From: Adrien Mazarguil <adrien.mazarguil@6wind.com>
To: Yuanhan Liu <yliu@fridaylinux.org>
Cc: stable@dpdk.org
Subject: [dpdk-stable] [PATCH 1/2] net/mlx: remove link update lock
Date: Thu, 31 Aug 2017 11:44:53 +0200 [thread overview]
Message-ID: <26f22c88d0b7481d6e365ca13cc8a275047454de.1504172212.git.adrien.mazarguil@6wind.com> (raw)
In-Reply-To: <cover.1504172212.git.adrien.mazarguil@6wind.com>
From: Matthieu Ternisien d'Ouville <matthieu.tdo@6wind.com>
[ backported from upstream commit 2c960a5116d05cc7c93d00df707655d72d73d230 ]
Retrieving link status information through the link update callback should
be quick and non-blocking.
Mellanox PMDs retrieve this information through ioctl() calls on the
related kernel netdevice. This appears to take a long time to
complete and may cause significant slowdowns in applications.
While these system calls cannot be accelerated, removing the lock on the
private structure allows applications to perform other control operations
from separate threads in the meantime. This function remains safe without
locking as it does not write the private structure, it is only used to
retrieve the name of the netdevice.
Signed-off-by: Matthieu Ternisien d'Ouville <matthieu.tdo@6wind.com>
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
---
drivers/net/mlx4/mlx4.c | 32 ++++++--------------------------
drivers/net/mlx5/mlx5.c | 2 +-
drivers/net/mlx5/mlx5.h | 1 -
drivers/net/mlx5/mlx5_ethdev.c | 28 +++++-----------------------
4 files changed, 12 insertions(+), 51 deletions(-)
diff --git a/drivers/net/mlx4/mlx4.c b/drivers/net/mlx4/mlx4.c
index 4a03d34..22af742 100644
--- a/drivers/net/mlx4/mlx4.c
+++ b/drivers/net/mlx4/mlx4.c
@@ -4836,7 +4836,7 @@ mlx4_allmulticast_disable(struct rte_eth_dev *dev)
}
/**
- * DPDK callback to retrieve physical link information (unlocked version).
+ * DPDK callback to retrieve physical link information.
*
* @param dev
* Pointer to Ethernet device structure.
@@ -4844,9 +4844,9 @@ mlx4_allmulticast_disable(struct rte_eth_dev *dev)
* Wait for request completion (ignored).
*/
static int
-mlx4_link_update_unlocked(struct rte_eth_dev *dev, int wait_to_complete)
+mlx4_link_update(struct rte_eth_dev *dev, int wait_to_complete)
{
- struct priv *priv = mlx4_get_priv(dev);
+ const struct priv *priv = mlx4_get_priv(dev);
struct ethtool_cmd edata = {
.cmd = ETHTOOL_GSET
};
@@ -4854,6 +4854,8 @@ mlx4_link_update_unlocked(struct rte_eth_dev *dev, int wait_to_complete)
struct rte_eth_link dev_link;
int link_speed = 0;
+ /* priv_lock() is not taken to allow concurrent calls. */
+
if (priv == NULL)
return -EINVAL;
(void)wait_to_complete;
@@ -4889,28 +4891,6 @@ mlx4_link_update_unlocked(struct rte_eth_dev *dev, int wait_to_complete)
}
/**
- * DPDK callback to retrieve physical link information.
- *
- * @param dev
- * Pointer to Ethernet device structure.
- * @param wait_to_complete
- * Wait for request completion (ignored).
- */
-static int
-mlx4_link_update(struct rte_eth_dev *dev, int wait_to_complete)
-{
- struct priv *priv = mlx4_get_priv(dev);
- int ret;
-
- if (priv == NULL)
- return -EINVAL;
- priv_lock(priv);
- ret = mlx4_link_update_unlocked(dev, wait_to_complete);
- priv_unlock(priv);
- return ret;
-}
-
-/**
* DPDK callback to change the MTU.
*
* Setting the MTU affects hardware MRU (packets larger than the MTU cannot be
@@ -5426,7 +5406,7 @@ priv_dev_link_status_handler(struct priv *priv, struct rte_eth_dev *dev)
struct rte_eth_link *link = &dev->data->dev_link;
priv->pending_alarm = 0;
- mlx4_link_update_unlocked(dev, 0);
+ mlx4_link_update(dev, 0);
if (((link->link_speed == 0) && link->link_status) ||
((link->link_speed != 0) && !link->link_status)) {
/* Inconsistent status, check again later. */
diff --git a/drivers/net/mlx5/mlx5.c b/drivers/net/mlx5/mlx5.c
index aa9d2dc..6f74e6d 100644
--- a/drivers/net/mlx5/mlx5.c
+++ b/drivers/net/mlx5/mlx5.c
@@ -668,7 +668,7 @@ mlx5_pci_probe(struct rte_pci_driver *pci_drv, struct rte_pci_device *pci_dev)
/* Bring Ethernet device up. */
DEBUG("forcing Ethernet interface up");
priv_set_flags(priv, ~IFF_UP, IFF_UP);
- mlx5_link_update_unlocked(priv->dev, 1);
+ mlx5_link_update(priv->dev, 1);
continue;
port_error:
diff --git a/drivers/net/mlx5/mlx5.h b/drivers/net/mlx5/mlx5.h
index 79b7a60..adb9d79 100644
--- a/drivers/net/mlx5/mlx5.h
+++ b/drivers/net/mlx5/mlx5.h
@@ -188,7 +188,6 @@ int priv_set_flags(struct priv *, unsigned int, unsigned int);
int mlx5_dev_configure(struct rte_eth_dev *);
void mlx5_dev_infos_get(struct rte_eth_dev *, struct rte_eth_dev_info *);
const uint32_t *mlx5_dev_supported_ptypes_get(struct rte_eth_dev *dev);
-int mlx5_link_update_unlocked(struct rte_eth_dev *, int);
int mlx5_link_update(struct rte_eth_dev *, int);
int mlx5_dev_set_mtu(struct rte_eth_dev *, uint16_t);
int mlx5_dev_get_flow_ctrl(struct rte_eth_dev *, struct rte_eth_fc_conf *);
diff --git a/drivers/net/mlx5/mlx5_ethdev.c b/drivers/net/mlx5/mlx5_ethdev.c
index ca981a5..279f6d8 100644
--- a/drivers/net/mlx5/mlx5_ethdev.c
+++ b/drivers/net/mlx5/mlx5_ethdev.c
@@ -680,7 +680,7 @@ mlx5_dev_supported_ptypes_get(struct rte_eth_dev *dev)
}
/**
- * Retrieve physical link information (unlocked version using legacy ioctl).
+ * DPDK callback to retrieve physical link information.
*
* @param dev
* Pointer to Ethernet device structure.
@@ -698,6 +698,8 @@ mlx5_link_update_unlocked_gset(struct rte_eth_dev *dev, int wait_to_complete)
struct rte_eth_link dev_link;
int link_speed = 0;
+ /* priv_lock() is not taken to allow concurrent calls. */
+
(void)wait_to_complete;
if (priv_ifreq(priv, SIOCGIFFLAGS, &ifr)) {
WARN("ioctl(SIOCGIFFLAGS) failed: %s", strerror(errno));
@@ -839,7 +841,7 @@ mlx5_link_update_unlocked_gs(struct rte_eth_dev *dev, int wait_to_complete)
}
/**
- * DPDK callback to retrieve physical link information (unlocked version).
+ * DPDK callback to retrieve physical link information.
*
* @param dev
* Pointer to Ethernet device structure.
@@ -847,7 +849,7 @@ mlx5_link_update_unlocked_gs(struct rte_eth_dev *dev, int wait_to_complete)
* Wait for request completion (ignored).
*/
int
-mlx5_link_update_unlocked(struct rte_eth_dev *dev, int wait_to_complete)
+mlx5_link_update(struct rte_eth_dev *dev, int wait_to_complete)
{
struct utsname utsname;
int ver[3];
@@ -861,26 +863,6 @@ mlx5_link_update_unlocked(struct rte_eth_dev *dev, int wait_to_complete)
}
/**
- * DPDK callback to retrieve physical link information.
- *
- * @param dev
- * Pointer to Ethernet device structure.
- * @param wait_to_complete
- * Wait for request completion (ignored).
- */
-int
-mlx5_link_update(struct rte_eth_dev *dev, int wait_to_complete)
-{
- struct priv *priv = mlx5_get_priv(dev);
- int ret;
-
- priv_lock(priv);
- ret = mlx5_link_update_unlocked(dev, wait_to_complete);
- priv_unlock(priv);
- return ret;
-}
-
-/**
* DPDK callback to change the MTU.
*
* Setting the MTU affects hardware MRU (packets larger than the MTU cannot be
--
2.1.4
next prev parent reply other threads:[~2017-08-31 9:45 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-08-14 12:18 [dpdk-stable] please help backporting some patches to LTS release 16.11.3 Yuanhan Liu
2017-08-31 9:44 ` [dpdk-stable] [PATCH 0/2] back-ported Mellanox fixes for stable Adrien Mazarguil
2017-08-31 9:44 ` Adrien Mazarguil [this message]
2017-08-31 9:44 ` [dpdk-stable] [PATCH 2/2] net/mlx4: fix assertion failure on link update Adrien Mazarguil
2017-09-01 2:00 ` [dpdk-stable] [PATCH 0/2] back-ported Mellanox fixes for stable Yuanhan Liu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=26f22c88d0b7481d6e365ca13cc8a275047454de.1504172212.git.adrien.mazarguil@6wind.com \
--to=adrien.mazarguil@6wind.com \
--cc=stable@dpdk.org \
--cc=yliu@fridaylinux.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).