* [dpdk-dev] [PATCH] net/mlx5: retry when link update fails
@ 2019-09-04 9:16 Moti Haimovsky
2019-10-16 7:34 ` [dpdk-dev] [PATCH v2] " Moti Haimovsky
0 siblings, 1 reply; 4+ messages in thread
From: Moti Haimovsky @ 2019-09-04 9:16 UTC (permalink / raw)
To: viacheslavo, rasland; +Cc: dev
mlx5_link_update immediately returns when called with no-wait parameter
and its call for retrieving the link status returns with EAGAIN error.
This is too harsh on busy systems where a first call fails with EAGAIN
from time to time.
This patch adds a (very limited) retry on such cases in order to allow
retrieving the link status.
Signed-off-by: Moti Haimovsky <motih@mellanox.com>
---
drivers/net/mlx5/mlx5_defs.h | 3 +++
drivers/net/mlx5/mlx5_ethdev.c | 5 +++--
2 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_defs.h b/drivers/net/mlx5/mlx5_defs.h
index d7440fd..06eb192 100644
--- a/drivers/net/mlx5/mlx5_defs.h
+++ b/drivers/net/mlx5/mlx5_defs.h
@@ -101,6 +101,9 @@
/* Timeout in seconds to get a valid link status. */
#define MLX5_LINK_STATUS_TIMEOUT 10
+/* Number of times to retry retrieving the physical link information. */
+#define MLX5_GET_LINK_STATUS_RETRY_COUNT 3
+
/* Maximum number of UAR pages used by a port,
* These are the size and mask for an array of mutexes used to synchronize
* the access to port's UARs on platforms that do not support 64 bit writes.
diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c
index 33a32fc..aace211 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -978,6 +978,7 @@ int mlx5_fw_version_get(struct rte_eth_dev *dev, char *fw_ver, size_t fw_size)
int ret;
struct rte_eth_link dev_link;
time_t start_time = time(NULL);
+ int retry = MLX5_GET_LINK_STATUS_RETRY_COUNT;
do {
ret = mlx5_link_update_unlocked_gs(dev, &dev_link);
@@ -986,7 +987,7 @@ int mlx5_fw_version_get(struct rte_eth_dev *dev, char *fw_ver, size_t fw_size)
if (ret == 0)
break;
/* Handle wait to complete situation. */
- if (wait_to_complete && ret == -EAGAIN) {
+ if ((wait_to_complete || retry) && ret == -EAGAIN) {
if (abs((int)difftime(time(NULL), start_time)) <
MLX5_LINK_STATUS_TIMEOUT) {
usleep(0);
@@ -998,7 +999,7 @@ int mlx5_fw_version_get(struct rte_eth_dev *dev, char *fw_ver, size_t fw_size)
} else if (ret < 0) {
return ret;
}
- } while (wait_to_complete);
+ } while (wait_to_complete || retry-- > 0);
ret = !!memcmp(&dev->data->dev_link, &dev_link,
sizeof(struct rte_eth_link));
dev->data->dev_link = dev_link;
--
1.8.3.1
^ permalink raw reply [flat|nested] 4+ messages in thread
* [dpdk-dev] [PATCH v2] net/mlx5: retry when link update fails
2019-09-04 9:16 [dpdk-dev] [PATCH] net/mlx5: retry when link update fails Moti Haimovsky
@ 2019-10-16 7:34 ` Moti Haimovsky
2019-10-16 7:42 ` Slava Ovsiienko
2019-10-20 10:07 ` Raslan Darawsheh
0 siblings, 2 replies; 4+ messages in thread
From: Moti Haimovsky @ 2019-10-16 7:34 UTC (permalink / raw)
To: viacheslavo, rasland; +Cc: dev
mlx5_link_update immediately returns when called with no-wait parameter
and its call for retrieving the link status returns with EAGAIN error.
This is too harsh on busy systems where a first call fails with EAGAIN
from time to time.
This patch adds a (very limited) retry on such cases in order to allow
retrieving the link status.
Signed-off-by: Moti Haimovsky <motih@mellanox.com>
---
V2:
* Code rebase
---
drivers/net/mlx5/mlx5_defs.h | 3 +++
drivers/net/mlx5/mlx5_ethdev.c | 5 +++--
2 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/drivers/net/mlx5/mlx5_defs.h b/drivers/net/mlx5/mlx5_defs.h
index 180122d..248ef3c 100644
--- a/drivers/net/mlx5/mlx5_defs.h
+++ b/drivers/net/mlx5/mlx5_defs.h
@@ -105,6 +105,9 @@
/* Timeout in seconds to get a valid link status. */
#define MLX5_LINK_STATUS_TIMEOUT 10
+/* Number of times to retry retrieving the physical link information. */
+#define MLX5_GET_LINK_STATUS_RETRY_COUNT 3
+
/* Maximum number of UAR pages used by a port,
* These are the size and mask for an array of mutexes used to synchronize
* the access to port's UARs on platforms that do not support 64 bit writes.
diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c
index f2b1752..5f05b2b 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -999,6 +999,7 @@ int mlx5_fw_version_get(struct rte_eth_dev *dev, char *fw_ver, size_t fw_size)
int ret;
struct rte_eth_link dev_link;
time_t start_time = time(NULL);
+ int retry = MLX5_GET_LINK_STATUS_RETRY_COUNT;
do {
ret = mlx5_link_update_unlocked_gs(dev, &dev_link);
@@ -1007,7 +1008,7 @@ int mlx5_fw_version_get(struct rte_eth_dev *dev, char *fw_ver, size_t fw_size)
if (ret == 0)
break;
/* Handle wait to complete situation. */
- if (wait_to_complete && ret == -EAGAIN) {
+ if ((wait_to_complete || retry) && ret == -EAGAIN) {
if (abs((int)difftime(time(NULL), start_time)) <
MLX5_LINK_STATUS_TIMEOUT) {
usleep(0);
@@ -1019,7 +1020,7 @@ int mlx5_fw_version_get(struct rte_eth_dev *dev, char *fw_ver, size_t fw_size)
} else if (ret < 0) {
return ret;
}
- } while (wait_to_complete);
+ } while (wait_to_complete || retry-- > 0);
ret = !!memcmp(&dev->data->dev_link, &dev_link,
sizeof(struct rte_eth_link));
dev->data->dev_link = dev_link;
--
1.8.3.1
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [dpdk-dev] [PATCH v2] net/mlx5: retry when link update fails
2019-10-16 7:34 ` [dpdk-dev] [PATCH v2] " Moti Haimovsky
@ 2019-10-16 7:42 ` Slava Ovsiienko
2019-10-20 10:07 ` Raslan Darawsheh
1 sibling, 0 replies; 4+ messages in thread
From: Slava Ovsiienko @ 2019-10-16 7:42 UTC (permalink / raw)
To: Moti Haimovsky, Raslan Darawsheh; +Cc: dev
> -----Original Message-----
> From: Moti Haimovsky <motih@mellanox.com>
> Sent: Wednesday, October 16, 2019 10:34
> To: Slava Ovsiienko <viacheslavo@mellanox.com>; Raslan Darawsheh
> <rasland@mellanox.com>
> Cc: dev@dpdk.org
> Subject: [PATCH v2] net/mlx5: retry when link update fails
>
> mlx5_link_update immediately returns when called with no-wait parameter
> and its call for retrieving the link status returns with EAGAIN error.
> This is too harsh on busy systems where a first call fails with EAGAIN from
> time to time.
> This patch adds a (very limited) retry on such cases in order to allow
> retrieving the link status.
>
> Signed-off-by: Moti Haimovsky <motih@mellanox.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
> ---
> V2:
> * Code rebase
> ---
> drivers/net/mlx5/mlx5_defs.h | 3 +++
> drivers/net/mlx5/mlx5_ethdev.c | 5 +++--
> 2 files changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/mlx5/mlx5_defs.h b/drivers/net/mlx5/mlx5_defs.h
> index 180122d..248ef3c 100644
> --- a/drivers/net/mlx5/mlx5_defs.h
> +++ b/drivers/net/mlx5/mlx5_defs.h
> @@ -105,6 +105,9 @@
> /* Timeout in seconds to get a valid link status. */ #define
> MLX5_LINK_STATUS_TIMEOUT 10
>
> +/* Number of times to retry retrieving the physical link information.
> +*/ #define MLX5_GET_LINK_STATUS_RETRY_COUNT 3
> +
> /* Maximum number of UAR pages used by a port,
> * These are the size and mask for an array of mutexes used to synchronize
> * the access to port's UARs on platforms that do not support 64 bit writes.
> diff --git a/drivers/net/mlx5/mlx5_ethdev.c
> b/drivers/net/mlx5/mlx5_ethdev.c index f2b1752..5f05b2b 100644
> --- a/drivers/net/mlx5/mlx5_ethdev.c
> +++ b/drivers/net/mlx5/mlx5_ethdev.c
> @@ -999,6 +999,7 @@ int mlx5_fw_version_get(struct rte_eth_dev *dev,
> char *fw_ver, size_t fw_size)
> int ret;
> struct rte_eth_link dev_link;
> time_t start_time = time(NULL);
> + int retry = MLX5_GET_LINK_STATUS_RETRY_COUNT;
>
> do {
> ret = mlx5_link_update_unlocked_gs(dev, &dev_link); @@ -
> 1007,7 +1008,7 @@ int mlx5_fw_version_get(struct rte_eth_dev *dev, char
> *fw_ver, size_t fw_size)
> if (ret == 0)
> break;
> /* Handle wait to complete situation. */
> - if (wait_to_complete && ret == -EAGAIN) {
> + if ((wait_to_complete || retry) && ret == -EAGAIN) {
> if (abs((int)difftime(time(NULL), start_time)) <
> MLX5_LINK_STATUS_TIMEOUT) {
> usleep(0);
> @@ -1019,7 +1020,7 @@ int mlx5_fw_version_get(struct rte_eth_dev *dev,
> char *fw_ver, size_t fw_size)
> } else if (ret < 0) {
> return ret;
> }
> - } while (wait_to_complete);
> + } while (wait_to_complete || retry-- > 0);
> ret = !!memcmp(&dev->data->dev_link, &dev_link,
> sizeof(struct rte_eth_link));
> dev->data->dev_link = dev_link;
> --
> 1.8.3.1
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [dpdk-dev] [PATCH v2] net/mlx5: retry when link update fails
2019-10-16 7:34 ` [dpdk-dev] [PATCH v2] " Moti Haimovsky
2019-10-16 7:42 ` Slava Ovsiienko
@ 2019-10-20 10:07 ` Raslan Darawsheh
1 sibling, 0 replies; 4+ messages in thread
From: Raslan Darawsheh @ 2019-10-20 10:07 UTC (permalink / raw)
To: Moti Haimovsky, Slava Ovsiienko; +Cc: dev
Hi,
> -----Original Message-----
> From: Moti Haimovsky <motih@mellanox.com>
> Sent: Wednesday, October 16, 2019 10:34 AM
> To: Slava Ovsiienko <viacheslavo@mellanox.com>; Raslan Darawsheh
> <rasland@mellanox.com>
> Cc: dev@dpdk.org
> Subject: [PATCH v2] net/mlx5: retry when link update fails
>
> mlx5_link_update immediately returns when called with no-wait parameter
> and its call for retrieving the link status returns with EAGAIN error.
> This is too harsh on busy systems where a first call fails with EAGAIN from
> time to time.
> This patch adds a (very limited) retry on such cases in order to allow
> retrieving the link status.
>
> Signed-off-by: Moti Haimovsky <motih@mellanox.com>
> ---
> V2:
> * Code rebase
Patch applied to next-net-mlx,
Kindest regards,
Raslan Darawsheh
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2019-10-20 10:07 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-09-04 9:16 [dpdk-dev] [PATCH] net/mlx5: retry when link update fails Moti Haimovsky
2019-10-16 7:34 ` [dpdk-dev] [PATCH v2] " Moti Haimovsky
2019-10-16 7:42 ` Slava Ovsiienko
2019-10-20 10:07 ` Raslan Darawsheh
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).