Commit 1d10da0e authored by Roi Dayan's avatar Roi Dayan Committed by Pablo Neira Ayuso

netfilter: flowtable: Remove WQ_MEM_RECLAIM from workqueue

This workqueue is in charge of handling offloaded flow tasks like
add/del/stats we should not use WQ_MEM_RECLAIM flag.
The flag can result in the following warning.

[  485.557189] ------------[ cut here ]------------
[  485.562976] workqueue: WQ_MEM_RECLAIM nf_flow_table_offload:flow_offload_worr
[  485.562985] WARNING: CPU: 7 PID: 3731 at kernel/workqueue.c:2610 check_flush0
[  485.590191] Kernel panic - not syncing: panic_on_warn set ...
[  485.597100] CPU: 7 PID: 3731 Comm: kworker/u112:8 Not tainted 5.7.0-rc1.21802
[  485.606629] Hardware name: Dell Inc. PowerEdge R730/072T6D, BIOS 2.4.3 01/177
[  485.615487] Workqueue: nf_flow_table_offload flow_offload_work_handler [nf_f]
[  485.624834] Call Trace:
[  485.628077]  dump_stack+0x50/0x70
[  485.632280]  panic+0xfb/0x2d7
[  485.636083]  ? check_flush_dependency+0x110/0x130
[  485.641830]  __warn.cold.12+0x20/0x2a
[  485.646405]  ? check_flush_dependency+0x110/0x130
[  485.652154]  ? check_flush_dependency+0x110/0x130
[  485.657900]  report_bug+0xb8/0x100
[  485.662187]  ? sched_clock_cpu+0xc/0xb0
[  485.666974]  do_error_trap+0x9f/0xc0
[  485.671464]  do_invalid_op+0x36/0x40
[  485.675950]  ? check_flush_dependency+0x110/0x130
[  485.681699]  invalid_op+0x28/0x30

Fixes: 7da182a9 ("netfilter: flowtable: Use work entry per offload command")
Reported-by: default avatarMarcelo Ricardo Leitner <mleitner@redhat.com>
Signed-off-by: default avatarRoi Dayan <roid@mellanox.com>
Reviewed-by: default avatarPaul Blakey <paulb@mellanox.com>
Signed-off-by: default avatarPablo Neira Ayuso <pablo@netfilter.org>
parent 2c889795
......@@ -1062,7 +1062,7 @@ static struct flow_indr_block_entry block_ing_entry = {
int nf_flow_table_offload_init(void)
{
nf_flow_offload_wq = alloc_workqueue("nf_flow_table_offload",
WQ_UNBOUND | WQ_MEM_RECLAIM, 0);
WQ_UNBOUND, 0);
if (!nf_flow_offload_wq)
return -ENOMEM;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment