From: Steve Wise <swise@opengridcomputing.com>
Date: Thu, 2 Nov 2017 14:11:03 -0700
Subject: iw_cxgb4: remove BUG_ON() usage.
Patch-mainline: v4.15-rc1
Git-commit: ba97b749979ef0ebb821e58ee8b16a84412922f6
References: bsc#1064802 bsc#1066129
iw_cxgb4 has many BUG_ON()s that were left over from various enhancemnets
made over the years. Almost all of them should just be removed. Some,
however indicate a ULP usage error and can be handled w/o bringing down
the system.
If the condition cannot happen with correctly implemented cxgb4 sw/fw,
then remove the BUG_ON.
If the condition indicates a misbehaving ULP (like CQ overflows), add
proper recovery logic.
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
drivers/infiniband/hw/cxgb4/cm.c | 25 +++++++++----------------
drivers/infiniband/hw/cxgb4/cq.c | 10 ----------
drivers/infiniband/hw/cxgb4/id_table.c | 1 -
drivers/infiniband/hw/cxgb4/iw_cxgb4.h | 1 -
drivers/infiniband/hw/cxgb4/provider.c | 4 ++--
drivers/infiniband/hw/cxgb4/qp.c | 3 ---
drivers/infiniband/hw/cxgb4/t4.h | 7 ++-----
7 files changed, 13 insertions(+), 38 deletions(-)
--- a/drivers/infiniband/hw/cxgb4/cm.c
+++ b/drivers/infiniband/hw/cxgb4/cm.c
@@ -914,8 +914,6 @@ static int send_mpa_req(struct c4iw_ep *
pr_debug("ep %p tid %u pd_len %d\n",
ep, ep->hwtid, ep->plen);
- BUG_ON(skb_cloned(skb));
-
mpalen = sizeof(*mpa) + ep->plen;
if (mpa_rev_to_use == 2)
mpalen += sizeof(struct mpa_v2_conn_params);
@@ -998,7 +996,6 @@ static int send_mpa_req(struct c4iw_ep *
*/
skb_get(skb);
t4_set_arp_err_handler(skb, NULL, arp_failure_discard);
- BUG_ON(ep->mpa_skb);
ep->mpa_skb = skb;
ret = c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
if (ret)
@@ -1084,7 +1081,6 @@ static int send_mpa_reject(struct c4iw_e
skb_get(skb);
set_wr_txq(skb, CPL_PRIORITY_DATA, ep->txq_idx);
t4_set_arp_err_handler(skb, NULL, mpa_start_arp_failure);
- BUG_ON(ep->mpa_skb);
ep->mpa_skb = skb;
ep->snd_seq += mpalen;
return c4iw_l2t_send(&ep->com.dev->rdev, skb, ep->l2t);
@@ -1838,7 +1834,6 @@ static int rx_data(struct c4iw_dev *dev,
struct c4iw_qp_attributes attrs;
update_rx_credits(ep, dlen);
- BUG_ON(!ep->com.qp);
if (status)
pr_err("%s Unexpected streaming data." \
" qpid %u ep %p state %d tid %u status %d\n",
@@ -2112,7 +2107,7 @@ static int c4iw_reconnect(struct c4iw_ep
* further connection establishment. As we are using the same EP pointer
* for reconnect, few skbs are used during the previous c4iw_connect(),
* which leaves the EP with inadequate skbs for further
- * c4iw_reconnect(), Further causing an assert BUG_ON() due to empty
+ * c4iw_reconnect(), Further causing a crash due to an empty
* skb_list() during peer_abort(). Allocate skbs which is already used.
*/
size = (CN_MAX_CON_BUF - skb_queue_len(&ep->com.ep_skb_list));
@@ -2359,7 +2354,6 @@ static int accept_cr(struct c4iw_ep *ep,
enum chip_type adapter_type = ep->com.dev->rdev.lldi.adapter_type;
pr_debug("ep %p tid %u\n", ep, ep->hwtid);
- BUG_ON(skb_cloned(skb));
skb_get(skb);
rpl = cplhdr(skb);
@@ -2443,7 +2437,6 @@ static int accept_cr(struct c4iw_ep *ep,
static void reject_cr(struct c4iw_dev *dev, u32 hwtid, struct sk_buff *skb)
{
pr_debug("c4iw_dev %p tid %u\n", dev, hwtid);
- BUG_ON(skb_cloned(skb));
skb_trim(skb, sizeof(struct cpl_tid_release));
release_tid(&dev->rdev, hwtid, skb);
return;
@@ -2716,7 +2709,7 @@ static int peer_close(struct c4iw_dev *d
disconnect = 0;
break;
default:
- BUG_ON(1);
+ WARN_ONCE(1, "Bad endpoint state %u\n", ep->com.state);
}
mutex_unlock(&ep->com.mutex);
if (disconnect)
@@ -2816,7 +2809,7 @@ static int peer_abort(struct c4iw_dev *d
mutex_unlock(&ep->com.mutex);
goto deref_ep;
default:
- BUG_ON(1);
+ WARN_ONCE(1, "Bad endpoint state %u\n", ep->com.state);
break;
}
dst_confirm(ep->dst);
@@ -2903,7 +2896,7 @@ static int close_con_rpl(struct c4iw_dev
case DEAD:
break;
default:
- BUG_ON(1);
+ WARN_ONCE(1, "Bad endpoint state %u\n", ep->com.state);
break;
}
mutex_unlock(&ep->com.mutex);
@@ -2921,7 +2914,6 @@ static int terminate(struct c4iw_dev *de
struct c4iw_qp_attributes attrs;
ep = get_ep_from_tid(dev, tid);
- BUG_ON(!ep);
if (ep && ep->com.qp) {
pr_warn("TERM received tid %u qpid %u\n",
@@ -3021,7 +3013,10 @@ int c4iw_accept_cr(struct iw_cm_id *cm_i
goto err_out;
}
- BUG_ON(!qp);
+ if (!qp) {
+ err = -EINVAL;
+ goto err_out;
+ }
set_bit(ULP_ACCEPT, &ep->com.history);
if ((conn_param->ord > cur_max_read_depth(ep->com.dev)) ||
@@ -3579,7 +3574,7 @@ int c4iw_ep_disconnect(struct c4iw_ep *e
__func__, ep, ep->com.state);
break;
default:
- BUG();
+ WARN_ONCE(1, "Bad endpoint state %u\n", ep->com.state);
break;
}
@@ -3679,7 +3674,6 @@ static void passive_ofld_conn_reply(stru
int ret;
rpl_skb = (struct sk_buff *)(unsigned long)req->cookie;
- BUG_ON(!rpl_skb);
if (req->retval) {
pr_err("%s passive open failure %d\n", __func__, req->retval);
mutex_lock(&dev->rdev.stats.lock);
@@ -4106,7 +4100,6 @@ static void process_work(struct work_str
dev = *((struct c4iw_dev **) (skb->cb + sizeof(void *)));
opcode = rpl->ot.opcode;
- BUG_ON(!work_handlers[opcode]);
ret = work_handlers[opcode](dev, skb);
if (!ret)
kfree_skb(skb);
--- a/drivers/infiniband/hw/cxgb4/cq.c
+++ b/drivers/infiniband/hw/cxgb4/cq.c
@@ -188,7 +188,6 @@ int c4iw_flush_rq(struct t4_wq *wq, stru
int flushed = 0;
int in_use = wq->rq.in_use - count;
- BUG_ON(in_use < 0);
pr_debug("wq %p cq %p rq.in_use %u skip count %u\n",
wq, cq, wq->rq.in_use, count);
while (in_use--) {
@@ -231,14 +230,11 @@ int c4iw_flush_sq(struct c4iw_qp *qhp)
if (wq->sq.flush_cidx == -1)
wq->sq.flush_cidx = wq->sq.cidx;
idx = wq->sq.flush_cidx;
- BUG_ON(idx >= wq->sq.size);
while (idx != wq->sq.pidx) {
swsqe = &wq->sq.sw_sq[idx];
- BUG_ON(swsqe->flushed);
swsqe->flushed = 1;
insert_sq_cqe(wq, cq, swsqe);
if (wq->sq.oldest_read == swsqe) {
- BUG_ON(swsqe->opcode != FW_RI_READ_REQ);
advance_oldest_read(wq);
}
flushed++;
@@ -259,7 +255,6 @@ static void flush_completed_wrs(struct t
if (wq->sq.flush_cidx == -1)
wq->sq.flush_cidx = wq->sq.cidx;
cidx = wq->sq.flush_cidx;
- BUG_ON(cidx > wq->sq.size);
while (cidx != wq->sq.pidx) {
swsqe = &wq->sq.sw_sq[cidx];
@@ -268,8 +263,6 @@ static void flush_completed_wrs(struct t
cidx = 0;
} else if (swsqe->complete) {
- BUG_ON(swsqe->flushed);
-
/*
* Insert this completed cqe into the swcq.
*/
@@ -613,7 +606,6 @@ proc_cqe:
*/
if (SQ_TYPE(hw_cqe)) {
int idx = CQE_WRID_SQ_IDX(hw_cqe);
- BUG_ON(idx >= wq->sq.size);
/*
* Account for any unsignaled completions completed by
@@ -627,7 +619,6 @@ proc_cqe:
wq->sq.in_use -= wq->sq.size + idx - wq->sq.cidx;
else
wq->sq.in_use -= idx - wq->sq.cidx;
- BUG_ON(wq->sq.in_use <= 0 && wq->sq.in_use >= wq->sq.size);
wq->sq.cidx = (uint16_t)idx;
pr_debug("completing sq idx %u\n", wq->sq.cidx);
@@ -638,7 +629,6 @@ proc_cqe:
} else {
pr_debug("completing rq idx %u\n", wq->rq.cidx);
*cookie = wq->rq.sw_rq[wq->rq.cidx].wr_id;
- BUG_ON(t4_rq_empty(wq));
if (c4iw_wr_log)
c4iw_log_wr_stats(wq, hw_cqe);
t4_rq_consume(wq);
--- a/drivers/infiniband/hw/cxgb4/id_table.c
+++ b/drivers/infiniband/hw/cxgb4/id_table.c
@@ -73,7 +73,6 @@ void c4iw_id_free(struct c4iw_id_table *
unsigned long flags;
obj -= alloc->start;
- BUG_ON((int)obj < 0);
spin_lock_irqsave(&alloc->lock, flags);
clear_bit(obj, alloc->table);
--- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
+++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
@@ -364,7 +364,6 @@ static inline int _insert_handle(struct
idr_preload_end();
}
- BUG_ON(ret == -ENOSPC);
return ret < 0 ? ret : 0;
}
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -310,8 +310,9 @@ static int c4iw_query_gid(struct ib_devi
pr_debug("ibdev %p, port %d, index %d, gid %p\n",
ibdev, port, index, gid);
+ if (!port)
+ return -EINVAL;
dev = to_c4iw_dev(ibdev);
- BUG_ON(port == 0);
memset(&(gid->raw[0]), 0, sizeof(gid->raw));
memcpy(&(gid->raw[0]), dev->rdev.lldi.ports[port-1]->dev_addr, 6);
return 0;
@@ -536,7 +537,6 @@ int c4iw_register_device(struct c4iw_dev
int i;
pr_debug("c4iw_dev %p\n", dev);
- BUG_ON(!dev->rdev.lldi.ports[0]);
strlcpy(dev->ibdev.name, "cxgb4_%d", IB_DEVICE_NAME_MAX);
memset(&dev->ibdev.node_guid, 0, sizeof(dev->ibdev.node_guid));
memcpy(&dev->ibdev.node_guid, dev->rdev.lldi.ports[0]->dev_addr, 6);
--- a/drivers/infiniband/hw/cxgb4/qp.c
+++ b/drivers/infiniband/hw/cxgb4/qp.c
@@ -689,7 +689,6 @@ static int build_memreg(struct t4_sq *sq
if (++p == (__be64 *)&sq->queue[sq->size])
p = (__be64 *)sq->queue;
}
- BUG_ON(rem < 0);
while (rem) {
*p = 0;
rem -= sizeof(*p);
@@ -1568,7 +1567,6 @@ int c4iw_modify_qp(struct c4iw_dev *rhp,
case C4IW_QP_STATE_RTS:
switch (attrs->next_state) {
case C4IW_QP_STATE_CLOSING:
- BUG_ON(kref_read(&qhp->ep->com.kref) < 2);
t4_set_wq_in_error(&qhp->wq);
set_state(qhp, C4IW_QP_STATE_CLOSING);
ep = qhp->ep;
@@ -1677,7 +1675,6 @@ err:
set_state(qhp, C4IW_QP_STATE_ERROR);
free = 1;
abort = 1;
- BUG_ON(!ep);
flush_qp(qhp);
wake_up(&qhp->wait);
out:
--- a/drivers/infiniband/hw/cxgb4/t4.h
+++ b/drivers/infiniband/hw/cxgb4/t4.h
@@ -425,7 +425,6 @@ static inline void t4_sq_produce(struct
static inline void t4_sq_consume(struct t4_wq *wq)
{
- BUG_ON(wq->sq.in_use < 1);
if (wq->sq.cidx == wq->sq.flush_cidx)
wq->sq.flush_cidx = -1;
wq->sq.in_use--;
@@ -600,7 +599,8 @@ static inline void t4_swcq_produce(struc
pr_warn("%s cxgb4 sw cq overflow cqid %u\n",
__func__, cq->cqid);
cq->error = 1;
- BUG_ON(1);
+ cq->sw_in_use--;
+ return;
}
if (++cq->sw_pidx == cq->size)
cq->sw_pidx = 0;
@@ -608,7 +608,6 @@ static inline void t4_swcq_produce(struc
static inline void t4_swcq_consume(struct t4_cq *cq)
{
- BUG_ON(cq->sw_in_use < 1);
cq->sw_in_use--;
if (++cq->sw_cidx == cq->size)
cq->sw_cidx = 0;
@@ -654,7 +653,6 @@ static inline int t4_next_hw_cqe(struct
ret = -EOVERFLOW;
cq->error = 1;
pr_err("cq overflow cqid %u\n", cq->cqid);
- BUG_ON(1);
} else if (t4_valid_cqe(cq, &cq->queue[cq->cidx])) {
/* Ensure CQE is flushed to memory */
@@ -672,7 +670,6 @@ static inline struct t4_cqe *t4_next_sw_
pr_warn("%s cxgb4 sw cq overflow cqid %u\n",
__func__, cq->cqid);
cq->error = 1;
- BUG_ON(1);
return NULL;
}
if (cq->sw_in_use)