IB/hfi1: Increment the retry timeout value for TID RDMA READ request

The RC retry timeout value is based on the estimated time for the
response packet to come back. However, for TID RDMA READ request, due
to the use of header suppression, the driver is normally not notified
for each incoming response packet until the last TID RDMA READ response
packet. Consequently, the retry timeout value should be extended to
cover the transaction time for the entire length of a segment (default
256K) instead of that for a single packet. This patch addresses the
issue by introducing new retry timer functions to account for multiple
packets and wrapper functions for backward compatibility.

Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This commit is contained in:
Kaike Wan
2019-01-23 19:31:57 -08:00
committed by Doug Ledford
parent b126078e89
commit 039cd3daf1
3 changed files with 20 additions and 9 deletions

View File

@@ -2393,11 +2393,12 @@ static inline unsigned long rvt_aeth_to_usec(u32 aeth)
}
/*
* rvt_add_retry_timer - add/start a retry timer
* rvt_add_retry_timer_ext - add/start a retry timer
* @qp - the QP
* @shift - timeout shift to wait for multiple packets
* add a retry timer on the QP
*/
void rvt_add_retry_timer(struct rvt_qp *qp)
void rvt_add_retry_timer_ext(struct rvt_qp *qp, u8 shift)
{
struct ib_qp *ibqp = &qp->ibqp;
struct rvt_dev_info *rdi = ib_to_rvt(ibqp->device);
@@ -2405,11 +2406,11 @@ void rvt_add_retry_timer(struct rvt_qp *qp)
lockdep_assert_held(&qp->s_lock);
qp->s_flags |= RVT_S_TIMER;
/* 4.096 usec. * (1 << qp->timeout) */
qp->s_timer.expires = jiffies + qp->timeout_jiffies +
rdi->busy_jiffies;
qp->s_timer.expires = jiffies + rdi->busy_jiffies +
(qp->timeout_jiffies << shift);
add_timer(&qp->s_timer);
}
EXPORT_SYMBOL(rvt_add_retry_timer);
EXPORT_SYMBOL(rvt_add_retry_timer_ext);
/**
* rvt_add_rnr_timer - add/start an rnr timer