[Bug 1068] [20.11] creating a rule with a shared action after port restart causes a segfault

[Bug 1068] [20.11] creating a rule with a shared action after port restart causes a segfault

[bugzilla]

https://bugs.dpdk.org/show_bug.cgi?id=1068

            Bug ID: 1068
           Summary: [20.11] creating a rule with a shared action after
                    port restart causes a segfault
           Product: DPDK
           Version: 20.11
          Hardware: All
                OS: All
            Status: UNCONFIRMED
          Severity: normal
          Priority: Normal
         Component: ethdev
          Assignee: dev@dpdk.org
          Reporter: alialnu@nvidia.com
  Target Milestone: ---

To reproduce:

$ sudo ./build/app/dpdk-testpmd -n 4  -a 0000:08:00.0 -- -i --rxq=4 --txq=4
--hairpinq=8 --hairpin-mode=0x12
[..]
testpmd> flow shared_action 0 destroy action_id 1
testpmd> flow shared_action 0 create action_id 1 ingress action rss queues 9 8
end level 2 types ipv6 l3-dst-only end func default key_len 40 / end
Shared action #1 created
testpmd> port stop 0
testpmd> port start 0
testpmd> flow shared_action 0 destroy action_id 1
port_flow_complain(): Caught PMD error type 16 (specific action): invalid
shared action: Invalid argument
testpmd> flow validate 0 priority 0 ingress group 0 pattern eth / end actions
set_mac_dst mac_addr 00:16:3e:3f:0f:4f / shared 1 / end
Segmentation fault

OS: Ubuntu 20.04
OFED: MLNX_OFED_LINUX-5.7-1.0.2.0
Device: ConnectX-5
Firmware: 16.34.1002

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 1068] [20.11] creating a rule with a shared action after port restart causes a segfault

[bugzilla]

https://bugs.dpdk.org/show_bug.cgi?id=1068

Dmitry Kozlyuk (dkozlyuk@nvidia.com) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |WONTFIX
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #1 from Dmitry Kozlyuk (dkozlyuk@nvidia.com) ---
Before 21.11 it was unspecified whether indirect actions are kept across port
restart. Since 21.11 there is RTE_ETH_DEV_CAPA_SHARED_OBJECT_KEEP to indicate
the behavior. This is a new feature that cannot be backported. The only way to
ensure predictable behavior with any PMD before 21.11 is to destroy indirect
actions before port stop.

-- 
You are receiving this mail because:
You are the assignee for the bug.