[v2] lib/distributor: fix livelock issue on flush
Checks
Commit Message
From: Liang Ma <liang.j.ma@intel.com>
The Distributor autotest can lock if ran enough times. Worker and
distributir threads get into a livelock situation waiting on each
other.
Issue first encountered by RedHat in Travis CI
https://bugs.dpdk.org/show_bug.cgi?id=316
To repeat:
`while sudo sh -c "echo 'distributor_autotest' |
./build/app/test/dpdk-test"; do :; done`
The root cause is where we are flushing on exit, and do not wait for
all worker packets to be returned before exiting.
Add a delay on flush so that all worker packets are returned before
completing the flush.
Bugzilla ID: 316
Fixes: 775003ad2f96 ("distributor: add new burst-capable library")
Cc: stable@dpdk.org
Tested-by: Michael Santana <msantana@redhat.com>
Signed-off-by: David Hunt <david.hunt@intel.com>
Signed-off-by: Liang Ma <liang.j.ma@intel.com>
---
v2 add correct patch attribution
---
lib/librte_distributor/rte_distributor.c | 3 +++
1 file changed, 3 insertions(+)
Comments
25/07/2019 11:03, David Hunt:
> From: Liang Ma <liang.j.ma@intel.com>
>
> The Distributor autotest can lock if ran enough times. Worker and
> distributir threads get into a livelock situation waiting on each
> other.
> Issue first encountered by RedHat in Travis CI
> https://bugs.dpdk.org/show_bug.cgi?id=316
>
> To repeat:
> `while sudo sh -c "echo 'distributor_autotest' |
> ./build/app/test/dpdk-test"; do :; done`
>
> The root cause is where we are flushing on exit, and do not wait for
> all worker packets to be returned before exiting.
> Add a delay on flush so that all worker packets are returned before
> completing the flush.
>
> Bugzilla ID: 316
> Fixes: 775003ad2f96 ("distributor: add new burst-capable library")
> Cc: stable@dpdk.org
>
> Tested-by: Michael Santana <msantana@redhat.com>
> Signed-off-by: David Hunt <david.hunt@intel.com>
> Signed-off-by: Liang Ma <liang.j.ma@intel.com>
Applied, thanks
@@ -542,6 +542,9 @@ rte_distributor_flush_v1705(struct rte_distributor *d)
while (total_outstanding(d) > 0)
rte_distributor_process(d, NULL, 0);
+ /* wait 10ms to allow all worker drain the pkts */
+ rte_delay_us(10000);
+
/*
* Send empty burst to all workers to allow them to exit
* gracefully, should they need to.