* [PATCH 09/16] ehca: event queue
From: Heiko J Schick @ 2006-05-15 17:42 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
drivers/infiniband/hw/ehca/ehca_eq.c | 222 +++++++++++++++++++++++++++++++++++
1 file changed, 222 insertions(+)
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_eq.c 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_eq.c 2006-05-15 13:29:49.000000000 +0200
@@ -0,0 +1,222 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * Event queue handling
+ *
+ * Authors: Waleri Fomin <fomin@de.ibm.com>
+ * Khadija Souissi <souissi@de.ibm.com>
+ * Reinhard Ernst <rernst@de.ibm.com>
+ * Heiko J Schick <schickhj@de.ibm.com>
+ * Hoang-Nam Nguyen <hnguyen@de.ibm.com>
+ *
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#define DEB_PREFIX "e_eq"
+
+#include "ehca_classes.h"
+#include "ehca_irq.h"
+#include "ehca_iverbs.h"
+#include "ehca_qes.h"
+#include "hcp_if.h"
+#include "ipz_pt_fn.h"
+
+int ehca_create_eq(struct ehca_shca *shca,
+ struct ehca_eq *eq,
+ const enum ehca_eq_type type, const u32 length)
+{
+ u64 ret = H_SUCCESS;
+ u32 nr_pages = 0;
+ u32 i;
+ void *vpage = NULL;
+
+ EDEB_EN(7, "shca=%p eq=%p length=%x", shca, eq, length);
+ EHCA_CHECK_ADR(shca);
+ EHCA_CHECK_ADR(eq);
+
+ spin_lock_init(&eq->spinlock);
+ eq->is_initialized = 0;
+
+ if (type != EHCA_EQ && type != EHCA_NEQ) {
+ EDEB_ERR(4, "Invalid EQ type %x. eq=%p", type, eq);
+ return -EINVAL;
+ }
+ if (length == 0) {
+ EDEB_ERR(4, "EQ length must not be zero. eq=%p", eq);
+ return -EINVAL;
+ }
+
+ ret = hipz_h_alloc_resource_eq(shca->ipz_hca_handle,
+ &eq->pf,
+ type,
+ length,
+ &eq->ipz_eq_handle,
+ &eq->length,
+ &nr_pages, &eq->ist);
+
+ if (ret != H_SUCCESS) {
+ EDEB_ERR(4, "Can't allocate EQ / NEQ. eq=%p", eq);
+ return -EINVAL;
+ }
+
+ ret = ipz_queue_ctor(&eq->ipz_queue, nr_pages,
+ EHCA_PAGESIZE, sizeof(struct ehca_eqe), 0);
+ if (!ret) {
+ EDEB_ERR(4, "Can't allocate EQ pages. eq=%p", eq);
+ goto create_eq_exit1;
+ }
+
+ for (i = 0; i < nr_pages; i++) {
+ u64 rpage;
+
+ if (!(vpage = ipz_qpageit_get_inc(&eq->ipz_queue))) {
+ ret = H_RESOURCE;
+ goto create_eq_exit2;
+ }
+
+ rpage = virt_to_abs(vpage);
+ ret = hipz_h_register_rpage_eq(shca->ipz_hca_handle,
+ eq->ipz_eq_handle,
+ &eq->pf,
+ 0, 0, rpage, 1);
+
+ if (i == (nr_pages - 1)) {
+ /* last page */
+ vpage = ipz_qpageit_get_inc(&eq->ipz_queue);
+ if (ret != H_SUCCESS || vpage)
+ goto create_eq_exit2;
+ } else {
+ if (ret != H_PAGE_REGISTERED || !vpage)
+ goto create_eq_exit2;
+ }
+ }
+
+ ipz_qeit_reset(&eq->ipz_queue);
+
+ /* register interrupt handlers and initialize work queues */
+ if (type == EHCA_EQ) {
+ ret = ibmebus_request_irq(NULL, eq->ist, ehca_interrupt_eq,
+ SA_INTERRUPT, "ehca_eq",
+ (void *)shca);
+ if (ret < 0)
+ EDEB_ERR(4, "Can't map interrupt handler.");
+
+ tasklet_init(&eq->interrupt_task, ehca_tasklet_eq, (long)shca);
+ } else if (type == EHCA_NEQ) {
+ ret = ibmebus_request_irq(NULL, eq->ist, ehca_interrupt_neq,
+ SA_INTERRUPT, "ehca_neq",
+ (void *)shca);
+ if (ret < 0)
+ EDEB_ERR(4, "Can't map interrupt handler.");
+
+ tasklet_init(&eq->interrupt_task, ehca_tasklet_neq, (long)shca);
+ }
+
+ eq->is_initialized = 1;
+
+ EDEB_EX(7, "ret=%lx", ret);
+
+ return 0;
+
+create_eq_exit2:
+ ipz_queue_dtor(&eq->ipz_queue);
+
+create_eq_exit1:
+ hipz_h_destroy_eq(shca->ipz_hca_handle, eq);
+
+ EDEB_EX(7, "ret=%lx", ret);
+
+ return -EINVAL;
+}
+
+void *ehca_poll_eq(struct ehca_shca *shca, struct ehca_eq *eq)
+{
+ unsigned long flags = 0;
+ void *eqe = NULL;
+
+ EDEB_EN(7, "shca=%p eq=%p", shca, eq);
+ EHCA_CHECK_ADR_P(shca);
+ EHCA_CHECK_EQ_P(eq);
+
+ spin_lock_irqsave(&eq->spinlock, flags);
+ eqe = ipz_eqit_eq_get_inc_valid(&eq->ipz_queue);
+ spin_unlock_irqrestore(&eq->spinlock, flags);
+
+ EDEB_EX(7, "eq=%p eqe=%p", eq, eqe);
+
+ return eqe;
+}
+
+void ehca_poll_eqs(unsigned long data)
+{
+ struct ehca_shca *shca;
+ struct ehca_module *module = (struct ehca_module*)data;
+
+ spin_lock(&module->shca_lock);
+ list_for_each_entry(shca, &module->shca_list, shca_list) {
+ if (shca->eq.is_initialized)
+ ehca_tasklet_eq((unsigned long)(void*)shca);
+ }
+ mod_timer(&module->timer, jiffies + HZ);
+ spin_unlock(&module->shca_lock);
+
+ return;
+}
+
+int ehca_destroy_eq(struct ehca_shca *shca, struct ehca_eq *eq)
+{
+ unsigned long flags = 0;
+ u64 h_ret = H_SUCCESS;
+
+ EDEB_EN(7, "shca=%p eq=%p", shca, eq);
+ EHCA_CHECK_ADR(shca);
+ EHCA_CHECK_EQ(eq);
+
+ spin_lock_irqsave(&eq->spinlock, flags);
+ ibmebus_free_irq(NULL, eq->ist, (void *)shca);
+
+ h_ret = hipz_h_destroy_eq(shca->ipz_hca_handle, eq);
+
+ spin_unlock_irqrestore(&eq->spinlock, flags);
+
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "Can't free EQ resources.");
+ return -EINVAL;
+ }
+ ipz_queue_dtor(&eq->ipz_queue);
+
+ EDEB_EX(7, "h_ret=%lx", h_ret);
+
+ return h_ret;
+}
^ permalink raw reply
* [PATCH 08/16] ehca: protection domain and address vector
From: Heiko J Schick @ 2006-05-15 17:42 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
drivers/infiniband/hw/ehca/ehca_av.c | 306 +++++++++++++++++++++++++++++++++++
drivers/infiniband/hw/ehca/ehca_pd.c | 118 +++++++++++++
2 files changed, 424 insertions(+)
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_av.c 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_av.c 2006-05-12 12:45:25.000000000 +0200
@@ -0,0 +1,306 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * adress vector functions
+ *
+ * Authors: Hoang-Nam Nguyen <hnguyen@de.ibm.com>
+ * Khadija Souissi <souissik@de.ibm.com>
+ * Reinhard Ernst <rernst@de.ibm.com>
+ * Christoph Raisch <raisch@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+
+#define DEB_PREFIX "ehav"
+
+#include <asm/current.h>
+
+#include "ehca_tools.h"
+#include "ehca_iverbs.h"
+#include "hcp_if.h"
+
+struct ib_ah *ehca_create_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr)
+{
+ extern struct ehca_module ehca_module;
+ extern int ehca_static_rate;
+ int ret = 0;
+ struct ehca_av *av = NULL;
+ struct ehca_shca *shca = NULL;
+
+ EHCA_CHECK_PD_P(pd);
+ EHCA_CHECK_ADR_P(ah_attr);
+
+ shca = container_of(pd->device, struct ehca_shca, ib_device);
+
+ EDEB_EN(7, "pd=%p ah_attr=%p", pd, ah_attr);
+
+ av = kmem_cache_alloc(ehca_module.cache_av, SLAB_KERNEL);
+ if (!av) {
+ EDEB_ERR(4, "Out of memory pd=%p ah_attr=%p", pd, ah_attr);
+ ret = -ENOMEM;
+ goto create_ah_exit0;
+ }
+
+ av->av.sl = ah_attr->sl;
+ av->av.dlid = ntohs(ah_attr->dlid);
+ av->av.slid_path_bits = ah_attr->src_path_bits;
+
+ if (ehca_static_rate < 0) {
+ int ah_mult = ib_rate_to_mult(ah_attr->static_rate);
+ int ehca_mult =
+ ib_rate_to_mult(shca->sport[ah_attr->port_num].rate );
+
+ if (ah_mult >= ehca_mult)
+ av->av.ipd = 0;
+ else
+ av->av.ipd = (ah_mult > 0) ?
+ ((ehca_mult - 1) / ah_mult) : 0;
+ } else
+ av->av.ipd = ehca_static_rate;
+
+ EDEB(7, "IPD av->av.ipd set =%x ah_attr->static_rate=%x "
+ "shca_ib_rate=%x ",av->av.ipd, ah_attr->static_rate,
+ shca->sport[ah_attr->port_num].rate);
+
+ av->av.lnh = ah_attr->ah_flags;
+ av->av.grh.word_0 |= EHCA_BMASK_SET(GRH_IPVERSION_MASK, 6);
+ av->av.grh.word_0 |= EHCA_BMASK_SET(GRH_TCLASS_MASK,
+ ah_attr->grh.traffic_class);
+ av->av.grh.word_0 |= EHCA_BMASK_SET(GRH_FLOWLABEL_MASK,
+ ah_attr->grh.flow_label);
+ av->av.grh.word_0 |= EHCA_BMASK_SET(GRH_HOPLIMIT_MASK,
+ ah_attr->grh.hop_limit);
+ av->av.grh.word_0 |= EHCA_BMASK_SET(GRH_NEXTHEADER_MASK, 0x1B);
+ /* IB transport */
+ av->av.grh.word_0 = be64_to_cpu(av->av.grh.word_0);
+ /* set sgid in grh.word_1 */
+ if (ah_attr->ah_flags & IB_AH_GRH) {
+ int rc = 0;
+ struct ib_port_attr port_attr;
+ union ib_gid gid;
+ memset(&port_attr, 0, sizeof(port_attr));
+ rc = ehca_query_port(pd->device, ah_attr->port_num,
+ &port_attr);
+ if (rc) { /* invalid port number */
+ ret = -EINVAL;
+ EDEB_ERR(4, "Invalid port number "
+ "ehca_query_port() returned %x "
+ "pd=%p ah_attr=%p", rc, pd, ah_attr);
+ goto create_ah_exit1;
+ }
+ memset(&gid, 0, sizeof(gid));
+ rc = ehca_query_gid(pd->device,
+ ah_attr->port_num,
+ ah_attr->grh.sgid_index, &gid);
+ if (rc) {
+ ret = -EINVAL;
+ EDEB_ERR(4, "Failed to retrieve sgid "
+ "ehca_query_gid() returned %x "
+ "pd=%p ah_attr=%p", rc, pd, ah_attr);
+ goto create_ah_exit1;
+ }
+ memcpy(&av->av.grh.word_1, &gid, sizeof(gid));
+ }
+ /* for the time being we use a hard coded PMTU of 2048 Bytes */
+ av->av.pmtu = 4;
+
+ /* dgid comes in grh.word_3 */
+ memcpy(&av->av.grh.word_3, &ah_attr->grh.dgid,
+ sizeof(ah_attr->grh.dgid));
+
+ EHCA_REGISTER_AV(device, pd);
+
+ EDEB_EX(7, "pd=%p ah_attr=%p av=%p", pd, ah_attr, av);
+ return &av->ib_ah;
+
+create_ah_exit1:
+ kmem_cache_free(ehca_module.cache_av, av);
+
+create_ah_exit0:
+ EDEB_EX(7, "ret=%x pd=%p ah_attr=%p", ret, pd, ah_attr);
+
+ return ERR_PTR(ret);
+}
+
+int ehca_modify_ah(struct ib_ah *ah, struct ib_ah_attr *ah_attr)
+{
+ struct ehca_av *av = NULL;
+ struct ehca_ud_av new_ehca_av;
+ struct ehca_pd *my_pd = NULL;
+ u32 cur_pid = current->tgid;
+ int ret = 0;
+
+ EHCA_CHECK_AV(ah);
+ EHCA_CHECK_ADR(ah_attr);
+
+ EDEB_EN(7, "ah=%p ah_attr=%p", ah, ah_attr);
+
+ my_pd = container_of(ah->pd, struct ehca_pd, ib_pd);
+ if (my_pd->ib_pd.uobject && my_pd->ib_pd.uobject->context &&
+ my_pd->ownpid != cur_pid) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, my_pd->ownpid);
+ return -EINVAL;
+ }
+
+ memset(&new_ehca_av, 0, sizeof(new_ehca_av));
+ new_ehca_av.sl = ah_attr->sl;
+ new_ehca_av.dlid = ntohs(ah_attr->dlid);
+ new_ehca_av.slid_path_bits = ah_attr->src_path_bits;
+ new_ehca_av.ipd = ah_attr->static_rate;
+ new_ehca_av.lnh = EHCA_BMASK_SET(GRH_FLAG_MASK,
+ ((ah_attr->ah_flags & IB_AH_GRH) > 0));
+ new_ehca_av.grh.word_0 = EHCA_BMASK_SET(GRH_TCLASS_MASK,
+ ah_attr->grh.traffic_class);
+ new_ehca_av.grh.word_0 |= EHCA_BMASK_SET(GRH_FLOWLABEL_MASK,
+ ah_attr->grh.flow_label);
+ new_ehca_av.grh.word_0 |= EHCA_BMASK_SET(GRH_HOPLIMIT_MASK,
+ ah_attr->grh.hop_limit);
+ new_ehca_av.grh.word_0 |= EHCA_BMASK_SET(GRH_NEXTHEADER_MASK, 0x1b);
+ new_ehca_av.grh.word_0 = be64_to_cpu(new_ehca_av.grh.word_0);
+
+ /* set sgid in grh.word_1 */
+ if (ah_attr->ah_flags & IB_AH_GRH) {
+ int rc = 0;
+ struct ib_port_attr port_attr;
+ union ib_gid gid;
+ memset(&port_attr, 0, sizeof(port_attr));
+ rc = ehca_query_port(ah->device, ah_attr->port_num,
+ &port_attr);
+ if (rc) { /* invalid port number */
+ ret = -EINVAL;
+ EDEB_ERR(4, "Invalid port number "
+ "ehca_query_port() returned %x "
+ "ah=%p ah_attr=%p port_num=%x",
+ rc, ah, ah_attr, ah_attr->port_num);
+ goto modify_ah_exit1;
+ }
+ memset(&gid, 0, sizeof(gid));
+ rc = ehca_query_gid(ah->device,
+ ah_attr->port_num,
+ ah_attr->grh.sgid_index, &gid);
+ if (rc) {
+ ret = -EINVAL;
+ EDEB_ERR(4, "Failed to retrieve sgid "
+ "ehca_query_gid() returned %x "
+ "ah=%p ah_attr=%p port_num=%x "
+ "sgid_index=%x",
+ rc, ah, ah_attr, ah_attr->port_num,
+ ah_attr->grh.sgid_index);
+ goto modify_ah_exit1;
+ }
+ memcpy(&new_ehca_av.grh.word_1, &gid, sizeof(gid));
+ }
+
+ new_ehca_av.pmtu = 4; /* see also comment in create_ah() */
+
+ memcpy(&new_ehca_av.grh.word_3, &ah_attr->grh.dgid,
+ sizeof(ah_attr->grh.dgid));
+
+ av = container_of(ah, struct ehca_av, ib_ah);
+ av->av = new_ehca_av;
+
+modify_ah_exit1:
+ EDEB_EX(7, "ret=%x ah=%p ah_attr=%p", ret, ah, ah_attr);
+
+ return ret;
+}
+
+int ehca_query_ah(struct ib_ah *ah, struct ib_ah_attr *ah_attr)
+{
+ int ret = 0;
+ struct ehca_av *av = NULL;
+ struct ehca_pd *my_pd = NULL;
+ u32 cur_pid = current->tgid;
+
+ EHCA_CHECK_AV(ah);
+ EHCA_CHECK_ADR(ah_attr);
+
+ EDEB_EN(7, "ah=%p ah_attr=%p", ah, ah_attr);
+
+ my_pd = container_of(ah->pd, struct ehca_pd, ib_pd);
+ if (my_pd->ib_pd.uobject && my_pd->ib_pd.uobject->context &&
+ my_pd->ownpid != cur_pid) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, my_pd->ownpid);
+ return -EINVAL;
+ }
+
+ av = container_of(ah, struct ehca_av, ib_ah);
+ memcpy(&ah_attr->grh.dgid, &av->av.grh.word_3,
+ sizeof(ah_attr->grh.dgid));
+ ah_attr->sl = av->av.sl;
+
+ ah_attr->dlid = av->av.dlid;
+
+ ah_attr->src_path_bits = av->av.slid_path_bits;
+ ah_attr->static_rate = av->av.ipd;
+ ah_attr->ah_flags = EHCA_BMASK_GET(GRH_FLAG_MASK, av->av.lnh);
+ ah_attr->grh.traffic_class = EHCA_BMASK_GET(GRH_TCLASS_MASK,
+ av->av.grh.word_0);
+ ah_attr->grh.hop_limit = EHCA_BMASK_GET(GRH_HOPLIMIT_MASK,
+ av->av.grh.word_0);
+ ah_attr->grh.flow_label = EHCA_BMASK_GET(GRH_FLOWLABEL_MASK,
+ av->av.grh.word_0);
+
+ EDEB_EX(7, "ah=%p ah_attr=%p ret=%x", ah, ah_attr, ret);
+ return ret;
+}
+
+int ehca_destroy_ah(struct ib_ah *ah)
+{
+ extern struct ehca_module ehca_module;
+ struct ehca_pd *my_pd = NULL;
+ u32 cur_pid = current->tgid;
+ int ret = 0;
+
+ EHCA_CHECK_AV(ah);
+ EHCA_DEREGISTER_AV(ah);
+
+ EDEB_EN(7, "ah=%p", ah);
+
+ my_pd = container_of(ah->pd, struct ehca_pd, ib_pd);
+ if (my_pd->ib_pd.uobject && my_pd->ib_pd.uobject->context &&
+ my_pd->ownpid != cur_pid) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, my_pd->ownpid);
+ return -EINVAL;
+ }
+
+ kmem_cache_free(ehca_module.cache_av,
+ container_of(ah, struct ehca_av, ib_ah));
+
+ EDEB_EX(7, "ret=%x ah=%p", ret, ah);
+ return ret;
+}
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_pd.c 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_pd.c 2006-05-15 13:35:24.000000000 +0200
@@ -0,0 +1,118 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * PD functions
+ *
+ * Authors: Christoph Raisch <raisch@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+
+#define DEB_PREFIX "vpd "
+
+#include <asm/current.h>
+
+#include "ehca_tools.h"
+#include "ehca_iverbs.h"
+
+struct ib_pd *ehca_alloc_pd(struct ib_device *device,
+ struct ib_ucontext *context, struct ib_udata *udata)
+{
+ extern struct ehca_module ehca_module;
+ struct ib_pd *mypd = NULL;
+ struct ehca_pd *pd = NULL;
+
+ EDEB_EN(7, "device=%p context=%p udata=%p", device, context, udata);
+
+ EHCA_CHECK_DEVICE_P(device);
+
+ pd = kmem_cache_alloc(ehca_module.cache_pd, SLAB_KERNEL);
+ if (!pd) {
+ EDEB_ERR(4, "ERROR device=%p context=%p pd=%p"
+ " out of memory", device, context, mypd);
+ return ERR_PTR(-ENOMEM);
+ }
+
+ memset(pd, 0, sizeof(struct ehca_pd));
+ pd->ownpid = current->tgid;
+
+ /* Kernel PD: when device = -1, 0
+ * User PD: when context != -1
+ */
+ if (!context) {
+ /* Kernel PDs after init reuses always
+ * the one created in ehca_shca_reopen()
+ */
+ struct ehca_shca *shca = container_of(device, struct ehca_shca,
+ ib_device);
+ pd->fw_pd.value = shca->pd->fw_pd.value;
+ } else
+ pd->fw_pd.value = (u64)pd;
+
+ mypd = &pd->ib_pd;
+
+ EHCA_REGISTER_PD(device, pd);
+
+ EDEB_EX(7, "device=%p context=%p pd=%p", device, context, mypd);
+
+ return mypd;
+}
+
+int ehca_dealloc_pd(struct ib_pd *pd)
+{
+ extern struct ehca_module ehca_module;
+ int ret = 0;
+ u32 cur_pid = current->tgid;
+ struct ehca_pd *my_pd = NULL;
+
+ EDEB_EN(7, "pd=%p", pd);
+
+ EHCA_CHECK_PD(pd);
+ my_pd = container_of(pd, struct ehca_pd, ib_pd);
+ if (my_pd->ib_pd.uobject && my_pd->ib_pd.uobject->context &&
+ my_pd->ownpid != cur_pid) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, my_pd->ownpid);
+ return -EINVAL;
+ }
+
+ EHCA_DEREGISTER_PD(pd);
+
+ kmem_cache_free(ehca_module.cache_pd,
+ container_of(pd, struct ehca_pd, ib_pd));
+
+ EDEB_EX(7, "pd=%p", pd);
+
+ return ret;
+}
^ permalink raw reply
* [PATCH 07/16] ehca: memory region
From: Heiko J Schick @ 2006-05-15 17:42 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
drivers/infiniband/hw/ehca/ehca_mrmw.c | 2474 +++++++++++++++++++++++++++++++++
drivers/infiniband/hw/ehca/ehca_mrmw.h | 143 +
2 files changed, 2617 insertions(+)
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_mrmw.h 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_mrmw.h 2006-04-28 15:18:05.000000000 +0200
@@ -0,0 +1,143 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * MR/MW declarations and inline functions
+ *
+ * Authors: Dietmar Decker <ddecker@de.ibm.com>
+ * Christoph Raisch <raisch@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef _EHCA_MRMW_H_
+#define _EHCA_MRMW_H_
+
+#undef DEB_PREFIX
+#define DEB_PREFIX "mrmw"
+
+int ehca_reg_mr(struct ehca_shca *shca,
+ struct ehca_mr *e_mr,
+ u64 *iova_start,
+ u64 size,
+ int acl,
+ struct ehca_pd *e_pd,
+ struct ehca_mr_pginfo *pginfo,
+ u32 *lkey,
+ u32 *rkey);
+
+int ehca_reg_mr_rpages(struct ehca_shca *shca,
+ struct ehca_mr *e_mr,
+ struct ehca_mr_pginfo *pginfo);
+
+int ehca_rereg_mr(struct ehca_shca *shca,
+ struct ehca_mr *e_mr,
+ u64 *iova_start,
+ u64 size,
+ int mr_access_flags,
+ struct ehca_pd *e_pd,
+ struct ehca_mr_pginfo *pginfo,
+ u32 *lkey,
+ u32 *rkey);
+
+int ehca_unmap_one_fmr(struct ehca_shca *shca,
+ struct ehca_mr *e_fmr);
+
+int ehca_reg_smr(struct ehca_shca *shca,
+ struct ehca_mr *e_origmr,
+ struct ehca_mr *e_newmr,
+ u64 *iova_start,
+ int acl,
+ struct ehca_pd *e_pd,
+ u32 *lkey,
+ u32 *rkey);
+
+int ehca_reg_internal_maxmr(struct ehca_shca *shca,
+ struct ehca_pd *e_pd,
+ struct ehca_mr **maxmr);
+
+int ehca_reg_maxmr(struct ehca_shca *shca,
+ struct ehca_mr *e_newmr,
+ u64 *iova_start,
+ int acl,
+ struct ehca_pd *e_pd,
+ u32 *lkey,
+ u32 *rkey);
+
+int ehca_dereg_internal_maxmr(struct ehca_shca *shca);
+
+int ehca_mr_chk_buf_and_calc_size(struct ib_phys_buf *phys_buf_array,
+ int num_phys_buf,
+ u64 *iova_start,
+ u64 *size);
+
+int ehca_fmr_check_page_list(struct ehca_mr *e_fmr,
+ u64 *page_list,
+ int list_len);
+
+int ehca_set_pagebuf(struct ehca_mr *e_mr,
+ struct ehca_mr_pginfo *pginfo,
+ u32 number,
+ u64 *kpage);
+
+int ehca_set_pagebuf_1(struct ehca_mr *e_mr,
+ struct ehca_mr_pginfo *pginfo,
+ u64 *rpage);
+
+int ehca_mr_is_maxmr(u64 size,
+ u64 *iova_start);
+
+void ehca_mrmw_map_acl(int ib_acl,
+ u32 *hipz_acl);
+
+void ehca_mrmw_set_pgsize_hipz_acl(u32 *hipz_acl);
+
+void ehca_mrmw_reverse_map_acl(const u32 *hipz_acl,
+ int *ib_acl);
+
+int ehca_mrmw_map_hrc_alloc(const u64 hipz_rc);
+
+int ehca_mrmw_map_hrc_rrpg_last(const u64 hipz_rc);
+
+int ehca_mrmw_map_hrc_rrpg_notlast(const u64 hipz_rc);
+
+int ehca_mrmw_map_hrc_query_mr(const u64 hipz_rc);
+
+int ehca_mrmw_map_hrc_free_mr(const u64 hipz_rc);
+
+int ehca_mrmw_map_hrc_free_mw(const u64 hipz_rc);
+
+int ehca_mrmw_map_hrc_reg_smr(const u64 hipz_rc);
+
+void ehca_mr_deletenew(struct ehca_mr *mr);
+
+#endif /*_EHCA_MRMW_H_*/
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_mrmw.c 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_mrmw.c 2006-05-15 15:43:31.000000000 +0200
@@ -0,0 +1,2474 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * MR/MW functions
+ *
+ * Authors: Dietmar Decker <ddecker@de.ibm.com>
+ * Christoph Raisch <raisch@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#undef DEB_PREFIX
+#define DEB_PREFIX "mrmw"
+
+#include <asm/current.h>
+
+#include "ehca_iverbs.h"
+#include "ehca_mrmw.h"
+#include "hcp_if.h"
+#include "hipz_hw.h"
+
+extern int ehca_use_hp_mr;
+
+static struct ehca_mr *ehca_mr_new(void)
+{
+ extern struct ehca_module ehca_module;
+ struct ehca_mr *me;
+
+ me = kmem_cache_alloc(ehca_module.cache_mr, SLAB_KERNEL);
+ if (me) {
+ memset(me, 0, sizeof(struct ehca_mr));
+ spin_lock_init(&me->mrlock);
+ EDEB_EX(7, "ehca_mr=%p sizeof(ehca_mr_t)=%x", me,
+ (u32) sizeof(struct ehca_mr));
+ } else {
+ EDEB_ERR(3, "alloc failed");
+ }
+
+ return me;
+}
+
+static void ehca_mr_delete(struct ehca_mr *me)
+{
+ extern struct ehca_module ehca_module;
+
+ kmem_cache_free(ehca_module.cache_mr, me);
+}
+
+static struct ehca_mw *ehca_mw_new(void)
+{
+ extern struct ehca_module ehca_module;
+ struct ehca_mw *me;
+
+ me = kmem_cache_alloc(ehca_module.cache_mw, SLAB_KERNEL);
+ if (me) {
+ memset(me, 0, sizeof(struct ehca_mw));
+ spin_lock_init(&me->mwlock);
+ EDEB_EX(7, "ehca_mw=%p sizeof(ehca_mw_t)=%x", me,
+ (u32) sizeof(struct ehca_mw));
+ } else {
+ EDEB_ERR(3, "alloc failed");
+ }
+
+ return me;
+}
+
+static void ehca_mw_delete(struct ehca_mw *me)
+{
+ extern struct ehca_module ehca_module;
+
+ kmem_cache_free(ehca_module.cache_mw, me);
+}
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+struct ib_mr *ehca_get_dma_mr(struct ib_pd *pd, int mr_access_flags)
+{
+ struct ib_mr *ib_mr = NULL;
+ int ret = 0;
+ struct ehca_mr *e_maxmr = NULL;
+ struct ehca_pd *e_pd = NULL;
+ struct ehca_shca *shca = NULL;
+
+ EDEB_EN(7, "pd=%p mr_access_flags=%x", pd, mr_access_flags);
+
+ EHCA_CHECK_PD_P(pd);
+ e_pd = container_of(pd, struct ehca_pd, ib_pd);
+ shca = container_of(pd->device, struct ehca_shca, ib_device);
+
+ if (shca->maxmr) {
+ e_maxmr = ehca_mr_new();
+ if (!e_maxmr) {
+ EDEB_ERR(4, "out of memory");
+ ib_mr = ERR_PTR(-ENOMEM);
+ goto get_dma_mr_exit0;
+ }
+
+ ret = ehca_reg_maxmr(shca, e_maxmr, (u64*)KERNELBASE,
+ mr_access_flags, e_pd,
+ &e_maxmr->ib.ib_mr.lkey,
+ &e_maxmr->ib.ib_mr.rkey);
+ if (ret) {
+ ib_mr = ERR_PTR(ret);
+ goto get_dma_mr_exit0;
+ }
+ ib_mr = &e_maxmr->ib.ib_mr;
+ } else {
+ EDEB_ERR(4, "no internal max-MR exist!");
+ ib_mr = ERR_PTR(-EINVAL);
+ goto get_dma_mr_exit0;
+ }
+
+get_dma_mr_exit0:
+ if (IS_ERR(ib_mr))
+ EDEB_EX(4, "rc=%lx pd=%p mr_access_flags=%x ",
+ PTR_ERR(ib_mr), pd, mr_access_flags);
+ else
+ EDEB_EX(7, "ib_mr=%p lkey=%x rkey=%x",
+ ib_mr, ib_mr->lkey, ib_mr->rkey);
+ return ib_mr;
+} /* end ehca_get_dma_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+struct ib_mr *ehca_reg_phys_mr(struct ib_pd *pd,
+ struct ib_phys_buf *phys_buf_array,
+ int num_phys_buf,
+ int mr_access_flags,
+ u64 *iova_start)
+{
+ struct ib_mr *ib_mr = NULL;
+ int ret = 0;
+ struct ehca_mr *e_mr = NULL;
+ struct ehca_shca *shca = NULL;
+ struct ehca_pd *e_pd = NULL;
+ u64 size = 0;
+ struct ehca_mr_pginfo pginfo={0,0,0,0,0,0,0,NULL,0,NULL,NULL,0,NULL,0};
+ u32 num_pages_mr = 0;
+ u32 num_pages_4k = 0; /* 4k portion "pages" */
+
+ EDEB_EN(7, "pd=%p phys_buf_array=%p num_phys_buf=%x "
+ "mr_access_flags=%x iova_start=%p", pd, phys_buf_array,
+ num_phys_buf, mr_access_flags, iova_start);
+
+ EHCA_CHECK_PD_P(pd);
+ if ((num_phys_buf <= 0) || ehca_adr_bad(phys_buf_array)) {
+ EDEB_ERR(4, "bad input values: num_phys_buf=%x "
+ "phys_buf_array=%p", num_phys_buf, phys_buf_array);
+ ib_mr = ERR_PTR(-EINVAL);
+ goto reg_phys_mr_exit0;
+ }
+ if (((mr_access_flags & IB_ACCESS_REMOTE_WRITE) &&
+ !(mr_access_flags & IB_ACCESS_LOCAL_WRITE)) ||
+ ((mr_access_flags & IB_ACCESS_REMOTE_ATOMIC) &&
+ !(mr_access_flags & IB_ACCESS_LOCAL_WRITE))) {
+ /* Remote Write Access requires Local Write Access */
+ /* Remote Atomic Access requires Local Write Access */
+ EDEB_ERR(4, "bad input values: mr_access_flags=%x",
+ mr_access_flags);
+ ib_mr = ERR_PTR(-EINVAL);
+ goto reg_phys_mr_exit0;
+ }
+
+ /* check physical buffer list and calculate size */
+ ret = ehca_mr_chk_buf_and_calc_size(phys_buf_array, num_phys_buf,
+ iova_start, &size);
+ if (ret) {
+ ib_mr = ERR_PTR(ret);
+ goto reg_phys_mr_exit0;
+ }
+ if ((size == 0) ||
+ (((u64)iova_start + size) < (u64)iova_start)) {
+ EDEB_ERR(4, "bad input values: size=%lx iova_start=%p",
+ size, iova_start);
+ ib_mr = ERR_PTR(-EINVAL);
+ goto reg_phys_mr_exit0;
+ }
+
+ e_pd = container_of(pd, struct ehca_pd, ib_pd);
+ shca = container_of(pd->device, struct ehca_shca, ib_device);
+
+ e_mr = ehca_mr_new();
+ if (!e_mr) {
+ EDEB_ERR(4, "out of memory");
+ ib_mr = ERR_PTR(-ENOMEM);
+ goto reg_phys_mr_exit0;
+ }
+
+ /* determine number of MR pages */
+ num_pages_mr = ((((u64)iova_start % PAGE_SIZE) + size +
+ PAGE_SIZE - 1) / PAGE_SIZE);
+ num_pages_4k = ((((u64)iova_start % EHCA_PAGESIZE) + size +
+ EHCA_PAGESIZE - 1) / EHCA_PAGESIZE);
+
+ /* register MR on HCA */
+ if (ehca_mr_is_maxmr(size, iova_start)) {
+ e_mr->flags |= EHCA_MR_FLAG_MAXMR;
+ ret = ehca_reg_maxmr(shca, e_mr, iova_start, mr_access_flags,
+ e_pd, &e_mr->ib.ib_mr.lkey,
+ &e_mr->ib.ib_mr.rkey);
+ if (ret) {
+ ib_mr = ERR_PTR(ret);
+ goto reg_phys_mr_exit1;
+ }
+ } else {
+ pginfo.type = EHCA_MR_PGI_PHYS;
+ pginfo.num_pages = num_pages_mr;
+ pginfo.num_4k = num_pages_4k;
+ pginfo.num_phys_buf = num_phys_buf;
+ pginfo.phys_buf_array = phys_buf_array;
+ pginfo.next_4k = (((u64)iova_start & ~PAGE_MASK) /
+ EHCA_PAGESIZE);
+
+ ret = ehca_reg_mr(shca, e_mr, iova_start, size, mr_access_flags,
+ e_pd, &pginfo, &e_mr->ib.ib_mr.lkey,
+ &e_mr->ib.ib_mr.rkey);
+ if (ret) {
+ ib_mr = ERR_PTR(ret);
+ goto reg_phys_mr_exit1;
+ }
+ }
+
+ /* successful registration of all pages */
+ ib_mr = &e_mr->ib.ib_mr;
+ goto reg_phys_mr_exit0;
+
+reg_phys_mr_exit1:
+ ehca_mr_delete(e_mr);
+reg_phys_mr_exit0:
+ if (IS_ERR(ib_mr))
+ EDEB_EX(4, "rc=%lx pd=%p phys_buf_array=%p "
+ "num_phys_buf=%x mr_access_flags=%x iova_start=%p",
+ PTR_ERR(ib_mr), pd, phys_buf_array,
+ num_phys_buf, mr_access_flags, iova_start);
+ else
+ EDEB_EX(7, "ib_mr=%p lkey=%x rkey=%x",
+ ib_mr, ib_mr->lkey, ib_mr->rkey);
+ return ib_mr;
+} /* end ehca_reg_phys_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+struct ib_mr *ehca_reg_user_mr(struct ib_pd *pd,
+ struct ib_umem *region,
+ int mr_access_flags,
+ struct ib_udata *udata)
+{
+ struct ib_mr *ib_mr = NULL;
+ struct ehca_mr *e_mr = NULL;
+ struct ehca_shca *shca = NULL;
+ struct ehca_pd *e_pd = NULL;
+ struct ehca_mr_pginfo pginfo={0,0,0,0,0,0,0,NULL,0,NULL,NULL,0,NULL,0};
+ int ret = 0;
+ u32 num_pages_mr = 0;
+ u32 num_pages_4k = 0; /* 4k portion "pages" */
+
+ EDEB_EN(7, "pd=%p region=%p mr_access_flags=%x udata=%p",
+ pd, region, mr_access_flags, udata);
+
+ EHCA_CHECK_PD_P(pd);
+ if (ehca_adr_bad(region)) {
+ EDEB_ERR(4, "bad input values: region=%p", region);
+ ib_mr = ERR_PTR(-EINVAL);
+ goto reg_user_mr_exit0;
+ }
+ if (((mr_access_flags & IB_ACCESS_REMOTE_WRITE) &&
+ !(mr_access_flags & IB_ACCESS_LOCAL_WRITE)) ||
+ ((mr_access_flags & IB_ACCESS_REMOTE_ATOMIC) &&
+ !(mr_access_flags & IB_ACCESS_LOCAL_WRITE))) {
+ /* Remote Write Access requires Local Write Access */
+ /* Remote Atomic Access requires Local Write Access */
+ EDEB_ERR(4, "bad input values: mr_access_flags=%x",
+ mr_access_flags);
+ ib_mr = ERR_PTR(-EINVAL);
+ goto reg_user_mr_exit0;
+ }
+ EDEB(7, "user_base=%lx virt_base=%lx length=%lx offset=%x page_size=%x "
+ "chunk_list.next=%p",
+ region->user_base, region->virt_base, region->length,
+ region->offset, region->page_size, region->chunk_list.next);
+ if (region->page_size != PAGE_SIZE) {
+ EDEB_ERR(4, "page size not supported, region->page_size=%x",
+ region->page_size);
+ ib_mr = ERR_PTR(-EINVAL);
+ goto reg_user_mr_exit0;
+ }
+
+ if ((region->length == 0) ||
+ ((region->virt_base + region->length) < region->virt_base)) {
+ EDEB_ERR(4, "bad input values: length=%lx virt_base=%lx",
+ region->length, region->virt_base);
+ ib_mr = ERR_PTR(-EINVAL);
+ goto reg_user_mr_exit0;
+ }
+
+ e_pd = container_of(pd, struct ehca_pd, ib_pd);
+ shca = container_of(pd->device, struct ehca_shca, ib_device);
+
+ e_mr = ehca_mr_new();
+ if (!e_mr) {
+ EDEB_ERR(4, "out of memory");
+ ib_mr = ERR_PTR(-ENOMEM);
+ goto reg_user_mr_exit0;
+ }
+
+ /* determine number of MR pages */
+ num_pages_mr = (((region->virt_base % PAGE_SIZE) + region->length +
+ PAGE_SIZE - 1) / PAGE_SIZE);
+ num_pages_4k = (((region->virt_base % EHCA_PAGESIZE) + region->length +
+ EHCA_PAGESIZE - 1) / EHCA_PAGESIZE);
+
+ /* register MR on HCA */
+ pginfo.type = EHCA_MR_PGI_USER;
+ pginfo.num_pages = num_pages_mr;
+ pginfo.num_4k = num_pages_4k;
+ pginfo.region = region;
+ pginfo.next_4k = region->offset / EHCA_PAGESIZE;
+ pginfo.next_chunk = list_prepare_entry(pginfo.next_chunk,
+ (®ion->chunk_list),
+ list);
+
+ ret = ehca_reg_mr(shca, e_mr, (u64*)region->virt_base,
+ region->length, mr_access_flags, e_pd, &pginfo,
+ &e_mr->ib.ib_mr.lkey, &e_mr->ib.ib_mr.rkey);
+ if (ret) {
+ ib_mr = ERR_PTR(ret);
+ goto reg_user_mr_exit1;
+ }
+
+ /* successful registration of all pages */
+ ib_mr = &e_mr->ib.ib_mr;
+ goto reg_user_mr_exit0;
+
+reg_user_mr_exit1:
+ ehca_mr_delete(e_mr);
+reg_user_mr_exit0:
+ if (IS_ERR(ib_mr))
+ EDEB_EX(4, "rc=%lx pd=%p region=%p mr_access_flags=%x "
+ "udata=%p",
+ PTR_ERR(ib_mr), pd, region, mr_access_flags, udata);
+ else
+ EDEB_EX(7, "ib_mr=%p lkey=%x rkey=%x",
+ ib_mr, ib_mr->lkey, ib_mr->rkey);
+ return ib_mr;
+} /* end ehca_reg_user_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_rereg_phys_mr(struct ib_mr *mr,
+ int mr_rereg_mask,
+ struct ib_pd *pd,
+ struct ib_phys_buf *phys_buf_array,
+ int num_phys_buf,
+ int mr_access_flags,
+ u64 *iova_start)
+{
+ int ret = 0;
+ struct ehca_shca *shca = NULL;
+ struct ehca_mr *e_mr = NULL;
+ u64 new_size = 0;
+ u64 *new_start = NULL;
+ u32 new_acl = 0;
+ struct ehca_pd *new_pd = NULL;
+ u32 tmp_lkey = 0;
+ u32 tmp_rkey = 0;
+ unsigned long sl_flags;
+ u32 num_pages_mr = 0;
+ u32 num_pages_4k = 0; /* 4k portion "pages" */
+ struct ehca_mr_pginfo pginfo={0,0,0,0,0,0,0,NULL,0,NULL,NULL,0,NULL,0};
+ struct ehca_pd *my_pd = NULL;
+ u32 cur_pid = current->tgid;
+
+ EDEB_EN(7, "mr=%p mr_rereg_mask=%x pd=%p phys_buf_array=%p "
+ "num_phys_buf=%x mr_access_flags=%x iova_start=%p",
+ mr, mr_rereg_mask, pd, phys_buf_array, num_phys_buf,
+ mr_access_flags, iova_start);
+
+ EHCA_CHECK_MR(mr);
+ my_pd = container_of(mr->pd, struct ehca_pd, ib_pd);
+ if (my_pd->ib_pd.uobject && my_pd->ib_pd.uobject->context &&
+ (my_pd->ownpid != cur_pid)) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, my_pd->ownpid);
+ ret = -EINVAL;
+ goto rereg_phys_mr_exit0;
+ }
+
+ if (!(mr_rereg_mask & IB_MR_REREG_TRANS)) {
+ /* TODO not supported, because PHYP rereg hCall needs pages*/
+ /* TODO: We will follow this with Tom ....*/
+ EDEB_ERR(4, "rereg without IB_MR_REREG_TRANS not supported yet,"
+ " mr_rereg_mask=%x", mr_rereg_mask);
+ ret = -EINVAL;
+ goto rereg_phys_mr_exit0;
+ }
+
+ e_mr = container_of(mr, struct ehca_mr, ib.ib_mr);
+ if (mr_rereg_mask & IB_MR_REREG_PD) {
+ EHCA_CHECK_PD(pd);
+ }
+
+ if ((mr_rereg_mask &
+ ~(IB_MR_REREG_TRANS | IB_MR_REREG_PD | IB_MR_REREG_ACCESS)) ||
+ (mr_rereg_mask == 0)) {
+ ret = -EINVAL;
+ goto rereg_phys_mr_exit0;
+ }
+
+ shca = container_of(mr->device, struct ehca_shca, ib_device);
+
+ /* check other parameters */
+ if (e_mr == shca->maxmr) {
+ /* should be impossible, however reject to be sure */
+ EDEB_ERR(3, "rereg internal max-MR impossible, mr=%p "
+ "shca->maxmr=%p mr->lkey=%x",
+ mr, shca->maxmr, mr->lkey);
+ ret = -EINVAL;
+ goto rereg_phys_mr_exit0;
+ }
+ if (mr_rereg_mask & IB_MR_REREG_TRANS) { /* transl., i.e. addr/size */
+ if (e_mr->flags & EHCA_MR_FLAG_FMR) {
+ EDEB_ERR(4, "not supported for FMR, mr=%p flags=%x",
+ mr, e_mr->flags);
+ ret = -EINVAL;
+ goto rereg_phys_mr_exit0;
+ }
+ if (ehca_adr_bad(phys_buf_array) || num_phys_buf <= 0) {
+ EDEB_ERR(4, "bad input values: mr_rereg_mask=%x "
+ "phys_buf_array=%p num_phys_buf=%x",
+ mr_rereg_mask, phys_buf_array, num_phys_buf);
+ ret = -EINVAL;
+ goto rereg_phys_mr_exit0;
+ }
+ }
+ if ((mr_rereg_mask & IB_MR_REREG_ACCESS) && /* change ACL */
+ (((mr_access_flags & IB_ACCESS_REMOTE_WRITE) &&
+ !(mr_access_flags & IB_ACCESS_LOCAL_WRITE)) ||
+ ((mr_access_flags & IB_ACCESS_REMOTE_ATOMIC) &&
+ !(mr_access_flags & IB_ACCESS_LOCAL_WRITE)))) {
+ /* Remote Write Access requires Local Write Access */
+ /* Remote Atomic Access requires Local Write Access */
+ EDEB_ERR(4, "bad input values: mr_rereg_mask=%x "
+ "mr_access_flags=%x", mr_rereg_mask, mr_access_flags);
+ ret = -EINVAL;
+ goto rereg_phys_mr_exit0;
+ }
+
+ /* set requested values dependent on rereg request */
+ spin_lock_irqsave(&e_mr->mrlock, sl_flags);
+ new_start = e_mr->start; /* new == old address */
+ new_size = e_mr->size; /* new == old length */
+ new_acl = e_mr->acl; /* new == old access control */
+ new_pd = container_of(mr->pd,struct ehca_pd,ib_pd); /*new == old PD*/
+
+ if (mr_rereg_mask & IB_MR_REREG_TRANS) {
+ new_start = iova_start; /* change address */
+ /* check physical buffer list and calculate size */
+ ret = ehca_mr_chk_buf_and_calc_size(phys_buf_array,
+ num_phys_buf, iova_start,
+ &new_size);
+ if (ret)
+ goto rereg_phys_mr_exit1;
+ if ((new_size == 0) ||
+ (((u64)iova_start + new_size) < (u64)iova_start)) {
+ EDEB_ERR(4, "bad input values: new_size=%lx "
+ "iova_start=%p", new_size, iova_start);
+ ret = -EINVAL;
+ goto rereg_phys_mr_exit1;
+ }
+ num_pages_mr = ((((u64)new_start % PAGE_SIZE) + new_size +
+ PAGE_SIZE - 1) / PAGE_SIZE);
+ num_pages_4k = ((((u64)new_start % EHCA_PAGESIZE) + new_size +
+ EHCA_PAGESIZE - 1) / EHCA_PAGESIZE);
+ pginfo.type = EHCA_MR_PGI_PHYS;
+ pginfo.num_pages = num_pages_mr;
+ pginfo.num_4k = num_pages_4k;
+ pginfo.num_phys_buf = num_phys_buf;
+ pginfo.phys_buf_array = phys_buf_array;
+ pginfo.next_4k = (((u64)iova_start & ~PAGE_MASK) /
+ EHCA_PAGESIZE);
+ }
+ if (mr_rereg_mask & IB_MR_REREG_ACCESS)
+ new_acl = mr_access_flags;
+ if (mr_rereg_mask & IB_MR_REREG_PD)
+ new_pd = container_of(pd, struct ehca_pd, ib_pd);
+
+ EDEB(7, "mr=%p new_start=%p new_size=%lx new_acl=%x new_pd=%p "
+ "num_pages_mr=%x num_pages_4k=%x", e_mr, new_start, new_size,
+ new_acl, new_pd, num_pages_mr, num_pages_4k);
+
+ ret = ehca_rereg_mr(shca, e_mr, new_start, new_size, new_acl,
+ new_pd, &pginfo, &tmp_lkey, &tmp_rkey);
+ if (ret)
+ goto rereg_phys_mr_exit1;
+
+ /* successful reregistration */
+ if (mr_rereg_mask & IB_MR_REREG_PD)
+ mr->pd = pd;
+ mr->lkey = tmp_lkey;
+ mr->rkey = tmp_rkey;
+
+rereg_phys_mr_exit1:
+ spin_unlock_irqrestore(&e_mr->mrlock, sl_flags);
+rereg_phys_mr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x mr=%p mr_rereg_mask=%x pd=%p "
+ "phys_buf_array=%p num_phys_buf=%x mr_access_flags=%x "
+ "iova_start=%p",
+ ret, mr, mr_rereg_mask, pd, phys_buf_array,
+ num_phys_buf, mr_access_flags, iova_start);
+ else
+ EDEB_EX(7, "mr=%p mr_rereg_mask=%x pd=%p phys_buf_array=%p "
+ "num_phys_buf=%x mr_access_flags=%x iova_start=%p",
+ mr, mr_rereg_mask, pd, phys_buf_array, num_phys_buf,
+ mr_access_flags, iova_start);
+
+ return ret;
+} /* end ehca_rereg_phys_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_query_mr(struct ib_mr *mr, struct ib_mr_attr *mr_attr)
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ struct ehca_shca *shca = NULL;
+ struct ehca_mr *e_mr = NULL;
+ struct ehca_pd *my_pd = NULL;
+ u32 cur_pid = current->tgid;
+ unsigned long sl_flags;
+ struct ehca_mr_hipzout_parms hipzout = {{0},0,0,0,0,0};
+
+ EDEB_EN(7, "mr=%p mr_attr=%p", mr, mr_attr);
+
+ EHCA_CHECK_MR(mr);
+
+ my_pd = container_of(mr->pd, struct ehca_pd, ib_pd);
+ if (my_pd->ib_pd.uobject && my_pd->ib_pd.uobject->context &&
+ (my_pd->ownpid != cur_pid)) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, my_pd->ownpid);
+ ret = -EINVAL;
+ goto query_mr_exit0;
+ }
+
+ e_mr = container_of(mr, struct ehca_mr, ib.ib_mr);
+ if (ehca_adr_bad(mr_attr)) {
+ EDEB_ERR(4, "bad input values: mr_attr=%p", mr_attr);
+ ret = -EINVAL;
+ goto query_mr_exit0;
+ }
+ if ((e_mr->flags & EHCA_MR_FLAG_FMR)) {
+ EDEB_ERR(4, "not supported for FMR, mr=%p e_mr=%p "
+ "e_mr->flags=%x", mr, e_mr, e_mr->flags);
+ ret = -EINVAL;
+ goto query_mr_exit0;
+ }
+
+ shca = container_of(mr->device, struct ehca_shca, ib_device);
+ memset(mr_attr, 0, sizeof(struct ib_mr_attr));
+ spin_lock_irqsave(&e_mr->mrlock, sl_flags);
+
+ h_ret = hipz_h_query_mr(shca->ipz_hca_handle, e_mr, &hipzout);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_mr_query failed, h_ret=%lx mr=%p "
+ "hca_hndl=%lx mr_hndl=%lx lkey=%x",
+ h_ret, mr, shca->ipz_hca_handle.handle,
+ e_mr->ipz_mr_handle.handle, mr->lkey);
+ ret = ehca_mrmw_map_hrc_query_mr(h_ret);
+ goto query_mr_exit1;
+ }
+ mr_attr->pd = mr->pd;
+ mr_attr->device_virt_addr = hipzout.vaddr;
+ mr_attr->size = hipzout.len;
+ mr_attr->lkey = hipzout.lkey;
+ mr_attr->rkey = hipzout.rkey;
+ ehca_mrmw_reverse_map_acl(&hipzout.acl, &mr_attr->mr_access_flags);
+
+query_mr_exit1:
+ spin_unlock_irqrestore(&e_mr->mrlock, sl_flags);
+query_mr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x mr=%p mr_attr=%p", ret, mr, mr_attr);
+ else
+ EDEB_EX(7, "pd=%p device_virt_addr=%lx size=%lx "
+ "mr_access_flags=%x lkey=%x rkey=%x",
+ mr_attr->pd, mr_attr->device_virt_addr,
+ mr_attr->size, mr_attr->mr_access_flags,
+ mr_attr->lkey, mr_attr->rkey);
+ return ret;
+} /* end ehca_query_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_dereg_mr(struct ib_mr *mr)
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ struct ehca_shca *shca = NULL;
+ struct ehca_mr *e_mr = NULL;
+ struct ehca_pd *my_pd = NULL;
+ u32 cur_pid = current->tgid;
+
+ EDEB_EN(7, "mr=%p", mr);
+
+ EHCA_CHECK_MR(mr);
+ my_pd = container_of(mr->pd, struct ehca_pd, ib_pd);
+ if (my_pd->ib_pd.uobject && my_pd->ib_pd.uobject->context &&
+ (my_pd->ownpid != cur_pid)) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, my_pd->ownpid);
+ ret = -EINVAL;
+ goto dereg_mr_exit0;
+ }
+
+ e_mr = container_of(mr, struct ehca_mr, ib.ib_mr);
+ shca = container_of(mr->device, struct ehca_shca, ib_device);
+
+ if ((e_mr->flags & EHCA_MR_FLAG_FMR)) {
+ EDEB_ERR(4, "not supported for FMR, mr=%p e_mr=%p "
+ "e_mr->flags=%x", mr, e_mr, e_mr->flags);
+ ret = -EINVAL;
+ goto dereg_mr_exit0;
+ } else if (e_mr == shca->maxmr) {
+ /* should be impossible, however reject to be sure */
+ EDEB_ERR(3, "dereg internal max-MR impossible, mr=%p "
+ "shca->maxmr=%p mr->lkey=%x",
+ mr, shca->maxmr, mr->lkey);
+ ret = -EINVAL;
+ goto dereg_mr_exit0;
+ }
+
+ /* TODO: BUSY: MR still has bound window(s) */
+ h_ret = hipz_h_free_resource_mr(shca->ipz_hca_handle, e_mr);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_free_mr failed, h_ret=%lx shca=%p e_mr=%p"
+ " hca_hndl=%lx mr_hndl=%lx mr->lkey=%x",
+ h_ret, shca, e_mr, shca->ipz_hca_handle.handle,
+ e_mr->ipz_mr_handle.handle, mr->lkey);
+ ret = ehca_mrmw_map_hrc_free_mr(h_ret);
+ goto dereg_mr_exit0;
+ }
+
+ /* successful deregistration */
+ ehca_mr_delete(e_mr);
+
+dereg_mr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x mr=%p", ret, mr);
+ else
+ EDEB_EX(7, "");
+ return ret;
+} /* end ehca_dereg_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+struct ib_mw *ehca_alloc_mw(struct ib_pd *pd)
+{
+ struct ib_mw *ib_mw = NULL;
+ u64 h_ret = H_SUCCESS;
+ struct ehca_shca *shca = NULL;
+ struct ehca_mw *e_mw = NULL;
+ struct ehca_pd *e_pd = NULL;
+ struct ehca_mw_hipzout_parms hipzout = {{0},0};
+
+ EDEB_EN(7, "pd=%p", pd);
+
+ EHCA_CHECK_PD_P(pd);
+ e_pd = container_of(pd, struct ehca_pd, ib_pd);
+ shca = container_of(pd->device, struct ehca_shca, ib_device);
+
+ e_mw = ehca_mw_new();
+ if (!e_mw) {
+ ib_mw = ERR_PTR(-ENOMEM);
+ goto alloc_mw_exit0;
+ }
+
+ h_ret = hipz_h_alloc_resource_mw(shca->ipz_hca_handle, e_mw,
+ e_pd->fw_pd, &hipzout);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_mw_allocate failed, h_ret=%lx shca=%p "
+ "hca_hndl=%lx mw=%p", h_ret, shca,
+ shca->ipz_hca_handle.handle, e_mw);
+ ib_mw = ERR_PTR(ehca_mrmw_map_hrc_alloc(h_ret));
+ goto alloc_mw_exit1;
+ }
+ /* successful MW allocation */
+ e_mw->ipz_mw_handle = hipzout.handle;
+ e_mw->ib_mw.rkey = hipzout.rkey;
+ ib_mw = &e_mw->ib_mw;
+ goto alloc_mw_exit0;
+
+alloc_mw_exit1:
+ ehca_mw_delete(e_mw);
+alloc_mw_exit0:
+ if (IS_ERR(ib_mw))
+ EDEB_EX(4, "rc=%lx pd=%p", PTR_ERR(ib_mw), pd);
+ else
+ EDEB_EX(7, "ib_mw=%p rkey=%x", ib_mw, ib_mw->rkey);
+ return ib_mw;
+} /* end ehca_alloc_mw() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_bind_mw(struct ib_qp *qp,
+ struct ib_mw *mw,
+ struct ib_mw_bind *mw_bind)
+{
+ int ret = 0;
+
+ /* TODO: not supported up to now */
+ EDEB_ERR(4, "bind MW currently not supported by HCAD");
+ ret = -EPERM;
+ goto bind_mw_exit0;
+
+bind_mw_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x qp=%p mw=%p mw_bind=%p",
+ ret, qp, mw, mw_bind);
+ else
+ EDEB_EX(7, "qp=%p mw=%p mw_bind=%p", qp, mw, mw_bind);
+ return ret;
+} /* end ehca_bind_mw() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_dealloc_mw(struct ib_mw *mw)
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ struct ehca_shca *shca = NULL;
+ struct ehca_mw *e_mw = NULL;
+
+ EDEB_EN(7, "mw=%p", mw);
+
+ EHCA_CHECK_MW(mw);
+ e_mw = container_of(mw, struct ehca_mw, ib_mw);
+ shca = container_of(mw->device, struct ehca_shca, ib_device);
+
+ h_ret = hipz_h_free_resource_mw(shca->ipz_hca_handle, e_mw);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_free_mw failed, h_ret=%lx shca=%p mw=%p "
+ "rkey=%x hca_hndl=%lx mw_hndl=%lx",
+ h_ret, shca, mw, mw->rkey, shca->ipz_hca_handle.handle,
+ e_mw->ipz_mw_handle.handle);
+ ret = ehca_mrmw_map_hrc_free_mw(h_ret);
+ goto dealloc_mw_exit0;
+ }
+ /* successful deallocation */
+ ehca_mw_delete(e_mw);
+
+dealloc_mw_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x mw=%p", ret, mw);
+ else
+ EDEB_EX(7, "");
+ return ret;
+} /* end ehca_dealloc_mw() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+struct ib_fmr *ehca_alloc_fmr(struct ib_pd *pd,
+ int mr_access_flags,
+ struct ib_fmr_attr *fmr_attr)
+{
+ struct ib_fmr *ib_fmr = NULL;
+ struct ehca_shca *shca = NULL;
+ struct ehca_mr *e_fmr = NULL;
+ int ret = 0;
+ struct ehca_pd *e_pd = NULL;
+ u32 tmp_lkey = 0;
+ u32 tmp_rkey = 0;
+ struct ehca_mr_pginfo pginfo={0,0,0,0,0,0,0,NULL,0,NULL,NULL,0,NULL,0};
+
+ EDEB_EN(7, "pd=%p mr_access_flags=%x fmr_attr=%p",
+ pd, mr_access_flags, fmr_attr);
+
+ EHCA_CHECK_PD_P(pd);
+ if (ehca_adr_bad(fmr_attr)) {
+ EDEB_ERR(4, "bad input values: fmr_attr=%p", fmr_attr);
+ ib_fmr = ERR_PTR(-EINVAL);
+ goto alloc_fmr_exit0;
+ }
+
+ EDEB(7, "max_pages=%x max_maps=%x page_shift=%x",
+ fmr_attr->max_pages, fmr_attr->max_maps, fmr_attr->page_shift);
+
+ /* check other parameters */
+ if (((mr_access_flags & IB_ACCESS_REMOTE_WRITE) &&
+ !(mr_access_flags & IB_ACCESS_LOCAL_WRITE)) ||
+ ((mr_access_flags & IB_ACCESS_REMOTE_ATOMIC) &&
+ !(mr_access_flags & IB_ACCESS_LOCAL_WRITE))) {
+ /* Remote Write Access requires Local Write Access */
+ /* Remote Atomic Access requires Local Write Access */
+ EDEB_ERR(4, "bad input values: mr_access_flags=%x",
+ mr_access_flags);
+ ib_fmr = ERR_PTR(-EINVAL);
+ goto alloc_fmr_exit0;
+ }
+ if (mr_access_flags & IB_ACCESS_MW_BIND) {
+ EDEB_ERR(4, "bad input values: mr_access_flags=%x",
+ mr_access_flags);
+ ib_fmr = ERR_PTR(-EINVAL);
+ goto alloc_fmr_exit0;
+ }
+ if ((fmr_attr->max_pages == 0) || (fmr_attr->max_maps == 0)) {
+ EDEB_ERR(4, "bad input values: fmr_attr->max_pages=%x "
+ "fmr_attr->max_maps=%x fmr_attr->page_shift=%x",
+ fmr_attr->max_pages, fmr_attr->max_maps,
+ fmr_attr->page_shift);
+ ib_fmr = ERR_PTR(-EINVAL);
+ goto alloc_fmr_exit0;
+ }
+ if (((1 << fmr_attr->page_shift) != EHCA_PAGESIZE) &&
+ ((1 << fmr_attr->page_shift) != PAGE_SIZE)) {
+ EDEB_ERR(4, "unsupported fmr_attr->page_shift=%x",
+ fmr_attr->page_shift);
+ ib_fmr = ERR_PTR(-EINVAL);
+ goto alloc_fmr_exit0;
+ }
+
+ e_pd = container_of(pd, struct ehca_pd, ib_pd);
+ shca = container_of(pd->device, struct ehca_shca, ib_device);
+
+ e_fmr = ehca_mr_new();
+ if (!e_fmr) {
+ ib_fmr = ERR_PTR(-ENOMEM);
+ goto alloc_fmr_exit0;
+ }
+ e_fmr->flags |= EHCA_MR_FLAG_FMR;
+
+ /* register MR on HCA */
+ ret = ehca_reg_mr(shca, e_fmr, NULL,
+ fmr_attr->max_pages * (1 << fmr_attr->page_shift),
+ mr_access_flags, e_pd, &pginfo,
+ &tmp_lkey, &tmp_rkey);
+ if (ret) {
+ ib_fmr = ERR_PTR(ret);
+ goto alloc_fmr_exit1;
+ }
+
+ /* successful */
+ e_fmr->fmr_page_size = 1 << fmr_attr->page_shift;
+ e_fmr->fmr_max_pages = fmr_attr->max_pages;
+ e_fmr->fmr_max_maps = fmr_attr->max_maps;
+ e_fmr->fmr_map_cnt = 0;
+ ib_fmr = &e_fmr->ib.ib_fmr;
+ goto alloc_fmr_exit0;
+
+alloc_fmr_exit1:
+ ehca_mr_delete(e_fmr);
+alloc_fmr_exit0:
+ if (IS_ERR(ib_fmr))
+ EDEB_EX(4, "rc=%lx pd=%p mr_access_flags=%x "
+ "fmr_attr=%p", PTR_ERR(ib_fmr), pd,
+ mr_access_flags, fmr_attr);
+ else
+ EDEB_EX(7, "ib_fmr=%p tmp_lkey=%x tmp_rkey=%x",
+ ib_fmr, tmp_lkey, tmp_rkey);
+ return ib_fmr;
+} /* end ehca_alloc_fmr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_map_phys_fmr(struct ib_fmr *fmr,
+ u64 *page_list,
+ int list_len,
+ u64 iova)
+{
+ int ret = 0;
+ struct ehca_shca *shca = NULL;
+ struct ehca_mr *e_fmr = NULL;
+ struct ehca_pd *e_pd = NULL;
+ struct ehca_mr_pginfo pginfo={0,0,0,0,0,0,0,NULL,0,NULL,NULL,0,NULL,0};
+ u32 tmp_lkey = 0;
+ u32 tmp_rkey = 0;
+
+ EDEB_EN(7, "fmr=%p page_list=%p list_len=%x iova=%lx",
+ fmr, page_list, list_len, iova);
+
+ EHCA_CHECK_FMR(fmr);
+ e_fmr = container_of(fmr, struct ehca_mr, ib.ib_fmr);
+ shca = container_of(fmr->device, struct ehca_shca, ib_device);
+ e_pd = container_of(fmr->pd, struct ehca_pd, ib_pd);
+
+ if (!(e_fmr->flags & EHCA_MR_FLAG_FMR)) {
+ EDEB_ERR(4, "not a FMR, e_fmr=%p e_fmr->flags=%x",
+ e_fmr, e_fmr->flags);
+ ret = -EINVAL;
+ goto map_phys_fmr_exit0;
+ }
+ ret = ehca_fmr_check_page_list(e_fmr, page_list, list_len);
+ if (ret)
+ goto map_phys_fmr_exit0;
+ if (iova % e_fmr->fmr_page_size) {
+ /* only whole-numbered pages */
+ EDEB_ERR(4, "bad iova, iova=%lx fmr_page_size=%x",
+ iova, e_fmr->fmr_page_size);
+ ret = -EINVAL;
+ goto map_phys_fmr_exit0;
+ }
+ if (e_fmr->fmr_map_cnt >= e_fmr->fmr_max_maps) {
+ /* HCAD does not limit the maps, however trace this anyway */
+ EDEB(6, "map limit exceeded, fmr=%p e_fmr->fmr_map_cnt=%x "
+ "e_fmr->fmr_max_maps=%x",
+ fmr, e_fmr->fmr_map_cnt, e_fmr->fmr_max_maps);
+ }
+
+ pginfo.type = EHCA_MR_PGI_FMR;
+ pginfo.num_pages = list_len;
+ pginfo.page_list = page_list;
+ pginfo.next_4k = ((iova & (e_fmr->fmr_page_size-1)) /
+ EHCA_PAGESIZE);
+
+ ret = ehca_rereg_mr(shca, e_fmr, (u64*)iova,
+ list_len * e_fmr->fmr_page_size,
+ e_fmr->acl, e_pd, &pginfo, &tmp_lkey, &tmp_rkey);
+ if (ret)
+ goto map_phys_fmr_exit0;
+
+ /* successful reregistration */
+ e_fmr->fmr_map_cnt++;
+ e_fmr->ib.ib_fmr.lkey = tmp_lkey;
+ e_fmr->ib.ib_fmr.rkey = tmp_rkey;
+
+map_phys_fmr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x fmr=%p page_list=%p list_len=%x iova=%lx",
+ ret, fmr, page_list, list_len, iova);
+ else
+ EDEB_EX(7, "lkey=%x rkey=%x",
+ e_fmr->ib.ib_fmr.lkey, e_fmr->ib.ib_fmr.rkey);
+ return ret;
+} /* end ehca_map_phys_fmr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_unmap_fmr(struct list_head *fmr_list)
+{
+ int ret = 0;
+ struct ib_fmr *ib_fmr = NULL;
+ struct ehca_shca *shca = NULL;
+ struct ehca_shca *prev_shca = NULL;
+ struct ehca_mr *e_fmr = NULL;
+ u32 num_fmr = 0;
+ u32 unmap_fmr_cnt = 0;
+
+ EDEB_EN(7, "fmr_list=%p", fmr_list);
+
+ /* check all FMR belong to same SHCA, and check internal flag */
+ list_for_each_entry(ib_fmr, fmr_list, list) {
+ prev_shca = shca;
+ shca = container_of(ib_fmr->device, struct ehca_shca,
+ ib_device);
+ EHCA_CHECK_FMR(ib_fmr);
+ e_fmr = container_of(ib_fmr, struct ehca_mr, ib.ib_fmr);
+ if ((shca != prev_shca) && prev_shca) {
+ EDEB_ERR(4, "SHCA mismatch, shca=%p prev_shca=%p "
+ "e_fmr=%p", shca, prev_shca, e_fmr);
+ ret = -EINVAL;
+ goto unmap_fmr_exit0;
+ }
+ if (!(e_fmr->flags & EHCA_MR_FLAG_FMR)) {
+ EDEB_ERR(4, "not a FMR, e_fmr=%p e_fmr->flags=%x",
+ e_fmr, e_fmr->flags);
+ ret = -EINVAL;
+ goto unmap_fmr_exit0;
+ }
+ num_fmr++;
+ }
+
+ /* loop over all FMRs to unmap */
+ list_for_each_entry(ib_fmr, fmr_list, list) {
+ unmap_fmr_cnt++;
+ e_fmr = container_of(ib_fmr, struct ehca_mr, ib.ib_fmr);
+ shca = container_of(ib_fmr->device, struct ehca_shca,
+ ib_device);
+ ret = ehca_unmap_one_fmr(shca, e_fmr);
+ if (ret) {
+ /* unmap failed, stop unmapping of rest of FMRs */
+ EDEB_ERR(4, "unmap of one FMR failed, stop rest, "
+ "e_fmr=%p num_fmr=%x unmap_fmr_cnt=%x lkey=%x",
+ e_fmr, num_fmr, unmap_fmr_cnt,
+ e_fmr->ib.ib_fmr.lkey);
+ goto unmap_fmr_exit0;
+ }
+ }
+
+unmap_fmr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x fmr_list=%p num_fmr=%x unmap_fmr_cnt=%x",
+ ret, fmr_list, num_fmr, unmap_fmr_cnt);
+ else
+ EDEB_EX(7, "num_fmr=%x", num_fmr);
+ return ret;
+} /* end ehca_unmap_fmr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_dealloc_fmr(struct ib_fmr *fmr)
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ struct ehca_shca *shca = NULL;
+ struct ehca_mr *e_fmr = NULL;
+
+ EDEB_EN(7, "fmr=%p", fmr);
+
+ EHCA_CHECK_FMR(fmr);
+ e_fmr = container_of(fmr, struct ehca_mr, ib.ib_fmr);
+ shca = container_of(fmr->device, struct ehca_shca, ib_device);
+
+ if (!(e_fmr->flags & EHCA_MR_FLAG_FMR)) {
+ EDEB_ERR(4, "not a FMR, e_fmr=%p e_fmr->flags=%x",
+ e_fmr, e_fmr->flags);
+ ret = -EINVAL;
+ goto free_fmr_exit0;
+ }
+
+ h_ret = hipz_h_free_resource_mr(shca->ipz_hca_handle, e_fmr);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_free_mr failed, h_ret=%lx e_fmr=%p "
+ "hca_hndl=%lx fmr_hndl=%lx fmr->lkey=%x",
+ h_ret, e_fmr, shca->ipz_hca_handle.handle,
+ e_fmr->ipz_mr_handle.handle, fmr->lkey);
+ ehca_mrmw_map_hrc_free_mr(h_ret);
+ goto free_fmr_exit0;
+ }
+ /* successful deregistration */
+ ehca_mr_delete(e_fmr);
+
+free_fmr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x fmr=%p", ret, fmr);
+ else
+ EDEB_EX(7, "");
+ return ret;
+} /* end ehca_dealloc_fmr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_reg_mr(struct ehca_shca *shca,
+ struct ehca_mr *e_mr,
+ u64 *iova_start,
+ u64 size,
+ int acl,
+ struct ehca_pd *e_pd,
+ struct ehca_mr_pginfo *pginfo,
+ u32 *lkey, /*OUT*/
+ u32 *rkey) /*OUT*/
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ u32 hipz_acl = 0;
+ struct ehca_mr_hipzout_parms hipzout = {{0},0,0,0,0,0};
+
+ EDEB_EN(7, "shca=%p e_mr=%p iova_start=%p size=%lx acl=%x e_pd=%p "
+ "pginfo=%p num_pages=%lx num_4k=%lx", shca, e_mr, iova_start,
+ size, acl, e_pd, pginfo, pginfo->num_pages, pginfo->num_4k);
+
+ ehca_mrmw_map_acl(acl, &hipz_acl);
+ ehca_mrmw_set_pgsize_hipz_acl(&hipz_acl);
+ if (ehca_use_hp_mr == 1)
+ hipz_acl |= 0x00000001;
+
+ h_ret = hipz_h_alloc_resource_mr(shca->ipz_hca_handle, e_mr,
+ (u64)iova_start, size, hipz_acl,
+ e_pd->fw_pd, &hipzout);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_alloc_mr failed, h_ret=%lx hca_hndl=%lx",
+ h_ret, shca->ipz_hca_handle.handle);
+ ret = ehca_mrmw_map_hrc_alloc(h_ret);
+ goto ehca_reg_mr_exit0;
+ }
+
+ e_mr->ipz_mr_handle = hipzout.handle;
+
+ ret = ehca_reg_mr_rpages(shca, e_mr, pginfo);
+ if (ret)
+ goto ehca_reg_mr_exit1;
+
+ /* successful registration */
+ e_mr->num_pages = pginfo->num_pages;
+ e_mr->num_4k = pginfo->num_4k;
+ e_mr->start = iova_start;
+ e_mr->size = size;
+ e_mr->acl = acl;
+ *lkey = hipzout.lkey;
+ *rkey = hipzout.rkey;
+ goto ehca_reg_mr_exit0;
+
+ehca_reg_mr_exit1:
+ h_ret = hipz_h_free_resource_mr(shca->ipz_hca_handle, e_mr);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(1, "h_ret=%lx shca=%p e_mr=%p iova_start=%p "
+ "size=%lx acl=%x e_pd=%p lkey=%x pginfo=%p "
+ "num_pages=%lx num_4k=%lx ret=%x", h_ret, shca, e_mr,
+ iova_start, size, acl, e_pd, hipzout.lkey, pginfo,
+ pginfo->num_pages, pginfo->num_4k, ret);
+ EDEB_ERR(1, "internal error in ehca_reg_mr, not recoverable");
+ }
+ehca_reg_mr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x shca=%p e_mr=%p iova_start=%p size=%lx "
+ "acl=%x e_pd=%p pginfo=%p num_pages=%lx num_4k=%lx",
+ ret, shca, e_mr, iova_start, size, acl, e_pd, pginfo,
+ pginfo->num_pages, pginfo->num_4k);
+ else
+ EDEB_EX(7, "ret=%x lkey=%x rkey=%x", ret, *lkey, *rkey);
+ return ret;
+} /* end ehca_reg_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_reg_mr_rpages(struct ehca_shca *shca,
+ struct ehca_mr *e_mr,
+ struct ehca_mr_pginfo *pginfo)
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ u32 rnum = 0;
+ u64 rpage = 0;
+ u32 i;
+ u64 *kpage = NULL;
+
+ EDEB_EN(7, "shca=%p e_mr=%p pginfo=%p num_pages=%lx num_4k=%lx",
+ shca, e_mr, pginfo, pginfo->num_pages, pginfo->num_4k);
+
+ kpage = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!kpage) {
+ EDEB_ERR(4, "kpage alloc failed");
+ ret = -ENOMEM;
+ goto ehca_reg_mr_rpages_exit0;
+ }
+
+ /* max 512 pages per shot */
+ for (i = 0; i < ((pginfo->num_4k + 512 - 1) / 512); i++) {
+
+ if (i == ((pginfo->num_4k + 512 - 1) / 512) - 1) {
+ rnum = pginfo->num_4k % 512; /* last shot */
+ if (rnum == 0)
+ rnum = 512; /* last shot is full */
+ } else
+ rnum = 512;
+
+ if (rnum > 1) {
+ ret = ehca_set_pagebuf(e_mr, pginfo, rnum, kpage);
+ if (ret) {
+ EDEB_ERR(4, "ehca_set_pagebuf bad rc, ret=%x "
+ "rnum=%x kpage=%p", ret, rnum, kpage);
+ ret = -EFAULT;
+ goto ehca_reg_mr_rpages_exit1;
+ }
+ rpage = virt_to_abs(kpage);
+ if (!rpage) {
+ EDEB_ERR(4, "kpage=%p i=%x", kpage, i);
+ ret = -EFAULT;
+ goto ehca_reg_mr_rpages_exit1;
+ }
+ } else { /* rnum==1 */
+ ret = ehca_set_pagebuf_1(e_mr, pginfo, &rpage);
+ if (ret) {
+ EDEB_ERR(4, "ehca_set_pagebuf_1 bad rc, "
+ "ret=%x i=%x", ret, i);
+ ret = -EFAULT;
+ goto ehca_reg_mr_rpages_exit1;
+ }
+ }
+
+ EDEB(9, "i=%x rnum=%x rpage=%lx", i, rnum, rpage);
+
+ h_ret = hipz_h_register_rpage_mr(shca->ipz_hca_handle, e_mr,
+ 0, /* pagesize 4k */
+ 0, rpage, rnum);
+
+ if (i == ((pginfo->num_4k + 512 - 1) / 512) - 1) {
+ /* check for 'registration complete'==H_SUCCESS */
+ /* and for 'page registered'==H_PAGE_REGISTERED */
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "last hipz_reg_rpage_mr failed, "
+ "h_ret=%lx e_mr=%p i=%x hca_hndl=%lx "
+ "mr_hndl=%lx lkey=%x", h_ret, e_mr, i,
+ shca->ipz_hca_handle.handle,
+ e_mr->ipz_mr_handle.handle,
+ e_mr->ib.ib_mr.lkey);
+ ret = ehca_mrmw_map_hrc_rrpg_last(h_ret);
+ break;
+ } else
+ ret = 0;
+ } else if (h_ret != H_PAGE_REGISTERED) {
+ EDEB_ERR(4, "hipz_reg_rpage_mr failed, h_ret=%lx "
+ "e_mr=%p i=%x lkey=%x hca_hndl=%lx "
+ "mr_hndl=%lx", h_ret, e_mr, i,
+ e_mr->ib.ib_mr.lkey,
+ shca->ipz_hca_handle.handle,
+ e_mr->ipz_mr_handle.handle);
+ ret = ehca_mrmw_map_hrc_rrpg_notlast(h_ret);
+ break;
+ } else
+ ret = 0;
+ } /* end for(i) */
+
+
+ehca_reg_mr_rpages_exit1:
+ kfree(kpage);
+ehca_reg_mr_rpages_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x shca=%p e_mr=%p pginfo=%p num_pages=%lx "
+ "num_4k=%lx", ret, shca, e_mr, pginfo,
+ pginfo->num_pages, pginfo->num_4k);
+ else
+ EDEB_EX(7, "ret=%x", ret);
+ return ret;
+} /* end ehca_reg_mr_rpages() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+inline int ehca_rereg_mr_rereg1(struct ehca_shca *shca,
+ struct ehca_mr *e_mr,
+ u64 *iova_start,
+ u64 size,
+ u32 acl,
+ struct ehca_pd *e_pd,
+ struct ehca_mr_pginfo *pginfo,
+ u32 *lkey, /*OUT*/
+ u32 *rkey) /*OUT*/
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ u32 hipz_acl = 0;
+ u64 *kpage = NULL;
+ u64 rpage = 0;
+ struct ehca_mr_pginfo pginfo_save;
+ struct ehca_mr_hipzout_parms hipzout = {{0},0,0,0,0,0};
+
+ EDEB_EN(7, "shca=%p e_mr=%p iova_start=%p size=%lx acl=%x "
+ "e_pd=%p pginfo=%p num_pages=%lx num_4k=%lx", shca, e_mr,
+ iova_start, size, acl, e_pd, pginfo, pginfo->num_pages,
+ pginfo->num_4k);
+
+ ehca_mrmw_map_acl(acl, &hipz_acl);
+ ehca_mrmw_set_pgsize_hipz_acl(&hipz_acl);
+
+ kpage = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!kpage) {
+ EDEB_ERR(4, "kpage alloc failed");
+ ret = -ENOMEM;
+ goto ehca_rereg_mr_rereg1_exit0;
+ }
+
+ pginfo_save = *pginfo;
+ ret = ehca_set_pagebuf(e_mr, pginfo, pginfo->num_4k, kpage);
+ if (ret) {
+ EDEB_ERR(4, "set pagebuf failed, e_mr=%p pginfo=%p type=%x "
+ "num_pages=%lx num_4k=%lx kpage=%p", e_mr, pginfo,
+ pginfo->type, pginfo->num_pages, pginfo->num_4k,kpage);
+ goto ehca_rereg_mr_rereg1_exit1;
+ }
+ rpage = virt_to_abs(kpage);
+ if (!rpage) {
+ EDEB_ERR(4, "kpage=%p", kpage);
+ ret = -EFAULT;
+ goto ehca_rereg_mr_rereg1_exit1;
+ }
+ h_ret = hipz_h_reregister_pmr(shca->ipz_hca_handle, e_mr,
+ (u64)iova_start, size, hipz_acl,
+ e_pd->fw_pd, rpage, &hipzout);
+ if (h_ret != H_SUCCESS) {
+ /* reregistration unsuccessful, */
+ /* try it again with the 3 hCalls, */
+ /* e.g. this is required in case H_MR_CONDITION */
+ /* (MW bound or MR is shared) */
+ EDEB(6, "hipz_h_reregister_pmr failed (Rereg1), h_ret=%lx "
+ "e_mr=%p", h_ret, e_mr);
+ *pginfo = pginfo_save;
+ ret = -EAGAIN;
+ } else if ((u64*)hipzout.vaddr != iova_start) {
+ EDEB_ERR(4, "PHYP changed iova_start in rereg_pmr, "
+ "iova_start=%p iova_start_out=%lx e_mr=%p "
+ "mr_handle=%lx lkey=%x lkey_out=%x", iova_start,
+ hipzout.vaddr, e_mr, e_mr->ipz_mr_handle.handle,
+ e_mr->ib.ib_mr.lkey, hipzout.lkey);
+ ret = -EFAULT;
+ } else {
+ /* successful reregistration */
+ /* note: start and start_out are identical for eServer HCAs */
+ e_mr->num_pages = pginfo->num_pages;
+ e_mr->num_4k = pginfo->num_4k;
+ e_mr->start = iova_start;
+ e_mr->size = size;
+ e_mr->acl = acl;
+ *lkey = hipzout.lkey;
+ *rkey = hipzout.rkey;
+ }
+
+ehca_rereg_mr_rereg1_exit1:
+ kfree(kpage);
+ehca_rereg_mr_rereg1_exit0:
+ if ( ret && (ret != -EAGAIN) )
+ EDEB_EX(4, "ret=%x h_ret=%lx lkey=%x rkey=%x pginfo=%p "
+ "num_pages=%lx num_4k=%lx", ret, h_ret, *lkey, *rkey,
+ pginfo, pginfo->num_pages, pginfo->num_4k);
+ else
+ EDEB_EX(7, "ret=%x h_ret=%lx lkey=%x rkey=%x pginfo=%p "
+ "num_pages=%lx num_4k=%lx", ret, h_ret, *lkey, *rkey,
+ pginfo, pginfo->num_pages, pginfo->num_4k);
+ return ret;
+} /* end ehca_rereg_mr_rereg1() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_rereg_mr(struct ehca_shca *shca,
+ struct ehca_mr *e_mr,
+ u64 *iova_start,
+ u64 size,
+ int acl,
+ struct ehca_pd *e_pd,
+ struct ehca_mr_pginfo *pginfo,
+ u32 *lkey,
+ u32 *rkey)
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ int rereg_1_hcall = 1; /* 1: use hipz_h_reregister_pmr directly */
+ int rereg_3_hcall = 0; /* 1: use 3 hipz calls for reregistration */
+
+ EDEB_EN(7, "shca=%p e_mr=%p iova_start=%p size=%lx acl=%x "
+ "e_pd=%p pginfo=%p num_pages=%lx num_4k=%lx", shca, e_mr,
+ iova_start, size, acl, e_pd, pginfo, pginfo->num_pages,
+ pginfo->num_4k);
+
+ /* first determine reregistration hCall(s) */
+ if ((pginfo->num_4k > 512) || (e_mr->num_4k > 512) ||
+ (pginfo->num_4k > e_mr->num_4k)) {
+ EDEB(7, "Rereg3 case, pginfo->num_4k=%lx "
+ "e_mr->num_4k=%x", pginfo->num_4k, e_mr->num_4k);
+ rereg_1_hcall = 0;
+ rereg_3_hcall = 1;
+ }
+
+ if (e_mr->flags & EHCA_MR_FLAG_MAXMR) { /* check for max-MR */
+ rereg_1_hcall = 0;
+ rereg_3_hcall = 1;
+ e_mr->flags &= ~EHCA_MR_FLAG_MAXMR;
+ EDEB(4, "Rereg MR for max-MR! e_mr=%p", e_mr);
+ }
+
+ if (rereg_1_hcall) {
+ ret = ehca_rereg_mr_rereg1(shca, e_mr, iova_start, size,
+ acl, e_pd, pginfo, lkey, rkey);
+ if (ret) {
+ if (ret == -EAGAIN)
+ rereg_3_hcall = 1;
+ else
+ goto ehca_rereg_mr_exit0;
+ }
+ }
+
+ if (rereg_3_hcall) {
+ struct ehca_mr save_mr;
+
+ /* first deregister old MR */
+ h_ret = hipz_h_free_resource_mr(shca->ipz_hca_handle, e_mr);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_free_mr failed, h_ret=%lx e_mr=%p "
+ "hca_hndl=%lx mr_hndl=%lx mr->lkey=%x",
+ h_ret, e_mr, shca->ipz_hca_handle.handle,
+ e_mr->ipz_mr_handle.handle,
+ e_mr->ib.ib_mr.lkey);
+ ret = ehca_mrmw_map_hrc_free_mr(h_ret);
+ goto ehca_rereg_mr_exit0;
+ }
+ /* clean ehca_mr_t, without changing struct ib_mr and lock */
+ save_mr = *e_mr;
+ ehca_mr_deletenew(e_mr);
+
+ /* set some MR values */
+ e_mr->flags = save_mr.flags;
+ e_mr->fmr_page_size = save_mr.fmr_page_size;
+ e_mr->fmr_max_pages = save_mr.fmr_max_pages;
+ e_mr->fmr_max_maps = save_mr.fmr_max_maps;
+ e_mr->fmr_map_cnt = save_mr.fmr_map_cnt;
+
+ ret = ehca_reg_mr(shca, e_mr, iova_start, size, acl,
+ e_pd, pginfo, lkey, rkey);
+ if (ret) {
+ u32 offset = (u64)(&e_mr->flags) - (u64)e_mr;
+ memcpy(&e_mr->flags, &(save_mr.flags),
+ sizeof(struct ehca_mr) - offset);
+ goto ehca_rereg_mr_exit0;
+ }
+ }
+
+ehca_rereg_mr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x shca=%p e_mr=%p iova_start=%p size=%lx "
+ "acl=%x e_pd=%p pginfo=%p num_pages=%lx lkey=%x rkey=%x"
+ " rereg_1_hcall=%x rereg_3_hcall=%x", ret, shca, e_mr,
+ iova_start, size, acl, e_pd, pginfo, pginfo->num_pages,
+ *lkey, *rkey, rereg_1_hcall, rereg_3_hcall);
+ else
+ EDEB_EX(7, "ret=%x shca=%p e_mr=%p iova_start=%p size=%lx "
+ "acl=%x e_pd=%p pginfo=%p num_pages=%lx lkey=%x rkey=%x"
+ " rereg_1_hcall=%x rereg_3_hcall=%x", ret, shca, e_mr,
+ iova_start, size, acl, e_pd, pginfo, pginfo->num_pages,
+ *lkey, *rkey, rereg_1_hcall, rereg_3_hcall);
+
+ return ret;
+} /* end ehca_rereg_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_unmap_one_fmr(struct ehca_shca *shca,
+ struct ehca_mr *e_fmr)
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ int rereg_1_hcall = 1; /* 1: use hipz_mr_reregister directly */
+ int rereg_3_hcall = 0; /* 1: use 3 hipz calls for unmapping */
+ struct ehca_pd *e_pd = NULL;
+ struct ehca_mr save_fmr;
+ u32 tmp_lkey = 0;
+ u32 tmp_rkey = 0;
+ struct ehca_mr_pginfo pginfo={0,0,0,0,0,0,0,NULL,0,NULL,NULL,0,NULL,0};
+ struct ehca_mr_hipzout_parms hipzout = {{0},0,0,0,0,0};
+
+ EDEB_EN(7, "shca=%p e_fmr=%p", shca, e_fmr);
+
+ /* first check if reregistration hCall can be used for unmap */
+ if (e_fmr->fmr_max_pages > 512) {
+ rereg_1_hcall = 0;
+ rereg_3_hcall = 1;
+ }
+
+ e_pd = container_of(e_fmr->ib.ib_fmr.pd, struct ehca_pd, ib_pd);
+
+ if (rereg_1_hcall) {
+ /* note: after using rereg hcall with len=0, */
+ /* rereg hcall must be used again for registering pages */
+ h_ret = hipz_h_reregister_pmr(shca->ipz_hca_handle, e_fmr, 0,
+ 0, 0, e_pd->fw_pd, 0, &hipzout);
+ if (h_ret != H_SUCCESS) {
+ /* should not happen, because length checked above, */
+ /* FMRs are not shared and no MW bound to FMRs */
+ EDEB_ERR(4, "hipz_reregister_pmr failed (Rereg1), "
+ "h_ret=%lx e_fmr=%p hca_hndl=%lx mr_hndl=%lx "
+ "lkey=%x lkey_out=%x", h_ret, e_fmr,
+ shca->ipz_hca_handle.handle,
+ e_fmr->ipz_mr_handle.handle,
+ e_fmr->ib.ib_fmr.lkey, hipzout.lkey);
+ rereg_3_hcall = 1;
+ } else {
+ /* successful reregistration */
+ e_fmr->start = NULL;
+ e_fmr->size = 0;
+ tmp_lkey = hipzout.lkey;
+ tmp_rkey = hipzout.rkey;
+ }
+ }
+
+ if (rereg_3_hcall) {
+ struct ehca_mr save_mr;
+
+ /* first free old FMR */
+ h_ret = hipz_h_free_resource_mr(shca->ipz_hca_handle, e_fmr);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_free_mr failed, h_ret=%lx e_fmr=%p "
+ "hca_hndl=%lx mr_hndl=%lx lkey=%x", h_ret,
+ e_fmr, shca->ipz_hca_handle.handle,
+ e_fmr->ipz_mr_handle.handle,
+ e_fmr->ib.ib_fmr.lkey);
+ ret = ehca_mrmw_map_hrc_free_mr(h_ret);
+ goto ehca_unmap_one_fmr_exit0;
+ }
+ /* clean ehca_mr_t, without changing lock */
+ save_fmr = *e_fmr;
+ ehca_mr_deletenew(e_fmr);
+
+ /* set some MR values */
+ e_fmr->flags = save_fmr.flags;
+ e_fmr->fmr_page_size = save_fmr.fmr_page_size;
+ e_fmr->fmr_max_pages = save_fmr.fmr_max_pages;
+ e_fmr->fmr_max_maps = save_fmr.fmr_max_maps;
+ e_fmr->fmr_map_cnt = save_fmr.fmr_map_cnt;
+ e_fmr->acl = save_fmr.acl;
+
+ pginfo.type = EHCA_MR_PGI_FMR;
+ pginfo.num_pages = 0;
+ pginfo.num_4k = 0;
+ ret = ehca_reg_mr(shca, e_fmr, NULL,
+ (e_fmr->fmr_max_pages * e_fmr->fmr_page_size),
+ e_fmr->acl, e_pd, &pginfo, &tmp_lkey,
+ &tmp_rkey);
+ if (ret) {
+ u32 offset = (u64)(&e_fmr->flags) - (u64)e_fmr;
+ memcpy(&e_fmr->flags, &(save_mr.flags),
+ sizeof(struct ehca_mr) - offset);
+ goto ehca_unmap_one_fmr_exit0;
+ }
+ }
+
+ehca_unmap_one_fmr_exit0:
+ EDEB_EX(7, "ret=%x tmp_lkey=%x tmp_rkey=%x fmr_max_pages=%x "
+ "rereg_1_hcall=%x rereg_3_hcall=%x", ret, tmp_lkey, tmp_rkey,
+ e_fmr->fmr_max_pages, rereg_1_hcall, rereg_3_hcall);
+ return ret;
+} /* end ehca_unmap_one_fmr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_reg_smr(struct ehca_shca *shca,
+ struct ehca_mr *e_origmr,
+ struct ehca_mr *e_newmr,
+ u64 *iova_start,
+ int acl,
+ struct ehca_pd *e_pd,
+ u32 *lkey, /*OUT*/
+ u32 *rkey) /*OUT*/
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ u32 hipz_acl = 0;
+ struct ehca_mr_hipzout_parms hipzout = {{0},0,0,0,0,0};
+
+ EDEB_EN(7,"shca=%p e_origmr=%p e_newmr=%p iova_start=%p acl=%x e_pd=%p",
+ shca, e_origmr, e_newmr, iova_start, acl, e_pd);
+
+ ehca_mrmw_map_acl(acl, &hipz_acl);
+ ehca_mrmw_set_pgsize_hipz_acl(&hipz_acl);
+
+ h_ret = hipz_h_register_smr(shca->ipz_hca_handle, e_newmr, e_origmr,
+ (u64)iova_start, hipz_acl, e_pd->fw_pd,
+ &hipzout);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_reg_smr failed, h_ret=%lx shca=%p e_origmr=%p"
+ " e_newmr=%p iova_start=%p acl=%x e_pd=%p hca_hndl=%lx"
+ " mr_hndl=%lx lkey=%x", h_ret, shca, e_origmr, e_newmr,
+ iova_start, acl, e_pd, shca->ipz_hca_handle.handle,
+ e_origmr->ipz_mr_handle.handle,
+ e_origmr->ib.ib_mr.lkey);
+ ret = ehca_mrmw_map_hrc_reg_smr(h_ret);
+ goto ehca_reg_smr_exit0;
+ }
+ /* successful registration */
+ e_newmr->num_pages = e_origmr->num_pages;
+ e_newmr->num_4k = e_origmr->num_4k;
+ e_newmr->start = iova_start;
+ e_newmr->size = e_origmr->size;
+ e_newmr->acl = acl;
+ e_newmr->ipz_mr_handle = hipzout.handle;
+ *lkey = hipzout.lkey;
+ *rkey = hipzout.rkey;
+ goto ehca_reg_smr_exit0;
+
+ehca_reg_smr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x shca=%p e_origmr=%p e_newmr=%p "
+ "iova_start=%p acl=%x e_pd=%p",
+ ret, shca, e_origmr, e_newmr, iova_start, acl, e_pd);
+ else
+ EDEB_EX(7, "ret=%x lkey=%x rkey=%x", ret, *lkey, *rkey);
+ return ret;
+} /* end ehca_reg_smr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* register internal max-MR to internal SHCA */
+int ehca_reg_internal_maxmr(
+ struct ehca_shca *shca,
+ struct ehca_pd *e_pd,
+ struct ehca_mr **e_maxmr) /*OUT*/
+{
+ int ret = 0;
+ struct ehca_mr *e_mr = NULL;
+ u64 *iova_start = NULL;
+ u64 size_maxmr = 0;
+ struct ehca_mr_pginfo pginfo={0,0,0,0,0,0,0,NULL,0,NULL,NULL,0,NULL,0};
+ struct ib_phys_buf ib_pbuf;
+ u32 num_pages_mr = 0;
+ u32 num_pages_4k = 0; /* 4k portion "pages" */
+
+ EDEB_EN(7, "shca=%p e_pd=%p e_maxmr=%p", shca, e_pd, e_maxmr);
+
+ if (ehca_adr_bad(shca) || ehca_adr_bad(e_pd) || ehca_adr_bad(e_maxmr)) {
+ EDEB_ERR(4, "bad input values: shca=%p e_pd=%p e_maxmr=%p",
+ shca, e_pd, e_maxmr);
+ ret = -EINVAL;
+ goto ehca_reg_internal_maxmr_exit0;
+ }
+
+ e_mr = ehca_mr_new();
+ if (!e_mr) {
+ EDEB_ERR(4, "out of memory");
+ ret = -ENOMEM;
+ goto ehca_reg_internal_maxmr_exit0;
+ }
+ e_mr->flags |= EHCA_MR_FLAG_MAXMR;
+
+ /* register internal max-MR on HCA */
+ size_maxmr = (u64)high_memory - PAGE_OFFSET;
+ EDEB(7, "high_memory=%p PAGE_OFFSET=%lx", high_memory, PAGE_OFFSET);
+ iova_start = (u64*)KERNELBASE;
+ ib_pbuf.addr = 0;
+ ib_pbuf.size = size_maxmr;
+ num_pages_mr = ((((u64)iova_start % PAGE_SIZE) + size_maxmr +
+ PAGE_SIZE - 1) / PAGE_SIZE);
+ num_pages_4k = ((((u64)iova_start % EHCA_PAGESIZE) + size_maxmr +
+ EHCA_PAGESIZE - 1) / EHCA_PAGESIZE);
+
+ pginfo.type = EHCA_MR_PGI_PHYS;
+ pginfo.num_pages = num_pages_mr;
+ pginfo.num_4k = num_pages_4k;
+ pginfo.num_phys_buf = 1;
+ pginfo.phys_buf_array = &ib_pbuf;
+
+ ret = ehca_reg_mr(shca, e_mr, iova_start, size_maxmr, 0, e_pd,
+ &pginfo, &e_mr->ib.ib_mr.lkey,
+ &e_mr->ib.ib_mr.rkey);
+ if (ret) {
+ EDEB_ERR(4, "reg of internal max MR failed, e_mr=%p "
+ "iova_start=%p size_maxmr=%lx num_pages_mr=%x "
+ "num_pages_4k=%x", e_mr, iova_start, size_maxmr,
+ num_pages_mr, num_pages_4k);
+ goto ehca_reg_internal_maxmr_exit1;
+ }
+
+ /* successful registration of all pages */
+ e_mr->ib.ib_mr.device = e_pd->ib_pd.device;
+ e_mr->ib.ib_mr.pd = &e_pd->ib_pd;
+ e_mr->ib.ib_mr.uobject = NULL;
+ atomic_inc(&(e_pd->ib_pd.usecnt));
+ atomic_set(&(e_mr->ib.ib_mr.usecnt), 0);
+ *e_maxmr = e_mr;
+ goto ehca_reg_internal_maxmr_exit0;
+
+ehca_reg_internal_maxmr_exit1:
+ ehca_mr_delete(e_mr);
+ehca_reg_internal_maxmr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x shca=%p e_pd=%p e_maxmr=%p",
+ ret, shca, e_pd, e_maxmr);
+ else
+ EDEB_EX(7, "*e_maxmr=%p lkey=%x rkey=%x",
+ *e_maxmr, (*e_maxmr)->ib.ib_mr.lkey,
+ (*e_maxmr)->ib.ib_mr.rkey);
+ return ret;
+} /* end ehca_reg_internal_maxmr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_reg_maxmr(struct ehca_shca *shca,
+ struct ehca_mr *e_newmr,
+ u64 *iova_start,
+ int acl,
+ struct ehca_pd *e_pd,
+ u32 *lkey,
+ u32 *rkey)
+{
+ int ret = 0;
+ u64 h_ret = H_SUCCESS;
+ struct ehca_mr *e_origmr = shca->maxmr;
+ u32 hipz_acl = 0;
+ struct ehca_mr_hipzout_parms hipzout = {{0},0,0,0,0,0};
+
+ EDEB_EN(7,"shca=%p e_origmr=%p e_newmr=%p iova_start=%p acl=%x e_pd=%p",
+ shca, e_origmr, e_newmr, iova_start, acl, e_pd);
+
+ ehca_mrmw_map_acl(acl, &hipz_acl);
+ ehca_mrmw_set_pgsize_hipz_acl(&hipz_acl);
+
+ h_ret = hipz_h_register_smr(shca->ipz_hca_handle, e_newmr, e_origmr,
+ (u64)iova_start, hipz_acl, e_pd->fw_pd,
+ &hipzout);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "hipz_reg_smr failed, h_ret=%lx e_origmr=%p "
+ "hca_hndl=%lx mr_hndl=%lx lkey=%x",
+ h_ret, e_origmr, shca->ipz_hca_handle.handle,
+ e_origmr->ipz_mr_handle.handle,
+ e_origmr->ib.ib_mr.lkey);
+ ret = ehca_mrmw_map_hrc_reg_smr(h_ret);
+ goto ehca_reg_maxmr_exit0;
+ }
+ /* successful registration */
+ e_newmr->num_pages = e_origmr->num_pages;
+ e_newmr->num_4k = e_origmr->num_4k;
+ e_newmr->start = iova_start;
+ e_newmr->size = e_origmr->size;
+ e_newmr->acl = acl;
+ e_newmr->ipz_mr_handle = hipzout.handle;
+ *lkey = hipzout.lkey;
+ *rkey = hipzout.rkey;
+
+ehca_reg_maxmr_exit0:
+ EDEB_EX(7, "ret=%x lkey=%x rkey=%x", ret, *lkey, *rkey);
+ return ret;
+} /* end ehca_reg_maxmr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+int ehca_dereg_internal_maxmr(struct ehca_shca *shca)
+{
+ int ret = 0;
+ struct ehca_mr *e_maxmr = NULL;
+ struct ib_pd *ib_pd = NULL;
+
+ EDEB_EN(7, "shca=%p shca->maxmr=%p", shca, shca->maxmr);
+
+ if (!shca->maxmr) {
+ EDEB_ERR(4, "bad call, shca=%p", shca);
+ ret = -EINVAL;
+ goto ehca_dereg_internal_maxmr_exit0;
+ }
+
+ e_maxmr = shca->maxmr;
+ ib_pd = e_maxmr->ib.ib_mr.pd;
+ shca->maxmr = NULL; /* remove internal max-MR indication from SHCA */
+
+ ret = ehca_dereg_mr(&e_maxmr->ib.ib_mr);
+ if (ret) {
+ EDEB_ERR(3, "dereg internal max-MR failed, "
+ "ret=%x e_maxmr=%p shca=%p lkey=%x",
+ ret, e_maxmr, shca, e_maxmr->ib.ib_mr.lkey);
+ shca->maxmr = e_maxmr;
+ goto ehca_dereg_internal_maxmr_exit0;
+ }
+
+ atomic_dec(&ib_pd->usecnt);
+
+ehca_dereg_internal_maxmr_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x shca=%p shca->maxmr=%p",
+ ret, shca, shca->maxmr);
+ else
+ EDEB_EX(7, "");
+ return ret;
+} /* end ehca_dereg_internal_maxmr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* check physical buffer array of MR verbs for validness and
+ * calculates MR size
+ */
+int ehca_mr_chk_buf_and_calc_size(struct ib_phys_buf *phys_buf_array,
+ int num_phys_buf,
+ u64 *iova_start,
+ u64 *size)
+{
+ struct ib_phys_buf *pbuf = phys_buf_array;
+ u64 size_count = 0;
+ u32 i;
+
+ if (num_phys_buf == 0) {
+ EDEB_ERR(4, "bad phys buf array len, num_phys_buf=0");
+ return -EINVAL;
+ }
+ /* check first buffer */
+ if (((u64)iova_start & ~PAGE_MASK) != (pbuf->addr & ~PAGE_MASK)) {
+ EDEB_ERR(4, "iova_start/addr mismatch, iova_start=%p "
+ "pbuf->addr=%lx pbuf->size=%lx",
+ iova_start, pbuf->addr, pbuf->size);
+ return -EINVAL;
+ }
+ if (((pbuf->addr + pbuf->size) % PAGE_SIZE) &&
+ (num_phys_buf > 1)) {
+ EDEB_ERR(4, "addr/size mismatch in 1st buf, pbuf->addr=%lx "
+ "pbuf->size=%lx", pbuf->addr, pbuf->size);
+ return -EINVAL;
+ }
+
+ for (i = 0; i < num_phys_buf; i++) {
+ if ((i > 0) && (pbuf->addr % PAGE_SIZE)) {
+ EDEB_ERR(4, "bad address, i=%x pbuf->addr=%lx "
+ "pbuf->size=%lx", i, pbuf->addr, pbuf->size);
+ return -EINVAL;
+ }
+ if (((i > 0) && /* not 1st */
+ (i < (num_phys_buf - 1)) && /* not last */
+ (pbuf->size % PAGE_SIZE)) || (pbuf->size == 0)) {
+ EDEB_ERR(4, "bad size, i=%x pbuf->size=%lx",
+ i, pbuf->size);
+ return -EINVAL;
+ }
+ size_count += pbuf->size;
+ pbuf++;
+ }
+
+ *size = size_count;
+ return 0;
+} /* end ehca_mr_chk_buf_and_calc_size() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* check page list of map FMR verb for validness */
+int ehca_fmr_check_page_list(struct ehca_mr *e_fmr,
+ u64 *page_list,
+ int list_len)
+{
+ u32 i;
+ u64 *page = NULL;
+
+ if (ehca_adr_bad(page_list)) {
+ EDEB_ERR(4, "bad page_list, page_list=%p fmr=%p",
+ page_list, e_fmr);
+ return -EINVAL;
+ }
+
+ if ((list_len == 0) || (list_len > e_fmr->fmr_max_pages)) {
+ EDEB_ERR(4, "bad list_len, list_len=%x e_fmr->fmr_max_pages=%x "
+ "fmr=%p", list_len, e_fmr->fmr_max_pages, e_fmr);
+ return -EINVAL;
+ }
+
+ /* each page must be aligned */
+ page = page_list;
+ for (i = 0; i < list_len; i++) {
+ if (*page % e_fmr->fmr_page_size) {
+ EDEB_ERR(4, "bad page, i=%x *page=%lx page=%p "
+ "fmr=%p fmr_page_size=%x",
+ i, *page, page, e_fmr, e_fmr->fmr_page_size);
+ return -EINVAL;
+ }
+ page++;
+ }
+
+ return 0;
+} /* end ehca_fmr_check_page_list() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* setup page buffer from page info */
+int ehca_set_pagebuf(struct ehca_mr *e_mr,
+ struct ehca_mr_pginfo *pginfo,
+ u32 number,
+ u64 *kpage)
+{
+ int ret = 0;
+ struct ib_umem_chunk *prev_chunk = NULL;
+ struct ib_umem_chunk *chunk = NULL;
+ struct ib_phys_buf *pbuf = NULL;
+ u64 *fmrlist = NULL;
+ u64 num4k = 0;
+ u64 pgaddr = 0;
+ u64 offs4k = 0;
+ u32 i = 0;
+ u32 j = 0;
+
+ EDEB_EN(7, "pginfo=%p type=%x num_pages=%lx num_4k=%lx next_buf=%lx "
+ "next_4k=%lx number=%x kpage=%p page_cnt=%lx page_4k_cnt=%lx "
+ "next_listelem=%lx region=%p next_chunk=%p next_nmap=%lx",
+ pginfo, pginfo->type, pginfo->num_pages, pginfo->num_4k,
+ pginfo->next_buf, pginfo->next_4k, number, kpage,
+ pginfo->page_cnt, pginfo->page_4k_cnt, pginfo->next_listelem,
+ pginfo->region, pginfo->next_chunk, pginfo->next_nmap);
+
+ if (pginfo->type == EHCA_MR_PGI_PHYS) {
+ /* loop over desired phys_buf_array entries */
+ while (i < number) {
+ pbuf = pginfo->phys_buf_array + pginfo->next_buf;
+ num4k = ((pbuf->addr % EHCA_PAGESIZE) + pbuf->size +
+ EHCA_PAGESIZE - 1) / EHCA_PAGESIZE;
+ offs4k = (pbuf->addr & ~PAGE_MASK) / EHCA_PAGESIZE;
+ while (pginfo->next_4k < offs4k + num4k) {
+ /* sanity check */
+ if ((pginfo->page_cnt >= pginfo->num_pages) ||
+ (pginfo->page_4k_cnt >= pginfo->num_4k)) {
+ EDEB_ERR(4, "page_cnt >= num_pages, "
+ "page_cnt=%lx num_pages=%lx "
+ "page_4k_cnt=%lx num_4k=%lx "
+ "i=%x", pginfo->page_cnt,
+ pginfo->num_pages,
+ pginfo->page_4k_cnt,
+ pginfo->num_4k, i);
+ ret = -EFAULT;
+ }
+ *kpage = phys_to_abs(
+ (pbuf->addr & (EHCA_PAGESIZE-1))
+ + (pginfo->next_4k * EHCA_PAGESIZE));
+ if ( !(*kpage) && pbuf->addr ) {
+ EDEB_ERR(4, "pbuf->addr=%lx "
+ "pbuf->size=%lx next_4k=%lx",
+ pbuf->addr, pbuf->size,
+ pginfo->next_4k);
+ ret = -EFAULT;
+ goto ehca_set_pagebuf_exit0;
+ }
+ (pginfo->page_4k_cnt)++;
+ (pginfo->next_4k)++;
+ if(pginfo->next_4k >= PAGE_SIZE/EHCA_PAGESIZE)
+ (pginfo->page_cnt)++;
+ kpage++;
+ i++;
+ if (i >= number) break;
+ }
+ if (pginfo->next_4k >= offs4k + num4k) {
+ (pginfo->next_buf)++;
+ pginfo->next_4k = 0;
+ }
+ }
+ } else if (pginfo->type == EHCA_MR_PGI_USER) {
+ /* loop over desired chunk entries */
+ chunk = pginfo->next_chunk;
+ prev_chunk = pginfo->next_chunk;
+ list_for_each_entry_continue(chunk,
+ (&(pginfo->region->chunk_list)),
+ list) {
+ EDEB(9, "chunk->page_list[0]=%lx",
+ (u64)sg_dma_address(&chunk->page_list[0]));
+ for (i = pginfo->next_nmap; i < chunk->nmap; ) {
+ pgaddr = ( page_to_pfn(chunk->page_list[i].page)
+ << PAGE_SHIFT );
+ *kpage = phys_to_abs(pgaddr +
+ (pginfo->next_4k *
+ EHCA_PAGESIZE));
+ EDEB(9,"pgaddr=%lx *kpage=%lx next_4k=%lx",
+ pgaddr, *kpage, pginfo->next_4k);
+ if ( !(*kpage) ) {
+ EDEB_ERR(4, "pgaddr=%lx "
+ "chunk->page_list[i]=%lx i=%x "
+ "next_4k=%lx mr=%p", pgaddr,
+ (u64)sg_dma_address(
+ &chunk->page_list[i]),
+ i, pginfo->next_4k, e_mr);
+ ret = -EFAULT;
+ goto ehca_set_pagebuf_exit0;
+ }
+ (pginfo->page_4k_cnt)++;
+ (pginfo->next_4k)++;
+ kpage++;
+ if (pginfo->next_4k >= PAGE_SIZE/EHCA_PAGESIZE) {
+ (pginfo->page_cnt)++;
+ (pginfo->next_nmap)++;
+ pginfo->next_4k = 0;
+ i++;
+ }
+ j++;
+ if (j >= number) break;
+ }
+ if ((pginfo->next_nmap >= chunk->nmap) &&
+ (j >= number)) {
+ pginfo->next_nmap = 0;
+ prev_chunk = chunk;
+ break;
+ } else if (pginfo->next_nmap >= chunk->nmap) {
+ pginfo->next_nmap = 0;
+ prev_chunk = chunk;
+ } else if (j >= number)
+ break;
+ else
+ prev_chunk = chunk;
+ }
+ pginfo->next_chunk =
+ list_prepare_entry(prev_chunk,
+ (&(pginfo->region->chunk_list)),
+ list);
+ } else if (pginfo->type == EHCA_MR_PGI_FMR) {
+ /* loop over desired page_list entries */
+ fmrlist = pginfo->page_list + pginfo->next_listelem;
+ for (i = 0; i < number; i++) {
+ *kpage = phys_to_abs((*fmrlist & (EHCA_PAGESIZE-1)) +
+ pginfo->next_4k * EHCA_PAGESIZE);
+ if ( !(*kpage) ) {
+ EDEB_ERR(4, "*fmrlist=%lx fmrlist=%p "
+ "next_listelem=%lx next_4k=%lx",
+ *fmrlist, fmrlist,
+ pginfo->next_listelem,pginfo->next_4k);
+ ret = -EFAULT;
+ goto ehca_set_pagebuf_exit0;
+ }
+ (pginfo->page_4k_cnt)++;
+ (pginfo->next_4k)++;
+ kpage++;
+ if ( pginfo->next_4k >=
+ ((e_mr->fmr_page_size) / EHCA_PAGESIZE) ) {
+ (pginfo->page_cnt)++;
+ (pginfo->next_listelem)++;
+ fmrlist++;
+ pginfo->next_4k = 0;
+ }
+ }
+ } else {
+ EDEB_ERR(4, "bad pginfo->type=%x", pginfo->type);
+ ret = -EFAULT;
+ goto ehca_set_pagebuf_exit0;
+ }
+
+ehca_set_pagebuf_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x e_mr=%p pginfo=%p type=%x num_pages=%lx "
+ "num_4k=%lx next_buf=%lx next_4k=%lx number=%x "
+ "kpage=%p page_cnt=%lx page_4k_cnt=%lx i=%x "
+ "next_listelem=%lx region=%p next_chunk=%p "
+ "next_nmap=%lx", ret, e_mr, pginfo, pginfo->type,
+ pginfo->num_pages, pginfo->num_4k, pginfo->next_buf,
+ pginfo->next_4k, number, kpage, pginfo->page_cnt,
+ pginfo->page_4k_cnt, i, pginfo->next_listelem,
+ pginfo->region, pginfo->next_chunk, pginfo->next_nmap);
+ else
+ EDEB_EX(7, "ret=%x e_mr=%p pginfo=%p type=%x num_pages=%lx "
+ "num_4k=%lx next_buf=%lx next_4k=%lx number=%x "
+ "kpage=%p page_cnt=%lx page_4k_cnt=%lx i=%x "
+ "next_listelem=%lx region=%p next_chunk=%p "
+ "next_nmap=%lx", ret, e_mr, pginfo, pginfo->type,
+ pginfo->num_pages, pginfo->num_4k, pginfo->next_buf,
+ pginfo->next_4k, number, kpage, pginfo->page_cnt,
+ pginfo->page_4k_cnt, i, pginfo->next_listelem,
+ pginfo->region, pginfo->next_chunk, pginfo->next_nmap);
+ return ret;
+} /* end ehca_set_pagebuf() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* setup 1 page from page info page buffer */
+int ehca_set_pagebuf_1(struct ehca_mr *e_mr,
+ struct ehca_mr_pginfo *pginfo,
+ u64 *rpage)
+{
+ int ret = 0;
+ struct ib_phys_buf *tmp_pbuf = NULL;
+ u64 *fmrlist = NULL;
+ struct ib_umem_chunk *chunk = NULL;
+ struct ib_umem_chunk *prev_chunk = NULL;
+ u64 pgaddr = 0;
+ u64 num4k = 0;
+ u64 offs4k = 0;
+
+ EDEB_EN(7, "pginfo=%p type=%x num_pages=%lx num_4k=%lx next_buf=%lx "
+ "next_4k=%lx rpage=%p page_cnt=%lx page_4k_cnt=%lx "
+ "next_listelem=%lx region=%p next_chunk=%p next_nmap=%lx",
+ pginfo, pginfo->type, pginfo->num_pages, pginfo->num_4k,
+ pginfo->next_buf, pginfo->next_4k, rpage, pginfo->page_cnt,
+ pginfo->page_4k_cnt, pginfo->next_listelem, pginfo->region,
+ pginfo->next_chunk, pginfo->next_nmap);
+
+ if (pginfo->type == EHCA_MR_PGI_PHYS) {
+ /* sanity check */
+ if ((pginfo->page_cnt >= pginfo->num_pages) ||
+ (pginfo->page_4k_cnt >= pginfo->num_4k)) {
+ EDEB_ERR(4, "page_cnt >= num_pages, page_cnt=%lx "
+ "num_pages=%lx page_4k_cnt=%lx num_4k=%lx",
+ pginfo->page_cnt, pginfo->num_pages,
+ pginfo->page_4k_cnt, pginfo->num_4k);
+ ret = -EFAULT;
+ goto ehca_set_pagebuf_1_exit0;
+ }
+ tmp_pbuf = pginfo->phys_buf_array + pginfo->next_buf;
+ num4k = ((tmp_pbuf->addr % EHCA_PAGESIZE) + tmp_pbuf->size +
+ EHCA_PAGESIZE - 1) / EHCA_PAGESIZE;
+ offs4k = (tmp_pbuf->addr & ~PAGE_MASK) / EHCA_PAGESIZE;
+ *rpage = phys_to_abs((tmp_pbuf->addr & (EHCA_PAGESIZE-1)) +
+ (pginfo->next_4k * EHCA_PAGESIZE));
+ if ( !(*rpage) && tmp_pbuf->addr ) {
+ EDEB_ERR(4, "tmp_pbuf->addr=%lx"
+ " tmp_pbuf->size=%lx next_4k=%lx",
+ tmp_pbuf->addr, tmp_pbuf->size,
+ pginfo->next_4k);
+ ret = -EFAULT;
+ goto ehca_set_pagebuf_1_exit0;
+ }
+ (pginfo->page_4k_cnt)++;
+ (pginfo->next_4k)++;
+ if(pginfo->next_4k >= PAGE_SIZE/EHCA_PAGESIZE)
+ (pginfo->page_cnt)++;
+ if (pginfo->next_4k >= offs4k + num4k) {
+ (pginfo->next_buf)++;
+ pginfo->next_4k = 0;
+ }
+ } else if (pginfo->type == EHCA_MR_PGI_USER) {
+ chunk = pginfo->next_chunk;
+ prev_chunk = pginfo->next_chunk;
+ list_for_each_entry_continue(chunk,
+ (&(pginfo->region->chunk_list)),
+ list) {
+ pgaddr = ( page_to_pfn(chunk->page_list[
+ pginfo->next_nmap].page)
+ << PAGE_SHIFT);
+ *rpage = phys_to_abs(pgaddr +
+ (pginfo->next_4k * EHCA_PAGESIZE));
+ EDEB(9,"pgaddr=%lx *rpage=%lx next_4k=%lx", pgaddr,
+ *rpage, pginfo->next_4k);
+ if ( !(*rpage) ) {
+ EDEB_ERR(4, "pgaddr=%lx chunk->page_list[]=%lx "
+ "next_nmap=%lx next_4k=%lx mr=%p",
+ pgaddr, (u64)sg_dma_address(
+ &chunk->page_list[
+ pginfo->next_nmap]),
+ pginfo->next_nmap, pginfo->next_4k,
+ e_mr);
+ ret = -EFAULT;
+ goto ehca_set_pagebuf_1_exit0;
+ }
+ (pginfo->page_4k_cnt)++;
+ (pginfo->next_4k)++;
+ if (pginfo->next_4k >= PAGE_SIZE/EHCA_PAGESIZE) {
+ (pginfo->page_cnt)++;
+ (pginfo->next_nmap)++;
+ pginfo->next_4k = 0;
+ }
+ if (pginfo->next_nmap >= chunk->nmap) {
+ pginfo->next_nmap = 0;
+ prev_chunk = chunk;
+ }
+ break;
+ }
+ pginfo->next_chunk =
+ list_prepare_entry(prev_chunk,
+ (&(pginfo->region->chunk_list)),
+ list);
+ } else if (pginfo->type == EHCA_MR_PGI_FMR) {
+ fmrlist = pginfo->page_list + pginfo->next_listelem;
+ *rpage = phys_to_abs((*fmrlist & (EHCA_PAGESIZE-1)) +
+ pginfo->next_4k * EHCA_PAGESIZE);
+ if ( !(*rpage) ) {
+ EDEB_ERR(4, "*fmrlist=%lx fmrlist=%p next_listelem=%lx "
+ "next_4k=%lx", *fmrlist, fmrlist,
+ pginfo->next_listelem, pginfo->next_4k);
+ ret = -EFAULT;
+ goto ehca_set_pagebuf_1_exit0;
+ }
+ (pginfo->page_4k_cnt)++;
+ (pginfo->next_4k)++;
+ if (pginfo->next_4k >= (e_mr->fmr_page_size)/EHCA_PAGESIZE) {
+ (pginfo->page_cnt)++;
+ (pginfo->next_listelem)++;
+ pginfo->next_4k = 0;
+ }
+ } else {
+ EDEB_ERR(4, "bad pginfo->type=%x", pginfo->type);
+ ret = -EFAULT;
+ goto ehca_set_pagebuf_1_exit0;
+ }
+
+ehca_set_pagebuf_1_exit0:
+ if (ret)
+ EDEB_EX(4, "ret=%x e_mr=%p pginfo=%p type=%x num_pages=%lx "
+ "num_4k=%lx next_buf=%lx next_4k=%lx rpage=%p "
+ "page_cnt=%lx page_4k_cnt=%lx next_listelem=%lx "
+ "region=%p next_chunk=%p next_nmap=%lx", ret, e_mr,
+ pginfo, pginfo->type, pginfo->num_pages, pginfo->num_4k,
+ pginfo->next_buf, pginfo->next_4k, rpage,
+ pginfo->page_cnt, pginfo->page_4k_cnt,
+ pginfo->next_listelem, pginfo->region,
+ pginfo->next_chunk, pginfo->next_nmap);
+ else
+ EDEB_EX(7, "ret=%x e_mr=%p pginfo=%p type=%x num_pages=%lx "
+ "num_4k=%lx next_buf=%lx next_4k=%lx rpage=%p "
+ "page_cnt=%lx page_4k_cnt=%lx next_listelem=%lx "
+ "region=%p next_chunk=%p next_nmap=%lx", ret, e_mr,
+ pginfo, pginfo->type, pginfo->num_pages, pginfo->num_4k,
+ pginfo->next_buf, pginfo->next_4k, rpage,
+ pginfo->page_cnt, pginfo->page_4k_cnt,
+ pginfo->next_listelem, pginfo->region,
+ pginfo->next_chunk, pginfo->next_nmap);
+ return ret;
+} /* end ehca_set_pagebuf_1() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* check MR if it is a max-MR, i.e. uses whole memory
+ * in case it's a max-MR 1 is returned, else 0
+ */
+int ehca_mr_is_maxmr(u64 size,
+ u64 *iova_start)
+{
+ /* a MR is treated as max-MR only if it fits following: */
+ if ((size == ((u64)high_memory - PAGE_OFFSET)) &&
+ (iova_start == (void*)KERNELBASE)) {
+ EDEB(6, "this is a max-MR");
+ return 1;
+ } else
+ return 0;
+} /* end ehca_mr_is_maxmr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+/* map access control for MR/MW. This routine is used for MR and MW. */
+void ehca_mrmw_map_acl(int ib_acl,
+ u32 *hipz_acl)
+{
+ *hipz_acl = 0;
+ if (ib_acl & IB_ACCESS_REMOTE_READ)
+ *hipz_acl |= HIPZ_ACCESSCTRL_R_READ;
+ if (ib_acl & IB_ACCESS_REMOTE_WRITE)
+ *hipz_acl |= HIPZ_ACCESSCTRL_R_WRITE;
+ if (ib_acl & IB_ACCESS_REMOTE_ATOMIC)
+ *hipz_acl |= HIPZ_ACCESSCTRL_R_ATOMIC;
+ if (ib_acl & IB_ACCESS_LOCAL_WRITE)
+ *hipz_acl |= HIPZ_ACCESSCTRL_L_WRITE;
+ if (ib_acl & IB_ACCESS_MW_BIND)
+ *hipz_acl |= HIPZ_ACCESSCTRL_MW_BIND;
+} /* end ehca_mrmw_map_acl() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* sets page size in hipz access control for MR/MW. */
+void ehca_mrmw_set_pgsize_hipz_acl(u32 *hipz_acl) /*INOUT*/
+{
+ return; /* HCA supports only 4k */
+} /* end ehca_mrmw_set_pgsize_hipz_acl() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* reverse map access control for MR/MW.
+ * This routine is used for MR and MW.
+ */
+void ehca_mrmw_reverse_map_acl(const u32 *hipz_acl,
+ int *ib_acl) /*OUT*/
+{
+ *ib_acl = 0;
+ if (*hipz_acl & HIPZ_ACCESSCTRL_R_READ)
+ *ib_acl |= IB_ACCESS_REMOTE_READ;
+ if (*hipz_acl & HIPZ_ACCESSCTRL_R_WRITE)
+ *ib_acl |= IB_ACCESS_REMOTE_WRITE;
+ if (*hipz_acl & HIPZ_ACCESSCTRL_R_ATOMIC)
+ *ib_acl |= IB_ACCESS_REMOTE_ATOMIC;
+ if (*hipz_acl & HIPZ_ACCESSCTRL_L_WRITE)
+ *ib_acl |= IB_ACCESS_LOCAL_WRITE;
+ if (*hipz_acl & HIPZ_ACCESSCTRL_MW_BIND)
+ *ib_acl |= IB_ACCESS_MW_BIND;
+} /* end ehca_mrmw_reverse_map_acl() */
+
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* map HIPZ rc to IB retcodes for MR/MW allocations
+ * Used for hipz_mr_reg_alloc and hipz_mw_alloc.
+ */
+int ehca_mrmw_map_hrc_alloc(const u64 hipz_rc)
+{
+ switch (hipz_rc) {
+ case H_SUCCESS: /* successful completion */
+ return 0;
+ case H_ADAPTER_PARM: /* invalid adapter handle */
+ case H_RT_PARM: /* invalid resource type */
+ case H_NOT_ENOUGH_RESOURCES: /* insufficient resources */
+ case H_MLENGTH_PARM: /* invalid memory length */
+ case H_MEM_ACCESS_PARM: /* invalid access controls */
+ case H_CONSTRAINED: /* resource constraint */
+ return -EINVAL;
+ case H_BUSY: /* long busy */
+ return -EBUSY;
+ default:
+ return -EINVAL;
+ }
+} /* end ehca_mrmw_map_hrc_alloc() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* map HIPZ rc to IB retcodes for MR register rpage
+ * Used for hipz_h_register_rpage_mr at registering last page
+ */
+int ehca_mrmw_map_hrc_rrpg_last(const u64 hipz_rc)
+{
+ switch (hipz_rc) {
+ case H_SUCCESS: /* registration complete */
+ return 0;
+ case H_PAGE_REGISTERED: /* page registered */
+ case H_ADAPTER_PARM: /* invalid adapter handle */
+ case H_RH_PARM: /* invalid resource handle */
+/* case H_QT_PARM: invalid queue type */
+ case H_PARAMETER: /* invalid logical address, */
+ /* or count zero or greater 512 */
+ case H_TABLE_FULL: /* page table full */
+ case H_HARDWARE: /* HCA not operational */
+ return -EINVAL;
+ case H_BUSY: /* long busy */
+ return -EBUSY;
+ default:
+ return -EINVAL;
+ }
+} /* end ehca_mrmw_map_hrc_rrpg_last() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* map HIPZ rc to IB retcodes for MR register rpage
+ * Used for hipz_h_register_rpage_mr at registering one page, but not last page
+ */
+int ehca_mrmw_map_hrc_rrpg_notlast(const u64 hipz_rc)
+{
+ switch (hipz_rc) {
+ case H_PAGE_REGISTERED: /* page registered */
+ return 0;
+ case H_SUCCESS: /* registration complete */
+ case H_ADAPTER_PARM: /* invalid adapter handle */
+ case H_RH_PARM: /* invalid resource handle */
+/* case H_QT_PARM: invalid queue type */
+ case H_PARAMETER: /* invalid logical address, */
+ /* or count zero or greater 512 */
+ case H_TABLE_FULL: /* page table full */
+ case H_HARDWARE: /* HCA not operational */
+ return -EINVAL;
+ case H_BUSY: /* long busy */
+ return -EBUSY;
+ default:
+ return -EINVAL;
+ }
+} /* end ehca_mrmw_map_hrc_rrpg_notlast() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* map HIPZ rc to IB retcodes for MR query. Used for hipz_mr_query. */
+int ehca_mrmw_map_hrc_query_mr(const u64 hipz_rc)
+{
+ switch (hipz_rc) {
+ case H_SUCCESS: /* successful completion */
+ return 0;
+ case H_ADAPTER_PARM: /* invalid adapter handle */
+ case H_RH_PARM: /* invalid resource handle */
+ return -EINVAL;
+ case H_BUSY: /* long busy */
+ return -EBUSY;
+ default:
+ return -EINVAL;
+ }
+} /* end ehca_mrmw_map_hrc_query_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* map HIPZ rc to IB retcodes for freeing MR resource
+ * Used for hipz_h_free_resource_mr
+ */
+int ehca_mrmw_map_hrc_free_mr(const u64 hipz_rc)
+{
+ switch (hipz_rc) {
+ case H_SUCCESS: /* resource freed */
+ return 0;
+ case H_ADAPTER_PARM: /* invalid adapter handle */
+ case H_RH_PARM: /* invalid resource handle */
+ case H_R_STATE: /* invalid resource state */
+ case H_HARDWARE: /* HCA not operational */
+ return -EINVAL;
+ case H_RESOURCE: /* Resource in use */
+ case H_BUSY: /* long busy */
+ return -EBUSY;
+ default:
+ return -EINVAL;
+ }
+} /* end ehca_mrmw_map_hrc_free_mr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* map HIPZ rc to IB retcodes for freeing MW resource
+ * Used for hipz_h_free_resource_mw
+ */
+int ehca_mrmw_map_hrc_free_mw(const u64 hipz_rc)
+{
+ switch (hipz_rc) {
+ case H_SUCCESS: /* resource freed */
+ return 0;
+ case H_ADAPTER_PARM: /* invalid adapter handle */
+ case H_RH_PARM: /* invalid resource handle */
+ case H_R_STATE: /* invalid resource state */
+ case H_HARDWARE: /* HCA not operational */
+ return -EINVAL;
+ case H_RESOURCE: /* Resource in use */
+ case H_BUSY: /* long busy */
+ return -EBUSY;
+ default:
+ return -EINVAL;
+ }
+} /* end ehca_mrmw_map_hrc_free_mw() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* map HIPZ rc to IB retcodes for SMR registrations
+ * Used for hipz_h_register_smr.
+ */
+int ehca_mrmw_map_hrc_reg_smr(const u64 hipz_rc)
+{
+ switch (hipz_rc) {
+ case H_SUCCESS: /* successful completion */
+ return 0;
+ case H_ADAPTER_PARM: /* invalid adapter handle */
+ case H_RH_PARM: /* invalid resource handle */
+ case H_MEM_PARM: /* invalid MR virtual address */
+ case H_MEM_ACCESS_PARM: /* invalid access controls */
+ case H_NOT_ENOUGH_RESOURCES: /* insufficient resources */
+ return -EINVAL;
+ case H_BUSY: /* long busy */
+ return -EBUSY;
+ default:
+ return -EINVAL;
+ }
+} /* end ehca_mrmw_map_hrc_reg_smr() */
+
+/*----------------------------------------------------------------------*/
+/*----------------------------------------------------------------------*/
+
+/* MR destructor and constructor
+ * used in Reregister MR verb, sets all fields in ehca_mr_t to 0,
+ * except struct ib_mr and spinlock
+ */
+void ehca_mr_deletenew(struct ehca_mr *mr)
+{
+ mr->flags = 0;
+ mr->num_pages = 0;
+ mr->num_4k = 0;
+ mr->acl = 0;
+ mr->start = NULL;
+ mr->fmr_page_size = 0;
+ mr->fmr_max_pages = 0;
+ mr->fmr_max_maps = 0;
+ mr->fmr_map_cnt = 0;
+ memset(&mr->ipz_mr_handle, 0, sizeof(mr->ipz_mr_handle));
+ memset(&mr->galpas, 0, sizeof(mr->galpas));
+ mr->nr_of_pages = 0;
+ mr->pagearray = NULL;
+ memset(&mr->pf, 0, sizeof(mr->pf));
+} /* end ehca_mr_deletenew() */
^ permalink raw reply
* [PATCH 06/16] ehca: interrupt handling routines
From: Heiko J Schick @ 2006-05-15 17:41 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
drivers/infiniband/hw/ehca/ehca_irq.c | 710 ++++++++++++++++++++++++++++++++++
drivers/infiniband/hw/ehca/ehca_irq.h | 77 +++
2 files changed, 787 insertions(+)
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_irq.h 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_irq.h 2006-05-10 08:07:51.000000000 +0200
@@ -0,0 +1,77 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * Function definitions and structs for EQs, NEQs and interrupts
+ *
+ * Authors: Heiko J Schick <schickhj@de.ibm.com>
+ * Khadija Souissi <souissi@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef __EHCA_IRQ_H
+#define __EHCA_IRQ_H
+
+
+struct ehca_shca;
+
+#include <linux/interrupt.h>
+#include <linux/types.h>
+#include <asm/atomic.h>
+
+int ehca_error_data(struct ehca_shca *shca, void *data, u64 resource);
+
+irqreturn_t ehca_interrupt_neq(int irq, void *dev_id, struct pt_regs *regs);
+void ehca_tasklet_neq(unsigned long data);
+
+irqreturn_t ehca_interrupt_eq(int irq, void *dev_id, struct pt_regs *regs);
+void ehca_tasklet_eq(unsigned long data);
+
+struct ehca_cpu_comp_task {
+ wait_queue_head_t wait_queue;
+ struct list_head cq_list;
+ struct task_struct *task;
+ spinlock_t task_lock;
+};
+
+struct ehca_comp_pool {
+ struct ehca_cpu_comp_task *cpu_comp_tasks;
+ int last_cpu;
+ spinlock_t last_cpu_lock;
+};
+
+struct ehca_comp_pool *ehca_create_comp_pool(void);
+void ehca_destroy_comp_pool(struct ehca_comp_pool *pool);
+void ehca_queue_comp_task(struct ehca_comp_pool *pool, struct ehca_cq *__cq);
+
+#endif
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_irq.c 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_irq.c 2006-05-15 13:29:49.000000000 +0200
@@ -0,0 +1,710 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * Functions for EQs, NEQs and interrupts
+ *
+ * Authors: Heiko J Schick <schickhj@de.ibm.com>
+ * Khadija Souissi <souissi@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#define DEB_PREFIX "eirq"
+
+#include "ehca_classes.h"
+#include "ehca_irq.h"
+#include "ehca_iverbs.h"
+#include "ehca_tools.h"
+#include "hcp_if.h"
+#include "hipz_fns.h"
+
+#define EQE_COMPLETION_EVENT EHCA_BMASK_IBM(1,1)
+#define EQE_CQ_QP_NUMBER EHCA_BMASK_IBM(8,31)
+#define EQE_EE_IDENTIFIER EHCA_BMASK_IBM(2,7)
+#define EQE_CQ_NUMBER EHCA_BMASK_IBM(8,31)
+#define EQE_QP_NUMBER EHCA_BMASK_IBM(8,31)
+#define EQE_QP_TOKEN EHCA_BMASK_IBM(32,63)
+#define EQE_CQ_TOKEN EHCA_BMASK_IBM(32,63)
+
+#define NEQE_COMPLETION_EVENT EHCA_BMASK_IBM(1,1)
+#define NEQE_EVENT_CODE EHCA_BMASK_IBM(2,7)
+#define NEQE_PORT_NUMBER EHCA_BMASK_IBM(8,15)
+#define NEQE_PORT_AVAILABILITY EHCA_BMASK_IBM(16,16)
+
+#define ERROR_DATA_LENGTH EHCA_BMASK_IBM(52,63)
+#define ERROR_DATA_TYPE EHCA_BMASK_IBM(0,7)
+
+static inline void comp_event_callback(struct ehca_cq *cq)
+{
+ EDEB_EN(7, "cq=%p", cq);
+
+ if (!cq->ib_cq.comp_handler)
+ return;
+
+ spin_lock(&cq->cb_lock);
+ cq->ib_cq.comp_handler(&cq->ib_cq, cq->ib_cq.cq_context);
+ spin_unlock(&cq->cb_lock);
+
+ EDEB_EX(7, "cq=%p", cq);
+
+ return;
+}
+
+static void print_error_data(struct ehca_shca * shca, void* data,
+ u64* rblock, int length)
+{
+ u64 type = EHCA_BMASK_GET(ERROR_DATA_TYPE, rblock[2]);
+ u64 resource = rblock[1];
+
+ EDEB_EN(7, "shca=%p data=%p rblock=%p length=%x",
+ shca, data, rblock, length);
+
+ switch (type) {
+ case 0x1: /* Queue Pair */
+ {
+ struct ehca_qp *qp = (struct ehca_qp*)data;
+
+ /* only print error data if AER is set */
+ if (rblock[6] == 0)
+ return;
+
+ EDEB_ERR(4, "QP 0x%x (resource=%lx) has errors.",
+ qp->ib_qp.qp_num, resource);
+ break;
+ }
+ case 0x4: /* Completion Queue */
+ {
+ struct ehca_cq *cq = (struct ehca_cq*)data;
+
+ EDEB_ERR(4, "CQ 0x%x (resource=%lx) has errors.",
+ cq->cq_number, resource);
+ break;
+ }
+ default:
+ EDEB_ERR(4, "Unknown errror type: %lx on %s.",
+ type, shca->ib_device.name);
+ break;
+ }
+
+ EDEB_ERR(4, "Error data is available: %lx.", resource);
+ EDEB_ERR(4, "EHCA ----- error data begin "
+ "---------------------------------------------------");
+ EDEB_DMP(4, rblock, length, "resource=%lx", resource);
+ EDEB_ERR(4, "EHCA ----- error data end "
+ "----------------------------------------------------");
+
+ EDEB_EX(7, "");
+
+ return;
+}
+
+int ehca_error_data(struct ehca_shca *shca, void *data,
+ u64 resource)
+{
+
+ unsigned long ret = 0;
+ u64 *rblock;
+ unsigned long block_count;
+
+ EDEB_EN(7, "shca=%p data=%p resource=%lx", shca, data, resource);
+
+ rblock = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!rblock) {
+ EDEB_ERR(4, "Cannot allocate rblock memory.");
+ ret = -ENOMEM;
+ goto error_data1;
+ }
+
+ ret = hipz_h_error_data(shca->ipz_hca_handle,
+ resource,
+ rblock,
+ &block_count);
+
+ if (ret == H_R_STATE) {
+ EDEB_ERR(4, "No error data is available: %lx.", resource);
+ }
+ else if (ret == H_SUCCESS) {
+ int length;
+
+ length = EHCA_BMASK_GET(ERROR_DATA_LENGTH, rblock[0]);
+
+ if (length > PAGE_SIZE)
+ length = PAGE_SIZE;
+
+ print_error_data(shca, data, rblock, length);
+ }
+ else {
+ EDEB_ERR(4, "Error data could not be fetched: %lx", resource);
+ }
+
+ kfree(rblock);
+
+error_data1:
+ return ret;
+
+}
+
+static void qp_event_callback(struct ehca_shca *shca,
+ u64 eqe,
+ enum ib_event_type event_type)
+{
+ struct ib_event event;
+ struct ehca_qp *qp;
+ unsigned long flags;
+ u32 token = EHCA_BMASK_GET(EQE_QP_TOKEN, eqe);
+
+ EDEB_EN(7, "eqe=%lx", eqe);
+
+ spin_lock_irqsave(&ehca_qp_idr_lock, flags);
+ qp = idr_find(&ehca_qp_idr, token);
+ spin_unlock_irqrestore(&ehca_qp_idr_lock, flags);
+
+
+ if (!qp)
+ return;
+
+ ehca_error_data(shca, qp, qp->ipz_qp_handle.handle);
+
+ if (!qp->ib_qp.event_handler)
+ return;
+
+ event.device = &shca->ib_device;
+ event.event = event_type;
+ event.element.qp = &qp->ib_qp;
+
+ qp->ib_qp.event_handler(&event, qp->ib_qp.qp_context);
+
+ EDEB_EX(7, "qp=%p", qp);
+
+ return;
+}
+
+static void cq_event_callback(struct ehca_shca *shca,
+ u64 eqe)
+{
+ struct ehca_cq *cq;
+ unsigned long flags;
+ u32 token = EHCA_BMASK_GET(EQE_CQ_TOKEN, eqe);
+
+ EDEB_EN(7, "eqe=%lx", eqe);
+
+ spin_lock_irqsave(&ehca_cq_idr_lock, flags);
+ cq = idr_find(&ehca_cq_idr, token);
+ spin_unlock_irqrestore(&ehca_cq_idr_lock, flags);
+
+ if (!cq)
+ return;
+
+ ehca_error_data(shca, cq, cq->ipz_cq_handle.handle);
+
+ EDEB_EX(7, "cq=%p", cq);
+
+ return;
+}
+
+static void parse_identifier(struct ehca_shca *shca, u64 eqe)
+{
+ u8 identifier = EHCA_BMASK_GET(EQE_EE_IDENTIFIER, eqe);
+
+ EDEB_EN(7, "shca=%p eqe=%lx", shca, eqe);
+
+ switch (identifier) {
+ case 0x02: /* path migrated */
+ qp_event_callback(shca, eqe, IB_EVENT_PATH_MIG);
+ break;
+ case 0x03: /* communication established */
+ qp_event_callback(shca, eqe, IB_EVENT_COMM_EST);
+ break;
+ case 0x04: /* send queue drained */
+ qp_event_callback(shca, eqe, IB_EVENT_SQ_DRAINED);
+ break;
+ case 0x05: /* QP error */
+ case 0x06: /* QP error */
+ qp_event_callback(shca, eqe, IB_EVENT_QP_FATAL);
+ break;
+ case 0x07: /* CQ error */
+ case 0x08: /* CQ error */
+ cq_event_callback(shca, eqe);
+ break;
+ case 0x09: /* MRMWPTE error */
+ EDEB_ERR(4, "MRMWPTE error.");
+ break;
+ case 0x0A: /* port event */
+ EDEB_ERR(4, "Port event.");
+ break;
+ case 0x0B: /* MR access error */
+ EDEB_ERR(4, "MR access error.");
+ break;
+ case 0x0C: /* EQ error */
+ EDEB_ERR(4, "EQ error.");
+ break;
+ case 0x0D: /* P/Q_Key mismatch */
+ EDEB_ERR(4, "P/Q_Key mismatch.");
+ break;
+ case 0x10: /* sampling complete */
+ EDEB_ERR(4, "Sampling complete.");
+ break;
+ case 0x11: /* unaffiliated access error */
+ EDEB_ERR(4, "Unaffiliated access error.");
+ break;
+ case 0x12: /* path migrating error */
+ EDEB_ERR(4, "Path migration error.");
+ break;
+ case 0x13: /* interface trace stopped */
+ EDEB_ERR(4, "Interface trace stopped.");
+ break;
+ case 0x14: /* first error capture info available */
+ default:
+ EDEB_ERR(4, "Unknown identifier: %x on %s.",
+ identifier, shca->ib_device.name);
+ break;
+ }
+
+ EDEB_EX(7, "eqe=%lx identifier=%x", eqe, identifier);
+
+ return;
+}
+
+static void parse_ec(struct ehca_shca *shca, u64 eqe)
+{
+ struct ib_event event;
+ u8 ec = EHCA_BMASK_GET(NEQE_EVENT_CODE, eqe);
+ u8 port = EHCA_BMASK_GET(NEQE_PORT_NUMBER, eqe);
+
+ EDEB_EN(7, "shca=%p eqe=%lx", shca, eqe);
+
+ switch (ec) {
+ case 0x30: /* port availability change */
+ if (EHCA_BMASK_GET(NEQE_PORT_AVAILABILITY, eqe)) {
+ EDEB(4, "%s: port %x is active.",
+ shca->ib_device.name, port);
+ event.device = &shca->ib_device;
+ event.event = IB_EVENT_PORT_ACTIVE;
+ event.element.port_num = port;
+ shca->sport[port - 1].port_state = IB_PORT_ACTIVE;
+ ib_dispatch_event(&event);
+ } else {
+ EDEB(4, "%s: port %x is inactive.",
+ shca->ib_device.name, port);
+ event.device = &shca->ib_device;
+ event.event = IB_EVENT_PORT_ERR;
+ event.element.port_num = port;
+ shca->sport[port - 1].port_state = IB_PORT_DOWN;
+ ib_dispatch_event(&event);
+ }
+ break;
+ case 0x31:
+ /* port configuration change */
+ /* disruptive change is caused by */
+ /* LID, PKEY or SM change */
+ EDEB(4, "EHCA disruptive port %x "
+ "configuration change.", port);
+
+ EDEB(4, "%s: port %x is inactive.",
+ shca->ib_device.name, port);
+ event.device = &shca->ib_device;
+ event.event = IB_EVENT_PORT_ERR;
+ event.element.port_num = port;
+ shca->sport[port - 1].port_state = IB_PORT_DOWN;
+ ib_dispatch_event(&event);
+
+ EDEB(4, "%s: port %x is active.",
+ shca->ib_device.name, port);
+ event.device = &shca->ib_device;
+ event.event = IB_EVENT_PORT_ACTIVE;
+ event.element.port_num = port;
+ shca->sport[port - 1].port_state = IB_PORT_ACTIVE;
+ ib_dispatch_event(&event);
+ break;
+ case 0x32: /* adapter malfunction */
+ EDEB_ERR(4, "Adapter malfunction.");
+ break;
+ case 0x33: /* trace stopped */
+ EDEB_ERR(4, "Traced stopped.");
+ break;
+ default:
+ EDEB_ERR(4, "Unknown event code: %x on %s.",
+ ec, shca->ib_device.name);
+ break;
+ }
+
+ EDEB_EN(7, "eqe=%lx ec=%x", eqe, ec);
+
+ return;
+}
+
+static inline void reset_eq_pending(struct ehca_cq *cq)
+{
+ u64 CQx_EP = 0;
+ struct h_galpa gal = cq->galpas.kernel;
+
+ EDEB_EN(7, "cq=%p", cq);
+
+ hipz_galpa_store_cq(gal, cqx_ep, 0x0);
+ CQx_EP = hipz_galpa_load(gal, CQTEMM_OFFSET(cqx_ep));
+ EDEB(7, "CQx_EP=%lx", CQx_EP);
+
+ EDEB_EX(7, "cq=%p", cq);
+
+ return;
+}
+
+irqreturn_t ehca_interrupt_neq(int irq, void *dev_id, struct pt_regs *regs)
+{
+ struct ehca_shca *shca = (struct ehca_shca*)dev_id;
+
+ EDEB_EN(7, "dev_id=%p", dev_id);
+
+ tasklet_hi_schedule(&shca->neq.interrupt_task);
+
+ EDEB_EX(7, "");
+
+ return IRQ_HANDLED;
+}
+
+void ehca_tasklet_neq(unsigned long data)
+{
+ struct ehca_shca *shca = (struct ehca_shca*)data;
+ struct ehca_eqe *eqe;
+ u64 ret = H_SUCCESS;
+
+ EDEB_EN(7, "shca=%p", shca);
+
+ eqe = (struct ehca_eqe *)ehca_poll_eq(shca, &shca->neq);
+
+ while (eqe) {
+ if (!EHCA_BMASK_GET(NEQE_COMPLETION_EVENT, eqe->entry))
+ parse_ec(shca, eqe->entry);
+
+ eqe = (struct ehca_eqe *)ehca_poll_eq(shca, &shca->neq);
+ }
+
+ ret = hipz_h_reset_event(shca->ipz_hca_handle,
+ shca->neq.ipz_eq_handle, 0xFFFFFFFFFFFFFFFFL);
+
+ if (ret != H_SUCCESS)
+ EDEB_ERR(4, "Can't clear notification events.");
+
+ EDEB_EX(7, "shca=%p", shca);
+
+ return;
+}
+
+irqreturn_t ehca_interrupt_eq(int irq, void *dev_id, struct pt_regs *regs)
+{
+ struct ehca_shca *shca = (struct ehca_shca*)dev_id;
+
+ EDEB_EN(7, "dev_id=%p", dev_id);
+
+ tasklet_hi_schedule(&shca->eq.interrupt_task);
+
+ EDEB_EX(7, "");
+
+ return IRQ_HANDLED;
+}
+
+void ehca_tasklet_eq(unsigned long data)
+{
+ struct ehca_shca *shca = (struct ehca_shca*)data;
+ struct ehca_eqe *eqe;
+ int int_state;
+
+ EDEB_EN(7, "shca=%p", shca);
+
+ do {
+ eqe = (struct ehca_eqe *)ehca_poll_eq(shca, &shca->eq);
+
+ if ((shca->hw_level >= 2) && eqe)
+ int_state = 1;
+ else
+ int_state = 0;
+
+ while ((int_state == 1) || eqe) {
+ while (eqe) {
+ u64 eqe_value = eqe->entry;
+
+ EDEB(7, "eqe_value=%lx", eqe_value);
+
+ /* TODO: better structure */
+ if (EHCA_BMASK_GET(EQE_COMPLETION_EVENT,
+ eqe_value)) {
+ extern struct ehca_comp_pool* ehca_pool;
+ extern struct idr ehca_cq_idr;
+ unsigned long flags;
+ u32 token;
+ struct ehca_cq *cq;
+
+ EDEB(6, "... completion event");
+ token =
+ EHCA_BMASK_GET(EQE_CQ_TOKEN,
+ eqe_value);
+ spin_lock_irqsave(&ehca_cq_idr_lock,
+ flags);
+ cq = idr_find(&ehca_cq_idr, token);
+
+ if (cq == NULL) {
+ spin_unlock(&ehca_cq_idr_lock);
+ break;
+ }
+
+ reset_eq_pending(cq);
+ ehca_queue_comp_task(ehca_pool, cq);
+ spin_unlock_irqrestore(&ehca_cq_idr_lock,
+ flags);
+ } else {
+ EDEB(6, "... non completion event");
+ parse_identifier(shca, eqe_value);
+ }
+ eqe =
+ (struct ehca_eqe *)ehca_poll_eq(shca,
+ &shca->eq);
+ }
+
+ if (shca->hw_level >= 2)
+ int_state =
+ hipz_h_query_int_state(shca->ipz_hca_handle,
+ shca->eq.ist);
+ eqe = (struct ehca_eqe *)ehca_poll_eq(shca, &shca->eq);
+
+ }
+ } while (int_state != 0);
+
+ EDEB_EX(7, "shca=%p", shca);
+
+ return;
+}
+
+static inline int find_next_online_cpu(struct ehca_comp_pool* pool)
+{
+ unsigned long flags_last_cpu;
+
+ spin_lock_irqsave(&pool->last_cpu_lock, flags_last_cpu);
+ pool->last_cpu = next_cpu(pool->last_cpu, cpu_online_map);
+
+ if (pool->last_cpu == NR_CPUS)
+ pool->last_cpu = 0;
+
+ spin_unlock_irqrestore(&pool->last_cpu_lock, flags_last_cpu);
+
+ return pool->last_cpu;
+}
+
+void ehca_queue_comp_task(struct ehca_comp_pool *pool, struct ehca_cq *__cq)
+{
+ int cpu;
+ int cpu_id;
+ struct ehca_cpu_comp_task *cct;
+ unsigned long flags_cct;
+ unsigned long flags_cq;
+
+ cpu = get_cpu();
+ cpu_id = find_next_online_cpu(pool);
+
+ EDEB_EN(7, "pool=%p cq=%p cq_nr=%x CPU=%x:%x:%x:%x",
+ pool, __cq, __cq->cq_number,
+ cpu, cpu_id, num_online_cpus(), num_possible_cpus());
+
+ BUG_ON(!cpu_online(cpu_id));
+
+ cct = per_cpu_ptr(pool->cpu_comp_tasks, cpu_id);
+
+ spin_lock_irqsave(&cct->task_lock, flags_cct);
+ spin_lock_irqsave(&__cq->task_lock, flags_cq);
+
+ if (__cq->nr_callbacks == 0) {
+ __cq->nr_callbacks++;
+ list_add_tail(&__cq->entry, &cct->cq_list);
+ wake_up(&cct->wait_queue);
+ }
+ else
+ __cq->nr_callbacks++;
+
+ spin_unlock_irqrestore(&__cq->task_lock, flags_cq);
+ spin_unlock_irqrestore(&cct->task_lock, flags_cct);
+
+ put_cpu();
+
+ EDEB_EX(7, "cct=%p", cct);
+
+ return;
+}
+
+static void run_comp_task(struct ehca_cpu_comp_task* cct)
+{
+ struct ehca_cq *cq = NULL;
+ unsigned long flags_cct;
+ unsigned long flags_cq;
+
+
+ EDEB_EN(7, "cct=%p", cct);
+
+ spin_lock_irqsave(&cct->task_lock, flags_cct);
+
+ while (!list_empty(&cct->cq_list)) {
+ cq = list_entry(cct->cq_list.next, struct ehca_cq, entry);
+ spin_unlock_irqrestore(&cct->task_lock, flags_cct);
+ comp_event_callback(cq);
+ spin_lock_irqsave(&cct->task_lock, flags_cct);
+
+ spin_lock_irqsave(&cq->task_lock, flags_cq);
+ cq->nr_callbacks--;
+ if (cq->nr_callbacks == 0)
+ list_del_init(cct->cq_list.next);
+ spin_unlock_irqrestore(&cq->task_lock, flags_cq);
+
+ }
+
+ spin_unlock_irqrestore(&cct->task_lock, flags_cct);
+
+ EDEB_EX(7, "cct=%p cq=%p", cct, cq);
+
+ return;
+}
+
+static int comp_task(void *__cct)
+{
+ struct ehca_cpu_comp_task* cct = __cct;
+ DECLARE_WAITQUEUE(wait, current);
+
+ EDEB_EN(7, "cct=%p", cct);
+
+ set_current_state(TASK_INTERRUPTIBLE);
+ while(!kthread_should_stop()) {
+ add_wait_queue(&cct->wait_queue, &wait);
+
+ if (list_empty(&cct->cq_list))
+ schedule();
+ else
+ __set_current_state(TASK_RUNNING);
+
+ remove_wait_queue(&cct->wait_queue, &wait);
+
+ if (!list_empty(&cct->cq_list))
+ run_comp_task(__cct);
+
+ set_current_state(TASK_INTERRUPTIBLE);
+ }
+ __set_current_state(TASK_RUNNING);
+
+ EDEB_EX(7, "");
+
+ return 0;
+}
+
+static struct task_struct *create_comp_task(struct ehca_comp_pool *pool,
+ int cpu)
+{
+ struct ehca_cpu_comp_task *cct;
+
+ EDEB_EN(7, "cpu=%d:%d", cpu, NR_CPUS);
+
+ cct = per_cpu_ptr(pool->cpu_comp_tasks, cpu);
+ spin_lock_init(&cct->task_lock);
+ INIT_LIST_HEAD(&cct->cq_list);
+ init_waitqueue_head(&cct->wait_queue);
+ cct->task = kthread_create(comp_task, cct, "ehca_comp/%d", cpu);
+
+ EDEB_EX(7, "cct/%d=%p", cpu, cct);
+
+ return cct->task;
+}
+
+static void destroy_comp_task(struct ehca_comp_pool *pool,
+ int cpu)
+{
+ struct ehca_cpu_comp_task *cct;
+ struct task_struct *task;
+
+ EDEB_EN(7, "pool=%p cpu=%d:%d", pool, cpu, NR_CPUS);
+
+ cct = per_cpu_ptr(pool->cpu_comp_tasks, cpu);
+ cct->task = NULL;
+ task = cct->task;
+
+ if (task)
+ kthread_stop(task);
+
+ EDEB_EX(7, "");
+
+ return;
+}
+
+struct ehca_comp_pool *ehca_create_comp_pool(void)
+{
+ struct ehca_comp_pool *pool;
+ int cpu;
+ struct task_struct *task;
+
+ EDEB_EN(7, "");
+
+ pool = kzalloc(sizeof(struct ehca_comp_pool), GFP_KERNEL);
+ if (pool == NULL)
+ return NULL;
+
+ spin_lock_init(&pool->last_cpu_lock);
+ pool->last_cpu = any_online_cpu(cpu_online_map);
+
+ pool->cpu_comp_tasks = alloc_percpu(struct ehca_cpu_comp_task);
+ if (pool->cpu_comp_tasks == NULL) {
+ kfree(pool);
+ return NULL;
+ }
+
+ for_each_online_cpu(cpu) {
+ task = create_comp_task(pool, cpu);
+ if (task) {
+ kthread_bind(task, cpu);
+ wake_up_process(task);
+ }
+ }
+
+ EDEB_EX(7, "pool=%p", pool);
+
+ return pool;
+}
+
+void ehca_destroy_comp_pool(struct ehca_comp_pool *pool)
+{
+ int i;
+
+ EDEB_EN(7, "pool=%p", pool);
+
+ for (i = 0; i < NR_CPUS; i++) {
+ if (cpu_online(i))
+ destroy_comp_task(pool, i);
+ }
+
+ EDEB_EN(7, "");
+
+ return;
+}
^ permalink raw reply
* [PATCH 05/16] ehca: common include files
From: Heiko J Schick @ 2006-05-15 17:41 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
drivers/infiniband/hw/ehca/ehca_iverbs.h | 181 +++++++++++++
drivers/infiniband/hw/ehca/ehca_tools.h | 411 +++++++++++++++++++++++++++++++
2 files changed, 592 insertions(+)
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_iverbs.h 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_iverbs.h 2006-04-28 14:20:08.000000000 +0200
@@ -0,0 +1,181 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * Function definitions for internal functions
+ *
+ * Authors: Heiko J Schick <schickhj@de.ibm.com>
+ * Dietmar Decker <ddecker@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef __EHCA_IVERBS_H__
+#define __EHCA_IVERBS_H__
+
+#include "ehca_classes.h"
+
+int ehca_query_device(struct ib_device *ibdev, struct ib_device_attr *props);
+
+int ehca_query_port(struct ib_device *ibdev, u8 port,
+ struct ib_port_attr *props);
+
+int ehca_query_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 * pkey);
+
+int ehca_query_gid(struct ib_device *ibdev, u8 port, int index,
+ union ib_gid *gid);
+
+int ehca_modify_port(struct ib_device *ibdev, u8 port, int port_modify_mask,
+ struct ib_port_modify *props);
+
+struct ib_pd *ehca_alloc_pd(struct ib_device *device,
+ struct ib_ucontext *context,
+ struct ib_udata *udata);
+
+int ehca_dealloc_pd(struct ib_pd *pd);
+
+struct ib_ah *ehca_create_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr);
+
+int ehca_modify_ah(struct ib_ah *ah, struct ib_ah_attr *ah_attr);
+
+int ehca_query_ah(struct ib_ah *ah, struct ib_ah_attr *ah_attr);
+
+int ehca_destroy_ah(struct ib_ah *ah);
+
+struct ib_mr *ehca_get_dma_mr(struct ib_pd *pd, int mr_access_flags);
+
+struct ib_mr *ehca_reg_phys_mr(struct ib_pd *pd,
+ struct ib_phys_buf *phys_buf_array,
+ int num_phys_buf,
+ int mr_access_flags, u64 *iova_start);
+
+struct ib_mr *ehca_reg_user_mr(struct ib_pd *pd,
+ struct ib_umem *region,
+ int mr_access_flags, struct ib_udata *udata);
+
+int ehca_rereg_phys_mr(struct ib_mr *mr,
+ int mr_rereg_mask,
+ struct ib_pd *pd,
+ struct ib_phys_buf *phys_buf_array,
+ int num_phys_buf, int mr_access_flags, u64 *iova_start);
+
+int ehca_query_mr(struct ib_mr *mr, struct ib_mr_attr *mr_attr);
+
+int ehca_dereg_mr(struct ib_mr *mr);
+
+struct ib_mw *ehca_alloc_mw(struct ib_pd *pd);
+
+int ehca_bind_mw(struct ib_qp *qp, struct ib_mw *mw,
+ struct ib_mw_bind *mw_bind);
+
+int ehca_dealloc_mw(struct ib_mw *mw);
+
+struct ib_fmr *ehca_alloc_fmr(struct ib_pd *pd,
+ int mr_access_flags,
+ struct ib_fmr_attr *fmr_attr);
+
+int ehca_map_phys_fmr(struct ib_fmr *fmr,
+ u64 *page_list, int list_len, u64 iova);
+
+int ehca_unmap_fmr(struct list_head *fmr_list);
+
+int ehca_dealloc_fmr(struct ib_fmr *fmr);
+
+enum ehca_eq_type {
+ EHCA_EQ = 0, /* Event Queue */
+ EHCA_NEQ /* Notification Event Queue */
+};
+
+int ehca_create_eq(struct ehca_shca *shca, struct ehca_eq *eq,
+ enum ehca_eq_type type, const u32 length);
+
+int ehca_destroy_eq(struct ehca_shca *shca, struct ehca_eq *eq);
+
+void *ehca_poll_eq(struct ehca_shca *shca, struct ehca_eq *eq);
+
+
+struct ib_cq *ehca_create_cq(struct ib_device *device, int cqe,
+ struct ib_ucontext *context,
+ struct ib_udata *udata);
+
+int ehca_destroy_cq(struct ib_cq *cq);
+
+int ehca_resize_cq(struct ib_cq *cq, int cqe, struct ib_udata *udata);
+
+int ehca_poll_cq(struct ib_cq *cq, int num_entries, struct ib_wc *wc);
+
+int ehca_peek_cq(struct ib_cq *cq, int wc_cnt);
+
+int ehca_req_notify_cq(struct ib_cq *cq, enum ib_cq_notify cq_notify);
+
+struct ib_qp *ehca_create_qp(struct ib_pd *pd,
+ struct ib_qp_init_attr *init_attr,
+ struct ib_udata *udata);
+
+int ehca_destroy_qp(struct ib_qp *qp);
+
+int ehca_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, int attr_mask);
+
+int ehca_query_qp(struct ib_qp *qp, struct ib_qp_attr *qp_attr,
+ int qp_attr_mask, struct ib_qp_init_attr *qp_init_attr);
+
+int ehca_post_send(struct ib_qp *qp, struct ib_send_wr *send_wr,
+ struct ib_send_wr **bad_send_wr);
+
+int ehca_post_recv(struct ib_qp *qp, struct ib_recv_wr *recv_wr,
+ struct ib_recv_wr **bad_recv_wr);
+
+u64 ehca_define_sqp(struct ehca_shca *shca, struct ehca_qp *ibqp,
+ struct ib_qp_init_attr *qp_init_attr);
+
+int ehca_attach_mcast(struct ib_qp *qp, union ib_gid *gid, u16 lid);
+
+int ehca_detach_mcast(struct ib_qp *qp, union ib_gid *gid, u16 lid);
+
+struct ib_ucontext *ehca_alloc_ucontext(struct ib_device *device,
+ struct ib_udata *udata);
+
+int ehca_dealloc_ucontext(struct ib_ucontext *context);
+
+int ehca_mmap(struct ib_ucontext *context, struct vm_area_struct *vma);
+
+void ehca_poll_eqs(unsigned long data);
+
+int ehca_mmap_nopage(u64 foffset,u64 length,void **mapped,
+ struct vm_area_struct **vma);
+
+int ehca_mmap_register(u64 physical,void **mapped,
+ struct vm_area_struct **vma);
+
+int ehca_munmap(unsigned long addr, size_t len);
+
+#endif
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_tools.h 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_tools.h 2006-05-03 13:44:15.000000000 +0200
@@ -0,0 +1,411 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * auxiliary functions
+ *
+ * Authors: Christoph Raisch <raisch@de.ibm.com>
+ * Hoang-Nam Nguyen <hnguyen@de.ibm.com>
+ * Khadija Souissi <souissik@de.ibm.com>
+ * Waleri Fomin <fomin@de.ibm.com>
+ * Heiko J Schick <schickhj@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+
+#ifndef EHCA_TOOLS_H
+#define EHCA_TOOLS_H
+
+#include <linux/kernel.h>
+#include <linux/spinlock.h>
+#include <linux/delay.h>
+#include <linux/idr.h>
+#include <linux/kthread.h>
+#include <linux/mm.h>
+#include <linux/mman.h>
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+#include <linux/vmalloc.h>
+#include <linux/version.h>
+
+#include <asm/abs_addr.h>
+#include <asm/ibmebus.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+
+#define EHCA_EDEB_TRACE_MASK_SIZE 32
+extern u8 ehca_edeb_mask[EHCA_EDEB_TRACE_MASK_SIZE];
+#define EDEB_ID_TO_U32(str4) (str4[3] | (str4[2] << 8) | (str4[1] << 16) | \
+ (str4[0] << 24))
+
+static inline u64 ehca_edeb_filter(const u32 level,
+ const u32 id, const u32 line)
+{
+ u64 ret = 0;
+ u32 filenr = 0;
+ u32 filter_level = 9;
+ u32 dynamic_level = 0;
+
+ /* This is code written for the gcc -O2 optimizer which should colapse
+ * to two single ints filter_level is the first level kicked out by
+ * compiler means trace everythin below 6. */
+ if (id == EDEB_ID_TO_U32("ehav")) {
+ filenr = 0x01;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("clas")) {
+ filenr = 0x02;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("cqeq")) {
+ filenr = 0x03;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("shca")) {
+ filenr = 0x05;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("eirq")) {
+ filenr = 0x06;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("lMad")) {
+ filenr = 0x07;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("mcas")) {
+ filenr = 0x08;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("mrmw")) {
+ filenr = 0x09;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("vpd ")) {
+ filenr = 0x0a;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("e_qp")) {
+ filenr = 0x0b;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("uqes")) {
+ filenr = 0x0c;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("PHYP")) {
+ filenr = 0x0d;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("hcpi")) {
+ filenr = 0x0e;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("iptz")) {
+ filenr = 0x0f;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("spta")) {
+ filenr = 0x10;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("simp")) {
+ filenr = 0x11;
+ filter_level = 8;
+ }
+ if (id == EDEB_ID_TO_U32("reqs")) {
+ filenr = 0x12;
+ filter_level = 8;
+ }
+
+ if ((filenr - 1) > sizeof(ehca_edeb_mask)) {
+ filenr = 0;
+ }
+
+ if (filenr == 0) {
+ filter_level = 9;
+ } /* default */
+ ret = filenr * 0x10000 + line;
+ if (filter_level <= level) {
+ return ret | 0x100000000L; /* this is the flag to not trace */
+ }
+ dynamic_level = ehca_edeb_mask[filenr];
+ if (likely(dynamic_level <= level)) {
+ ret = ret | 0x100000000L;
+ };
+ return ret;
+}
+
+#ifdef EHCA_USE_HCALL_KERNEL
+#ifdef CONFIG_PPC_PSERIES
+
+#include <asm/paca.h>
+
+/**
+ * IS_EDEB_ON - Checks if debug is on for the given level.
+ */
+#define IS_EDEB_ON(level) \
+ ((ehca_edeb_filter(level, EDEB_ID_TO_U32(DEB_PREFIX), __LINE__) & 0x100000000L)==0)
+
+#define EDEB_P_GENERIC(level,idstring,format,args...) \
+do { \
+ u64 ehca_edeb_filterresult = \
+ ehca_edeb_filter(level, EDEB_ID_TO_U32(DEB_PREFIX), __LINE__);\
+ if ((ehca_edeb_filterresult & 0x100000000L) == 0) \
+ printk("PU%04x %08x:%s " idstring " "format "\n", \
+ get_paca()->paca_index, (u32)(ehca_edeb_filterresult), \
+ __func__, ##args); \
+} while (1 == 0)
+
+#elif REAL_HCALL
+
+#define EDEB_P_GENERIC(level,idstring,format,args...) \
+do { \
+ u64 ehca_edeb_filterresult = \
+ ehca_edeb_filter(level, EDEB_ID_TO_U32(DEB_PREFIX), __LINE__); \
+ if ((ehca_edeb_filterresult & 0x100000000L) == 0) \
+ printk("%08x:%s " idstring " "format "\n", \
+ (u32)(ehca_edeb_filterresult), \
+ __func__, ##args); \
+} while (1 == 0)
+
+#endif
+#else
+
+#define IS_EDEB_ON(level) (1)
+
+#define EDEB_P_GENERIC(level,idstring,format,args...) \
+do { \
+ printk("%s " idstring " "format "\n", \
+ __func__, ##args); \
+} while (1 == 0)
+
+#endif
+
+/**
+ * EDEB - Trace output macro.
+ * @level tracelevel
+ * @format optional format string, use "" if not desired
+ * @args printf like arguments for trace, use %Lx for u64, %x for u32
+ * %p for pointer
+ */
+#define EDEB(level,format,args...) \
+ EDEB_P_GENERIC(level,"",format,##args)
+#define EDEB_ERR(level,format,args...) \
+ EDEB_P_GENERIC(level,"HCAD_ERROR ",format,##args)
+#define EDEB_EN(level,format,args...) \
+ EDEB_P_GENERIC(level,">>>",format,##args)
+#define EDEB_EX(level,format,args...) \
+ EDEB_P_GENERIC(level,"<<<",format,##args)
+
+/**
+ * EDEB macro to dump a memory block, whose length is n*8 bytes.
+ * Each line has the following layout:
+ * <format string> adr=X ofs=Y <8 bytes hex> <8 bytes hex>
+ */
+#define EDEB_DMP(level,adr,len,format,args...) \
+ do { \
+ unsigned int x; \
+ unsigned int l = (unsigned int)(len); \
+ unsigned char *deb = (unsigned char*)(adr); \
+ for (x = 0; x < l; x += 16) { \
+ EDEB(level, format " adr=%p ofs=%04x %016lx %016lx", \
+ ##args, deb, x, *((u64 *)&deb[0]), *((u64 *)&deb[8])); \
+ deb += 16; \
+ } \
+ } while (0)
+
+/* define a bitmask, little endian version */
+#define EHCA_BMASK(pos,length) (((pos)<<16)+(length))
+/* define a bitmask, the ibm way... */
+#define EHCA_BMASK_IBM(from,to) (((63-to)<<16)+((to)-(from)+1))
+/* internal function, don't use */
+#define EHCA_BMASK_SHIFTPOS(mask) (((mask)>>16)&0xffff)
+/* internal function, don't use */
+#define EHCA_BMASK_MASK(mask) (0xffffffffffffffffULL >> ((64-(mask))&0xffff))
+/* return value shifted and masked by mask\n
+ * variable|=HCA_BMASK_SET(MY_MASK,0x4711) ORs the bits in variable\n
+ * variable&=~HCA_BMASK_SET(MY_MASK,-1) clears the bits from the mask
+ * in variable
+ */
+#define EHCA_BMASK_SET(mask,value) \
+ ((EHCA_BMASK_MASK(mask) & ((u64)(value)))<<EHCA_BMASK_SHIFTPOS(mask))
+/* extract a parameter from value by mask\n
+ * param=EHCA_BMASK_GET(MY_MASK,value)
+ */
+#define EHCA_BMASK_GET(mask,value) \
+ ( EHCA_BMASK_MASK(mask)& (((u64)(value))>>EHCA_BMASK_SHIFTPOS(mask)))
+
+#define PARANOIA_MODE
+#ifdef PARANOIA_MODE
+
+#define EHCA_CHECK_ADR_P(adr) \
+ if (unlikely(adr == 0)) { \
+ EDEB_ERR(4, "adr=%p check failed line %i", adr, \
+ __LINE__); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_CHECK_ADR(adr) \
+ if (unlikely(adr == 0)) { \
+ EDEB_ERR(4, "adr=%p check failed line %i", adr, \
+ __LINE__); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_DEVICE_P(device) \
+ if (unlikely(device == 0)) { \
+ EDEB_ERR(4, "device=%p check failed", device); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_CHECK_DEVICE(device) \
+ if (unlikely(device == 0)) { \
+ EDEB_ERR(4, "device=%p check failed", device); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_PD(pd) \
+ if (unlikely(pd == 0)) { \
+ EDEB_ERR(4, "pd=%p check failed", pd); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_PD_P(pd) \
+ if (unlikely(pd == 0)) { \
+ EDEB_ERR(4, "pd=%p check failed", pd); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_CHECK_AV(av) \
+ if (unlikely(av == 0)) { \
+ EDEB_ERR(4, "av=%p check failed", av); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_AV_P(av) \
+ if (unlikely(av == 0)) { \
+ EDEB_ERR(4, "av=%p check failed", av); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_CHECK_CQ(cq) \
+ if (unlikely(cq == 0)) { \
+ EDEB_ERR(4, "cq=%p check failed", cq); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_CQ_P(cq) \
+ if (unlikely(cq == 0)) { \
+ EDEB_ERR(4, "cq=%p check failed", cq); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_CHECK_EQ(eq) \
+ if (unlikely(eq == 0)) { \
+ EDEB_ERR(4, "eq=%p check failed", eq); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_EQ_P(eq) \
+ if (unlikely(eq == 0)) { \
+ EDEB_ERR(4, "eq=%p check failed", eq); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_CHECK_QP(qp) \
+ if (unlikely(qp == 0)) { \
+ EDEB_ERR(4, "qp=%p check failed", qp); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_QP_P(qp) \
+ if (unlikely(qp == 0)) { \
+ EDEB_ERR(4, "qp=%p check failed", qp); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_CHECK_MR(mr) \
+ if (unlikely(mr == 0)) { \
+ EDEB_ERR(4, "mr=%p check failed", mr); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_MR_P(mr) \
+ if (unlikely(mr == 0)) { \
+ EDEB_ERR(4, "mr=%p check failed", mr); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_CHECK_MW(mw) \
+ if (unlikely(mw == 0)) { \
+ EDEB_ERR(4, "mw=%p check failed", mw); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_MW_P(mw) \
+ if (unlikely(mw == 0)) { \
+ EDEB_ERR(4, "mw=%p check failed", mw); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_CHECK_FMR(fmr) \
+ if (unlikely(fmr == 0)) { \
+ EDEB_ERR(4, "fmr=%p check failed", fmr); \
+ return -EFAULT; }
+
+#define EHCA_CHECK_FMR_P(fmr) \
+ if (unlikely(fmr == 0)) { \
+ EDEB_ERR(4, "fmr=%p check failed", fmr); \
+ return ERR_PTR(-EFAULT); }
+
+#define EHCA_REGISTER_PD(device,pd)
+#define EHCA_REGISTER_AV(pd,av)
+#define EHCA_DEREGISTER_PD(PD)
+#define EHCA_DEREGISTER_AV(av)
+#else
+#define EHCA_CHECK_DEVICE_P(device)
+
+#define EHCA_CHECK_PD(pd)
+#define EHCA_REGISTER_PD(device,pd)
+#define EHCA_DEREGISTER_PD(PD)
+#endif
+
+/**
+ * ehca_adr_bad - Handle to be used for adress translation mechanisms,
+ * currently a placeholder.
+ */
+static inline int ehca_adr_bad(void *adr)
+{
+ return !adr;
+}
+
+/**
+ * ehca2ib_return_code - Returns ib return code corresponding to the given
+ * ehca return code.
+ */
+static inline int ehca2ib_return_code(u64 ehca_rc)
+{
+ switch (ehca_rc) {
+ case H_SUCCESS:
+ return 0;
+ case H_BUSY:
+ return -EBUSY;
+ case H_NO_MEM:
+ return -ENOMEM;
+ default:
+ return -EINVAL;
+ }
+}
+
+#endif /* EHCA_TOOLS_H */
^ permalink raw reply
* [PATCH 04/16] ehca: InfiniBand query and multicast functionality
From: Heiko J Schick @ 2006-05-15 17:41 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
drivers/infiniband/hw/ehca/ehca_hca.c | 282 ++++++++++++++++++++++++++++++++
drivers/infiniband/hw/ehca/ehca_mcast.c | 194 ++++++++++++++++++++++
2 files changed, 476 insertions(+)
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_hca.c 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_hca.c 2006-05-02 10:55:26.000000000 +0200
@@ -0,0 +1,282 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * HCA query functions
+ *
+ * Authors: Heiko J Schick <schickhj@de.ibm.com>
+ * Christoph Raisch <raisch@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#undef DEB_PREFIX
+#define DEB_PREFIX "shca"
+
+#include "ehca_tools.h"
+
+#include "hcp_if.h"
+
+int ehca_query_device(struct ib_device *ibdev, struct ib_device_attr *props)
+{
+ int ret = 0;
+ struct ehca_shca *shca;
+ struct hipz_query_hca *rblock;
+
+ EDEB_EN(7, "");
+
+ memset(props, 0, sizeof(struct ib_device_attr));
+ shca = container_of(ibdev, struct ehca_shca, ib_device);
+
+ rblock = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!rblock) {
+ EDEB_ERR(4, "Can't allocate rblock memory.");
+ ret = -ENOMEM;
+ goto query_device0;
+ }
+
+ if (hipz_h_query_hca(shca->ipz_hca_handle, rblock) != H_SUCCESS) {
+ EDEB_ERR(4, "Can't query device properties");
+ ret = -EINVAL;
+ goto query_device1;
+ }
+ props->fw_ver = rblock->hw_ver;
+ props->max_mr_size = rblock->max_mr_size;
+ props->vendor_id = rblock->vendor_id >> 8;
+ props->vendor_part_id = rblock->vendor_part_id >> 16;
+ props->hw_ver = rblock->hw_ver;
+ props->max_qp = min_t(int, rblock->max_qp, INT_MAX);
+ props->max_qp_wr = min_t(int, rblock->max_wqes_wq, INT_MAX);
+ props->max_sge = min_t(int, rblock->max_sge, INT_MAX);
+ props->max_sge_rd = min_t(int, rblock->max_sge_rd, INT_MAX);
+ props->max_cq = min_t(int, rblock->max_cq, INT_MAX);
+ props->max_cqe = min_t(int, rblock->max_cqe, INT_MAX);
+ props->max_mr = min_t(int, rblock->max_mr, INT_MAX);
+ props->max_mw = min_t(int, rblock->max_mw, INT_MAX);
+ props->max_pd = min_t(int, rblock->max_pd, INT_MAX);
+ props->max_ah = min_t(int, rblock->max_ah, INT_MAX);
+ props->max_fmr = min_t(int, rblock->max_mr, INT_MAX);
+ props->max_srq = 0;
+ props->max_srq_wr = 0;
+ props->max_srq_sge = 0;
+ props->max_pkeys = 16;
+ props->local_ca_ack_delay
+ = rblock->local_ca_ack_delay;
+ props->max_raw_ipv6_qp
+ = min_t(int, rblock->max_raw_ipv6_qp, INT_MAX);
+ props->max_raw_ethy_qp
+ = min_t(int, rblock->max_raw_ethy_qp, INT_MAX);
+ props->max_mcast_grp
+ = min_t(int, rblock->max_mcast_grp, INT_MAX);
+ props->max_mcast_qp_attach
+ = min_t(int, rblock->max_mcast_qp_attach, INT_MAX);
+ props->max_total_mcast_qp_attach
+ = min_t(int, rblock->max_total_mcast_qp_attach, INT_MAX);
+
+query_device1:
+ kfree(rblock);
+
+query_device0:
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
+
+int ehca_query_port(struct ib_device *ibdev,
+ u8 port, struct ib_port_attr *props)
+{
+ int ret = 0;
+ struct ehca_shca *shca;
+ struct hipz_query_port *rblock;
+
+ EDEB_EN(7, "port=%x", port);
+
+ memset(props, 0, sizeof(struct ib_port_attr));
+ shca = container_of(ibdev, struct ehca_shca, ib_device);
+
+ rblock = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!rblock) {
+ EDEB_ERR(4, "Can't allocate rblock memory.");
+ ret = -ENOMEM;
+ goto query_port0;
+ }
+
+ if (hipz_h_query_port(shca->ipz_hca_handle, port, rblock) != H_SUCCESS) {
+ EDEB_ERR(4, "Can't query port properties");
+ ret = -EINVAL;
+ goto query_port1;
+ }
+
+ props->state = rblock->state;
+
+ switch (rblock->max_mtu) {
+ case 0x1:
+ props->active_mtu = props->max_mtu = IB_MTU_256;
+ break;
+ case 0x2:
+ props->active_mtu = props->max_mtu = IB_MTU_512;
+ break;
+ case 0x3:
+ props->active_mtu = props->max_mtu = IB_MTU_1024;
+ break;
+ case 0x4:
+ props->active_mtu = props->max_mtu = IB_MTU_2048;
+ break;
+ case 0x5:
+ props->active_mtu = props->max_mtu = IB_MTU_4096;
+ break;
+ default:
+ EDEB_ERR(4, "Unknown MTU size: %x.", rblock->max_mtu);
+ }
+
+ props->gid_tbl_len = rblock->gid_tbl_len;
+ props->max_msg_sz = rblock->max_msg_sz;
+ props->bad_pkey_cntr = rblock->bad_pkey_cntr;
+ props->qkey_viol_cntr = rblock->qkey_viol_cntr;
+ props->pkey_tbl_len = rblock->pkey_tbl_len;
+ props->lid = rblock->lid;
+ props->sm_lid = rblock->sm_lid;
+ props->lmc = rblock->lmc;
+ props->sm_sl = rblock->sm_sl;
+ props->subnet_timeout = rblock->subnet_timeout;
+ props->init_type_reply = rblock->init_type_reply;
+
+ props->active_width = IB_WIDTH_12X;
+ props->active_speed = 0x1;
+
+query_port1:
+ kfree(rblock);
+
+query_port0:
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
+
+int ehca_query_pkey(struct ib_device *ibdev, u8 port, u16 index, u16 *pkey)
+{
+ int ret = 0;
+ struct ehca_shca *shca;
+ struct hipz_query_port *rblock;
+
+ EDEB_EN(7, "port=%x index=%x", port, index);
+
+ if (index > 16) {
+ EDEB_ERR(4, "Invalid index: %x.", index);
+ ret = -EINVAL;
+ goto query_pkey0;
+ }
+
+ shca = container_of(ibdev, struct ehca_shca, ib_device);
+
+ rblock = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!rblock) {
+ EDEB_ERR(4, "Can't allocate rblock memory.");
+ ret = -ENOMEM;
+ goto query_pkey0;
+ }
+
+ if (hipz_h_query_port(shca->ipz_hca_handle, port, rblock) != H_SUCCESS) {
+ EDEB_ERR(4, "Can't query port properties");
+ ret = -EINVAL;
+ goto query_pkey1;
+ }
+
+ memcpy(pkey, &rblock->pkey_entries + index, sizeof(u16));
+
+query_pkey1:
+ kfree(rblock);
+
+query_pkey0:
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
+
+int ehca_query_gid(struct ib_device *ibdev, u8 port,
+ int index, union ib_gid *gid)
+{
+ int ret = 0;
+ struct ehca_shca *shca;
+ struct hipz_query_port *rblock;
+
+ EDEB_EN(7, "port=%x index=%x", port, index);
+
+ if (index > 255) {
+ EDEB_ERR(4, "Invalid index: %x.", index);
+ ret = -EINVAL;
+ goto query_gid0;
+ }
+
+ shca = container_of(ibdev, struct ehca_shca, ib_device);
+
+ rblock = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!rblock) {
+ EDEB_ERR(4, "Can't allocate rblock memory.");
+ ret = -ENOMEM;
+ goto query_gid0;
+ }
+
+ if (hipz_h_query_port(shca->ipz_hca_handle, port, rblock) != H_SUCCESS) {
+ EDEB_ERR(4, "Can't query port properties");
+ ret = -EINVAL;
+ goto query_gid1;
+ }
+
+ memcpy(&gid->raw[0], &rblock->gid_prefix, sizeof(u64));
+ memcpy(&gid->raw[8], &rblock->guid_entries[index], sizeof(u64));
+
+query_gid1:
+ kfree(rblock);
+
+query_gid0:
+ EDEB_EX(7, "ret=%x GID=%lx%lx", ret,
+ *(u64 *) & gid->raw[0],
+ *(u64 *) & gid->raw[8]);
+
+ return ret;
+}
+
+int ehca_modify_port(struct ib_device *ibdev,
+ u8 port, int port_modify_mask,
+ struct ib_port_modify *props)
+{
+ int ret = 0;
+
+ EDEB_EN(7, "port=%x", port);
+
+ /* Not implemented yet. */
+
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_mcast.c 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_mcast.c 2006-05-15 09:54:24.000000000 +0200
@@ -0,0 +1,194 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * mcast functions
+ *
+ * Authors: Khadija Souissi <souissik@de.ibm.com>
+ * Waleri Fomin <fomin@de.ibm.com>
+ * Reinhard Ernst <rernst@de.ibm.com>
+ * Hoang-Nam Nguyen <hnguyen@de.ibm.com>
+ * Heiko J Schick <schickhj@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#define DEB_PREFIX "mcas"
+
+#include <linux/module.h>
+#include <linux/err.h>
+#include "ehca_classes.h"
+#include "ehca_tools.h"
+#include "ehca_qes.h"
+#include "ehca_iverbs.h"
+
+#include "hcp_if.h"
+
+#define MAX_MC_LID 0xFFFE
+#define MIN_MC_LID 0xC000 /* Multicast limits */
+#define EHCA_VALID_MULTICAST_GID(gid) ((gid)[0] == 0xFF)
+#define EHCA_VALID_MULTICAST_LID(lid) (((lid) >= MIN_MC_LID) && ((lid) <= MAX_MC_LID))
+
+int ehca_attach_mcast(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
+{
+ struct ehca_qp *my_qp = NULL;
+ struct ehca_shca *shca = NULL;
+ union ib_gid my_gid;
+ u64 h_ret = H_SUCCESS;
+ int ret = 0;
+
+ EHCA_CHECK_ADR(ibqp);
+ EHCA_CHECK_ADR(gid);
+
+ my_qp = container_of(ibqp, struct ehca_qp, ib_qp);
+
+ EHCA_CHECK_QP(my_qp);
+ if (ibqp->qp_type != IB_QPT_UD) {
+ EDEB_ERR(4, "invalid qp_type %x gid, ret=%x",
+ ibqp->qp_type, EINVAL);
+ return -EINVAL;
+ }
+
+ shca = container_of(ibqp->pd->device, struct ehca_shca, ib_device);
+ EHCA_CHECK_ADR(shca);
+
+ if (!(EHCA_VALID_MULTICAST_GID(gid->raw))) {
+ EDEB_ERR(4, "gid is not valid mulitcast gid ret=%x",
+ EINVAL);
+ return -EINVAL;
+ } else if ((lid < MIN_MC_LID) || (lid > MAX_MC_LID)) {
+ EDEB_ERR(4, "lid=%x is not valid mulitcast lid ret=%x",
+ lid, EINVAL);
+ return -EINVAL;
+ }
+
+ memcpy(&my_gid.raw, gid->raw, sizeof(union ib_gid));
+
+ h_ret = hipz_h_attach_mcqp(shca->ipz_hca_handle,
+ my_qp->ipz_qp_handle,
+ my_qp->galpas.kernel,
+ lid, my_gid.global.subnet_prefix,
+ my_gid.global.interface_id);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4,
+ "ehca_qp=%p qp_num=%x hipz_h_attach_mcqp() failed "
+ "h_ret=%lx", my_qp, ibqp->qp_num, h_ret);
+ }
+ ret = ehca2ib_return_code(h_ret);
+
+ EDEB_EX(7, "mcast attach ret=%x\n"
+ "ehca_qp=%p qp_num=%x lid=%x\n"
+ "my_gid= %x %x %x %x\n"
+ " %x %x %x %x\n"
+ " %x %x %x %x\n"
+ " %x %x %x %x\n",
+ ret, my_qp, ibqp->qp_num, lid,
+ my_gid.raw[0], my_gid.raw[1],
+ my_gid.raw[2], my_gid.raw[3],
+ my_gid.raw[4], my_gid.raw[5],
+ my_gid.raw[6], my_gid.raw[7],
+ my_gid.raw[8], my_gid.raw[9],
+ my_gid.raw[10], my_gid.raw[11],
+ my_gid.raw[12], my_gid.raw[13],
+ my_gid.raw[14], my_gid.raw[15]);
+
+ return ret;
+}
+
+int ehca_detach_mcast(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
+{
+ struct ehca_qp *my_qp = NULL;
+ struct ehca_shca *shca = NULL;
+ union ib_gid my_gid;
+ u64 h_ret = H_SUCCESS;
+ int ret = 0;
+
+ EHCA_CHECK_ADR(ibqp);
+ EHCA_CHECK_ADR(gid);
+
+ my_qp = container_of(ibqp, struct ehca_qp, ib_qp);
+
+ EHCA_CHECK_QP(my_qp);
+ if (ibqp->qp_type != IB_QPT_UD) {
+ EDEB_ERR(4, "invalid qp_type %x gid, ret=%x",
+ ibqp->qp_type, EINVAL);
+ return -EINVAL;
+ }
+
+ shca = container_of(ibqp->pd->device, struct ehca_shca, ib_device);
+ EHCA_CHECK_ADR(shca);
+
+ if (!(EHCA_VALID_MULTICAST_GID(gid->raw))) {
+ EDEB_ERR(4, "gid is not valid mulitcast gid ret=%x",
+ EINVAL);
+ return -EINVAL;
+ } else if ((lid < MIN_MC_LID) || (lid > MAX_MC_LID)) {
+ EDEB_ERR(4, "lid=%x is not valid mulitcast lid ret=%x",
+ lid, EINVAL);
+ return -EINVAL;
+ }
+
+ EDEB_EN(7, "dgid=%p qp_numl=%x lid=%x",
+ gid, ibqp->qp_num, lid);
+
+ memcpy(&my_gid.raw, gid->raw, sizeof(union ib_gid));
+
+ h_ret = hipz_h_detach_mcqp(shca->ipz_hca_handle,
+ my_qp->ipz_qp_handle,
+ my_qp->galpas.kernel,
+ lid, my_gid.global.subnet_prefix,
+ my_gid.global.interface_id);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4,
+ "ehca_qp=%p qp_num=%x hipz_h_detach_mcqp() failed "
+ "h_ret=%lx", my_qp, ibqp->qp_num, h_ret);
+ }
+ ret = ehca2ib_return_code(h_ret);
+
+ EDEB_EX(7, "mcast detach ret=%x\n"
+ "ehca_qp=%p qp_num=%x lid=%x\n"
+ "my_gid= %x %x %x %x\n"
+ " %x %x %x %x\n"
+ " %x %x %x %x\n"
+ " %x %x %x %x\n",
+ ret, my_qp, ibqp->qp_num, lid,
+ my_gid.raw[0], my_gid.raw[1],
+ my_gid.raw[2], my_gid.raw[3],
+ my_gid.raw[4], my_gid.raw[5],
+ my_gid.raw[6], my_gid.raw[7],
+ my_gid.raw[8], my_gid.raw[9],
+ my_gid.raw[10], my_gid.raw[11],
+ my_gid.raw[12], my_gid.raw[13],
+ my_gid.raw[14], my_gid.raw[15]);
+
+ return ret;
+}
^ permalink raw reply
* [PATCH 03/16] ehca: userspace support
From: Heiko J Schick @ 2006-05-15 17:41 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
drivers/infiniband/hw/ehca/ehca_uverbs.c | 391 +++++++++++++++++++++++++++++++
1 file changed, 391 insertions(+)
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_uverbs.c 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_uverbs.c 2006-05-12 12:31:52.000000000 +0200
@@ -0,0 +1,391 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * userspace support verbs
+ *
+ * Authors: Christoph Raisch <raisch@de.ibm.com>
+ * Hoang-Nam Nguyen <hnguyen@de.ibm.com>
+ * Heiko J Schick <schickhj@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#undef DEB_PREFIX
+#define DEB_PREFIX "uver"
+
+#include <asm/current.h>
+
+#include "ehca_classes.h"
+#include "ehca_iverbs.h"
+#include "ehca_mrmw.h"
+#include "ehca_tools.h"
+#include "hcp_if.h"
+
+struct ib_ucontext *ehca_alloc_ucontext(struct ib_device *device,
+ struct ib_udata *udata)
+{
+ struct ehca_ucontext *my_context = NULL;
+
+ EHCA_CHECK_ADR_P(device);
+ EDEB_EN(7, "device=%p name=%s", device, device->name);
+
+ my_context = kzalloc(sizeof *my_context, GFP_KERNEL);
+ if (!my_context) {
+ EDEB_ERR(4, "Out of memory device=%p", device);
+ return ERR_PTR(-ENOMEM);
+ }
+
+ EDEB_EX(7, "device=%p ucontext=%p", device, my_context);
+
+ return &my_context->ib_ucontext;
+}
+
+int ehca_dealloc_ucontext(struct ib_ucontext *context)
+{
+ struct ehca_ucontext *my_context = NULL;
+ EHCA_CHECK_ADR(context);
+ EDEB_EN(7, "ucontext=%p", context);
+ my_context = container_of(context, struct ehca_ucontext, ib_ucontext);
+ kfree(my_context);
+ EDEB_EN(7, "ucontext=%p", context);
+ return 0;
+}
+
+struct page *ehca_nopage(struct vm_area_struct *vma,
+ unsigned long address, int *type)
+{
+ struct page *mypage = NULL;
+ u64 fileoffset = vma->vm_pgoff << PAGE_SHIFT;
+ u32 idr_handle = fileoffset >> 32;
+ u32 q_type = (fileoffset >> 28) & 0xF; /* CQ, QP,... */
+ u32 rsrc_type = (fileoffset >> 24) & 0xF; /* sq,rq,cmnd_window */
+ u32 cur_pid = current->tgid;
+ unsigned long flags;
+
+ EDEB_EN(7, "vm_start=%lx vm_end=%lx vm_page_prot=%lx vm_fileoff=%lx "
+ "address=%lx",
+ vma->vm_start, vma->vm_end, vma->vm_page_prot, fileoffset,
+ address);
+
+ if (q_type == 1) { /* CQ */
+ struct ehca_cq *cq = NULL;
+ u64 offset;
+ void *vaddr = NULL;
+
+ spin_lock_irqsave(&ehca_cq_idr_lock, flags);
+ cq = idr_find(&ehca_cq_idr, idr_handle);
+ spin_unlock_irqrestore(&ehca_cq_idr_lock, flags);
+
+ if (cq->ownpid != cur_pid) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, cq->ownpid);
+ return NOPAGE_SIGBUS;
+ }
+
+ /* make sure this mmap really belongs to the authorized user */
+ if (!cq) {
+ EDEB_ERR(4, "cq is NULL ret=NOPAGE_SIGBUS");
+ return NOPAGE_SIGBUS;
+ }
+ if (rsrc_type == 2) {
+ EDEB(6, "cq=%p cq queuearea", cq);
+ offset = address - vma->vm_start;
+ vaddr = ipz_qeit_calc(&cq->ipz_queue, offset);
+ EDEB(6, "offset=%lx vaddr=%p", offset, vaddr);
+ mypage = virt_to_page(vaddr);
+ }
+ } else if (q_type == 2) { /* QP */
+ struct ehca_qp *qp = NULL;
+ struct ehca_pd *pd = NULL;
+ u64 offset;
+ void *vaddr = NULL;
+
+ spin_lock_irqsave(&ehca_qp_idr_lock, flags);
+ qp = idr_find(&ehca_qp_idr, idr_handle);
+ spin_unlock_irqrestore(&ehca_qp_idr_lock, flags);
+
+
+ pd = container_of(qp->ib_qp.pd, struct ehca_pd, ib_pd);
+ if (pd->ownpid != cur_pid) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, pd->ownpid);
+ return NOPAGE_SIGBUS;
+ }
+
+ /* make sure this mmap really belongs to the authorized user */
+ if (!qp) {
+ EDEB_ERR(4, "qp is NULL ret=NOPAGE_SIGBUS");
+ return NOPAGE_SIGBUS;
+ }
+ if (rsrc_type == 2) { /* rqueue */
+ EDEB(6, "qp=%p qp rqueuearea", qp);
+ offset = address - vma->vm_start;
+ vaddr = ipz_qeit_calc(&qp->ipz_rqueue, offset);
+ EDEB(6, "offset=%lx vaddr=%p", offset, vaddr);
+ mypage = virt_to_page(vaddr);
+ } else if (rsrc_type == 3) { /* squeue */
+ EDEB(6, "qp=%p qp squeuearea", qp);
+ offset = address - vma->vm_start;
+ vaddr = ipz_qeit_calc(&qp->ipz_squeue, offset);
+ EDEB(6, "offset=%lx vaddr=%p", offset, vaddr);
+ mypage = virt_to_page(vaddr);
+ }
+ }
+
+ if (!mypage) {
+ EDEB_ERR(4, "Invalid page adr==NULL ret=NOPAGE_SIGBUS");
+ return NOPAGE_SIGBUS;
+ }
+ get_page(mypage);
+ EDEB_EX(7, "page adr=%p", mypage);
+ return mypage;
+}
+
+static struct vm_operations_struct ehcau_vm_ops = {
+ .nopage = ehca_nopage,
+};
+
+int ehca_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
+{
+ u64 fileoffset = vma->vm_pgoff << PAGE_SHIFT;
+ u32 idr_handle = fileoffset >> 32;
+ u32 q_type = (fileoffset >> 28) & 0xF; /* CQ, QP,... */
+ u32 rsrc_type = (fileoffset >> 24) & 0xF; /* sq,rq,cmnd_window */
+ u32 ret = -EFAULT; /* assume the worst */
+ u64 vsize = 0; /* must be calculated/set below */
+ u64 physical = 0; /* must be calculated/set below */
+ u32 cur_pid = current->tgid;
+ unsigned long flags;
+
+ EDEB_EN(7, "vm_start=%lx vm_end=%lx vm_page_prot=%lx vm_fileoff=%lx",
+ vma->vm_start, vma->vm_end, vma->vm_page_prot, fileoffset);
+
+ if (q_type == 1) { /* CQ */
+ struct ehca_cq *cq;
+
+ spin_lock_irqsave(&ehca_cq_idr_lock, flags);
+ cq = idr_find(&ehca_cq_idr, idr_handle);
+ spin_unlock_irqrestore(&ehca_cq_idr_lock, flags);
+
+ if (cq->ownpid != cur_pid) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, cq->ownpid);
+ return -ENOMEM;
+ }
+
+ /* make sure this mmap really belongs to the authorized user */
+ if (!cq)
+ return -EINVAL;
+ if (!cq->ib_cq.uobject)
+ return -EINVAL;
+ if (cq->ib_cq.uobject->context != context)
+ return -EINVAL;
+ if (rsrc_type == 1) { /* galpa fw handle */
+ EDEB(6, "cq=%p cq triggerarea", cq);
+ vma->vm_flags |= VM_RESERVED;
+ vsize = vma->vm_end - vma->vm_start;
+ if (vsize != EHCA_PAGESIZE) {
+ EDEB_ERR(4, "invalid vsize=%lx",
+ vma->vm_end - vma->vm_start);
+ ret = -EINVAL;
+ goto mmap_exit0;
+ }
+
+ physical = cq->galpas.user.fw_handle;
+ vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+ vma->vm_flags |= VM_IO | VM_RESERVED;
+
+ EDEB(6, "vsize=%lx physical=%lx", vsize, physical);
+ ret = remap_pfn_range(vma, vma->vm_start,
+ physical >> PAGE_SHIFT, vsize,
+ vma->vm_page_prot);
+ if (ret) {
+ EDEB_ERR(4, "remap_pfn_range() failed ret=%x",
+ ret);
+ ret = -ENOMEM;
+ }
+ goto mmap_exit0;
+ } else if (rsrc_type == 2) { /* cq queue_addr */
+ EDEB(6, "cq=%p cq q_addr", cq);
+ /* vma->vm_page_prot =
+ * pgprot_noncached(vma->vm_page_prot); */
+ vma->vm_flags |= VM_RESERVED;
+ vma->vm_ops = &ehcau_vm_ops;
+ ret = 0;
+ goto mmap_exit0;
+ } else {
+ EDEB_ERR(6, "bad resource type %x", rsrc_type);
+ ret = -EINVAL;
+ goto mmap_exit0;
+ }
+ } else if (q_type == 2) { /* QP */
+ struct ehca_qp *qp = NULL;
+ struct ehca_pd *pd = NULL;
+
+ spin_lock_irqsave(&ehca_qp_idr_lock, flags);
+ qp = idr_find(&ehca_qp_idr, idr_handle);
+ spin_unlock_irqrestore(&ehca_qp_idr_lock, flags);
+
+ pd = container_of(qp->ib_qp.pd, struct ehca_pd, ib_pd);
+ if (pd->ownpid != cur_pid) {
+ EDEB_ERR(4, "Invalid caller pid=%x ownpid=%x",
+ cur_pid, pd->ownpid);
+ return -ENOMEM;
+ }
+
+ /* make sure this mmap really belongs to the authorized user */
+ if (!qp || !qp->ib_qp.uobject ||
+ qp->ib_qp.uobject->context != context) {
+ EDEB(6, "qp=%p, uobject=%p, context=%p",
+ qp, qp->ib_qp.uobject, qp->ib_qp.uobject->context);
+ ret = -EINVAL;
+ goto mmap_exit0;
+ }
+ if (rsrc_type == 1) { /* galpa fw handle */
+ EDEB(6, "qp=%p qp triggerarea", qp);
+ vma->vm_flags |= VM_RESERVED;
+ vsize = vma->vm_end - vma->vm_start;
+ if (vsize != EHCA_PAGESIZE) {
+ EDEB_ERR(4, "invalid vsize=%lx",
+ vma->vm_end - vma->vm_start);
+ ret = -EINVAL;
+ goto mmap_exit0;
+ }
+
+ physical = qp->galpas.user.fw_handle;
+ vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+ vma->vm_flags |= VM_IO | VM_RESERVED;
+
+ EDEB(6, "vsize=%lx physical=%lx", vsize, physical);
+ ret = remap_pfn_range(vma, vma->vm_start,
+ physical >> PAGE_SHIFT, vsize,
+ vma->vm_page_prot);
+ if (ret) {
+ EDEB_ERR(4, "remap_pfn_range() failed ret=%x",
+ ret);
+ ret = -ENOMEM;
+ }
+ goto mmap_exit0;
+ } else if (rsrc_type == 2) { /* qp rqueue_addr */
+ EDEB(6, "qp=%p qp rqueue_addr", qp);
+ vma->vm_flags |= VM_RESERVED;
+ vma->vm_ops = &ehcau_vm_ops;
+ ret = 0;
+ goto mmap_exit0;
+ } else if (rsrc_type == 3) { /* qp squeue_addr */
+ EDEB(6, "qp=%p qp squeue_addr", qp);
+ vma->vm_flags |= VM_RESERVED;
+ vma->vm_ops = &ehcau_vm_ops;
+ ret = 0;
+ goto mmap_exit0;
+ } else {
+ EDEB_ERR(4, "bad resource type %x", rsrc_type);
+ ret = -EINVAL;
+ goto mmap_exit0;
+ }
+ } else {
+ EDEB_ERR(4, "bad queue type %x", q_type);
+ ret = -EINVAL;
+ goto mmap_exit0;
+ }
+
+mmap_exit0:
+ EDEB_EX(7, "ret=%x", ret);
+ return ret;
+}
+
+int ehca_mmap_nopage(u64 foffset, u64 length, void ** mapped,
+ struct vm_area_struct ** vma)
+{
+ EDEB_EN(7, "foffset=%lx length=%lx", foffset, length);
+ down_write(¤t->mm->mmap_sem);
+ *mapped = (void*)
+ do_mmap(NULL,0,
+ length,
+ PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS,
+ foffset);
+ up_write(¤t->mm->mmap_sem);
+ if (*mapped) {
+ *vma = find_vma(current->mm,(u64)*mapped);
+ if (*vma) {
+ (*vma)->vm_flags |= VM_RESERVED;
+ (*vma)->vm_ops = &ehcau_vm_ops;
+ } else
+ EDEB_ERR(4, "couldn't find queue vma queue=%p", *mapped);
+ } else
+ EDEB_ERR(4, "couldn't create mmap length=%lx", length);
+ EDEB_EX(7, "mapped=%p", *mapped);
+ return 0;
+}
+
+int ehca_mmap_register(u64 physical, void ** mapped,
+ struct vm_area_struct ** vma)
+{
+ int ret = 0;
+ unsigned long vsize;
+ /* ehca hw supports only 4k page */
+ ehca_mmap_nopage(0, EHCA_PAGESIZE, mapped, vma);
+ (*vma)->vm_flags |= VM_RESERVED;
+ vsize = (*vma)->vm_end - (*vma)->vm_start;
+ if (vsize != EHCA_PAGESIZE) {
+ EDEB_ERR(4, "invalid vsize=%lx",
+ (*vma)->vm_end - (*vma)->vm_start);
+ ret = -EINVAL;
+ return ret;
+ }
+
+ (*vma)->vm_page_prot = pgprot_noncached((*vma)->vm_page_prot);
+ (*vma)->vm_flags |= VM_IO | VM_RESERVED;
+
+ EDEB(6, "vsize=%lx physical=%lx", vsize, physical);
+ ret = remap_pfn_range((*vma), (*vma)->vm_start,
+ physical >> PAGE_SHIFT, vsize,
+ (*vma)->vm_page_prot);
+ if (ret) {
+ EDEB_ERR(4, "remap_pfn_range() failed ret=%x", ret);
+ ret = -ENOMEM;
+ }
+ return ret;
+
+}
+
+int ehca_munmap(unsigned long addr, size_t len) {
+ int ret = 0;
+ struct mm_struct *mm = current->mm;
+ if (mm) {
+ down_write(&mm->mmap_sem);
+ ret = do_munmap(mm, addr, len);
+ up_write(&mm->mmap_sem);
+ }
+ return ret;
+}
^ permalink raw reply
* [PATCH 02/16] ehca: structure definitions
From: Heiko J Schick @ 2006-05-15 17:41 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
drivers/infiniband/hw/ehca/ehca_classes.h | 350 ++++++++++++++++++++++
drivers/infiniband/hw/ehca/ehca_classes_pSeries.h | 251 +++++++++++++++
2 files changed, 601 insertions(+)
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_classes.h 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_classes.h 2006-05-12 12:48:21.000000000 +0200
@@ -0,0 +1,350 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * Struct definition for eHCA internal structures
+ *
+ * Authors: Heiko J Schick <schickhj@de.ibm.com>
+ * Christoph Raisch <raisch@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef __EHCA_CLASSES_H__
+#define __EHCA_CLASSES_H__
+
+#include "ehca_classes.h"
+#include "ipz_pt_fn.h"
+
+struct ehca_module;
+struct ehca_qp;
+struct ehca_cq;
+struct ehca_eq;
+struct ehca_mr;
+struct ehca_mw;
+struct ehca_pd;
+struct ehca_av;
+
+#ifdef CONFIG_PPC64
+#include "ehca_classes_pSeries.h"
+#endif
+
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_user_verbs.h>
+
+#include "ehca_irq.h"
+
+struct ehca_module {
+ struct list_head shca_list;
+ spinlock_t shca_lock;
+ struct timer_list timer;
+ kmem_cache_t *cache_pd;
+ kmem_cache_t *cache_cq;
+ kmem_cache_t *cache_qp;
+ kmem_cache_t *cache_av;
+ kmem_cache_t *cache_mr;
+ kmem_cache_t *cache_mw;
+ struct ehca_pfmodule pf;
+};
+
+struct ehca_eq {
+ u32 length;
+ struct ipz_queue ipz_queue;
+ struct ipz_eq_handle ipz_eq_handle;
+ struct work_struct work;
+ struct h_galpas galpas;
+ int is_initialized;
+ struct ehca_pfeq pf;
+ spinlock_t spinlock;
+ struct tasklet_struct interrupt_task;
+ u32 ist;
+};
+
+struct ehca_sport {
+ struct ib_cq *ibcq_aqp1;
+ struct ib_qp *ibqp_aqp1;
+ enum ib_rate rate;
+ enum ib_port_state port_state;
+};
+
+struct ehca_shca {
+ struct ib_device ib_device;
+ struct ibmebus_dev *ibmebus_dev;
+ u8 num_ports;
+ int hw_level;
+ struct list_head shca_list;
+ struct ipz_adapter_handle ipz_hca_handle;
+ struct ehca_sport sport[2];
+ struct ehca_eq eq;
+ struct ehca_eq neq;
+ struct ehca_mr *maxmr;
+ struct ehca_pd *pd;
+ struct ehca_pfshca pf;
+ struct h_galpas galpas;
+};
+
+struct ehca_pd {
+ struct ib_pd ib_pd;
+ struct ipz_pd fw_pd;
+ struct ehca_pfpd pf;
+ u32 ownpid;
+};
+
+struct ehca_qp {
+ struct ib_qp ib_qp;
+ u32 qp_type;
+ struct ipz_queue ipz_squeue;
+ struct ipz_queue ipz_rqueue;
+ struct h_galpas galpas;
+ u32 qkey;
+ u32 real_qp_num;
+ u32 token;
+ spinlock_t spinlock_s;
+ spinlock_t spinlock_r;
+ u32 sq_max_inline_data_size;
+ struct ipz_qp_handle ipz_qp_handle;
+ struct ehca_pfqp pf;
+ struct ib_qp_init_attr init_attr;
+ u64 uspace_squeue;
+ u64 uspace_rqueue;
+ u64 uspace_fwh;
+ struct ehca_cq *send_cq;
+ struct ehca_cq *recv_cq;
+ unsigned int sqerr_purgeflag;
+ struct hlist_node list_entries;
+};
+
+/* must be power of 2 */
+#define QP_HASHTAB_LEN 8
+
+struct ehca_cq {
+ struct ib_cq ib_cq;
+ struct ipz_queue ipz_queue;
+ struct h_galpas galpas;
+ spinlock_t spinlock;
+ u32 cq_number;
+ u32 token;
+ u32 nr_of_entries;
+ struct ipz_cq_handle ipz_cq_handle;
+ struct ehca_pfcq pf;
+ spinlock_t cb_lock;
+ u64 uspace_queue;
+ u64 uspace_fwh;
+ struct hlist_head qp_hashtab[QP_HASHTAB_LEN];
+ struct list_head entry;
+ u32 nr_callbacks;
+ spinlock_t task_lock;
+ u32 ownpid;
+};
+
+enum ehca_mr_flag {
+ EHCA_MR_FLAG_FMR = 0x80000000, /* FMR, created with ehca_alloc_fmr */
+ EHCA_MR_FLAG_MAXMR = 0x40000000, /* max-MR */
+};
+
+struct ehca_mr {
+ union {
+ struct ib_mr ib_mr; /* must always be first in ehca_mr */
+ struct ib_fmr ib_fmr; /* must always be first in ehca_mr */
+ } ib;
+ spinlock_t mrlock;
+
+ enum ehca_mr_flag flags;
+ u32 num_pages; /* number of MR pages */
+ u32 num_4k; /* number of 4k "page" portions to form MR */
+ int acl; /* ACL (stored here for usage in reregister) */
+ u64 *start; /* virtual start address (stored here for */
+ /* usage in reregister) */
+ u64 size; /* size (stored here for usage in reregister) */
+ u32 fmr_page_size; /* page size for FMR */
+ u32 fmr_max_pages; /* max pages for FMR */
+ u32 fmr_max_maps; /* max outstanding maps for FMR */
+ u32 fmr_map_cnt; /* map counter for FMR */
+ /* fw specific data */
+ struct ipz_mrmw_handle ipz_mr_handle; /* MR handle for h-calls */
+ struct h_galpas galpas;
+ /* data for userspace bridge */
+ u32 nr_of_pages;
+ void *pagearray;
+
+ struct ehca_pfmr pf; /* platform specific part of MR */
+};
+
+struct ehca_mw {
+ struct ib_mw ib_mw; /* gen2 mw, must always be first in ehca_mw */
+ spinlock_t mwlock;
+
+ u8 never_bound; /* indication MW was never bound */
+ struct ipz_mrmw_handle ipz_mw_handle; /* MW handle for h-calls */
+ struct h_galpas galpas;
+
+ struct ehca_pfmw pf; /* platform specific part of MW */
+};
+
+enum ehca_mr_pgi_type {
+ EHCA_MR_PGI_PHYS = 1, /* type of ehca_reg_phys_mr,
+ * ehca_rereg_phys_mr,
+ * ehca_reg_internal_maxmr */
+ EHCA_MR_PGI_USER = 2, /* type of ehca_reg_user_mr */
+ EHCA_MR_PGI_FMR = 3 /* type of ehca_map_phys_fmr */
+};
+
+struct ehca_mr_pginfo {
+ enum ehca_mr_pgi_type type;
+ u64 num_pages;
+ u64 page_cnt;
+ u64 num_4k; /* number of 4k "page" portions */
+ u64 page_4k_cnt; /* counter for 4k "page" portions */
+ u64 next_4k; /* next 4k "page" portion in buffer/chunk/listelem */
+
+ /* type EHCA_MR_PGI_PHYS section */
+ int num_phys_buf;
+ struct ib_phys_buf *phys_buf_array;
+ u64 next_buf;
+
+ /* type EHCA_MR_PGI_USER section */
+ struct ib_umem *region;
+ struct ib_umem_chunk *next_chunk;
+ u64 next_nmap;
+
+ /* type EHCA_MR_PGI_FMR section */
+ u64 *page_list;
+ u64 next_listelem;
+ /* next_4k also used within EHCA_MR_PGI_FMR */
+};
+
+/* output parameters for MR/FMR hipz calls */
+struct ehca_mr_hipzout_parms {
+ struct ipz_mrmw_handle handle;
+ u32 lkey;
+ u32 rkey;
+ u64 len;
+ u64 vaddr;
+ u32 acl;
+};
+
+/* output parameters for MW hipz calls */
+struct ehca_mw_hipzout_parms {
+ struct ipz_mrmw_handle handle;
+ u32 rkey;
+};
+
+struct ehca_av {
+ struct ib_ah ib_ah;
+ struct ehca_ud_av av;
+};
+
+struct ehca_ucontext {
+ struct ib_ucontext ib_ucontext;
+};
+
+struct ehca_module *ehca_module_new(void);
+
+int ehca_module_delete(struct ehca_module *me);
+
+int ehca_eq_ctor(struct ehca_eq *eq);
+
+int ehca_eq_dtor(struct ehca_eq *eq);
+
+struct ehca_shca *ehca_shca_new(void);
+
+int ehca_shca_delete(struct ehca_shca *me);
+
+struct ehca_sport *ehca_sport_new(struct ehca_shca *anchor);
+
+extern spinlock_t ehca_qp_idr_lock;
+extern spinlock_t ehca_cq_idr_lock;
+extern struct idr ehca_qp_idr;
+extern struct idr ehca_cq_idr;
+
+struct ipzu_queue_resp {
+ u64 queue; /* points to first queue entry */
+ u32 qe_size; /* queue entry size */
+ u32 act_nr_of_sg;
+ u32 queue_length; /* queue length allocated in bytes */
+ u32 pagesize;
+ u32 toggle_state;
+ u32 dummy; /* padding for 8 byte alignment */
+};
+
+struct ehca_create_cq_resp {
+ u32 cq_number;
+ u32 token;
+ struct ipzu_queue_resp ipz_queue;
+ struct h_galpas galpas;
+};
+
+struct ehca_create_qp_resp {
+ u32 qp_num;
+ u32 token;
+ u32 qp_type;
+ u32 qkey;
+ /* qp_num assigned by ehca: sqp0/1 may have got different numbers */
+ u32 real_qp_num;
+ u32 dummy; /* padding for 8 byte alignment */
+ struct ipzu_queue_resp ipz_squeue;
+ struct ipzu_queue_resp ipz_rqueue;
+ struct h_galpas galpas;
+};
+
+struct ehca_alloc_cq_parms {
+ u32 nr_cqe;
+ u32 act_nr_of_entries;
+ u32 act_pages;
+ struct ipz_eq_handle eq_handle;
+};
+
+struct ehca_alloc_qp_parms {
+ int servicetype;
+ int sigtype;
+ int daqp_ctrl;
+ int max_send_sge;
+ int max_recv_sge;
+ int ud_av_l_key_ctl;
+
+ u16 act_nr_send_wqes;
+ u16 act_nr_recv_wqes;
+ u8 act_nr_recv_sges;
+ u8 act_nr_send_sges;
+
+ u32 nr_rq_pages;
+ u32 nr_sq_pages;
+
+ struct ipz_eq_handle ipz_eq_handle;
+ struct ipz_pd pd;
+};
+
+int ehca_cq_assign_qp(struct ehca_cq *cq, struct ehca_qp *qp);
+int ehca_cq_unassign_qp(struct ehca_cq *cq, unsigned int qp_num);
+struct ehca_qp* ehca_cq_get_qp(struct ehca_cq *cq, int qp_num);
+
+#endif
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_classes_pSeries.h 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_classes_pSeries.h 2006-04-28 14:20:07.000000000 +0200
@@ -0,0 +1,251 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * pSeries interface definitions
+ *
+ * Authors: Waleri Fomin <fomin@de.ibm.com>
+ * Christoph Raisch <raisch@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef __EHCA_CLASSES_PSERIES_H__
+#define __EHCA_CLASSES_PSERIES_H__
+
+#include "hcp_phyp.h"
+#include "ipz_pt_fn.h"
+
+
+struct ehca_pfmodule {
+};
+
+struct ehca_pfshca {
+};
+
+struct ehca_pfqp {
+ struct ipz_qpt sqpt;
+ struct ipz_qpt rqpt;
+};
+
+struct ehca_pfcq {
+ struct ipz_qpt qpt;
+ u32 cqnr;
+};
+
+struct ehca_pfeq {
+ struct ipz_qpt qpt;
+ struct h_galpa galpa;
+ u32 eqnr;
+};
+
+struct ehca_pfpd {
+};
+
+struct ehca_pfmr {
+};
+
+struct ehca_pfmw {
+};
+
+struct ipz_adapter_handle {
+ u64 handle;
+};
+
+struct ipz_cq_handle {
+ u64 handle;
+};
+
+struct ipz_eq_handle {
+ u64 handle;
+};
+
+struct ipz_qp_handle {
+ u64 handle;
+};
+struct ipz_mrmw_handle {
+ u64 handle;
+};
+
+struct ipz_pd {
+ u32 value;
+};
+
+struct hcp_modify_qp_control_block {
+ u32 qkey; /* 00 */
+ u32 rdd; /* reliable datagram domain */
+ u32 send_psn; /* 02 */
+ u32 receive_psn; /* 03 */
+ u32 prim_phys_port; /* 04 */
+ u32 alt_phys_port; /* 05 */
+ u32 prim_p_key_idx; /* 06 */
+ u32 alt_p_key_idx; /* 07 */
+ u32 rdma_atomic_ctrl; /* 08 */
+ u32 qp_state; /* 09 */
+ u32 reserved_10; /* 10 */
+ u32 rdma_nr_atomic_resp_res; /* 11 */
+ u32 path_migration_state; /* 12 */
+ u32 rdma_atomic_outst_dest_qp; /* 13 */
+ u32 dest_qp_nr; /* 14 */
+ u32 min_rnr_nak_timer_field; /* 15 */
+ u32 service_level; /* 16 */
+ u32 send_grh_flag; /* 17 */
+ u32 retry_count; /* 18 */
+ u32 timeout; /* 19 */
+ u32 path_mtu; /* 20 */
+ u32 max_static_rate; /* 21 */
+ u32 dlid; /* 22 */
+ u32 rnr_retry_count; /* 23 */
+ u32 source_path_bits; /* 24 */
+ u32 traffic_class; /* 25 */
+ u32 hop_limit; /* 26 */
+ u32 source_gid_idx; /* 27 */
+ u32 flow_label; /* 28 */
+ u32 reserved_29; /* 29 */
+ union { /* 30 */
+ u64 dw[2];
+ u8 byte[16];
+ } dest_gid;
+ u32 service_level_al; /* 34 */
+ u32 send_grh_flag_al; /* 35 */
+ u32 retry_count_al; /* 36 */
+ u32 timeout_al; /* 37 */
+ u32 max_static_rate_al; /* 38 */
+ u32 dlid_al; /* 39 */
+ u32 rnr_retry_count_al; /* 40 */
+ u32 source_path_bits_al; /* 41 */
+ u32 traffic_class_al; /* 42 */
+ u32 hop_limit_al; /* 43 */
+ u32 source_gid_idx_al; /* 44 */
+ u32 flow_label_al; /* 45 */
+ u32 reserved_46; /* 46 */
+ u32 reserved_47; /* 47 */
+ union { /* 48 */
+ u64 dw[2];
+ u8 byte[16];
+ } dest_gid_al;
+ u32 max_nr_outst_send_wr; /* 52 */
+ u32 max_nr_outst_recv_wr; /* 53 */
+ u32 disable_ete_credit_check; /* 54 */
+ u32 qp_number; /* 55 */
+ u64 send_queue_handle; /* 56 */
+ u64 recv_queue_handle; /* 58 */
+ u32 actual_nr_sges_in_sq_wqe; /* 60 */
+ u32 actual_nr_sges_in_rq_wqe; /* 61 */
+ u32 qp_enable; /* 62 */
+ u32 curr_srq_limit; /* 63 */
+ u64 qp_aff_asyn_ev_log_reg; /* 64 */
+ u64 shared_rq_hndl; /* 66 */
+ u64 trigg_doorbell_qp_hndl; /* 68 */
+ u32 reserved_70_127[58]; /* 70 */
+};
+
+#define MQPCB_MASK_QKEY EHCA_BMASK_IBM(0,0)
+#define MQPCB_MASK_SEND_PSN EHCA_BMASK_IBM(2,2)
+#define MQPCB_MASK_RECEIVE_PSN EHCA_BMASK_IBM(3,3)
+#define MQPCB_MASK_PRIM_PHYS_PORT EHCA_BMASK_IBM(4,4)
+#define MQPCB_PRIM_PHYS_PORT EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_ALT_PHYS_PORT EHCA_BMASK_IBM(5,5)
+#define MQPCB_MASK_PRIM_P_KEY_IDX EHCA_BMASK_IBM(6,6)
+#define MQPCB_PRIM_P_KEY_IDX EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_ALT_P_KEY_IDX EHCA_BMASK_IBM(7,7)
+#define MQPCB_MASK_RDMA_ATOMIC_CTRL EHCA_BMASK_IBM(8,8)
+#define MQPCB_MASK_QP_STATE EHCA_BMASK_IBM(9,9)
+#define MQPCB_QP_STATE EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_RDMA_NR_ATOMIC_RESP_RES EHCA_BMASK_IBM(11,11)
+#define MQPCB_MASK_PATH_MIGRATION_STATE EHCA_BMASK_IBM(12,12)
+#define MQPCB_MASK_RDMA_ATOMIC_OUTST_DEST_QP EHCA_BMASK_IBM(13,13)
+#define MQPCB_MASK_DEST_QP_NR EHCA_BMASK_IBM(14,14)
+#define MQPCB_MASK_MIN_RNR_NAK_TIMER_FIELD EHCA_BMASK_IBM(15,15)
+#define MQPCB_MASK_SERVICE_LEVEL EHCA_BMASK_IBM(16,16)
+#define MQPCB_MASK_SEND_GRH_FLAG EHCA_BMASK_IBM(17,17)
+#define MQPCB_MASK_RETRY_COUNT EHCA_BMASK_IBM(18,18)
+#define MQPCB_MASK_TIMEOUT EHCA_BMASK_IBM(19,19)
+#define MQPCB_MASK_PATH_MTU EHCA_BMASK_IBM(20,20)
+#define MQPCB_PATH_MTU EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_MAX_STATIC_RATE EHCA_BMASK_IBM(21,21)
+#define MQPCB_MAX_STATIC_RATE EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_DLID EHCA_BMASK_IBM(22,22)
+#define MQPCB_DLID EHCA_BMASK_IBM(16,31)
+#define MQPCB_MASK_RNR_RETRY_COUNT EHCA_BMASK_IBM(23,23)
+#define MQPCB_RNR_RETRY_COUNT EHCA_BMASK_IBM(29,31)
+#define MQPCB_MASK_SOURCE_PATH_BITS EHCA_BMASK_IBM(24,24)
+#define MQPCB_SOURCE_PATH_BITS EHCA_BMASK_IBM(25,31)
+#define MQPCB_MASK_TRAFFIC_CLASS EHCA_BMASK_IBM(25,25)
+#define MQPCB_TRAFFIC_CLASS EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_HOP_LIMIT EHCA_BMASK_IBM(26,26)
+#define MQPCB_HOP_LIMIT EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_SOURCE_GID_IDX EHCA_BMASK_IBM(27,27)
+#define MQPCB_SOURCE_GID_IDX EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_FLOW_LABEL EHCA_BMASK_IBM(28,28)
+#define MQPCB_FLOW_LABEL EHCA_BMASK_IBM(12,31)
+#define MQPCB_MASK_DEST_GID EHCA_BMASK_IBM(30,30)
+#define MQPCB_MASK_SERVICE_LEVEL_AL EHCA_BMASK_IBM(31,31)
+#define MQPCB_SERVICE_LEVEL_AL EHCA_BMASK_IBM(28,31)
+#define MQPCB_MASK_SEND_GRH_FLAG_AL EHCA_BMASK_IBM(32,32)
+#define MQPCB_SEND_GRH_FLAG_AL EHCA_BMASK_IBM(31,31)
+#define MQPCB_MASK_RETRY_COUNT_AL EHCA_BMASK_IBM(33,33)
+#define MQPCB_RETRY_COUNT_AL EHCA_BMASK_IBM(29,31)
+#define MQPCB_MASK_TIMEOUT_AL EHCA_BMASK_IBM(34,34)
+#define MQPCB_TIMEOUT_AL EHCA_BMASK_IBM(27,31)
+#define MQPCB_MASK_MAX_STATIC_RATE_AL EHCA_BMASK_IBM(35,35)
+#define MQPCB_MAX_STATIC_RATE_AL EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_DLID_AL EHCA_BMASK_IBM(36,36)
+#define MQPCB_DLID_AL EHCA_BMASK_IBM(16,31)
+#define MQPCB_MASK_RNR_RETRY_COUNT_AL EHCA_BMASK_IBM(37,37)
+#define MQPCB_RNR_RETRY_COUNT_AL EHCA_BMASK_IBM(29,31)
+#define MQPCB_MASK_SOURCE_PATH_BITS_AL EHCA_BMASK_IBM(38,38)
+#define MQPCB_SOURCE_PATH_BITS_AL EHCA_BMASK_IBM(25,31)
+#define MQPCB_MASK_TRAFFIC_CLASS_AL EHCA_BMASK_IBM(39,39)
+#define MQPCB_TRAFFIC_CLASS_AL EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_HOP_LIMIT_AL EHCA_BMASK_IBM(40,40)
+#define MQPCB_HOP_LIMIT_AL EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_SOURCE_GID_IDX_AL EHCA_BMASK_IBM(41,41)
+#define MQPCB_SOURCE_GID_IDX_AL EHCA_BMASK_IBM(24,31)
+#define MQPCB_MASK_FLOW_LABEL_AL EHCA_BMASK_IBM(42,42)
+#define MQPCB_FLOW_LABEL_AL EHCA_BMASK_IBM(12,31)
+#define MQPCB_MASK_DEST_GID_AL EHCA_BMASK_IBM(44,44)
+#define MQPCB_MASK_MAX_NR_OUTST_SEND_WR EHCA_BMASK_IBM(45,45)
+#define MQPCB_MAX_NR_OUTST_SEND_WR EHCA_BMASK_IBM(16,31)
+#define MQPCB_MASK_MAX_NR_OUTST_RECV_WR EHCA_BMASK_IBM(46,46)
+#define MQPCB_MAX_NR_OUTST_RECV_WR EHCA_BMASK_IBM(16,31)
+#define MQPCB_MASK_DISABLE_ETE_CREDIT_CHECK EHCA_BMASK_IBM(47,47)
+#define MQPCB_DISABLE_ETE_CREDIT_CHECK EHCA_BMASK_IBM(31,31)
+#define MQPCB_QP_NUMBER EHCA_BMASK_IBM(8,31)
+#define MQPCB_MASK_QP_ENABLE EHCA_BMASK_IBM(48,48)
+#define MQPCB_QP_ENABLE EHCA_BMASK_IBM(31,31)
+#define MQPCB_MASK_CURR_SQR_LIMIT EHCA_BMASK_IBM(49,49)
+#define MQPCB_CURR_SQR_LIMIT EHCA_BMASK_IBM(15,31)
+#define MQPCB_MASK_QP_AFF_ASYN_EV_LOG_REG EHCA_BMASK_IBM(50,50)
+#define MQPCB_MASK_SHARED_RQ_HNDL EHCA_BMASK_IBM(51,51)
+
+#endif /* __EHCA_CLASSES_PSERIES_H__ */
^ permalink raw reply
* [PATCH 01/16] ehca: module infrastructure
From: Heiko J Schick @ 2006-05-15 17:41 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
drivers/infiniband/hw/ehca/ehca_main.c | 966 +++++++++++++++++++++++++++++++++
1 file changed, 966 insertions(+)
--- linux-2.6.17-rc2-orig/drivers/infiniband/hw/ehca/ehca_main.c 1970-01-01 01:00:00.000000000 +0100
+++ linux-2.6.17-rc2/drivers/infiniband/hw/ehca/ehca_main.c 2006-05-15 19:17:26.000000000 +0200
@@ -0,0 +1,966 @@
+/*
+ * IBM eServer eHCA Infiniband device driver for Linux on POWER
+ *
+ * module start stop, hca detection
+ *
+ * Authors: Heiko J Schick <schickhj@de.ibm.com>
+ *
+ * Copyright (c) 2005 IBM Corporation
+ *
+ * All rights reserved.
+ *
+ * This source code is distributed under a dual license of GPL v2.0 and OpenIB
+ * BSD.
+ *
+ * OpenIB BSD License
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * Redistributions of source code must retain the above copyright notice, this
+ * list of conditions and the following disclaimer.
+ *
+ * Redistributions in binary form must reproduce the above copyright notice,
+ * this list of conditions and the following disclaimer in the documentation
+ * and/or other materials
+ * provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
+ * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#define DEB_PREFIX "shca"
+
+#include "ehca_classes.h"
+#include "ehca_iverbs.h"
+#include "ehca_mrmw.h"
+#include "ehca_tools.h"
+#include "hcp_if.h"
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_AUTHOR("Christoph Raisch <raisch@de.ibm.com>");
+MODULE_DESCRIPTION("IBM eServer HCA InfiniBand Device Driver");
+MODULE_VERSION("SVNEHCA_0006");
+
+struct ehca_comp_pool* ehca_pool;
+
+int ehca_open_aqp1 = 0;
+int ehca_debug_level = -1;
+int ehca_hw_level = 0;
+int ehca_nr_ports = 2;
+int ehca_use_hp_mr = 0;
+int ehca_port_act_time = 30;
+int ehca_poll_all_eqs = 1;
+int ehca_static_rate = -1;
+
+module_param_named(open_aqp1, ehca_open_aqp1, int, 0);
+module_param_named(debug_level, ehca_debug_level, int, 0);
+module_param_named(hw_level, ehca_hw_level, int, 0);
+module_param_named(nr_ports, ehca_nr_ports, int, 0);
+module_param_named(use_hp_mr, ehca_use_hp_mr, int, 0);
+module_param_named(port_act_time, ehca_port_act_time, int, 0);
+module_param_named(poll_all_eqs, ehca_poll_all_eqs, int, 0);
+module_param_named(static_rate, ehca_static_rate, int, 0);
+
+MODULE_PARM_DESC(open_aqp1,
+ "AQP1 on startup (0: no (default), 1: yes)");
+MODULE_PARM_DESC(debug_level,
+ "debug level"
+ " (0: node, 6: only errors (default), 9: all)");
+MODULE_PARM_DESC(hw_level,
+ "hardware level"
+ " (0: autosensing (default), 1: v. 0.20, 2: v. 0.21)");
+MODULE_PARM_DESC(nr_ports,
+ "number of connected ports (default: 2)");
+MODULE_PARM_DESC(use_hp_mr,
+ "high performance MRs (0: no (default), 1: yes)");
+MODULE_PARM_DESC(port_act_time,
+ "time to wait for port activation (default: 30 sec)");
+MODULE_PARM_DESC(poll_all_eqs,
+ "polls all event queues periodically"
+ " (0: no, 1: yes (default))");
+MODULE_PARM_DESC(static_rate,
+ "set permanent static rate (default: disabled)");
+
+/* This external trace mask controls what will end up in the
+ * kernel ring buffer. Number 6 means, that everything between
+ * 0 and 5 will be stored.
+ */
+u8 ehca_edeb_mask[EHCA_EDEB_TRACE_MASK_SIZE]={6, 6, 6, 6,
+ 6, 6, 6, 6,
+ 6, 6, 6, 6,
+ 6, 6, 6, 6,
+ 6, 6, 6, 6,
+ 6, 6, 6, 6,
+ 6, 6, 6, 6,
+ 6, 6, 0, 0};
+
+spinlock_t ehca_qp_idr_lock;
+spinlock_t ehca_cq_idr_lock;
+DEFINE_IDR(ehca_qp_idr);
+DEFINE_IDR(ehca_cq_idr);
+
+struct ehca_module ehca_module;
+
+void ehca_init_trace(void)
+{
+ EDEB_EN(7, "");
+
+ if (ehca_debug_level != -1) {
+ int i;
+ for (i = 0; i < EHCA_EDEB_TRACE_MASK_SIZE; i++)
+ ehca_edeb_mask[i] = ehca_debug_level;
+ }
+
+ EDEB_EX(7, "");
+}
+
+int ehca_create_slab_caches(struct ehca_module *ehca_module)
+{
+ int ret = 0;
+
+ EDEB_EN(7, "");
+
+ ehca_module->cache_pd =
+ kmem_cache_create("ehca_cache_pd",
+ sizeof(struct ehca_pd),
+ 0, SLAB_HWCACHE_ALIGN,
+ NULL, NULL);
+ if (!ehca_module->cache_pd) {
+ EDEB_ERR(4, "Cannot create PD SLAB cache.");
+ ret = -ENOMEM;
+ goto create_slab_caches1;
+ }
+
+ ehca_module->cache_cq =
+ kmem_cache_create("ehca_cache_cq",
+ sizeof(struct ehca_cq),
+ 0, SLAB_HWCACHE_ALIGN,
+ NULL, NULL);
+ if (!ehca_module->cache_cq) {
+ EDEB_ERR(4, "Cannot create CQ SLAB cache.");
+ ret = -ENOMEM;
+ goto create_slab_caches2;
+ }
+
+ ehca_module->cache_qp =
+ kmem_cache_create("ehca_cache_qp",
+ sizeof(struct ehca_qp),
+ 0, SLAB_HWCACHE_ALIGN,
+ NULL, NULL);
+ if (!ehca_module->cache_qp) {
+ EDEB_ERR(4, "Cannot create QP SLAB cache.");
+ ret = -ENOMEM;
+ goto create_slab_caches3;
+ }
+
+ ehca_module->cache_av =
+ kmem_cache_create("ehca_cache_av",
+ sizeof(struct ehca_av),
+ 0, SLAB_HWCACHE_ALIGN,
+ NULL, NULL);
+ if (!ehca_module->cache_av) {
+ EDEB_ERR(4, "Cannot create AV SLAB cache.");
+ ret = -ENOMEM;
+ goto create_slab_caches4;
+ }
+
+ ehca_module->cache_mw =
+ kmem_cache_create("ehca_cache_mw",
+ sizeof(struct ehca_mw),
+ 0, SLAB_HWCACHE_ALIGN,
+ NULL, NULL);
+ if (!ehca_module->cache_mw) {
+ EDEB_ERR(4, "Cannot create MW SLAB cache.");
+ ret = -ENOMEM;
+ goto create_slab_caches5;
+ }
+
+ ehca_module->cache_mr =
+ kmem_cache_create("ehca_cache_mr",
+ sizeof(struct ehca_mr),
+ 0, SLAB_HWCACHE_ALIGN,
+ NULL, NULL);
+ if (!ehca_module->cache_mr) {
+ EDEB_ERR(4, "Cannot create MR SLAB cache.");
+ ret = -ENOMEM;
+ goto create_slab_caches6;
+ }
+
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+
+create_slab_caches6:
+ kmem_cache_destroy(ehca_module->cache_mw);
+
+create_slab_caches5:
+ kmem_cache_destroy(ehca_module->cache_av);
+
+create_slab_caches4:
+ kmem_cache_destroy(ehca_module->cache_qp);
+
+create_slab_caches3:
+ kmem_cache_destroy(ehca_module->cache_cq);
+
+create_slab_caches2:
+ kmem_cache_destroy(ehca_module->cache_pd);
+
+create_slab_caches1:
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
+
+int ehca_destroy_slab_caches(struct ehca_module *ehca_module)
+{
+ int ret;
+
+ EDEB_EN(7, "");
+
+ ret = kmem_cache_destroy(ehca_module->cache_pd);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy PD SLAB cache. ret=%x", ret);
+
+ ret = kmem_cache_destroy(ehca_module->cache_cq);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy CQ SLAB cache. ret=%x", ret);
+
+ ret = kmem_cache_destroy(ehca_module->cache_qp);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy QP SLAB cache. ret=%x", ret);
+
+ ret = kmem_cache_destroy(ehca_module->cache_av);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy AV SLAB cache. ret=%x", ret);
+
+ ret = kmem_cache_destroy(ehca_module->cache_mw);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy MW SLAB cache. ret=%x", ret);
+
+ ret = kmem_cache_destroy(ehca_module->cache_mr);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy MR SLAB cache. ret=%x", ret);
+
+ EDEB_EX(7, "");
+
+ return 0;
+}
+
+#define EHCA_HCAAVER EHCA_BMASK_IBM(32,39)
+#define EHCA_REVID EHCA_BMASK_IBM(40,63)
+
+int ehca_sense_attributes(struct ehca_shca *shca)
+{
+ int ret = -EINVAL;
+ u64 h_ret = H_SUCCESS;
+ struct hipz_query_hca *rblock;
+
+ EDEB_EN(7, "shca=%p", shca);
+
+ rblock = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!rblock) {
+ EDEB_ERR(4, "Cannot allocate rblock memory.");
+ ret = -ENOMEM;
+ goto num_ports0;
+ }
+
+ h_ret = hipz_h_query_hca(shca->ipz_hca_handle, rblock);
+ if (h_ret != H_SUCCESS) {
+ EDEB_ERR(4, "Cannot query device properties. h_ret=%lx", h_ret);
+ ret = -EPERM;
+ goto num_ports1;
+ }
+
+ if (ehca_nr_ports == 1)
+ shca->num_ports = 1;
+ else
+ shca->num_ports = (u8)rblock->num_ports;
+
+ EDEB(6, " ... found %x ports", rblock->num_ports);
+
+ if (ehca_hw_level == 0) {
+ u32 hcaaver;
+ u32 revid;
+
+ hcaaver = EHCA_BMASK_GET(EHCA_HCAAVER, rblock->hw_ver);
+ revid = EHCA_BMASK_GET(EHCA_REVID, rblock->hw_ver);
+
+ EDEB(6, " ... hardware version=%x:%x",
+ hcaaver, revid);
+
+ if ((hcaaver == 1) && (revid == 0))
+ shca->hw_level = 0;
+ else if ((hcaaver == 1) && (revid == 1))
+ shca->hw_level = 1;
+ else if ((hcaaver == 1) && (revid == 2))
+ shca->hw_level = 2;
+ }
+ EDEB(6, " ... hardware level=%x", shca->hw_level);
+
+ shca->sport[0].rate = IB_RATE_30_GBPS;
+ shca->sport[1].rate = IB_RATE_30_GBPS;
+
+ ret = 0;
+
+num_ports1:
+ kfree(rblock);
+
+num_ports0:
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
+
+static int init_node_guid(struct ehca_shca* shca)
+{
+ int ret = 0;
+ struct hipz_query_hca *rblock;
+
+ EDEB_EN(7, "");
+
+ rblock = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL);
+ if (!rblock) {
+ EDEB_ERR(4, "Can't allocate rblock memory.");
+ ret = -ENOMEM;
+ goto init_node_guid0;
+ }
+
+ if (hipz_h_query_hca(shca->ipz_hca_handle, rblock) != H_SUCCESS) {
+ EDEB_ERR(4, "Can't query device properties");
+ ret = -EINVAL;
+ goto init_node_guid1;
+ }
+
+ memcpy(&shca->ib_device.node_guid, &rblock->node_guid, (sizeof(u64)));
+
+init_node_guid1:
+ kfree(rblock);
+
+init_node_guid0:
+ EDEB_EX(7, "node_guid=%lx ret=%x", shca->ib_device.node_guid, ret);
+
+ return ret;
+}
+
+int ehca_register_device(struct ehca_shca *shca)
+{
+ int ret = 0;
+
+ EDEB_EN(7, "shca=%p", shca);
+
+ ret = init_node_guid(shca);
+ if (ret)
+ return ret;
+
+ strlcpy(shca->ib_device.name, "ehca%d", IB_DEVICE_NAME_MAX);
+ shca->ib_device.owner = THIS_MODULE;
+
+ shca->ib_device.uverbs_abi_ver = 5;
+ shca->ib_device.uverbs_cmd_mask =
+ (1ull << IB_USER_VERBS_CMD_GET_CONTEXT) |
+ (1ull << IB_USER_VERBS_CMD_QUERY_DEVICE) |
+ (1ull << IB_USER_VERBS_CMD_QUERY_PORT) |
+ (1ull << IB_USER_VERBS_CMD_ALLOC_PD) |
+ (1ull << IB_USER_VERBS_CMD_DEALLOC_PD) |
+ (1ull << IB_USER_VERBS_CMD_REG_MR) |
+ (1ull << IB_USER_VERBS_CMD_DEREG_MR) |
+ (1ull << IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL) |
+ (1ull << IB_USER_VERBS_CMD_CREATE_CQ) |
+ (1ull << IB_USER_VERBS_CMD_DESTROY_CQ) |
+ (1ull << IB_USER_VERBS_CMD_CREATE_QP) |
+ (1ull << IB_USER_VERBS_CMD_MODIFY_QP) |
+ (1ull << IB_USER_VERBS_CMD_QUERY_QP) |
+ (1ull << IB_USER_VERBS_CMD_DESTROY_QP) |
+ (1ull << IB_USER_VERBS_CMD_ATTACH_MCAST) |
+ (1ull << IB_USER_VERBS_CMD_DETACH_MCAST);
+
+ shca->ib_device.node_type = RDMA_NODE_IB_CA;
+ shca->ib_device.phys_port_cnt = shca->num_ports;
+ shca->ib_device.dma_device = &shca->ibmebus_dev->ofdev.dev;
+ shca->ib_device.query_device = ehca_query_device;
+ shca->ib_device.query_port = ehca_query_port;
+ shca->ib_device.query_gid = ehca_query_gid;
+ shca->ib_device.query_pkey = ehca_query_pkey;
+ /* shca->in_device.modify_device = ehca_modify_device */
+ shca->ib_device.modify_port = ehca_modify_port;
+ shca->ib_device.alloc_ucontext = ehca_alloc_ucontext;
+ shca->ib_device.dealloc_ucontext = ehca_dealloc_ucontext;
+ shca->ib_device.alloc_pd = ehca_alloc_pd;
+ shca->ib_device.dealloc_pd = ehca_dealloc_pd;
+ shca->ib_device.create_ah = ehca_create_ah;
+ /* shca->ib_device.modify_ah = ehca_modify_ah; */
+ shca->ib_device.query_ah = ehca_query_ah;
+ shca->ib_device.destroy_ah = ehca_destroy_ah;
+ shca->ib_device.create_qp = ehca_create_qp;
+ shca->ib_device.modify_qp = ehca_modify_qp;
+ shca->ib_device.query_qp = ehca_query_qp;
+ shca->ib_device.destroy_qp = ehca_destroy_qp;
+ shca->ib_device.post_send = ehca_post_send;
+ shca->ib_device.post_recv = ehca_post_recv;
+ shca->ib_device.create_cq = ehca_create_cq;
+ shca->ib_device.destroy_cq = ehca_destroy_cq;
+ shca->ib_device.resize_cq = ehca_resize_cq;
+ shca->ib_device.poll_cq = ehca_poll_cq;
+ /* shca->ib_device.peek_cq = ehca_peek_cq; */
+ shca->ib_device.req_notify_cq = ehca_req_notify_cq;
+ /* shca->ib_device.req_ncomp_notif = ehca_req_ncomp_notif; */
+ shca->ib_device.get_dma_mr = ehca_get_dma_mr;
+ shca->ib_device.reg_phys_mr = ehca_reg_phys_mr;
+ shca->ib_device.reg_user_mr = ehca_reg_user_mr;
+ shca->ib_device.query_mr = ehca_query_mr;
+ shca->ib_device.dereg_mr = ehca_dereg_mr;
+ shca->ib_device.rereg_phys_mr = ehca_rereg_phys_mr;
+ shca->ib_device.alloc_mw = ehca_alloc_mw;
+ shca->ib_device.bind_mw = ehca_bind_mw;
+ shca->ib_device.dealloc_mw = ehca_dealloc_mw;
+ shca->ib_device.alloc_fmr = ehca_alloc_fmr;
+ shca->ib_device.map_phys_fmr = ehca_map_phys_fmr;
+ shca->ib_device.unmap_fmr = ehca_unmap_fmr;
+ shca->ib_device.dealloc_fmr = ehca_dealloc_fmr;
+ shca->ib_device.attach_mcast = ehca_attach_mcast;
+ shca->ib_device.detach_mcast = ehca_detach_mcast;
+ /* shca->ib_device.process_mad = ehca_process_mad; */
+ shca->ib_device.mmap = ehca_mmap;
+
+ ret = ib_register_device(&shca->ib_device);
+
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
+
+static int ehca_create_aqp1(struct ehca_shca *shca, u32 port)
+{
+ struct ehca_sport *sport;
+ struct ib_cq *ibcq;
+ struct ib_qp *ibqp;
+ struct ib_qp_init_attr qp_init_attr;
+ int ret = 0;
+
+ EDEB_EN(7, "shca=%p port=%x", shca, port);
+
+ sport = &shca->sport[port - 1];
+
+ if (sport->ibcq_aqp1) {
+ EDEB_ERR(4, "AQP1 CQ is already created.");
+ return -EPERM;
+ }
+
+ ibcq = ib_create_cq(&shca->ib_device, NULL, NULL, (void*)(-1), 10);
+ if (IS_ERR(ibcq)) {
+ EDEB_ERR(4, "Cannot create AQP1 CQ.");
+ return PTR_ERR(ibcq);
+ }
+ sport->ibcq_aqp1 = ibcq;
+
+ if (sport->ibqp_aqp1) {
+ EDEB_ERR(4, "AQP1 QP is already created.");
+ ret = -EPERM;
+ goto create_aqp1;
+ }
+
+ memset(&qp_init_attr, 0, sizeof(struct ib_qp_init_attr));
+ qp_init_attr.send_cq = ibcq;
+ qp_init_attr.recv_cq = ibcq;
+ qp_init_attr.sq_sig_type = IB_SIGNAL_ALL_WR;
+ qp_init_attr.cap.max_send_wr = 100;
+ qp_init_attr.cap.max_recv_wr = 100;
+ qp_init_attr.cap.max_send_sge = 2;
+ qp_init_attr.cap.max_recv_sge = 1;
+ qp_init_attr.qp_type = IB_QPT_GSI;
+ qp_init_attr.port_num = port;
+ qp_init_attr.qp_context = NULL;
+ qp_init_attr.event_handler = NULL;
+ qp_init_attr.srq = NULL;
+
+ ibqp = ib_create_qp(&shca->pd->ib_pd, &qp_init_attr);
+ if (IS_ERR(ibqp)) {
+ EDEB_ERR(4, "Cannot create AQP1 QP.");
+ ret = PTR_ERR(ibqp);
+ goto create_aqp1;
+ }
+ sport->ibqp_aqp1 = ibqp;
+
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+
+create_aqp1:
+ ib_destroy_cq(sport->ibcq_aqp1);
+
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
+
+static int ehca_destroy_aqp1(struct ehca_sport *sport)
+{
+ int ret = 0;
+
+ EDEB_EN(7, "sport=%p", sport);
+
+ ret = ib_destroy_qp(sport->ibqp_aqp1);
+ if (ret) {
+ EDEB_ERR(4, "Cannot destroy AQP1 QP. ret=%x", ret);
+ goto destroy_aqp1;
+ }
+
+ ret = ib_destroy_cq(sport->ibcq_aqp1);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy AQP1 CQ. ret=%x", ret);
+
+destroy_aqp1:
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
+
+static ssize_t ehca_show_debug_mask(struct device_driver *ddp, char *buf)
+{
+ int i;
+ int total = 0;
+ total += snprintf(buf + total, PAGE_SIZE - total, "%d",
+ ehca_edeb_mask[0]);
+ for (i = 1; i < EHCA_EDEB_TRACE_MASK_SIZE; i++) {
+ total += snprintf(buf + total, PAGE_SIZE - total, "%d",
+ ehca_edeb_mask[i]);
+ }
+
+ total += snprintf(buf + total, PAGE_SIZE - total, "\n");
+
+ return total;
+}
+
+static ssize_t ehca_store_debug_mask(struct device_driver *ddp,
+ const char *buf, size_t count)
+{
+ int i;
+ for (i = 0; i < EHCA_EDEB_TRACE_MASK_SIZE; i++) {
+ char value = buf[i] - '0';
+ if ((value <= 9) && (count >= i)) {
+ ehca_edeb_mask[i] = value;
+ }
+ }
+ return count;
+}
+DRIVER_ATTR(debug_mask, S_IRUSR | S_IWUSR,
+ ehca_show_debug_mask, ehca_store_debug_mask);
+
+void ehca_create_driver_sysfs(struct ibmebus_driver *drv)
+{
+ driver_create_file(&drv->driver, &driver_attr_debug_mask);
+}
+
+void ehca_remove_driver_sysfs(struct ibmebus_driver *drv)
+{
+ driver_remove_file(&drv->driver, &driver_attr_debug_mask);
+}
+
+#define EHCA_RESOURCE_ATTR(name) \
+static ssize_t ehca_show_##name(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ \
+ struct ehca_shca *shca; \
+ struct hipz_query_hca *rblock; \
+ int data; \
+ \
+ shca = dev->driver_data; \
+ \
+ rblock = kzalloc(H_CB_ALIGNMENT, GFP_KERNEL); \
+ if (!rblock) { \
+ EDEB_ERR(4, "Can't allocate rblock memory."); \
+ return 0; \
+ } \
+ \
+ if (hipz_h_query_hca(shca->ipz_hca_handle, rblock) != H_SUCCESS) { \
+ EDEB_ERR(4, "Can't query device properties"); \
+ kfree(rblock); \
+ return 0; \
+ } \
+ \
+ data = rblock->name; \
+ kfree(rblock); \
+ \
+ if ((strcmp(#name, "num_ports") == 0) && (ehca_nr_ports == 1)) \
+ return snprintf(buf, 256, "1\n"); \
+ else \
+ return snprintf(buf, 256, "%d\n", data); \
+ \
+} \
+static DEVICE_ATTR(name, S_IRUGO, ehca_show_##name, NULL);
+
+EHCA_RESOURCE_ATTR(num_ports);
+EHCA_RESOURCE_ATTR(hw_ver);
+EHCA_RESOURCE_ATTR(max_eq);
+EHCA_RESOURCE_ATTR(cur_eq);
+EHCA_RESOURCE_ATTR(max_cq);
+EHCA_RESOURCE_ATTR(cur_cq);
+EHCA_RESOURCE_ATTR(max_qp);
+EHCA_RESOURCE_ATTR(cur_qp);
+EHCA_RESOURCE_ATTR(max_mr);
+EHCA_RESOURCE_ATTR(cur_mr);
+EHCA_RESOURCE_ATTR(max_mw);
+EHCA_RESOURCE_ATTR(cur_mw);
+EHCA_RESOURCE_ATTR(max_pd);
+EHCA_RESOURCE_ATTR(max_ah);
+
+static ssize_t ehca_show_adapter_handle(struct device *dev,
+ struct device_attribute *attr,
+ char *buf)
+{
+ struct ehca_shca *shca = dev->driver_data;
+
+ return sprintf(buf, "%lx\n", shca->ipz_hca_handle.handle);
+
+}
+static DEVICE_ATTR(adapter_handle, S_IRUGO, ehca_show_adapter_handle, NULL);
+
+
+
+void ehca_create_device_sysfs(struct ibmebus_dev *dev)
+{
+ device_create_file(&dev->ofdev.dev, &dev_attr_adapter_handle);
+ device_create_file(&dev->ofdev.dev, &dev_attr_num_ports);
+ device_create_file(&dev->ofdev.dev, &dev_attr_hw_ver);
+ device_create_file(&dev->ofdev.dev, &dev_attr_max_eq);
+ device_create_file(&dev->ofdev.dev, &dev_attr_cur_eq);
+ device_create_file(&dev->ofdev.dev, &dev_attr_max_cq);
+ device_create_file(&dev->ofdev.dev, &dev_attr_cur_cq);
+ device_create_file(&dev->ofdev.dev, &dev_attr_max_qp);
+ device_create_file(&dev->ofdev.dev, &dev_attr_cur_qp);
+ device_create_file(&dev->ofdev.dev, &dev_attr_max_mr);
+ device_create_file(&dev->ofdev.dev, &dev_attr_cur_mr);
+ device_create_file(&dev->ofdev.dev, &dev_attr_max_mw);
+ device_create_file(&dev->ofdev.dev, &dev_attr_cur_mw);
+ device_create_file(&dev->ofdev.dev, &dev_attr_max_pd);
+ device_create_file(&dev->ofdev.dev, &dev_attr_max_ah);
+}
+
+void ehca_remove_device_sysfs(struct ibmebus_dev *dev)
+{
+ device_remove_file(&dev->ofdev.dev, &dev_attr_adapter_handle);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_num_ports);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_hw_ver);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_max_eq);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_cur_eq);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_max_cq);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_cur_cq);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_max_qp);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_cur_qp);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_max_mr);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_cur_mr);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_max_mw);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_cur_mw);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_max_pd);
+ device_remove_file(&dev->ofdev.dev, &dev_attr_max_ah);
+}
+
+static int __devinit ehca_probe(struct ibmebus_dev *dev,
+ const struct of_device_id *id)
+{
+ struct ehca_shca *shca;
+ u64 *handle;
+ struct ib_pd *ibpd;
+ int ret = 0;
+
+ EDEB_EN(7, "name=%s", dev->name);
+
+ handle = (u64 *)get_property(dev->ofdev.node, "ibm,hca-handle", NULL);
+ if (!handle) {
+ EDEB_ERR(4, "Cannot get eHCA handle for adapter: %s.",
+ dev->ofdev.node->full_name);
+ return -ENODEV;
+ }
+
+ if (!(*handle)) {
+ EDEB_ERR(4, "Wrong eHCA handle for adapter: %s.",
+ dev->ofdev.node->full_name);
+ return -ENODEV;
+ }
+
+ shca = (struct ehca_shca *)ib_alloc_device(sizeof(*shca));
+ if (shca == NULL) {
+ EDEB_ERR(4, "Cannot allocate shca memory.");
+ return -ENOMEM;
+ }
+
+ shca->ibmebus_dev = dev;
+ shca->ipz_hca_handle.handle = *handle;
+ dev->ofdev.dev.driver_data = shca;
+
+ ret = ehca_sense_attributes(shca);
+ if (ret < 0) {
+ EDEB_ERR(4, "Cannot sense eHCA attributes.");
+ goto probe1;
+ }
+
+ /* create event queues */
+ ret = ehca_create_eq(shca, &shca->eq, EHCA_EQ, 2048);
+ if (ret) {
+ EDEB_ERR(4, "Cannot create EQ.");
+ goto probe1;
+ }
+
+ ret = ehca_create_eq(shca, &shca->neq, EHCA_NEQ, 513);
+ if (ret) {
+ EDEB_ERR(4, "Cannot create NEQ.");
+ goto probe2;
+ }
+
+ /* create internal protection domain */
+ ibpd = ehca_alloc_pd(&shca->ib_device, (void*)(-1), NULL);
+ if (IS_ERR(ibpd)) {
+ EDEB_ERR(4, "Cannot create internal PD.");
+ ret = PTR_ERR(ibpd);
+ goto probe3;
+ }
+
+ shca->pd = container_of(ibpd, struct ehca_pd, ib_pd);
+ shca->pd->ib_pd.device = &shca->ib_device;
+
+ /* create internal max MR */
+ ret = ehca_reg_internal_maxmr(shca, shca->pd, &shca->maxmr);
+ if (ret) {
+ EDEB_ERR(4, "Cannot create internal MR. ret=%x", ret);
+ goto probe4;
+ }
+
+ ret = ehca_register_device(shca);
+ if (ret) {
+ EDEB_ERR(4, "Cannot register Infiniband device.");
+ goto probe5;
+ }
+
+ /* create AQP1 for port 1 */
+ if (ehca_open_aqp1 == 1) {
+ shca->sport[0].port_state = IB_PORT_DOWN;
+ ret = ehca_create_aqp1(shca, 1);
+ if (ret) {
+ EDEB_ERR(4, "Cannot create AQP1 for port 1.");
+ goto probe6;
+ }
+ }
+
+ /* create AQP1 for port 2 */
+ if ((ehca_open_aqp1 == 1) && (shca->num_ports == 2)) {
+ shca->sport[1].port_state = IB_PORT_DOWN;
+ ret = ehca_create_aqp1(shca, 2);
+ if (ret) {
+ EDEB_ERR(4, "Cannot create AQP1 for port 2.");
+ goto probe7;
+ }
+ }
+
+ ehca_create_device_sysfs(dev);
+
+ spin_lock(&ehca_module.shca_lock);
+ list_add(&shca->shca_list, &ehca_module.shca_list);
+ spin_unlock(&ehca_module.shca_lock);
+
+ EDEB_EX(7, "ret=%x", ret);
+
+ return 0;
+
+probe7:
+ ret = ehca_destroy_aqp1(&shca->sport[0]);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy AQP1 for port 1. ret=%x", ret);
+
+probe6:
+ ib_unregister_device(&shca->ib_device);
+
+probe5:
+ ret = ehca_dereg_internal_maxmr(shca);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy internal MR. ret=%x", ret);
+
+probe4:
+ ret = ehca_dealloc_pd(&shca->pd->ib_pd);
+ if (ret != 0)
+ EDEB_ERR(4, "Cannot destroy internal PD. ret=%x", ret);
+
+probe3:
+ ret = ehca_destroy_eq(shca, &shca->neq);
+ if (ret != 0)
+ EDEB_ERR(4, "Cannot destroy NEQ. ret=%x", ret);
+
+probe2:
+ ret = ehca_destroy_eq(shca, &shca->eq);
+ if (ret != 0)
+ EDEB_ERR(4, "Cannot destroy EQ. ret=%x", ret);
+
+probe1:
+ ib_dealloc_device(&shca->ib_device);
+
+ EDEB_EX(4, "ret=%x", ret);
+
+ return -EINVAL;
+}
+
+static int __devexit ehca_remove(struct ibmebus_dev *dev)
+{
+ struct ehca_shca *shca = dev->ofdev.dev.driver_data;
+ int ret;
+
+ EDEB_EN(7, "shca=%p", shca);
+
+ ehca_remove_device_sysfs(dev);
+
+ if (ehca_open_aqp1 == 1) {
+ int i;
+
+ for (i = 0; i < shca->num_ports; i++) {
+ ret = ehca_destroy_aqp1(&shca->sport[i]);
+ if (ret != 0)
+ EDEB_ERR(4, "Cannot destroy AQP1 for port %x."
+ " ret=%x", ret, i);
+ }
+ }
+
+ ib_unregister_device(&shca->ib_device);
+
+ ret = ehca_dereg_internal_maxmr(shca);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy internal MR. ret=%x", ret);
+
+ ret = ehca_dealloc_pd(&shca->pd->ib_pd);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy internal PD. ret=%x", ret);
+
+ ret = ehca_destroy_eq(shca, &shca->eq);
+ if (ret)
+ EDEB_ERR(4, "Cannot destroy EQ. ret=%x", ret);
+
+ ret = ehca_destroy_eq(shca, &shca->neq);
+ if (ret)
+ EDEB_ERR(4, "Canot destroy NEQ. ret=%x", ret);
+
+ ib_dealloc_device(&shca->ib_device);
+
+ spin_lock(&ehca_module.shca_lock);
+ list_del(&shca->shca_list);
+ spin_unlock(&ehca_module.shca_lock);
+
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+}
+
+static struct of_device_id ehca_device_table[] =
+{
+ {
+ .name = "lhca",
+ .compatible = "IBM,lhca",
+ },
+ {},
+};
+
+static struct ibmebus_driver ehca_driver = {
+ .name = "ehca",
+ .id_table = ehca_device_table,
+ .probe = ehca_probe,
+ .remove = ehca_remove,
+};
+
+int __init ehca_module_init(void)
+{
+ int ret = 0;
+
+ printk(KERN_INFO "eHCA Infiniband Device Driver "
+ "(Rel.: SVNEHCA_0006)\n");
+ EDEB_EN(7, "");
+
+ idr_init(&ehca_qp_idr);
+ idr_init(&ehca_cq_idr);
+ spin_lock_init(&ehca_qp_idr_lock);
+ spin_lock_init(&ehca_cq_idr_lock);
+
+ INIT_LIST_HEAD(&ehca_module.shca_list);
+ spin_lock_init(&ehca_module.shca_lock);
+
+ ehca_init_trace();
+
+ ehca_pool = ehca_create_comp_pool();
+ if (ehca_pool == NULL) {
+ EDEB_ERR(4, "Cannot create comp pool.");
+ ret = -EINVAL;
+ goto module_init0;
+ }
+
+ if ((ret = ehca_create_slab_caches(&ehca_module))) {
+ EDEB_ERR(4, "Cannot create SLAB caches");
+ ret = -ENOMEM;
+ goto module_init1;
+ }
+
+ if ((ret = ibmebus_register_driver(&ehca_driver))) {
+ EDEB_ERR(4, "Cannot register eHCA device driver");
+ ret = -EINVAL;
+ goto module_init2;
+ }
+
+ ehca_create_driver_sysfs(&ehca_driver);
+
+ if (ehca_poll_all_eqs != 1) {
+ EDEB_ERR(4, "WARNING!!!");
+ EDEB_ERR(4, "It is possible to lose interrupts.");
+
+ return 0;
+ }
+
+ init_timer(&ehca_module.timer);
+ ehca_module.timer.function = ehca_poll_eqs;
+ ehca_module.timer.data = (unsigned long)(void*)&ehca_module;
+ ehca_module.timer.expires = jiffies + HZ;
+ add_timer(&ehca_module.timer);
+
+ EDEB_EX(7, "ret=%x", ret);
+
+ return 0;
+
+module_init2:
+ ehca_destroy_slab_caches(&ehca_module);
+
+module_init1:
+ ehca_destroy_comp_pool(ehca_pool);
+
+module_init0:
+ EDEB_EX(7, "ret=%x", ret);
+
+ return ret;
+};
+
+void __exit ehca_module_exit(void)
+{
+ EDEB_EN(7, "");
+
+ if (ehca_poll_all_eqs == 1)
+ del_timer_sync(&ehca_module.timer);
+
+ ehca_remove_driver_sysfs(&ehca_driver);
+ ibmebus_unregister_driver(&ehca_driver);
+
+ if (ehca_destroy_slab_caches(&ehca_module) != 0)
+ EDEB_ERR(4, "Cannot destroy SLAB caches");
+
+ ehca_destroy_comp_pool(ehca_pool);
+
+ idr_destroy(&ehca_cq_idr);
+ idr_destroy(&ehca_qp_idr);
+
+ EDEB_EX(7, "");
+};
+
+module_init(ehca_module_init);
+module_exit(ehca_module_exit);
^ permalink raw reply
* [PATCH 00/16] ehca: IBM eHCA InfiniBand Device Driver
From: Heiko J Schick @ 2006-05-15 17:41 UTC (permalink / raw)
To: openib-general, Christoph Raisch, Hoang-Nam Nguyen, Marcus Eder,
schihei, linux-kernel, linuxppc-dev
Hello,
many thanks for your comments. They are very helpful for us. All
17 patches have to be applied, otherwise the driver won't compile.
We would appreciate for any comments and feedbacks.
Signed-off-by: Heiko J Schick <schickhj@de.ibm.com>
Changelog-by: Heiko J Schick <schickhj@de.ibm.com>
Changelog:
Differences to PatchSet http://openib.org/pipermail/openib-general/2006-April/020584.html
Differences to PatchSet http://openib.org/pipermail/openib-general/2006-March/018144.html
Differences to PatchSet http://openib.org/pipermail/openib-general/2006-March/017412.html
- Linux kernel coding style
- Reduce number of parameters passed to firmware interface wrappers
- Remove ehca_kernel.h
- Remove implementation of plpar_hcall_7arg_7ret() and plpar_hcall_9arg_9ret(),
which are now included in kernel code
- Remove simulation stub
drivers/infiniband/hw/ehca/Kconfig | 6
drivers/infiniband/hw/ehca/Makefile | 16
drivers/infiniband/hw/ehca/ehca_av.c | 306 ++
drivers/infiniband/hw/ehca/ehca_classes.h | 350 +++
drivers/infiniband/hw/ehca/ehca_classes_pSeries.h | 251 ++
drivers/infiniband/hw/ehca/ehca_cq.c | 431 +++
drivers/infiniband/hw/ehca/ehca_eq.c | 222 +
drivers/infiniband/hw/ehca/ehca_hca.c | 282 ++
drivers/infiniband/hw/ehca/ehca_irq.c | 710 ++++++
drivers/infiniband/hw/ehca/ehca_irq.h | 77
drivers/infiniband/hw/ehca/ehca_iverbs.h | 181 +
drivers/infiniband/hw/ehca/ehca_main.c | 966 ++++++++
drivers/infiniband/hw/ehca/ehca_mcast.c | 194 +
drivers/infiniband/hw/ehca/ehca_mrmw.c | 2474 ++++++++++++++++++++++
drivers/infiniband/hw/ehca/ehca_mrmw.h | 143 +
drivers/infiniband/hw/ehca/ehca_pd.c | 118 +
drivers/infiniband/hw/ehca/ehca_qes.h | 274 ++
drivers/infiniband/hw/ehca/ehca_qp.c | 1565 +++++++++++++
drivers/infiniband/hw/ehca/ehca_reqs.c | 683 ++++++
drivers/infiniband/hw/ehca/ehca_sqp.c | 123 +
drivers/infiniband/hw/ehca/ehca_tools.h | 411 +++
drivers/infiniband/hw/ehca/ehca_uverbs.c | 391 +++
drivers/infiniband/hw/ehca/hcp_if.c | 1476 +++++++++++++
drivers/infiniband/hw/ehca/hcp_if.h | 330 ++
drivers/infiniband/hw/ehca/hcp_phyp.c | 92
drivers/infiniband/hw/ehca/hcp_phyp.h | 95
drivers/infiniband/hw/ehca/hipz_fns.h | 68
drivers/infiniband/hw/ehca/hipz_fns_core.h | 122 +
drivers/infiniband/hw/ehca/hipz_hw.h | 395 +++
drivers/infiniband/hw/ehca/ipz_pt_fn.c | 177 +
drivers/infiniband/hw/ehca/ipz_pt_fn.h | 254 ++
31 files changed, 13183 insertions(+)
^ permalink raw reply
* [PATCH] Export PowerPC atomic operations to userspace
From: Brent Cook @ 2006-05-15 16:54 UTC (permalink / raw)
To: linuxppc-embedded
The atomic operations in asm/atomic.h are really useful from userspace too.
Other architectures (i386, x86_64, mips) export these to userspace, but the
powerpc versions are guarded by __KERNEL__ for some reason. Can we remove
these if there is no good reason to guard them?
Signed-off-by: Brent Cook <bcook@bpointsys.com>
Index: linux-2.6-bps/include/asm-powerpc/atomic.h
===================================================================
--- linux-2.6-bps/include/asm-powerpc/atomic.h (revision 77)
+++ linux-2.6-bps/include/asm-powerpc/atomic.h (working copy)
@@ -7,7 +7,6 @@
typedef struct { volatile int counter; } atomic_t;
-#ifdef __KERNEL__
#include <linux/compiler.h>
#include <asm/synch.h>
#include <asm/asm-compat.h>
@@ -414,5 +413,4 @@
#endif /* __powerpc64__ */
#include <asm-generic/atomic.h>
-#endif /* __KERNEL__ */
#endif /* _ASM_POWERPC_ATOMIC_H_ */
^ permalink raw reply
* MPC8248 goes berserk when printing messages on the SMC console
From: Laurent Pinchart @ 2006-05-15 16:28 UTC (permalink / raw)
To: linuxppc-embedded
Hi everybody,
I'm trying to convince an MPC8248 board to boot Linux, and I'm having problems
with the SMC console. I posted an e-mail on linuxppc-embedded regarding
BDI2000 problems earlier today. The problem hasn't been solved, but I've been
able to work around it by adding an infinite loop in the kernel code at the
point where I want to break, halting the execution when the loop has been
reached, and using gdb from there.
The SMC console works fine in U-Boot. 115200 8N1, no problem there.
When booting a Linux 2.6.16 kernel, the processor doesn't print anything on
the console and seems to hang. Further analysis with a BDI2000 reveals the
following.
- The CPU is stuck in cpm_uart_console_write, checking for
while ((bdp->cbd_sc & BD_SC_READY) != 0);
At that point, most of the external and internal memory read as 0 (SDRAM,
flash, ...). A few memory locations contain values different than 0, and I
suspect those to come directly from the cache.
- If I break right before setting the BD_SC_READY flag and run using stepi,
random characters are sent on the serial port. Measuring the baudrate using
an oscilloscope confirms that it has been correctly configured. Only the data
is wrong.
- If I remove the lines
bdp->cbd_sc |= BD_SC_READY;
the kernel runs until it panics when mounting the root partition.
- If I disable the SMC transmitter (by not setting SMCMR_TEN in SMCMR), the
kernel hangs when wrapping around to the first buffer descriptor, waiting for
it to be ready, which of course never happens.
This seems to confirm that the CPM does something very armful to the CPU. I'm
completely lost here, and would appreciate any help.
Best regards,
Laurent Pinchart
^ permalink raw reply
* Re: [PATCH] Fix pSeries identification in prom_init.c
From: Michael Neuling @ 2006-05-15 16:17 UTC (permalink / raw)
To: Benjamin Herrenschmidt; +Cc: linuxppc-dev list, Paul Mackerras, segher
In-Reply-To: <1147672282.21291.92.camel@localhost.localdomain>
> > The OF trampoline code prom_init.c still needs to identify IBM
> > pSeries (PAPR) machines in order to run some platform specific code
> > on them like instanciating the TCE tables. The code doing that
> > detection was changed recently in 2.6.17 early stages but was done
> > slightly incorrectly. It should be testing for an exact match of
> > "chrp" and it currently tests for anything that begins with
> > "chrp". That means it will incorrectly match with platforms using
> > Maple-like device-trees and have open firmware. This fixes it by
> > using strcmp instead of strncmp to match what the actual platform
> > detection code does.
>
> Michael, I noticed you changed strcmp to strncmp, any reason why you
> did that ?
To be safe if we are returned a non terminated string. I'd not realised
the case you've mentioned.
How much we should trust firmware? With strcpy, should we explicitly
terminate the string first (I removed one of these originally)? Patch
below, compiled not run.
-
The OF trampoline code prom_init.c still needs to identify IBM pSeries
(PAPR) machines in order to run some platform specific code on them like
instantiating the TCE tables. The code doing that detection was changed
recently in 2.6.17 early stages but was done slightly incorrectly. It
should be testing for an exact match of "chrp" and it currently tests
for anything that begins with "chrp". That means it will incorrectly
match with platforms using Maple-like device-trees and have open
firmware. This fixes it by using strcmp instead of strncmp to match what
the actual platform detection code does.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michael Neuling <mikey@neuling.org>
---
arch/powerpc/kernel/prom_init.c | 3 ++-
1 files changed, 2 insertions(+), 1 deletion(-)
Index: linux-2.6-powerpc/arch/powerpc/kernel/prom_init.c
===================================================================
--- linux-2.6-powerpc.orig/arch/powerpc/kernel/prom_init.c
+++ linux-2.6-powerpc/arch/powerpc/kernel/prom_init.c
@@ -1636,7 +1636,8 @@ static int __init prom_find_machine_type
compat, sizeof(compat)-1);
if (len <= 0)
return PLATFORM_GENERIC;
- if (strncmp(compat, RELOC("chrp"), 4))
+ compat[len] = 0;
+ if (strcmp(compat, RELOC("chrp")))
return PLATFORM_GENERIC;
/* Default to pSeries. We need to know if we are running LPAR */
^ permalink raw reply
* [dtc][PATCH] Fix ftdump data walking
From: Jimi Xenidis @ 2006-05-15 15:47 UTC (permalink / raw)
To: linuxppc-dev
This patch fixes a pointer addition bug in ftdump and handles endian
issue.
Signed-off-by: Jimi Xenidis <jimix@watson.ibm.com>
---
diff --git a/ftdump.c b/ftdump.c
index 4d4354b..7e643c8 100644
--- a/ftdump.c
+++ b/ftdump.c
@@ -67,7 +67,8 @@ static void print_data(const void *data,
} else if ((len % 4) == 0) {
printf(" = <");
for (i = 0; i < len; i += 4)
- printf("%08x%s", *((uint32_t *)data + i),
+ printf("%08x%s",
+ be32_to_cpu(*((uint32_t *)((ulong)data + i))),
i < (len - 4) ? " " : "");
printf(">");
} else {
^ permalink raw reply related
* [PATCH] Create /proc/rtas,/proc/ppc64/rtas if RTAS exists.
From: mostrows @ 2006-05-15 12:51 UTC (permalink / raw)
To: paulus, benh, linuxppc-dev
Use the existence of RTAS device tree node to determine if
/proc/rtas. /proc/ppc64/rtas are to be created. Using machine type
is not reliable (i.e. Maple-like machines may have RTAS).
--
Signed-off-by: Michal Ostrowski <mostrows@watson.ibm.com>
---
arch/powerpc/kernel/proc_ppc64.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
d8054b07bbf12e0178ec5ad7ce6a750a3af53064
diff --git a/arch/powerpc/kernel/proc_ppc64.c b/arch/powerpc/kernel/proc_ppc64.c
index 3c2cf66..2b87f82 100644
--- a/arch/powerpc/kernel/proc_ppc64.c
+++ b/arch/powerpc/kernel/proc_ppc64.c
@@ -52,7 +52,7 @@ static int __init proc_ppc64_create(void
if (!root)
return 1;
- if (!machine_is(pseries) && !machine_is(cell))
+ if (!of_find_node_by_path("/rtas"))
return 0;
if (!proc_mkdir("rtas", root))
--
1.1.4.g0b63-dirty
^ permalink raw reply related
* Re: [PATCH 5/6] Have ia64 use add_active_range() and free_area_init_nodes
From: Mel Gorman @ 2006-05-15 12:27 UTC (permalink / raw)
To: Andrew Morton
Cc: davej, tony.luck, linuxppc-dev, ak, bob.picco, linux-kernel,
linux-mm
In-Reply-To: <20060514203158.216a966e.akpm@osdl.org>
On (14/05/06 20:31), Andrew Morton didst pronounce:
> Mel Gorman <mel@csn.ul.ie> wrote:
> >
> > Size zones and holes in an architecture independent manner for ia64.
> >
>
> This one makes my ia64 die very early in boot. The trace is pretty useless.
>
> config at http://www.zip.com.au/~akpm/linux/patches/stuff/config-ia64
>
> <log snipped>
Curses. When I tried to reproduce this, the machine booted with my default
config but died before initialising the console with your config. The machine
is far away so I can't see the screen or restart the machine remotely so
I can only assume it is dying for the same reasons yours did.
> Note the misaligned pfns.
>
> Andy's (misspelled) CONFIG_UNALIGNED_ZONE_BOUNDRIES patch didn't actually
> include an update to any Kconfig files. But hacking that in by hand didn't
> help.
It would not have helped in this case because the zone boundaries would still
be in the wrong place for ia64. Below is a patch that aligns the zones on
all architectures that use CONFIG_ARCH_POPULATES_NODE_MAP . That is currently
i386, x86_64, powerpc, ppc and ia64. It does *not* align pgdat->node_start_pfn
but I don't believe that it is necessary.
I can't test it on ia64 until I get someone to restart the machine. The patch
compiles and is currently boot-testing on a range of other machines. I hope
to know within 5-6 hours if everything is ok.
diff -rup -X /usr/src/patchset-0.5/bin//dontdiff linux-2.6.17-rc4-mm4-clean/mm/page_alloc.c linux-2.6.17-rc4-mm4-ia64_force_alignment/mm/page_alloc.c
--- linux-2.6.17-rc4-mm4-clean/mm/page_alloc.c 2006-05-15 10:37:55.000000000 +0100
+++ linux-2.6.17-rc4-mm4-ia64_force_alignment/mm/page_alloc.c 2006-05-15 13:10:42.000000000 +0100
@@ -2640,14 +2640,20 @@ void __init free_area_init_nodes(unsigne
{
unsigned long nid;
int zone_index;
+ unsigned long lowest_pfn = find_min_pfn_with_active_regions();
+
+ lowest_pfn = zone_boundary_align_pfn(lowest_pfn);
+ arch_max_dma_pfn = zone_boundary_align_pfn(arch_max_dma_pfn);
+ arch_max_dma32_pfn = zone_boundary_align_pfn(arch_max_dma32_pfn);
+ arch_max_low_pfn = zone_boundary_align_pfn(arch_max_low_pfn);
+ arch_max_high_pfn = zone_boundary_align_pfn(arch_max_high_pfn);
/* Record where the zone boundaries are */
memset(arch_zone_lowest_possible_pfn, 0,
sizeof(arch_zone_lowest_possible_pfn));
memset(arch_zone_highest_possible_pfn, 0,
sizeof(arch_zone_highest_possible_pfn));
- arch_zone_lowest_possible_pfn[ZONE_DMA] =
- find_min_pfn_with_active_regions();
+ arch_zone_lowest_possible_pfn[ZONE_DMA] = lowest_pfn;
arch_zone_highest_possible_pfn[ZONE_DMA] = arch_max_dma_pfn;
arch_zone_highest_possible_pfn[ZONE_DMA32] = arch_max_dma32_pfn;
arch_zone_highest_possible_pfn[ZONE_NORMAL] = arch_max_low_pfn;
^ permalink raw reply
* Re: windfarm for PM72/PM73/RM31
From: Étienne Bersac @ 2006-05-15 11:39 UTC (permalink / raw)
To: Robin H. Johnson; +Cc: linuxppc-dev
In-Reply-To: <20060515035151.GB25086@curie-int.vc.shawcable.net>
Ok,
Let me wish you a good recovering :)
Étienne.
^ permalink raw reply
* Re: [PATCH 5/6] Have ia64 use add_active_range() and free_area_init_nodes
From: Andy Whitcroft @ 2006-05-15 11:02 UTC (permalink / raw)
To: KAMEZAWA Hiroyuki
Cc: akpm, nickpiggin, tony.luck, davej, mel, linux-kernel, bob.picco,
ak, linuxppc-dev, linux-mm
In-Reply-To: <20060515192918.c3e2e895.kamezawa.hiroyu@jp.fujitsu.com>
KAMEZAWA Hiroyuki wrote:
> On Mon, 15 May 2006 11:19:27 +0100
> Andy Whitcroft <apw@shadowen.org> wrote:
>
>
>>Nick Piggin wrote:
>>
>>>Andy Whitcroft wrote:
>>>
>>>
>>>>Interesting. You are correct there was no config component, at the time
>>>>I didn't have direct evidence that any architecture needed it, only that
>>>>we had an unchecked requirement on zones, a requirement that had only
>>>>recently arrived with the changes to free buddy detection. I note that
>>>
>>>
>>>Recently arrived? Over a year ago with the no-buddy-bitmap patches,
>>>right? Just checking because I that's what I'm assuming broke it...
>>
>>Yep, sorry I forget I was out of the game for 6 months! And yes that
>>was when the requirements were altered.
>>
>
> When no-bitmap-buddy patches was included,
>
> 1. bad_range() is not covered by CONFIG_VM_DEBUG. It always worked.
> ==
> static int bad_range(struct zone *zone, struct page *page)
> {
> if (page_to_pfn(page) >= zone->zone_start_pfn + zone->spanned_pages)
> return 1;
> if (page_to_pfn(page) < zone->zone_start_pfn)
> return 1;
> ==
> And , this code
> ==
> buddy = __page_find_buddy(page, page_idx, order);
>
> if (bad_range(zone, buddy))
> break;
> ==
>
> checked whether buddy is in zone and guarantees it to have page struct.
>
>
> But clean-up/speed-up codes vanished these checks. (I don't know when this occurs)
> Sorry for misses these things.
Heh, sorry to make it sound like it was you who was responsible.
-apw
^ permalink raw reply
* booting linux on IBM750 evaluation board
From: sharath kumar @ 2006-05-15 10:41 UTC (permalink / raw)
To: linuxppc-embedded
HI
i am working on IBM750 evaluation baord from IBM. It
has Marvel chipset.
I only have pibs running on this board.
I want to boot linux on this baord.
can anyone tell the procedure
Thanks and Regards
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
^ permalink raw reply
* Re: [PATCH 5/6] Have ia64 use add_active_range() and free_area_init_nodes
From: KAMEZAWA Hiroyuki @ 2006-05-15 10:47 UTC (permalink / raw)
To: KAMEZAWA Hiroyuki
Cc: akpm, nickpiggin, tony.luck, davej, mel, linux-kernel, bob.picco,
ak, linuxppc-dev, linux-mm
In-Reply-To: <20060515192918.c3e2e895.kamezawa.hiroyu@jp.fujitsu.com>
On Mon, 15 May 2006 19:29:18 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:
> On Mon, 15 May 2006 11:19:27 +0100
> Andy Whitcroft <apw@shadowen.org> wrote:
>
> > Nick Piggin wrote:
> > > Andy Whitcroft wrote:
> > >
> > >> Interesting. You are correct there was no config component, at the time
> > >> I didn't have direct evidence that any architecture needed it, only that
> > >> we had an unchecked requirement on zones, a requirement that had only
> > >> recently arrived with the changes to free buddy detection. I note that
> > >
> > >
> > > Recently arrived? Over a year ago with the no-buddy-bitmap patches,
> > > right? Just checking because I that's what I'm assuming broke it...
> >
> > Yep, sorry I forget I was out of the game for 6 months! And yes that
> > was when the requirements were altered.
> >
> When no-bitmap-buddy patches was included,
>
> 1. bad_range() is not covered by CONFIG_VM_DEBUG. It always worked.
> ==
> static int bad_range(struct zone *zone, struct page *page)
> {
> if (page_to_pfn(page) >= zone->zone_start_pfn + zone->spanned_pages)
> return 1;
> if (page_to_pfn(page) < zone->zone_start_pfn)
> return 1;
> ==
> And , this code
> ==
> buddy = __page_find_buddy(page, page_idx, order);
>
> if (bad_range(zone, buddy))
> break;
> ==
>
> checked whether buddy is in zone and guarantees it to have page struct.
>
>
> But clean-up/speed-up codes vanished these checks. (I don't know when this occurs)
> Sorry for misses these things.
>
One more point
When above no-bitmap patches was included, the user of not-aligned zones
are only ia64, I think. Because ia64 used virtual mem_map, page_to_pfn(page)
on CONFIG_DISCONTIG_MEM doesn't access page struct itself.
#define page_to_pfn(page) (page - vmemmap)
So, it didn't panic. ia64/vmemmap was safe.
If other archs used not-aligned zone + CONFIG_DISCONTIGMEM,
not-aligned-zones problem would come out earlier.
-Kame
^ permalink raw reply
* Re: [PATCH 5/6] Have ia64 use add_active_range() and free_area_init_nodes
From: KAMEZAWA Hiroyuki @ 2006-05-15 10:29 UTC (permalink / raw)
To: Andy Whitcroft
Cc: akpm, nickpiggin, tony.luck, davej, mel, linux-kernel, bob.picco,
ak, linuxppc-dev, linux-mm
In-Reply-To: <446855AF.1090100@shadowen.org>
On Mon, 15 May 2006 11:19:27 +0100
Andy Whitcroft <apw@shadowen.org> wrote:
> Nick Piggin wrote:
> > Andy Whitcroft wrote:
> >
> >> Interesting. You are correct there was no config component, at the time
> >> I didn't have direct evidence that any architecture needed it, only that
> >> we had an unchecked requirement on zones, a requirement that had only
> >> recently arrived with the changes to free buddy detection. I note that
> >
> >
> > Recently arrived? Over a year ago with the no-buddy-bitmap patches,
> > right? Just checking because I that's what I'm assuming broke it...
>
> Yep, sorry I forget I was out of the game for 6 months! And yes that
> was when the requirements were altered.
>
When no-bitmap-buddy patches was included,
1. bad_range() is not covered by CONFIG_VM_DEBUG. It always worked.
==
static int bad_range(struct zone *zone, struct page *page)
{
if (page_to_pfn(page) >= zone->zone_start_pfn + zone->spanned_pages)
return 1;
if (page_to_pfn(page) < zone->zone_start_pfn)
return 1;
==
And , this code
==
buddy = __page_find_buddy(page, page_idx, order);
if (bad_range(zone, buddy))
break;
==
checked whether buddy is in zone and guarantees it to have page struct.
But clean-up/speed-up codes vanished these checks. (I don't know when this occurs)
Sorry for misses these things.
-Kame
^ permalink raw reply
* Re: [PATCH 5/6] Have ia64 use add_active_range() and free_area_init_nodes
From: Andy Whitcroft @ 2006-05-15 10:19 UTC (permalink / raw)
To: Nick Piggin
Cc: Andrew Morton, davej, tony.luck, linuxppc-dev, Mel Gorman,
linux-kernel, bob.picco, ak, linux-mm
In-Reply-To: <44685123.7040501@yahoo.com.au>
Nick Piggin wrote:
> Andy Whitcroft wrote:
>
>> Interesting. You are correct there was no config component, at the time
>> I didn't have direct evidence that any architecture needed it, only that
>> we had an unchecked requirement on zones, a requirement that had only
>> recently arrived with the changes to free buddy detection. I note that
>
>
> Recently arrived? Over a year ago with the no-buddy-bitmap patches,
> right? Just checking because I that's what I'm assuming broke it...
Yep, sorry I forget I was out of the game for 6 months! And yes that
was when the requirements were altered.
-apw
^ permalink raw reply
* Page fault when debugging a 2.6.16 kernel on MPC8248
From: Laurent Pinchart @ 2006-05-15 10:10 UTC (permalink / raw)
To: linuxppc-embedded
Hello everybody,
I'm trying to bring up Linux on a custom MPC8248 board.
I've been able to setup U-Boot correctly up to the point where I boot the
Linux kernel using bootm. The kernel doesn't print anything on the serial
console and hangs.
Thanks to a BDI2000, I've been able to get the following backtrace.
#0 cpm_uart_console_write (co=0xd,
s=0xc0203e4b "x version 2.6.16-dirty (laurent@pclaurent) (gcc version
3.4.1) #422 Thu May 11 13:30:44 CEST 2006\n<6>Technotrade S.A.",
count=1) at drivers/serial/cpm_uart/cpm_uart_core.c:1023
#1 0xc0014144 in __call_console_drivers (start=3, end=105) at
kernel/printk.c:368
#2 0xc00144f4 in release_console_sem () at kernel/printk.c:430
#3 0xc0014ce4 in register_console (console=0xc01c2e60) at kernel/printk.c:946
#4 0xc01f7308 in cpm_uart_console_init () at
drivers/serial/cpm_uart/cpm_uart_core.c:1149
#5 0xc01f673c in console_init () at drivers/char/tty_io.c:3181
#6 0xc01e85b8 in start_kernel () at init/main.c:495
#7 0x00000000 in ?? ()
At that point, all the memory read with the BDI2000 (internal memory, SDRAM,
SRAM, flash, ...) reads as 0.
Knowing that something was wrong with the memory, I tried to execute the
kernel step by step. Unfortunately, after breaking in start_kernel, the first
gdb "next" command to step over the call to printk(KERN_NOTICE) hung. Hitting
Ctrl-C in gdb returned the following backtrace.
#0 0xc0005f4c in __delay ()
#1 0xc0013b18 in panic (fmt=0x487ab0 <Address 0x487ab0 out of bounds>) at
include/asm/delay.h:42
#2 0xc001679c in do_exit (code=11) at kernel/exit.c:809
#3 0x2400c022 in ?? ()
#4 0xc0004b8c in die (str=0x3ff <Address 0x3ff out of bounds>, fp=0xc01e7c60,
err=11) at arch/ppc/kernel/traps.c:101
#5 0xc000bf3c in bad_page_fault (regs=0xc01e7c60, address=0, sig=11) at
arch/ppc/mm/fault.c:336
#6 0xc00046d0 in handle_page_fault ()
#7 0xc00046d0 in handle_page_fault ()
I'm lost here. Why does Linux page faults when executed step by step (or
rather using "next") but doesn't when running normally ? I suspect that this
could be related to the memory reading as 0 in cpm_uart_console_write.
Thanks in advance for all the help you can provide.
Best regards,
Laurent Pinchart
^ permalink raw reply
* Re: how to mount /dev/ram0 to /
From: tony @ 2006-05-15 9:45 UTC (permalink / raw)
To: linuxppc-embedded@ozlabs.org
SSByZWJ1aWxkIHRoZSBidXp5Ym94LGFuZCBhZGRlZCBzb21lIHRvb2xzKGxpa2UgZG9zMnVuaXgs
c3R0eSx1bml4MmRvcyxkYyxsb2Fka21wKSAsdGhlbiBpdCB3b3Jrcyhtb3VudCAvZGV2L3JhbTAg
YXV0b21hdGljYWxseSksbWF5YmUgaXQncyBzb21ldGhpbmcgYWJvdXQgdGhlIHRvb2xzIGFib3Zl
Lg0KDQp0aGFuayBXb2xmZ2FuZyBEZW5rIGFuZCBhbGwuLg0KDQqhoaGhoaGhoaGhoaGhoaGhdG9u
eQ0KoaGhoaGhoaGhoaGhoaGhoWhhbmd0b29AMTYzLmNvbQ0KoaGhoaGhoaGhoaGhoaGhoaGhoaEy
MDA2LTA1LTE1DQo=
^ permalink raw reply
* Re: [PATCH 5/6] Have ia64 use add_active_range() and free_area_init_nodes
From: Nick Piggin @ 2006-05-15 10:00 UTC (permalink / raw)
To: Andy Whitcroft
Cc: Andrew Morton, davej, tony.luck, linuxppc-dev, Mel Gorman,
linux-kernel, bob.picco, ak, linux-mm
In-Reply-To: <44683A09.2060404@shadowen.org>
Andy Whitcroft wrote:
> Interesting. You are correct there was no config component, at the time
> I didn't have direct evidence that any architecture needed it, only that
> we had an unchecked requirement on zones, a requirement that had only
> recently arrived with the changes to free buddy detection. I note that
Recently arrived? Over a year ago with the no-buddy-bitmap patches,
right? Just checking because I that's what I'm assuming broke it...
> MAX_ORDER is 17 for ia64 so that probabally accounts for the
> missalignment. It is clear that the reporting is slightly over-zelous
> as I am reporting zero-sized zones. I'll get that fixed and patch to
> you. I'll also have a look at the patch as added to -mm and try and get
> the rest of the spelling sorted :-/.
>
> I'll go see if we currently have a machine to test this config on.
--
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox