* [dpdk-dev] [PATCH] net/mlx5:fix storing the synced MAC to internal table
@ 2020-12-09 13:42 Souvik Dey
2020-12-09 13:42 ` Souvik Dey
0 siblings, 1 reply; 5+ messages in thread
From: Souvik Dey @ 2020-12-09 13:42 UTC (permalink / raw)
To: rasland, matan, shahafs, viacheslavo; +Cc: dev, Souvik Dey
During the mlx5_dev_spawn(), it tries to sync the MAC address of the
Device to the internal MAC table. In case of SR-IOV VF , the
mlx5_nl_mac_addr_sync() gets, 2 MAC address in the fresh start.
One the unicast MAC assigned to the VF on the host and the second
the multicast mac of 33:33:00:00:00:01. Currently without check the
type of MAC we try to add the MAC 33:33:00:00:00:01 to the first free
place in the MAC table which will be mostly index 1, as index 0 stores
the unicast MAC assigned to the VF. This causes issues subsequently
when we try to add multicast address list. As every IPv6 IP generates
differnet multicast MAC, to support multiple IP on a single interface,
we need to add a list of multicast MAC. As an example, if we try to
add 3 multicast MAC as mentioned below
mcast addr 0 - 33:33:00:00:00:01
mcast addr 1 - 01:00:5e:00:00:01
mcast addr 2 - 33:33:ff:68:fe:7d
then, the addition of these MAC will fail with error -EADDRINUSE in
mlx5_internal_mac_addr_add(), and none of the MAC address will be
added to the device, which will lead of packet drop in case multicast
promiscuous is not enable.
To make this work, we should store the synced MAC to the proper section
of the internal table, after checking the type of the MAC synced. This
check will make the MAC of 33:33:00:00:00:01(synced MAC) to be stored at
index 128(starting of MLX5_MAX_MC_MAC_ADDRESSES), which will be removed
and added again during the mlx5_set_mc_addr_list(). This will make sure
that all the multicast MACs are added properly to both the internal table
and also the device, specially in case of SR-IOV VF.
Souvik Dey (1):
net/mlx5:fix storing the synced MAC to internal table
drivers/common/mlx5/linux/mlx5_nl.c | 20 +++++++++++++++-----
1 file changed, 15 insertions(+), 5 deletions(-)
--
2.9.3.windows.1
^ permalink raw reply [flat|nested] 5+ messages in thread
* [dpdk-dev] [PATCH] net/mlx5:fix storing the synced MAC to internal table
2020-12-09 13:42 [dpdk-dev] [PATCH] net/mlx5:fix storing the synced MAC to internal table Souvik Dey
@ 2020-12-09 13:42 ` Souvik Dey
0 siblings, 0 replies; 5+ messages in thread
From: Souvik Dey @ 2020-12-09 13:42 UTC (permalink / raw)
To: rasland, matan, shahafs, viacheslavo; +Cc: dev, Souvik Dey
As the internal MAC table is divided into Unicast and Multicast address
section, we should check the type of synced MAC address before storing
it to the internal table. Currently the check is not done, and the
synced MAC of 33:33:00:00:00:01 gets stored in the unicast section
(mostly index 1) which causes all subsequent mlx5_set_mc_addr_list()
to fail with error -EADDRINUSE, as the mac_list conatins the MAC
33:33:00:00:00:01. This denies adding of any new multicast address to
the internal list and also fails to add the MAC address to the device
in case of SR-IOV VF case.
Signed-off-by: Souvik Dey <sodey@rbbn.com>
---
drivers/common/mlx5/linux/mlx5_nl.c | 20 +++++++++++++++-----
1 file changed, 15 insertions(+), 5 deletions(-)
diff --git a/drivers/common/mlx5/linux/mlx5_nl.c b/drivers/common/mlx5/linux/mlx5_nl.c
index 40d8620..da56add 100644
--- a/drivers/common/mlx5/linux/mlx5_nl.c
+++ b/drivers/common/mlx5/linux/mlx5_nl.c
@@ -758,11 +758,21 @@ mlx5_nl_mac_addr_sync(int nlsk_fd, unsigned int iface_idx,
break;
if (j != n)
continue;
- /* Find the first entry available. */
- for (j = 0; j != n; ++j) {
- if (rte_is_zero_ether_addr(&mac_addrs[j])) {
- mac_addrs[j] = macs[i];
- break;
+ if (is_multicast_ether_addr(&macs[i])) {
+ /* Find the first entry available. */
+ for (j = MLX5_MAX_UC_MAC_ADDRESSES; j != n; ++j) {
+ if (rte_is_zero_ether_addr(&mac_addrs[j])) {
+ mac_addrs[j] = macs[i];
+ break;
+ }
+ }
+ } else {
+ /* Find the first entry available. */
+ for (j = 0; j != MLX5_MAX_UC_MAC_ADDRESSES; ++j) {
+ if (rte_is_zero_ether_addr(&mac_addrs[j])) {
+ mac_addrs[j] = macs[i];
+ break;
+ }
}
}
}
--
2.9.3.windows.1
^ permalink raw reply [flat|nested] 5+ messages in thread
* [dpdk-dev] [PATCH] net/mlx5:fix storing the synced MAC to internal table
@ 2020-12-09 15:11 Souvik Dey
2020-12-09 15:11 ` Souvik Dey
0 siblings, 1 reply; 5+ messages in thread
From: Souvik Dey @ 2020-12-09 15:11 UTC (permalink / raw)
To: rasland, matan, shahafs, viacheslavo; +Cc: dev, Souvik Dey
During the mlx5_dev_spawn(), it tries to sync the MAC address of the
Device to the internal MAC table. In case of SR-IOV VF , the
mlx5_nl_mac_addr_sync() gets, 2 MAC address in the fresh start.
One the unicast MAC assigned to the VF on the host and the second
the multicast mac of 33:33:00:00:00:01. Currently without check the
type of MAC we try to add the MAC 33:33:00:00:00:01 to the first free
place in the MAC table which will be mostly index 1, as index 0 stores
the unicast MAC assigned to the VF. This causes issues subsequently
when we try to add multicast address list. As every IPv6 IP generates
differnet multicast MAC, to support multiple IP on a single interface,
we need to add a list of multicast MAC. As an example, if we try to
add 3 multicast MAC as mentioned below
mcast addr 0 - 33:33:00:00:00:01
mcast addr 1 - 01:00:5e:00:00:01
mcast addr 2 - 33:33:ff:68:fe:7d
then, the addition of these MAC will fail with error -EADDRINUSE in
mlx5_internal_mac_addr_add(), and none of the MAC address will be
added to the device, which will lead of packet drop in case multicast
promiscuous is not enable.
To make this work, we should store the synced MAC to the proper section
of the internal table, after checking the type of the MAC synced. This
check will make the MAC of 33:33:00:00:00:01(synced MAC) to be stored at
index 128(starting of MLX5_MAX_MC_MAC_ADDRESSES), which will be removed
and added again during the mlx5_set_mc_addr_list(). This will make sure
that all the multicast MACs are added properly to both the internal table
and also the device, specially in case of SR-IOV VF.
Souvik Dey (1):
net/mlx5:fix storing the synced MAC to internal table
drivers/common/mlx5/linux/mlx5_nl.c | 20 +++++++++++++++-----
1 file changed, 15 insertions(+), 5 deletions(-)
--
2.9.3.windows.1
^ permalink raw reply [flat|nested] 5+ messages in thread
* [dpdk-dev] [PATCH] net/mlx5:fix storing the synced MAC to internal table
2020-12-09 15:11 Souvik Dey
@ 2020-12-09 15:11 ` Souvik Dey
2021-01-21 16:56 ` Slava Ovsiienko
0 siblings, 1 reply; 5+ messages in thread
From: Souvik Dey @ 2020-12-09 15:11 UTC (permalink / raw)
To: rasland, matan, shahafs, viacheslavo; +Cc: dev, Souvik Dey
As the internal MAC table is divided into Unicast and Multicast address
section, we should check the type of synced MAC address before storing
it to the internal table. Currently the check is not done, and the
synced MAC of 33:33:00:00:00:01 gets stored in the unicast section
(mostly index 1) which causes all subsequent mlx5_set_mc_addr_list()
to fail with error -EADDRINUSE, as the mac_list contains the MAC
33:33:00:00:00:01. This denies adding of any new multicast address to
the internal list and also fails to add the MAC address to the device
in case of SR-IOV VF case.
Signed-off-by: Souvik Dey <sodey@rbbn.com>
---
drivers/common/mlx5/linux/mlx5_nl.c | 20 +++++++++++++++-----
1 file changed, 15 insertions(+), 5 deletions(-)
diff --git a/drivers/common/mlx5/linux/mlx5_nl.c b/drivers/common/mlx5/linux/mlx5_nl.c
index 40d8620..ef7a521 100644
--- a/drivers/common/mlx5/linux/mlx5_nl.c
+++ b/drivers/common/mlx5/linux/mlx5_nl.c
@@ -758,11 +758,21 @@ mlx5_nl_mac_addr_sync(int nlsk_fd, unsigned int iface_idx,
break;
if (j != n)
continue;
- /* Find the first entry available. */
- for (j = 0; j != n; ++j) {
- if (rte_is_zero_ether_addr(&mac_addrs[j])) {
- mac_addrs[j] = macs[i];
- break;
+ if (rte_is_multicast_ether_addr(&macs[i])) {
+ /* Find the first entry available. */
+ for (j = MLX5_MAX_UC_MAC_ADDRESSES; j != n; ++j) {
+ if (rte_is_zero_ether_addr(&mac_addrs[j])) {
+ mac_addrs[j] = macs[i];
+ break;
+ }
+ }
+ } else {
+ /* Find the first entry available. */
+ for (j = 0; j != MLX5_MAX_UC_MAC_ADDRESSES; ++j) {
+ if (rte_is_zero_ether_addr(&mac_addrs[j])) {
+ mac_addrs[j] = macs[i];
+ break;
+ }
}
}
}
--
2.9.3.windows.1
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [dpdk-dev] [PATCH] net/mlx5:fix storing the synced MAC to internal table
2020-12-09 15:11 ` Souvik Dey
@ 2021-01-21 16:56 ` Slava Ovsiienko
0 siblings, 0 replies; 5+ messages in thread
From: Slava Ovsiienko @ 2021-01-21 16:56 UTC (permalink / raw)
To: Souvik Dey, Raslan Darawsheh, Matan Azrad, Shahaf Shuler; +Cc: dev
Hi, Souvik
Thank you for the patch, please see my comments below.
>From: Souvik Dey <sodey@rbbn.com>
>Sent: Wednesday, December 9, 2020 17:11
>To: Raslan Darawsheh <rasland@nvidia.com>; Matan Azrad <matan@nvidia.com>; Shahaf Shuler ><shahafs@nvidia.com>; Slava Ovsiienko <viacheslavo@nvidia.com>
>Cc: dev@dpdk.org; Souvik Dey <sodey@rbbn.com>
>Subject: [PATCH] net/mlx5:fix storing the synced MAC to internal table
We have the warnings:
Wrong headline format:
net/mlx5:fix storing the synced MAC to internal table
Wrong headline prefix:
net/mlx5:fix storing the synced MAC to internal table
Missing 'Fixes' tag:
net/mlx5:fix storing the synced MAC to internal table
The patch headline should be:
common/mlx5: fix storing the synched MAC to internal table
- net/ -> common/
- space after mlx5:
- synced -> synched
The patch should contain the Fixes tag:
Fixes: f22442cb5d42 (“net/mlx5: reduce Netlink commands dependencies”)
Fixes: ccdcba53a3f4 (“net/mlx5: use Netlink to add/remove MAC addresses”)
The patch should contain (to be sent to LTS ML):
Cc: stable@dpdk.org
>As the internal MAC table is divided into Unicast and Multicast address
>section, we should check the type of synced MAC address before storing
section -> sections
synced -> synched
>it to the internal table. Currently the check is not done, and the
>synced MAC of 33:33:00:00:00:01 gets stored in the unicast section
sync ->synched
>(mostly index 1) which causes all subsequent mlx5_set_mc_addr_list()
reword? : which causes -> causing
>to fail with error -EADDRINUSE, as the mac_list contains the MAC
>33:33:00:00:00:01. This denies adding of any new multicast address to
>the internal list and also fails to add the MAC address to the device
>in case of SR-IOV VF case.
typo: case (to remove)
With best regards,
Slava
diff --git a/drivers/common/mlx5/linux/mlx5_nl.c b/drivers/common/mlx5/linux/mlx5_nl.c
index 40d8620..ef7a521 100644
--- a/drivers/common/mlx5/linux/mlx5_nl.c
+++ b/drivers/common/mlx5/linux/mlx5_nl.c
@@ -758,11 +758,21 @@ mlx5_nl_mac_addr_sync(int nlsk_fd, unsigned int iface_idx,
break;
if (j != n)
continue;
- /* Find the first entry available. */
- for (j = 0; j != n; ++j) {
- if (rte_is_zero_ether_addr(&mac_addrs[j])) {
- mac_addrs[j] = macs[i];
- break;
+ if (rte_is_multicast_ether_addr(&macs[i])) {
+ /* Find the first entry available. */
+ for (j = MLX5_MAX_UC_MAC_ADDRESSES; j != n; ++j) {
+ if (rte_is_zero_ether_addr(&mac_addrs[j])) {
+ mac_addrs[j] = macs[i];
+ break;
+ }
+ }
+ } else {
+ /* Find the first entry available. */
+ for (j = 0; j != MLX5_MAX_UC_MAC_ADDRESSES; ++j) {
+ if (rte_is_zero_ether_addr(&mac_addrs[j])) {
+ mac_addrs[j] = macs[i];
+ break;
+ }
}
}
}
--
2.9.3.windows.1
________________________________________
Notice: This e-mail together with any attachments may contain information of Ribbon Communications Inc. that is confidential and/or proprietary for the sole use of the intended recipient. Any review, disclosure, reliance or distribution by others or forwarding without express permission is strictly prohibited. If you are not the intended recipient, please notify the sender immediately and then delete all copies, including any attachments.
________________________________________
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2021-01-21 16:56 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-12-09 13:42 [dpdk-dev] [PATCH] net/mlx5:fix storing the synced MAC to internal table Souvik Dey
2020-12-09 13:42 ` Souvik Dey
2020-12-09 15:11 Souvik Dey
2020-12-09 15:11 ` Souvik Dey
2021-01-21 16:56 ` Slava Ovsiienko
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).