[PATCH RESEND 00/11] staging/rdma/hfi1: Miscellaneous Fixes

[PATCH RESEND 00/11] staging/rdma/hfi1: Miscellaneous Fixes

[Jubin John]

Resending these previously sent hfi1 driver patches as a series, instead
of individual patches, based on feedback about order of patch application
from a friendly patch-bot of Greg KH.


Dean Luick (3):
  staging/rdma/hfi1: Always download SBus firmware
  staging/rdma/hfi1: Select only devices with active links
  staging/rdma/hfi1: use one-shot LCB write

Easwar Hariharan (1):
  staging/rdma/hfi1: Clear the QSFP reset that is asserted on FLR

Ignacio Hernandez (1):
  staging/rdma/hfi1: Remove spurious error messages

Ira Weiny (1):
  staging/rdma/hfi1: Eliminate WARN_ON when VL is invalid

Jareer Abdel-Qader (1):
  staging/rdma/hfi1: Disable thermal polling before sensor
    initialization

Mark F. Brown (1):
  staging/rdma/hfi1: Workaround to prevent corruption during packet
    delivery

Sebastian Sanchez (2):
  staging/rdma/hfi1: Convert dd_dev_info() to hfi1_cdbg() in process
    startup
  staging/rdma/hfi1: Fix for opaportconfig ledon by not checking for
    portNum

Vennila Megavannan (1):
  staging/rdma/hfi1: Enable WFR PCIe extended tags from the driver

 drivers/staging/rdma/hfi1/chip.c     |   79 ++++++++++++++++++++++++++--------
 drivers/staging/rdma/hfi1/chip.h     |    1 +
 drivers/staging/rdma/hfi1/driver.c   |   26 ++++++------
 drivers/staging/rdma/hfi1/file_ops.c |   31 ++++++++++---
 drivers/staging/rdma/hfi1/firmware.c |    2 +-
 drivers/staging/rdma/hfi1/hfi.h      |    4 ++
 drivers/staging/rdma/hfi1/init.c     |   59 ++++++++++++++++++++------
 drivers/staging/rdma/hfi1/mad.c      |    4 +-
 drivers/staging/rdma/hfi1/pcie.c     |   24 +++++-----
 drivers/staging/rdma/hfi1/pio.c      |   19 ++++----
 drivers/staging/rdma/hfi1/sdma.c     |   12 ++++-
 11 files changed, 183 insertions(+), 78 deletions(-)

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 01/11] staging/rdma/hfi1: Convert dd_dev_info() to hfi1_cdbg() in process startup

[Jubin John]

From: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Replacing dd_dev_info() for hfi1_cdbg() to avoid generating syslog
output for every context that is open by PSM.

Reviewed-by: Mitko Haralanov <mitko.haralanov-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/file_ops.c |   10 +++++-----
 drivers/staging/rdma/hfi1/init.c     |   31 ++++++++++++++++++-------------
 drivers/staging/rdma/hfi1/pio.c      |   19 ++++++++++---------
 3 files changed, 33 insertions(+), 27 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/file_ops.c b/drivers/staging/rdma/hfi1/file_ops.c
index aae9826..fba5889 100644
--- a/drivers/staging/rdma/hfi1/file_ops.c
+++ b/drivers/staging/rdma/hfi1/file_ops.c
@@ -687,9 +687,9 @@ static int hfi1_file_mmap(struct file *fp, struct vm_area_struct *vma)
 	}
 
 	vma->vm_flags = flags;
-	dd_dev_info(dd,
-		    "%s: %u:%u type:%u io/vf:%d/%d, addr:0x%llx, len:%lu(%lu), flags:0x%lx\n",
-		    __func__, ctxt, subctxt, type, mapio, vmf, memaddr, memlen,
+	hfi1_cdbg(PROC,
+		  "%u:%u type:%u io/vf:%d/%d, addr:0x%llx, len:%lu(%lu), flags:0x%lx\n",
+		    ctxt, subctxt, type, mapio, vmf, memaddr, memlen,
 		    vma->vm_end - vma->vm_start, vma->vm_flags);
 	pfn = (unsigned long)(memaddr >> PAGE_SHIFT);
 	if (vmf) {
@@ -1011,8 +1011,8 @@ static int allocate_ctxt(struct file *fp, struct hfi1_devdata *dd,
 	if (!uctxt->sc)
 		return -ENOMEM;
 
-	dbg("allocated send context %u(%u)\n", uctxt->sc->sw_index,
-		uctxt->sc->hw_context);
+	hfi1_cdbg(PROC, "allocated send context %u(%u)\n", uctxt->sc->sw_index,
+		  uctxt->sc->hw_context);
 	ret = sc_enable(uctxt->sc);
 	if (ret)
 		return ret;
diff --git a/drivers/staging/rdma/hfi1/init.c b/drivers/staging/rdma/hfi1/init.c
index 8666f3a..4a79744 100644
--- a/drivers/staging/rdma/hfi1/init.c
+++ b/drivers/staging/rdma/hfi1/init.c
@@ -60,6 +60,7 @@
 #include "hfi.h"
 #include "device.h"
 #include "common.h"
+#include "trace.h"
 #include "mad.h"
 #include "sdma.h"
 #include "debugfs.h"
@@ -208,7 +209,7 @@ struct hfi1_ctxtdata *hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, u32 ctxt)
 	if (rcd) {
 		u32 rcvtids, max_entries;
 
-		dd_dev_info(dd, "%s: setting up context %u\n", __func__, ctxt);
+		hfi1_cdbg(PROC, "setting up context %u\n", ctxt);
 
 		INIT_LIST_HEAD(&rcd->qp_wait_list);
 		rcd->ppd = ppd;
@@ -279,8 +280,9 @@ struct hfi1_ctxtdata *hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, u32 ctxt)
 				   rcd->ctxt);
 			rcd->egrbufs.count = MAX_EAGER_ENTRIES;
 		}
-		dd_dev_info(dd, "ctxt%u: max Eager buffer RcvArray entries: %u\n",
-			    rcd->ctxt, rcd->egrbufs.count);
+		hfi1_cdbg(PROC,
+			  "ctxt%u: max Eager buffer RcvArray entries: %u\n",
+			  rcd->ctxt, rcd->egrbufs.count);
 
 		/*
 		 * Allocate array that will hold the eager buffer accounting
@@ -308,8 +310,8 @@ struct hfi1_ctxtdata *hfi1_create_ctxtdata(struct hfi1_pportdata *ppd, u32 ctxt)
 		 */
 		if (rcd->egrbufs.size < hfi1_max_mtu) {
 			rcd->egrbufs.size = __roundup_pow_of_two(hfi1_max_mtu);
-			dd_dev_info(dd,
-				    "ctxt%u: eager bufs size too small. Adjusting to %zu\n",
+			hfi1_cdbg(PROC,
+				  "ctxt%u: eager bufs size too small. Adjusting to %zu\n",
 				    rcd->ctxt, rcd->egrbufs.size);
 		}
 		rcd->egrbufs.rcvtid_size = HFI1_MAX_EAGER_BUFFER_SIZE;
@@ -1660,9 +1662,11 @@ int hfi1_setup_eagerbufs(struct hfi1_ctxtdata *rcd)
 	rcd->egrbufs.numbufs = idx;
 	rcd->egrbufs.size = alloced_bytes;
 
-	dd_dev_info(dd, "ctxt%u: Alloced %u rcv tid entries @ %uKB, total %zuKB\n",
-		rcd->ctxt, rcd->egrbufs.alloced, rcd->egrbufs.rcvtid_size,
-		rcd->egrbufs.size);
+	hfi1_cdbg(PROC,
+		  "ctxt%u: Alloced %u rcv tid entries @ %uKB, total %zuKB\n",
+		  rcd->ctxt, rcd->egrbufs.alloced, rcd->egrbufs.rcvtid_size,
+		  rcd->egrbufs.size);
+
 
 	/*
 	 * Set the contexts rcv array head update threshold to the closest
@@ -1683,13 +1687,14 @@ int hfi1_setup_eagerbufs(struct hfi1_ctxtdata *rcd)
 		rcd->expected_count = MAX_TID_PAIR_ENTRIES * 2;
 
 	rcd->expected_base = rcd->eager_base + egrtop;
-	dd_dev_info(dd, "ctxt%u: eager:%u, exp:%u, egrbase:%u, expbase:%u\n",
-		    rcd->ctxt, rcd->egrbufs.alloced, rcd->expected_count,
-		    rcd->eager_base, rcd->expected_base);
+	hfi1_cdbg(PROC, "ctxt%u: eager:%u, exp:%u, egrbase:%u, expbase:%u\n",
+		  rcd->ctxt, rcd->egrbufs.alloced, rcd->expected_count,
+		  rcd->eager_base, rcd->expected_base);
 
 	if (!hfi1_rcvbuf_validate(rcd->egrbufs.rcvtid_size, PT_EAGER, &order)) {
-		dd_dev_err(dd, "ctxt%u: current Eager buffer size is invalid %u\n",
-			   rcd->ctxt, rcd->egrbufs.rcvtid_size);
+		hfi1_cdbg(PROC,
+			  "ctxt%u: current Eager buffer size is invalid %u\n",
+			  rcd->ctxt, rcd->egrbufs.rcvtid_size);
 		ret = -EINVAL;
 		goto bail;
 	}
diff --git a/drivers/staging/rdma/hfi1/pio.c b/drivers/staging/rdma/hfi1/pio.c
index e5c32db..eab58c1 100644
--- a/drivers/staging/rdma/hfi1/pio.c
+++ b/drivers/staging/rdma/hfi1/pio.c
@@ -815,15 +815,16 @@ struct send_context *sc_alloc(struct hfi1_devdata *dd, int type,
 		}
 	}
 
-	dd_dev_info(dd,
-		"Send context %u(%u) %s group %u credits %u credit_ctrl 0x%llx threshold %u\n",
-		sw_index,
-		hw_context,
-		sc_type_name(type),
-		sc->group,
-		sc->credits,
-		sc->credit_ctrl,
-		thresh);
+	hfi1_cdbg(PIO,
+		  "Send context %u(%u) %s group %u credits %u credit_ctrl 0x%llx threshold %u\n",
+		  sw_index,
+		  hw_context,
+		  sc_type_name(type),
+		  sc->group,
+		  sc->credits,
+		  sc->credit_ctrl,
+		  thresh);
+
 
 	return sc;
 }
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 02/11] staging/rdma/hfi1: Clear the QSFP reset that is asserted on FLR

[Jubin John]

From: Easwar Hariharan <easwar.hariharan-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

The FLR on driver load asserts the QSFP reset pin and the driver does
not deassert it after. This patch allows the external QSFP cable to exit
reset by writing 1 to all the QSFP pins.

Reviewed-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Easwar Hariharan <easwar.hariharan-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/chip.c |   13 +++++--------
 1 files changed, 5 insertions(+), 8 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/chip.c b/drivers/staging/rdma/hfi1/chip.c
index e489819..e309828 100644
--- a/drivers/staging/rdma/hfi1/chip.c
+++ b/drivers/staging/rdma/hfi1/chip.c
@@ -9929,19 +9929,16 @@ static void init_chip(struct hfi1_devdata *dd)
 		setextled(dd, 0);
 	/*
 	 * Clear the QSFP reset.
-	 * A0 leaves the out lines floating on power on, then on an FLR
-	 * enforces a 0 on all out pins.  The driver does not touch
+	 * An FLR enforces a 0 on all out pins. The driver does not touch
 	 * ASIC_QSFPn_OUT otherwise.  This leaves RESET_N low and
-	 * anything  plugged constantly in reset, if it pays attention
+	 * anything plugged constantly in reset, if it pays attention
 	 * to RESET_N.
-	 * A prime example of this is SiPh. For now, set all pins high.
+	 * Prime examples of this are optical cables. Set all pins high.
 	 * I2CCLK and I2CDAT will change per direction, and INT_N and
 	 * MODPRS_N are input only and their value is ignored.
 	 */
-	if (is_a0(dd)) {
-		write_csr(dd, ASIC_QSFP1_OUT, 0x1f);
-		write_csr(dd, ASIC_QSFP2_OUT, 0x1f);
-	}
+	write_csr(dd, ASIC_QSFP1_OUT, 0x1f);
+	write_csr(dd, ASIC_QSFP2_OUT, 0x1f);
 }
 
 static void init_early_variables(struct hfi1_devdata *dd)
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 03/11] staging/rdma/hfi1: Enable WFR PCIe extended tags from the driver

[Jubin John]

From: Vennila Megavannan <vennila.megavannan-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Some BIOS implementations turn off extended tags in DevCtl (a RW
field) even though it was originally set and is advertised in DevCap
Fix is to set it in the driver

Reviewed-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Vennila Megavannan <vennila.megavannan-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/pcie.c |   12 +++++++++++-
 1 files changed, 11 insertions(+), 1 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/pcie.c b/drivers/staging/rdma/hfi1/pcie.c
index a956044..430e95a 100644
--- a/drivers/staging/rdma/hfi1/pcie.c
+++ b/drivers/staging/rdma/hfi1/pcie.c
@@ -475,8 +475,18 @@ static void tune_pcie_caps(struct hfi1_devdata *dd)
 {
 	struct pci_dev *parent;
 	u16 rc_mpss, rc_mps, ep_mpss, ep_mps;
-	u16 rc_mrrs, ep_mrrs, max_mrrs;
+	u16 rc_mrrs, ep_mrrs, max_mrrs, ectl;
 
+	/*
+	 * Turn on extended tags in DevCtl in case the BIOS has turned it off
+	 * to improve WFR SDMA bandwidth
+	 */
+	pcie_capability_read_word(dd->pcidev, PCI_EXP_DEVCTL, &ectl);
+	if (!(ectl & PCI_EXP_DEVCTL_EXT_TAG)) {
+		dd_dev_info(dd, "Enabling PCIe extended tags\n");
+		ectl |= PCI_EXP_DEVCTL_EXT_TAG;
+		pcie_capability_write_word(dd->pcidev, PCI_EXP_DEVCTL, ectl);
+	}
 	/* Find out supported and configured values for parent (root) */
 	parent = dd->pcidev->bus->self;
 	if (!pci_is_root_bus(parent->bus)) {
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 04/11] staging/rdma/hfi1: Always download SBus firmware

[Jubin John]

From: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

B0 dual port parts require the SBus firmware to always be
downloaded.

Remove reset of the SBus Master spico.  It is not necessary
since the SBus firmware download already does that.

Reviewed-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/firmware.c |    2 +-
 drivers/staging/rdma/hfi1/pcie.c     |   12 +-----------
 2 files changed, 2 insertions(+), 12 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/firmware.c b/drivers/staging/rdma/hfi1/firmware.c
index b4bdcf3..f311282 100644
--- a/drivers/staging/rdma/hfi1/firmware.c
+++ b/drivers/staging/rdma/hfi1/firmware.c
@@ -1568,7 +1568,7 @@ int load_pcie_firmware(struct hfi1_devdata *dd)
 	/* both firmware loads below use the SBus */
 	set_sbus_fast_mode(dd);
 
-	if (fw_sbus_load && (dd->flags & HFI1_DO_INIT_ASIC)) {
+	if (fw_sbus_load) {
 		turn_off_spicos(dd, SPICO_SBUS);
 		ret = load_sbus_firmware(dd, &fw_sbus);
 		if (ret)
diff --git a/drivers/staging/rdma/hfi1/pcie.c b/drivers/staging/rdma/hfi1/pcie.c
index 430e95a..01c0a5a 100644
--- a/drivers/staging/rdma/hfi1/pcie.c
+++ b/drivers/staging/rdma/hfi1/pcie.c
@@ -957,17 +957,7 @@ int do_pcie_gen3_transition(struct hfi1_devdata *dd)
 	}
 
 retry:
-
-	if (therm) {
-		/*
-		 * toggle SPICO_ENABLE to get back to the state
-		 * just after the firmware load
-		 */
-		sbus_request(dd, SBUS_MASTER_BROADCAST, 0x01,
-			WRITE_SBUS_RECEIVER, 0x00000040);
-		sbus_request(dd, SBUS_MASTER_BROADCAST, 0x01,
-			WRITE_SBUS_RECEIVER, 0x00000140);
-	}
+	/* the SBus download will reset the spico for thermal */
 
 	/* step 3: download SBus Master firmware */
 	/* step 4: download PCIe Gen3 SerDes firmware */
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 05/11] staging/rdma/hfi1: Disable thermal polling before sensor initialization

[Jubin John]

From: Jareer Abdel-Qader <jareer.h.abdel-qader-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

During driver load the thermal sensor needs to be reset prior
to initialization of the sensor. This prevents a possible sensor lock
up which can cause the wrong temperature value to be reported.
This fix leads to remove disabling thermal polling from
reset_asic_csrs() function.

Reviewed by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed by: Easwar Hariharan <easwar.hariharan-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jareer Abdel-Qader <jareer.h.abdel-qader-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Ira Weiny <ira.weiny-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/chip.c |    6 ++++--
 1 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/chip.c b/drivers/staging/rdma/hfi1/chip.c
index e309828..bf7cc2d 100644
--- a/drivers/staging/rdma/hfi1/chip.c
+++ b/drivers/staging/rdma/hfi1/chip.c
@@ -9455,7 +9455,7 @@ static void reset_asic_csrs(struct hfi1_devdata *dd)
 	/* We might want to retain this state across FLR if we ever use it */
 	write_csr(dd, ASIC_CFG_DRV_STR, 0);
 
-	write_csr(dd, ASIC_CFG_THERM_POLL_EN, 0);
+	/* ASIC_CFG_THERM_POLL_EN leave alone */
 	/* ASIC_STS_THERM read-only */
 	/* ASIC_CFG_RESET leave alone */
 
@@ -10800,7 +10800,9 @@ static int thermal_init(struct hfi1_devdata *dd)
 
 	acquire_hw_mutex(dd);
 	dd_dev_info(dd, "Initializing thermal sensor\n");
-
+	/* Disable polling of thermal readings */
+	write_csr(dd, ASIC_CFG_THERM_POLL_EN, 0x0);
+	msleep(100);
 	/* Thermal Sensor Initialization */
 	/*    Step 1: Reset the Thermal SBus Receiver */
 	ret = sbus_request_slow(dd, SBUS_THERMAL, 0x0,
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 06/11] staging/rdma/hfi1: Select only devices with active links

[Jubin John]

From: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

When looking for or validating a user device, only use devices
that are currently active.

Reviewed-by: Mitko Haralanov <mitko.haralanov-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/file_ops.c |   21 ++++++++++++++++++---
 1 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/file_ops.c b/drivers/staging/rdma/hfi1/file_ops.c
index fba5889..6fc7691 100644
--- a/drivers/staging/rdma/hfi1/file_ops.c
+++ b/drivers/staging/rdma/hfi1/file_ops.c
@@ -874,6 +874,14 @@ done:
 	return ret;
 }
 
+/* return true if the device available for general use */
+static int usable_device(struct hfi1_devdata *dd)
+{
+	struct hfi1_pportdata *ppd = dd->pport;
+
+	return driver_lstate(ppd) == IB_PORT_ACTIVE;
+}
+
 static int get_user_context(struct file *fp, struct hfi1_user_info *uinfo,
 			    int devno, unsigned alg)
 {
@@ -903,7 +911,11 @@ static int get_user_context(struct file *fp, struct hfi1_user_info *uinfo,
 
 			for (dev = 0; dev < devmax; dev++) {
 				pdd = hfi1_lookup(dev);
-				if (pdd && pdd->freectxts &&
+				if (!pdd)
+					continue;
+				if (!usable_device(pdd))
+					continue;
+				if (pdd->freectxts &&
 				    pdd->freectxts > free) {
 					dd = pdd;
 					free = pdd->freectxts;
@@ -912,7 +924,11 @@ static int get_user_context(struct file *fp, struct hfi1_user_info *uinfo,
 		} else {
 			for (dev = 0; dev < devmax; dev++) {
 				pdd = hfi1_lookup(dev);
-				if (pdd && pdd->freectxts) {
+				if (!pdd)
+					continue;
+				if (!usable_device(pdd))
+					continue;
+				if (pdd->freectxts) {
 					dd = pdd;
 					break;
 				}
@@ -936,7 +952,6 @@ static int find_shared_ctxt(struct file *fp,
 	for (ndev = 0; ndev < devmax; ndev++) {
 		struct hfi1_devdata *dd = hfi1_lookup(ndev);
 
-		/* device portion of usable() */
 		if (!(dd && (dd->flags & HFI1_PRESENT) && dd->kregbase))
 			continue;
 		for (i = dd->first_user_ctxt; i < dd->num_rcv_contexts; i++) {
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 07/11] staging/rdma/hfi1: Fix for opaportconfig ledon by not checking for portNum

[Jubin John]

From: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

opaportconfig ledon fails with error message due to port
number being checked in the attr modifier. This change
removes the check for the port number in AttrMod, so the
P field is ignored.

Reviewed-by: Easwar Hariharan <easwar.hariharan-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Sebastian Sanchez <sebastian.sanchez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/mad.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/mad.c b/drivers/staging/rdma/hfi1/mad.c
index 32f7037..a122565 100644
--- a/drivers/staging/rdma/hfi1/mad.c
+++ b/drivers/staging/rdma/hfi1/mad.c
@@ -3458,7 +3458,7 @@ static int __subn_get_opa_led_info(struct opa_smp *smp, u32 am, u8 *data,
 	u32 nport = OPA_AM_NPORT(am);
 	u64 reg;
 
-	if (nport != 1 || OPA_AM_PORTNUM(am)) {
+	if (nport != 1) {
 		smp->status |= IB_SMP_INVALID_FIELD;
 		return reply((struct ib_mad_hdr *)smp);
 	}
@@ -3483,7 +3483,7 @@ static int __subn_set_opa_led_info(struct opa_smp *smp, u32 am, u8 *data,
 	u32 nport = OPA_AM_NPORT(am);
 	int on = !!(be32_to_cpu(p->rsvd_led_mask) & OPA_LED_MASK);
 
-	if (nport != 1 || OPA_AM_PORTNUM(am)) {
+	if (nport != 1) {
 		smp->status |= IB_SMP_INVALID_FIELD;
 		return reply((struct ib_mad_hdr *)smp);
 	}
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 08/11] staging/rdma/hfi1: Remove spurious error messages

[Jubin John]

From: Ignacio Hernandez <ignacio.hernandez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Changed the order in which diagnostics messages are printed,
taking into account the cases where the errors are handled in
rcv_hdrerr() and no further message is needed to report.

Reviewed-by: Mark Debbage <mark.debbage-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Arthur Kepner <arthur.kepner-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Ignacio Hernandez <ignacio.hernandez-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/driver.c |   26 +++++++++++++-------------
 1 files changed, 13 insertions(+), 13 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/driver.c b/drivers/staging/rdma/hfi1/driver.c
index ce69141..61fc626 100644
--- a/drivers/staging/rdma/hfi1/driver.c
+++ b/drivers/staging/rdma/hfi1/driver.c
@@ -1163,20 +1163,20 @@ void handle_eflags(struct hfi1_packet *packet)
 	struct hfi1_ctxtdata *rcd = packet->rcd;
 	u32 rte = rhf_rcv_type_err(packet->rhf);
 
-	dd_dev_err(rcd->dd,
-		"receive context %d: rhf 0x%016llx, errs [ %s%s%s%s%s%s%s%s] rte 0x%x\n",
-		rcd->ctxt, packet->rhf,
-		packet->rhf & RHF_K_HDR_LEN_ERR ? "k_hdr_len " : "",
-		packet->rhf & RHF_DC_UNC_ERR ? "dc_unc " : "",
-		packet->rhf & RHF_DC_ERR ? "dc " : "",
-		packet->rhf & RHF_TID_ERR ? "tid " : "",
-		packet->rhf & RHF_LEN_ERR ? "len " : "",
-		packet->rhf & RHF_ECC_ERR ? "ecc " : "",
-		packet->rhf & RHF_VCRC_ERR ? "vcrc " : "",
-		packet->rhf & RHF_ICRC_ERR ? "icrc " : "",
-		rte);
-
 	rcv_hdrerr(rcd, rcd->ppd, packet);
+	if (rhf_err_flags(packet->rhf))
+		dd_dev_err(rcd->dd,
+			   "receive context %d: rhf 0x%016llx, errs [ %s%s%s%s%s%s%s%s] rte 0x%x\n",
+			   rcd->ctxt, packet->rhf,
+			   packet->rhf & RHF_K_HDR_LEN_ERR ? "k_hdr_len " : "",
+			   packet->rhf & RHF_DC_UNC_ERR ? "dc_unc " : "",
+			   packet->rhf & RHF_DC_ERR ? "dc " : "",
+			   packet->rhf & RHF_TID_ERR ? "tid " : "",
+			   packet->rhf & RHF_LEN_ERR ? "len " : "",
+			   packet->rhf & RHF_ECC_ERR ? "ecc " : "",
+			   packet->rhf & RHF_VCRC_ERR ? "vcrc " : "",
+			   packet->rhf & RHF_ICRC_ERR ? "icrc " : "",
+			   rte);
 }
 
 /*
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 09/11] staging/rdma/hfi1: use one-shot LCB write

[Jubin John]

From: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Use the one-shot LCB write implemented in the 8051 firmware.
This speeds up 8051 LCB writes by 2x.  Use old method for older
firmwares.

Reviewed-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/chip.c |   42 +++++++++++++++++++++++++++++++++----
 drivers/staging/rdma/hfi1/chip.h |    1 +
 2 files changed, 38 insertions(+), 5 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/chip.c b/drivers/staging/rdma/hfi1/chip.c
index bf7cc2d..7658c76 100644
--- a/drivers/staging/rdma/hfi1/chip.c
+++ b/drivers/staging/rdma/hfi1/chip.c
@@ -4774,13 +4774,25 @@ int read_lcb_csr(struct hfi1_devdata *dd, u32 addr, u64 *data)
  */
 static int write_lcb_via_8051(struct hfi1_devdata *dd, u32 addr, u64 data)
 {
+	u32 regno;
+	int ret;
 
-	if (acquire_lcb_access(dd, 0) == 0) {
-		write_csr(dd, addr, data);
-		release_lcb_access(dd, 0);
-		return 0;
+	if (dd->icode == ICODE_FUNCTIONAL_SIMULATOR ||
+	    (dd->dc8051_ver < dc8051_ver(0, 20))) {
+		if (acquire_lcb_access(dd, 0) == 0) {
+			write_csr(dd, addr, data);
+			release_lcb_access(dd, 0);
+			return 0;
+		}
+		return -EBUSY;
 	}
-	return -EBUSY;
+
+	/* register is an index of LCB registers: (offset - base) / 8 */
+	regno = (addr - DC_LCB_CFG_RUN) >> 3;
+	ret = do_8051_command(dd, HCMD_WRITE_LCB_CSR, regno, &data);
+	if (ret != HCMD_SUCCESS)
+		return -EBUSY;
+	return 0;
 }
 
 /*
@@ -4862,6 +4874,26 @@ static int do_8051_command(
 	 */
 
 	/*
+	 * When writing a LCB CSR, out_data contains the full value to
+	 * to be written, while in_data contains the relative LCB
+	 * address in 7:0.  Do the work here, rather than the caller,
+	 * of distrubting the write data to where it needs to go:
+	 *
+	 * Write data
+	 *   39:00 -> in_data[47:8]
+	 *   47:40 -> DC8051_CFG_EXT_DEV_0.RETURN_CODE
+	 *   63:48 -> DC8051_CFG_EXT_DEV_0.RSP_DATA
+	 */
+	if (type == HCMD_WRITE_LCB_CSR) {
+		in_data |= ((*out_data) & 0xffffffffffull) << 8;
+		reg = ((((*out_data) >> 40) & 0xff) <<
+				DC_DC8051_CFG_EXT_DEV_0_RETURN_CODE_SHIFT)
+		      | ((((*out_data) >> 48) & 0xffff) <<
+				DC_DC8051_CFG_EXT_DEV_0_RSP_DATA_SHIFT);
+		write_csr(dd, DC_DC8051_CFG_EXT_DEV_0, reg);
+	}
+
+	/*
 	 * Do two writes: the first to stabilize the type and req_data, the
 	 * second to activate.
 	 */
diff --git a/drivers/staging/rdma/hfi1/chip.h b/drivers/staging/rdma/hfi1/chip.h
index ebf9041..d74aed8 100644
--- a/drivers/staging/rdma/hfi1/chip.h
+++ b/drivers/staging/rdma/hfi1/chip.h
@@ -235,6 +235,7 @@
 #define HCMD_MISC		   0x05
 #define HCMD_READ_LCB_IDLE_MSG 0x06
 #define HCMD_READ_LCB_CSR      0x07
+#define HCMD_WRITE_LCB_CSR     0x08
 #define HCMD_INTERFACE_TEST	   0xff
 
 /* DC_DC8051_CFG_HOST_CMD_1.RETURN_CODE - 8051 host command return */
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 10/11] staging/rdma/hfi1: Eliminate WARN_ON when VL is invalid

[Jubin John]

From: Ira Weiny <iweiny-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>

sdma_select_engine_vl only needs to protect itself from an invalid VL.
Something higher up the stack should be warning the user when they try
to use an SL which maps to an invalid VL.

Reviewed-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Kaike Wan <kaike.wan-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Ira Weiny <iweiny-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/sdma.c |   12 ++++++++++--
 1 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/sdma.c b/drivers/staging/rdma/hfi1/sdma.c
index 2a1da21..998bf43 100644
--- a/drivers/staging/rdma/hfi1/sdma.c
+++ b/drivers/staging/rdma/hfi1/sdma.c
@@ -777,8 +777,14 @@ struct sdma_engine *sdma_select_engine_vl(
 	struct sdma_map_elem *e;
 	struct sdma_engine *rval;
 
-	if (WARN_ON(vl > 8))
-		return NULL;
+	/* NOTE This should only happen if SC->VL changed after the initial
+	 *      checks on the QP/AH
+	 *      Default will return engine 0 below
+	 */
+	if (unlikely(vl >= num_vls)) {
+		rval = NULL;
+		goto done;
+	}
 
 	rcu_read_lock();
 	m = rcu_dereference(dd->sdma_map);
@@ -790,6 +796,8 @@ struct sdma_engine *sdma_select_engine_vl(
 	rval = e->sde[selector & e->mask];
 	rcu_read_unlock();
 
+done:
+	rval =  !rval ? &dd->per_sdma[0] : rval;
 	trace_hfi1_sdma_engine_select(dd, selector, vl, rval->this_idx);
 	return rval;
 }
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH RESEND 11/11] staging/rdma/hfi1: Workaround to prevent corruption during packet delivery

[Jubin John]

From: Mark F. Brown <mark.f.brown-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>

Disabling one receive context when RX_DMA is receiving a packet can cause
incorrect packet delivery for a subsequent packet on another receive
context.

This is resolved by doing the following:
1. Programming dummy tail address for every receive context
   before enabling it
2. While deallocating receive context resetting tail address
   to dummy address
3. Leaving the dummy address in when disabling tail update
4. When disabling receive context leaving tail update enabled

Reviewed-by: Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Reviewed-by: Mitko Haralanov <mitko.haralanov-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Mark F. Brown <mark.f.brown-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/staging/rdma/hfi1/chip.c |   18 +++++++++++++++---
 drivers/staging/rdma/hfi1/hfi.h  |    4 ++++
 drivers/staging/rdma/hfi1/init.c |   28 ++++++++++++++++++++++++++++
 3 files changed, 47 insertions(+), 3 deletions(-)

diff --git a/drivers/staging/rdma/hfi1/chip.c b/drivers/staging/rdma/hfi1/chip.c
index 7658c76..e74c32e 100644
--- a/drivers/staging/rdma/hfi1/chip.c
+++ b/drivers/staging/rdma/hfi1/chip.c
@@ -7785,6 +7785,17 @@ void hfi1_rcvctrl(struct hfi1_devdata *dd, unsigned int op, int ctxt)
 	}
 	if (op & HFI1_RCVCTRL_CTXT_DIS) {
 		write_csr(dd, RCV_VL15, 0);
+		/*
+		 * When receive context is being disabled turn on tail
+		 * update with a dummy tail address and then disable
+		 * receive context.
+		 */
+		if (dd->rcvhdrtail_dummy_physaddr) {
+			write_kctxt_csr(dd, ctxt, RCV_HDR_TAIL_ADDR,
+					dd->rcvhdrtail_dummy_physaddr);
+			rcvctrl |= RCV_CTXT_CTRL_TAIL_UPD_SMASK;
+		}
+
 		rcvctrl &= ~RCV_CTXT_CTRL_ENABLE_SMASK;
 	}
 	if (op & HFI1_RCVCTRL_INTRAVAIL_ENB)
@@ -7854,10 +7865,11 @@ void hfi1_rcvctrl(struct hfi1_devdata *dd, unsigned int op, int ctxt)
 	if (op & (HFI1_RCVCTRL_TAILUPD_DIS | HFI1_RCVCTRL_CTXT_DIS))
 		/*
 		 * If the context has been disabled and the Tail Update has
-		 * been cleared, clear the RCV_HDR_TAIL_ADDR CSR so
-		 * it doesn't contain an address that is invalid.
+		 * been cleared, set the RCV_HDR_TAIL_ADDR CSR to dummy address
+		 * so it doesn't contain an address that is invalid.
 		 */
-		write_kctxt_csr(dd, ctxt, RCV_HDR_TAIL_ADDR, 0);
+		write_kctxt_csr(dd, ctxt, RCV_HDR_TAIL_ADDR,
+				dd->rcvhdrtail_dummy_physaddr);
 }
 
 u32 hfi1_read_cntrs(struct hfi1_devdata *dd, loff_t pos, char **namep,
diff --git a/drivers/staging/rdma/hfi1/hfi.h b/drivers/staging/rdma/hfi1/hfi.h
index 190f7a2..02f2deb 100644
--- a/drivers/staging/rdma/hfi1/hfi.h
+++ b/drivers/staging/rdma/hfi1/hfi.h
@@ -1084,6 +1084,10 @@ struct hfi1_devdata {
 	/* Save the enabled LCB error bits */
 	u64 lcb_err_en;
 	u8 dc_shutdown;
+
+	/* receive context tail dummy address */
+	volatile __le64 *rcvhdrtail_dummy_kvaddr;
+	dma_addr_t rcvhdrtail_dummy_physaddr;
 };
 
 /* 8051 firmware version helper */
diff --git a/drivers/staging/rdma/hfi1/init.c b/drivers/staging/rdma/hfi1/init.c
index 4a79744..36a3065 100644
--- a/drivers/staging/rdma/hfi1/init.c
+++ b/drivers/staging/rdma/hfi1/init.c
@@ -693,6 +693,18 @@ int hfi1_init(struct hfi1_devdata *dd, int reinit)
 	if (ret)
 		goto done;
 
+	/* allocate dummy tail memory for all receive contexts */
+	dd->rcvhdrtail_dummy_kvaddr = dma_zalloc_coherent(
+		&dd->pcidev->dev, sizeof(u64),
+		&dd->rcvhdrtail_dummy_physaddr,
+		GFP_KERNEL);
+
+	if (!dd->rcvhdrtail_dummy_kvaddr) {
+		dd_dev_err(dd, "cannot allocate dummy tail memory\n");
+		ret = -ENOMEM;
+		goto done;
+	}
+
 	/* dd->rcd can be NULL if early initialization failed */
 	for (i = 0; dd->rcd && i < dd->first_user_ctxt; ++i) {
 		/*
@@ -1268,6 +1280,14 @@ static void cleanup_device_data(struct hfi1_devdata *dd)
 	tmp = dd->rcd;
 	dd->rcd = NULL;
 	spin_unlock_irqrestore(&dd->uctxt_lock, flags);
+
+	if (dd->rcvhdrtail_dummy_kvaddr) {
+		dma_free_coherent(&dd->pcidev->dev, sizeof(u64),
+				  (void *)dd->rcvhdrtail_dummy_kvaddr,
+				  dd->rcvhdrtail_dummy_physaddr);
+				  dd->rcvhdrtail_dummy_kvaddr = NULL;
+	}
+
 	for (ctxt = 0; tmp && ctxt < dd->num_rcv_contexts; ctxt++) {
 		struct hfi1_ctxtdata *rcd = tmp[ctxt];
 
@@ -1523,6 +1543,14 @@ int hfi1_create_rcvhdrq(struct hfi1_devdata *dd, struct hfi1_ctxtdata *rcd)
 	reg = (dd->rcvhdrsize & RCV_HDR_SIZE_HDR_SIZE_MASK)
 		<< RCV_HDR_SIZE_HDR_SIZE_SHIFT;
 	write_kctxt_csr(dd, rcd->ctxt, RCV_HDR_SIZE, reg);
+
+	/*
+	 * Program dummy tail address for every receive context
+	 * before enabling any receive context
+	 */
+	write_kctxt_csr(dd, rcd->ctxt, RCV_HDR_TAIL_ADDR,
+			dd->rcvhdrtail_dummy_physaddr);
+
 	return 0;
 
 bail_free:
-- 
1.7.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH RESEND 11/11] staging/rdma/hfi1: Workaround to prevent corruption during packet delivery

[Greg KH]

On Fri, Nov 06, 2015 at 08:07:06PM -0500, Jubin John wrote:
> --- a/drivers/staging/rdma/hfi1/hfi.h
> +++ b/drivers/staging/rdma/hfi1/hfi.h
> @@ -1084,6 +1084,10 @@ struct hfi1_devdata {
>  	/* Save the enabled LCB error bits */
>  	u64 lcb_err_en;
>  	u8 dc_shutdown;
> +
> +	/* receive context tail dummy address */
> +	volatile __le64 *rcvhdrtail_dummy_kvaddr;

There's no way that volatile here is actually correct, please fix.

Re: [PATCH RESEND 10/11] staging/rdma/hfi1: Eliminate WARN_ON when VL is invalid

[Greg KH]

On Fri, Nov 06, 2015 at 08:07:05PM -0500, Jubin John wrote:
> From: Ira Weiny <iweiny-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> 
> sdma_select_engine_vl only needs to protect itself from an invalid VL.
> Something higher up the stack should be warning the user when they try
> to use an SL which maps to an invalid VL.
> 
> Reviewed-by: Dean Luick <dean.luick-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> Reviewed-by: Mike Marciniszyn <mike.marciniszyn-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> Reviewed-by: Kaike Wan <kaike.wan-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> Signed-off-by: Ira Weiny <iweiny-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> Signed-off-by: Jubin John <jubin.john-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> ---
>  drivers/staging/rdma/hfi1/sdma.c |   12 ++++++++++--
>  1 files changed, 10 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/staging/rdma/hfi1/sdma.c b/drivers/staging/rdma/hfi1/sdma.c
> index 2a1da21..998bf43 100644
> --- a/drivers/staging/rdma/hfi1/sdma.c
> +++ b/drivers/staging/rdma/hfi1/sdma.c
> @@ -777,8 +777,14 @@ struct sdma_engine *sdma_select_engine_vl(
>  	struct sdma_map_elem *e;
>  	struct sdma_engine *rval;
>  
> -	if (WARN_ON(vl > 8))
> -		return NULL;
> +	/* NOTE This should only happen if SC->VL changed after the initial
> +	 *      checks on the QP/AH
> +	 *      Default will return engine 0 below
> +	 */
> +	if (unlikely(vl >= num_vls)) {

Can you prove that unlikely() makes a measured difference here?  If not,
please remove it.  If you can, please provide the proof in the changelog
when you resend it.

thanks,

greg k-h
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH RESEND 11/11] staging/rdma/hfi1: Workaround to prevent corruption during packet delivery

[Jubin John]

On Fri, Nov 06, 2015 at 07:25:31PM -0800, Greg KH wrote:
> On Fri, Nov 06, 2015 at 08:07:06PM -0500, Jubin John wrote:
> > --- a/drivers/staging/rdma/hfi1/hfi.h
> > +++ b/drivers/staging/rdma/hfi1/hfi.h
> > @@ -1084,6 +1084,10 @@ struct hfi1_devdata {
> >  	/* Save the enabled LCB error bits */
> >  	u64 lcb_err_en;
> >  	u8 dc_shutdown;
> > +
> > +	/* receive context tail dummy address */
> > +	volatile __le64 *rcvhdrtail_dummy_kvaddr;
> 
> There's no way that volatile here is actually correct, please fix.
> 
You're right. Will fix in v2.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH RESEND 10/11] staging/rdma/hfi1: Eliminate WARN_ON when VL is invalid

[Jubin John]

> > -	if (WARN_ON(vl > 8))
> > -		return NULL;
> > +	/* NOTE This should only happen if SC->VL changed after the initial
> > +	 *      checks on the QP/AH
> > +	 *      Default will return engine 0 below
> > +	 */
> > +	if (unlikely(vl >= num_vls)) {
> 
> Can you prove that unlikely() makes a measured difference here?  If not,
> please remove it.  If you can, please provide the proof in the changelog
> when you resend it.

Unfortunately, I can't prove this so will remove in v2.

Thanks,
Jubin John
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html