iommu/arm-smmu-v3: Add and use static helper function arm_smmu_cmdq_issue_cmd_with_sync()

The obvious key to the performance optimization of commit 587e6c10a7ce
("iommu/arm-smmu-v3: Reduce contention during command-queue insertion") is
to allow multiple cores to insert commands in parallel after a brief mutex
contention.

Obviously, inserting as many commands at a time as possible can reduce the
number of times the mutex contention participates, thereby improving the
overall performance. At least it reduces the number of calls to function
arm_smmu_cmdq_issue_cmdlist().

Therefore, function arm_smmu_cmdq_issue_cmd_with_sync() is added to insert
the 'cmd+sync' commands at a time.

Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
Link: https://lore.kernel.org/r/20210811114852.2429-3-thunder.leizhen@huawei.com
Signed-off-by: Will Deacon <will@kernel.org>

1 files changed, 19 insertions, 16 deletions

diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
index ac0c4b21cfa8..3cdb01f93f1b 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
@@ -845,8 +845,9 @@ static int arm_smmu_cmdq_issue_cmdlist(struct arm_smmu_device *smmu,
 	return ret;
 }
 
-static int arm_smmu_cmdq_issue_cmd(struct arm_smmu_device *smmu,
-				   struct arm_smmu_cmdq_ent *ent)
+static int __arm_smmu_cmdq_issue_cmd(struct arm_smmu_device *smmu,
+				     struct arm_smmu_cmdq_ent *ent,
+				     bool sync)
 {
 	u64 cmd[CMDQ_ENT_DWORDS];
 
@@ -856,12 +857,19 @@ static int arm_smmu_cmdq_issue_cmd(struct arm_smmu_device *smmu,
 		return -EINVAL;
 	}
 
-	return arm_smmu_cmdq_issue_cmdlist(smmu, cmd, 1, false);
+	return arm_smmu_cmdq_issue_cmdlist(smmu, cmd, 1, sync);
 }
 
-static int arm_smmu_cmdq_issue_sync(struct arm_smmu_device *smmu)
+static int arm_smmu_cmdq_issue_cmd(struct arm_smmu_device *smmu,
+				   struct arm_smmu_cmdq_ent *ent)
 {
-	return arm_smmu_cmdq_issue_cmdlist(smmu, NULL, 0, true);
+	return __arm_smmu_cmdq_issue_cmd(smmu, ent, false);
+}
+
+static int arm_smmu_cmdq_issue_cmd_with_sync(struct arm_smmu_device *smmu,
+					     struct arm_smmu_cmdq_ent *ent)
+{
+	return __arm_smmu_cmdq_issue_cmd(smmu, ent, true);
 }
 
 static void arm_smmu_cmdq_batch_add(struct arm_smmu_device *smmu,
@@ -929,8 +937,7 @@ void arm_smmu_tlb_inv_asid(struct arm_smmu_device *smmu, u16 asid)
 		.tlbi.asid = asid,
 	};
 
-	arm_smmu_cmdq_issue_cmd(smmu, &cmd);
-	arm_smmu_cmdq_issue_sync(smmu);
+	arm_smmu_cmdq_issue_cmd_with_sync(smmu, &cmd);
 }
 
 static void arm_smmu_sync_cd(struct arm_smmu_domain *smmu_domain,
@@ -1211,8 +1218,7 @@ static void arm_smmu_sync_ste_for_sid(struct arm_smmu_device *smmu, u32 sid)
 		},
 	};
 
-	arm_smmu_cmdq_issue_cmd(smmu, &cmd);
-	arm_smmu_cmdq_issue_sync(smmu);
+	arm_smmu_cmdq_issue_cmd_with_sync(smmu, &cmd);
 }
 
 static void arm_smmu_write_strtab_ent(struct arm_smmu_master *master, u32 sid,
@@ -1824,8 +1830,7 @@ static void arm_smmu_tlb_inv_context(void *cookie)
 	} else {
 		cmd.opcode	= CMDQ_OP_TLBI_S12_VMALL;
 		cmd.tlbi.vmid	= smmu_domain->s2_cfg.vmid;
-		arm_smmu_cmdq_issue_cmd(smmu, &cmd);
-		arm_smmu_cmdq_issue_sync(smmu);
+		arm_smmu_cmdq_issue_cmd_with_sync(smmu, &cmd);
 	}
 	arm_smmu_atc_inv_domain(smmu_domain, 0, 0, 0);
 }
@@ -3339,18 +3344,16 @@ static int arm_smmu_device_reset(struct arm_smmu_device *smmu, bool bypass)
 
 	/* Invalidate any cached configuration */
 	cmd.opcode = CMDQ_OP_CFGI_ALL;
-	arm_smmu_cmdq_issue_cmd(smmu, &cmd);
-	arm_smmu_cmdq_issue_sync(smmu);
+	arm_smmu_cmdq_issue_cmd_with_sync(smmu, &cmd);
 
 	/* Invalidate any stale TLB entries */
 	if (smmu->features & ARM_SMMU_FEAT_HYP) {
 		cmd.opcode = CMDQ_OP_TLBI_EL2_ALL;
-		arm_smmu_cmdq_issue_cmd(smmu, &cmd);
+		arm_smmu_cmdq_issue_cmd_with_sync(smmu, &cmd);
 	}
 
 	cmd.opcode = CMDQ_OP_TLBI_NSNH_ALL;
-	arm_smmu_cmdq_issue_cmd(smmu, &cmd);
-	arm_smmu_cmdq_issue_sync(smmu);
+	arm_smmu_cmdq_issue_cmd_with_sync(smmu, &cmd);
 
 	/* Event queue */
 	writeq_relaxed(smmu->evtq.q.q_base, smmu->base + ARM_SMMU_EVTQ_BASE);