The patch below does not apply to the 4.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to stable@vger.kernel.org.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From a93a0a31111231bb1949f4a83b17238f0fa32d6a Mon Sep 17 00:00:00 2001
From: "Michael J. Ruhl" michael.j.ruhl@intel.com Date: Wed, 2 May 2018 06:43:07 -0700 Subject: [PATCH] IB/hfi1: Reorder incorrect send context disable
User send context integrity bits are cleared before the context is disabled. If the send context is still processing data, any packets that need those integrity bits will cause an error and halt the send context.
During the disable handling, the driver waits for the context to drain. If the context is halted, the driver will eventually timeout because the context won't drain and then incorrectly bounce the link.
Reorder the bit clearing and the context disable.
Examine the software state and send context status as well as the egress status to determine if a send context is in the halted state.
Promote the check macros to static functions for consistency with the new check and to follow kernel style.
Remove an unused define that refers to the egress timeout.
Cc: stable@vger.kernel.org # 4.9.x Reviewed-by: Mitko Haralanov mitko.haralanov@intel.com Reviewed-by: Mike Marciniszyn mike.marciniszyn@intel.com Signed-off-by: Michael J. Ruhl michael.j.ruhl@intel.com Signed-off-by: Dennis Dalessandro dennis.dalessandro@intel.com Signed-off-by: Doug Ledford dledford@redhat.com
diff --git a/drivers/infiniband/hw/hfi1/file_ops.c b/drivers/infiniband/hw/hfi1/file_ops.c index 1b778fd16a32..c9d23c37a371 100644 --- a/drivers/infiniband/hw/hfi1/file_ops.c +++ b/drivers/infiniband/hw/hfi1/file_ops.c @@ -689,8 +689,8 @@ static int hfi1_file_close(struct inode *inode, struct file *fp) * checks to default and disable the send context. */ if (uctxt->sc) { - set_pio_integrity(uctxt->sc); sc_disable(uctxt->sc); + set_pio_integrity(uctxt->sc); }
hfi1_free_ctxt_rcv_groups(uctxt); diff --git a/drivers/infiniband/hw/hfi1/pio.c b/drivers/infiniband/hw/hfi1/pio.c index 40dac4d16eb8..9cac15d10c4f 100644 --- a/drivers/infiniband/hw/hfi1/pio.c +++ b/drivers/infiniband/hw/hfi1/pio.c @@ -50,8 +50,6 @@ #include "qp.h" #include "trace.h"
-#define SC_CTXT_PACKET_EGRESS_TIMEOUT 350 /* in chip cycles */ - #define SC(name) SEND_CTXT_##name /* * Send Context functions @@ -961,15 +959,40 @@ void sc_disable(struct send_context *sc) }
/* return SendEgressCtxtStatus.PacketOccupancy */ -#define packet_occupancy(r) \ - (((r) & SEND_EGRESS_CTXT_STATUS_CTXT_EGRESS_PACKET_OCCUPANCY_SMASK)\ - >> SEND_EGRESS_CTXT_STATUS_CTXT_EGRESS_PACKET_OCCUPANCY_SHIFT) +static u64 packet_occupancy(u64 reg) +{ + return (reg & + SEND_EGRESS_CTXT_STATUS_CTXT_EGRESS_PACKET_OCCUPANCY_SMASK) + >> SEND_EGRESS_CTXT_STATUS_CTXT_EGRESS_PACKET_OCCUPANCY_SHIFT; +}
/* is egress halted on the context? */ -#define egress_halted(r) \ - ((r) & SEND_EGRESS_CTXT_STATUS_CTXT_EGRESS_HALT_STATUS_SMASK) +static bool egress_halted(u64 reg) +{ + return !!(reg & SEND_EGRESS_CTXT_STATUS_CTXT_EGRESS_HALT_STATUS_SMASK); +}
-/* wait for packet egress, optionally pause for credit return */ +/* is the send context halted? */ +static bool is_sc_halted(struct hfi1_devdata *dd, u32 hw_context) +{ + return !!(read_kctxt_csr(dd, hw_context, SC(STATUS)) & + SC(STATUS_CTXT_HALTED_SMASK)); +} + +/** + * sc_wait_for_packet_egress + * @sc: valid send context + * @pause: wait for credit return + * + * Wait for packet egress, optionally pause for credit return + * + * Egress halt and Context halt are not necessarily the same thing, so + * check for both. + * + * NOTE: The context halt bit may not be set immediately. Because of this, + * it is necessary to check the SW SFC_HALTED bit (set in the IRQ) and the HW + * context bit to determine if the context is halted. + */ static void sc_wait_for_packet_egress(struct send_context *sc, int pause) { struct hfi1_devdata *dd = sc->dd; @@ -981,8 +1004,9 @@ static void sc_wait_for_packet_egress(struct send_context *sc, int pause) reg_prev = reg; reg = read_csr(dd, sc->hw_context * 8 + SEND_EGRESS_CTXT_STATUS); - /* done if egress is stopped */ - if (egress_halted(reg)) + /* done if any halt bits, SW or HW are set */ + if (sc->flags & SCF_HALTED || + is_sc_halted(dd, sc->hw_context) || egress_halted(reg)) break; reg = packet_occupancy(reg); if (reg == 0)