linux-kernel - [PATCH] iommu/arm-smmu-v3: remove the approach of MSI polling for CMD SYNC

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 17 Jul 2020 11:07:09 +1200
From:   Barry Song <song.bao.hua@...ilicon.com>
To:     <robin.murphy@....com>, <will@...nel.org>, <joro@...tes.org>
CC:     <linux-kernel@...r.kernel.org>, <linuxarm@...wei.com>,
        <linux-arm-kernel@...ts.infradead.org>,
        <iommu@...ts.linux-foundation.org>,
        Barry Song <song.bao.hua@...ilicon.com>,
        Prime Zeng <prime.zeng@...ilicon.com>
Subject: [PATCH] iommu/arm-smmu-v3: remove the approach of MSI polling for CMD SYNC

Before commit 587e6c10a7ce ("iommu/arm-smmu-v3: Reduce contention during
command-queue insertion"), msi polling perhaps performed better since
it could run outside the spin_lock_irqsave() while the code polling cons
reg was running in the lock.

But after the great reorganization of smmu queue, neither of these two
polling methods are running in a spinlock. And real tests show polling
cons reg via sev means smaller latency. It is probably because polling
by msi will ask hardware to write memory but sev polling depends on the
update of register only.

Using 16 threads to run netperf on hns3 100G NIC with UDP packet size
in 32768bytes and set iommu to strict, TX throughput can improve from
25227.74Mbps to 27145.59Mbps by this patch. In this case, SMMU is super
busy as hns3 sends map/unmap requests extremely frequently.

Cc: Prime Zeng <prime.zeng@...ilicon.com>
Signed-off-by: Barry Song <song.bao.hua@...ilicon.com>
---
 drivers/iommu/arm-smmu-v3.c | 46 +------------------------------------
 1 file changed, 1 insertion(+), 45 deletions(-)

diff --git a/drivers/iommu/arm-smmu-v3.c b/drivers/iommu/arm-smmu-v3.c
index f578677a5c41..e55282a636c8 100644
--- a/drivers/iommu/arm-smmu-v3.c
+++ b/drivers/iommu/arm-smmu-v3.c
@@ -964,12 +964,7 @@ static int arm_smmu_cmdq_build_cmd(u64 *cmd, struct arm_smmu_cmdq_ent *ent)
 		cmd[1] |= FIELD_PREP(CMDQ_PRI_1_RESP, ent->pri.resp);
 		break;
 	case CMDQ_OP_CMD_SYNC:
-		if (ent->sync.msiaddr) {
-			cmd[0] |= FIELD_PREP(CMDQ_SYNC_0_CS, CMDQ_SYNC_0_CS_IRQ);
-			cmd[1] |= ent->sync.msiaddr & CMDQ_SYNC_1_MSIADDR_MASK;
-		} else {
-			cmd[0] |= FIELD_PREP(CMDQ_SYNC_0_CS, CMDQ_SYNC_0_CS_SEV);
-		}
+		cmd[0] |= FIELD_PREP(CMDQ_SYNC_0_CS, CMDQ_SYNC_0_CS_SEV);
 		cmd[0] |= FIELD_PREP(CMDQ_SYNC_0_MSH, ARM_SMMU_SH_ISH);
 		cmd[0] |= FIELD_PREP(CMDQ_SYNC_0_MSIATTR, ARM_SMMU_MEMATTR_OIWB);
 		break;
@@ -983,21 +978,10 @@ static int arm_smmu_cmdq_build_cmd(u64 *cmd, struct arm_smmu_cmdq_ent *ent)
 static void arm_smmu_cmdq_build_sync_cmd(u64 *cmd, struct arm_smmu_device *smmu,
 					 u32 prod)
 {
-	struct arm_smmu_queue *q = &smmu->cmdq.q;
 	struct arm_smmu_cmdq_ent ent = {
 		.opcode = CMDQ_OP_CMD_SYNC,
 	};
 
-	/*
-	 * Beware that Hi16xx adds an extra 32 bits of goodness to its MSI
-	 * payload, so the write will zero the entire command on that platform.
-	 */
-	if (smmu->features & ARM_SMMU_FEAT_MSI &&
-	    smmu->features & ARM_SMMU_FEAT_COHERENCY) {
-		ent.sync.msiaddr = q->base_dma + Q_IDX(&q->llq, prod) *
-				   q->ent_dwords * 8;
-	}
-
 	arm_smmu_cmdq_build_cmd(cmd, &ent);
 }
 
@@ -1251,30 +1235,6 @@ static int arm_smmu_cmdq_poll_until_not_full(struct arm_smmu_device *smmu,
 	return ret;
 }
 
-/*
- * Wait until the SMMU signals a CMD_SYNC completion MSI.
- * Must be called with the cmdq lock held in some capacity.
- */
-static int __arm_smmu_cmdq_poll_until_msi(struct arm_smmu_device *smmu,
-					  struct arm_smmu_ll_queue *llq)
-{
-	int ret = 0;
-	struct arm_smmu_queue_poll qp;
-	struct arm_smmu_cmdq *cmdq = &smmu->cmdq;
-	u32 *cmd = (u32 *)(Q_ENT(&cmdq->q, llq->prod));
-
-	queue_poll_init(smmu, &qp);
-
-	/*
-	 * The MSI won't generate an event, since it's being written back
-	 * into the command queue.
-	 */
-	qp.wfe = false;
-	smp_cond_load_relaxed(cmd, !VAL || (ret = queue_poll(&qp)));
-	llq->cons = ret ? llq->prod : queue_inc_prod_n(llq, 1);
-	return ret;
-}
-
 /*
  * Wait until the SMMU cons index passes llq->prod.
  * Must be called with the cmdq lock held in some capacity.
@@ -1332,10 +1292,6 @@ static int __arm_smmu_cmdq_poll_until_consumed(struct arm_smmu_device *smmu,
 static int arm_smmu_cmdq_poll_until_sync(struct arm_smmu_device *smmu,
 					 struct arm_smmu_ll_queue *llq)
 {
-	if (smmu->features & ARM_SMMU_FEAT_MSI &&
-	    smmu->features & ARM_SMMU_FEAT_COHERENCY)
-		return __arm_smmu_cmdq_poll_until_msi(smmu, llq);
-
 	return __arm_smmu_cmdq_poll_until_consumed(smmu, llq);
 }
 
-- 
2.27.0