March 2021 - Linuxarm - mailweb.openeuler.org

[PATCH for-next 0/2] RDMA/hns: Support to select congestion control algorithm
by Weihang Li 24 Mar '21

24 Mar '21

The configuration of congestion control algorithm is recorded in the firmware, the driver queries it and then sets it to the hardware. Wei Xu (1): RDMA/hns: Support query information of functions from FW Yangyang Li (1): RDMA/hns: Support congestion control type selection according to the FW drivers/infiniband/hw/hns/hns_roce_device.h | 11 ++ drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 192 ++++++++++++++++++++++++++++ drivers/infiniband/hw/hns/hns_roce_hw_v2.h | 34 ++++- drivers/infiniband/hw/hns/hns_roce_main.c | 2 + 4 files changed, 236 insertions(+), 3 deletions(-) -- 2.8.1

3 6

Re: [bug report] net: hns3: vf indexing in hclge_add_fd_entry()
by shenjian (K) 24 Mar '21

24 Mar '21

Hello Dan, Thanks for your report! 在 2021/3/24 15:01, Dan Carpenter 写道: > Hello Jian Shen, > > The patch 5f2b1238b33c: "net: hns3: refactor out > hclge_add_fd_entry()" from Mar 22, 2021, leads to the following > static checker warning: > > drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c:6512 hclge_fd_parse_ring_cookie() > warn: array off by one? 'hdev->vport[vf]' > > drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c > 6493 static int hclge_fd_parse_ring_cookie(struct hclge_dev *hdev, u64 ring_cookie, > 6494 u16 *vport_id, u8 *action, u16 *queue_id) > 6495 { > 6496 struct hclge_vport *vport = hdev->vport; > 6497 > 6498 if (ring_cookie == RX_CLS_FLOW_DISC) { > 6499 *action = HCLGE_FD_ACTION_DROP_PACKET; > 6500 } else { > 6501 u32 ring = ethtool_get_flow_spec_ring(ring_cookie); > 6502 u8 vf = ethtool_get_flow_spec_ring_vf(ring_cookie); > 6503 u16 tqps; > 6504 > 6505 if (vf > hdev->num_req_vfs) { > ^^^^^^^^^^^^^^^^^^^^^^ > > The is off by one but checking hdev->num_req_vfs in this context doesn't > make sense. Should it instead be check hdev->num_alloc_vport? Also > should we add HCLGE_VF_VPORT_START_NUM? > > vf = ethtool_get_flow_spec_ring_vf(ring_cookie); > vf += HCLGE_VF_VPORT_START_NUM; > if (vf >= hdev->num_alloc_vport) > return -EINVAL; > > 6506 dev_err(&hdev->pdev->dev, > 6507 "Error: vf id (%u) > max vf num (%u)\n", > ^^ > Use >= > > 6508 vf, hdev->num_req_vfs); > 6509 return -EINVAL; > 6510 } > 6511 > 6512 *vport_id = vf ? hdev->vport[vf].vport_id : vport->vport_id; > ^^^^^^^^^ > 6513 tqps = hdev->vport[vf].nic.kinfo.num_tqps; > ^^^ > The vport array has hdev->num_vmdq_vport + hdev->num_req_vfs + 1; > elements. ->vport[0] is tqp_main_vport. The next elements are > hdev->num_vmdq_vport and the last part of the array is hdev->num_req_vfs. > > Another possibility is that perhaps this is what was intended? > > idx = vf + 1 + hdev->num_vmdq_vport; > > *vport_id = vf ? vport[idx].vport_id : vport[0].vport_id; > tqps = vport[idx].nic.kinfo.num_tqps; > > There is related code that offers clues but I'm not sure what to do. > > 6514 > 6515 if (ring >= tqps) { > 6516 dev_err(&hdev->pdev->dev, > 6517 "Error: queue id (%u) > max tqp num (%u)\n", > 6518 ring, tqps - 1); > 6519 return -EINVAL; > 6520 } > 6521 > 6522 *action = HCLGE_FD_ACTION_SELECT_QUEUE; > 6523 *queue_id = ring; > 6524 } > 6525 > 6526 return 0; > 6527 } > > [ snip ] > > 9111 static bool hclge_check_vf_mac_exist(struct hclge_vport *vport, int vf_idx, > 9112 u8 *mac_addr) > 9113 { > 9114 struct hclge_mac_vlan_tbl_entry_cmd req; > 9115 struct hclge_dev *hdev = vport->back; > 9116 struct hclge_desc desc; > 9117 u16 egress_port = 0; > 9118 int i; > 9119 > 9120 if (is_zero_ether_addr(mac_addr)) > 9121 return false; > 9122 > 9123 memset(&req, 0, sizeof(req)); > 9124 hnae3_set_field(egress_port, HCLGE_MAC_EPORT_VFID_M, > 9125 HCLGE_MAC_EPORT_VFID_S, vport->vport_id); > 9126 req.egress_port = cpu_to_le16(egress_port); > 9127 hclge_prepare_mac_addr(&req, mac_addr, false); > 9128 > 9129 if (hclge_lookup_mac_vlan_tbl(vport, &req, &desc, false) != -ENOENT) > 9130 return true; > 9131 > 9132 vf_idx += HCLGE_VF_VPORT_START_NUM; > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > We're are skipping the first element. Should it instead be?: > > vf_idx += hdev->num_vmdq_vport + 1; > > 9133 for (i = hdev->num_vmdq_vport + 1; i < hdev->num_alloc_vport; i++) > ^^^^^^^^^^^^^^^^^^^^^^^^ > We're only checking the last part of the array. > > 9134 if (i != vf_idx && > 9135 ether_addr_equal(mac_addr, hdev->vport[i].vf_info.mac)) > 9136 return true; > 9137 > 9138 return false; > 9139 } > > Another thing that's not clear to me is how pci_num_vf() relates to > this. I suspect that it is the same as hdev->num_vmdq_vport, but I > can't be sure. > > regards, > dan carpenter > . The use for num_vmdq_vport is confusing. At the beginning, the HNS3 driver is planed to support VMDQ . Whereafter the hardware supports SR-IOV with better performance. So VMDQ feature is discarded, but the codes of vmdq is remained. For the value of num_vmdq_vport is always 0, so the vport id of VF is actually start from 1. That's why the driver still work. As the num_vmdq_vport is actually useless and confusing, I will send a patch to remove it soon. regards, Jian Shen

1 0

[PATCH RFC] net: sched: implement TCQ_F_CAN_BYPASS for lockless qdisc
by Yunsheng Lin 24 Mar '21

24 Mar '21

Currently pfifo_fast has both TCQ_F_CAN_BYPASS and TCQ_F_NOLOCK flag set, but queue discipline by-pass does not work for lockless qdisc because skb is always enqueued to qdisc even when the qdisc is empty, see __dev_xmit_skb(). This patch calles sch_direct_xmit() to transmit the skb directly to the driver for empty lockless qdisc too, which aviod enqueuing and dequeuing operation. qdisc->empty is set to false whenever a skb is enqueued, and is set to true when skb dequeuing return NULL, see pfifo_fast_dequeue(). Also, qdisc is scheduled at the end of qdisc_run_end() when q->empty is false to avoid packet stuck problem. The performance for ip_forward test increases about 10% with this patch. Signed-off-by: Yunsheng Lin <linyunsheng(a)huawei.com> --- include/net/sch_generic.h | 7 +++++-- net/core/dev.c | 11 +++++++++++ 2 files changed, 16 insertions(+), 2 deletions(-) diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h index 2d6eb60..6591356 100644 --- a/include/net/sch_generic.h +++ b/include/net/sch_generic.h @@ -161,7 +161,6 @@ static inline bool qdisc_run_begin(struct Qdisc *qdisc) if (qdisc->flags & TCQ_F_NOLOCK) { if (!spin_trylock(&qdisc->seqlock)) return false; - WRITE_ONCE(qdisc->empty, false); } else if (qdisc_is_running(qdisc)) { return false; } @@ -176,8 +175,12 @@ static inline bool qdisc_run_begin(struct Qdisc *qdisc) static inline void qdisc_run_end(struct Qdisc *qdisc) { write_seqcount_end(&qdisc->running); - if (qdisc->flags & TCQ_F_NOLOCK) + if (qdisc->flags & TCQ_F_NOLOCK) { spin_unlock(&qdisc->seqlock); + + if (unlikely(!READ_ONCE(qdisc->empty))) + __netif_schedule(qdisc); + } } static inline bool qdisc_may_bulk(const struct Qdisc *qdisc) diff --git a/net/core/dev.c b/net/core/dev.c index 2bfdd52..fa8504d 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -3791,7 +3791,18 @@ static inline int __dev_xmit_skb(struct sk_buff *skb, struct Qdisc *q, qdisc_calculate_pkt_len(skb, q); if (q->flags & TCQ_F_NOLOCK) { + if (q->flags & TCQ_F_CAN_BYPASS && READ_ONCE(q->empty) && qdisc_run_begin(q)) { + qdisc_bstats_cpu_update(q, skb); + + if (sch_direct_xmit(skb, q, dev, txq, NULL, true) && !READ_ONCE(q->empty)) + __qdisc_run(q); + + qdisc_run_end(q); + return NET_XMIT_SUCCESS; + } + rc = q->enqueue(skb, q, &to_free) & NET_XMIT_MASK; + WRITE_ONCE(q->empty, false); qdisc_run(q); if (unlikely(to_free)) -- 2.7.4

9 25

[PATCH] dma-mapping: make map_benchmark compile into module
by Tian Tao 24 Mar '21

24 Mar '21

under some scenarios, it is necessary to compile map_benchmark into module to test iommu, so this patch changed Kconfig and export_symbol to implement map_benchmark compiled into module. Signed-off-by: Tian Tao <tiantao6(a)hisilicon.com> --- kernel/dma/Kconfig | 2 +- kernel/kthread.c | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/kernel/dma/Kconfig b/kernel/dma/Kconfig index 77b4055..0468293 100644 --- a/kernel/dma/Kconfig +++ b/kernel/dma/Kconfig @@ -223,7 +223,7 @@ config DMA_API_DEBUG_SG If unsure, say N. config DMA_MAP_BENCHMARK - bool "Enable benchmarking of streaming DMA mapping" + tristate "Enable benchmarking of streaming DMA mapping" depends on DEBUG_FS help Provides /sys/kernel/debug/dma_map_benchmark that helps with testing diff --git a/kernel/kthread.c b/kernel/kthread.c index 1578973..fa4736f 100644 --- a/kernel/kthread.c +++ b/kernel/kthread.c @@ -455,6 +455,7 @@ void kthread_bind_mask(struct task_struct *p, const struct cpumask *mask) { __kthread_bind_mask(p, mask, TASK_UNINTERRUPTIBLE); } +EXPORT_SYMBOL(kthread_bind_mask); /** * kthread_bind - bind a just-created kthread to a cpu. -- 2.7.4

2 1

[PATCH net] net: sched: fix packet stuck problem for lockless qdisc
by Yunsheng Lin 23 Mar '21

23 Mar '21

Lockless qdisc has below concurrent problem: cpu0 cpu1 . . q->enqueue . . . qdisc_run_begin() . . . dequeue_skb() . . . sch_direct_xmit() . . . . q->enqueue . qdisc_run_begin() . return and do nothing . . qdisc_run_end() . cpu1 enqueue a skb without calling __qdisc_run() because cpu0 has not released the lock yet and spin_trylock() return false for cpu1 in qdisc_run_begin(), and cpu0 do not see the skb enqueued by cpu1 when calling dequeue_skb() because cpu1 may enqueue the skb after cpu0 calling dequeue_skb() and before cpu0 calling qdisc_run_end(). Lockless qdisc has another concurrent problem when tx_action is involved: cpu0(serving tx_action) cpu1 cpu2 . . . . q->enqueue . . qdisc_run_begin() . . dequeue_skb() . . . q->enqueue . . . . sch_direct_xmit() . . . qdisc_run_begin() . . return and do nothing . . . clear __QDISC_STATE_SCHED . . qdisc_run_begin() . . return and do nothing . . . . . . qdisc_run_begin() . This patch fixes the above data race by: 1. Set a flag after spin_trylock() return false. 2. Retry a spin_trylock() in case other CPU may not see the new flag after it releases the lock. 3. reschedule if the flag is set after the lock is released at the end of qdisc_run_end(). For tx_action case, the flags is also set when cpu1 is at the end if qdisc_run_begin(), so tx_action will be rescheduled again to dequeue the skb enqueued by cpu2. Also clear the flag before dequeuing in order to reduce the overhead of the above process, and aviod doing the heavy test_and_clear_bit() at the end of qdisc_run_end(). Fixes: 6b3ba9146fe6 ("net: sched: allow qdiscs to handle locking") Signed-off-by: Yunsheng Lin <linyunsheng(a)huawei.com> --- For those who has not been following the qdsic scheduling discussion, there is packet stuck problem for lockless qdisc, see [1], and I has done some cleanup and added some enhanced features too, see [2] [3]. While I was doing the optimization for lockless qdisc, it accurred to me that these optimization is useless if there is still basic bug in lockless qdisc, even the bug is not easily reproducible. So look through [1] again, I found that the data race for tx action mentioned by Michael, and thought deep about it and came up with this patch trying to fix it. So I am really appreciated some who still has the reproducer can try this patch and report back. 1. https://lore.kernel.org/netdev/d102074f-7489-e35a-98cf-e2cad7efd8a2@netroun… 2. https://patchwork.kernel.org/project/netdevbpf/patch/1615777818-13969-1-git… 3. https://patchwork.kernel.org/project/netdevbpf/patch/1615800610-34700-1-git… --- include/net/sch_generic.h | 23 ++++++++++++++++++++--- net/sched/sch_generic.c | 1 + 2 files changed, 21 insertions(+), 3 deletions(-) diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h index f7a6e14..4220eab 100644 --- a/include/net/sch_generic.h +++ b/include/net/sch_generic.h @@ -36,6 +36,7 @@ struct qdisc_rate_table { enum qdisc_state_t { __QDISC_STATE_SCHED, __QDISC_STATE_DEACTIVATED, + __QDISC_STATE_NEED_RESCHEDULE, }; struct qdisc_size_table { @@ -159,8 +160,17 @@ static inline bool qdisc_is_empty(const struct Qdisc *qdisc) static inline bool qdisc_run_begin(struct Qdisc *qdisc) { if (qdisc->flags & TCQ_F_NOLOCK) { - if (!spin_trylock(&qdisc->seqlock)) - return false; + if (!spin_trylock(&qdisc->seqlock)) { + set_bit(__QDISC_STATE_NEED_RESCHEDULE, + &qdisc->state); + + /* Retry again in case other CPU may not see the + * new flags after it releases the lock at the + * end of qdisc_run_end(). + */ + if (!spin_trylock(&qdisc->seqlock)) + return false; + } WRITE_ONCE(qdisc->empty, false); } else if (qdisc_is_running(qdisc)) { return false; @@ -176,8 +186,15 @@ static inline bool qdisc_run_begin(struct Qdisc *qdisc) static inline void qdisc_run_end(struct Qdisc *qdisc) { write_seqcount_end(&qdisc->running); - if (qdisc->flags & TCQ_F_NOLOCK) + if (qdisc->flags & TCQ_F_NOLOCK) { spin_unlock(&qdisc->seqlock); + + if (unlikely(test_bit(__QDISC_STATE_NEED_RESCHEDULE, + &qdisc->state) && + !test_bit(__QDISC_STATE_DEACTIVATED, + &qdisc->state))) + __netif_schedule(qdisc); + } } static inline bool qdisc_may_bulk(const struct Qdisc *qdisc) diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c index 44991ea..25d75d8 100644 --- a/net/sched/sch_generic.c +++ b/net/sched/sch_generic.c @@ -205,6 +205,7 @@ static struct sk_buff *dequeue_skb(struct Qdisc *q, bool *validate, const struct netdev_queue *txq = q->dev_queue; struct sk_buff *skb = NULL; + clear_bit(__QDISC_STATE_NEED_RESCHEDULE, &q->state); *packets = 1; if (unlikely(!skb_queue_empty(&q->gso_skb))) { spinlock_t *lock = NULL; -- 2.7.4

4 11

[PATCH net-next 0/8] net: hns: add some cleanups
by Huazhong Tan 23 Mar '21

23 Mar '21

This series includes some cleanups for the HNS ethernet driver. Huazhong Tan (6): net: hns: remove unused get_autoneg() net: hns: remove unused set_autoneg() net: hns: remove unused set_rx_ignore_pause_frames() net: hns: remove unused config_half_duplex() net: hns: remove unused NIC_LB_TEST_RX_PKG_ERR net: hns: remove unused HNS_LED_PC_REG Yonglong Liu (2): net: hns: remove unnecessary !! operation in hns_mac_config_sds_loopback_acpi() net: hns: remove redundant variable initialization drivers/net/ethernet/hisilicon/hns/hnae.h | 6 ------ drivers/net/ethernet/hisilicon/hns/hns_ae_adapt.c | 22 +--------------------- drivers/net/ethernet/hisilicon/hns/hns_dsaf_gmac.c | 18 ------------------ drivers/net/ethernet/hisilicon/hns/hns_dsaf_mac.c | 2 +- drivers/net/ethernet/hisilicon/hns/hns_dsaf_mac.h | 4 ---- drivers/net/ethernet/hisilicon/hns/hns_dsaf_main.c | 16 ++++++++-------- drivers/net/ethernet/hisilicon/hns/hns_dsaf_misc.c | 2 +- drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c | 4 ++-- .../net/ethernet/hisilicon/hns/hns_dsaf_xgmac.c | 18 +----------------- drivers/net/ethernet/hisilicon/hns/hns_enet.c | 4 ++-- drivers/net/ethernet/hisilicon/hns/hns_ethtool.c | 3 +-- drivers/net/ethernet/hisilicon/hns_mdio.c | 4 ++-- 12 files changed, 19 insertions(+), 84 deletions(-) -- 2.7.4

1 8

[PATCH] coresight: core: Fix typo in coresight-core.c
by Qi Liu 23 Mar '21

23 Mar '21

Fix up one typo: compoment->component. Fixes: 8e264c52e1da ("coresight: core: Allow the coresight core driver to be built as a module") Signed-off-by: Qi Liu <liuqi115(a)huawei.com> --- drivers/hwtracing/coresight/coresight-core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/hwtracing/coresight/coresight-core.c b/drivers/hwtracing/coresight/coresight-core.c index 0062c89..bc397a7 100644 --- a/drivers/hwtracing/coresight/coresight-core.c +++ b/drivers/hwtracing/coresight/coresight-core.c @@ -86,7 +86,7 @@ static int coresight_id_match(struct device *dev, void *data) i_csdev->type != CORESIGHT_DEV_TYPE_SOURCE) return 0; - /* Get the source ID for both compoment */ + /* Get the source ID for both component */ trace_id = source_ops(csdev)->trace_id(csdev); i_trace_id = source_ops(i_csdev)->trace_id(i_csdev); -- 2.8.1

4 4

Re: [PATCH v14 07/13] iommu/smmuv3: Implement cache_invalidate
by chenxiang (M) 23 Mar '21

23 Mar '21

Hi Eric, 在 2021/2/24 4:56, Eric Auger 写道: > Implement domain-selective, pasid selective and page-selective > IOTLB invalidations. > > Signed-off-by: Eric Auger <eric.auger(a)redhat.com> > > --- > > v13 -> v14: > - Add domain invalidation > - do global inval when asid is not provided with addr > granularity > > v7 -> v8: > - ASID based invalidation using iommu_inv_pasid_info > - check ARCHID/PASID flags in addr based invalidation > - use __arm_smmu_tlb_inv_context and __arm_smmu_tlb_inv_range_nosync > > v6 -> v7 > - check the uapi version > > v3 -> v4: > - adapt to changes in the uapi > - add support for leaf parameter > - do not use arm_smmu_tlb_inv_range_nosync or arm_smmu_tlb_inv_context > anymore > > v2 -> v3: > - replace __arm_smmu_tlb_sync by arm_smmu_cmdq_issue_sync > > v1 -> v2: > - properly pass the asid > --- > drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 74 +++++++++++++++++++++ > 1 file changed, 74 insertions(+) > > 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 4c19a1114de4..df3adc49111c 100644 > --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c > +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c > @@ -2949,6 +2949,79 @@ static void arm_smmu_detach_pasid_table(struct iommu_domain *domain) > mutex_unlock(&smmu_domain->init_mutex); > } > > +static int > +arm_smmu_cache_invalidate(struct iommu_domain *domain, struct device *dev, > + struct iommu_cache_invalidate_info *inv_info) > +{ > + struct arm_smmu_cmdq_ent cmd = {.opcode = CMDQ_OP_TLBI_NSNH_ALL}; > + struct arm_smmu_domain *smmu_domain = to_smmu_domain(domain); > + struct arm_smmu_device *smmu = smmu_domain->smmu; > + > + if (smmu_domain->stage != ARM_SMMU_DOMAIN_NESTED) > + return -EINVAL; > + > + if (!smmu) > + return -EINVAL; > + > + if (inv_info->version != IOMMU_CACHE_INVALIDATE_INFO_VERSION_1) > + return -EINVAL; > + > + if (inv_info->cache & IOMMU_CACHE_INV_TYPE_PASID || > + inv_info->cache & IOMMU_CACHE_INV_TYPE_DEV_IOTLB) { > + return -ENOENT; > + } > + > + if (!(inv_info->cache & IOMMU_CACHE_INV_TYPE_IOTLB)) > + return -EINVAL; > + > + /* IOTLB invalidation */ > + > + switch (inv_info->granularity) { > + case IOMMU_INV_GRANU_PASID: > + { > + struct iommu_inv_pasid_info *info = > + &inv_info->granu.pasid_info; > + > + if (info->flags & IOMMU_INV_ADDR_FLAGS_PASID) > + return -ENOENT; > + if (!(info->flags & IOMMU_INV_PASID_FLAGS_ARCHID)) > + return -EINVAL; > + > + __arm_smmu_tlb_inv_context(smmu_domain, info->archid); > + return 0; > + } > + case IOMMU_INV_GRANU_ADDR: > + { > + struct iommu_inv_addr_info *info = &inv_info->granu.addr_info; > + size_t size = info->nb_granules * info->granule_size; > + bool leaf = info->flags & IOMMU_INV_ADDR_FLAGS_LEAF; > + > + if (info->flags & IOMMU_INV_ADDR_FLAGS_PASID) > + return -ENOENT; > + > + if (!(info->flags & IOMMU_INV_ADDR_FLAGS_ARCHID)) > + break; > + > + arm_smmu_tlb_inv_range_domain(info->addr, size, > + info->granule_size, leaf, > + info->archid, smmu_domain); Is it possible to add a check before the function to make sure that info->granule_size can be recognized by SMMU? There is a scenario which will cause TLBI issue: RIL feature is enabled on guest but is disabled on host, and then on host it just invalidate 4K/2M/1G once a time, but from QEMU, info->nb_granules is always 1 and info->granule_size = size, if size is not equal to 4K or 2M or 1G (for example size = granule_size is 5M), it will only part of the size it wants to invalidate. I think maybe we can add a check here: if RIL is not enabled and also size is not the granule_size (4K/2M/1G) supported by SMMU hardware, can we just simply use the smallest granule_size supported by hardware all the time? > + > + arm_smmu_cmdq_issue_sync(smmu); > + return 0; > + } > + case IOMMU_INV_GRANU_DOMAIN: > + break; I check the qemu code (https://github.com/eauger/qemu/tree/v5.2.0-2stage-rfcv8) for opcode CMD_TLBI_NH_ALL or CMD_TLBI_NSNH_ALL from guest OS it calls smmu_inv_notifiers_all() to unamp all notifiers of all mr's, but it seems not set event.entry.granularity which i think it should set IOMMU_INV_GRAN_ADDR. BTW, for opcode CMD_TLBI_NH_ALL or CMD_TLBI_NSNH_ALL, it needs to unmap size = 0x1000000000000 on 48bit system (it may spend much time), maybe it is better to set it as IOMMU_INV_GRANU_DOMAIN, then in host OS, send TLBI_NH_ALL directly when IOMMU_INV_GRANU_DOMAIN. > + default: > + return -EINVAL; > + } > + > + /* Global S1 invalidation */ > + cmd.tlbi.vmid = smmu_domain->s2_cfg.vmid; > + arm_smmu_cmdq_issue_cmd(smmu, &cmd); > + arm_smmu_cmdq_issue_sync(smmu); > + return 0; > +} > + > static bool arm_smmu_dev_has_feature(struct device *dev, > enum iommu_dev_features feat) > { > @@ -3048,6 +3121,7 @@ static struct iommu_ops arm_smmu_ops = { > .put_resv_regions = generic_iommu_put_resv_regions, > .attach_pasid_table = arm_smmu_attach_pasid_table, > .detach_pasid_table = arm_smmu_detach_pasid_table, > + .cache_invalidate = arm_smmu_cache_invalidate, > .dev_has_feat = arm_smmu_dev_has_feature, > .dev_feat_enabled = arm_smmu_dev_feature_enabled, > .dev_enable_feat = arm_smmu_dev_enable_feature,

2 4

[PATCH V2 net-next 0/7] net: hns3: refactor and new features for flow director
by Huazhong Tan 23 Mar '21

23 Mar '21

This patchset refactor some functions and add some new features for flow director. patch 1~3: refactor large functions patch 4, 7: add traffic class and user-def field support for ethtool patch 5: refactor flow director configuration patch 6: clean up for hns3_del_all_fd_entries() change log: V1->V2: modifies patch 5 as Jakub suggested, keep configuring ethtool/tc flower rules synchronously while aRFS asynchronously. changes the usecnt of user-def rule checking in patch 7. removes previous patches 8 and 9 from this series, since there are issues that need further discussion. previous version: V1: https://patchwork.kernel.org/project/netdevbpf/cover/1615811031-55209-1-git… Jian Shen (7): net: hns3: refactor out hclge_add_fd_entry() net: hns3: refactor out hclge_fd_get_tuple() net: hns3: refactor for function hclge_fd_convert_tuple net: hns3: add support for traffic class tuple support for flow director by ethtool net: hns3: refactor flow director configuration net: hns3: refine for hns3_del_all_fd_entries() net: hns3: add support for user-def data of flow director drivers/net/ethernet/hisilicon/hns3/hnae3.h | 2 - drivers/net/ethernet/hisilicon/hns3/hns3_enet.c | 10 - .../net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h | 14 + .../ethernet/hisilicon/hns3/hns3pf/hclge_main.c | 1316 +++++++++++++------- .../ethernet/hisilicon/hns3/hns3pf/hclge_main.h | 58 + 5 files changed, 936 insertions(+), 464 deletions(-) -- 2.7.4

2 8

[PATCH] sched/fair: remove redundant test_idle_cores for non-smt
by Barry Song 22 Mar '21

22 Mar '21

update_idle_core() is only done for the case of sched_smt_present. but test_idle_cores() is done for all machines even those without smt. this could contribute to up 8%+ hackbench performance loss on a machine like kunpeng 920 which has no smt. this patch removes the redundant test_idle_cores() for non-smt machines. we run the below hackbench with different -g parameter from 2 to 14, for each different g, we run the command 10 times and get the average time: $ numactl -N 0 hackbench -p -T -l 20000 -g $1 hackbench will report the time which is needed to complete a certain number of messages transmissions between a certain number of tasks, for example: $ numactl -N 0 hackbench -p -T -l 20000 -g 10 Running in threaded mode with 10 groups using 40 file descriptors each (== 400 tasks) Each sender will pass 20000 messages of 100 bytes The below is the result of hackbench w/ and w/o this patch: g= 2 4 6 8 10 12 14 w/o: 1.8151 3.8499 5.5142 7.2491 9.0340 10.7345 12.0929 w/ : 1.8428 3.7436 5.4501 6.9522 8.2882 9.9535 11.3367 +4.1% +8.3% +7.3% +6.3% Signed-off-by: Barry Song <song.bao.hua(a)hisilicon.com> --- kernel/sched/fair.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 2e2ab1e..de42a32 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -6038,9 +6038,11 @@ static inline bool test_idle_cores(int cpu, bool def) { struct sched_domain_shared *sds; - sds = rcu_dereference(per_cpu(sd_llc_shared, cpu)); - if (sds) - return READ_ONCE(sds->has_idle_cores); + if (static_branch_likely(&sched_smt_present)) { + sds = rcu_dereference(per_cpu(sd_llc_shared, cpu)); + if (sds) + return READ_ONCE(sds->has_idle_cores); + } return def; } -- 1.8.3.1

6 5