From: Mike Marciniszyn mike.marciniszyn@cornelisnetworks.com
[ Upstream commit 6b13215df1d37f5be23fc4a01a915a287b25ce15 ]
Add traces that were vital in isolating an issue with pq waitlist in commit fa8dac396863 ("IB/hfi1: Fix another case where pq is left on waitlist")
Link: https://lore.kernel.org/r/1617026056-50483-8-git-send-email-dennis.dalessand... Reviewed-by: Kaike Wan kaike.wan@intel.com Signed-off-by: Mike Marciniszyn mike.marciniszyn@cornelisnetworks.com Signed-off-by: Dennis Dalessandro dennis.dalessandro@cornelisnetworks.com Signed-off-by: Jason Gunthorpe jgg@nvidia.com Stable-dep-of: 00cbce5cbf88 ("IB/hfi1: Fix bugs with non-PAGE_SIZE-end multi-iovec user SDMA requests") Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/infiniband/hw/hfi1/trace_tx.h | 75 ++++++++++++++++++++++++++ drivers/infiniband/hw/hfi1/user_sdma.c | 12 +++-- drivers/infiniband/hw/hfi1/user_sdma.h | 1 + 3 files changed, 85 insertions(+), 3 deletions(-)
diff --git a/drivers/infiniband/hw/hfi1/trace_tx.h b/drivers/infiniband/hw/hfi1/trace_tx.h index 847654196dd34..d44fc54858b90 100644 --- a/drivers/infiniband/hw/hfi1/trace_tx.h +++ b/drivers/infiniband/hw/hfi1/trace_tx.h @@ -54,6 +54,7 @@ #include "mad.h" #include "sdma.h" #include "ipoib.h" +#include "user_sdma.h"
const char *parse_sdma_flags(struct trace_seq *p, u64 desc0, u64 desc1);
@@ -654,6 +655,80 @@ TRACE_EVENT(hfi1_sdma_user_completion, __entry->code) );
+TRACE_EVENT(hfi1_usdma_defer, + TP_PROTO(struct hfi1_user_sdma_pkt_q *pq, + struct sdma_engine *sde, + struct iowait *wait), + TP_ARGS(pq, sde, wait), + TP_STRUCT__entry(DD_DEV_ENTRY(pq->dd) + __field(struct hfi1_user_sdma_pkt_q *, pq) + __field(struct sdma_engine *, sde) + __field(struct iowait *, wait) + __field(int, engine) + __field(int, empty) + ), + TP_fast_assign(DD_DEV_ASSIGN(pq->dd); + __entry->pq = pq; + __entry->sde = sde; + __entry->wait = wait; + __entry->engine = sde->this_idx; + __entry->empty = list_empty(&__entry->wait->list); + ), + TP_printk("[%s] pq %llx sde %llx wait %llx engine %d empty %d", + __get_str(dev), + (unsigned long long)__entry->pq, + (unsigned long long)__entry->sde, + (unsigned long long)__entry->wait, + __entry->engine, + __entry->empty + ) +); + +TRACE_EVENT(hfi1_usdma_activate, + TP_PROTO(struct hfi1_user_sdma_pkt_q *pq, + struct iowait *wait, + int reason), + TP_ARGS(pq, wait, reason), + TP_STRUCT__entry(DD_DEV_ENTRY(pq->dd) + __field(struct hfi1_user_sdma_pkt_q *, pq) + __field(struct iowait *, wait) + __field(int, reason) + ), + TP_fast_assign(DD_DEV_ASSIGN(pq->dd); + __entry->pq = pq; + __entry->wait = wait; + __entry->reason = reason; + ), + TP_printk("[%s] pq %llx wait %llx reason %d", + __get_str(dev), + (unsigned long long)__entry->pq, + (unsigned long long)__entry->wait, + __entry->reason + ) +); + +TRACE_EVENT(hfi1_usdma_we, + TP_PROTO(struct hfi1_user_sdma_pkt_q *pq, + int we_ret), + TP_ARGS(pq, we_ret), + TP_STRUCT__entry(DD_DEV_ENTRY(pq->dd) + __field(struct hfi1_user_sdma_pkt_q *, pq) + __field(int, state) + __field(int, we_ret) + ), + TP_fast_assign(DD_DEV_ASSIGN(pq->dd); + __entry->pq = pq; + __entry->state = pq->state; + __entry->we_ret = we_ret; + ), + TP_printk("[%s] pq %llx state %d we_ret %d", + __get_str(dev), + (unsigned long long)__entry->pq, + __entry->state, + __entry->we_ret + ) +); + const char *print_u32_array(struct trace_seq *, u32 *, int); #define __print_u32_hex(arr, len) print_u32_array(p, arr, len)
diff --git a/drivers/infiniband/hw/hfi1/user_sdma.c b/drivers/infiniband/hw/hfi1/user_sdma.c index 4a4956f96a7eb..da5b2e37355ab 100644 --- a/drivers/infiniband/hw/hfi1/user_sdma.c +++ b/drivers/infiniband/hw/hfi1/user_sdma.c @@ -133,6 +133,7 @@ static int defer_packet_queue( container_of(wait->iow, struct hfi1_user_sdma_pkt_q, busy);
write_seqlock(&sde->waitlock); + trace_hfi1_usdma_defer(pq, sde, &pq->busy); if (sdma_progress(sde, seq, txreq)) goto eagain; /* @@ -157,7 +158,8 @@ static void activate_packet_queue(struct iowait *wait, int reason) { struct hfi1_user_sdma_pkt_q *pq = container_of(wait, struct hfi1_user_sdma_pkt_q, busy); - pq->busy.lock = NULL; + + trace_hfi1_usdma_activate(pq, wait, reason); xchg(&pq->state, SDMA_PKT_Q_ACTIVE); wake_up(&wait->wait_dma); }; @@ -599,13 +601,17 @@ int hfi1_user_sdma_process_request(struct hfi1_filedata *fd, while (req->seqsubmitted != req->info.npkts) { ret = user_sdma_send_pkts(req, pcount); if (ret < 0) { + int we_ret; + if (ret != -EBUSY) goto free_req; - if (wait_event_interruptible_timeout( + we_ret = wait_event_interruptible_timeout( pq->busy.wait_dma, pq->state == SDMA_PKT_Q_ACTIVE, msecs_to_jiffies( - SDMA_IOWAIT_TIMEOUT)) <= 0) + SDMA_IOWAIT_TIMEOUT)); + trace_hfi1_usdma_we(pq, we_ret); + if (we_ret <= 0) flush_pq_iowait(pq); } } diff --git a/drivers/infiniband/hw/hfi1/user_sdma.h b/drivers/infiniband/hw/hfi1/user_sdma.h index 1e8c02fe8ad1d..fabe581399068 100644 --- a/drivers/infiniband/hw/hfi1/user_sdma.h +++ b/drivers/infiniband/hw/hfi1/user_sdma.h @@ -53,6 +53,7 @@ #include "common.h" #include "iowait.h" #include "user_exp_rcv.h" +#include "mmu_rb.h"
/* The maximum number of Data io vectors per message/request */ #define MAX_VECTORS_PER_REQ 8