Hi Sasha, please consider taking a small fix for this one (also useful for 4.15):
commit d3b9e8ad425cfd5b9116732e057f1b48e4d3bcb8 Author: Max Gurtovoy maxg@mellanox.com Date: Mon Mar 5 20:09:48 2018 +0200
RDMA/core: Reduce poll batch for direct cq polling
Fix warning limit for kernel stack consumption:
drivers/infiniband/core/cq.c: In function 'ib_process_cq_direct': drivers/infiniband/core/cq.c:78:1: error: the frame size of 1032 bytes is larger than 1024 bytes [-Werror=frame-larger-than=]
Using smaller ib_wc array on the stack brings us comfortably below that limit again.
Fixes: 246d8b184c10 ("IB/cq: Don't force IB_POLL_DIRECT poll context for ib_process_cq_direct") Reported-by: Arnd Bergmann arnd@arndb.de Reviewed-by: Sergey Gorenko sergeygo@mellanox.com Signed-off-by: Max Gurtovoy maxg@mellanox.com Signed-off-by: Leon Romanovsky leonro@mellanox.com Reviewed-by: Bart Van Assche bart.vanassche@wdc.com Acked-by: Arnd Bergmann arnd@arndb.de Signed-off-by: Jason Gunthorpe jgg@mellanox.com
-Max.
On 4/9/2018 3:20 AM, Sasha Levin wrote:
From: Sagi Grimberg sagi@grimberg.me
[ Upstream commit 246d8b184c100e8eb6b4e8c88f232c2ed2a4e672 ]
polling the completion queue directly does not interfere with the existing polling logic, hence drop the requirement. Be aware that running ib_process_cq_direct with non IB_POLL_DIRECT CQ may trigger concurrent CQ processing.
This can be used for polling mode ULPs.
Cc: Bart Van Assche bart.vanassche@wdc.com Reported-by: Steve Wise swise@opengridcomputing.com Signed-off-by: Sagi Grimberg sagi@grimberg.me [maxg: added wcs array argument to __ib_process_cq] Signed-off-by: Max Gurtovoy maxg@mellanox.com Signed-off-by: Doug Ledford dledford@redhat.com Signed-off-by: Sasha Levin alexander.levin@microsoft.com
drivers/infiniband/core/cq.c | 23 +++++++++++++---------- 1 file changed, 13 insertions(+), 10 deletions(-)
diff --git a/drivers/infiniband/core/cq.c b/drivers/infiniband/core/cq.c index f2ae75fa3128..c8c5a5a7f433 100644 --- a/drivers/infiniband/core/cq.c +++ b/drivers/infiniband/core/cq.c @@ -25,9 +25,10 @@ #define IB_POLL_FLAGS \ (IB_CQ_NEXT_COMP | IB_CQ_REPORT_MISSED_EVENTS) -static int __ib_process_cq(struct ib_cq *cq, int budget) +static int __ib_process_cq(struct ib_cq *cq, int budget, struct ib_wc *poll_wc) { int i, n, completed = 0;
- struct ib_wc *wcs = poll_wc ? : cq->wc;
/* * budget might be (-1) if the caller does not @@ -35,9 +36,9 @@ static int __ib_process_cq(struct ib_cq *cq, int budget) * minimum here. */ while ((n = ib_poll_cq(cq, min_t(u32, IB_POLL_BATCH,
budget - completed), cq->wc)) > 0) {
budget - completed), wcs)) > 0) {
struct ib_wc *wc = &cq->wc[i];
struct ib_wc *wc = &wcs[i];if (wc->wr_cqe) wc->wr_cqe->done(cq, wc); @@ -60,18 +61,20 @@ static int __ib_process_cq(struct ib_cq *cq, int budget)
- @cq: CQ to process
- @budget: number of CQEs to poll for
- This function is used to process all outstanding CQ entries on a
- %IB_POLL_DIRECT CQ. It does not offload CQ processing to a different
- context and does not ask for completion interrupts from the HCA.
- This function is used to process all outstanding CQ entries.
- It does not offload CQ processing to a different context and does
- not ask for completion interrupts from the HCA.
- Using direct processing on CQ with non IB_POLL_DIRECT type may trigger
*/ int ib_process_cq_direct(struct ib_cq *cq, int budget) {
- concurrent processing.
- Note: do not pass -1 as %budget unless it is guaranteed that the number
- of completions that will be processed is small.
- WARN_ON_ONCE(cq->poll_ctx != IB_POLL_DIRECT);
- struct ib_wc wcs[IB_POLL_BATCH];
- return __ib_process_cq(cq, budget);
- return __ib_process_cq(cq, budget, wcs); } EXPORT_SYMBOL(ib_process_cq_direct);
@@ -85,7 +88,7 @@ static int ib_poll_handler(struct irq_poll *iop, int budget) struct ib_cq *cq = container_of(iop, struct ib_cq, iop); int completed;
- completed = __ib_process_cq(cq, budget);
- completed = __ib_process_cq(cq, budget, NULL); if (completed < budget) { irq_poll_complete(&cq->iop); if (ib_req_notify_cq(cq, IB_POLL_FLAGS) > 0)
@@ -105,7 +108,7 @@ static void ib_cq_poll_work(struct work_struct *work) struct ib_cq *cq = container_of(work, struct ib_cq, work); int completed;
- completed = __ib_process_cq(cq, IB_POLL_BUDGET_WORKQUEUE);
- completed = __ib_process_cq(cq, IB_POLL_BUDGET_WORKQUEUE, NULL); if (completed >= IB_POLL_BUDGET_WORKQUEUE || ib_req_notify_cq(cq, IB_POLL_FLAGS) > 0) queue_work(ib_comp_wq, &cq->work);