microsoft
/
WSL2-Linux-Kernel
зеркало из https://github.com/microsoft/WSL2-Linux-Kernel.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность

RDMA/irdma: optimize rx path by removing unnecessary copy 

In the function irdma_post_recv, the function irdma_copy_sg_list is
not needed since the struct irdma_sge and ib_sge have the similar
member variables. The struct irdma_sge can be replaced with the
struct ib_sge totally.

This can increase the rx performance of irdma.

Link: https://lore.kernel.org/r/20211030104226.253346-1-yanjun.zhu@linux.dev
Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev>
Reviewed-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
This commit is contained in:
Zhu Yanjun 2021-10-30 06:42:26 -04:00 коммит произвёл Jason Gunthorpe
Родитель 4e446714fb
Коммит 9ed8110c9b
3 изменённых файлов: 39 добавлений и 64 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

38
drivers/infiniband/hw/irdma/uk.c
Просмотреть файл

@ -13,16 +13,16 @@
* @sge: sge length and stag
* @valid: The wqe valid
*/
static void irdma_set_fragment(__le64 *wqe, u32 offset, struct irdma_sge *sge,
static void irdma_set_fragment(__le64 *wqe, u32 offset, struct ib_sge *sge,
u8 valid)
{
if (sge) {
set_64bit_val(wqe, offset,
FIELD_PREP(IRDMAQPSQ_FRAG_TO, sge->tag_off));
FIELD_PREP(IRDMAQPSQ_FRAG_TO, sge->addr));
set_64bit_val(wqe, offset + 8,
FIELD_PREP(IRDMAQPSQ_VALID, valid) |
FIELD_PREP(IRDMAQPSQ_FRAG_LEN, sge->len) |
FIELD_PREP(IRDMAQPSQ_FRAG_STAG, sge->stag));
FIELD_PREP(IRDMAQPSQ_FRAG_LEN, sge->length) |
FIELD_PREP(IRDMAQPSQ_FRAG_STAG, sge->lkey));
} else {
set_64bit_val(wqe, offset, 0);
set_64bit_val(wqe, offset + 8,
@ -38,14 +38,14 @@ static void irdma_set_fragment(__le64 *wqe, u32 offset, struct irdma_sge *sge,
* @valid: wqe valid flag
*/
static void irdma_set_fragment_gen_1(__le64 *wqe, u32 offset,
struct irdma_sge *sge, u8 valid)
struct ib_sge *sge, u8 valid)
{
if (sge) {
set_64bit_val(wqe, offset,
FIELD_PREP(IRDMAQPSQ_FRAG_TO, sge->tag_off));
FIELD_PREP(IRDMAQPSQ_FRAG_TO, sge->addr));
set_64bit_val(wqe, offset + 8,
FIELD_PREP(IRDMAQPSQ_GEN1_FRAG_LEN, sge->len) |
FIELD_PREP(IRDMAQPSQ_GEN1_FRAG_STAG, sge->stag));
FIELD_PREP(IRDMAQPSQ_GEN1_FRAG_LEN, sge->length) |
FIELD_PREP(IRDMAQPSQ_GEN1_FRAG_STAG, sge->lkey));
} else {
set_64bit_val(wqe, offset, 0);
set_64bit_val(wqe, offset + 8, 0);
@ -289,7 +289,7 @@ enum irdma_status_code irdma_uk_rdma_write(struct irdma_qp_uk *qp,
return IRDMA_ERR_INVALID_FRAG_COUNT;
for (i = 0; i < op_info->num_lo_sges; i++)
total_size += op_info->lo_sg_list[i].len;
total_size += op_info->lo_sg_list[i].length;
read_fence |= info->read_fence;
@ -310,7 +310,7 @@ enum irdma_status_code irdma_uk_rdma_write(struct irdma_qp_uk *qp,
irdma_clr_wqes(qp, wqe_idx);
set_64bit_val(wqe, 16,
FIELD_PREP(IRDMAQPSQ_FRAG_TO, op_info->rem_addr.tag_off));
FIELD_PREP(IRDMAQPSQ_FRAG_TO, op_info->rem_addr.addr));
if (info->imm_data_valid) {
set_64bit_val(wqe, 0,
@ -339,7 +339,7 @@ enum irdma_status_code irdma_uk_rdma_write(struct irdma_qp_uk *qp,
++addl_frag_cnt;
}
hdr = FIELD_PREP(IRDMAQPSQ_REMSTAG, op_info->rem_addr.stag) |
hdr = FIELD_PREP(IRDMAQPSQ_REMSTAG, op_info->rem_addr.lkey) |
FIELD_PREP(IRDMAQPSQ_OPCODE, info->op_type) |
FIELD_PREP(IRDMAQPSQ_IMMDATAFLAG, info->imm_data_valid) |
FIELD_PREP(IRDMAQPSQ_REPORTRTT, info->report_rtt) |
@ -391,7 +391,7 @@ enum irdma_status_code irdma_uk_rdma_read(struct irdma_qp_uk *qp,
return IRDMA_ERR_INVALID_FRAG_COUNT;
for (i = 0; i < op_info->num_lo_sges; i++)
total_size += op_info->lo_sg_list[i].len;
total_size += op_info->lo_sg_list[i].length;
ret_code = irdma_fragcnt_to_quanta_sq(op_info->num_lo_sges, &quanta);
if (ret_code)
@ -426,8 +426,8 @@ enum irdma_status_code irdma_uk_rdma_read(struct irdma_qp_uk *qp,
++addl_frag_cnt;
}
set_64bit_val(wqe, 16,
FIELD_PREP(IRDMAQPSQ_FRAG_TO, op_info->rem_addr.tag_off));
hdr = FIELD_PREP(IRDMAQPSQ_REMSTAG, op_info->rem_addr.stag) |
FIELD_PREP(IRDMAQPSQ_FRAG_TO, op_info->rem_addr.addr));
hdr = FIELD_PREP(IRDMAQPSQ_REMSTAG, op_info->rem_addr.lkey) |
FIELD_PREP(IRDMAQPSQ_REPORTRTT, (info->report_rtt ? 1 : 0)) |
FIELD_PREP(IRDMAQPSQ_ADDFRAGCNT, addl_frag_cnt) |
FIELD_PREP(IRDMAQPSQ_OPCODE,
@ -477,7 +477,7 @@ enum irdma_status_code irdma_uk_send(struct irdma_qp_uk *qp,
return IRDMA_ERR_INVALID_FRAG_COUNT;
for (i = 0; i < op_info->num_sges; i++)
total_size += op_info->sg_list[i].len;
total_size += op_info->sg_list[i].length;
if (info->imm_data_valid)
frag_cnt = op_info->num_sges + 1;
@ -705,9 +705,9 @@ irdma_uk_inline_rdma_write(struct irdma_qp_uk *qp, struct irdma_post_sq_info *in
read_fence |= info->read_fence;
set_64bit_val(wqe, 16,
FIELD_PREP(IRDMAQPSQ_FRAG_TO, op_info->rem_addr.tag_off));
FIELD_PREP(IRDMAQPSQ_FRAG_TO, op_info->rem_addr.addr));
hdr = FIELD_PREP(IRDMAQPSQ_REMSTAG, op_info->rem_addr.stag) |
hdr = FIELD_PREP(IRDMAQPSQ_REMSTAG, op_info->rem_addr.lkey) |
FIELD_PREP(IRDMAQPSQ_OPCODE, info->op_type) |
FIELD_PREP(IRDMAQPSQ_INLINEDATALEN, op_info->len) |
FIELD_PREP(IRDMAQPSQ_REPORTRTT, info->report_rtt ? 1 : 0) |
@ -826,7 +826,7 @@ irdma_uk_stag_local_invalidate(struct irdma_qp_uk *qp,
u64 hdr;
u32 wqe_idx;
bool local_fence = false;
struct irdma_sge sge = {};
struct ib_sge sge = {};
info->push_wqe = qp->push_db ? true : false;
op_info = &info->op.inv_local_stag;
@ -839,7 +839,7 @@ irdma_uk_stag_local_invalidate(struct irdma_qp_uk *qp,
irdma_clr_wqes(qp, wqe_idx);
sge.stag = op_info->target_stag;
sge.lkey = op_info->target_stag;
qp->wqe_ops.iw_set_fragment(wqe, 0, &sge, 0);
set_64bit_val(wqe, 16, 0);

23
drivers/infiniband/hw/irdma/user.h
Просмотреть файл

@ -16,7 +16,6 @@
#define irdma_access_privileges u32
#define irdma_physical_fragment u64
#define irdma_address_list u64 *
#define irdma_sgl struct irdma_sge *
#define IRDMA_MAX_MR_SIZE 0x200000000000ULL
@ -151,12 +150,6 @@ struct irdma_cq_uk;
struct irdma_qp_uk_init_info;
struct irdma_cq_uk_init_info;
struct irdma_sge {
irdma_tagged_offset tag_off;
u32 len;
irdma_stag stag;
};
struct irdma_ring {
u32 head;
u32 tail;
@ -172,7 +165,7 @@ struct irdma_extended_cqe {
};
struct irdma_post_send {
irdma_sgl sg_list;
struct ib_sge *sg_list;
u32 num_sges;
u32 qkey;
u32 dest_qp;
@ -189,26 +182,26 @@ struct irdma_post_inline_send {
struct irdma_post_rq_info {
u64 wr_id;
irdma_sgl sg_list;
struct ib_sge *sg_list;
u32 num_sges;
};
struct irdma_rdma_write {
irdma_sgl lo_sg_list;
struct ib_sge *lo_sg_list;
u32 num_lo_sges;
struct irdma_sge rem_addr;
struct ib_sge rem_addr;
};
struct irdma_inline_rdma_write {
void *data;
u32 len;
struct irdma_sge rem_addr;
struct ib_sge rem_addr;
};
struct irdma_rdma_read {
irdma_sgl lo_sg_list;
struct ib_sge *lo_sg_list;
u32 num_lo_sges;
struct irdma_sge rem_addr;
struct ib_sge rem_addr;
};
struct irdma_bind_window {
@ -304,7 +297,7 @@ enum irdma_status_code irdma_uk_stag_local_invalidate(struct irdma_qp_uk *qp,
struct irdma_wqe_uk_ops {
void (*iw_copy_inline_data)(u8 *dest, u8 *src, u32 len, u8 polarity);
u16 (*iw_inline_data_size_to_quanta)(u32 data_size);
void (*iw_set_fragment)(__le64 *wqe, u32 offset, struct irdma_sge *sge,
void (*iw_set_fragment)(__le64 *wqe, u32 offset, struct ib_sge *sge,
u8 valid);
void (*iw_set_mw_bind_wqe)(__le64 *wqe,
struct irdma_bind_window *op_info);

42
drivers/infiniband/hw/irdma/verbs.c
Просмотреть файл

@ -3038,24 +3038,6 @@ done:
return 0;
}
/**
* irdma_copy_sg_list - copy sg list for qp
* @sg_list: copied into sg_list
* @sgl: copy from sgl
* @num_sges: count of sg entries
*/
static void irdma_copy_sg_list(struct irdma_sge *sg_list, struct ib_sge *sgl,
int num_sges)
{
unsigned int i;
for (i = 0; (i < num_sges) && (i < IRDMA_MAX_WQ_FRAGMENT_COUNT); i++) {
sg_list[i].tag_off = sgl[i].addr;
sg_list[i].len = sgl[i].length;
sg_list[i].stag = sgl[i].lkey;
}
}
/**
* irdma_post_send - kernel application wr
* @ibqp: qp ptr for wr
@ -3132,8 +3114,7 @@ static int irdma_post_send(struct ib_qp *ibqp,
ret = irdma_uk_inline_send(ukqp, &info, false);
} else {
info.op.send.num_sges = ib_wr->num_sge;
info.op.send.sg_list = (struct irdma_sge *)
ib_wr->sg_list;
info.op.send.sg_list = ib_wr->sg_list;
if (iwqp->ibqp.qp_type == IB_QPT_UD ||
iwqp->ibqp.qp_type == IB_QPT_GSI) {
ah = to_iwah(ud_wr(ib_wr)->ah);
@ -3168,15 +3149,18 @@ static int irdma_post_send(struct ib_qp *ibqp,
if (ib_wr->send_flags & IB_SEND_INLINE) {
info.op.inline_rdma_write.data = (void *)(uintptr_t)ib_wr->sg_list[0].addr;
info.op.inline_rdma_write.len = ib_wr->sg_list[0].length;
info.op.inline_rdma_write.rem_addr.tag_off = rdma_wr(ib_wr)->remote_addr;
info.op.inline_rdma_write.rem_addr.stag = rdma_wr(ib_wr)->rkey;
info.op.inline_rdma_write.len =
ib_wr->sg_list[0].length;
info.op.inline_rdma_write.rem_addr.addr =
rdma_wr(ib_wr)->remote_addr;
info.op.inline_rdma_write.rem_addr.lkey =
rdma_wr(ib_wr)->rkey;
ret = irdma_uk_inline_rdma_write(ukqp, &info, false);
} else {
info.op.rdma_write.lo_sg_list = (void *)ib_wr->sg_list;
info.op.rdma_write.num_lo_sges = ib_wr->num_sge;
info.op.rdma_write.rem_addr.tag_off = rdma_wr(ib_wr)->remote_addr;
info.op.rdma_write.rem_addr.stag = rdma_wr(ib_wr)->rkey;
info.op.rdma_write.rem_addr.addr = rdma_wr(ib_wr)->remote_addr;
info.op.rdma_write.rem_addr.lkey = rdma_wr(ib_wr)->rkey;
ret = irdma_uk_rdma_write(ukqp, &info, false);
}
@ -3197,8 +3181,8 @@ static int irdma_post_send(struct ib_qp *ibqp,
break;
}
info.op_type = IRDMA_OP_TYPE_RDMA_READ;
info.op.rdma_read.rem_addr.tag_off = rdma_wr(ib_wr)->remote_addr;
info.op.rdma_read.rem_addr.stag = rdma_wr(ib_wr)->rkey;
info.op.rdma_read.rem_addr.addr = rdma_wr(ib_wr)->remote_addr;
info.op.rdma_read.rem_addr.lkey = rdma_wr(ib_wr)->rkey;
info.op.rdma_read.lo_sg_list = (void *)ib_wr->sg_list;
info.op.rdma_read.num_lo_sges = ib_wr->num_sge;
@ -3285,7 +3269,6 @@ static int irdma_post_recv(struct ib_qp *ibqp,
struct irdma_qp *iwqp;
struct irdma_qp_uk *ukqp;
struct irdma_post_rq_info post_recv = {};
struct irdma_sge sg_list[IRDMA_MAX_WQ_FRAGMENT_COUNT];
enum irdma_status_code ret = 0;
unsigned long flags;
int err = 0;
@ -3300,8 +3283,7 @@ static int irdma_post_recv(struct ib_qp *ibqp,
while (ib_wr) {
post_recv.num_sges = ib_wr->num_sge;
post_recv.wr_id = ib_wr->wr_id;
irdma_copy_sg_list(sg_list, ib_wr->sg_list, ib_wr->num_sge);
post_recv.sg_list = sg_list;
post_recv.sg_list = ib_wr->sg_list;
ret = irdma_uk_post_receive(ukqp, &post_recv);
if (ret) {
ibdev_dbg(&iwqp->iwdev->ibdev,