating link status

[Ori Kam]

> -----Original Message-----
> From: Beilei Xing [mailto:beilei.xing@intel.com]
> Sent: Wednesday, January 24, 2018 12:16 PM
> To: Ori Kam <orika@mellanox.com>
> Cc: dev@dpdk.org
> Subject: [PATCH v3] examples/flow_filtering: add delay during updating link
> status
> 
> Add up to 9s delay for getting link status to make sure NIC updates
> link status successfully, just like other applications such as
> testpmd and l2fwd.
> 
> Signed-off-by: Beilei Xing <beilei.xing@intel.com>
> ---
> v3 changes:
>  - Modify MAX_REPEAT_TIME with MAX_REPEAT_TIMES
> v2 changes:
>  - Add rte_delay_ms(CHECK_INTERVAL) which is missed in v1.
> 
>  examples/flow_filtering/main.c | 13 ++++++++++++-
>  1 file changed, 12 insertions(+), 1 deletion(-)
> 
> diff --git a/examples/flow_filtering/main.c b/examples/flow_filtering/main.c
> index 4a07b63..d16a0a5 100644
> --- a/examples/flow_filtering/main.c
> +++ b/examples/flow_filtering/main.c
> @@ -55,6 +55,7 @@
>  #include <rte_mbuf.h>
>  #include <rte_net.h>
>  #include <rte_flow.h>
> +#include <rte_cycles.h>
> 
>  static volatile bool force_quit;
> 
> @@ -119,13 +120,23 @@ main_loop(void)
>  	rte_eth_dev_close(port_id);
>  }
> 
> +#define CHECK_INTERVAL 1000  /* 100ms */
> +#define MAX_REPEAT_TIMES 90  /* 9s (90 * 100ms) in total */
> +
>  static void
>  assert_link_status(void)
>  {
>  	struct rte_eth_link link;
> +	uint8_t rep_cnt = MAX_REPEAT_TIMES;
> 
>  	memset(&link, 0, sizeof(link));
> -	rte_eth_link_get(port_id, &link);
> +	do {
> +		rte_eth_link_get(port_id, &link);
> +		if (link.link_status == ETH_LINK_UP)
> +			break;
> +		rte_delay_ms(CHECK_INTERVAL);
> +	} while (--rep_cnt);
> +
>  	if (link.link_status == ETH_LINK_DOWN)
>  		rte_exit(EXIT_FAILURE, ":: error: link is still down\n");
>  }
> --
> 2.5.5

Acked-by: Ori Kam <orika@mellanox.com>

Thanks,
Ori

Re: [dpdk-dev] [PATCH v3] examples/flow_filtering: add delay during updating link status

[Thomas Monjalon]

> > Add up to 9s delay for getting link status to make sure NIC updates
> > link status successfully, just like other applications such as
> > testpmd and l2fwd.
> > 
> > Signed-off-by: Beilei Xing <beilei.xing@intel.com>
> 
> Acked-by: Ori Kam <orika@mellanox.com>

Applied, thanks