lib/distributor: fix livelock issue on flush
PATCH 1/1 lib/distributor: fix livelock issue on flush
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 my 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
--- lib/librte_distributor/rte_distributor.c | 3 +++ 1 file changed, 3 insertions(+)
lib/librte_distributor/rte_distributor.c +3 -0
@@ -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.
Signature · 1 line
2.17.1
PATCH v2 1/1 lib/distributor: fix livelock issue on flush
From: Liang Ma <[email protected]>
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
---
v2 add correct patch attribution
--- lib/librte_distributor/rte_distributor.c | 3 +++ 1 file changed, 3 insertions(+)
lib/librte_distributor/rte_distributor.c +3 -0
@@ -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.
Signature · 1 line
2.17.1
Thomas Monjalon
Applied25/07/2019 11:03, David Hunt:
Quoted 17 lines
From: Liang Ma <[email protected]>
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=316To 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.
Quoted 7 lines
Bugzilla ID: 316
Fixes 775003ad2f96 ("distributor: add new burst-capable library")Tested-by Michael Santana <[email protected]>Signed-off-by David Hunt <[email protected]>Signed-off-by Liang Ma <[email protected]>
Applied, thanks