Re: [PATCH] ixgbe: eliminate duplicate filterlist symbols

On 8/28/2017 6:13 PM, David Harton (dharton) wrote:

quoted

-----Original Message-----
From: Ferruh Yigit [mailto:ferruh.yigit@intel.com]
Sent: Monday, August 28, 2017 9:27 AM
To: David Harton (dharton) <redacted>;
konstantin.ananyev@intel.com
Cc: dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH] ixgbe: eliminate duplicate filterlist
symbols

On 8/25/2017 8:21 PM, David Harton wrote:

quoted

Some compilers generate warnings for duplicate symbols for the set of
filter lists current defined in ixgbe_ethdev.h.
This commits moves the defintion and declaration to the source file
that actually uses them and provides a function to initialize the
values akin to its flush function.

Signed-off-by: David Harton <redacted>
---
 drivers/net/ixgbe/ixgbe_ethdev.c |  8 ++------
drivers/net/ixgbe/ixgbe_ethdev.h |  7 +------
 drivers/net/ixgbe/ixgbe_flow.c   | 18 ++++++++++++++++++
 3 files changed, 21 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c

b/drivers/net/ixgbe/ixgbe_ethdev.c
index 1ec5aaf..ed21af5 100644

--- a/drivers/net/ixgbe/ixgbe_ethdev.c
+++ b/drivers/net/ixgbe/ixgbe_ethdev.c

@@ -1332,12 +1332,8 @@ struct rte_ixgbe_xstats_name_off {
 	/* initialize l2 tunnel filter list & hash */
 	ixgbe_l2_tn_filter_init(eth_dev);

-	TAILQ_INIT(&filter_ntuple_list);
-	TAILQ_INIT(&filter_ethertype_list);
-	TAILQ_INIT(&filter_syn_list);
-	TAILQ_INIT(&filter_fdir_list);
-	TAILQ_INIT(&filter_l2_tunnel_list);
-	TAILQ_INIT(&ixgbe_flow_list);
+	/* initialize flow filter lists */
+	ixgbe_filterlist_init();

Will it work if list initialization converted to the static
initialization? That would remove requirement of this function call.

I was trying to keep the behavior the same:
  .probe -> eth_ixgbe_dev_init -> ixgbe_filterlist_init
  .remove -> eth_ixgbe_dev_uninit -> ixgbe_filterlist_flush

While I think the below would work I always like mirror/parallel relationships like ctor/dtor for objects.

I would agree if ixgbe_filterlist_init() initializes something more than
list itself, but for this case initialization can be avoided completely
which I would prefer.

Driver maintainer also cc'ed for more comment.

Is that reasonable?

Thanks,
Dave

quoted

something like:

diff --git a/drivers/net/ixgbe/ixgbe_flow.c

b/drivers/net/ixgbe/ixgbe_flow.c index c8645f056..6184eb4af 100644

--- a/drivers/net/ixgbe/ixgbe_flow.c
+++ b/drivers/net/ixgbe/ixgbe_flow.c

@@ -79,6 +79,24 @@
 #define IXGBE_MAX_N_TUPLE_PRIO 7
 #define IXGBE_MAX_FLX_SOURCE_OFF 62

+static TAILQ_HEAD(ixgbe_ntuple_filter_list, ixgbe_ntuple_filter_ele)
+       filter_ntuple_list = TAILQ_HEAD_INITIALIZER(filter_ntuple_list);
+
+static TAILQ_HEAD(ixgbe_ethertype_filter_list,

ixgbe_ethertype_filter_ele)
+       filter_ethertype_list =
TAILQ_HEAD_INITIALIZER(filter_ethertype_list);
+
+static TAILQ_HEAD(ixgbe_syn_filter_list, ixgbe_eth_syn_filter_ele)
+       filter_syn_list = TAILQ_HEAD_INITIALIZER(filter_syn_list);
+
+static TAILQ_HEAD(ixgbe_fdir_rule_filter_list, ixgbe_fdir_rule_ele)
+       filter_fdir_list = TAILQ_HEAD_INITIALIZER(filter_fdir_list);
+
+static TAILQ_HEAD(ixgbe_l2_tunnel_filter_list,
ixgbe_eth_l2_tunnel_conf_ele)
+       filter_l2_tunnel_list =
TAILQ_HEAD_INITIALIZER(filter_l2_tunnel_list);
+
+static TAILQ_HEAD(ixgbe_flow_mem_list, ixgbe_flow_mem)
+       ixgbe_flow_list = TAILQ_HEAD_INITIALIZER(ixgbe_flow_list);
+