- Kernel - mailweb.openeuler.org

[PATCH OLK-6.6 0/4] cgroup v2 cpuset supports exclusive cpumask.
by Chen Ridong 01 Mar '24

01 Mar '24

Chen Ridong (2): cgroup/cpuset: Add cpuset.cpus.exclusive.effective for v2 cgroup/cpuset: fix update_cpumask return value Waiman Long (2): cgroup/cpuset: Add cpuset.cpus.exclusive for v2 cgroup/cpuset: Introduce remote partition kernel/cgroup/cpuset.c | 1249 +++++++++++++++++++++++++++++----------- 1 file changed, 923 insertions(+), 326 deletions(-) -- 2.34.1

2 5

[PATCH OLK-6.6] PCI: Avoid potential out-of-bounds read in pci_dev_for_each_resource()
by Jialin Zhang 01 Mar '24

01 Mar '24

From: Andy Shevchenko <andriy.shevchenko(a)linux.intel.com> stable inclusion from stable-v6.6.14 commit 5b3e25efe16e06779a9a7c7610217c1b921ec179 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I93L7H CVE: CVE-2023-52466 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit 3171e46d677a668eed3086da78671f1e4f5b8405 ] Coverity complains that pointer in the pci_dev_for_each_resource() may be wrong, i.e., might be used for the out-of-bounds read. There is no actual issue right now because we have another check afterwards and the out-of-bounds read is not being performed. In any case it's better code with this fixed, hence the proposed change. As Jonas pointed out "It probably makes the code slightly less performant as res will now be checked for being not NULL (which will always be true), but I doubt it will be significant (or in any hot paths)." Fixes: 09cc90063240 ("PCI: Introduce pci_dev_for_each_resource()") Reported-by: Bjorn Helgaas <bhelgaas(a)google.com> Closes: https://lore.kernel.org/r/20230509182122.GA1259567@bhelgaas Suggested-by: Jonas Gorski <jonas.gorski(a)gmail.com> Link: https://lore.kernel.org/r/20231030114218.2752236-1-andriy.shevchenko@linux.… Signed-off-by: Andy Shevchenko <andriy.shevchenko(a)linux.intel.com> Signed-off-by: Bjorn Helgaas <bhelgaas(a)google.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Jialin Zhang <zhangjialin11(a)huawei.com> --- include/linux/pci.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/include/linux/pci.h b/include/linux/pci.h index 9aa545843712..6f96facca51a 100644 --- a/include/linux/pci.h +++ b/include/linux/pci.h @@ -2141,14 +2141,14 @@ int pci_iobar_pfn(struct pci_dev *pdev, int bar, struct vm_area_struct *vma); (pci_resource_end((dev), (bar)) ? \ resource_size(pci_resource_n((dev), (bar))) : 0) -#define __pci_dev_for_each_res0(dev, res, ...) \ - for (unsigned int __b = 0; \ - res = pci_resource_n(dev, __b), __b < PCI_NUM_RESOURCES; \ +#define __pci_dev_for_each_res0(dev, res, ...) \ + for (unsigned int __b = 0; \ + __b < PCI_NUM_RESOURCES && (res = pci_resource_n(dev, __b)); \ __b++) -#define __pci_dev_for_each_res1(dev, res, __b) \ - for (__b = 0; \ - res = pci_resource_n(dev, __b), __b < PCI_NUM_RESOURCES; \ +#define __pci_dev_for_each_res1(dev, res, __b) \ + for (__b = 0; \ + __b < PCI_NUM_RESOURCES && (res = pci_resource_n(dev, __b)); \ __b++) #define pci_dev_for_each_resource(dev, res, ...) \ -- 2.25.1

2 1

[OLK-6.6 v5 1/3] RDMA/hns: Support congestion control algorithm parameter configuration
by Chengchang Tang 01 Mar '24

01 Mar '24

driver inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I6J5O7 ---------------------------------------------------------------------- hns roce support 4 congestion control algorithms. Each algorihm involves multiple parameters. This patch add port sysfs directory for each algorithm, which allows users to modify the parameters of these algorithms. Signed-off-by: Chengchang Tang <tangchengchang(a)huawei.com> Signed-off-by: Juan Zhou <zhoujuan51(a)h-partners.com> --- drivers/infiniband/hw/hns/Makefile | 2 +- drivers/infiniband/hw/hns/hns_roce_device.h | 27 ++ drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 63 ++++ drivers/infiniband/hw/hns/hns_roce_hw_v2.h | 132 ++++++++ drivers/infiniband/hw/hns/hns_roce_main.c | 3 + drivers/infiniband/hw/hns/hns_roce_sysfs.c | 346 ++++++++++++++++++++ 6 files changed, 572 insertions(+), 1 deletion(-) create mode 100644 drivers/infiniband/hw/hns/hns_roce_sysfs.c diff --git a/drivers/infiniband/hw/hns/Makefile b/drivers/infiniband/hw/hns/Makefile index be1e1cdbcfa8..161615fda3ae 100644 --- a/drivers/infiniband/hw/hns/Makefile +++ b/drivers/infiniband/hw/hns/Makefile @@ -8,7 +8,7 @@ ccflags-y := -I $(srctree)/drivers/net/ethernet/hisilicon/hns3 hns-roce-objs := hns_roce_main.o hns_roce_cmd.o hns_roce_pd.o \ hns_roce_ah.o hns_roce_hem.o hns_roce_mr.o hns_roce_qp.o \ hns_roce_cq.o hns_roce_alloc.o hns_roce_db.o hns_roce_srq.o hns_roce_restrack.o \ - hns_roce_debugfs.o + hns_roce_debugfs.o hns_roce_sysfs.o ifdef CONFIG_INFINIBAND_HNS_HIP08 hns-roce-hw-v2-objs := hns_roce_hw_v2.o $(hns-roce-objs) diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h index d8887b5d9882..d6c33bdf56d3 100644 --- a/drivers/infiniband/hw/hns/hns_roce_device.h +++ b/drivers/infiniband/hw/hns/hns_roce_device.h @@ -710,6 +710,14 @@ struct hns_roce_eq_table { struct hns_roce_eq *eq; }; +enum hns_roce_scc_algo { + HNS_ROCE_SCC_ALGO_DCQCN = 0, + HNS_ROCE_SCC_ALGO_LDCP, + HNS_ROCE_SCC_ALGO_HC3, + HNS_ROCE_SCC_ALGO_DIP, + HNS_ROCE_SCC_ALGO_TOTAL, +}; + enum cong_type { CONG_TYPE_DCQCN, CONG_TYPE_LDCP, @@ -954,6 +962,20 @@ struct hns_roce_hw { u64 *stats, u32 port, int *hw_counters); const struct ib_device_ops *hns_roce_dev_ops; const struct ib_device_ops *hns_roce_dev_srq_ops; + int (*config_scc_param)(struct hns_roce_dev *hr_dev, + enum hns_roce_scc_algo algo); + int (*query_scc_param)(struct hns_roce_dev *hr_dev, + enum hns_roce_scc_algo alog); +}; + +#define HNS_ROCE_SCC_PARAM_SIZE 4 +struct hns_roce_scc_param { + __le32 param[HNS_ROCE_SCC_PARAM_SIZE]; + u32 lifespan; + unsigned long timestamp; + enum hns_roce_scc_algo algo_type; + struct delayed_work scc_cfg_dwork; + struct hns_roce_dev *hr_dev; }; struct hns_roce_dev { @@ -1019,6 +1041,7 @@ struct hns_roce_dev { u64 dwqe_page; struct hns_roce_dev_debugfs dbgfs; atomic64_t *dfx_cnt; + struct hns_roce_scc_param *scc_param; }; static inline struct hns_roce_dev *to_hr_dev(struct ib_device *ib_dev) @@ -1154,6 +1177,8 @@ static inline u8 get_tclass(const struct ib_global_route *grh) grh->traffic_class >> DSCP_SHIFT : grh->traffic_class; } +extern const struct attribute_group *hns_attr_port_groups[]; + void hns_roce_init_uar_table(struct hns_roce_dev *dev); int hns_roce_uar_alloc(struct hns_roce_dev *dev, struct hns_roce_uar *uar); @@ -1294,4 +1319,6 @@ struct hns_user_mmap_entry * hns_roce_user_mmap_entry_insert(struct ib_ucontext *ucontext, u64 address, size_t length, enum hns_roce_mmap_type mmap_type); +void hns_roce_register_sysfs(struct hns_roce_dev *hr_dev); +void hns_roce_unregister_sysfs(struct hns_roce_dev *hr_dev); #endif /* _HNS_ROCE_DEVICE_H */ diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c index 4d6a64df5a03..905040e9ccbe 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c @@ -6733,6 +6733,67 @@ static void hns_roce_v2_cleanup_eq_table(struct hns_roce_dev *hr_dev) kfree(eq_table->eq); } +static const enum hns_roce_opcode_type scc_opcode[] = { + HNS_ROCE_OPC_CFG_DCQCN_PARAM, + HNS_ROCE_OPC_CFG_LDCP_PARAM, + HNS_ROCE_OPC_CFG_HC3_PARAM, + HNS_ROCE_OPC_CFG_DIP_PARAM, +}; + +static int hns_roce_v2_config_scc_param(struct hns_roce_dev *hr_dev, + enum hns_roce_scc_algo algo) +{ + struct hns_roce_scc_param *scc_param; + struct hns_roce_cmq_desc desc; + int ret; + + if (algo >= HNS_ROCE_SCC_ALGO_TOTAL) { + ibdev_err_ratelimited(&hr_dev->ib_dev, "invalid SCC algo.\n"); + return -EINVAL; + } + + hns_roce_cmq_setup_basic_desc(&desc, scc_opcode[algo], false); + scc_param = &hr_dev->scc_param[algo]; + memcpy(&desc.data, scc_param, sizeof(scc_param->param)); + + ret = hns_roce_cmq_send(hr_dev, &desc, 1); + if (ret) + ibdev_err_ratelimited(&hr_dev->ib_dev, + "failed to configure scc param, opcode: 0x%x, ret = %d.\n", + le16_to_cpu(desc.opcode), ret); + return ret; +} + +static int hns_roce_v2_query_scc_param(struct hns_roce_dev *hr_dev, + enum hns_roce_scc_algo algo) +{ + struct hns_roce_scc_param *scc_param; + struct hns_roce_cmq_desc desc; + int ret; + + if (hr_dev->pci_dev->revision <= PCI_REVISION_ID_HIP08 || hr_dev->is_vf) + return -EOPNOTSUPP; + + if (algo >= HNS_ROCE_SCC_ALGO_TOTAL) { + ibdev_err_ratelimited(&hr_dev->ib_dev, "invalid SCC algo.\n"); + return -EINVAL; + } + + hns_roce_cmq_setup_basic_desc(&desc, scc_opcode[algo], true); + ret = hns_roce_cmq_send(hr_dev, &desc, 1); + if (ret) { + ibdev_err_ratelimited(&hr_dev->ib_dev, + "failed to query scc param, opcode: 0x%x, ret = %d.\n", + le16_to_cpu(desc.opcode), ret); + return ret; + } + + scc_param = &hr_dev->scc_param[algo]; + memcpy(scc_param, &desc.data, sizeof(scc_param->param)); + + return 0; +} + static const struct ib_device_ops hns_roce_v2_dev_ops = { .destroy_qp = hns_roce_v2_destroy_qp, .modify_cq = hns_roce_v2_modify_cq, @@ -6782,6 +6843,8 @@ static const struct hns_roce_hw hns_roce_hw_v2 = { .query_hw_counter = hns_roce_hw_v2_query_counter, .hns_roce_dev_ops = &hns_roce_v2_dev_ops, .hns_roce_dev_srq_ops = &hns_roce_v2_dev_srq_ops, + .config_scc_param = hns_roce_v2_config_scc_param, + .query_scc_param = hns_roce_v2_query_scc_param, }; static const struct pci_device_id hns_roce_hw_v2_pci_tbl[] = { diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h index 3fe2d9467612..741be211ed7e 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h @@ -195,6 +195,10 @@ enum { /* CMQ command */ enum hns_roce_opcode_type { HNS_QUERY_FW_VER = 0x0001, + HNS_ROCE_OPC_CFG_DCQCN_PARAM = 0x1A80, + HNS_ROCE_OPC_CFG_LDCP_PARAM = 0x1A81, + HNS_ROCE_OPC_CFG_HC3_PARAM = 0x1A82, + HNS_ROCE_OPC_CFG_DIP_PARAM = 0x1A83, HNS_ROCE_OPC_QUERY_HW_VER = 0x8000, HNS_ROCE_OPC_CFG_GLOBAL_PARAM = 0x8001, HNS_ROCE_OPC_ALLOC_PF_RES = 0x8004, @@ -1428,6 +1432,134 @@ struct hns_roce_wqe_atomic_seg { __le64 cmp_data; }; +#define HNS_ROCE_DCQCN_AI_OFS 0 +#define HNS_ROCE_DCQCN_AI_SZ sizeof(u16) +#define HNS_ROCE_DCQCN_AI_MAX ((u16)(~0U)) +#define HNS_ROCE_DCQCN_F_OFS (HNS_ROCE_DCQCN_AI_OFS + HNS_ROCE_DCQCN_AI_SZ) +#define HNS_ROCE_DCQCN_F_SZ sizeof(u8) +#define HNS_ROCE_DCQCN_F_MAX ((u8)(~0U)) +#define HNS_ROCE_DCQCN_TKP_OFS (HNS_ROCE_DCQCN_F_OFS + HNS_ROCE_DCQCN_F_SZ) +#define HNS_ROCE_DCQCN_TKP_SZ sizeof(u8) +#define HNS_ROCE_DCQCN_TKP_MAX 15 +#define HNS_ROCE_DCQCN_TMP_OFS (HNS_ROCE_DCQCN_TKP_OFS + HNS_ROCE_DCQCN_TKP_SZ) +#define HNS_ROCE_DCQCN_TMP_SZ sizeof(u16) +#define HNS_ROCE_DCQCN_TMP_MAX 15 +#define HNS_ROCE_DCQCN_ALP_OFS (HNS_ROCE_DCQCN_TMP_OFS + HNS_ROCE_DCQCN_TMP_SZ) +#define HNS_ROCE_DCQCN_ALP_SZ sizeof(u16) +#define HNS_ROCE_DCQCN_ALP_MAX ((u16)(~0U)) +#define HNS_ROCE_DCQCN_MAX_SPEED_OFS (HNS_ROCE_DCQCN_ALP_OFS + \ + HNS_ROCE_DCQCN_ALP_SZ) +#define HNS_ROCE_DCQCN_MAX_SPEED_SZ sizeof(u32) +#define HNS_ROCE_DCQCN_MAX_SPEED_MAX ((u32)(~0U)) +#define HNS_ROCE_DCQCN_G_OFS (HNS_ROCE_DCQCN_MAX_SPEED_OFS + \ + HNS_ROCE_DCQCN_MAX_SPEED_SZ) +#define HNS_ROCE_DCQCN_G_SZ sizeof(u8) +#define HNS_ROCE_DCQCN_G_MAX 15 +#define HNS_ROCE_DCQCN_AL_OFS (HNS_ROCE_DCQCN_G_OFS + HNS_ROCE_DCQCN_G_SZ) +#define HNS_ROCE_DCQCN_AL_SZ sizeof(u8) +#define HNS_ROCE_DCQCN_AL_MAX ((u8)(~0U)) +#define HNS_ROCE_DCQCN_CNP_TIME_OFS (HNS_ROCE_DCQCN_AL_OFS + \ + HNS_ROCE_DCQCN_AL_SZ) +#define HNS_ROCE_DCQCN_CNP_TIME_SZ sizeof(u8) +#define HNS_ROCE_DCQCN_CNP_TIME_MAX ((u8)(~0U)) +#define HNS_ROCE_DCQCN_ASHIFT_OFS (HNS_ROCE_DCQCN_CNP_TIME_OFS + \ + HNS_ROCE_DCQCN_CNP_TIME_SZ) +#define HNS_ROCE_DCQCN_ASHIFT_SZ sizeof(u8) +#define HNS_ROCE_DCQCN_ASHIFT_MAX 15 +#define HNS_ROCE_DCQCN_LIFESPAN_OFS (HNS_ROCE_DCQCN_ASHIFT_OFS + \ + HNS_ROCE_DCQCN_ASHIFT_SZ) +#define HNS_ROCE_DCQCN_LIFESPAN_SZ sizeof(u32) +#define HNS_ROCE_DCQCN_LIFESPAN_MAX 1000 + +#define HNS_ROCE_LDCP_CWD0_OFS 0 +#define HNS_ROCE_LDCP_CWD0_SZ sizeof(u32) +#define HNS_ROCE_LDCP_CWD0_MAX ((u32)(~0U)) +#define HNS_ROCE_LDCP_ALPHA_OFS (HNS_ROCE_LDCP_CWD0_OFS + HNS_ROCE_LDCP_CWD0_SZ) +#define HNS_ROCE_LDCP_ALPHA_SZ sizeof(u8) +#define HNS_ROCE_LDCP_ALPHA_MAX ((u8)(~0U)) +#define HNS_ROCE_LDCP_GAMMA_OFS (HNS_ROCE_LDCP_ALPHA_OFS + \ + HNS_ROCE_LDCP_ALPHA_SZ) +#define HNS_ROCE_LDCP_GAMMA_SZ sizeof(u8) +#define HNS_ROCE_LDCP_GAMMA_MAX ((u8)(~0U)) +#define HNS_ROCE_LDCP_BETA_OFS (HNS_ROCE_LDCP_GAMMA_OFS + \ + HNS_ROCE_LDCP_GAMMA_SZ) +#define HNS_ROCE_LDCP_BETA_SZ sizeof(u8) +#define HNS_ROCE_LDCP_BETA_MAX ((u8)(~0U)) +#define HNS_ROCE_LDCP_ETA_OFS (HNS_ROCE_LDCP_BETA_OFS + HNS_ROCE_LDCP_BETA_SZ) +#define HNS_ROCE_LDCP_ETA_SZ sizeof(u8) +#define HNS_ROCE_LDCP_ETA_MAX ((u8)(~0U)) +#define HNS_ROCE_LDCP_LIFESPAN_OFS (4 * sizeof(u32)) +#define HNS_ROCE_LDCP_LIFESPAN_SZ sizeof(u32) +#define HNS_ROCE_LDCP_LIFESPAN_MAX 1000 + +#define HNS_ROCE_HC3_INITIAL_WINDOW_OFS 0 +#define HNS_ROCE_HC3_INITIAL_WINDOW_SZ sizeof(u32) +#define HNS_ROCE_HC3_INITIAL_WINDOW_MAX ((u32)(~0U)) +#define HNS_ROCE_HC3_BANDWIDTH_OFS (HNS_ROCE_HC3_INITIAL_WINDOW_OFS + \ + HNS_ROCE_HC3_INITIAL_WINDOW_SZ) +#define HNS_ROCE_HC3_BANDWIDTH_SZ sizeof(u32) +#define HNS_ROCE_HC3_BANDWIDTH_MAX ((u32)(~0U)) +#define HNS_ROCE_HC3_QLEN_SHIFT_OFS (HNS_ROCE_HC3_BANDWIDTH_OFS + \ + HNS_ROCE_HC3_BANDWIDTH_SZ) +#define HNS_ROCE_HC3_QLEN_SHIFT_SZ sizeof(u8) +#define HNS_ROCE_HC3_QLEN_SHIFT_MAX ((u8)(~0U)) +#define HNS_ROCE_HC3_PORT_USAGE_SHIFT_OFS (HNS_ROCE_HC3_QLEN_SHIFT_OFS + \ + HNS_ROCE_HC3_QLEN_SHIFT_SZ) +#define HNS_ROCE_HC3_PORT_USAGE_SHIFT_SZ sizeof(u8) +#define HNS_ROCE_HC3_PORT_USAGE_SHIFT_MAX ((u8)(~0U)) +#define HNS_ROCE_HC3_OVER_PERIOD_OFS (HNS_ROCE_HC3_PORT_USAGE_SHIFT_OFS + \ + HNS_ROCE_HC3_PORT_USAGE_SHIFT_SZ) +#define HNS_ROCE_HC3_OVER_PERIOD_SZ sizeof(u8) +#define HNS_ROCE_HC3_OVER_PERIOD_MAX ((u8)(~0U)) +#define HNS_ROCE_HC3_MAX_STAGE_OFS (HNS_ROCE_HC3_OVER_PERIOD_OFS + \ + HNS_ROCE_HC3_OVER_PERIOD_SZ) +#define HNS_ROCE_HC3_MAX_STAGE_SZ sizeof(u8) +#define HNS_ROCE_HC3_MAX_STAGE_MAX ((u8)(~0U)) +#define HNS_ROCE_HC3_GAMMA_SHIFT_OFS (HNS_ROCE_HC3_MAX_STAGE_OFS + \ + HNS_ROCE_HC3_MAX_STAGE_SZ) +#define HNS_ROCE_HC3_GAMMA_SHIFT_SZ sizeof(u8) +#define HNS_ROCE_HC3_GAMMA_SHIFT_MAX 15 +#define HNS_ROCE_HC3_LIFESPAN_OFS (4 * sizeof(u32)) +#define HNS_ROCE_HC3_LIFESPAN_SZ sizeof(u32) +#define HNS_ROCE_HC3_LIFESPAN_MAX 1000 + +#define HNS_ROCE_DIP_AI_OFS 0 +#define HNS_ROCE_DIP_AI_SZ sizeof(u16) +#define HNS_ROCE_DIP_AI_MAX ((u16)(~0U)) +#define HNS_ROCE_DIP_F_OFS (HNS_ROCE_DIP_AI_OFS + HNS_ROCE_DIP_AI_SZ) +#define HNS_ROCE_DIP_F_SZ sizeof(u8) +#define HNS_ROCE_DIP_F_MAX ((u8)(~0U)) +#define HNS_ROCE_DIP_TKP_OFS (HNS_ROCE_DIP_F_OFS + HNS_ROCE_DIP_F_SZ) +#define HNS_ROCE_DIP_TKP_SZ sizeof(u8) +#define HNS_ROCE_DIP_TKP_MAX 15 +#define HNS_ROCE_DIP_TMP_OFS (HNS_ROCE_DIP_TKP_OFS + HNS_ROCE_DIP_TKP_SZ) +#define HNS_ROCE_DIP_TMP_SZ sizeof(u16) +#define HNS_ROCE_DIP_TMP_MAX 15 +#define HNS_ROCE_DIP_ALP_OFS (HNS_ROCE_DIP_TMP_OFS + HNS_ROCE_DIP_TMP_SZ) +#define HNS_ROCE_DIP_ALP_SZ sizeof(u16) +#define HNS_ROCE_DIP_ALP_MAX ((u16)(~0U)) +#define HNS_ROCE_DIP_MAX_SPEED_OFS (HNS_ROCE_DIP_ALP_OFS + HNS_ROCE_DIP_ALP_SZ) +#define HNS_ROCE_DIP_MAX_SPEED_SZ sizeof(u32) +#define HNS_ROCE_DIP_MAX_SPEED_MAX ((u32)(~0U)) +#define HNS_ROCE_DIP_G_OFS (HNS_ROCE_DIP_MAX_SPEED_OFS + \ + HNS_ROCE_DIP_MAX_SPEED_SZ) +#define HNS_ROCE_DIP_G_SZ sizeof(u8) +#define HNS_ROCE_DIP_G_MAX 15 +#define HNS_ROCE_DIP_AL_OFS (HNS_ROCE_DIP_G_OFS + HNS_ROCE_DIP_G_SZ) +#define HNS_ROCE_DIP_AL_SZ sizeof(u8) +#define HNS_ROCE_DIP_AL_MAX ((u8)(~0U)) +#define HNS_ROCE_DIP_CNP_TIME_OFS (HNS_ROCE_DIP_AL_OFS + HNS_ROCE_DIP_AL_SZ) +#define HNS_ROCE_DIP_CNP_TIME_SZ sizeof(u8) +#define HNS_ROCE_DIP_CNP_TIME_MAX ((u8)(~0U)) +#define HNS_ROCE_DIP_ASHIFT_OFS (HNS_ROCE_DIP_CNP_TIME_OFS + \ + HNS_ROCE_DIP_CNP_TIME_SZ) +#define HNS_ROCE_DIP_ASHIFT_SZ sizeof(u8) +#define HNS_ROCE_DIP_ASHIFT_MAX 15 +#define HNS_ROCE_DIP_LIFESPAN_OFS (HNS_ROCE_DIP_ASHIFT_OFS + \ + HNS_ROCE_DIP_ASHIFT_SZ) +#define HNS_ROCE_DIP_LIFESPAN_SZ sizeof(u32) +#define HNS_ROCE_DIP_LIFESPAN_MAX 1000 + struct hns_roce_sccc_clr { __le32 qpn; __le32 rsv[5]; diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c index b55fe6911f9f..18a346c5a6eb 100644 --- a/drivers/infiniband/hw/hns/hns_roce_main.c +++ b/drivers/infiniband/hw/hns/hns_roce_main.c @@ -633,6 +633,7 @@ static const struct ib_device_ops hns_roce_dev_ops = { .query_pkey = hns_roce_query_pkey, .query_port = hns_roce_query_port, .reg_user_mr = hns_roce_reg_user_mr, + .port_groups = hns_attr_port_groups, INIT_RDMA_OBJ_SIZE(ib_ah, hns_roce_ah, ibah), INIT_RDMA_OBJ_SIZE(ib_cq, hns_roce_cq, ib_cq), @@ -1106,6 +1107,7 @@ int hns_roce_init(struct hns_roce_dev *hr_dev) if (ret) goto error_failed_register_device; + hns_roce_register_sysfs(hr_dev); hns_roce_register_debugfs(hr_dev); return 0; @@ -1140,6 +1142,7 @@ int hns_roce_init(struct hns_roce_dev *hr_dev) void hns_roce_exit(struct hns_roce_dev *hr_dev) { + hns_roce_unregister_sysfs(hr_dev); hns_roce_unregister_debugfs(hr_dev); hns_roce_unregister_device(hr_dev); diff --git a/drivers/infiniband/hw/hns/hns_roce_sysfs.c b/drivers/infiniband/hw/hns/hns_roce_sysfs.c new file mode 100644 index 000000000000..63a50180c4f2 --- /dev/null +++ b/drivers/infiniband/hw/hns/hns_roce_sysfs.c @@ -0,0 +1,346 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2023 Hisilicon Limited. + */ + +#include <rdma/ib_sysfs.h> + +#include "hnae3.h" +#include "hns_roce_device.h" +#include "hns_roce_hw_v2.h" + +static void scc_param_config_work(struct work_struct *work) +{ + struct hns_roce_scc_param *scc_param = container_of(work, + struct hns_roce_scc_param, scc_cfg_dwork.work); + struct hns_roce_dev *hr_dev = scc_param->hr_dev; + + hr_dev->hw->config_scc_param(hr_dev, scc_param->algo_type); +} + +static void get_default_scc_param(struct hns_roce_dev *hr_dev) +{ + int ret; + int i; + + for (i = 0; i < HNS_ROCE_SCC_ALGO_TOTAL; i++) { + hr_dev->scc_param[i].timestamp = jiffies; + ret = hr_dev->hw->query_scc_param(hr_dev, i); + if (ret && ret != -EOPNOTSUPP) + ibdev_warn_ratelimited(&hr_dev->ib_dev, + "failed to get default parameters of scc algo %d, ret = %d.\n", + i, ret); + } +} + +static int alloc_scc_param(struct hns_roce_dev *hr_dev) +{ + struct hns_roce_scc_param *scc_param; + int i; + + scc_param = kvcalloc(HNS_ROCE_SCC_ALGO_TOTAL, sizeof(*scc_param), + GFP_KERNEL); + if (!scc_param) + return -ENOMEM; + + for (i = 0; i < HNS_ROCE_SCC_ALGO_TOTAL; i++) { + scc_param[i].algo_type = i; + scc_param[i].hr_dev = hr_dev; + INIT_DELAYED_WORK(&scc_param[i].scc_cfg_dwork, + scc_param_config_work); + } + + hr_dev->scc_param = scc_param; + + get_default_scc_param(hr_dev); + + return 0; +} + +struct hns_port_cc_attr { + struct ib_port_attribute port_attr; + enum hns_roce_scc_algo algo_type; + u32 offset; + u32 size; + u32 max; + u32 min; +}; + +static int scc_attr_check(struct hns_roce_dev *hr_dev, + struct hns_port_cc_attr *scc_attr, u32 port_num) +{ + if (port_num > hr_dev->caps.num_ports) + return -ENODEV; + + if (WARN_ON(scc_attr->size > sizeof(u32))) + return -EINVAL; + + if (WARN_ON(scc_attr->algo_type >= HNS_ROCE_SCC_ALGO_TOTAL)) + return -EINVAL; + + return 0; +} + +static ssize_t scc_attr_show(struct ib_device *ibdev, u32 port_num, + struct ib_port_attribute *attr, char *buf) +{ + struct hns_port_cc_attr *scc_attr = + container_of(attr, struct hns_port_cc_attr, port_attr); + struct hns_roce_dev *hr_dev = to_hr_dev(ibdev); + struct hns_roce_scc_param *scc_param; + __le32 val = 0; + int ret; + + ret = scc_attr_check(hr_dev, scc_attr, port_num); + if (ret) + return ret; + + scc_param = &hr_dev->scc_param[scc_attr->algo_type]; + + memcpy(&val, (void *)scc_param + scc_attr->offset, scc_attr->size); + + return sysfs_emit(buf, "%u\n", le32_to_cpu(val)); +} + +static ssize_t scc_attr_store(struct ib_device *ibdev, u32 port_num, + struct ib_port_attribute *attr, const char *buf, + size_t count) +{ + struct hns_port_cc_attr *scc_attr = + container_of(attr, struct hns_port_cc_attr, port_attr); + struct hns_roce_dev *hr_dev = to_hr_dev(ibdev); + struct hns_roce_scc_param *scc_param; + unsigned long lifespan_jiffies; + unsigned long exp_time; + __le32 attr_val; + u32 val; + int ret; + + ret = scc_attr_check(hr_dev, scc_attr, port_num); + if (ret) + return ret; + + if (kstrtou32(buf, 0, &val)) + return -EINVAL; + + if (val > scc_attr->max || val < scc_attr->min) + return -EINVAL; + + attr_val = cpu_to_le32(val); + scc_param = &hr_dev->scc_param[scc_attr->algo_type]; + memcpy((void *)scc_param + scc_attr->offset, &attr_val, + scc_attr->size); + + /* lifespan is only used for driver */ + if (scc_attr->offset >= offsetof(typeof(*scc_param), lifespan)) + return count; + + lifespan_jiffies = msecs_to_jiffies(scc_param->lifespan); + exp_time = scc_param->timestamp + lifespan_jiffies; + + if (time_is_before_eq_jiffies(exp_time)) { + scc_param->timestamp = jiffies; + queue_delayed_work(hr_dev->irq_workq, &scc_param->scc_cfg_dwork, + lifespan_jiffies); + } + + return count; +} + +static umode_t scc_attr_is_visible(struct kobject *kobj, + struct attribute *attr, int i) +{ + struct ib_port_attribute *port_attr = + container_of(attr, struct ib_port_attribute, attr); + struct hns_port_cc_attr *scc_attr = + container_of(port_attr, struct hns_port_cc_attr, port_attr); + u32 port_num; + struct ib_device *ibdev = ib_port_sysfs_get_ibdev_kobj(kobj, &port_num); + struct hns_roce_dev *hr_dev = to_hr_dev(ibdev); + + if (hr_dev->is_vf || + !(hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_QP_FLOW_CTRL)) + return 0; + + if (!(hr_dev->caps.cong_type & (1 << scc_attr->algo_type))) + return 0; + + return 0644; +} + +#define __HNS_SCC_ATTR(_name, _type, _offset, _size, _min, _max) { \ + .port_attr = __ATTR(_name, 0644, scc_attr_show, scc_attr_store), \ + .algo_type = _type, \ + .offset = _offset, \ + .size = _size, \ + .min = _min, \ + .max = _max, \ +} + +#define HNS_PORT_DCQCN_CC_ATTR_RW(_name, NAME) \ + struct hns_port_cc_attr hns_roce_port_attr_dcqcn_##_name = \ + __HNS_SCC_ATTR(_name, HNS_ROCE_SCC_ALGO_DCQCN, \ + HNS_ROCE_DCQCN_##NAME##_OFS, \ + HNS_ROCE_DCQCN_##NAME##_SZ, \ + 0, HNS_ROCE_DCQCN_##NAME##_MAX) + +HNS_PORT_DCQCN_CC_ATTR_RW(ai, AI); +HNS_PORT_DCQCN_CC_ATTR_RW(f, F); +HNS_PORT_DCQCN_CC_ATTR_RW(tkp, TKP); +HNS_PORT_DCQCN_CC_ATTR_RW(tmp, TMP); +HNS_PORT_DCQCN_CC_ATTR_RW(alp, ALP); +HNS_PORT_DCQCN_CC_ATTR_RW(max_speed, MAX_SPEED); +HNS_PORT_DCQCN_CC_ATTR_RW(g, G); +HNS_PORT_DCQCN_CC_ATTR_RW(al, AL); +HNS_PORT_DCQCN_CC_ATTR_RW(cnp_time, CNP_TIME); +HNS_PORT_DCQCN_CC_ATTR_RW(ashift, ASHIFT); +HNS_PORT_DCQCN_CC_ATTR_RW(lifespan, LIFESPAN); + +static struct attribute *dcqcn_param_attrs[] = { + &hns_roce_port_attr_dcqcn_ai.port_attr.attr, + &hns_roce_port_attr_dcqcn_f.port_attr.attr, + &hns_roce_port_attr_dcqcn_tkp.port_attr.attr, + &hns_roce_port_attr_dcqcn_tmp.port_attr.attr, + &hns_roce_port_attr_dcqcn_alp.port_attr.attr, + &hns_roce_port_attr_dcqcn_max_speed.port_attr.attr, + &hns_roce_port_attr_dcqcn_g.port_attr.attr, + &hns_roce_port_attr_dcqcn_al.port_attr.attr, + &hns_roce_port_attr_dcqcn_cnp_time.port_attr.attr, + &hns_roce_port_attr_dcqcn_ashift.port_attr.attr, + &hns_roce_port_attr_dcqcn_lifespan.port_attr.attr, + NULL, +}; + +static const struct attribute_group dcqcn_cc_param_group = { + .name = "dcqcn_cc_param", + .attrs = dcqcn_param_attrs, + .is_visible = scc_attr_is_visible, +}; + +#define HNS_PORT_LDCP_CC_ATTR_RW(_name, NAME) \ + struct hns_port_cc_attr hns_roce_port_attr_ldcp_##_name = \ + __HNS_SCC_ATTR(_name, HNS_ROCE_SCC_ALGO_LDCP, \ + HNS_ROCE_LDCP_##NAME##_OFS, \ + HNS_ROCE_LDCP_##NAME##_SZ, \ + 0, HNS_ROCE_LDCP_##NAME##_MAX) + +HNS_PORT_LDCP_CC_ATTR_RW(cwd0, CWD0); +HNS_PORT_LDCP_CC_ATTR_RW(alpha, ALPHA); +HNS_PORT_LDCP_CC_ATTR_RW(gamma, GAMMA); +HNS_PORT_LDCP_CC_ATTR_RW(beta, BETA); +HNS_PORT_LDCP_CC_ATTR_RW(eta, ETA); +HNS_PORT_LDCP_CC_ATTR_RW(lifespan, LIFESPAN); + +static struct attribute *ldcp_param_attrs[] = { + &hns_roce_port_attr_ldcp_cwd0.port_attr.attr, + &hns_roce_port_attr_ldcp_alpha.port_attr.attr, + &hns_roce_port_attr_ldcp_gamma.port_attr.attr, + &hns_roce_port_attr_ldcp_beta.port_attr.attr, + &hns_roce_port_attr_ldcp_eta.port_attr.attr, + &hns_roce_port_attr_ldcp_lifespan.port_attr.attr, + NULL, +}; + +static const struct attribute_group ldcp_cc_param_group = { + .name = "ldcp_cc_param", + .attrs = ldcp_param_attrs, + .is_visible = scc_attr_is_visible, +}; + +#define HNS_PORT_HC3_CC_ATTR_RW(_name, NAME) \ + struct hns_port_cc_attr hns_roce_port_attr_hc3_##_name = \ + __HNS_SCC_ATTR(_name, HNS_ROCE_SCC_ALGO_HC3, \ + HNS_ROCE_HC3_##NAME##_OFS, \ + HNS_ROCE_HC3_##NAME##_SZ, \ + 0, HNS_ROCE_HC3_##NAME##_MAX) + +HNS_PORT_HC3_CC_ATTR_RW(initial_window, INITIAL_WINDOW); +HNS_PORT_HC3_CC_ATTR_RW(bandwidth, BANDWIDTH); +HNS_PORT_HC3_CC_ATTR_RW(qlen_shift, QLEN_SHIFT); +HNS_PORT_HC3_CC_ATTR_RW(port_usage_shift, PORT_USAGE_SHIFT); +HNS_PORT_HC3_CC_ATTR_RW(over_period, OVER_PERIOD); +HNS_PORT_HC3_CC_ATTR_RW(max_stage, MAX_STAGE); +HNS_PORT_HC3_CC_ATTR_RW(gamma_shift, GAMMA_SHIFT); +HNS_PORT_HC3_CC_ATTR_RW(lifespan, LIFESPAN); + +static struct attribute *hc3_param_attrs[] = { + &hns_roce_port_attr_hc3_initial_window.port_attr.attr, + &hns_roce_port_attr_hc3_bandwidth.port_attr.attr, + &hns_roce_port_attr_hc3_qlen_shift.port_attr.attr, + &hns_roce_port_attr_hc3_port_usage_shift.port_attr.attr, + &hns_roce_port_attr_hc3_over_period.port_attr.attr, + &hns_roce_port_attr_hc3_max_stage.port_attr.attr, + &hns_roce_port_attr_hc3_gamma_shift.port_attr.attr, + &hns_roce_port_attr_hc3_lifespan.port_attr.attr, + NULL, +}; + +static const struct attribute_group hc3_cc_param_group = { + .name = "hc3_cc_param", + .attrs = hc3_param_attrs, + .is_visible = scc_attr_is_visible, +}; + +#define HNS_PORT_DIP_CC_ATTR_RW(_name, NAME) \ + struct hns_port_cc_attr hns_roce_port_attr_dip_##_name = \ + __HNS_SCC_ATTR(_name, HNS_ROCE_SCC_ALGO_DIP, \ + HNS_ROCE_DIP_##NAME##_OFS, \ + HNS_ROCE_DIP_##NAME##_SZ, \ + 0, HNS_ROCE_DIP_##NAME##_MAX) + +HNS_PORT_DIP_CC_ATTR_RW(ai, AI); +HNS_PORT_DIP_CC_ATTR_RW(f, F); +HNS_PORT_DIP_CC_ATTR_RW(tkp, TKP); +HNS_PORT_DIP_CC_ATTR_RW(tmp, TMP); +HNS_PORT_DIP_CC_ATTR_RW(alp, ALP); +HNS_PORT_DIP_CC_ATTR_RW(max_speed, MAX_SPEED); +HNS_PORT_DIP_CC_ATTR_RW(g, G); +HNS_PORT_DIP_CC_ATTR_RW(al, AL); +HNS_PORT_DIP_CC_ATTR_RW(cnp_time, CNP_TIME); +HNS_PORT_DIP_CC_ATTR_RW(ashift, ASHIFT); +HNS_PORT_DIP_CC_ATTR_RW(lifespan, LIFESPAN); + +static struct attribute *dip_param_attrs[] = { + &hns_roce_port_attr_dip_ai.port_attr.attr, + &hns_roce_port_attr_dip_f.port_attr.attr, + &hns_roce_port_attr_dip_tkp.port_attr.attr, + &hns_roce_port_attr_dip_tmp.port_attr.attr, + &hns_roce_port_attr_dip_alp.port_attr.attr, + &hns_roce_port_attr_dip_max_speed.port_attr.attr, + &hns_roce_port_attr_dip_g.port_attr.attr, + &hns_roce_port_attr_dip_al.port_attr.attr, + &hns_roce_port_attr_dip_cnp_time.port_attr.attr, + &hns_roce_port_attr_dip_ashift.port_attr.attr, + &hns_roce_port_attr_dip_lifespan.port_attr.attr, + NULL, +}; + +static const struct attribute_group dip_cc_param_group = { + .name = "dip_cc_param", + .attrs = dip_param_attrs, + .is_visible = scc_attr_is_visible, +}; + +const struct attribute_group *hns_attr_port_groups[] = { + &dcqcn_cc_param_group, + &ldcp_cc_param_group, + &hc3_cc_param_group, + &dip_cc_param_group, + NULL, +}; + +void hns_roce_register_sysfs(struct hns_roce_dev *hr_dev) +{ + int ret; + + ret = alloc_scc_param(hr_dev); + if (ret) + dev_err(hr_dev->dev, "alloc scc param failed, ret = %d!\n", + ret); +} + +void hns_roce_unregister_sysfs(struct hns_roce_dev *hr_dev) +{ + if (hr_dev->scc_param) + kvfree(hr_dev->scc_param); +} -- 2.30.0

1 0

[openeuler:OLK-6.6 1683/3700] mm/memcontrol.c:3147:6: sparse: sparse: symbol 'hisi_oom_recover' was not declared. Should it be static?
by kernel test robot 01 Mar '24

01 Mar '24

tree: https://gitee.com/openeuler/kernel.git OLK-6.6 head: 7f1397339b784f46c104e2e1257952c885c84128 commit: b498d9f1bacd40d583c0970cab02ad522127a7e7 [1683/3700] arm64/ascend: Add new enable_oom_killer interface for oom contrl config: arm64-randconfig-r111-20240227 (https://download.01.org/0day-ci/archive/20240301/202403011613.m2fhHDHp-lkp@…) compiler: aarch64-linux-gcc (GCC) 13.2.0 reproduce: (https://download.01.org/0day-ci/archive/20240301/202403011613.m2fhHDHp-lkp@…) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp(a)intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202403011613.m2fhHDHp-lkp@intel.com/ sparse warnings: (new ones prefixed by >>) >> mm/memcontrol.c:3147:6: sparse: sparse: symbol 'hisi_oom_recover' was not declared. Should it be static? mm/memcontrol.c:4228:21: sparse: sparse: incompatible types in comparison expression (different address spaces): mm/memcontrol.c:4228:21: sparse: struct mem_cgroup_threshold_ary [noderef] __rcu * mm/memcontrol.c:4228:21: sparse: struct mem_cgroup_threshold_ary * mm/memcontrol.c:4230:21: sparse: sparse: incompatible types in comparison expression (different address spaces): mm/memcontrol.c:4230:21: sparse: struct mem_cgroup_threshold_ary [noderef] __rcu * mm/memcontrol.c:4230:21: sparse: struct mem_cgroup_threshold_ary * mm/memcontrol.c:4386:9: sparse: sparse: incompatible types in comparison expression (different address spaces): mm/memcontrol.c:4386:9: sparse: struct mem_cgroup_threshold_ary [noderef] __rcu * mm/memcontrol.c:4386:9: sparse: struct mem_cgroup_threshold_ary * mm/memcontrol.c:4480:9: sparse: sparse: incompatible types in comparison expression (different address spaces): mm/memcontrol.c:4480:9: sparse: struct mem_cgroup_threshold_ary [noderef] __rcu * mm/memcontrol.c:4480:9: sparse: struct mem_cgroup_threshold_ary * mm/memcontrol.c:6198:23: sparse: sparse: incompatible types in comparison expression (different address spaces): mm/memcontrol.c:6198:23: sparse: struct task_struct [noderef] __rcu * mm/memcontrol.c:6198:23: sparse: struct task_struct * mm/memcontrol.c: note: in included file (through include/linux/rculist.h, include/linux/pid.h, include/linux/sched.h, ...): include/linux/list.h:83:21: sparse: sparse: self-comparison always evaluates to true mm/memcontrol.c: note: in included file: include/linux/memcontrol.h:751:9: sparse: sparse: context imbalance in 'folio_lruvec_lock' - wrong count at exit include/linux/memcontrol.h:751:9: sparse: sparse: context imbalance in 'folio_lruvec_lock_irq' - wrong count at exit include/linux/memcontrol.h:751:9: sparse: sparse: context imbalance in 'folio_lruvec_lock_irqsave' - wrong count at exit mm/memcontrol.c:2111:6: sparse: sparse: context imbalance in 'folio_memcg_lock' - wrong count at exit mm/memcontrol.c:2158:17: sparse: sparse: context imbalance in '__folio_memcg_unlock' - unexpected unlock mm/memcontrol.c: note: in included file (through include/linux/rculist.h, include/linux/pid.h, include/linux/sched.h, ...): include/linux/list.h:83:21: sparse: sparse: self-comparison always evaluates to true include/linux/list.h:83:21: sparse: sparse: self-comparison always evaluates to true mm/memcontrol.c: note: in included file (through include/linux/rculist.h, include/linux/pid.h, include/linux/sched.h, ...): include/linux/rcupdate.h:780:9: sparse: sparse: context imbalance in 'mem_cgroup_count_precharge_pte_range' - unexpected unlock include/linux/rcupdate.h:780:9: sparse: sparse: context imbalance in 'mem_cgroup_move_charge_pte_range' - unexpected unlock vim +/hisi_oom_recover +3147 mm/memcontrol.c 3145 3146 #ifdef CONFIG_ASCEND_OOM > 3147 void hisi_oom_recover(struct obj_cgroup *objcg) 3148 { 3149 struct mem_cgroup *memcg; 3150 3151 memcg = get_mem_cgroup_from_objcg(objcg); 3152 if (!mem_cgroup_is_root(memcg)) 3153 memcg_oom_recover(memcg); 3154 css_put(&memcg->css); 3155 } 3156 #else 3157 static inline void hisi_oom_recover(struct obj_cgroup *objcg) { } 3158 #endif 3159 -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0

[PATCH OLK-5.10 0/2] btrfs: mainline backport
by Yifan Qiao 01 Mar '24

01 Mar '24

Filipe Manana (2): btrfs: do not start and wait for delalloc on snapshot roots on transaction commit btrfs: fix deadlock when cloning inline extents and using qgroups fs/btrfs/ctree.h | 2 +- fs/btrfs/inode.c | 4 ++-- fs/btrfs/ioctl.c | 2 +- fs/btrfs/qgroup.c | 2 +- fs/btrfs/send.c | 4 ++-- fs/btrfs/transaction.c | 49 ++++++------------------------------------ 6 files changed, 13 insertions(+), 50 deletions(-) -- 2.39.2

2 3

[PATCH OLK-5.10 0/2] btrfs: mainline backport
by Yifan Qiao 01 Mar '24

01 Mar '24

Filipe Manana (2): btrfs: do not start and wait for delalloc on snapshot roots on transaction commit btrfs: fix deadlock when cloning inline extents and using qgroups fs/btrfs/ctree.h | 2 +- fs/btrfs/inode.c | 4 ++-- fs/btrfs/ioctl.c | 2 +- fs/btrfs/qgroup.c | 2 +- fs/btrfs/send.c | 4 ++-- fs/btrfs/transaction.c | 49 ++++++------------------------------------ 6 files changed, 13 insertions(+), 50 deletions(-) -- 2.39.2

2 3

[PATCH OLK-5.10 1/2] btrfs: do not start and wait for delalloc on snapshot roots on transaction commit
by Yifan Qiao 01 Mar '24

01 Mar '24

From: Filipe Manana <fdmanana(a)suse.com> mainline inclusion from mainline-v5.11-rc1 commit 88090ad36a64af1eb5b78d26b2ccd07eedae80b5 category: bugfix bugzilla: N/A Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… ------------------------------------------------------ We do not need anymore to start writeback for delalloc of roots that are being snapshotted and wait for it to complete. This was done in commit 609e804d771f59 ("Btrfs: fix file corruption after snapshotting due to mix of buffered/DIO writes") to fix a type of file corruption where files in a snapshot end up having their i_size updated in a non-ordered way, leaving implicit file holes, when buffered IO writes that increase a file's size are followed by direct IO writes that also increase the file's size. This is not needed anymore because we now have a more generic mechanism to prevent a non-ordered i_size update since commit 9ddc959e802bf7 ("btrfs: use the file extent tree infrastructure"), which addresses this scenario involving snapshots as well. Reviewed-by: Josef Bacik <josef(a)toxicpanda.com> Signed-off-by: Filipe Manana <fdmanana(a)suse.com> Reviewed-by: David Sterba <dsterba(a)suse.com> Signed-off-by: David Sterba <dsterba(a)suse.com> --- fs/btrfs/transaction.c | 49 ++++++------------------------------------ 1 file changed, 6 insertions(+), 43 deletions(-) diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index abd67f984fbc..2f89ac8fd5eb 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -2015,10 +2015,8 @@ static void btrfs_cleanup_pending_block_groups(struct btrfs_trans_handle *trans) } } -static inline int btrfs_start_delalloc_flush(struct btrfs_trans_handle *trans) +static inline int btrfs_start_delalloc_flush(struct btrfs_fs_info *fs_info) { - struct btrfs_fs_info *fs_info = trans->fs_info; - /* * We use writeback_inodes_sb here because if we used * btrfs_start_delalloc_roots we would deadlock with fs freeze. @@ -2028,50 +2026,15 @@ static inline int btrfs_start_delalloc_flush(struct btrfs_trans_handle *trans) * from already being in a transaction and our join_transaction doesn't * have to re-take the fs freeze lock. */ - if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) { + if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) writeback_inodes_sb(fs_info->sb, WB_REASON_SYNC); - } else { - struct btrfs_pending_snapshot *pending; - struct list_head *head = &trans->transaction->pending_snapshots; - - /* - * Flush dellaloc for any root that is going to be snapshotted. - * This is done to avoid a corrupted version of files, in the - * snapshots, that had both buffered and direct IO writes (even - * if they were done sequentially) due to an unordered update of - * the inode's size on disk. - */ - list_for_each_entry(pending, head, list) { - int ret; - - ret = btrfs_start_delalloc_snapshot(pending->root); - if (ret) - return ret; - } - } return 0; } -static inline void btrfs_wait_delalloc_flush(struct btrfs_trans_handle *trans) +static inline void btrfs_wait_delalloc_flush(struct btrfs_fs_info *fs_info) { - struct btrfs_fs_info *fs_info = trans->fs_info; - - if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) { + if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1); - } else { - struct btrfs_pending_snapshot *pending; - struct list_head *head = &trans->transaction->pending_snapshots; - - /* - * Wait for any dellaloc that we started previously for the roots - * that are going to be snapshotted. This is to avoid a corrupted - * version of files in the snapshots that had both buffered and - * direct IO writes (even if they were done sequentially). - */ - list_for_each_entry(pending, head, list) - btrfs_wait_ordered_extents(pending->root, - U64_MAX, 0, U64_MAX); - } } int btrfs_commit_transaction(struct btrfs_trans_handle *trans) @@ -2209,7 +2172,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans) extwriter_counter_dec(cur_trans, trans->type); - ret = btrfs_start_delalloc_flush(trans); + ret = btrfs_start_delalloc_flush(fs_info); if (ret) goto cleanup_transaction; @@ -2225,7 +2188,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans) if (ret) goto cleanup_transaction; - btrfs_wait_delalloc_flush(trans); + btrfs_wait_delalloc_flush(fs_info); /* * Wait for all ordered extents started by a fast fsync that joined this -- 2.39.2

2 2

[PATCH OLK-5.10 1/2] btrfs: do not start and wait for delalloc on snapshot roots on transaction commit
by q00831684 01 Mar '24

01 Mar '24

From: Filipe Manana <fdmanana(a)suse.com> mainline inclusion from mainline-v5.11-rc1 commit 88090ad36a64af1eb5b78d26b2ccd07eedae80b5 category: bugfix bugzilla: N/A Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… ------------------------------------------------------ We do not need anymore to start writeback for delalloc of roots that are being snapshotted and wait for it to complete. This was done in commit 609e804d771f59 ("Btrfs: fix file corruption after snapshotting due to mix of buffered/DIO writes") to fix a type of file corruption where files in a snapshot end up having their i_size updated in a non-ordered way, leaving implicit file holes, when buffered IO writes that increase a file's size are followed by direct IO writes that also increase the file's size. This is not needed anymore because we now have a more generic mechanism to prevent a non-ordered i_size update since commit 9ddc959e802bf7 ("btrfs: use the file extent tree infrastructure"), which addresses this scenario involving snapshots as well. Reviewed-by: Josef Bacik <josef(a)toxicpanda.com> Signed-off-by: Filipe Manana <fdmanana(a)suse.com> Reviewed-by: David Sterba <dsterba(a)suse.com> Signed-off-by: David Sterba <dsterba(a)suse.com> --- fs/btrfs/transaction.c | 49 ++++++------------------------------------ 1 file changed, 6 insertions(+), 43 deletions(-) diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index abd67f984fbc..2f89ac8fd5eb 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -2015,10 +2015,8 @@ static void btrfs_cleanup_pending_block_groups(struct btrfs_trans_handle *trans) } } -static inline int btrfs_start_delalloc_flush(struct btrfs_trans_handle *trans) +static inline int btrfs_start_delalloc_flush(struct btrfs_fs_info *fs_info) { - struct btrfs_fs_info *fs_info = trans->fs_info; - /* * We use writeback_inodes_sb here because if we used * btrfs_start_delalloc_roots we would deadlock with fs freeze. @@ -2028,50 +2026,15 @@ static inline int btrfs_start_delalloc_flush(struct btrfs_trans_handle *trans) * from already being in a transaction and our join_transaction doesn't * have to re-take the fs freeze lock. */ - if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) { + if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) writeback_inodes_sb(fs_info->sb, WB_REASON_SYNC); - } else { - struct btrfs_pending_snapshot *pending; - struct list_head *head = &trans->transaction->pending_snapshots; - - /* - * Flush dellaloc for any root that is going to be snapshotted. - * This is done to avoid a corrupted version of files, in the - * snapshots, that had both buffered and direct IO writes (even - * if they were done sequentially) due to an unordered update of - * the inode's size on disk. - */ - list_for_each_entry(pending, head, list) { - int ret; - - ret = btrfs_start_delalloc_snapshot(pending->root); - if (ret) - return ret; - } - } return 0; } -static inline void btrfs_wait_delalloc_flush(struct btrfs_trans_handle *trans) +static inline void btrfs_wait_delalloc_flush(struct btrfs_fs_info *fs_info) { - struct btrfs_fs_info *fs_info = trans->fs_info; - - if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) { + if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1); - } else { - struct btrfs_pending_snapshot *pending; - struct list_head *head = &trans->transaction->pending_snapshots; - - /* - * Wait for any dellaloc that we started previously for the roots - * that are going to be snapshotted. This is to avoid a corrupted - * version of files in the snapshots that had both buffered and - * direct IO writes (even if they were done sequentially). - */ - list_for_each_entry(pending, head, list) - btrfs_wait_ordered_extents(pending->root, - U64_MAX, 0, U64_MAX); - } } int btrfs_commit_transaction(struct btrfs_trans_handle *trans) @@ -2209,7 +2172,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans) extwriter_counter_dec(cur_trans, trans->type); - ret = btrfs_start_delalloc_flush(trans); + ret = btrfs_start_delalloc_flush(fs_info); if (ret) goto cleanup_transaction; @@ -2225,7 +2188,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans) if (ret) goto cleanup_transaction; - btrfs_wait_delalloc_flush(trans); + btrfs_wait_delalloc_flush(fs_info); /* * Wait for all ordered extents started by a fast fsync that joined this -- 2.39.2

2 2

[PATCH OLK-5.10 1/2] btrfs: do not start and wait for delalloc on snapshot roots on transaction commit
by q00831684 01 Mar '24

01 Mar '24

From: Filipe Manana <fdmanana(a)suse.com> mainline inclusion from mainline-v5.11-rc1 commit 88090ad36a64af1eb5b78d26b2ccd07eedae80b5 category: bugfix bugzilla: N/A Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… ------------------------------------------------------ We do not need anymore to start writeback for delalloc of roots that are being snapshotted and wait for it to complete. This was done in commit 609e804d771f59 ("Btrfs: fix file corruption after snapshotting due to mix of buffered/DIO writes") to fix a type of file corruption where files in a snapshot end up having their i_size updated in a non-ordered way, leaving implicit file holes, when buffered IO writes that increase a file's size are followed by direct IO writes that also increase the file's size. This is not needed anymore because we now have a more generic mechanism to prevent a non-ordered i_size update since commit 9ddc959e802bf7 ("btrfs: use the file extent tree infrastructure"), which addresses this scenario involving snapshots as well. Reviewed-by: Josef Bacik <josef(a)toxicpanda.com> Signed-off-by: Filipe Manana <fdmanana(a)suse.com> Reviewed-by: David Sterba <dsterba(a)suse.com> Signed-off-by: David Sterba <dsterba(a)suse.com> --- fs/btrfs/transaction.c | 49 ++++++------------------------------------ 1 file changed, 6 insertions(+), 43 deletions(-) diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index abd67f984fbc..2f89ac8fd5eb 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -2015,10 +2015,8 @@ static void btrfs_cleanup_pending_block_groups(struct btrfs_trans_handle *trans) } } -static inline int btrfs_start_delalloc_flush(struct btrfs_trans_handle *trans) +static inline int btrfs_start_delalloc_flush(struct btrfs_fs_info *fs_info) { - struct btrfs_fs_info *fs_info = trans->fs_info; - /* * We use writeback_inodes_sb here because if we used * btrfs_start_delalloc_roots we would deadlock with fs freeze. @@ -2028,50 +2026,15 @@ static inline int btrfs_start_delalloc_flush(struct btrfs_trans_handle *trans) * from already being in a transaction and our join_transaction doesn't * have to re-take the fs freeze lock. */ - if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) { + if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) writeback_inodes_sb(fs_info->sb, WB_REASON_SYNC); - } else { - struct btrfs_pending_snapshot *pending; - struct list_head *head = &trans->transaction->pending_snapshots; - - /* - * Flush dellaloc for any root that is going to be snapshotted. - * This is done to avoid a corrupted version of files, in the - * snapshots, that had both buffered and direct IO writes (even - * if they were done sequentially) due to an unordered update of - * the inode's size on disk. - */ - list_for_each_entry(pending, head, list) { - int ret; - - ret = btrfs_start_delalloc_snapshot(pending->root); - if (ret) - return ret; - } - } return 0; } -static inline void btrfs_wait_delalloc_flush(struct btrfs_trans_handle *trans) +static inline void btrfs_wait_delalloc_flush(struct btrfs_fs_info *fs_info) { - struct btrfs_fs_info *fs_info = trans->fs_info; - - if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) { + if (btrfs_test_opt(fs_info, FLUSHONCOMMIT)) btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1); - } else { - struct btrfs_pending_snapshot *pending; - struct list_head *head = &trans->transaction->pending_snapshots; - - /* - * Wait for any dellaloc that we started previously for the roots - * that are going to be snapshotted. This is to avoid a corrupted - * version of files in the snapshots that had both buffered and - * direct IO writes (even if they were done sequentially). - */ - list_for_each_entry(pending, head, list) - btrfs_wait_ordered_extents(pending->root, - U64_MAX, 0, U64_MAX); - } } int btrfs_commit_transaction(struct btrfs_trans_handle *trans) @@ -2209,7 +2172,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans) extwriter_counter_dec(cur_trans, trans->type); - ret = btrfs_start_delalloc_flush(trans); + ret = btrfs_start_delalloc_flush(fs_info); if (ret) goto cleanup_transaction; @@ -2225,7 +2188,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans) if (ret) goto cleanup_transaction; - btrfs_wait_delalloc_flush(trans); + btrfs_wait_delalloc_flush(fs_info); /* * Wait for all ordered extents started by a fast fsync that joined this -- 2.39.2

2 3

[openeuler:OLK-5.10 7311/30000] kernel/workqueue.o: warning: objtool: pwq_dec_nr_in_flight()+0x441: unreachable instruction
by kernel test robot 01 Mar '24

01 Mar '24

tree: https://gitee.com/openeuler/kernel.git OLK-5.10 head: ebba55154f6b47b476dbcd26c12f31ead8bcd4c6 commit: 003a421b04ba85abaf24a92736dc47f76fda0803 [7311/30000] disable OPTIMIZE_INLINING by default config: x86_64-buildonly-randconfig-001-20240226 (https://download.01.org/0day-ci/archive/20240301/202403011347.BSVD2waD-lkp@…) compiler: clang version 17.0.6 (https://github.com/llvm/llvm-project 6009708b4367171ccdbf4b5905cb6a803753fe18) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240301/202403011347.BSVD2waD-lkp@…) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp(a)intel.com> | Closes: https://lore.kernel.org/oe-kbuild-all/202403011347.BSVD2waD-lkp@intel.com/ All warnings (new ones prefixed by >>): >> kernel/workqueue.o: warning: objtool: pwq_dec_nr_in_flight()+0x441: unreachable instruction -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0

2024

2023

2022

2021

2020

2019

Kernel