[PATCH][7/12] InfiniBand/mthca: optimize event queue handling

From: Roland Dreier
Date: Mon Jan 24 2005 - 01:35:34 EST


From: "Michael S. Tsirkin" <mst@xxxxxxxxxxxxxx>

Event queue handling performance improvements:
- Only calculate EQ entry address once, and don't truncate the
consumer index until we really need to.
- Only read ECR once. If a new event occurs while we're in the
interrupt handler, we'll get another interrupt anyway, since we
only clear events once.

Signed-off-by: Michael S. Tsirkin <mst@xxxxxxxxxxxxxx>
Signed-off-by: Roland Dreier <roland@xxxxxxxxxxx>

--- linux-bk.orig/drivers/infiniband/hw/mthca/mthca_provider.h 2005-01-23 08:30:27.000000000 -0800
+++ linux-bk/drivers/infiniband/hw/mthca/mthca_provider.h 2005-01-23 20:51:23.739805744 -0800
@@ -66,11 +66,11 @@
struct mthca_dev *dev;
int eqn;
u32 ecr_mask;
+ u32 cons_index;
u16 msi_x_vector;
u16 msi_x_entry;
int have_irq;
int nent;
- int cons_index;
struct mthca_buf_list *page_list;
struct mthca_mr mr;
};
--- linux-bk.orig/drivers/infiniband/hw/mthca/mthca_eq.c 2005-01-23 20:47:40.946675448 -0800
+++ linux-bk/drivers/infiniband/hw/mthca/mthca_eq.c 2005-01-23 20:51:23.740805592 -0800
@@ -164,12 +164,12 @@
MTHCA_ASYNC_EVENT_MASK;
}

-static inline void set_eq_ci(struct mthca_dev *dev, int eqn, int ci)
+static inline void set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u32 ci)
{
u32 doorbell[2];

- doorbell[0] = cpu_to_be32(MTHCA_EQ_DB_SET_CI | eqn);
- doorbell[1] = cpu_to_be32(ci);
+ doorbell[0] = cpu_to_be32(MTHCA_EQ_DB_SET_CI | eq->eqn);
+ doorbell[1] = cpu_to_be32(ci & (eq->nent - 1));

mthca_write64(doorbell,
dev->kar + MTHCA_EQ_DOORBELL,
@@ -200,21 +200,22 @@
MTHCA_GET_DOORBELL_LOCK(&dev->doorbell_lock));
}

-static inline struct mthca_eqe *get_eqe(struct mthca_eq *eq, int entry)
+static inline struct mthca_eqe *get_eqe(struct mthca_eq *eq, u32 entry)
{
- return eq->page_list[entry * MTHCA_EQ_ENTRY_SIZE / PAGE_SIZE].buf
- + (entry * MTHCA_EQ_ENTRY_SIZE) % PAGE_SIZE;
+ unsigned long off = (entry & (eq->nent - 1)) * MTHCA_EQ_ENTRY_SIZE;
+ return eq->page_list[off / PAGE_SIZE].buf + off % PAGE_SIZE;
}

-static inline int next_eqe_sw(struct mthca_eq *eq)
+static inline struct mthca_eqe* next_eqe_sw(struct mthca_eq *eq)
{
- return !(MTHCA_EQ_ENTRY_OWNER_HW &
- get_eqe(eq, eq->cons_index)->owner);
+ struct mthca_eqe* eqe;
+ eqe = get_eqe(eq, eq->cons_index);
+ return (MTHCA_EQ_ENTRY_OWNER_HW & eqe->owner) ? NULL : eqe;
}

-static inline void set_eqe_hw(struct mthca_eq *eq, int entry)
+static inline void set_eqe_hw(struct mthca_eqe *eqe)
{
- get_eqe(eq, entry)->owner = MTHCA_EQ_ENTRY_OWNER_HW;
+ eqe->owner = MTHCA_EQ_ENTRY_OWNER_HW;
}

static void port_change(struct mthca_dev *dev, int port, int active)
@@ -235,10 +236,10 @@
{
struct mthca_eqe *eqe;
int disarm_cqn;
+ int eqes_found = 0;

- while (next_eqe_sw(eq)) {
+ while ((eqe = next_eqe_sw(eq))) {
int set_ci = 0;
- eqe = get_eqe(eq, eq->cons_index);

/*
* Make sure we read EQ entry contents after we've
@@ -328,12 +329,13 @@
break;
};

- set_eqe_hw(eq, eq->cons_index);
- eq->cons_index = (eq->cons_index + 1) & (eq->nent - 1);
+ set_eqe_hw(eqe);
+ ++eq->cons_index;
+ eqes_found = 1;

if (set_ci) {
wmb(); /* see comment below */
- set_eq_ci(dev, eq->eqn, eq->cons_index);
+ set_eq_ci(dev, eq, eq->cons_index);
set_ci = 0;
}
}
@@ -347,8 +349,10 @@
* possibility of the HCA writing an entry and then
* having set_eqe_hw() overwrite the owner field.
*/
- wmb();
- set_eq_ci(dev, eq->eqn, eq->cons_index);
+ if (likely(eqes_found)) {
+ wmb();
+ set_eq_ci(dev, eq, eq->cons_index);
+ }
eq_req_not(dev, eq->eqn);
}

@@ -362,7 +366,7 @@
if (dev->eq_table.clr_mask)
writel(dev->eq_table.clr_mask, dev->eq_table.clr_int);

- while ((ecr = readl(dev->hcr + MTHCA_ECR_OFFSET + 4)) != 0) {
+ if ((ecr = readl(dev->hcr + MTHCA_ECR_OFFSET + 4)) != 0) {
work = 1;

writel(ecr, dev->hcr + MTHCA_ECR_CLR_OFFSET + 4);
@@ -440,7 +444,7 @@
}

for (i = 0; i < nent; ++i)
- set_eqe_hw(eq, i);
+ set_eqe_hw(get_eqe(eq, i));

eq->eqn = mthca_alloc(&dev->eq_table.alloc);
if (eq->eqn == -1)

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/