crypto: hisilicon/qm - add the save operation of eqe and aeqe

The eqe and aeqe are device updated values that include the
valid bit and queue number. In the current process, there is no
memory barrier added, so it cannot be guaranteed that the valid
bit is read before other processes are executed. Since eqe and aeqe
are only 4 bytes and the device writes them to memory in a single
operation, saving the values of eqe and aeqe ensures that the valid
bit and queue number read by the CPU were written by the device
simultaneously.

Signed-off-by: nieweiqiang <nieweiqiang@huawei.com>
Signed-off-by: Chenghai Huang <huangchenghai2@huawei.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

diff --git a/drivers/crypto/hisilicon/qm.c b/drivers/crypto/hisilicon/qm.c
index 30e44cfb57ee7fe61072cc38dbb44f06b46fc7a7..8274da8b37ca66b323f3d215cb724c1ffe52f31e 100644 (file)
--- a/drivers/crypto/hisilicon/qm.c
+++ b/drivers/crypto/hisilicon/qm.c
@@ -64,10 +64,10 @@
 #define QM_EQE_AEQE_SIZE               (2UL << 12)
 #define QM_EQC_PHASE_SHIFT             16
 
-#define QM_EQE_PHASE(eqe)              ((le32_to_cpu((eqe)->dw0) >> 16) & 0x1)
+#define QM_EQE_PHASE(dw0)              (((dw0) >> 16) & 0x1)
 #define QM_EQE_CQN_MASK                        GENMASK(15, 0)
 
-#define QM_AEQE_PHASE(aeqe)            ((le32_to_cpu((aeqe)->dw0) >> 16) & 0x1)
+#define QM_AEQE_PHASE(dw0)             (((dw0) >> 16) & 0x1)
 #define QM_AEQE_TYPE_SHIFT             17
 #define QM_AEQE_TYPE_MASK              0xf
 #define QM_AEQE_CQN_MASK               GENMASK(15, 0)
@@ -976,23 +976,23 @@ static void qm_get_complete_eqe_num(struct hisi_qm *qm)
 {
        struct qm_eqe *eqe = qm->eqe + qm->status.eq_head;
        struct hisi_qm_poll_data *poll_data = NULL;
+       u32 dw0 = le32_to_cpu(eqe->dw0);
        u16 eq_depth = qm->eq_depth;
        u16 cqn, eqe_num = 0;
 
-       if (QM_EQE_PHASE(eqe) != qm->status.eqc_phase) {
+       if (QM_EQE_PHASE(dw0) != qm->status.eqc_phase) {
                atomic64_inc(&qm->debug.dfx.err_irq_cnt);
                qm_db(qm, 0, QM_DOORBELL_CMD_EQ, qm->status.eq_head, 0);
                return;
        }
 
-       cqn = le32_to_cpu(eqe->dw0) & QM_EQE_CQN_MASK;
+       cqn = dw0 & QM_EQE_CQN_MASK;
        if (unlikely(cqn >= qm->qp_num))
                return;
        poll_data = &qm->poll_data[cqn];
 
-       while (QM_EQE_PHASE(eqe) == qm->status.eqc_phase) {
-               cqn = le32_to_cpu(eqe->dw0) & QM_EQE_CQN_MASK;
-               poll_data->qp_finish_id[eqe_num] = cqn;
+       while (QM_EQE_PHASE(dw0) != qm->status.eqc_phase) {
+               poll_data->qp_finish_id[eqe_num] = dw0 & QM_EQE_CQN_MASK;
                eqe_num++;
 
                if (qm->status.eq_head == eq_depth - 1) {
@@ -1006,6 +1006,8 @@ static void qm_get_complete_eqe_num(struct hisi_qm *qm)
 
                if (eqe_num == (eq_depth >> 1) - 1)
                        break;
+
+               dw0 = le32_to_cpu(eqe->dw0);
        }
 
        poll_data->eqe_num = eqe_num;
@@ -1098,15 +1100,15 @@ static irqreturn_t qm_aeq_thread(int irq, void *data)
 {
        struct hisi_qm *qm = data;
        struct qm_aeqe *aeqe = qm->aeqe + qm->status.aeq_head;
+       u32 dw0 = le32_to_cpu(aeqe->dw0);
        u16 aeq_depth = qm->aeq_depth;
        u32 type, qp_id;
 
        atomic64_inc(&qm->debug.dfx.aeq_irq_cnt);
 
-       while (QM_AEQE_PHASE(aeqe) == qm->status.aeqc_phase) {
-               type = (le32_to_cpu(aeqe->dw0) >> QM_AEQE_TYPE_SHIFT) &
-                       QM_AEQE_TYPE_MASK;
-               qp_id = le32_to_cpu(aeqe->dw0) & QM_AEQE_CQN_MASK;
+       while (QM_AEQE_PHASE(dw0) == qm->status.aeqc_phase) {
+               type = (dw0 >> QM_AEQE_TYPE_SHIFT) & QM_AEQE_TYPE_MASK;
+               qp_id = dw0 & QM_AEQE_CQN_MASK;
 
                switch (type) {
                case QM_EQ_OVERFLOW:
@@ -1134,6 +1136,7 @@ static irqreturn_t qm_aeq_thread(int irq, void *data)
                        aeqe++;
                        qm->status.aeq_head++;
                }
+               dw0 = le32_to_cpu(aeqe->dw0);
        }
 
        qm_db(qm, 0, QM_DOORBELL_CMD_AEQ, qm->status.aeq_head, 0);