Commit 84172dee authored by Steve Wise's avatar Steve Wise Committed by Roland Dreier

RDMA/cxgb4: Optimize CQ overflow detection

1) save the timestamp flit in the cq when we consume a CQE.

2) always compare the saved flit with the previous entry flit when
   reading the next CQE entry.  If the flits don't compare, then we
   have overflowed.
Signed-off-by: default avatarSteve Wise <swise@opengridcomputing.com>
Signed-off-by: default avatarRoland Dreier <rolandd@cisco.com>
parent 895cf5f3
......@@ -373,6 +373,7 @@ static void create_read_req_cqe(struct t4_wq *wq, struct t4_cqe *hw_cqe,
V_CQE_SWCQE(SW_CQE(hw_cqe)) |
V_CQE_OPCODE(FW_RI_READ_REQ) |
V_CQE_TYPE(1));
read_cqe->bits_type_ts = hw_cqe->bits_type_ts;
}
/*
......
......@@ -434,7 +434,7 @@ struct t4_cq {
struct c4iw_rdev *rdev;
u64 ugts;
size_t memsize;
u64 timestamp;
__be64 bits_type_ts;
u32 cqid;
u16 size; /* including status page */
u16 cidx;
......@@ -487,6 +487,7 @@ static inline void t4_swcq_consume(struct t4_cq *cq)
static inline void t4_hwcq_consume(struct t4_cq *cq)
{
cq->bits_type_ts = cq->queue[cq->cidx].bits_type_ts;
cq->cidx_inc++;
if (++cq->cidx == cq->size) {
cq->cidx = 0;
......@@ -501,20 +502,23 @@ static inline int t4_valid_cqe(struct t4_cq *cq, struct t4_cqe *cqe)
static inline int t4_next_hw_cqe(struct t4_cq *cq, struct t4_cqe **cqe)
{
int ret = 0;
u64 bits_type_ts = be64_to_cpu(cq->queue[cq->cidx].bits_type_ts);
int ret;
u16 prev_cidx;
if (G_CQE_GENBIT(bits_type_ts) == cq->gen) {
*cqe = &cq->queue[cq->cidx];
cq->timestamp = G_CQE_TS(bits_type_ts);
} else if (G_CQE_TS(bits_type_ts) > cq->timestamp)
ret = -EOVERFLOW;
if (cq->cidx == 0)
prev_cidx = cq->size - 1;
else
ret = -ENODATA;
if (ret == -EOVERFLOW) {
printk(KERN_ERR MOD "cq overflow cqid %u\n", cq->cqid);
prev_cidx = cq->cidx - 1;
if (cq->queue[prev_cidx].bits_type_ts != cq->bits_type_ts) {
ret = -EOVERFLOW;
cq->error = 1;
}
printk(KERN_ERR MOD "cq overflow cqid %u\n", cq->cqid);
} else if (t4_valid_cqe(cq, &cq->queue[cq->cidx])) {
*cqe = &cq->queue[cq->cidx];
ret = 0;
} else
ret = -ENODATA;
return ret;
}
......
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