DPDK patches and discussions
 help / color / mirror / Atom feed
From: Andrew Rybchenko <arybchenko@solarflare.com>
To: Olivier Matz <olivier.matz@6wind.com>, <dev@dpdk.org>
Cc: Thomas Monjalon <thomas@monjalon.net>,
	Ferruh Yigit <ferruh.yigit@intel.com>,
	Ivan Malov <Ivan.Malov@oktetlabs.ru>
Subject: Re: [dpdk-dev] [PATCH] ethdev: return diagnostic when setting MAC address
Date: Fri, 16 Mar 2018 18:41:51 +0300	[thread overview]
Message-ID: <749ef7af-dde0-529e-f80c-e0b76408cd4c@solarflare.com> (raw)
In-Reply-To: <20180227151129.30387-1-olivier.matz@6wind.com>

Hi Olivier,

On 02/27/2018 06:11 PM, Olivier Matz wrote:
> Change the prototype and the behavior of dev_ops->eth_mac_addr_set(): a
> return code is added to notify the caller (librte_ether) if an error
> occurred in the PMD.
>
> The new default MAC address is now copied in dev->data->mac_addrs[0]
> only if the operation is successful.
>
> The patch also updates all the PMDs accordingly.
>
> Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
> ---
>
> Hi,
>
> This patch is the following of the discussion we had in this thread:
> https://dpdk.org/dev/patchwork/patch/32284/
>
> I did my best to keep the consistency inside the PMDs. The behavior
> of eth_mac_addr_set() is inspired from other fonctions in the same
> PMD, usually eth_mac_addr_add(). For instance:
> - dpaa and dpaa2 return 0 on error.
> - some PMDs (bnxt, mlx5, ...?) do not return a -errno code (-1 or
>    positive values).
> - some PMDs (avf, tap) check if the address is the same and return 0
>    in that case. This could go in generic code?
>
> I tried to use the following errors when relevant:
> - -EPERM when a VF is not allowed to do a change
> - -ENOTSUP if the function is not supported
> - -EIO if this is an unknown error from lower layer (hw or sdk)
> - -EINVAL for other unknown errors
>
> Please, PMD maintainers, feel free to comment if you ahve specific
> needs for your driver.
>
> Thanks
> Olivier
>
>
>   doc/guides/rel_notes/deprecation.rst    |  8 --------
>   drivers/net/ark/ark_ethdev.c            |  9 ++++++---
>   drivers/net/avf/avf_ethdev.c            | 12 ++++++++----
>   drivers/net/bnxt/bnxt_ethdev.c          | 10 ++++++----
>   drivers/net/bonding/rte_eth_bond_pmd.c  |  8 ++++++--
>   drivers/net/dpaa/dpaa_ethdev.c          |  4 +++-
>   drivers/net/dpaa2/dpaa2_ethdev.c        |  6 ++++--
>   drivers/net/e1000/igb_ethdev.c          | 12 +++++++-----
>   drivers/net/failsafe/failsafe_ops.c     | 16 +++++++++++++---
>   drivers/net/i40e/i40e_ethdev.c          | 24 ++++++++++++++---------
>   drivers/net/i40e/i40e_ethdev_vf.c       | 12 +++++++-----
>   drivers/net/ixgbe/ixgbe_ethdev.c        | 13 ++++++++-----
>   drivers/net/mlx4/mlx4.h                 |  2 +-
>   drivers/net/mlx4/mlx4_ethdev.c          |  7 +++++--
>   drivers/net/mlx5/mlx5.h                 |  2 +-
>   drivers/net/mlx5/mlx5_mac.c             |  7 +++++--
>   drivers/net/mrvl/mrvl_ethdev.c          |  7 ++++++-
>   drivers/net/null/rte_eth_null.c         |  3 ++-
>   drivers/net/octeontx/octeontx_ethdev.c  |  4 +++-
>   drivers/net/qede/qede_ethdev.c          |  7 +++----
>   drivers/net/sfc/sfc_ethdev.c            | 14 +++++---------
>   drivers/net/szedata2/rte_eth_szedata2.c |  3 ++-
>   drivers/net/tap/rte_eth_tap.c           | 34 +++++++++++++++++++++------------
>   drivers/net/virtio/virtio_ethdev.c      | 15 ++++++++++-----
>   drivers/net/vmxnet3/vmxnet3_ethdev.c    |  5 +++--
>   lib/librte_ether/rte_ethdev.c           |  7 +++++--
>   lib/librte_ether/rte_ethdev_core.h      |  2 +-
>   test/test/virtual_pmd.c                 |  3 ++-
>   28 files changed, 159 insertions(+), 97 deletions(-)

<snip>

> diff --git a/drivers/net/sfc/sfc_ethdev.c b/drivers/net/sfc/sfc_ethdev.c
> index 89a452907..b76d4bb2c 100644
> --- a/drivers/net/sfc/sfc_ethdev.c
> +++ b/drivers/net/sfc/sfc_ethdev.c
> @@ -920,7 +920,7 @@ sfc_dev_set_mtu(struct rte_eth_dev *dev, uint16_t mtu)
>   	SFC_ASSERT(rc > 0);
>   	return -rc;
>   }
> -static void
> +static int
>   sfc_mac_addr_set(struct rte_eth_dev *dev, struct ether_addr *mac_addr)
>   {
>   	struct sfc_adapter *sa = dev->data->dev_private;
> @@ -939,13 +939,14 @@ sfc_mac_addr_set(struct rte_eth_dev *dev, struct ether_addr *mac_addr)
>   	if (port->isolated) {
>   		sfc_err(sa, "isolated mode is active on the port");
>   		sfc_err(sa, "will not set MAC address");
> +		rc = ENOTSUP;
>   		goto unlock;
>   	}
>   
>   	if (sa->state != SFC_ADAPTER_STARTED) {
>   		sfc_info(sa, "the port is not started");
>   		sfc_info(sa, "the new MAC address will be set on port start");
> -
> +		rc = EINVAL;
>   		goto unlock;
>   	}
>   
> @@ -982,14 +983,9 @@ sfc_mac_addr_set(struct rte_eth_dev *dev, struct ether_addr *mac_addr)
>   	}
>   
>   unlock:
> -	/*
> -	 * In the case of failure sa->port->default_mac_addr does not
> -	 * need rollback since no error code is returned, and the upper
> -	 * API will anyway update the external MAC address storage.
> -	 * To be consistent with that new value it is better to keep
> -	 * the device private value the same.
> -	 */
>   	sfc_adapter_unlock(sa);
> +	SFC_ASSERT(rc >= 0);
> +	return -rc;
>   }
>   
>   

It is a bit more complicated for sfc. Please, find the patch from Ivan 
below:
  - it is not an error if isolated mode - we just need to save a new 
value locally
  - it requires careful rollback in the case of failure

===

diff --git a/drivers/net/sfc/sfc_ethdev.c b/drivers/net/sfc/sfc_ethdev.c
index 89a4529..f5670a9 100644
--- a/drivers/net/sfc/sfc_ethdev.c
+++ b/drivers/net/sfc/sfc_ethdev.c
@@ -920,13 +920,14 @@
         SFC_ASSERT(rc > 0);
         return -rc;
  }
-static void
+static int
  sfc_mac_addr_set(struct rte_eth_dev *dev, struct ether_addr *mac_addr)
  {
         struct sfc_adapter *sa = dev->data->dev_private;
         const efx_nic_cfg_t *encp = efx_nic_cfg_get(sa->nic);
         struct sfc_port *port = &sa->port;
-       int rc;
+       struct ether_addr *old_addr = &dev->data->mac_addrs[0];
+       int rc = 0;

         sfc_adapter_lock(sa);

@@ -936,9 +937,16 @@
          */
         ether_addr_copy(mac_addr, &port->default_mac_addr);

+       /*
+        * Neither of the two following checks can return
+        * an error. The new MAC address is preserved in
+        * the device private data and can be activated
+        * on the next port start if the user prevents
+        * isolated mode from being enabled.
+        */
         if (port->isolated) {
-               sfc_err(sa, "isolated mode is active on the port");
-               sfc_err(sa, "will not set MAC address");
+               sfc_warn(sa, "isolated mode is active on the port");
+               sfc_warn(sa, "will not set MAC address");
                 goto unlock;
         }

@@ -962,8 +970,13 @@
                  * we also need to update unicast filters
                  */
                 rc = sfc_set_rx_mode(sa);
-               if (rc != 0)
+               if (rc != 0) {
                         sfc_err(sa, "cannot set filter (rc = %u)", rc);
+
+                       /* Rollback the old address */
+                       (void)efx_mac_addr_set(sa->nic, 
old_addr->addr_bytes);
+                       (void)sfc_set_rx_mode(sa);
+               }
         } else {
                 sfc_warn(sa, "cannot set MAC address with filters 
installed");
                 sfc_warn(sa, "adapter will be restarted to pick the new 
MAC");
@@ -982,14 +995,13 @@
         }

  unlock:
-       /*
-        * In the case of failure sa->port->default_mac_addr does not
-        * need rollback since no error code is returned, and the upper
-        * API will anyway update the external MAC address storage.
-        * To be consistent with that new value it is better to keep
-        * the device private value the same.
-        */
+       if (rc != 0)
+               ether_addr_copy(old_addr, &port->default_mac_addr);
+
         sfc_adapter_unlock(sa);
+
+       SFC_ASSERT(rc >= 0);
+       return -rc;
  }


===

<snip>

  parent reply	other threads:[~2018-03-16 15:42 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-02-27 15:11 Olivier Matz
2018-03-05 10:29 ` Adrien Mazarguil
2018-03-06  9:37 ` Tomasz Duszynski
2018-03-16 15:41 ` Andrew Rybchenko [this message]
2018-03-26 18:39 ` Ferruh Yigit
2018-03-28  8:24   ` Olivier Matz
2018-04-03 12:41 ` [dpdk-dev] [PATCH v2] " Olivier Matz
2018-04-03 13:02   ` Andrew Rybchenko
2018-04-03 14:57   ` Adrien Mazarguil
2018-04-03 16:26     ` Olivier Matz
2018-04-03 16:19   ` Ferruh Yigit
2018-04-03 16:25     ` Olivier Matz
2018-04-04  8:19   ` Shreyansh Jain
2018-04-06 15:21   ` [dpdk-dev] [PATCH] " Olivier Matz
2018-04-06 15:34     ` Olivier Matz
2018-04-09  8:57     ` Nélio Laranjeiro
2018-04-06 15:34   ` [dpdk-dev] [PATCH v3] " Olivier Matz
2018-04-06 16:03     ` Ferruh Yigit
2018-04-10 13:19       ` Thomas Monjalon
2018-04-11 16:32     ` [dpdk-dev] [PATCH v4] " Olivier Matz
2018-04-11 19:30       ` Ferruh Yigit

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=749ef7af-dde0-529e-f80c-e0b76408cd4c@solarflare.com \
    --to=arybchenko@solarflare.com \
    --cc=Ivan.Malov@oktetlabs.ru \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=olivier.matz@6wind.com \
    --cc=thomas@monjalon.net \
    /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).