From: Mike Marciniszyn <mike.marciniszyn@xxxxxxxxxxxxxxxxxxxx> Despite the patch noted below, the warning still happens with certain kernel configs. It appears that either check_preemption_disabled() is inconsistent with with debug_rcu_read_lock() or the patch incorrectly assumes that an RCU critical section will prevent the current cpu from changing. A clarification has been solicited via: https://lore.kernel.org/linux-rdma/CH0PR01MB71536FB1BD5ECF16E65CB3BFF26F9@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/T/#u This patch will silence the warning for now by using get_cpu()/put_cpu(). Fixes: b6d57e24ce6c ("IB/hfi1: Insure use of smp_processor_id() is preempt disabled") Cc: stable@xxxxxxxxxxxxxxx Reviewed-by: Dennis Dalessandro <dennis.dalessandro@xxxxxxxxxxxxxxxxxxxx> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@xxxxxxxxxxxxxxxxxxxx> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@xxxxxxxxxxxxxxxxxxxx> --- drivers/infiniband/hw/hfi1/sdma.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c index f07d328..809096d 100644 --- a/drivers/infiniband/hw/hfi1/sdma.c +++ b/drivers/infiniband/hw/hfi1/sdma.c @@ -839,15 +839,15 @@ struct sdma_engine *sdma_select_user_engine(struct hfi1_devdata *dd, goto out; rcu_read_lock(); - cpu_id = smp_processor_id(); + cpu_id = get_cpu(); rht_node = rhashtable_lookup(dd->sdma_rht, &cpu_id, sdma_rht_params); - if (rht_node && rht_node->map[vl]) { struct sdma_rht_map_elem *map = rht_node->map[vl]; sde = map->sde[selector & map->mask]; } + put_cpu(); rcu_read_unlock(); if (sde)
