aboutsummaryrefslogtreecommitdiff
path: root/drivers/infiniband/hw/cxgb3
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/infiniband/hw/cxgb3')
-rw-r--r--drivers/infiniband/hw/cxgb3/Kconfig2
-rw-r--r--drivers/infiniband/hw/cxgb3/Makefile6
-rw-r--r--drivers/infiniband/hw/cxgb3/cxio_dbg.c1
-rw-r--r--drivers/infiniband/hw/cxgb3/cxio_hal.c92
-rw-r--r--drivers/infiniband/hw/cxgb3/cxio_hal.h26
-rw-r--r--drivers/infiniband/hw/cxgb3/cxio_resource.c79
-rw-r--r--drivers/infiniband/hw/cxgb3/cxio_wr.h53
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch.c117
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch.h31
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_cm.c469
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_cm.h8
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_ev.c30
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_mem.c22
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_provider.c198
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_provider.h12
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_qp.c235
-rw-r--r--drivers/infiniband/hw/cxgb3/iwch_user.h8
17 files changed, 850 insertions, 539 deletions
diff --git a/drivers/infiniband/hw/cxgb3/Kconfig b/drivers/infiniband/hw/cxgb3/Kconfig
index 2acec3fadf6..2b6352b8548 100644
--- a/drivers/infiniband/hw/cxgb3/Kconfig
+++ b/drivers/infiniband/hw/cxgb3/Kconfig
@@ -10,7 +10,7 @@ config INFINIBAND_CXGB3
our website at <http://www.chelsio.com>.
For customer support, please visit our customer support page at
- <http://www.chelsio.com/support.htm>.
+ <http://www.chelsio.com/support.html>.
Please send feedback to <linux-bugs@chelsio.com>.
diff --git a/drivers/infiniband/hw/cxgb3/Makefile b/drivers/infiniband/hw/cxgb3/Makefile
index 7e7b5a66f04..2761364185a 100644
--- a/drivers/infiniband/hw/cxgb3/Makefile
+++ b/drivers/infiniband/hw/cxgb3/Makefile
@@ -1,10 +1,8 @@
-EXTRA_CFLAGS += -Idrivers/net/cxgb3
+ccflags-y := -Idrivers/net/ethernet/chelsio/cxgb3
obj-$(CONFIG_INFINIBAND_CXGB3) += iw_cxgb3.o
iw_cxgb3-y := iwch_cm.o iwch_ev.o iwch_cq.o iwch_qp.o iwch_mem.o \
iwch_provider.o iwch.o cxio_hal.o cxio_resource.o
-ifdef CONFIG_INFINIBAND_CXGB3_DEBUG
-EXTRA_CFLAGS += -DDEBUG
-endif
+ccflags-$(CONFIG_INFINIBAND_CXGB3_DEBUG) += -DDEBUG
diff --git a/drivers/infiniband/hw/cxgb3/cxio_dbg.c b/drivers/infiniband/hw/cxgb3/cxio_dbg.c
index a8d24d53f30..8bca6b4ec9a 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_dbg.c
+++ b/drivers/infiniband/hw/cxgb3/cxio_dbg.c
@@ -31,6 +31,7 @@
*/
#ifdef DEBUG
#include <linux/types.h>
+#include <linux/slab.h>
#include "common.h"
#include "cxgb3_ioctl.h"
#include "cxio_hal.h"
diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.c b/drivers/infiniband/hw/cxgb3/cxio_hal.c
index f6d5747153a..de1c61b417d 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.c
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.c
@@ -37,6 +37,7 @@
#include <linux/spinlock.h>
#include <linux/pci.h>
#include <linux/dma-mapping.h>
+#include <linux/slab.h>
#include <net/net_namespace.h>
#include "cxio_resource.h"
@@ -109,7 +110,6 @@ int cxio_hal_cq_op(struct cxio_rdev *rdev_p, struct t3_cq *cq,
while (!CQ_VLD_ENTRY(rptr, cq->size_log2, cqe)) {
udelay(1);
if (i++ > 1000000) {
- BUG_ON(1);
printk(KERN_ERR "%s: stalled rnic\n",
rdev_p->dev_name);
return -EIO;
@@ -152,29 +152,30 @@ static int cxio_hal_clear_qp_ctx(struct cxio_rdev *rdev_p, u32 qpid)
sge_cmd = qpid << 8 | 3;
wqe->sge_cmd = cpu_to_be64(sge_cmd);
skb->priority = CPL_PRIORITY_CONTROL;
- return (cxgb3_ofld_send(rdev_p->t3cdev_p, skb));
+ return iwch_cxgb3_ofld_send(rdev_p->t3cdev_p, skb);
}
-int cxio_create_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq)
+int cxio_create_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq, int kernel)
{
struct rdma_cq_setup setup;
int size = (1UL << (cq->size_log2)) * sizeof(struct t3_cqe);
+ size += 1; /* one extra page for storing cq-in-err state */
cq->cqid = cxio_hal_get_cqid(rdev_p->rscp);
if (!cq->cqid)
return -ENOMEM;
- cq->sw_queue = kzalloc(size, GFP_KERNEL);
- if (!cq->sw_queue)
- return -ENOMEM;
- cq->queue = dma_alloc_coherent(&(rdev_p->rnic_info.pdev->dev),
- (1UL << (cq->size_log2)) *
- sizeof(struct t3_cqe),
+ if (kernel) {
+ cq->sw_queue = kzalloc(size, GFP_KERNEL);
+ if (!cq->sw_queue)
+ return -ENOMEM;
+ }
+ cq->queue = dma_alloc_coherent(&(rdev_p->rnic_info.pdev->dev), size,
&(cq->dma_addr), GFP_KERNEL);
if (!cq->queue) {
kfree(cq->sw_queue);
return -ENOMEM;
}
- pci_unmap_addr_set(cq, mapping, cq->dma_addr);
+ dma_unmap_addr_set(cq, mapping, cq->dma_addr);
memset(cq->queue, 0, size);
setup.id = cq->cqid;
setup.base_addr = (u64) (cq->dma_addr);
@@ -188,6 +189,7 @@ int cxio_create_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq)
return (rdev_p->t3cdev_p->ctl(rdev_p->t3cdev_p, RDMA_CQ_SETUP, &setup));
}
+#ifdef notyet
int cxio_resize_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq)
{
struct rdma_cq_setup setup;
@@ -199,6 +201,7 @@ int cxio_resize_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq)
setup.ovfl_mode = 1;
return (rdev_p->t3cdev_p->ctl(rdev_p->t3cdev_p, RDMA_CQ_SETUP, &setup));
}
+#endif
static u32 get_qpid(struct cxio_rdev *rdev_p, struct cxio_ucontext *uctx)
{
@@ -297,7 +300,7 @@ int cxio_create_qp(struct cxio_rdev *rdev_p, u32 kernel_domain,
goto err4;
memset(wq->queue, 0, depth * sizeof(union t3_wr));
- pci_unmap_addr_set(wq, mapping, wq->dma_addr);
+ dma_unmap_addr_set(wq, mapping, wq->dma_addr);
wq->doorbell = (void __iomem *)rdev_p->rnic_info.kdb_addr;
if (!kernel_domain)
wq->udb = (u64)rdev_p->rnic_info.udbell_physbase +
@@ -325,7 +328,7 @@ int cxio_destroy_cq(struct cxio_rdev *rdev_p, struct t3_cq *cq)
dma_free_coherent(&(rdev_p->rnic_info.pdev->dev),
(1UL << (cq->size_log2))
* sizeof(struct t3_cqe), cq->queue,
- pci_unmap_addr(cq, mapping));
+ dma_unmap_addr(cq, mapping));
cxio_hal_put_cqid(rdev_p->rscp, cq->cqid);
return err;
}
@@ -336,7 +339,7 @@ int cxio_destroy_qp(struct cxio_rdev *rdev_p, struct t3_wq *wq,
dma_free_coherent(&(rdev_p->rnic_info.pdev->dev),
(1UL << (wq->size_log2))
* sizeof(union t3_wr), wq->queue,
- pci_unmap_addr(wq, mapping));
+ dma_unmap_addr(wq, mapping));
kfree(wq->sq);
cxio_hal_rqtpool_free(rdev_p, wq->rq_addr, (1UL << wq->rq_size_log2));
kfree(wq->rq);
@@ -410,6 +413,7 @@ int cxio_flush_sq(struct t3_wq *wq, struct t3_cq *cq, int count)
ptr = wq->sq_rptr + count;
sqp = wq->sq + Q_PTR2IDX(ptr, wq->sq_size_log2);
while (ptr != wq->sq_wptr) {
+ sqp->signaled = 0;
insert_sq_cqe(wq, cq, sqp);
ptr++;
sqp = wq->sq + Q_PTR2IDX(ptr, wq->sq_size_log2);
@@ -450,7 +454,7 @@ static int cqe_completes_wr(struct t3_cqe *cqe, struct t3_wq *wq)
if ((CQE_OPCODE(*cqe) == T3_READ_RESP) && SQ_TYPE(*cqe))
return 0;
- if ((CQE_OPCODE(*cqe) == T3_SEND) && RQ_TYPE(*cqe) &&
+ if (CQE_SEND_OPCODE(*cqe) && RQ_TYPE(*cqe) &&
Q_EMPTY(wq->rq_rptr, wq->rq_wptr))
return 0;
@@ -536,7 +540,7 @@ static int cxio_hal_init_ctrl_qp(struct cxio_rdev *rdev_p)
err = -ENOMEM;
goto err;
}
- pci_unmap_addr_set(&rdev_p->ctrl_qp, mapping,
+ dma_unmap_addr_set(&rdev_p->ctrl_qp, mapping,
rdev_p->ctrl_qp.dma_addr);
rdev_p->ctrl_qp.doorbell = (void __iomem *)rdev_p->rnic_info.kdb_addr;
memset(rdev_p->ctrl_qp.workq, 0,
@@ -571,7 +575,7 @@ static int cxio_hal_init_ctrl_qp(struct cxio_rdev *rdev_p)
(unsigned long long) rdev_p->ctrl_qp.dma_addr,
rdev_p->ctrl_qp.workq, 1 << T3_CTRL_QP_SIZE_LOG2);
skb->priority = CPL_PRIORITY_CONTROL;
- return (cxgb3_ofld_send(rdev_p->t3cdev_p, skb));
+ return iwch_cxgb3_ofld_send(rdev_p->t3cdev_p, skb);
err:
kfree_skb(skb);
return err;
@@ -582,13 +586,13 @@ static int cxio_hal_destroy_ctrl_qp(struct cxio_rdev *rdev_p)
dma_free_coherent(&(rdev_p->rnic_info.pdev->dev),
(1UL << T3_CTRL_QP_SIZE_LOG2)
* sizeof(union t3_wr), rdev_p->ctrl_qp.workq,
- pci_unmap_addr(&rdev_p->ctrl_qp, mapping));
+ dma_unmap_addr(&rdev_p->ctrl_qp, mapping));
return cxio_hal_clear_qp_ctx(rdev_p, T3_CTRL_QP_ID);
}
/* write len bytes of data into addr (32B aligned address)
* If data is NULL, clear len byte of memory to zero.
- * caller aquires the ctrl_qp lock before the call
+ * caller acquires the ctrl_qp lock before the call
*/
static int cxio_hal_ctrl_qp_write_mem(struct cxio_rdev *rdev_p, u32 addr,
u32 len, void *data)
@@ -701,6 +705,9 @@ static int __cxio_tpt_op(struct cxio_rdev *rdev_p, u32 reset_tpt_entry,
u32 stag_idx;
u32 wptr;
+ if (cxio_fatal_error(rdev_p))
+ return -EIO;
+
stag_state = stag_state > 0;
stag_idx = (*stag) >> 8;
@@ -725,17 +732,15 @@ static int __cxio_tpt_op(struct cxio_rdev *rdev_p, u32 reset_tpt_entry,
V_TPT_STAG_TYPE(type) | V_TPT_PDID(pdid));
BUG_ON(page_size >= 28);
tpt.flags_pagesize_qpid = cpu_to_be32(V_TPT_PERM(perm) |
- F_TPT_MW_BIND_ENABLE |
- V_TPT_ADDR_TYPE((zbva ? TPT_ZBTO : TPT_VATO)) |
- V_TPT_PAGE_SIZE(page_size));
- tpt.rsvd_pbl_addr = reset_tpt_entry ? 0 :
- cpu_to_be32(V_TPT_PBL_ADDR(PBL_OFF(rdev_p, pbl_addr)>>3));
+ ((perm & TPT_MW_BIND) ? F_TPT_MW_BIND_ENABLE : 0) |
+ V_TPT_ADDR_TYPE((zbva ? TPT_ZBTO : TPT_VATO)) |
+ V_TPT_PAGE_SIZE(page_size));
+ tpt.rsvd_pbl_addr = cpu_to_be32(V_TPT_PBL_ADDR(PBL_OFF(rdev_p, pbl_addr)>>3));
tpt.len = cpu_to_be32(len);
tpt.va_hi = cpu_to_be32((u32) (to >> 32));
tpt.va_low_or_fbo = cpu_to_be32((u32) (to & 0xFFFFFFFFULL));
tpt.rsvd_bind_cnt_or_pstag = 0;
- tpt.rsvd_pbl_size = reset_tpt_entry ? 0 :
- cpu_to_be32(V_TPT_PBL_SIZE(pbl_size >> 2));
+ tpt.rsvd_pbl_size = cpu_to_be32(V_TPT_PBL_SIZE(pbl_size >> 2));
}
err = cxio_hal_ctrl_qp_write_mem(rdev_p,
stag_idx +
@@ -848,14 +853,16 @@ int cxio_rdma_init(struct cxio_rdev *rdev_p, struct t3_rdma_init_attr *attr)
wqe->qpcaps = attr->qpcaps;
wqe->ulpdu_size = cpu_to_be16(attr->tcp_emss);
wqe->rqe_count = cpu_to_be16(attr->rqe_count);
- wqe->flags_rtr_type = cpu_to_be16(attr->flags|V_RTR_TYPE(attr->rtr_type));
+ wqe->flags_rtr_type = cpu_to_be16(attr->flags |
+ V_RTR_TYPE(attr->rtr_type) |
+ V_CHAN(attr->chan));
wqe->ord = cpu_to_be32(attr->ord);
wqe->ird = cpu_to_be32(attr->ird);
wqe->qp_dma_addr = cpu_to_be64(attr->qp_dma_addr);
wqe->qp_dma_size = cpu_to_be32(attr->qp_dma_size);
wqe->irs = cpu_to_be32(attr->irs);
skb->priority = 0; /* 0=>ToeQ; 1=>CtrlQ */
- return (cxgb3_ofld_send(rdev_p->t3cdev_p, skb));
+ return iwch_cxgb3_ofld_send(rdev_p->t3cdev_p, skb);
}
void cxio_register_ev_cb(cxio_hal_ev_callback_func_t ev_cb)
@@ -938,6 +945,23 @@ int cxio_rdev_open(struct cxio_rdev *rdev_p)
if (!rdev_p->t3cdev_p)
rdev_p->t3cdev_p = dev2t3cdev(netdev_p);
rdev_p->t3cdev_p->ulp = (void *) rdev_p;
+
+ err = rdev_p->t3cdev_p->ctl(rdev_p->t3cdev_p, GET_EMBEDDED_INFO,
+ &(rdev_p->fw_info));
+ if (err) {
+ printk(KERN_ERR "%s t3cdev_p(%p)->ctl returned error %d.\n",
+ __func__, rdev_p->t3cdev_p, err);
+ goto err1;
+ }
+ if (G_FW_VERSION_MAJOR(rdev_p->fw_info.fw_vers) != CXIO_FW_MAJ) {
+ printk(KERN_ERR MOD "fatal firmware version mismatch: "
+ "need version %u but adapter has version %u\n",
+ CXIO_FW_MAJ,
+ G_FW_VERSION_MAJOR(rdev_p->fw_info.fw_vers));
+ err = -EINVAL;
+ goto err1;
+ }
+
err = rdev_p->t3cdev_p->ctl(rdev_p->t3cdev_p, RDMA_GET_PARAMS,
&(rdev_p->rnic_info));
if (err) {
@@ -1011,6 +1035,7 @@ err3:
err2:
cxio_hal_destroy_ctrl_qp(rdev_p);
err1:
+ rdev_p->t3cdev_p->ulp = NULL;
list_del(&rdev_p->entry);
return err;
}
@@ -1021,9 +1046,9 @@ void cxio_rdev_close(struct cxio_rdev *rdev_p)
cxio_hal_pblpool_destroy(rdev_p);
cxio_hal_rqtpool_destroy(rdev_p);
list_del(&rdev_p->entry);
- rdev_p->t3cdev_p->ulp = NULL;
cxio_hal_destroy_ctrl_qp(rdev_p);
cxio_hal_destroy_resource(rdev_p->rscp);
+ rdev_p->t3cdev_p->ulp = NULL;
}
}
@@ -1204,11 +1229,12 @@ int cxio_poll_cq(struct t3_wq *wq, struct t3_cq *cq, struct t3_cqe *cqe,
}
/* incoming SEND with no receive posted failures */
- if ((CQE_OPCODE(*hw_cqe) == T3_SEND) && RQ_TYPE(*hw_cqe) &&
+ if (CQE_SEND_OPCODE(*hw_cqe) && RQ_TYPE(*hw_cqe) &&
Q_EMPTY(wq->rq_rptr, wq->rq_wptr)) {
ret = -1;
goto skip_cqe;
}
+ BUG_ON((*cqe_flushed == 0) && !SW_CQE(*hw_cqe));
goto proc_cqe;
}
@@ -1223,6 +1249,13 @@ int cxio_poll_cq(struct t3_wq *wq, struct t3_cq *cq, struct t3_cqe *cqe,
* then we complete this with TPT_ERR_MSN and mark the wq in
* error.
*/
+
+ if (Q_EMPTY(wq->rq_rptr, wq->rq_wptr)) {
+ wq->error = 1;
+ ret = -1;
+ goto skip_cqe;
+ }
+
if (unlikely((CQE_WRID_MSN(*hw_cqe) != (wq->rq_rptr + 1)))) {
wq->error = 1;
hw_cqe->header |= htonl(V_CQE_STATUS(TPT_ERR_MSN));
@@ -1277,6 +1310,7 @@ proc_cqe:
cxio_hal_pblpool_free(wq->rdev,
wq->rq[Q_PTR2IDX(wq->rq_rptr,
wq->rq_size_log2)].pbl_addr, T3_STAG0_PBL_SIZE);
+ BUG_ON(Q_EMPTY(wq->rq_rptr, wq->rq_wptr));
wq->rq_rptr++;
}
diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.h b/drivers/infiniband/hw/cxgb3/cxio_hal.h
index 656fe47bc84..78fbe9ffe7f 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.h
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.h
@@ -34,6 +34,7 @@
#include <linux/list.h>
#include <linux/mutex.h>
+#include <linux/kfifo.h>
#include "t3_cpl.h"
#include "t3cdev.h"
@@ -52,7 +53,7 @@
#define T3_MAX_PBL_SIZE 256
#define T3_MAX_RQ_SIZE 1024
#define T3_MAX_QP_DEPTH (T3_MAX_RQ_SIZE-1)
-#define T3_MAX_CQ_DEPTH 8192
+#define T3_MAX_CQ_DEPTH 65536
#define T3_MAX_NUM_STAG (1<<15)
#define T3_MAX_MR_SIZE 0x100000000ULL
#define T3_PAGESIZE_MASK 0xffff000 /* 4KB-128MB */
@@ -61,6 +62,8 @@
#define T3_MAX_DEV_NAME_LEN 32
+#define CXIO_FW_MAJ 7
+
struct cxio_hal_ctrl_qp {
u32 wptr;
u32 rptr;
@@ -68,18 +71,18 @@ struct cxio_hal_ctrl_qp {
wait_queue_head_t waitq;/* wait for RspQ/CQE msg */
union t3_wr *workq; /* the work request queue */
dma_addr_t dma_addr; /* pci bus address of the workq */
- DECLARE_PCI_UNMAP_ADDR(mapping)
+ DEFINE_DMA_UNMAP_ADDR(mapping);
void __iomem *doorbell;
};
struct cxio_hal_resource {
- struct kfifo *tpt_fifo;
+ struct kfifo tpt_fifo;
spinlock_t tpt_fifo_lock;
- struct kfifo *qpid_fifo;
+ struct kfifo qpid_fifo;
spinlock_t qpid_fifo_lock;
- struct kfifo *cqid_fifo;
+ struct kfifo cqid_fifo;
spinlock_t cqid_fifo_lock;
- struct kfifo *pdid_fifo;
+ struct kfifo pdid_fifo;
spinlock_t pdid_fifo_lock;
};
@@ -108,8 +111,16 @@ struct cxio_rdev {
struct gen_pool *pbl_pool;
struct gen_pool *rqt_pool;
struct list_head entry;
+ struct ch_embedded_info fw_info;
+ u32 flags;
+#define CXIO_ERROR_FATAL 1
};
+static inline int cxio_fatal_error(struct cxio_rdev *rdev_p)
+{
+ return rdev_p->flags & CXIO_ERROR_FATAL;
+}
+
static inline int cxio_num_stags(struct cxio_rdev *rdev_p)
{
return min((int)T3_MAX_NUM_STAG, (int)((rdev_p->rnic_info.tpt_top - rdev_p->rnic_info.tpt_base) >> 5));
@@ -146,7 +157,7 @@ int cxio_rdev_open(struct cxio_rdev *rdev);
void cxio_rdev_close(struct cxio_rdev *rdev);
int cxio_hal_cq_op(struct cxio_rdev *rdev, struct t3_cq *cq,
enum t3_cq_opcode op, u32 credit);
-int cxio_create_cq(struct cxio_rdev *rdev, struct t3_cq *cq);
+int cxio_create_cq(struct cxio_rdev *rdev, struct t3_cq *cq, int kernel);
int cxio_destroy_cq(struct cxio_rdev *rdev, struct t3_cq *cq);
int cxio_resize_cq(struct cxio_rdev *rdev, struct t3_cq *cq);
void cxio_release_ucontext(struct cxio_rdev *rdev, struct cxio_ucontext *uctx);
@@ -183,6 +194,7 @@ void cxio_count_scqes(struct t3_cq *cq, struct t3_wq *wq, int *count);
void cxio_flush_hw_cq(struct t3_cq *cq);
int cxio_poll_cq(struct t3_wq *wq, struct t3_cq *cq, struct t3_cqe *cqe,
u8 *cqe_flushed, u64 *cookie, u32 *credit);
+int iwch_cxgb3_ofld_send(struct t3cdev *tdev, struct sk_buff *skb);
#define MOD "iw_cxgb3: "
#define PDBG(fmt, args...) pr_debug(MOD fmt, ## args)
diff --git a/drivers/infiniband/hw/cxgb3/cxio_resource.c b/drivers/infiniband/hw/cxgb3/cxio_resource.c
index bd233c08765..c40088ecf9f 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_resource.c
+++ b/drivers/infiniband/hw/cxgb3/cxio_resource.c
@@ -39,12 +39,12 @@
#include "cxio_resource.h"
#include "cxio_hal.h"
-static struct kfifo *rhdl_fifo;
+static struct kfifo rhdl_fifo;
static spinlock_t rhdl_fifo_lock;
#define RANDOM_SIZE 16
-static int __cxio_init_resource_fifo(struct kfifo **fifo,
+static int __cxio_init_resource_fifo(struct kfifo *fifo,
spinlock_t *fifo_lock,
u32 nr, u32 skip_low,
u32 skip_high,
@@ -55,50 +55,51 @@ static int __cxio_init_resource_fifo(struct kfifo **fifo,
u32 rarray[16];
spin_lock_init(fifo_lock);
- *fifo = kfifo_alloc(nr * sizeof(u32), GFP_KERNEL, fifo_lock);
- if (IS_ERR(*fifo))
+ if (kfifo_alloc(fifo, nr * sizeof(u32), GFP_KERNEL))
return -ENOMEM;
for (i = 0; i < skip_low + skip_high; i++)
- __kfifo_put(*fifo, (unsigned char *) &entry, sizeof(u32));
+ kfifo_in(fifo, (unsigned char *) &entry, sizeof(u32));
if (random) {
j = 0;
- random_bytes = random32();
+ random_bytes = prandom_u32();
for (i = 0; i < RANDOM_SIZE; i++)
rarray[i] = i + skip_low;
for (i = skip_low + RANDOM_SIZE; i < nr - skip_high; i++) {
if (j >= RANDOM_SIZE) {
j = 0;
- random_bytes = random32();
+ random_bytes = prandom_u32();
}
idx = (random_bytes >> (j * 2)) & 0xF;
- __kfifo_put(*fifo,
+ kfifo_in(fifo,
(unsigned char *) &rarray[idx],
sizeof(u32));
rarray[idx] = i;
j++;
}
for (i = 0; i < RANDOM_SIZE; i++)
- __kfifo_put(*fifo,
+ kfifo_in(fifo,
(unsigned char *) &rarray[i],
sizeof(u32));
} else
for (i = skip_low; i < nr - skip_high; i++)
- __kfifo_put(*fifo, (unsigned char *) &i, sizeof(u32));
+ kfifo_in(fifo, (unsigned char *) &i, sizeof(u32));
for (i = 0; i < skip_low + skip_high; i++)
- kfifo_get(*fifo, (unsigned char *) &entry, sizeof(u32));
+ if (kfifo_out_locked(fifo, (unsigned char *) &entry,
+ sizeof(u32), fifo_lock) != sizeof(u32))
+ break;
return 0;
}
-static int cxio_init_resource_fifo(struct kfifo **fifo, spinlock_t * fifo_lock,
+static int cxio_init_resource_fifo(struct kfifo *fifo, spinlock_t * fifo_lock,
u32 nr, u32 skip_low, u32 skip_high)
{
return (__cxio_init_resource_fifo(fifo, fifo_lock, nr, skip_low,
skip_high, 0));
}
-static int cxio_init_resource_fifo_random(struct kfifo **fifo,
+static int cxio_init_resource_fifo_random(struct kfifo *fifo,
spinlock_t * fifo_lock,
u32 nr, u32 skip_low, u32 skip_high)
{
@@ -113,15 +114,13 @@ static int cxio_init_qpid_fifo(struct cxio_rdev *rdev_p)
spin_lock_init(&rdev_p->rscp->qpid_fifo_lock);
- rdev_p->rscp->qpid_fifo = kfifo_alloc(T3_MAX_NUM_QP * sizeof(u32),
- GFP_KERNEL,
- &rdev_p->rscp->qpid_fifo_lock);
- if (IS_ERR(rdev_p->rscp->qpid_fifo))
+ if (kfifo_alloc(&rdev_p->rscp->qpid_fifo, T3_MAX_NUM_QP * sizeof(u32),
+ GFP_KERNEL))
return -ENOMEM;
for (i = 16; i < T3_MAX_NUM_QP; i++)
if (!(i & rdev_p->qpmask))
- __kfifo_put(rdev_p->rscp->qpid_fifo,
+ kfifo_in(&rdev_p->rscp->qpid_fifo,
(unsigned char *) &i, sizeof(u32));
return 0;
}
@@ -134,7 +133,7 @@ int cxio_hal_init_rhdl_resource(u32 nr_rhdl)
void cxio_hal_destroy_rhdl_resource(void)
{
- kfifo_free(rhdl_fifo);
+ kfifo_free(&rhdl_fifo);
}
/* nr_* must be power of 2 */
@@ -167,11 +166,11 @@ int cxio_hal_init_resource(struct cxio_rdev *rdev_p,
goto pdid_err;
return 0;
pdid_err:
- kfifo_free(rscp->cqid_fifo);
+ kfifo_free(&rscp->cqid_fifo);
cqid_err:
- kfifo_free(rscp->qpid_fifo);
+ kfifo_free(&rscp->qpid_fifo);
qpid_err:
- kfifo_free(rscp->tpt_fifo);
+ kfifo_free(&rscp->tpt_fifo);
tpt_err:
return -ENOMEM;
}
@@ -179,33 +178,37 @@ tpt_err:
/*
* returns 0 if no resource available
*/
-static u32 cxio_hal_get_resource(struct kfifo *fifo)
+static u32 cxio_hal_get_resource(struct kfifo *fifo, spinlock_t * lock)
{
u32 entry;
- if (kfifo_get(fifo, (unsigned char *) &entry, sizeof(u32)))
+ if (kfifo_out_locked(fifo, (unsigned char *) &entry, sizeof(u32), lock))
return entry;
else
return 0; /* fifo emptry */
}
-static void cxio_hal_put_resource(struct kfifo *fifo, u32 entry)
+static void cxio_hal_put_resource(struct kfifo *fifo, spinlock_t * lock,
+ u32 entry)
{
- BUG_ON(kfifo_put(fifo, (unsigned char *) &entry, sizeof(u32)) == 0);
+ BUG_ON(
+ kfifo_in_locked(fifo, (unsigned char *) &entry, sizeof(u32), lock)
+ == 0);
}
u32 cxio_hal_get_stag(struct cxio_hal_resource *rscp)
{
- return cxio_hal_get_resource(rscp->tpt_fifo);
+ return cxio_hal_get_resource(&rscp->tpt_fifo, &rscp->tpt_fifo_lock);
}
void cxio_hal_put_stag(struct cxio_hal_resource *rscp, u32 stag)
{
- cxio_hal_put_resource(rscp->tpt_fifo, stag);
+ cxio_hal_put_resource(&rscp->tpt_fifo, &rscp->tpt_fifo_lock, stag);
}
u32 cxio_hal_get_qpid(struct cxio_hal_resource *rscp)
{
- u32 qpid = cxio_hal_get_resource(rscp->qpid_fifo);
+ u32 qpid = cxio_hal_get_resource(&rscp->qpid_fifo,
+ &rscp->qpid_fifo_lock);
PDBG("%s qpid 0x%x\n", __func__, qpid);
return qpid;
}
@@ -213,35 +216,35 @@ u32 cxio_hal_get_qpid(struct cxio_hal_resource *rscp)
void cxio_hal_put_qpid(struct cxio_hal_resource *rscp, u32 qpid)
{
PDBG("%s qpid 0x%x\n", __func__, qpid);
- cxio_hal_put_resource(rscp->qpid_fifo, qpid);
+ cxio_hal_put_resource(&rscp->qpid_fifo, &rscp->qpid_fifo_lock, qpid);
}
u32 cxio_hal_get_cqid(struct cxio_hal_resource *rscp)
{
- return cxio_hal_get_resource(rscp->cqid_fifo);
+ return cxio_hal_get_resource(&rscp->cqid_fifo, &rscp->cqid_fifo_lock);
}
void cxio_hal_put_cqid(struct cxio_hal_resource *rscp, u32 cqid)
{
- cxio_hal_put_resource(rscp->cqid_fifo, cqid);
+ cxio_hal_put_resource(&rscp->cqid_fifo, &rscp->cqid_fifo_lock, cqid);
}
u32 cxio_hal_get_pdid(struct cxio_hal_resource *rscp)
{
- return cxio_hal_get_resource(rscp->pdid_fifo);
+ return cxio_hal_get_resource(&rscp->pdid_fifo, &rscp->pdid_fifo_lock);
}
void cxio_hal_put_pdid(struct cxio_hal_resource *rscp, u32 pdid)
{
- cxio_hal_put_resource(rscp->pdid_fifo, pdid);
+ cxio_hal_put_resource(&rscp->pdid_fifo, &rscp->pdid_fifo_lock, pdid);
}
void cxio_hal_destroy_resource(struct cxio_hal_resource *rscp)
{
- kfifo_free(rscp->tpt_fifo);
- kfifo_free(rscp->cqid_fifo);
- kfifo_free(rscp->qpid_fifo);
- kfifo_free(rscp->pdid_fifo);
+ kfifo_free(&rscp->tpt_fifo);
+ kfifo_free(&rscp->cqid_fifo);
+ kfifo_free(&rscp->qpid_fifo);
+ kfifo_free(&rscp->pdid_fifo);
kfree(rscp);
}
diff --git a/drivers/infiniband/hw/cxgb3/cxio_wr.h b/drivers/infiniband/hw/cxgb3/cxio_wr.h
index 04618f7bfbb..83d2e19d31a 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_wr.h
+++ b/drivers/infiniband/hw/cxgb3/cxio_wr.h
@@ -176,7 +176,7 @@ struct t3_send_wr {
struct t3_sge sgl[T3_MAX_SGE]; /* 4+ */
};
-#define T3_MAX_FASTREG_DEPTH 24
+#define T3_MAX_FASTREG_DEPTH 10
#define T3_MAX_FASTREG_FRAG 10
struct t3_fastreg_wr {
@@ -327,6 +327,11 @@ enum rdma_init_rtr_types {
#define V_RTR_TYPE(x) ((x) << S_RTR_TYPE)
#define G_RTR_TYPE(x) ((((x) >> S_RTR_TYPE)) & M_RTR_TYPE)
+#define S_CHAN 4
+#define M_CHAN 0x3
+#define V_CHAN(x) ((x) << S_CHAN)
+#define G_CHAN(x) ((((x) >> S_CHAN)) & M_CHAN)
+
struct t3_rdma_init_attr {
u32 tid;
u32 qpid;
@@ -346,6 +351,7 @@ struct t3_rdma_init_attr {
u16 flags;
u16 rqe_count;
u32 irs;
+ u32 chan;
};
struct t3_rdma_init_wr {
@@ -604,6 +610,12 @@ struct t3_cqe {
#define CQE_STATUS(x) (G_CQE_STATUS(be32_to_cpu((x).header)))
#define CQE_OPCODE(x) (G_CQE_OPCODE(be32_to_cpu((x).header)))
+#define CQE_SEND_OPCODE(x)( \
+ (G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND) || \
+ (G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND_WITH_SE) || \
+ (G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND_WITH_INV) || \
+ (G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND_WITH_SE_INV))
+
#define CQE_LEN(x) (be32_to_cpu((x).len))
/* used for RQ completion processing */
@@ -677,9 +689,9 @@ struct t3_swrq {
* A T3 WQ implements both the SQ and RQ.
*/
struct t3_wq {
- union t3_wr *queue; /* DMA accessable memory */
+ union t3_wr *queue; /* DMA accessible memory */
dma_addr_t dma_addr; /* DMA address for HW */
- DECLARE_PCI_UNMAP_ADDR(mapping) /* unmap kruft */
+ DEFINE_DMA_UNMAP_ADDR(mapping); /* unmap kruft */
u32 error; /* 1 once we go to ERROR */
u32 qpid;
u32 wptr; /* idx to next available WR slot */
@@ -706,7 +718,7 @@ struct t3_cq {
u32 wptr;
u32 size_log2;
dma_addr_t dma_addr;
- DECLARE_PCI_UNMAP_ADDR(mapping)
+ DEFINE_DMA_UNMAP_ADDR(mapping);
struct t3_cqe *queue;
struct t3_cqe *sw_queue;
u32 sw_rptr;
@@ -716,9 +728,40 @@ struct t3_cq {
#define CQ_VLD_ENTRY(ptr,size_log2,cqe) (Q_GENBIT(ptr,size_log2) == \
CQE_GENBIT(*cqe))
+struct t3_cq_status_page {
+ u32 cq_err;
+};
+
+static inline int cxio_cq_in_error(struct t3_cq *cq)
+{
+ return ((struct t3_cq_status_page *)
+ &cq->queue[1 << cq->size_log2])->cq_err;
+}
+
+static inline void cxio_set_cq_in_error(struct t3_cq *cq)
+{
+ ((struct t3_cq_status_page *)
+ &cq->queue[1 << cq->size_log2])->cq_err = 1;
+}
+
static inline void cxio_set_wq_in_error(struct t3_wq *wq)
{
- wq->queue->wq_in_err.err = 1;
+ wq->queue->wq_in_err.err |= 1;
+}
+
+static inline void cxio_disable_wq_db(struct t3_wq *wq)
+{
+ wq->queue->wq_in_err.err |= 2;
+}
+
+static inline void cxio_enable_wq_db(struct t3_wq *wq)
+{
+ wq->queue->wq_in_err.err &= ~2;
+}
+
+static inline int cxio_wq_db_enabled(struct t3_wq *wq)
+{
+ return !(wq->queue->wq_in_err.err & 2);
}
static inline struct t3_cqe *cxio_next_hw_cqe(struct t3_cq *cq)
diff --git a/drivers/infiniband/hw/cxgb3/iwch.c b/drivers/infiniband/hw/cxgb3/iwch.c
index 4489c89d671..8e77dc543dd 100644
--- a/drivers/infiniband/hw/cxgb3/iwch.c
+++ b/drivers/infiniband/hw/cxgb3/iwch.c
@@ -47,22 +47,62 @@ MODULE_DESCRIPTION("Chelsio T3 RDMA Driver");
MODULE_LICENSE("Dual BSD/GPL");
MODULE_VERSION(DRV_VERSION);
-cxgb3_cpl_handler_func t3c_handlers[NUM_CPL_CMDS];
-
static void open_rnic_dev(struct t3cdev *);
static void close_rnic_dev(struct t3cdev *);
+static void iwch_event_handler(struct t3cdev *, u32, u32);
struct cxgb3_client t3c_client = {
.name = "iw_cxgb3",
.add = open_rnic_dev,
.remove = close_rnic_dev,
.handlers = t3c_handlers,
- .redirect = iwch_ep_redirect
+ .redirect = iwch_ep_redirect,
+ .event_handler = iwch_event_handler
};
static LIST_HEAD(dev_list);
static DEFINE_MUTEX(dev_mutex);
+static int disable_qp_db(int id, void *p, void *data)
+{
+ struct iwch_qp *qhp = p;
+
+ cxio_disable_wq_db(&qhp->wq);
+ return 0;
+}
+
+static int enable_qp_db(int id, void *p, void *data)
+{
+ struct iwch_qp *qhp = p;
+
+ if (data)
+ ring_doorbell(qhp->rhp->rdev.ctrl_qp.doorbell, qhp->wq.qpid);
+ cxio_enable_wq_db(&qhp->wq);
+ return 0;
+}
+
+static void disable_dbs(struct iwch_dev *rnicp)
+{
+ spin_lock_irq(&rnicp->lock);
+ idr_for_each(&rnicp->qpidr, disable_qp_db, NULL);
+ spin_unlock_irq(&rnicp->lock);
+}
+
+static void enable_dbs(struct iwch_dev *rnicp, int ring_db)
+{
+ spin_lock_irq(&rnicp->lock);
+ idr_for_each(&rnicp->qpidr, enable_qp_db,
+ (void *)(unsigned long)ring_db);
+ spin_unlock_irq(&rnicp->lock);
+}
+
+static void iwch_db_drop_task(struct work_struct *work)
+{
+ struct iwch_dev *rnicp = container_of(work, struct iwch_dev,
+ db_drop_task.work);
+ enable_dbs(rnicp, 1);
+}
+
static void rnic_init(struct iwch_dev *rnicp)
{
PDBG("%s iwch_dev %p\n", __func__, rnicp);
@@ -70,6 +110,7 @@ static void rnic_init(struct iwch_dev *rnicp)
idr_init(&rnicp->qpidr);
idr_init(&rnicp->mmidr);
spin_lock_init(&rnicp->lock);
+ INIT_DELAYED_WORK(&rnicp->db_drop_task, iwch_db_drop_task);
rnicp->attr.max_qps = T3_MAX_NUM_QP - 32;
rnicp->attr.max_wrs = T3_MAX_QP_DEPTH;
@@ -103,11 +144,9 @@ static void rnic_init(struct iwch_dev *rnicp)
static void open_rnic_dev(struct t3cdev *tdev)
{
struct iwch_dev *rnicp;
- static int vers_printed;
PDBG("%s t3cdev %p\n", __func__, tdev);
- if (!vers_printed++)
- printk(KERN_INFO MOD "Chelsio T3 RDMA Driver - version %s\n",
+ printk_once(KERN_INFO MOD "Chelsio T3 RDMA Driver - version %s\n",
DRV_VERSION);
rnicp = (struct iwch_dev *)ib_alloc_device(sizeof(*rnicp));
if (!rnicp) {
@@ -147,6 +186,9 @@ static void close_rnic_dev(struct t3cdev *tdev)
mutex_lock(&dev_mutex);
list_for_each_entry_safe(dev, tmp, &dev_list, entry) {
if (dev->rdev.t3cdev_p == tdev) {
+ dev->rdev.flags = CXIO_ERROR_FATAL;
+ synchronize_net();
+ cancel_delayed_work_sync(&dev->db_drop_task);
list_del(&dev->entry);
iwch_unregister_device(dev);
cxio_rdev_close(&dev->rdev);
@@ -160,6 +202,69 @@ static void close_rnic_dev(struct t3cdev *tdev)
mutex_unlock(&dev_mutex);
}
+static void iwch_event_handler(struct t3cdev *tdev, u32 evt, u32 port_id)
+{
+ struct cxio_rdev *rdev = tdev->ulp;
+ struct iwch_dev *rnicp;
+ struct ib_event event;
+ u32 portnum = port_id + 1;
+ int dispatch = 0;
+
+ if (!rdev)
+ return;
+ rnicp = rdev_to_iwch_dev(rdev);
+ switch (evt) {
+ case OFFLOAD_STATUS_DOWN: {
+ rdev->flags = CXIO_ERROR_FATAL;
+ synchronize_net();
+ event.event = IB_EVENT_DEVICE_FATAL;
+ dispatch = 1;
+ break;
+ }
+ case OFFLOAD_PORT_DOWN: {
+ event.event = IB_EVENT_PORT_ERR;
+ dispatch = 1;
+ break;
+ }
+ case OFFLOAD_PORT_UP: {
+ event.event = IB_EVENT_PORT_ACTIVE;
+ dispatch = 1;
+ break;
+ }
+ case OFFLOAD_DB_FULL: {
+ disable_dbs(rnicp);
+ break;
+ }
+ case OFFLOAD_DB_EMPTY: {
+ enable_dbs(rnicp, 1);
+ break;
+ }
+ case OFFLOAD_DB_DROP: {
+ unsigned long delay = 1000;
+ unsigned short r;
+
+ disable_dbs(rnicp);
+ get_random_bytes(&r, 2);
+ delay += r & 1023;
+
+ /*
+ * delay is between 1000-2023 usecs.
+ */
+ schedule_delayed_work(&rnicp->db_drop_task,
+ usecs_to_jiffies(delay));
+ break;
+ }
+ }
+
+ if (dispatch) {
+ event.device = &rnicp->ibdev;
+ event.element.port_num = portnum;
+ ib_dispatch_event(&event);
+ }
+
+ return;
+}
+
static int __init iwch_init_module(void)
{
int err;
diff --git a/drivers/infiniband/hw/cxgb3/iwch.h b/drivers/infiniband/hw/cxgb3/iwch.h
index 3773453b2cf..837862287a2 100644
--- a/drivers/infiniband/hw/cxgb3/iwch.h
+++ b/drivers/infiniband/hw/cxgb3/iwch.h
@@ -36,6 +36,7 @@
#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/idr.h>
+#include <linux/workqueue.h>
#include <rdma/ib_verbs.h>
@@ -110,6 +111,7 @@ struct iwch_dev {
struct idr mmidr;
spinlock_t lock;
struct list_head entry;
+ struct delayed_work db_drop_task;
};
static inline struct iwch_dev *to_iwch_dev(struct ib_device *ibdev)
@@ -117,6 +119,11 @@ static inline struct iwch_dev *to_iwch_dev(struct ib_device *ibdev)
return container_of(ibdev, struct iwch_dev, ibdev);
}
+static inline struct iwch_dev *rdev_to_iwch_dev(struct cxio_rdev *rdev)
+{
+ return container_of(rdev, struct iwch_dev, rdev);
+}
+
static inline int t3b_device(const struct iwch_dev *rhp)
{
return rhp->rdev.t3cdev_p->type == T3B;
@@ -146,19 +153,17 @@ static inline int insert_handle(struct iwch_dev *rhp, struct idr *idr,
void *handle, u32 id)
{
int ret;
- int newid;
-
- do {
- if (!idr_pre_get(idr, GFP_KERNEL)) {
- return -ENOMEM;
- }
- spin_lock_irq(&rhp->lock);
- ret = idr_get_new_above(idr, handle, id, &newid);
- BUG_ON(newid != id);
- spin_unlock_irq(&rhp->lock);
- } while (ret == -EAGAIN);
-
- return ret;
+
+ idr_preload(GFP_KERNEL);
+ spin_lock_irq(&rhp->lock);
+
+ ret = idr_alloc(idr, handle, id, id + 1, GFP_NOWAIT);
+
+ spin_unlock_irq(&rhp->lock);
+ idr_preload_end();
+
+ BUG_ON(ret == -ENOSPC);
+ return ret < 0 ? ret : 0;
}
static inline void remove_handle(struct iwch_dev *rhp, struct idr *idr, u32 id)
diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.c b/drivers/infiniband/hw/cxgb3/iwch_cm.c
index c325c44807e..cb78b1e9bcd 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_cm.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_cm.c
@@ -31,6 +31,7 @@
*/
#include <linux/module.h>
#include <linux/list.h>
+#include <linux/slab.h>
#include <linux/workqueue.h>
#include <linux/skbuff.h>
#include <linux/timer.h>
@@ -101,12 +102,9 @@ static unsigned int cong_flavor = 1;
module_param(cong_flavor, uint, 0644);
MODULE_PARM_DESC(cong_flavor, "TCP Congestion control flavor (default=1)");
-static void process_work(struct work_struct *work);
static struct workqueue_struct *workq;
-static DECLARE_WORK(skb_work, process_work);
static struct sk_buff_head rxq;
-static cxgb3_cpl_handler_func work_handlers[NUM_CPL_CMDS];
static struct sk_buff *get_skb(struct sk_buff *skb, int len, gfp_t gfp);
static void ep_timeout(unsigned long arg);
@@ -130,15 +128,46 @@ static void stop_ep_timer(struct iwch_ep *ep)
{
PDBG("%s ep %p\n", __func__, ep);
if (!timer_pending(&ep->timer)) {
- printk(KERN_ERR "%s timer stopped when its not running! ep %p state %u\n",
+ WARN(1, "%s timer stopped when its not running! ep %p state %u\n",
__func__, ep, ep->com.state);
- WARN_ON(1);
return;
}
del_timer_sync(&ep->timer);
put_ep(&ep->com);
}
+static int iwch_l2t_send(struct t3cdev *tdev, struct sk_buff *skb, struct l2t_entry *l2e)
+{
+ int error = 0;
+ struct cxio_rdev *rdev;
+
+ rdev = (struct cxio_rdev *)tdev->ulp;
+ if (cxio_fatal_error(rdev)) {
+ kfree_skb(skb);
+ return -EIO;
+ }
+ error = l2t_send(tdev, skb, l2e);
+ if (error < 0)
+ kfree_skb(skb);
+ return error;
+}
+
+int iwch_cxgb3_ofld_send(struct t3cdev *tdev, struct sk_buff *skb)
+{
+ int error = 0;
+ struct cxio_rdev *rdev;
+
+ rdev = (struct cxio_rdev *)tdev->ulp;
+ if (cxio_fatal_error(rdev)) {
+ kfree_skb(skb);
+ return -EIO;
+ }
+ error = cxgb3_ofld_send(tdev, skb);
+ if (error < 0)
+ kfree_skb(skb);
+ return error;
+}
+
static void release_tid(struct t3cdev *tdev, u32 hwtid, struct sk_buff *skb)
{
struct cpl_tid_release *req;
@@ -150,7 +179,7 @@ static void release_tid(struct t3cdev *tdev, u32 hwtid, struct sk_buff *skb)
req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD));
OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_TID_RELEASE, hwtid));
skb->priority = CPL_PRIORITY_SETUP;
- cxgb3_ofld_send(tdev, skb);
+ iwch_cxgb3_ofld_send(tdev, skb);
return;
}
@@ -172,8 +201,7 @@ int iwch_quiesce_tid(struct iwch_ep *ep)
req->val = cpu_to_be64(1 << S_TCB_RX_QUIESCE);
skb->priority = CPL_PRIORITY_DATA;
- cxgb3_ofld_send(ep->com.tdev, skb);
- return 0;
+ return iwch_cxgb3_ofld_send(ep->com.tdev, skb);
}
int iwch_resume_tid(struct iwch_ep *ep)
@@ -194,8 +222,7 @@ int iwch_resume_tid(struct iwch_ep *ep)
req->val = 0;
skb->priority = CPL_PRIORITY_DATA;
- cxgb3_ofld_send(ep->com.tdev, skb);
- return 0;
+ return iwch_cxgb3_ofld_send(ep->com.tdev, skb);
}
static void set_emss(struct iwch_ep *ep, u16 opt)
@@ -252,42 +279,25 @@ static void *alloc_ep(int size, gfp_t gfp)
void __free_ep(struct kref *kref)
{
- struct iwch_ep_common *epc;
- epc = container_of(kref, struct iwch_ep_common, kref);
- PDBG("%s ep %p state %s\n", __func__, epc, states[state_read(epc)]);
- kfree(epc);
+ struct iwch_ep *ep;
+ ep = container_of(container_of(kref, struct iwch_ep_common, kref),
+ struct iwch_ep, com);
+ PDBG("%s ep %p state %s\n", __func__, ep, states[state_read(&ep->com)]);
+ if (test_bit(RELEASE_RESOURCES, &ep->com.flags)) {
+ cxgb3_remove_tid(ep->com.tdev, (void *)ep, ep->hwtid);
+ dst_release(ep->dst);
+ l2t_release(ep->com.tdev, ep->l2t);
+ }
+ kfree(ep);
}
static void release_ep_resources(struct iwch_ep *ep)
{
PDBG("%s ep %p tid %d\n", __func__, ep, ep->hwtid);
- cxgb3_remove_tid(ep->com.tdev, (void *)ep, ep->hwtid);
- dst_release(ep->dst);
- l2t_release(L2DATA(ep->com.tdev), ep->l2t);
+ set_bit(RELEASE_RESOURCES, &ep->com.flags);
put_ep(&ep->com);
}
-static void process_work(struct work_struct *work)
-{
- struct sk_buff *skb = NULL;
- void *ep;
- struct t3cdev *tdev;
- int ret;
-
- while ((skb = skb_dequeue(&rxq))) {
- ep = *((void **) (skb->cb));
- tdev = *((struct t3cdev **) (skb->cb + sizeof(void *)));
- ret = work_handlers[G_OPCODE(ntohl((__force __be32)skb->csum))](tdev, skb, ep);
- if (ret & CPL_RET_BUF_DONE)
- kfree_skb(skb);
-
- /*
- * ep was referenced in sched(), and is freed here.
- */
- put_ep((struct iwch_ep_common *)ep);
- }
-}
-
static int status2errno(int status)
{
switch (status) {
@@ -327,23 +337,12 @@ static struct rtable *find_route(struct t3cdev *dev, __be32 local_ip,
__be16 peer_port, u8 tos)
{
struct rtable *rt;
- struct flowi fl = {
- .oif = 0,
- .nl_u = {
- .ip4_u = {
- .daddr = peer_ip,
- .saddr = local_ip,
- .tos = tos}
- },
- .proto = IPPROTO_TCP,
- .uli_u = {
- .ports = {
- .sport = local_port,
- .dport = peer_port}
- }
- };
-
- if (ip_route_output_flow(&init_net, &rt, &fl, NULL, 0))
+ struct flowi4 fl4;
+
+ rt = ip_route_output_ports(&init_net, &fl4, NULL, peer_ip, local_ip,
+ peer_port, local_port, IPPROTO_TCP,
+ tos, 0);
+ if (IS_ERR(rt))
return NULL;
return rt;
}
@@ -382,7 +381,7 @@ static void abort_arp_failure(struct t3cdev *dev, struct sk_buff *skb)
PDBG("%s t3cdev %p\n", __func__, dev);
req->cmd = CPL_ABORT_NO_RST;
- cxgb3_ofld_send(dev, skb);
+ iwch_cxgb3_ofld_send(dev, skb);
}
static int send_halfclose(struct iwch_ep *ep, gfp_t gfp)
@@ -402,8 +401,7 @@ static int send_halfclose(struct iwch_ep *ep, gfp_t gfp)
req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_CLOSE_CON));
req->wr.wr_lo = htonl(V_WR_TID(ep->hwtid));
OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_CLOSE_CON_REQ, ep->hwtid));
- l2t_send(ep->com.tdev, skb, ep->l2t);
- return 0;
+ return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
}
static int send_abort(struct iwch_ep *ep, struct sk_buff *skb, gfp_t gfp)
@@ -420,12 +418,12 @@ static int send_abort(struct iwch_ep *ep, struct sk_buff *skb, gfp_t gfp)
skb->priority = CPL_PRIORITY_DATA;
set_arp_failure_handler(skb, abort_arp_failure);
req = (struct cpl_abort_req *) skb_put(skb, sizeof(*req));
+ memset(req, 0, sizeof(*req));
req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_HOST_ABORT_CON_REQ));
req->wr.wr_lo = htonl(V_WR_TID(ep->hwtid));
OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_ABORT_REQ, ep->hwtid));
req->cmd = CPL_ABORT_SEND_RST;
- l2t_send(ep->com.tdev, skb, ep->l2t);
- return 0;
+ return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
}
static int send_connect(struct iwch_ep *ep)
@@ -454,7 +452,8 @@ static int send_connect(struct iwch_ep *ep)
V_MSS_IDX(mtu_idx) |
V_L2T_IDX(ep->l2t->idx) | V_TX_CHANNEL(ep->l2t->smt_idx);
opt0l = V_TOS((ep->tos >> 2) & M_TOS) | V_RCV_BUFSIZ(rcv_win>>10);
- opt2 = V_FLAVORS_VALID(1) | V_CONG_CONTROL_FLAVOR(cong_flavor);
+ opt2 = F_RX_COALESCE_VALID | V_RX_COALESCE(0) | V_FLAVORS_VALID(1) |
+ V_CONG_CONTROL_FLAVOR(cong_flavor);
skb->priority = CPL_PRIORITY_SETUP;
set_arp_failure_handler(skb, act_open_req_arp_failure);
@@ -469,8 +468,7 @@ static int send_connect(struct iwch_ep *ep)
req->opt0l = htonl(opt0l);
req->params = 0;
req->opt2 = htonl(opt2);
- l2t_send(ep->com.tdev, skb, ep->l2t);
- return 0;
+ return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
}
static void send_mpa_req(struct iwch_ep *ep, struct sk_buff *skb)
@@ -527,7 +525,7 @@ static void send_mpa_req(struct iwch_ep *ep, struct sk_buff *skb)
req->sndseq = htonl(ep->snd_seq);
BUG_ON(ep->mpa_skb);
ep->mpa_skb = skb;
- l2t_send(ep->com.tdev, skb, ep->l2t);
+ iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
start_ep_timer(ep);
state_set(&ep->com, MPA_REQ_SENT);
return;
@@ -578,8 +576,7 @@ static int send_mpa_reject(struct iwch_ep *ep, const void *pdata, u8 plen)
req->sndseq = htonl(ep->snd_seq);
BUG_ON(ep->mpa_skb);
ep->mpa_skb = skb;
- l2t_send(ep->com.tdev, skb, ep->l2t);
- return 0;
+ return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
}
static int send_mpa_reply(struct iwch_ep *ep, const void *pdata, u8 plen)
@@ -630,8 +627,7 @@ static int send_mpa_reply(struct iwch_ep *ep, const void *pdata, u8 plen)
req->sndseq = htonl(ep->snd_seq);
ep->mpa_skb = skb;
state_set(&ep->com, MPA_REP_SENT);
- l2t_send(ep->com.tdev, skb, ep->l2t);
- return 0;
+ return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
}
static int act_establish(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
@@ -726,8 +722,10 @@ static void connect_reply_upcall(struct iwch_ep *ep, int status)
memset(&event, 0, sizeof(event));
event.event = IW_CM_EVENT_CONNECT_REPLY;
event.status = status;
- event.local_addr = ep->com.local_addr;
- event.remote_addr = ep->com.remote_addr;
+ memcpy(&event.local_addr, &ep->com.local_addr,
+ sizeof(ep->com.local_addr));
+ memcpy(&event.remote_addr, &ep->com.remote_addr,
+ sizeof(ep->com.remote_addr));
if ((status == 0) || (status == -ECONNREFUSED)) {
event.private_data_len = ep->plen;
@@ -752,15 +750,24 @@ static void connect_request_upcall(struct iwch_ep *ep)
PDBG("%s ep %p tid %d\n", __func__, ep, ep->hwtid);
memset(&event, 0, sizeof(event));
event.event = IW_CM_EVENT_CONNECT_REQUEST;
- event.local_addr = ep->com.local_addr;
- event.remote_addr = ep->com.remote_addr;
+ memcpy(&event.local_addr, &ep->com.local_addr,
+ sizeof(ep->com.local_addr));
+ memcpy(&event.remote_addr, &ep->com.remote_addr,
+ sizeof(ep->com.local_addr));
event.private_data_len = ep->plen;
event.private_data = ep->mpa_pkt + sizeof(struct mpa_message);
event.provider_data = ep;
- if (state_read(&ep->parent_ep->com) != DEAD)
+ /*
+ * Until ird/ord negotiation via MPAv2 support is added, send max
+ * supported values
+ */
+ event.ird = event.ord = 8;
+ if (state_read(&ep->parent_ep->com) != DEAD) {
+ get_ep(&ep->com);
ep->parent_ep->com.cm_id->event_handler(
ep->parent_ep->com.cm_id,
&event);
+ }
put_ep(&ep->parent_ep->com);
ep->parent_ep = NULL;
}
@@ -772,6 +779,11 @@ static void established_upcall(struct iwch_ep *ep)
PDBG("%s ep %p\n", __func__, ep);
memset(&event, 0, sizeof(event));
event.event = IW_CM_EVENT_ESTABLISHED;
+ /*
+ * Until ird/ord negotiation via MPAv2 support is added, send max
+ * supported values
+ */
+ event.ird = event.ord = 8;
if (ep->com.cm_id) {
PDBG("%s ep %p tid %d\n", __func__, ep, ep->hwtid);
ep->com.cm_id->event_handler(ep->com.cm_id, &event);
@@ -795,7 +807,7 @@ static int update_rx_credits(struct iwch_ep *ep, u32 credits)
OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_RX_DATA_ACK, ep->hwtid));
req->credit_dack = htonl(V_RX_CREDITS(credits) | V_RX_FORCE_ACK(1));
skb->priority = CPL_PRIORITY_ACK;
- cxgb3_ofld_send(ep->com.tdev, skb);
+ iwch_cxgb3_ofld_send(ep->com.tdev, skb);
return credits;
}
@@ -916,7 +928,7 @@ static void process_mpa_reply(struct iwch_ep *ep, struct sk_buff *skb)
goto err;
if (peer2peer && iwch_rqes_posted(ep->com.qp) == 0) {
- iwch_post_zb_read(ep->com.qp);
+ iwch_post_zb_read(ep);
}
goto out;
@@ -1080,37 +1092,45 @@ static int tx_ack(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
struct iwch_ep *ep = ctx;
struct cpl_wr_ack *hdr = cplhdr(skb);
unsigned int credits = ntohs(hdr->credits);
+ unsigned long flags;
+ int post_zb = 0;
PDBG("%s ep %p credits %u\n", __func__, ep, credits);
if (credits == 0) {
- PDBG(KERN_ERR "%s 0 credit ack ep %p state %u\n",
- __func__, ep, state_read(&ep->com));
+ PDBG("%s 0 credit ack ep %p state %u\n",
+ __func__, ep, state_read(&ep->com));
return CPL_RET_BUF_DONE;
}
+ spin_lock_irqsave(&ep->com.lock, flags);
BUG_ON(credits != 1);
dst_confirm(ep->dst);
if (!ep->mpa_skb) {
PDBG("%s rdma_init wr_ack ep %p state %u\n",
- __func__, ep, state_read(&ep->com));
+ __func__, ep, ep->com.state);
if (ep->mpa_attr.initiator) {
PDBG("%s initiator ep %p state %u\n",
- __func__, ep, state_read(&ep->com));
- if (peer2peer)
- iwch_post_zb_read(ep->com.qp);
+ __func__, ep, ep->com.state);
+ if (peer2peer && ep->com.state == FPDU_MODE)
+ post_zb = 1;
} else {
PDBG("%s responder ep %p state %u\n",
- __func__, ep, state_read(&ep->com));
- ep->com.rpl_done = 1;
- wake_up(&ep->com.waitq);
+ __func__, ep, ep->com.state);
+ if (ep->com.state == MPA_REQ_RCVD) {
+ ep->com.rpl_done = 1;
+ wake_up(&ep->com.waitq);
+ }
}
} else {
PDBG("%s lsm ack ep %p state %u freeing skb\n",
- __func__, ep, state_read(&ep->com));
+ __func__, ep, ep->com.state);
kfree_skb(ep->mpa_skb);
ep->mpa_skb = NULL;
}
+ spin_unlock_irqrestore(&ep->com.lock, flags);
+ if (post_zb)
+ iwch_post_zb_read(ep);
return CPL_RET_BUF_DONE;
}
@@ -1127,8 +1147,7 @@ static int abort_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
* We get 2 abort replies from the HW. The first one must
* be ignored except for scribbling that we need one more.
*/
- if (!(ep->flags & ABORT_REQ_IN_PROGRESS)) {
- ep->flags |= ABORT_REQ_IN_PROGRESS;
+ if (!test_and_set_bit(ABORT_REQ_IN_PROGRESS, &ep->com.flags)) {
return CPL_RET_BUF_DONE;
}
@@ -1173,7 +1192,7 @@ static int act_open_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
release_tid(ep->com.tdev, GET_TID(rpl), NULL);
cxgb3_free_atid(ep->com.tdev, ep->atid);
dst_release(ep->dst);
- l2t_release(L2DATA(ep->com.tdev), ep->l2t);
+ l2t_release(ep->com.tdev, ep->l2t);
put_ep(&ep->com);
return CPL_RET_BUF_DONE;
}
@@ -1203,8 +1222,7 @@ static int listen_start(struct iwch_listen_ep *ep)
req->opt1 = htonl(V_CONN_POLICY(CPL_CONN_POLICY_ASK));
skb->priority = 1;
- cxgb3_ofld_send(ep->com.tdev, skb);
- return 0;
+ return iwch_cxgb3_ofld_send(ep->com.tdev, skb);
}
static int pass_open_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
@@ -1237,8 +1255,7 @@ static int listen_stop(struct iwch_listen_ep *ep)
req->cpu_idx = 0;
OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_CLOSE_LISTSRV_REQ, ep->stid));
skb->priority = 1;
- cxgb3_ofld_send(ep->com.tdev, skb);
- return 0;
+ return iwch_cxgb3_ofld_send(ep->com.tdev, skb);
}
static int close_listsrv_rpl(struct t3cdev *tdev, struct sk_buff *skb,
@@ -1275,7 +1292,8 @@ static void accept_cr(struct iwch_ep *ep, __be32 peer_ip, struct sk_buff *skb)
V_MSS_IDX(mtu_idx) |
V_L2T_IDX(ep->l2t->idx) | V_TX_CHANNEL(ep->l2t->smt_idx);
opt0l = V_TOS((ep->tos >> 2) & M_TOS) | V_RCV_BUFSIZ(rcv_win>>10);
- opt2 = V_FLAVORS_VALID(1) | V_CONG_CONTROL_FLAVOR(cong_flavor);
+ opt2 = F_RX_COALESCE_VALID | V_RX_COALESCE(0) | V_FLAVORS_VALID(1) |
+ V_CONG_CONTROL_FLAVOR(cong_flavor);
rpl = cplhdr(skb);
rpl->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD));
@@ -1286,7 +1304,7 @@ static void accept_cr(struct iwch_ep *ep, __be32 peer_ip, struct sk_buff *skb)
rpl->opt2 = htonl(opt2);
rpl->rsvd = rpl->opt2; /* workaround for HW bug */
skb->priority = CPL_PRIORITY_SETUP;
- l2t_send(ep->com.tdev, skb, ep->l2t);
+ iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
return;
}
@@ -1315,7 +1333,7 @@ static void reject_cr(struct t3cdev *tdev, u32 hwtid, __be32 peer_ip,
rpl->opt0l_status = htonl(CPL_PASS_OPEN_REJECT);
rpl->opt2 = 0;
rpl->rsvd = rpl->opt2;
- cxgb3_ofld_send(tdev, skb);
+ iwch_cxgb3_ofld_send(tdev, skb);
}
}
@@ -1343,15 +1361,8 @@ static int pass_accept_req(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
tim.mac_addr = req->dst_mac;
tim.vlan_tag = ntohs(req->vlan_tag);
if (tdev->ctl(tdev, GET_IFF_FROM_MAC, &tim) < 0 || !tim.dev) {
- printk(KERN_ERR
- "%s bad dst mac %02x %02x %02x %02x %02x %02x\n",
- __func__,
- req->dst_mac[0],
- req->dst_mac[1],
- req->dst_mac[2],
- req->dst_mac[3],
- req->dst_mac[4],
- req->dst_mac[5]);
+ printk(KERN_ERR "%s bad dst mac %pM\n",
+ __func__, req->dst_mac);
goto reject;
}
@@ -1366,8 +1377,8 @@ static int pass_accept_req(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
__func__);
goto reject;
}
- dst = &rt->u.dst;
- l2t = t3_l2t_get(tdev, dst->neighbour, dst->neighbour->dev);
+ dst = &rt->dst;
+ l2t = t3_l2t_get(tdev, dst, NULL, &req->peer_ip);
if (!l2t) {
printk(KERN_ERR MOD "%s - failed to allocate l2t entry!\n",
__func__);
@@ -1378,7 +1389,7 @@ static int pass_accept_req(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
if (!child_ep) {
printk(KERN_ERR MOD "%s - failed to allocate ep entry!\n",
__func__);
- l2t_release(L2DATA(tdev), l2t);
+ l2t_release(tdev, l2t);
dst_release(dst);
goto reject;
}
@@ -1450,10 +1461,14 @@ static int peer_close(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
/*
* We're gonna mark this puppy DEAD, but keep
* the reference on it until the ULP accepts or
- * rejects the CR.
+ * rejects the CR. Also wake up anyone waiting
+ * in rdma connection migration (see iwch_accept_cr()).
*/
__state_set(&ep->com, CLOSING);
- get_ep(&ep->com);
+ ep->com.rpl_done = 1;
+ ep->com.rpl_err = -ECONNRESET;
+ PDBG("waking up ep %p\n", ep);
+ wake_up(&ep->com.waitq);
break;
case MPA_REP_SENT:
__state_set(&ep->com, CLOSING);
@@ -1534,8 +1549,7 @@ static int peer_abort(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
* We get 2 peer aborts from the HW. The first one must
* be ignored except for scribbling that we need one more.
*/
- if (!(ep->flags & PEER_ABORT_IN_PROGRESS)) {
- ep->flags |= PEER_ABORT_IN_PROGRESS;
+ if (!test_and_set_bit(PEER_ABORT_IN_PROGRESS, &ep->com.flags)) {
return CPL_RET_BUF_DONE;
}
@@ -1562,9 +1576,13 @@ static int peer_abort(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
/*
* We're gonna mark this puppy DEAD, but keep
* the reference on it until the ULP accepts or
- * rejects the CR.
+ * rejects the CR. Also wake up anyone waiting
+ * in rdma connection migration (see iwch_accept_cr()).
*/
- get_ep(&ep->com);
+ ep->com.rpl_done = 1;
+ ep->com.rpl_err = -ECONNRESET;
+ PDBG("waking up ep %p\n", ep);
+ wake_up(&ep->com.waitq);
break;
case MORIBUND:
case CLOSING:
@@ -1613,7 +1631,7 @@ static int peer_abort(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
rpl->wr.wr_lo = htonl(V_WR_TID(ep->hwtid));
OPCODE_TID(rpl) = htonl(MK_OPCODE_TID(CPL_ABORT_RPL, ep->hwtid));
rpl->cmd = CPL_ABORT_NO_RST;
- cxgb3_ofld_send(ep->com.tdev, rpl_skb);
+ iwch_cxgb3_ofld_send(ep->com.tdev, rpl_skb);
out:
if (release)
release_ep_resources(ep);
@@ -1666,7 +1684,7 @@ static int close_con_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
* T3A does 3 things when a TERM is received:
* 1) send up a CPL_RDMA_TERMINATE message with the TERM packet
* 2) generate an async event on the QP with the TERMINATE opcode
- * 3) post a TERMINATE opcde cqe into the associated CQ.
+ * 3) post a TERMINATE opcode cqe into the associated CQ.
*
* For (1), we save the message in the qp for later consumer consumption.
* For (2), we move the QP into TERMINATE, post a QP event and disconnect.
@@ -1678,6 +1696,9 @@ static int terminate(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
{
struct iwch_ep *ep = ctx;
+ if (state_read(&ep->com) != FPDU_MODE)
+ return CPL_RET_BUF_DONE;
+
PDBG("%s ep %p\n", __func__, ep);
skb_pull(skb, sizeof(struct cpl_rdma_terminate));
PDBG("%s saving %d bytes of term msg\n", __func__, skb->len);
@@ -1739,9 +1760,8 @@ static void ep_timeout(unsigned long arg)
__state_set(&ep->com, ABORTING);
break;
default:
- printk(KERN_ERR "%s unexpected state ep %p state %u\n",
+ WARN(1, "%s unexpected state ep %p state %u\n",
__func__, ep, ep->com.state);
- WARN_ON(1);
abort = 0;
}
spin_unlock_irqrestore(&ep->com.lock, flags);
@@ -1767,6 +1787,7 @@ int iwch_reject_cr(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len)
err = send_mpa_reject(ep, pdata, pdata_len);
err = iwch_ep_disconnect(ep, 0, GFP_KERNEL);
}
+ put_ep(&ep->com);
return 0;
}
@@ -1780,8 +1801,10 @@ int iwch_accept_cr(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
struct iwch_qp *qp = get_qhp(h, conn_param->qpn);
PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
- if (state_read(&ep->com) == DEAD)
- return -ECONNRESET;
+ if (state_read(&ep->com) == DEAD) {
+ err = -ECONNRESET;
+ goto err;
+ }
BUG_ON(state_read(&ep->com) != MPA_REQ_RCVD);
BUG_ON(!qp);
@@ -1789,20 +1812,21 @@ int iwch_accept_cr(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
if ((conn_param->ord > qp->rhp->attr.max_rdma_read_qp_depth) ||
(conn_param->ird > qp->rhp->attr.max_rdma_reads_per_qp)) {
abort_connection(ep, NULL, GFP_KERNEL);
- return -EINVAL;
+ err = -EINVAL;
+ goto err;
}
cm_id->add_ref(cm_id);
ep->com.cm_id = cm_id;
ep->com.qp = qp;
- ep->com.rpl_done = 0;
- ep->com.rpl_err = 0;
ep->ird = conn_param->ird;
ep->ord = conn_param->ord;
- PDBG("%s %d ird %d ord %d\n", __func__, __LINE__, ep->ird, ep->ord);
- get_ep(&ep->com);
+ if (peer2peer && ep->ird == 0)
+ ep->ird = 1;
+
+ PDBG("%s %d ird %d ord %d\n", __func__, __LINE__, ep->ird, ep->ord);
/* bind QP to EP and move to RTS */
attrs.mpa_attr = ep->mpa_attr;
@@ -1821,30 +1845,31 @@ int iwch_accept_cr(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
err = iwch_modify_qp(ep->com.qp->rhp,
ep->com.qp, mask, &attrs, 1);
if (err)
- goto err;
+ goto err1;
/* if needed, wait for wr_ack */
if (iwch_rqes_posted(qp)) {
wait_event(ep->com.waitq, ep->com.rpl_done);
err = ep->com.rpl_err;
if (err)
- goto err;
+ goto err1;
}
err = send_mpa_reply(ep, conn_param->private_data,
conn_param->private_data_len);
if (err)
- goto err;
+ goto err1;
state_set(&ep->com, FPDU_MODE);
established_upcall(ep);
put_ep(&ep->com);
return 0;
-err:
+err1:
ep->com.cm_id = NULL;
ep->com.qp = NULL;
cm_id->rem_ref(cm_id);
+err:
put_ep(&ep->com);
return err;
}
@@ -1852,8 +1877,9 @@ err:
static int is_loopback_dst(struct iw_cm_id *cm_id)
{
struct net_device *dev;
+ struct sockaddr_in *raddr = (struct sockaddr_in *)&cm_id->remote_addr;
- dev = ip_dev_find(&init_net, cm_id->remote_addr.sin_addr.s_addr);
+ dev = ip_dev_find(&init_net, raddr->sin_addr.s_addr);
if (!dev)
return 0;
dev_put(dev);
@@ -1862,10 +1888,17 @@ static int is_loopback_dst(struct iw_cm_id *cm_id)
int iwch_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
{
- int err = 0;
struct iwch_dev *h = to_iwch_dev(cm_id->device);
struct iwch_ep *ep;
struct rtable *rt;
+ int err = 0;
+ struct sockaddr_in *laddr = (struct sockaddr_in *)&cm_id->local_addr;
+ struct sockaddr_in *raddr = (struct sockaddr_in *)&cm_id->remote_addr;
+
+ if (cm_id->remote_addr.ss_family != PF_INET) {
+ err = -ENOSYS;
+ goto out;
+ }
if (is_loopback_dst(cm_id)) {
err = -ENOSYS;
@@ -1885,6 +1918,10 @@ int iwch_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
conn_param->private_data, ep->plen);
ep->ird = conn_param->ird;
ep->ord = conn_param->ord;
+
+ if (peer2peer && ep->ord == 0)
+ ep->ord = 1;
+
ep->com.tdev = h->rdev.t3cdev_p;
cm_id->add_ref(cm_id);
@@ -1905,21 +1942,17 @@ int iwch_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
}
/* find a route */
- rt = find_route(h->rdev.t3cdev_p,
- cm_id->local_addr.sin_addr.s_addr,
- cm_id->remote_addr.sin_addr.s_addr,
- cm_id->local_addr.sin_port,
- cm_id->remote_addr.sin_port, IPTOS_LOWDELAY);
+ rt = find_route(h->rdev.t3cdev_p, laddr->sin_addr.s_addr,
+ raddr->sin_addr.s_addr, laddr->sin_port,
+ raddr->sin_port, IPTOS_LOWDELAY);
if (!rt) {
printk(KERN_ERR MOD "%s - cannot find route.\n", __func__);
err = -EHOSTUNREACH;
goto fail3;
}
- ep->dst = &rt->u.dst;
-
- /* get a l2t entry */
- ep->l2t = t3_l2t_get(ep->com.tdev, ep->dst->neighbour,
- ep->dst->neighbour->dev);
+ ep->dst = &rt->dst;
+ ep->l2t = t3_l2t_get(ep->com.tdev, ep->dst, NULL,
+ &raddr->sin_addr.s_addr);
if (!ep->l2t) {
printk(KERN_ERR MOD "%s - cannot alloc l2e.\n", __func__);
err = -ENOMEM;
@@ -1928,20 +1961,23 @@ int iwch_connect(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param)
state_set(&ep->com, CONNECTING);
ep->tos = IPTOS_LOWDELAY;
- ep->com.local_addr = cm_id->local_addr;
- ep->com.remote_addr = cm_id->remote_addr;
+ memcpy(&ep->com.local_addr, &cm_id->local_addr,
+ sizeof(ep->com.local_addr));
+ memcpy(&ep->com.remote_addr, &cm_id->remote_addr,
+ sizeof(ep->com.remote_addr));
/* send connect request to rnic */
err = send_connect(ep);
if (!err)
goto out;
- l2t_release(L2DATA(h->rdev.t3cdev_p), ep->l2t);
+ l2t_release(h->rdev.t3cdev_p, ep->l2t);
fail4:
dst_release(ep->dst);
fail3:
cxgb3_free_atid(ep->com.tdev, ep->atid);
fail2:
+ cm_id->rem_ref(cm_id);
put_ep(&ep->com);
out:
return err;
@@ -1956,6 +1992,11 @@ int iwch_create_listen(struct iw_cm_id *cm_id, int backlog)
might_sleep();
+ if (cm_id->local_addr.ss_family != PF_INET) {
+ err = -ENOSYS;
+ goto fail1;
+ }
+
ep = alloc_ep(sizeof(*ep), GFP_KERNEL);
if (!ep) {
printk(KERN_ERR MOD "%s - cannot alloc ep.\n", __func__);
@@ -1967,7 +2008,8 @@ int iwch_create_listen(struct iw_cm_id *cm_id, int backlog)
cm_id->add_ref(cm_id);
ep->com.cm_id = cm_id;
ep->backlog = backlog;
- ep->com.local_addr = cm_id->local_addr;
+ memcpy(&ep->com.local_addr, &cm_id->local_addr,
+ sizeof(ep->com.local_addr));
/*
* Allocate a server TID.
@@ -2013,8 +2055,11 @@ int iwch_destroy_listen(struct iw_cm_id *cm_id)
ep->com.rpl_done = 0;
ep->com.rpl_err = 0;
err = listen_stop(ep);
+ if (err)
+ goto done;
wait_event(ep->com.waitq, ep->com.rpl_done);
cxgb3_free_stid(ep->com.tdev, ep->stid);
+done:
err = ep->com.rpl_err;
cm_id->rem_ref(cm_id);
put_ep(&ep->com);
@@ -2026,12 +2071,22 @@ int iwch_ep_disconnect(struct iwch_ep *ep, int abrupt, gfp_t gfp)
int ret=0;
unsigned long flags;
int close = 0;
+ int fatal = 0;
+ struct t3cdev *tdev;
+ struct cxio_rdev *rdev;
spin_lock_irqsave(&ep->com.lock, flags);
PDBG("%s ep %p state %s, abrupt %d\n", __func__, ep,
states[ep->com.state], abrupt);
+ tdev = (struct t3cdev *)ep->com.tdev;
+ rdev = (struct cxio_rdev *)tdev->ulp;
+ if (cxio_fatal_error(rdev)) {
+ fatal = 1;
+ close_complete_upcall(ep);
+ ep->com.state = DEAD;
+ }
switch (ep->com.state) {
case MPA_REQ_WAIT:
case MPA_REQ_SENT:
@@ -2045,14 +2100,17 @@ int iwch_ep_disconnect(struct iwch_ep *ep, int abrupt, gfp_t gfp)
ep->com.state = CLOSING;
start_ep_timer(ep);
}
+ set_bit(CLOSE_SENT, &ep->com.flags);
break;
case CLOSING:
- close = 1;
- if (abrupt) {
- stop_ep_timer(ep);
- ep->com.state = ABORTING;
- } else
- ep->com.state = MORIBUND;
+ if (!test_and_set_bit(CLOSE_SENT, &ep->com.flags)) {
+ close = 1;
+ if (abrupt) {
+ stop_ep_timer(ep);
+ ep->com.state = ABORTING;
+ } else
+ ep->com.state = MORIBUND;
+ }
break;
case MORIBUND:
case ABORTING:
@@ -2071,7 +2129,11 @@ int iwch_ep_disconnect(struct iwch_ep *ep, int abrupt, gfp_t gfp)
ret = send_abort(ep, NULL, gfp);
else
ret = send_halfclose(ep, gfp);
+ if (ret)
+ fatal = 1;
}
+ if (fatal)
+ release_ep_resources(ep);
return ret;
}
@@ -2086,7 +2148,7 @@ int iwch_ep_redirect(void *ctx, struct dst_entry *old, struct dst_entry *new,
PDBG("%s ep %p redirect to dst %p l2t %p\n", __func__, ep, new,
l2t);
dst_hold(new);
- l2t_release(L2DATA(ep->com.tdev), ep->l2t);
+ l2t_release(ep->com.tdev, ep->l2t);
ep->l2t = l2t;
dst_release(old);
ep->dst = new;
@@ -2095,7 +2157,49 @@ int iwch_ep_redirect(void *ctx, struct dst_entry *old, struct dst_entry *new,
/*
* All the CM events are handled on a work queue to have a safe context.
+ * These are the real handlers that are called from the work queue.
*/
+static const cxgb3_cpl_handler_func work_handlers[NUM_CPL_CMDS] = {
+ [CPL_ACT_ESTABLISH] = act_establish,
+ [CPL_ACT_OPEN_RPL] = act_open_rpl,
+ [CPL_RX_DATA] = rx_data,
+ [CPL_TX_DMA_ACK] = tx_ack,
+ [CPL_ABORT_RPL_RSS] = abort_rpl,
+ [CPL_ABORT_RPL] = abort_rpl,
+ [CPL_PASS_OPEN_RPL] = pass_open_rpl,
+ [CPL_CLOSE_LISTSRV_RPL] = close_listsrv_rpl,
+ [CPL_PASS_ACCEPT_REQ] = pass_accept_req,
+ [CPL_PASS_ESTABLISH] = pass_establish,
+ [CPL_PEER_CLOSE] = peer_close,
+ [CPL_ABORT_REQ_RSS] = peer_abort,
+ [CPL_CLOSE_CON_RPL] = close_con_rpl,
+ [CPL_RDMA_TERMINATE] = terminate,
+ [CPL_RDMA_EC_STATUS] = ec_status,
+};
+
+static void process_work(struct work_struct *work)
+{
+ struct sk_buff *skb = NULL;
+ void *ep;
+ struct t3cdev *tdev;
+ int ret;
+
+ while ((skb = skb_dequeue(&rxq))) {
+ ep = *((void **) (skb->cb));
+ tdev = *((struct t3cdev **) (skb->cb + sizeof(void *)));
+ ret = work_handlers[G_OPCODE(ntohl((__force __be32)skb->csum))](tdev, skb, ep);
+ if (ret & CPL_RET_BUF_DONE)
+ kfree_skb(skb);
+
+ /*
+ * ep was referenced in sched(), and is freed here.
+ */
+ put_ep((struct iwch_ep_common *)ep);
+ }
+}
+
+static DECLARE_WORK(skb_work, process_work);
+
static int sched(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
{
struct iwch_ep_common *epc = ctx;
@@ -2127,6 +2231,29 @@ static int set_tcb_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
return CPL_RET_BUF_DONE;
}
+/*
+ * All upcalls from the T3 Core go to sched() to schedule the
+ * processing on a work queue.
+ */
+cxgb3_cpl_handler_func t3c_handlers[NUM_CPL_CMDS] = {
+ [CPL_ACT_ESTABLISH] = sched,
+ [CPL_ACT_OPEN_RPL] = sched,
+ [CPL_RX_DATA] = sched,
+ [CPL_TX_DMA_ACK] = sched,
+ [CPL_ABORT_RPL_RSS] = sched,
+ [CPL_ABORT_RPL] = sched,
+ [CPL_PASS_OPEN_RPL] = sched,
+ [CPL_CLOSE_LISTSRV_RPL] = sched,
+ [CPL_PASS_ACCEPT_REQ] = sched,
+ [CPL_PASS_ESTABLISH] = sched,
+ [CPL_PEER_CLOSE] = sched,
+ [CPL_CLOSE_CON_RPL] = sched,
+ [CPL_ABORT_REQ_RSS] = sched,
+ [CPL_RDMA_TERMINATE] = sched,
+ [CPL_RDMA_EC_STATUS] = sched,
+ [CPL_SET_TCB_RPL] = set_tcb_rpl,
+};
+
int __init iwch_cm_init(void)
{
skb_queue_head_init(&rxq);
@@ -2135,46 +2262,6 @@ int __init iwch_cm_init(void)
if (!workq)
return -ENOMEM;
- /*
- * All upcalls from the T3 Core go to sched() to
- * schedule the processing on a work queue.
- */
- t3c_handlers[CPL_ACT_ESTABLISH] = sched;
- t3c_handlers[CPL_ACT_OPEN_RPL] = sched;
- t3c_handlers[CPL_RX_DATA] = sched;
- t3c_handlers[CPL_TX_DMA_ACK] = sched;
- t3c_handlers[CPL_ABORT_RPL_RSS] = sched;
- t3c_handlers[CPL_ABORT_RPL] = sched;
- t3c_handlers[CPL_PASS_OPEN_RPL] = sched;
- t3c_handlers[CPL_CLOSE_LISTSRV_RPL] = sched;
- t3c_handlers[CPL_PASS_ACCEPT_REQ] = sched;
- t3c_handlers[CPL_PASS_ESTABLISH] = sched;
- t3c_handlers[CPL_PEER_CLOSE] = sched;
- t3c_handlers[CPL_CLOSE_CON_RPL] = sched;
- t3c_handlers[CPL_ABORT_REQ_RSS] = sched;
- t3c_handlers[CPL_RDMA_TERMINATE] = sched;
- t3c_handlers[CPL_RDMA_EC_STATUS] = sched;
- t3c_handlers[CPL_SET_TCB_RPL] = set_tcb_rpl;
-
- /*
- * These are the real handlers that are called from a
- * work queue.
- */
- work_handlers[CPL_ACT_ESTABLISH] = act_establish;
- work_handlers[CPL_ACT_OPEN_RPL] = act_open_rpl;
- work_handlers[CPL_RX_DATA] = rx_data;
- work_handlers[CPL_TX_DMA_ACK] = tx_ack;
- work_handlers[CPL_ABORT_RPL_RSS] = abort_rpl;
- work_handlers[CPL_ABORT_RPL] = abort_rpl;
- work_handlers[CPL_PASS_OPEN_RPL] = pass_open_rpl;
- work_handlers[CPL_CLOSE_LISTSRV_RPL] = close_listsrv_rpl;
- work_handlers[CPL_PASS_ACCEPT_REQ] = pass_accept_req;
- work_handlers[CPL_PASS_ESTABLISH] = pass_establish;
- work_handlers[CPL_PEER_CLOSE] = peer_close;
- work_handlers[CPL_ABORT_REQ_RSS] = peer_abort;
- work_handlers[CPL_CLOSE_CON_RPL] = close_con_rpl;
- work_handlers[CPL_RDMA_TERMINATE] = terminate;
- work_handlers[CPL_RDMA_EC_STATUS] = ec_status;
return 0;
}
diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.h b/drivers/infiniband/hw/cxgb3/iwch_cm.h
index d7c7e09f099..b9efadfffb4 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_cm.h
+++ b/drivers/infiniband/hw/cxgb3/iwch_cm.h
@@ -145,8 +145,10 @@ enum iwch_ep_state {
};
enum iwch_ep_flags {
- PEER_ABORT_IN_PROGRESS = (1 << 0),
- ABORT_REQ_IN_PROGRESS = (1 << 1),
+ PEER_ABORT_IN_PROGRESS = 0,
+ ABORT_REQ_IN_PROGRESS = 1,
+ RELEASE_RESOURCES = 2,
+ CLOSE_SENT = 3,
};
struct iwch_ep_common {
@@ -161,6 +163,7 @@ struct iwch_ep_common {
wait_queue_head_t waitq;
int rpl_done;
int rpl_err;
+ unsigned long flags;
};
struct iwch_listen_ep {
@@ -188,7 +191,6 @@ struct iwch_ep {
u16 plen;
u32 ird;
u32 ord;
- u32 flags;
};
static inline struct iwch_ep *to_ep(struct iw_cm_id *cm_id)
diff --git a/drivers/infiniband/hw/cxgb3/iwch_ev.c b/drivers/infiniband/hw/cxgb3/iwch_ev.c
index 7b67a677172..abcc9e76962 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_ev.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_ev.c
@@ -29,7 +29,7 @@
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
-#include <linux/slab.h>
+#include <linux/gfp.h>
#include <linux/mman.h>
#include <net/sock.h>
#include "iwch_provider.h"
@@ -46,6 +46,7 @@ static void post_qp_event(struct iwch_dev *rnicp, struct iwch_cq *chp,
struct ib_event event;
struct iwch_qp_attributes attrs;
struct iwch_qp *qhp;
+ unsigned long flag;
spin_lock(&rnicp->lock);
qhp = get_qhp(rnicp, CQE_QPID(rsp_msg->cqe));
@@ -76,6 +77,14 @@ static void post_qp_event(struct iwch_dev *rnicp, struct iwch_cq *chp,
atomic_inc(&qhp->refcnt);
spin_unlock(&rnicp->lock);
+ if (qhp->attr.state == IWCH_QP_STATE_RTS) {
+ attrs.next_state = IWCH_QP_STATE_TERMINATE;
+ iwch_modify_qp(qhp->rhp, qhp, IWCH_QP_ATTR_NEXT_STATE,
+ &attrs, 1);
+ if (send_term)
+ iwch_post_terminate(qhp, rsp_msg);
+ }
+
event.event = ib_event;
event.device = chp->ibcq.device;
if (ib_event == IB_EVENT_CQ_ERR)
@@ -86,13 +95,9 @@ static void post_qp_event(struct iwch_dev *rnicp, struct iwch_cq *chp,
if (qhp->ibqp.event_handler)
(*qhp->ibqp.event_handler)(&event, qhp->ibqp.qp_context);
- if (qhp->attr.state == IWCH_QP_STATE_RTS) {
- attrs.next_state = IWCH_QP_STATE_TERMINATE;
- iwch_modify_qp(qhp->rhp, qhp, IWCH_QP_ATTR_NEXT_STATE,
- &attrs, 1);
- if (send_term)
- iwch_post_terminate(qhp, rsp_msg);
- }
+ spin_lock_irqsave(&chp->comp_handler_lock, flag);
+ (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context);
+ spin_unlock_irqrestore(&chp->comp_handler_lock, flag);
if (atomic_dec_and_test(&qhp->refcnt))
wake_up(&qhp->wait);
@@ -105,6 +110,7 @@ void iwch_ev_dispatch(struct cxio_rdev *rdev_p, struct sk_buff *skb)
struct iwch_cq *chp;
struct iwch_qp *qhp;
u32 cqid = RSPQ_CQID(rsp_msg);
+ unsigned long flag;
rnicp = (struct iwch_dev *) rdev_p->ulp;
spin_lock(&rnicp->lock);
@@ -168,7 +174,9 @@ void iwch_ev_dispatch(struct cxio_rdev *rdev_p, struct sk_buff *skb)
*/
if (qhp->ep && SQ_TYPE(rsp_msg->cqe))
dst_confirm(qhp->ep->dst);
+ spin_lock_irqsave(&chp->comp_handler_lock, flag);
(*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context);
+ spin_unlock_irqrestore(&chp->comp_handler_lock, flag);
break;
case TPT_ERR_STAG:
@@ -179,12 +187,6 @@ void iwch_ev_dispatch(struct cxio_rdev *rdev_p, struct sk_buff *skb)
case TPT_ERR_BOUND:
case TPT_ERR_INVALIDATE_SHARED_MR:
case TPT_ERR_INVALIDATE_MR_WITH_MW_BOUND:
- printk(KERN_ERR "%s - CQE Err qpid 0x%x opcode %d status 0x%x "
- "type %d wrid.hi 0x%x wrid.lo 0x%x \n", __func__,
- CQE_QPID(rsp_msg->cqe), CQE_OPCODE(rsp_msg->cqe),
- CQE_STATUS(rsp_msg->cqe), CQE_TYPE(rsp_msg->cqe),
- CQE_WRID_HI(rsp_msg->cqe), CQE_WRID_LOW(rsp_msg->cqe));
- (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context);
post_qp_event(rnicp, chp, rsp_msg, IB_EVENT_QP_ACCESS_ERR, 1);
break;
diff --git a/drivers/infiniband/hw/cxgb3/iwch_mem.c b/drivers/infiniband/hw/cxgb3/iwch_mem.c
index ec49a5cbdeb..5c36ee2809a 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_mem.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_mem.c
@@ -29,6 +29,7 @@
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
+#include <linux/slab.h>
#include <asm/byteorder.h>
#include <rdma/iw_cm.h>
@@ -39,7 +40,7 @@
#include "iwch.h"
#include "iwch_provider.h"
-static void iwch_finish_mem_reg(struct iwch_mr *mhp, u32 stag)
+static int iwch_finish_mem_reg(struct iwch_mr *mhp, u32 stag)
{
u32 mmid;
@@ -47,14 +48,15 @@ static void iwch_finish_mem_reg(struct iwch_mr *mhp, u32 stag)
mhp->attr.stag = stag;
mmid = stag >> 8;
mhp->ibmr.rkey = mhp->ibmr.lkey = stag;
- insert_handle(mhp->rhp, &mhp->rhp->mmidr, mhp, mmid);
PDBG("%s mmid 0x%x mhp %p\n", __func__, mmid, mhp);
+ return insert_handle(mhp->rhp, &mhp->rhp->mmidr, mhp, mmid);
}
int iwch_register_mem(struct iwch_dev *rhp, struct iwch_pd *php,
struct iwch_mr *mhp, int shift)
{
u32 stag;
+ int ret;
if (cxio_register_phys_mem(&rhp->rdev,
&stag, mhp->attr.pdid,
@@ -66,9 +68,11 @@ int iwch_register_mem(struct iwch_dev *rhp, struct iwch_pd *php,
mhp->attr.pbl_size, mhp->attr.pbl_addr))
return -ENOMEM;
- iwch_finish_mem_reg(mhp, stag);
-
- return 0;
+ ret = iwch_finish_mem_reg(mhp, stag);
+ if (ret)
+ cxio_dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size,
+ mhp->attr.pbl_addr);
+ return ret;
}
int iwch_reregister_mem(struct iwch_dev *rhp, struct iwch_pd *php,
@@ -77,6 +81,7 @@ int iwch_reregister_mem(struct iwch_dev *rhp, struct iwch_pd *php,
int npages)
{
u32 stag;
+ int ret;
/* We could support this... */
if (npages > mhp->attr.pbl_size)
@@ -93,9 +98,12 @@ int iwch_reregister_mem(struct iwch_dev *rhp, struct iwch_pd *php,
mhp->attr.pbl_size, mhp->attr.pbl_addr))
return -ENOMEM;
- iwch_finish_mem_reg(mhp, stag);
+ ret = iwch_finish_mem_reg(mhp, stag);
+ if (ret)
+ cxio_dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size,
+ mhp->attr.pbl_addr);
- return 0;
+ return ret;
}
int iwch_alloc_pbl(struct iwch_mr *mhp, int npages)
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index b89640aa6e1..811b24a539c 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -37,9 +37,12 @@
#include <linux/delay.h>
#include <linux/errno.h>
#include <linux/list.h>
+#include <linux/sched.h>
#include <linux/spinlock.h>
#include <linux/ethtool.h>
#include <linux/rtnetlink.h>
+#include <linux/inetdevice.h>
+#include <linux/slab.h>
#include <asm/io.h>
#include <asm/irq.h>
@@ -58,13 +61,6 @@
#include "iwch_user.h"
#include "common.h"
-static int iwch_modify_port(struct ib_device *ibdev,
- u8 port, int port_modify_mask,
- struct ib_port_modify *props)
-{
- return -ENOSYS;
-}
-
static struct ib_ah *iwch_ah_create(struct ib_pd *pd,
struct ib_ah_attr *ah_attr)
{
@@ -151,6 +147,8 @@ static struct ib_cq *iwch_create_cq(struct ib_device *ibdev, int entries, int ve
struct iwch_create_cq_resp uresp;
struct iwch_create_cq_req ureq;
struct iwch_ucontext *ucontext = NULL;
+ static int warned;
+ size_t resplen;
PDBG("%s ib_dev %p entries %d\n", __func__, ibdev, entries);
rhp = to_iwch_dev(ibdev);
@@ -185,16 +183,21 @@ static struct ib_cq *iwch_create_cq(struct ib_device *ibdev, int entries, int ve
entries = roundup_pow_of_two(entries);
chp->cq.size_log2 = ilog2(entries);
- if (cxio_create_cq(&rhp->rdev, &chp->cq)) {
+ if (cxio_create_cq(&rhp->rdev, &chp->cq, !ucontext)) {
kfree(chp);
return ERR_PTR(-ENOMEM);
}
chp->rhp = rhp;
chp->ibcq.cqe = 1 << chp->cq.size_log2;
spin_lock_init(&chp->lock);
+ spin_lock_init(&chp->comp_handler_lock);
atomic_set(&chp->refcnt, 1);
init_waitqueue_head(&chp->wait);
- insert_handle(rhp, &rhp->cqidr, chp, chp->cq.cqid);
+ if (insert_handle(rhp, &rhp->cqidr, chp, chp->cq.cqid)) {
+ cxio_destroy_cq(&chp->rhp->rdev, &chp->cq);
+ kfree(chp);
+ return ERR_PTR(-ENOMEM);
+ }
if (ucontext) {
struct iwch_mm_entry *mm;
@@ -210,15 +213,27 @@ static struct ib_cq *iwch_create_cq(struct ib_device *ibdev, int entries, int ve
uresp.key = ucontext->key;
ucontext->key += PAGE_SIZE;
spin_unlock(&ucontext->mmap_lock);
- if (ib_copy_to_udata(udata, &uresp, sizeof (uresp))) {
+ mm->key = uresp.key;
+ mm->addr = virt_to_phys(chp->cq.queue);
+ if (udata->outlen < sizeof uresp) {
+ if (!warned++)
+ printk(KERN_WARNING MOD "Warning - "
+ "downlevel libcxgb3 (non-fatal).\n");
+ mm->len = PAGE_ALIGN((1UL << uresp.size_log2) *
+ sizeof(struct t3_cqe));
+ resplen = sizeof(struct iwch_create_cq_resp_v0);
+ } else {
+ mm->len = PAGE_ALIGN(((1UL << uresp.size_log2) + 1) *
+ sizeof(struct t3_cqe));
+ uresp.memsize = mm->len;
+ uresp.reserved = 0;
+ resplen = sizeof uresp;
+ }
+ if (ib_copy_to_udata(udata, &uresp, resplen)) {
kfree(mm);
iwch_destroy_cq(&chp->ibcq);
return ERR_PTR(-EFAULT);
}
- mm->key = uresp.key;
- mm->addr = virt_to_phys(chp->cq.queue);
- mm->len = PAGE_ALIGN((1UL << uresp.size_log2) *
- sizeof (struct t3_cqe));
insert_mmap(ucontext, mm);
}
PDBG("created cqid 0x%0x chp %p size 0x%0x, dma_addr 0x%0llx\n",
@@ -545,7 +560,7 @@ static int iwch_reregister_phys_mem(struct ib_mr *mr,
__be64 *page_list = NULL;
int shift = 0;
u64 total_size;
- int npages;
+ int npages = 0;
int ret;
PDBG("%s ib_mr %p ib_pd %p\n", __func__, mr, pd);
@@ -603,14 +618,13 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
{
__be64 *pages;
int shift, n, len;
- int i, j, k;
+ int i, k, entry;
int err = 0;
- struct ib_umem_chunk *chunk;
struct iwch_dev *rhp;
struct iwch_pd *php;
struct iwch_mr *mhp;
struct iwch_reg_user_mr_resp uresp;
-
+ struct scatterlist *sg;
PDBG("%s ib_pd %p\n", __func__, pd);
php = to_iwch_pd(pd);
@@ -630,9 +644,7 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
shift = ffs(mhp->umem->page_size) - 1;
- n = 0;
- list_for_each_entry(chunk, &mhp->umem->chunk_list, list)
- n += chunk->nents;
+ n = mhp->umem->nmap;
err = iwch_alloc_pbl(mhp, n);
if (err)
@@ -646,12 +658,10 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
i = n = 0;
- list_for_each_entry(chunk, &mhp->umem->chunk_list, list)
- for (j = 0; j < chunk->nmap; ++j) {
- len = sg_dma_len(&chunk->page_list[j]) >> shift;
+ for_each_sg(mhp->umem->sg_head.sgl, sg, mhp->umem->nmap, entry) {
+ len = sg_dma_len(sg) >> shift;
for (k = 0; k < len; ++k) {
- pages[i++] = cpu_to_be64(sg_dma_address(
- &chunk->page_list[j]) +
+ pages[i++] = cpu_to_be64(sg_dma_address(sg) +
mhp->umem->page_size * k);
if (i == PAGE_SIZE / sizeof *pages) {
err = iwch_write_pbl(mhp, pages, i, n);
@@ -661,7 +671,7 @@ static struct ib_mr *iwch_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
i = 0;
}
}
- }
+ }
if (i)
err = iwch_write_pbl(mhp, pages, i, n);
@@ -724,7 +734,7 @@ static struct ib_mr *iwch_get_dma_mr(struct ib_pd *pd, int acc)
return ibmr;
}
-static struct ib_mw *iwch_alloc_mw(struct ib_pd *pd)
+static struct ib_mw *iwch_alloc_mw(struct ib_pd *pd, enum ib_mw_type type)
{
struct iwch_dev *rhp;
struct iwch_pd *php;
@@ -733,6 +743,9 @@ static struct ib_mw *iwch_alloc_mw(struct ib_pd *pd)
u32 stag = 0;
int ret;
+ if (type != IB_MW_TYPE_1)
+ return ERR_PTR(-EINVAL);
+
php = to_iwch_pd(pd);
rhp = php->rhp;
mhp = kzalloc(sizeof(*mhp), GFP_KERNEL);
@@ -749,7 +762,11 @@ static struct ib_mw *iwch_alloc_mw(struct ib_pd *pd)
mhp->attr.stag = stag;
mmid = (stag) >> 8;
mhp->ibmw.rkey = stag;
- insert_handle(rhp, &rhp->mmidr, mhp, mmid);
+ if (insert_handle(rhp, &rhp->mmidr, mhp, mmid)) {
+ cxio_deallocate_window(&rhp->rdev, mhp->attr.stag);
+ kfree(mhp);
+ return ERR_PTR(-ENOMEM);
+ }
PDBG("%s mmid 0x%x mhp %p stag 0x%x\n", __func__, mmid, mhp, stag);
return &(mhp->ibmw);
}
@@ -765,8 +782,8 @@ static int iwch_dealloc_mw(struct ib_mw *mw)
mmid = (mw->rkey) >> 8;
cxio_deallocate_window(&rhp->rdev, mhp->attr.stag);
remove_handle(rhp, &rhp->mmidr, mmid);
- kfree(mhp);
PDBG("%s ib_mw %p mmid 0x%x ptr %p\n", __func__, mw, mmid, mhp);
+ kfree(mhp);
return 0;
}
@@ -777,37 +794,43 @@ static struct ib_mr *iwch_alloc_fast_reg_mr(struct ib_pd *pd, int pbl_depth)
struct iwch_mr *mhp;
u32 mmid;
u32 stag = 0;
- int ret;
+ int ret = 0;
php = to_iwch_pd(pd);
rhp = php->rhp;
mhp = kzalloc(sizeof(*mhp), GFP_KERNEL);
if (!mhp)
- return ERR_PTR(-ENOMEM);
+ goto err;
mhp->rhp = rhp;
ret = iwch_alloc_pbl(mhp, pbl_depth);
- if (ret) {
- kfree(mhp);
- return ERR_PTR(ret);
- }
+ if (ret)
+ goto err1;
mhp->attr.pbl_size = pbl_depth;
ret = cxio_allocate_stag(&rhp->rdev, &stag, php->pdid,
mhp->attr.pbl_size, mhp->attr.pbl_addr);
- if (ret) {
- iwch_free_pbl(mhp);
- kfree(mhp);
- return ERR_PTR(ret);
- }
+ if (ret)
+ goto err2;
mhp->attr.pdid = php->pdid;
mhp->attr.type = TPT_NON_SHARED_MR;
mhp->attr.stag = stag;
mhp->attr.state = 1;
mmid = (stag) >> 8;
mhp->ibmr.rkey = mhp->ibmr.lkey = stag;
- insert_handle(rhp, &rhp->mmidr, mhp, mmid);
+ if (insert_handle(rhp, &rhp->mmidr, mhp, mmid))
+ goto err3;
+
PDBG("%s mmid 0x%x mhp %p stag 0x%x\n", __func__, mmid, mhp, stag);
return &(mhp->ibmr);
+err3:
+ cxio_dereg_mem(&rhp->rdev, stag, mhp->attr.pbl_size,
+ mhp->attr.pbl_addr);
+err2:
+ iwch_free_pbl(mhp);
+err1:
+ kfree(mhp);
+err:
+ return ERR_PTR(ret);
}
static struct ib_fast_reg_page_list *iwch_alloc_fastreg_pbl(
@@ -960,7 +983,13 @@ static struct ib_qp *iwch_create_qp(struct ib_pd *pd,
spin_lock_init(&qhp->lock);
init_waitqueue_head(&qhp->wait);
atomic_set(&qhp->refcnt, 1);
- insert_handle(rhp, &rhp->qpidr, qhp, qhp->wq.qpid);
+
+ if (insert_handle(rhp, &rhp->qpidr, qhp, qhp->wq.qpid)) {
+ cxio_destroy_qp(&rhp->rdev, &qhp->wq,
+ ucontext ? &ucontext->uctx : &rhp->rdev.uctx);
+ kfree(qhp);
+ return ERR_PTR(-ENOMEM);
+ }
if (udata) {
@@ -1102,9 +1131,7 @@ static u64 fw_vers_string_to_u64(struct iwch_dev *iwch_dev)
char *cp, *next;
unsigned fw_maj, fw_min, fw_mic;
- rtnl_lock();
lldev->ethtool_ops->get_drvinfo(lldev, &info);
- rtnl_unlock();
next = info.fw_version + 1;
cp = strsep(&next, ".");
@@ -1154,14 +1181,42 @@ static int iwch_query_device(struct ib_device *ibdev,
static int iwch_query_port(struct ib_device *ibdev,
u8 port, struct ib_port_attr *props)
{
+ struct iwch_dev *dev;
+ struct net_device *netdev;
+ struct in_device *inetdev;
+
PDBG("%s ibdev %p\n", __func__, ibdev);
+
+ dev = to_iwch_dev(ibdev);
+ netdev = dev->rdev.port_info.lldevs[port-1];
+
+ memset(props, 0, sizeof(struct ib_port_attr));
props->max_mtu = IB_MTU_4096;
- props->lid = 0;
- props->lmc = 0;
- props->sm_lid = 0;
- props->sm_sl = 0;
- props->state = IB_PORT_ACTIVE;
- props->phys_state = 0;
+ if (netdev->mtu >= 4096)
+ props->active_mtu = IB_MTU_4096;
+ else if (netdev->mtu >= 2048)
+ props->active_mtu = IB_MTU_2048;
+ else if (netdev->mtu >= 1024)
+ props->active_mtu = IB_MTU_1024;
+ else if (netdev->mtu >= 512)
+ props->active_mtu = IB_MTU_512;
+ else
+ props->active_mtu = IB_MTU_256;
+
+ if (!netif_carrier_ok(netdev))
+ props->state = IB_PORT_DOWN;
+ else {
+ inetdev = in_dev_get(netdev);
+ if (inetdev) {
+ if (inetdev->ifa_list)
+ props->state = IB_PORT_ACTIVE;
+ else
+ props->state = IB_PORT_INIT;
+ in_dev_put(inetdev);
+ } else
+ props->state = IB_PORT_INIT;
+ }
+
props->port_cap_flags =
IB_PORT_CM_SUP |
IB_PORT_SNMP_TUNNEL_SUP |
@@ -1170,9 +1225,8 @@ static int iwch_query_port(struct ib_device *ibdev,
IB_PORT_VENDOR_CLASS_SUP | IB_PORT_BOOT_MGMT_SUP;
props->gid_tbl_len = 1;
props->pkey_tbl_len = 1;
- props->qkey_viol_cntr = 0;
props->active_width = 2;
- props->active_speed = 2;
+ props->active_speed = IB_SPEED_DDR;
props->max_msg_sz = -1;
return 0;
@@ -1187,28 +1241,6 @@ static ssize_t show_rev(struct device *dev, struct device_attribute *attr,
return sprintf(buf, "%d\n", iwch_dev->rdev.t3cdev_p->type);
}
-static int fw_supports_fastreg(struct iwch_dev *iwch_dev)
-{
- struct ethtool_drvinfo info;
- struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev;
- char *cp, *next;
- unsigned fw_maj, fw_min;
-
- rtnl_lock();
- lldev->ethtool_ops->get_drvinfo(lldev, &info);
- rtnl_unlock();
-
- next = info.fw_version+1;
- cp = strsep(&next, ".");
- sscanf(cp, "%i", &fw_maj);
- cp = strsep(&next, ".");
- sscanf(cp, "%i", &fw_min);
-
- PDBG("%s maj %u min %u\n", __func__, fw_maj, fw_min);
-
- return fw_maj > 6 || (fw_maj == 6 && fw_min > 0);
-}
-
static ssize_t show_fw_ver(struct device *dev, struct device_attribute *attr, char *buf)
{
struct iwch_dev *iwch_dev = container_of(dev, struct iwch_dev,
@@ -1217,9 +1249,7 @@ static ssize_t show_fw_ver(struct device *dev, struct device_attribute *attr, ch
struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev;
PDBG("%s dev 0x%p\n", __func__, dev);
- rtnl_lock();
lldev->ethtool_ops->get_drvinfo(lldev, &info);
- rtnl_unlock();
return sprintf(buf, "%s\n", info.fw_version);
}
@@ -1232,9 +1262,7 @@ static ssize_t show_hca(struct device *dev, struct device_attribute *attr,
struct net_device *lldev = iwch_dev->rdev.t3cdev_p->lldev;
PDBG("%s dev 0x%p\n", __func__, dev);
- rtnl_lock();
lldev->ethtool_ops->get_drvinfo(lldev, &info);
- rtnl_unlock();
return sprintf(buf, "%s\n", info.driver);
}
@@ -1325,12 +1353,12 @@ int iwch_register_device(struct iwch_dev *dev)
memset(&dev->ibdev.node_guid, 0, sizeof(dev->ibdev.node_guid));
memcpy(&dev->ibdev.node_guid, dev->rdev.t3cdev_p->lldev->dev_addr, 6);
dev->ibdev.owner = THIS_MODULE;
- dev->device_cap_flags = IB_DEVICE_LOCAL_DMA_LKEY | IB_DEVICE_MEM_WINDOW;
+ dev->device_cap_flags = IB_DEVICE_LOCAL_DMA_LKEY |
+ IB_DEVICE_MEM_WINDOW |
+ IB_DEVICE_MEM_MGT_EXTENSIONS;
/* cxgb3 supports STag 0. */
dev->ibdev.local_dma_lkey = 0;
- if (fw_supports_fastreg(dev))
- dev->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS;
dev->ibdev.uverbs_cmd_mask =
(1ull << IB_USER_VERBS_CMD_GET_CONTEXT) |
@@ -1357,7 +1385,6 @@ int iwch_register_device(struct iwch_dev *dev)
dev->ibdev.dma_device = &(dev->rdev.rnic_info.pdev->dev);
dev->ibdev.query_device = iwch_query_device;
dev->ibdev.query_port = iwch_query_port;
- dev->ibdev.modify_port = iwch_modify_port;
dev->ibdev.query_pkey = iwch_query_pkey;
dev->ibdev.query_gid = iwch_query_gid;
dev->ibdev.alloc_ucontext = iwch_alloc_ucontext;
@@ -1392,6 +1419,7 @@ int iwch_register_device(struct iwch_dev *dev)
dev->ibdev.post_send = iwch_post_send;
dev->ibdev.post_recv = iwch_post_receive;
dev->ibdev.get_protocol_stats = iwch_get_mib;
+ dev->ibdev.uverbs_abi_ver = IWCH_UVERBS_ABI_VERSION;
dev->ibdev.iwcm = kmalloc(sizeof(struct iw_cm_verbs), GFP_KERNEL);
if (!dev->ibdev.iwcm)
@@ -1406,7 +1434,7 @@ int iwch_register_device(struct iwch_dev *dev)
dev->ibdev.iwcm->rem_ref = iwch_qp_rem_ref;
dev->ibdev.iwcm->get_qp = iwch_get_qp;
- ret = ib_register_device(&dev->ibdev);
+ ret = ib_register_device(&dev->ibdev, NULL);
if (ret)
goto bail1;
@@ -1421,6 +1449,7 @@ int iwch_register_device(struct iwch_dev *dev)
bail2:
ib_unregister_device(&dev->ibdev);
bail1:
+ kfree(dev->ibdev.iwcm);
return ret;
}
@@ -1433,5 +1462,6 @@ void iwch_unregister_device(struct iwch_dev *dev)
device_remove_file(&dev->ibdev.dev,
iwch_class_attributes[i]);
ib_unregister_device(&dev->ibdev);
+ kfree(dev->ibdev.iwcm);
return;
}
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.h b/drivers/infiniband/hw/cxgb3/iwch_provider.h
index f5ceca05c43..87c14b0c5ac 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.h
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.h
@@ -103,6 +103,7 @@ struct iwch_cq {
struct iwch_dev *rhp;
struct t3_cq cq;
spinlock_t lock;
+ spinlock_t comp_handler_lock;
atomic_t refcnt;
wait_queue_head_t wait;
u32 __user *user_rptr_addr;
@@ -293,9 +294,16 @@ static inline u32 iwch_ib_to_tpt_access(int acc)
return (acc & IB_ACCESS_REMOTE_WRITE ? TPT_REMOTE_WRITE : 0) |
(acc & IB_ACCESS_REMOTE_READ ? TPT_REMOTE_READ : 0) |
(acc & IB_ACCESS_LOCAL_WRITE ? TPT_LOCAL_WRITE : 0) |
+ (acc & IB_ACCESS_MW_BIND ? TPT_MW_BIND : 0) |
TPT_LOCAL_READ;
}
+static inline u32 iwch_ib_to_tpt_bind_access(int acc)
+{
+ return (acc & IB_ACCESS_REMOTE_WRITE ? TPT_REMOTE_WRITE : 0) |
+ (acc & IB_ACCESS_REMOTE_READ ? TPT_REMOTE_READ : 0);
+}
+
enum iwch_mmid_state {
IWCH_STAG_STATE_VALID,
IWCH_STAG_STATE_INVALID
@@ -325,11 +333,9 @@ int iwch_bind_mw(struct ib_qp *qp,
struct ib_mw_bind *mw_bind);
int iwch_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc);
int iwch_post_terminate(struct iwch_qp *qhp, struct respQ_msg_t *rsp_msg);
-int iwch_post_zb_read(struct iwch_qp *qhp);
+int iwch_post_zb_read(struct iwch_ep *ep);
int iwch_register_device(struct iwch_dev *dev);
void iwch_unregister_device(struct iwch_dev *dev);
-int iwch_quiesce_qps(struct iwch_cq *chp);
-int iwch_resume_qps(struct iwch_cq *chp);
void stop_read_rep_timer(struct iwch_qp *qhp);
int iwch_register_mem(struct iwch_dev *rhp, struct iwch_pd *php,
struct iwch_mr *mhp, int shift);
diff --git a/drivers/infiniband/hw/cxgb3/iwch_qp.c b/drivers/infiniband/hw/cxgb3/iwch_qp.c
index 9a3be3a9d5d..b57c0befd96 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_qp.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_qp.c
@@ -29,6 +29,8 @@
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
+#include <linux/sched.h>
+#include <linux/gfp.h>
#include "iwch_provider.h"
#include "iwch.h"
#include "iwch_cm.h"
@@ -99,8 +101,8 @@ static int build_rdma_write(union t3_wr *wqe, struct ib_send_wr *wr,
if (wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM) {
plen = 4;
wqe->write.sgl[0].stag = wr->ex.imm_data;
- wqe->write.sgl[0].len = __constant_cpu_to_be32(0);
- wqe->write.num_sgle = __constant_cpu_to_be32(0);
+ wqe->write.sgl[0].len = cpu_to_be32(0);
+ wqe->write.num_sgle = cpu_to_be32(0);
*flit_cnt = 6;
} else {
plen = 0;
@@ -195,15 +197,12 @@ static int build_inv_stag(union t3_wr *wqe, struct ib_send_wr *wr,
return 0;
}
-/*
- * TBD: this is going to be moved to firmware. Missing pdid/qpid check for now.
- */
static int iwch_sgl2pbl_map(struct iwch_dev *rhp, struct ib_sge *sg_list,
u32 num_sgle, u32 * pbl_addr, u8 * page_size)
{
int i;
struct iwch_mr *mhp;
- u32 offset;
+ u64 offset;
for (i = 0; i < num_sgle; i++) {
mhp = get_mhp(rhp, (sg_list[i].lkey) >> 8);
@@ -235,8 +234,8 @@ static int iwch_sgl2pbl_map(struct iwch_dev *rhp, struct ib_sge *sg_list,
return -EINVAL;
}
offset = sg_list[i].addr - mhp->attr.va_fbo;
- offset += ((u32) mhp->attr.va_fbo) %
- (1UL << (12 + mhp->attr.page_size));
+ offset += mhp->attr.va_fbo &
+ ((1UL << (12 + mhp->attr.page_size)) - 1);
pbl_addr[i] = ((mhp->attr.pbl_addr -
rhp->rdev.rnic_info.pbl_base) >> 3) +
(offset >> (12 + mhp->attr.page_size));
@@ -266,8 +265,8 @@ static int build_rdma_recv(struct iwch_qp *qhp, union t3_wr *wqe,
wqe->recv.sgl[i].len = cpu_to_be32(wr->sg_list[i].length);
/* to in the WQE == the offset into the page */
- wqe->recv.sgl[i].to = cpu_to_be64(((u32) wr->sg_list[i].addr) %
- (1UL << (12 + page_size[i])));
+ wqe->recv.sgl[i].to = cpu_to_be64(((u32)wr->sg_list[i].addr) &
+ ((1UL << (12 + page_size[i])) - 1));
/* pbl_addr is the adapters address in the PBL */
wqe->recv.pbl_addr[i] = cpu_to_be32(pbl_addr[i]);
@@ -367,18 +366,19 @@ int iwch_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
spin_lock_irqsave(&qhp->lock, flag);
if (qhp->attr.state > IWCH_QP_STATE_RTS) {
spin_unlock_irqrestore(&qhp->lock, flag);
- return -EINVAL;
+ err = -EINVAL;
+ goto out;
}
num_wrs = Q_FREECNT(qhp->wq.sq_rptr, qhp->wq.sq_wptr,
qhp->wq.sq_size_log2);
- if (num_wrs <= 0) {
+ if (num_wrs == 0) {
spin_unlock_irqrestore(&qhp->lock, flag);
- return -ENOMEM;
+ err = -ENOMEM;
+ goto out;
}
while (wr) {
if (num_wrs == 0) {
err = -ENOMEM;
- *bad_wr = wr;
break;
}
idx = Q_PTR2IDX(qhp->wq.wptr, qhp->wq.size_log2);
@@ -430,10 +430,8 @@ int iwch_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
wr->opcode);
err = -EINVAL;
}
- if (err) {
- *bad_wr = wr;
+ if (err)
break;
- }
wqe->send.wrid.id0.hi = qhp->wq.sq_wptr;
sqp->wr_id = wr->wr_id;
sqp->opcode = wr2opcode(t3_wr_opcode);
@@ -455,7 +453,12 @@ int iwch_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
++(qhp->wq.sq_wptr);
}
spin_unlock_irqrestore(&qhp->lock, flag);
- ring_doorbell(qhp->wq.doorbell, qhp->wq.qpid);
+ if (cxio_wq_db_enabled(&qhp->wq))
+ ring_doorbell(qhp->wq.doorbell, qhp->wq.qpid);
+
+out:
+ if (err)
+ *bad_wr = wr;
return err;
}
@@ -473,18 +476,19 @@ int iwch_post_receive(struct ib_qp *ibqp, struct ib_recv_wr *wr,
spin_lock_irqsave(&qhp->lock, flag);
if (qhp->attr.state > IWCH_QP_STATE_RTS) {
spin_unlock_irqrestore(&qhp->lock, flag);
- return -EINVAL;
+ err = -EINVAL;
+ goto out;
}
num_wrs = Q_FREECNT(qhp->wq.rq_rptr, qhp->wq.rq_wptr,
qhp->wq.rq_size_log2) - 1;
if (!wr) {
spin_unlock_irqrestore(&qhp->lock, flag);
- return -EINVAL;
+ err = -ENOMEM;
+ goto out;
}
while (wr) {
if (wr->num_sge > T3_MAX_SGE) {
err = -EINVAL;
- *bad_wr = wr;
break;
}
idx = Q_PTR2IDX(qhp->wq.wptr, qhp->wq.size_log2);
@@ -496,10 +500,10 @@ int iwch_post_receive(struct ib_qp *ibqp, struct ib_recv_wr *wr,
err = build_zero_stag_recv(qhp, wqe, wr);
else
err = -ENOMEM;
- if (err) {
- *bad_wr = wr;
+
+ if (err)
break;
- }
+
build_fw_riwrh((void *) wqe, T3_WR_RCV, T3_COMPLETION_FLAG,
Q_GENBIT(qhp->wq.wptr, qhp->wq.size_log2),
0, sizeof(struct t3_receive_wr) >> 3, T3_SOPEOP);
@@ -512,7 +516,12 @@ int iwch_post_receive(struct ib_qp *ibqp, struct ib_recv_wr *wr,
num_wrs--;
}
spin_unlock_irqrestore(&qhp->lock, flag);
- ring_doorbell(qhp->wq.doorbell, qhp->wq.qpid);
+ if (cxio_wq_db_enabled(&qhp->wq))
+ ring_doorbell(qhp->wq.doorbell, qhp->wq.qpid);
+
+out:
+ if (err)
+ *bad_wr = wr;
return err;
}
@@ -545,7 +554,7 @@ int iwch_bind_mw(struct ib_qp *qp,
}
num_wrs = Q_FREECNT(qhp->wq.sq_rptr, qhp->wq.sq_wptr,
qhp->wq.sq_size_log2);
- if ((num_wrs) <= 0) {
+ if (num_wrs == 0) {
spin_unlock_irqrestore(&qhp->lock, flag);
return -ENOMEM;
}
@@ -558,18 +567,19 @@ int iwch_bind_mw(struct ib_qp *qp,
if (mw_bind->send_flags & IB_SEND_SIGNALED)
t3_wr_flags = T3_COMPLETION_FLAG;
- sgl.addr = mw_bind->addr;
- sgl.lkey = mw_bind->mr->lkey;
- sgl.length = mw_bind->length;
+ sgl.addr = mw_bind->bind_info.addr;
+ sgl.lkey = mw_bind->bind_info.mr->lkey;
+ sgl.length = mw_bind->bind_info.length;
wqe->bind.reserved = 0;
wqe->bind.type = TPT_VATO;
/* TBD: check perms */
- wqe->bind.perms = iwch_ib_to_tpt_access(mw_bind->mw_access_flags);
- wqe->bind.mr_stag = cpu_to_be32(mw_bind->mr->lkey);
+ wqe->bind.perms = iwch_ib_to_tpt_bind_access(
+ mw_bind->bind_info.mw_access_flags);
+ wqe->bind.mr_stag = cpu_to_be32(mw_bind->bind_info.mr->lkey);
wqe->bind.mw_stag = cpu_to_be32(mw->rkey);
- wqe->bind.mw_len = cpu_to_be32(mw_bind->length);
- wqe->bind.mw_va = cpu_to_be64(mw_bind->addr);
+ wqe->bind.mw_len = cpu_to_be32(mw_bind->bind_info.length);
+ wqe->bind.mw_va = cpu_to_be64(mw_bind->bind_info.addr);
err = iwch_sgl2pbl_map(rhp, &sgl, 1, &pbl_addr, &page_size);
if (err) {
spin_unlock_irqrestore(&qhp->lock, flag);
@@ -591,7 +601,8 @@ int iwch_bind_mw(struct ib_qp *qp,
++(qhp->wq.sq_wptr);
spin_unlock_irqrestore(&qhp->lock, flag);
- ring_doorbell(qhp->wq.doorbell, qhp->wq.qpid);
+ if (cxio_wq_db_enabled(&qhp->wq))
+ ring_doorbell(qhp->wq.doorbell, qhp->wq.qpid);
return err;
}
@@ -728,7 +739,7 @@ static inline void build_term_codes(struct respQ_msg_t *rsp_msg,
}
}
-int iwch_post_zb_read(struct iwch_qp *qhp)
+int iwch_post_zb_read(struct iwch_ep *ep)
{
union t3_wr *wqe;
struct sk_buff *skb;
@@ -745,17 +756,16 @@ int iwch_post_zb_read(struct iwch_qp *qhp)
wqe->read.rdmaop = T3_READ_REQ;
wqe->read.reserved[0] = 0;
wqe->read.reserved[1] = 0;
- wqe->read.reserved[2] = 0;
wqe->read.rem_stag = cpu_to_be32(1);
wqe->read.rem_to = cpu_to_be64(1);
wqe->read.local_stag = cpu_to_be32(1);
wqe->read.local_len = cpu_to_be32(0);
wqe->read.local_to = cpu_to_be64(1);
wqe->send.wrh.op_seop_flags = cpu_to_be32(V_FW_RIWR_OP(T3_WR_READ));
- wqe->send.wrh.gen_tid_len = cpu_to_be32(V_FW_RIWR_TID(qhp->ep->hwtid)|
+ wqe->send.wrh.gen_tid_len = cpu_to_be32(V_FW_RIWR_TID(ep->hwtid)|
V_FW_RIWR_LEN(flit_cnt));
skb->priority = CPL_PRIORITY_DATA;
- return cxgb3_ofld_send(qhp->rhp->rdev.t3cdev_p, skb);
+ return iwch_cxgb3_ofld_send(ep->com.qp->rhp->rdev.t3cdev_p, skb);
}
/*
@@ -787,61 +797,82 @@ int iwch_post_terminate(struct iwch_qp *qhp, struct respQ_msg_t *rsp_msg)
V_FW_RIWR_FLAGS(T3_COMPLETION_FLAG | T3_NOTIFY_FLAG));
wqe->send.wrh.gen_tid_len = cpu_to_be32(V_FW_RIWR_TID(qhp->ep->hwtid));
skb->priority = CPL_PRIORITY_DATA;
- return cxgb3_ofld_send(qhp->rhp->rdev.t3cdev_p, skb);
+ return iwch_cxgb3_ofld_send(qhp->rhp->rdev.t3cdev_p, skb);
}
/*
* Assumes qhp lock is held.
*/
-static void __flush_qp(struct iwch_qp *qhp, unsigned long *flag)
+static void __flush_qp(struct iwch_qp *qhp, struct iwch_cq *rchp,
+ struct iwch_cq *schp)
{
- struct iwch_cq *rchp, *schp;
int count;
int flushed;
- rchp = get_chp(qhp->rhp, qhp->attr.rcq);
- schp = get_chp(qhp->rhp, qhp->attr.scq);
PDBG("%s qhp %p rchp %p schp %p\n", __func__, qhp, rchp, schp);
/* take a ref on the qhp since we must release the lock */
atomic_inc(&qhp->refcnt);
- spin_unlock_irqrestore(&qhp->lock, *flag);
+ spin_unlock(&qhp->lock);
- /* locking heirarchy: cq lock first, then qp lock. */
- spin_lock_irqsave(&rchp->lock, *flag);
+ /* locking hierarchy: cq lock first, then qp lock. */
+ spin_lock(&rchp->lock);
spin_lock(&qhp->lock);
cxio_flush_hw_cq(&rchp->cq);
cxio_count_rcqes(&rchp->cq, &qhp->wq, &count);
flushed = cxio_flush_rq(&qhp->wq, &rchp->cq, count);
spin_unlock(&qhp->lock);
- spin_unlock_irqrestore(&rchp->lock, *flag);
- if (flushed)
+ spin_unlock(&rchp->lock);
+ if (flushed) {
+ spin_lock(&rchp->comp_handler_lock);
(*rchp->ibcq.comp_handler)(&rchp->ibcq, rchp->ibcq.cq_context);
+ spin_unlock(&rchp->comp_handler_lock);
+ }
- /* locking heirarchy: cq lock first, then qp lock. */
- spin_lock_irqsave(&schp->lock, *flag);
+ /* locking hierarchy: cq lock first, then qp lock. */
+ spin_lock(&schp->lock);
spin_lock(&qhp->lock);
cxio_flush_hw_cq(&schp->cq);
cxio_count_scqes(&schp->cq, &qhp->wq, &count);
flushed = cxio_flush_sq(&qhp->wq, &schp->cq, count);
spin_unlock(&qhp->lock);
- spin_unlock_irqrestore(&schp->lock, *flag);
- if (flushed)
+ spin_unlock(&schp->lock);
+ if (flushed) {
+ spin_lock(&schp->comp_handler_lock);
(*schp->ibcq.comp_handler)(&schp->ibcq, schp->ibcq.cq_context);
+ spin_unlock(&schp->comp_handler_lock);
+ }
/* deref */
if (atomic_dec_and_test(&qhp->refcnt))
wake_up(&qhp->wait);
- spin_lock_irqsave(&qhp->lock, *flag);
+ spin_lock(&qhp->lock);
}
-static void flush_qp(struct iwch_qp *qhp, unsigned long *flag)
+static void flush_qp(struct iwch_qp *qhp)
{
- if (qhp->ibqp.uobject)
+ struct iwch_cq *rchp, *schp;
+
+ rchp = get_chp(qhp->rhp, qhp->attr.rcq);
+ schp = get_chp(qhp->rhp, qhp->attr.scq);
+
+ if (qhp->ibqp.uobject) {
cxio_set_wq_in_error(&qhp->wq);
- else
- __flush_qp(qhp, flag);
+ cxio_set_cq_in_error(&rchp->cq);
+ spin_lock(&rchp->comp_handler_lock);
+ (*rchp->ibcq.comp_handler)(&rchp->ibcq, rchp->ibcq.cq_context);
+ spin_unlock(&rchp->comp_handler_lock);
+ if (schp != rchp) {
+ cxio_set_cq_in_error(&schp->cq);
+ spin_lock(&schp->comp_handler_lock);
+ (*schp->ibcq.comp_handler)(&schp->ibcq,
+ schp->ibcq.cq_context);
+ spin_unlock(&schp->comp_handler_lock);
+ }
+ return;
+ }
+ __flush_qp(qhp, rchp, schp);
}
@@ -852,7 +883,8 @@ u16 iwch_rqes_posted(struct iwch_qp *qhp)
{
union t3_wr *wqe = qhp->wq.queue;
u16 count = 0;
- while ((count+1) != 0 && fw_riwrh_opcode((struct fw_riwrh *)wqe) == T3_WR_RCV) {
+
+ while (count < USHRT_MAX && fw_riwrh_opcode((struct fw_riwrh *)wqe) == T3_WR_RCV) {
count++;
wqe++;
}
@@ -879,20 +911,13 @@ static int rdma_init(struct iwch_dev *rhp, struct iwch_qp *qhp,
(qhp->attr.mpa_attr.xmit_marker_enabled << 1) |
(qhp->attr.mpa_attr.crc_enabled << 2);
- /*
- * XXX - The IWCM doesn't quite handle getting these
- * attrs set before going into RTS. For now, just turn
- * them on always...
- */
-#if 0
- init_attr.qpcaps = qhp->attr.enableRdmaRead |
- (qhp->attr.enableRdmaWrite << 1) |
- (qhp->attr.enableBind << 2) |
- (qhp->attr.enable_stag0_fastreg << 3) |
- (qhp->attr.enable_stag0_fastreg << 4);
-#else
- init_attr.qpcaps = 0x1f;
-#endif
+ init_attr.qpcaps = uP_RI_QP_RDMA_READ_ENABLE |
+ uP_RI_QP_RDMA_WRITE_ENABLE |
+ uP_RI_QP_BIND_ENABLE;
+ if (!qhp->ibqp.uobject)
+ init_attr.qpcaps |= uP_RI_QP_STAG0_ENABLE |
+ uP_RI_QP_FAST_REGISTER_ENABLE;
+
init_attr.tcp_emss = qhp->ep->emss;
init_attr.ord = qhp->attr.max_ord;
init_attr.ird = qhp->attr.max_ird;
@@ -900,8 +925,7 @@ static int rdma_init(struct iwch_dev *rhp, struct iwch_qp *qhp,
init_attr.qp_dma_size = (1UL << qhp->wq.size_log2);
init_attr.rqe_count = iwch_rqes_posted(qhp);
init_attr.flags = qhp->attr.mpa_attr.initiator ? MPA_INITIATOR : 0;
- if (!qhp->ibqp.uobject)
- init_attr.flags |= PRIV_QP;
+ init_attr.chan = qhp->ep->l2t->smt_idx;
if (peer2peer) {
init_attr.rtr_type = RTR_READ;
if (init_attr.ord == 0 && qhp->attr.mpa_attr.initiator)
@@ -1008,7 +1032,7 @@ int iwch_modify_qp(struct iwch_dev *rhp, struct iwch_qp *qhp,
break;
case IWCH_QP_STATE_ERROR:
qhp->attr.state = IWCH_QP_STATE_ERROR;
- flush_qp(qhp, &flag);
+ flush_qp(qhp);
break;
default:
ret = -EINVAL;
@@ -1056,7 +1080,7 @@ int iwch_modify_qp(struct iwch_dev *rhp, struct iwch_qp *qhp,
}
switch (attrs->next_state) {
case IWCH_QP_STATE_IDLE:
- flush_qp(qhp, &flag);
+ flush_qp(qhp);
qhp->attr.state = IWCH_QP_STATE_IDLE;
qhp->attr.llp_stream_handle = NULL;
put_ep(&qhp->ep->com);
@@ -1082,7 +1106,6 @@ int iwch_modify_qp(struct iwch_dev *rhp, struct iwch_qp *qhp,
goto out;
}
qhp->attr.state = IWCH_QP_STATE_IDLE;
- memset(&qhp->attr, 0, sizeof(qhp->attr));
break;
case IWCH_QP_STATE_TERMINATE:
if (!internal) {
@@ -1111,7 +1134,7 @@ err:
free=1;
wake_up(&qhp->wait);
BUG_ON(!ep);
- flush_qp(qhp, &flag);
+ flush_qp(qhp);
out:
spin_unlock_irqrestore(&qhp->lock, flag);
@@ -1138,59 +1161,3 @@ out:
PDBG("%s exit state %d\n", __func__, qhp->attr.state);
return ret;
}
-
-static int quiesce_qp(struct iwch_qp *qhp)
-{
- spin_lock_irq(&qhp->lock);
- iwch_quiesce_tid(qhp->ep);
- qhp->flags |= QP_QUIESCED;
- spin_unlock_irq(&qhp->lock);
- return 0;
-}
-
-static int resume_qp(struct iwch_qp *qhp)
-{
- spin_lock_irq(&qhp->lock);
- iwch_resume_tid(qhp->ep);
- qhp->flags &= ~QP_QUIESCED;
- spin_unlock_irq(&qhp->lock);
- return 0;
-}
-
-int iwch_quiesce_qps(struct iwch_cq *chp)
-{
- int i;
- struct iwch_qp *qhp;
-
- for (i=0; i < T3_MAX_NUM_QP; i++) {
- qhp = get_qhp(chp->rhp, i);
- if (!qhp)
- continue;
- if ((qhp->attr.rcq == chp->cq.cqid) && !qp_quiesced(qhp)) {
- quiesce_qp(qhp);
- continue;
- }
- if ((qhp->attr.scq == chp->cq.cqid) && !qp_quiesced(qhp))
- quiesce_qp(qhp);
- }
- return 0;
-}
-
-int iwch_resume_qps(struct iwch_cq *chp)
-{
- int i;
- struct iwch_qp *qhp;
-
- for (i=0; i < T3_MAX_NUM_QP; i++) {
- qhp = get_qhp(chp->rhp, i);
- if (!qhp)
- continue;
- if ((qhp->attr.rcq == chp->cq.cqid) && qp_quiesced(qhp)) {
- resume_qp(qhp);
- continue;
- }
- if ((qhp->attr.scq == chp->cq.cqid) && qp_quiesced(qhp))
- resume_qp(qhp);
- }
- return 0;
-}
diff --git a/drivers/infiniband/hw/cxgb3/iwch_user.h b/drivers/infiniband/hw/cxgb3/iwch_user.h
index cb7086f558c..a277c31fcaf 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_user.h
+++ b/drivers/infiniband/hw/cxgb3/iwch_user.h
@@ -45,10 +45,18 @@ struct iwch_create_cq_req {
__u64 user_rptr_addr;
};
+struct iwch_create_cq_resp_v0 {
+ __u64 key;
+ __u32 cqid;
+ __u32 size_log2;
+};
+
struct iwch_create_cq_resp {
__u64 key;
__u32 cqid;
__u32 size_log2;
+ __u32 memsize;
+ __u32 reserved;
};
struct iwch_create_qp_resp {
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-12-09 09:07:20 +0000