- Kernel - mailweb.openeuler.org

[PATCH OLK-5.10 v4 0/8] Introduce soft domain v4
by Zhang Qiao 30 Jun '25

30 Jun '25

v3: - Rebase OLK-5.10 - Add a bugfix for atomic sleep Zhang Qiao (8): sched: topology: Build soft domain for LLC sched: Attach task group to soft domain sched: fair: Select idle cpu in soft domain sched: fair: Disable numa migrate for soft domian task sched: Add cmdline sched_soft_domain switch for soft domain feature config: Configurate CONFIG_SCHED_SOFT_DOMAIN sched: Balance newly task to soft domain sched: Fix might sleep in atomic section issue arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + include/linux/sched/topology.h | 21 + init/Kconfig | 12 + kernel/sched/Makefile | 1 + kernel/sched/core.c | 85 +++++ kernel/sched/fair.c | 156 +++++++- kernel/sched/features.h | 4 + kernel/sched/sched.h | 43 +++ kernel/sched/soft_domain.c | 505 +++++++++++++++++++++++++ 10 files changed, 827 insertions(+), 2 deletions(-) create mode 100644 kernel/sched/soft_domain.c -- 2.18.0.huawei.25

2 9

[PATCH 00/13] arm64/nmi: Support for FEAT_NMI
by Yang Yingliang 30 Jun '25

30 Jun '25

Jinjie Ruan (1): irqchip/gic-v3: Fix hard LOCKUP caused by NMI being masked Lorenzo Pieralisi (1): irqchip/gic-v3: Implement FEAT_GICv3_NMI support Mark Brown (11): arm64/booting: Document boot requirements for FEAT_NMI arm64/sysreg: Add definitions for immediate versions of MSR ALLINT arm64/asm: Introduce assembly macros for managing ALLINT arm64/hyp-stub: Enable access to ALLINT arm64/cpufeature: Detect PE support for FEAT_NMI KVM: arm64: Hide FEAT_NMI from guests arm64/nmi: Manage masking for superpriority interrupts along with DAIF arm64/entry: Don't call preempt_schedule_irq() with NMIs masked arm64/irq: Document handling of FEAT_NMI in irqflags.h arm64/nmi: Add handling of superpriority interrupts as NMIs arm64/nmi: Add Kconfig for NMI Documentation/arm64/booting.rst | 6 ++ arch/arm64/Kconfig | 17 ++++ arch/arm64/include/asm/assembler.h | 18 +++++ arch/arm64/include/asm/cpucaps.h | 2 + arch/arm64/include/asm/cpufeature.h | 29 +++++++ arch/arm64/include/asm/daifflags.h | 20 +++++ arch/arm64/include/asm/irqflags.h | 10 +++ arch/arm64/include/asm/ptrace.h | 5 +- arch/arm64/include/asm/sysreg.h | 26 ++++++ arch/arm64/include/uapi/asm/ptrace.h | 1 + arch/arm64/kernel/cpufeature.c | 64 ++++++++++++++- arch/arm64/kernel/head.S | 13 +++ arch/arm64/kernel/process.c | 9 +++ arch/arm64/kvm/hyp/switch.c | 7 ++ arch/arm64/kvm/sys_regs.c | 2 + drivers/irqchip/irq-gic-v3.c | 116 ++++++++++++++++++++++++--- include/linux/irqchip/arm-gic-v3.h | 4 + 17 files changed, 335 insertions(+), 14 deletions(-) -- 2.25.1

1 13

[PATCH OLK-6.6] tracing: probes: Fix a possible race in trace_probe_log APIs
by Tengda Wu 30 Jun '25

30 Jun '25

From: "Masami Hiramatsu (Google)" <mhiramat(a)kernel.org> stable inclusion from stable-6.6.94 commit 70417bada46dcfb91b59757bad343f753d9d9345 category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/ICIOVW Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit fd837de3c9cb1a162c69bc1fb1f438467fe7f2f5 ] Since the shared trace_probe_log variable can be accessed and modified via probe event create operation of kprobe_events, uprobe_events, and dynamic_events, it should be protected. In the dynamic_events, all operations are serialized by `dyn_event_ops_mutex`. But kprobe_events and uprobe_events interfaces are not serialized. To solve this issue, introduces dyn_event_create(), which runs create() operation under the mutex, for kprobe_events and uprobe_events. This also uses lockdep to check the mutex is held when using trace_probe_log* APIs. Link: https://lore.kernel.org/all/174684868120.551552.3068655787654268804.stgit@d… Reported-by: Paul Cacheux <paulcacheux(a)gmail.com> Closes: https://lore.kernel.org/all/20250510074456.805a16872b591e2971a4d221@kernel.… Fixes: ab105a4fb894 ("tracing: Use tracing error_log with probe events") Signed-off-by: Masami Hiramatsu (Google) <mhiramat(a)kernel.org> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Tengda Wu <wutengda2(a)huawei.com> --- kernel/trace/trace_dynevent.c | 16 +++++++++++++++- kernel/trace/trace_dynevent.h | 1 + kernel/trace/trace_kprobe.c | 2 +- kernel/trace/trace_probe.c | 9 +++++++++ kernel/trace/trace_uprobe.c | 2 +- 5 files changed, 27 insertions(+), 3 deletions(-) diff --git a/kernel/trace/trace_dynevent.c b/kernel/trace/trace_dynevent.c index 4376887e0d8a..c9b0533407ed 100644 --- a/kernel/trace/trace_dynevent.c +++ b/kernel/trace/trace_dynevent.c @@ -16,7 +16,7 @@ #include "trace_output.h" /* for trace_event_sem */ #include "trace_dynevent.h" -static DEFINE_MUTEX(dyn_event_ops_mutex); +DEFINE_MUTEX(dyn_event_ops_mutex); static LIST_HEAD(dyn_event_ops_list); bool trace_event_dyn_try_get_ref(struct trace_event_call *dyn_call) @@ -125,6 +125,20 @@ int dyn_event_release(const char *raw_command, struct dyn_event_operations *type return ret; } +/* + * Locked version of event creation. The event creation must be protected by + * dyn_event_ops_mutex because of protecting trace_probe_log. + */ +int dyn_event_create(const char *raw_command, struct dyn_event_operations *type) +{ + int ret; + + mutex_lock(&dyn_event_ops_mutex); + ret = type->create(raw_command); + mutex_unlock(&dyn_event_ops_mutex); + return ret; +} + static int create_dyn_event(const char *raw_command) { struct dyn_event_operations *ops; diff --git a/kernel/trace/trace_dynevent.h b/kernel/trace/trace_dynevent.h index 936477a111d3..beee3f8d7544 100644 --- a/kernel/trace/trace_dynevent.h +++ b/kernel/trace/trace_dynevent.h @@ -100,6 +100,7 @@ void *dyn_event_seq_next(struct seq_file *m, void *v, loff_t *pos); void dyn_event_seq_stop(struct seq_file *m, void *v); int dyn_events_release_all(struct dyn_event_operations *type); int dyn_event_release(const char *raw_command, struct dyn_event_operations *type); +int dyn_event_create(const char *raw_command, struct dyn_event_operations *type); /* * for_each_dyn_event - iterate over the dyn_event list diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index 5d056ec79119..73ebe6700eb9 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -1002,7 +1002,7 @@ static int create_or_delete_trace_kprobe(const char *raw_command) if (raw_command[0] == '-') return dyn_event_release(raw_command, &trace_kprobe_ops); - ret = trace_kprobe_create(raw_command); + ret = dyn_event_create(raw_command, &trace_kprobe_ops); return ret == -ECANCELED ? -EINVAL : ret; } diff --git a/kernel/trace/trace_probe.c b/kernel/trace/trace_probe.c index 5390e5306913..3a7fbe1da879 100644 --- a/kernel/trace/trace_probe.c +++ b/kernel/trace/trace_probe.c @@ -154,9 +154,12 @@ static const struct fetch_type *find_fetch_type(const char *type, unsigned long } static struct trace_probe_log trace_probe_log; +extern struct mutex dyn_event_ops_mutex; void trace_probe_log_init(const char *subsystem, int argc, const char **argv) { + lockdep_assert_held(&dyn_event_ops_mutex); + trace_probe_log.subsystem = subsystem; trace_probe_log.argc = argc; trace_probe_log.argv = argv; @@ -165,11 +168,15 @@ void trace_probe_log_init(const char *subsystem, int argc, const char **argv) void trace_probe_log_clear(void) { + lockdep_assert_held(&dyn_event_ops_mutex); + memset(&trace_probe_log, 0, sizeof(trace_probe_log)); } void trace_probe_log_set_index(int index) { + lockdep_assert_held(&dyn_event_ops_mutex); + trace_probe_log.index = index; } @@ -178,6 +185,8 @@ void __trace_probe_log_err(int offset, int err_type) char *command, *p; int i, len = 0, pos = 0; + lockdep_assert_held(&dyn_event_ops_mutex); + if (!trace_probe_log.argv) return; diff --git a/kernel/trace/trace_uprobe.c b/kernel/trace/trace_uprobe.c index 9110a759e882..89b467c6560a 100644 --- a/kernel/trace/trace_uprobe.c +++ b/kernel/trace/trace_uprobe.c @@ -730,7 +730,7 @@ static int create_or_delete_trace_uprobe(const char *raw_command) if (raw_command[0] == '-') return dyn_event_release(raw_command, &trace_uprobe_ops); - ret = trace_uprobe_create(raw_command); + ret = dyn_event_create(raw_command, &trace_uprobe_ops); return ret == -ECANCELED ? -EINVAL : ret; } -- 2.34.1

2 1

[openeuler:openEuler-1.0-LTS 1673/1673] block/blk-mq-sched.o: warning: objtool: blk_mq_sched_insert_request()+0x13f: unreachable instruction
by kernel test robot 30 Jun '25

30 Jun '25

Hi Ming, First bad commit (maybe != root cause): tree: https://gitee.com/openeuler/kernel.git openEuler-1.0-LTS head: a9d08fabefcfd9cb8ebe45f8b79a321c0f349d40 commit: 756a9563633e27b424df25eba9c7975967892ccd [1673/1673] blk-mq: insert passthrough request into hctx->dispatch directly config: x86_64-buildonly-randconfig-002-20250628 (https://download.01.org/0day-ci/archive/20250630/202506301025.8TcXBis8-lkp@…) compiler: clang version 20.1.7 (https://github.com/llvm/llvm-project 6146a88f60492b520a36f8f8f3231e15f3cc6082) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250630/202506301025.8TcXBis8-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/202506301025.8TcXBis8-lkp@intel.com/ All warnings (new ones prefixed by >>): >> block/blk-mq-sched.o: warning: objtool: blk_mq_sched_insert_request()+0x13f: unreachable instruction -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0

[PATCH OLK-5.10] net/oenetcls: Support cluster sched
by Yue Haibing 30 Jun '25

30 Jun '25

hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/ICBFCS CVE: NA -------------------------------- Configure NIC irqs as cluster or custom for support cluster sched. Signed-off-by: Yue Haibing <yuehaibing(a)huawei.com> --- net/oenetcls/oenetcls.h | 19 ++- net/oenetcls/oenetcls_main.c | 241 +++++++++++++++++++++++---------- net/oenetcls/oenetcls_ntuple.c | 84 ++++++++++-- 3 files changed, 260 insertions(+), 84 deletions(-) diff --git a/net/oenetcls/oenetcls.h b/net/oenetcls/oenetcls.h index 72aeedcaf0a0..b2c4a3c61956 100644 --- a/net/oenetcls/oenetcls.h +++ b/net/oenetcls/oenetcls.h @@ -7,7 +7,6 @@ #define OECLS_MAX_NETDEV_NUM 8 #define OECLS_MAX_RXQ_NUM_PER_DEV 256 -#define OECLS_MAX_NUMA_NUM 16 #define OECLS_MAX_CPU_NUM 1024 #define OECLS_TIMEOUT (5 * HZ) @@ -27,8 +26,20 @@ struct oecls_netdev_info { int old_filter_state; }; +struct oecls_rxq { + int rxq_id; + int status; +}; + +struct oecls_numa_clusterinfo { + int cluster_id; + int cur_freeidx; + struct oecls_rxq rxqs[OECLS_MAX_RXQ_NUM_PER_DEV]; +}; + struct oecls_numa_bound_dev_info { DECLARE_BITMAP(bitmap_rxq, OECLS_MAX_RXQ_NUM_PER_DEV); + struct oecls_numa_clusterinfo *cluster_info; }; struct oecls_numa_info { @@ -66,6 +77,12 @@ struct oecls_sk_rule { int nid; }; +struct oecls_sk_entry { + struct hlist_node node; + void *sk; + u32 sk_rule_hash; +}; + struct oecls_dev_flow { unsigned short cpu; unsigned short filter; diff --git a/net/oenetcls/oenetcls_main.c b/net/oenetcls/oenetcls_main.c index 8cf5d02bb5b9..1c40357ac58b 100644 --- a/net/oenetcls/oenetcls_main.c +++ b/net/oenetcls/oenetcls_main.c @@ -12,7 +12,8 @@ int oecls_netdev_num; static struct oecls_netdev_info oecls_netdev_info_table[OECLS_MAX_NETDEV_NUM]; int oecls_numa_num; -static struct oecls_numa_info oecls_numa_info_table[OECLS_MAX_NUMA_NUM]; +static int oecls_cluster_cpu_num, oecls_cluster_per_numa; +static struct oecls_numa_info *oecls_numa_info_table; int debug; module_param(debug, int, 0644); @@ -51,8 +52,7 @@ static bool check_params(void) int check_appname(char *task_name) { - char *start = appname; - char *end; + char *start = appname, *end; if (!strlen(appname)) return 0; @@ -384,7 +384,7 @@ static void get_netdev_queue_info(struct oecls_netdev_info *oecls_dev) { struct oecls_netdev_queue_info *rxq_info; struct irq_desc *desc; - int irq; + int irq, cpu; for_each_irq_desc(irq, desc) { if (!desc->action) @@ -393,14 +393,14 @@ static void get_netdev_queue_info(struct oecls_netdev_info *oecls_dev) continue; if (!check_irq_name(desc->action->name, oecls_dev)) continue; - - oecls_debug("irq=%d, [%s], rxq_id=%d\n", irq, desc->action->name, - oecls_dev->rxq_num); - if (oecls_dev->rxq_num >= OECLS_MAX_RXQ_NUM_PER_DEV) break; rxq_info = &oecls_dev->rxq[oecls_dev->rxq_num++]; rxq_info->irq = irq; + cpu = cpumask_first(irq_data_get_effective_affinity_mask(&desc->irq_data)); + rxq_info->affinity_cpu = cpu; + oecls_debug("irq=%d, [%s], rxq_id=%d affinity_cpu:%d\n", + irq, desc->action->name, oecls_dev->rxq_num, cpu); } } @@ -555,8 +555,7 @@ static void clean_oecls_netdev_info(void) static int init_oecls_netdev_info(char *netdev_str) { - char *start = netdev_str; - char *end; + char *start = netdev_str, *end; int err = -ENODEV; while (*start != '\0') { @@ -584,7 +583,7 @@ static int init_oecls_netdev_info(char *netdev_str) struct oecls_numa_info *get_oecls_numa_info(unsigned int nid) { - if (nid >= OECLS_MAX_NUMA_NUM) + if (nid >= oecls_numa_num) return NULL; return &oecls_numa_info_table[nid]; } @@ -609,25 +608,52 @@ static void init_numa_avail_cpus(int nid, struct oecls_numa_info *numa_info) } } -static void init_numa_rxq_bitmap(int nid, struct oecls_numa_info *numa_info) +static void clean_oecls_rxq(void) { struct oecls_numa_bound_dev_info *bound_dev; struct oecls_netdev_info *oecls_dev; - int bound_rxq_num; - int rxq_id; - int devid; - int cpu; + struct oecls_numa_info *numa_info; + int nid, devid; + + for_each_oecls_numa(nid, numa_info) { + for_each_oecls_netdev(devid, oecls_dev) { + bound_dev = &numa_info->bound_dev[devid]; + free_to_l0(bound_dev->cluster_info); + } + } +} + +static int init_numa_rxq_bitmap(int nid, struct oecls_numa_info *numa_info) +{ + int bound_rxq_num, cluster_id, cluster_idx, cur_idx; + struct oecls_numa_bound_dev_info *bound_dev; + struct oecls_netdev_info *oecls_dev; + int rxq_id, devid, cpu, ret = 0; for_each_oecls_netdev(devid, oecls_dev) { bound_rxq_num = 0; bound_dev = &numa_info->bound_dev[devid]; bitmap_zero(bound_dev->bitmap_rxq, OECLS_MAX_RXQ_NUM_PER_DEV); + bound_dev->cluster_info = alloc_from_l0(sizeof(struct oecls_numa_clusterinfo) + * oecls_cluster_per_numa); + if (!bound_dev->cluster_info) { + ret = -ENOMEM; + goto out; + } for (rxq_id = 0; rxq_id < oecls_dev->rxq_num; rxq_id++) { cpu = oecls_dev->rxq[rxq_id].affinity_cpu; if (cpu_to_node(cpu) == nid) { set_bit(rxq_id, bound_dev->bitmap_rxq); + cluster_id = cpu / oecls_cluster_cpu_num; + cluster_idx = cluster_id % oecls_cluster_per_numa; + bound_dev->cluster_info[cluster_idx].cluster_id = cluster_id; + cur_idx = bound_dev->cluster_info[cluster_idx].cur_freeidx++; + bound_dev->cluster_info[cluster_idx].rxqs[cur_idx].rxq_id = rxq_id; + bound_dev->cluster_info[cluster_idx].rxqs[cur_idx].status = 1; bound_rxq_num++; + oecls_debug("cpu:%d cluster_id:%d cluster_idx:%d rxq_id:%d cur_idx:%d\n", + cpu, cluster_id, cluster_idx, rxq_id, cur_idx); } } @@ -635,6 +661,52 @@ static void init_numa_rxq_bitmap(int nid, struct oecls_numa_info *numa_info) nid, devid, oecls_dev->dev_name, oecls_dev->rxq_num, bound_rxq_num, OECLS_MAX_RXQ_NUM_PER_DEV, bound_dev->bitmap_rxq); } + return ret; + +out: + clean_oecls_rxq(); + return ret; +} + +static int get_cluster_rxq(struct oecls_numa_bound_dev_info *bound_dev) +{ + int cpu = smp_processor_id(); + int cluster_id = cpu / oecls_cluster_cpu_num; + int i, j, rxq_id; + + for (i = 0; i < oecls_cluster_per_numa; i++) { + if (cluster_id != bound_dev->cluster_info[i].cluster_id) + continue; + for (j = 0; j < OECLS_MAX_RXQ_NUM_PER_DEV; j++) { + if (bound_dev->cluster_info[i].rxqs[j].status == 1) { + bound_dev->cluster_info[i].rxqs[j].status = 2; + rxq_id = bound_dev->cluster_info[i].rxqs[j].rxq_id; + oecls_debug("cluster:%d cpu:%d alloc rxq_id:%d\n", + cluster_id, cpu, rxq_id); + return rxq_id; + } + } + } + oecls_debug("cluster:%d no free rxq for cpu:%d\n", cluster_id, cpu); + return -1; +} + +static int put_cluster_rxq(struct oecls_numa_bound_dev_info *bound_dev, int rxq_id) +{ + int i, j; + + for (i = 0; i < oecls_cluster_per_numa; i++) { + for (j = 0; j < OECLS_MAX_RXQ_NUM_PER_DEV; j++) { + if (bound_dev->cluster_info[i].rxqs[j].status == 2 && + bound_dev->cluster_info[i].rxqs[j].rxq_id == rxq_id) { + bound_dev->cluster_info[i].rxqs[j].status = 1; + oecls_debug("free rxq_id:%d\n", rxq_id); + return 0; + } + } + } + oecls_debug("no match malloced rxq_id:%d\n", rxq_id); + return -1; } int alloc_rxq_id(int nid, int devid) @@ -655,12 +727,21 @@ int alloc_rxq_id(int nid, int devid) } bound_dev = &numa_info->bound_dev[devid]; + if (strategy == 1) { + rxq_id = get_cluster_rxq(bound_dev); + if (rxq_id < 0 || rxq_id >= OECLS_MAX_RXQ_NUM_PER_DEV) + pr_info("failed to get rxq_id:%d in cluster, try numa\n", rxq_id); + else + goto found; + } + rxq_id = find_first_bit(bound_dev->bitmap_rxq, OECLS_MAX_RXQ_NUM_PER_DEV); if (rxq_id >= OECLS_MAX_RXQ_NUM_PER_DEV) { oecls_error("error rxq_id:%d\n", rxq_id); return -EINVAL; } +found: clear_bit(rxq_id, bound_dev->bitmap_rxq); oecls_debug("alloc nid:%d, dev_id:%d, rxq_id:%d\n", nid, devid, rxq_id); return rxq_id; @@ -688,6 +769,9 @@ void free_rxq_id(int nid, int devid, int rxq_id) return; } + if (strategy == 1) + put_cluster_rxq(bound_dev, rxq_id); + if (test_bit(rxq_id, bound_dev->bitmap_rxq)) { oecls_error("error nid:%d, devid:%d, rxq_id:%d\n", nid, devid, rxq_id); return; @@ -697,22 +781,28 @@ void free_rxq_id(int nid, int devid, int rxq_id) oecls_debug("free nid:%d, dev_id:%d, rxq_id:%d\n", nid, devid, rxq_id); } -static void init_oecls_numa_info(void) +static int init_oecls_numa_info(void) { struct oecls_numa_info *numa_info; - unsigned int numa_num; - int nid; + int nid, ret = 0; - numa_num = num_online_nodes(); - if (numa_num > OECLS_MAX_NUMA_NUM) { - oecls_error("online numa num:%d is too much!\n", numa_num); - numa_num = OECLS_MAX_NUMA_NUM; + oecls_numa_num = num_online_nodes(); + oecls_numa_info_table = alloc_from_l0(sizeof(struct oecls_numa_info) * oecls_numa_num); + if (!oecls_numa_info_table) { + ret = -ENOMEM; + oecls_error("oecls_numa_info_table alloc failed:%d\n", ret); + return ret; } - oecls_numa_num = numa_num; - oecls_debug("set oecls_numa_num=%d\n", numa_num); + + oecls_cluster_cpu_num = cpumask_weight(topology_cluster_cpumask(smp_processor_id())); + oecls_cluster_per_numa = (nr_cpu_ids / oecls_cluster_cpu_num) / oecls_numa_num; + oecls_debug("oecls_numa_num=%d cluster_cpu_num:%d cluster_cpu_num:%d\n", + oecls_numa_num, oecls_cluster_per_numa, oecls_cluster_cpu_num); for_each_oecls_numa(nid, numa_info) init_numa_avail_cpus(nid, numa_info); + + return ret; } static int alloc_available_cpu(int nid, struct oecls_numa_info *numa_info) @@ -746,10 +836,7 @@ static void config_affinity_strategy_default(struct oecls_netdev_info *oecls_dev int rxq_num = oecls_dev->rxq_num; int rxq_per_numa = rxq_num / oecls_numa_num; int remain = rxq_num - rxq_per_numa * oecls_numa_num; - int numa_rxq_id; - int rxq_id; - int nid; - int cpu; + int numa_rxq_id, rxq_id, nid, cpu; oecls_debug("dev=%s, rxq_num=%d, rxq_per_numa=%d, remain=%d\n", oecls_dev->dev_name, rxq_num, rxq_per_numa, remain); @@ -786,47 +873,62 @@ static void config_affinity_strategy_default(struct oecls_netdev_info *oecls_dev static void config_affinity_strategy_cluster(struct oecls_netdev_info *oecls_dev) { - int cluster_cpu_num = 8; - int cluster_num = num_online_cpus() / cluster_cpu_num; - int cluster_cpu_id = 0; - int rxq_id = 0; - int cluster; - int cpu; + int rxq_num = oecls_dev->rxq_num; + int rxq_per_numa = rxq_num / oecls_numa_num; + int remain = rxq_num - rxq_per_numa * oecls_numa_num; + int cpu_idx = oecls_cluster_cpu_num - 1; + int cluster, cpu, rxq_id = 0, round; + + round = rxq_per_numa < oecls_cluster_per_numa ? rxq_per_numa : oecls_cluster_per_numa; + if (remain > 0) + round++; + oecls_debug("round=%d\n", round); - // average config rxq to every cluster while (rxq_id < oecls_dev->rxq_num) { - for (cluster = 0; cluster < cluster_num; cluster++) { - cpu = cluster * cluster_cpu_num + cluster_cpu_id; + for (cluster = 0; cluster < oecls_cluster_per_numa * oecls_numa_num; cluster++) { + if (cluster % oecls_cluster_per_numa >= round) + continue; + cpu = cluster * oecls_cluster_cpu_num + cpu_idx; if (rxq_id >= oecls_dev->rxq_num) break; add_netdev_irq_affinity_cpu(oecls_dev, rxq_id++, cpu); } - cluster_cpu_id++; + cpu_idx--; + if (--cpu_idx < 0) + cpu_idx = oecls_cluster_cpu_num - 1; } } -static void config_affinity_strategy_16cores(struct oecls_netdev_info *oecls_dev) +static void config_affinity_strategy_numa(struct oecls_netdev_info *oecls_dev) { + int rxq_num = oecls_dev->rxq_num; + int rxq_per_numa = rxq_num / oecls_numa_num; + int cpu_per_numa = nr_cpu_ids /oecls_numa_num; + int remain = rxq_num - rxq_per_numa * oecls_numa_num; struct oecls_numa_info *numa_info; - int numa_start_cpu; - int numa_cpu_id; - int rxq_id = 0; - int nid; - int cpu; + int numa_start_cpu, numa_cpu_id; + int rxq_id = 0, nid, cpu; - // only use 16 cores of one numa for_each_oecls_numa(nid, numa_info) { numa_start_cpu = find_first_bit(numa_info->avail_cpus, OECLS_MAX_CPU_NUM); - for (numa_cpu_id = 0; numa_cpu_id < 16; numa_cpu_id++) { - cpu = numa_start_cpu + numa_cpu_id; - + for (numa_cpu_id = 0; numa_cpu_id < rxq_per_numa; numa_cpu_id++) { + cpu = numa_start_cpu + (numa_cpu_id % cpu_per_numa); if (rxq_id >= oecls_dev->rxq_num) break; add_netdev_irq_affinity_cpu(oecls_dev, rxq_id++, cpu); } + if (remain-- > 0) { + cpu = numa_start_cpu + (numa_cpu_id % cpu_per_numa); + add_netdev_irq_affinity_cpu(oecls_dev, rxq_id++, cpu); + } } } +static void config_affinity_strategy_custom(struct oecls_netdev_info *oecls_dev) +{ + oecls_debug("dev=%s\n", oecls_dev->dev_name); +} + static void config_affinity_strategy(void) { struct oecls_netdev_info *oecls_dev; @@ -834,15 +936,16 @@ static void config_affinity_strategy(void) for_each_oecls_netdev(devid, oecls_dev) { switch (strategy) { - case 0: - config_affinity_strategy_default(oecls_dev); - break; case 1: config_affinity_strategy_cluster(oecls_dev); break; case 2: - config_affinity_strategy_16cores(oecls_dev); + config_affinity_strategy_numa(oecls_dev); + break; + case 3: + config_affinity_strategy_custom(oecls_dev); break; + case 0: default: config_affinity_strategy_default(oecls_dev); break; @@ -862,8 +965,7 @@ static void enable_affinity_strategy(void) { struct oecls_netdev_queue_info *rxq_info; struct oecls_netdev_info *oecls_dev; - int rxq_id; - int devid; + int rxq_id, devid; for_each_oecls_netdev(devid, oecls_dev) { for (rxq_id = 0; rxq_id < oecls_dev->rxq_num; rxq_id++) { @@ -888,10 +990,7 @@ static void set_netdev_xps_queue(bool enable) const struct cpumask clear_mask = { 0 }; struct oecls_netdev_info *oecls_dev; const struct cpumask *cpu_mask; - int rxq_id; - int devid; - int cpu; - int nid; + int rxq_id, devid, cpu, nid; for_each_oecls_netdev(devid, oecls_dev) { for (rxq_id = 0; rxq_id < oecls_dev->rxq_num; rxq_id++) { @@ -910,26 +1009,29 @@ static void set_netdev_xps_queue(bool enable) static __init int oecls_init(void) { struct oecls_numa_info *numa_info; - int nid; - int err; + int nid, err; - oecls_debug("[init] mode=%d, ifname=[%s]\n", mode, ifname); if (!check_params()) return -EINVAL; init_oecls_l0_cache(); - init_oecls_numa_info(); + err = init_oecls_numa_info(); + if (err) + goto clean_l0; err = init_oecls_netdev_info(ifname); if (err) - goto out; + goto clean_numa; // Set irq affinity config_affinity_strategy(); enable_affinity_strategy(); // Calculate rxq bounded to one numa - for_each_oecls_numa(nid, numa_info) - init_numa_rxq_bitmap(nid, numa_info); + for_each_oecls_numa(nid, numa_info) { + err = init_numa_rxq_bitmap(nid, numa_info); + if (err) + goto clean_rxq; + } #ifdef CONFIG_XPS set_netdev_xps_queue(true); @@ -941,16 +1043,18 @@ static __init int oecls_init(void) oecls_flow_res_init(); return 0; -out: + +clean_rxq: +clean_numa: clean_oecls_netdev_info(); clean_oecls_numa_info(); +clean_l0: clean_oecls_l0_cache(); return err; } static __exit void oecls_exit(void) { - oecls_debug("[exit] mode=%d\n", mode); if (mode == 0) oecls_ntuple_res_clean(); else @@ -960,6 +1064,7 @@ static __exit void oecls_exit(void) set_netdev_xps_queue(false); #endif + clean_oecls_rxq(); clean_oecls_netdev_info(); clean_oecls_numa_info(); clean_oecls_l0_cache(); diff --git a/net/oenetcls/oenetcls_ntuple.c b/net/oenetcls/oenetcls_ntuple.c index 38d1f5df6ff1..0ac69eb08f24 100644 --- a/net/oenetcls/oenetcls_ntuple.c +++ b/net/oenetcls/oenetcls_ntuple.c @@ -10,7 +10,7 @@ #include <trace/hooks/oenetcls.h> #include "oenetcls.h" -struct oecls_sk_rule_list oecls_sk_rules; +struct oecls_sk_rule_list oecls_sk_rules, oecls_sk_list; static void init_oecls_sk_rules(void) { @@ -21,21 +21,28 @@ static void init_oecls_sk_rules(void) mutex_init(&oecls_sk_rules.mutex); } -static struct hlist_head *oecls_sk_rule_hash(u32 dip4, u16 dport) +static inline struct hlist_head *get_rule_hashlist(u32 dip4, u16 dport) { return oecls_sk_rules.hash + (jhash_2words(dip4, dport, 0) & OECLS_SK_RULE_HASHMASK); } +static inline struct hlist_head *get_sk_hashlist(void *sk) +{ + return oecls_sk_list.hash + (jhash(sk, sizeof(sk), 0) & OECLS_SK_RULE_HASHMASK); +} + static void add_sk_rule(int devid, u32 dip4, u16 dport, void *sk, int action, int ruleid, int nid) { - struct hlist_head *hlist = oecls_sk_rule_hash(dip4, dport); + struct hlist_head *hlist = get_rule_hashlist(dip4, dport); + struct hlist_head *sk_hlist = get_sk_hashlist(sk); struct oecls_sk_rule *rule; + struct oecls_sk_entry *entry; rule = alloc_from_l0(sizeof(struct oecls_sk_rule)); - if (!rule) - return; - oecls_debug("alloc rule=%p\n", rule); + entry = alloc_from_l0(sizeof(struct oecls_sk_entry)); + if (!rule || !entry) + goto out; rule->sk = sk; rule->dip4 = dip4; @@ -45,18 +52,47 @@ static void add_sk_rule(int devid, u32 dip4, u16 dport, void *sk, int action, rule->ruleid = ruleid; rule->nid = nid; hlist_add_head(&rule->node, hlist); + + entry->sk = sk; + entry->sk_rule_hash = jhash_2words(dip4, dport, 0); + hlist_add_head(&entry->node, sk_hlist); + return; +out: + oecls_debug("alloc failed rule:%p entry:%p\n", rule, entry); + free_to_l0(entry); + free_to_l0(rule); +} + +static struct oecls_sk_entry *get_sk_entry(void *sk) +{ + struct hlist_head *sk_hlist = get_sk_hashlist(sk); + struct oecls_sk_entry *entry = NULL; + + hlist_for_each_entry(entry, sk_hlist, node) { + if (entry->sk == sk) + break; + } + return entry; } static void del_sk_rule(struct oecls_sk_rule *rule) { - hlist_del_init(&rule->node); + struct oecls_sk_entry *entry; + + entry = get_sk_entry(rule->sk); + if (!entry) + return; + hlist_del_init(&entry->node); + free_to_l0(entry); + oecls_debug("del rule=%p\n", rule); + hlist_del_init(&rule->node); free_to_l0(rule); } static struct oecls_sk_rule *get_sk_rule(int devid, u32 dip4, u16 dport) { - struct hlist_head *hlist = oecls_sk_rule_hash(dip4, dport); + struct hlist_head *hlist = get_rule_hashlist(dip4, dport); struct oecls_sk_rule *rule = NULL; hlist_for_each_entry(rule, hlist, node) { @@ -66,7 +102,25 @@ static struct oecls_sk_rule *get_sk_rule(int devid, u32 dip4, u16 dport) return rule; } -static bool reuseport_check(int devid, u32 dip4, u16 dport) +static struct oecls_sk_rule *get_rule_from_sk(int devid, void *sk) +{ + struct oecls_sk_rule *rule = NULL; + struct oecls_sk_entry *entry; + struct hlist_head *hlist; + + entry = get_sk_entry(sk); + if (!entry) + return NULL; + + hlist = oecls_sk_rules.hash + (entry->sk_rule_hash & OECLS_SK_RULE_HASHMASK); + hlist_for_each_entry(rule, hlist, node) { + if (rule->devid == devid && rule->sk == sk) + break; + } + return rule; +} + +static inline bool reuseport_check(int devid, u32 dip4, u16 dport) { return !!get_sk_rule(devid, dip4, dport); } @@ -90,7 +144,7 @@ static u32 get_first_ip4_addr(struct net *net) in_dev_for_each_ifa_rcu(ifa, in_dev) { if (!strcmp(dev->name, ifa->ifa_label)) { dip4 = ifa->ifa_local; - oecls_debug("dev: %s, dip4: 0x%x\n", dev->name, dip4); + oecls_debug("dev: %s, dip4:%pI4\n", dev->name, &dip4); goto out; } } @@ -382,10 +436,10 @@ static void del_ntuple_rule(struct sock *sk) mutex_lock(&oecls_sk_rules.mutex); for_each_oecls_netdev(devid, oecls_dev) { strncpy(ctx.netdev, oecls_dev->dev_name, IFNAMSIZ); - rule = get_sk_rule(devid, dip4, dport); + rule = get_rule_from_sk(devid, sk); if (!rule) { - oecls_debug("rule not found! sk:%p, devid:%d, dip4:0x%x, dport:%d\n", sk, - devid, dip4, dport); + oecls_debug("rule not found! sk:%p, devid:%d, dip4:%pI4, dport:%d\n", + sk, devid, &dip4, ntohs(dport)); continue; } @@ -456,8 +510,8 @@ static void ethtool_cfg_rxcls(void *data, struct sock *sk, int is_del) if (sk->sk_family != AF_INET && sk->sk_family != AF_INET6) return; - oecls_debug("[cpu:%d] app:%s, sk:%p, is_del:%d, ip:0x%x, port:0x%x\n", smp_processor_id(), - current->comm, sk, is_del, sk->sk_rcv_saddr, sk->sk_num); + oecls_debug("[cpu:%d] app:%s, sk:%p, is_del:%d, ip:%pI4, port:%d\n", smp_processor_id(), + current->comm, sk, is_del, &sk->sk_rcv_saddr, (u16)sk->sk_num); if (is_del) del_ntuple_rule(sk); -- 2.34.1

2 1

[PATCH openEuler-1.0-LTS 0/2] Fix CVE-2025-38058
by Long Li 30 Jun '25

30 Jun '25

This patch set fix CVE-2025-38058. Al Viro (2): do_umount(): add missing barrier before refcount checks in sync case __legitimize_mnt(): check for MNT_SYNC_UMOUNT should be under mount_lock fs/namespace.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) -- 2.39.2

2 3

[PATCH OLK-6.6] bpf: Avoid __bpf_prog_ret0_warn when jit fails
by Pu Lehui 30 Jun '25

30 Jun '25

From: KaFai Wan <mannkafai(a)gmail.com> mainline inclusion from mainline-v6.16-rc1 commit 86bc9c742426a16b52a10ef61f5b721aecca2344 category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/ICIMYR Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- syzkaller reported an issue: WARNING: CPU: 3 PID: 217 at kernel/bpf/core.c:2357 __bpf_prog_ret0_warn+0xa/0x20 kernel/bpf/core.c:2357 Modules linked in: CPU: 3 UID: 0 PID: 217 Comm: kworker/u32:6 Not tainted 6.15.0-rc4-syzkaller-00040-g8bac8898fe39 RIP: 0010:__bpf_prog_ret0_warn+0xa/0x20 kernel/bpf/core.c:2357 Call Trace: <TASK> bpf_dispatcher_nop_func include/linux/bpf.h:1316 [inline] __bpf_prog_run include/linux/filter.h:718 [inline] bpf_prog_run include/linux/filter.h:725 [inline] cls_bpf_classify+0x74a/0x1110 net/sched/cls_bpf.c:105 ... When creating bpf program, 'fp->jit_requested' depends on bpf_jit_enable. This issue is triggered because of CONFIG_BPF_JIT_ALWAYS_ON is not set and bpf_jit_enable is set to 1, causing the arch to attempt JIT the prog, but jit failed due to FAULT_INJECTION. As a result, incorrectly treats the program as valid, when the program runs it calls `__bpf_prog_ret0_warn` and triggers the WARN_ON_ONCE(1). Reported-by: syzbot+0903f6d7f285e41cdf10(a)syzkaller.appspotmail.com Closes: https://lore.kernel.org/bpf/6816e34e.a70a0220.254cdc.002c.GAE@google.com Fixes: fa9dd599b4da ("bpf: get rid of pure_initcall dependency to enable jits") Signed-off-by: KaFai Wan <mannkafai(a)gmail.com> Link: https://lore.kernel.org/r/20250526133358.2594176-1-mannkafai@gmail.com Signed-off-by: Alexei Starovoitov <ast(a)kernel.org> Signed-off-by: Pu Lehui <pulehui(a)huawei.com> --- kernel/bpf/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c index e968b07fd6b4..2ecaf891a167 100644 --- a/kernel/bpf/core.c +++ b/kernel/bpf/core.c @@ -2360,7 +2360,7 @@ struct bpf_prog *bpf_prog_select_runtime(struct bpf_prog *fp, int *err) /* In case of BPF to BPF calls, verifier did all the prep * work with regards to JITing, etc. */ - bool jit_needed = false; + bool jit_needed = fp->jit_requested; if (fp->bpf_func) goto finalize; -- 2.34.1

2 1

[openeuler:openEuler-1.0-LTS 1673/1673] fs/btrfs/transaction.o: warning: objtool: start_transaction()+0x3bd: unreachable instruction
by kernel test robot 30 Jun '25

30 Jun '25

Hi Filipe, First bad commit (maybe != root cause): tree: https://gitee.com/openeuler/kernel.git openEuler-1.0-LTS head: a9d08fabefcfd9cb8ebe45f8b79a321c0f349d40 commit: 2582bb7bd5a9278555265a99a94a8416e5eb7d3c [1673/1673] Btrfs: fix deadlock between fiemap and transaction commits config: x86_64-buildonly-randconfig-002-20250628 (https://download.01.org/0day-ci/archive/20250630/202506300834.dp152UeL-lkp@…) compiler: clang version 20.1.7 (https://github.com/llvm/llvm-project 6146a88f60492b520a36f8f8f3231e15f3cc6082) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250630/202506300834.dp152UeL-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/202506300834.dp152UeL-lkp@intel.com/ All warnings (new ones prefixed by >>): >> fs/btrfs/transaction.o: warning: objtool: start_transaction()+0x3bd: unreachable instruction -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0

[openeuler:OLK-6.6 2411/2411] arch/x86/kvm/../../../virt/kvm/eventfd.c:719: warning: cannot understand function prototype: 'struct eventfd_shadow '
by kernel test robot 30 Jun '25

30 Jun '25

tree: https://gitee.com/openeuler/kernel.git OLK-6.6 head: c02455c4cce8052f74fc11c09e9073c01c7f33a4 commit: 2dcf53278783921137dbb4eb4628f2c96cac65e2 [2411/2411] eventfd:Introduce the eventfd batch processing interface config: x86_64-buildonly-randconfig-2001-20250626 (https://download.01.org/0day-ci/archive/20250629/202506291906.a435dLWR-lkp@…) compiler: gcc-12 (Debian 12.2.0-14) 12.2.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250629/202506291906.a435dLWR-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/202506291906.a435dLWR-lkp@intel.com/ All warnings (new ones prefixed by >>): >> arch/x86/kvm/../../../virt/kvm/eventfd.c:719: warning: cannot understand function prototype: 'struct eventfd_shadow ' >> arch/x86/kvm/../../../virt/kvm/eventfd.c:1102: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst * If an exception occurs, arch/x86/kvm/../../../virt/kvm/eventfd.c:1131: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst * If the temporarily stored eventfd cannot be successfully updated arch/x86/kvm/../../../virt/kvm/eventfd.c:1176: warning: This comment starts with '/**', but isn't a kernel-doc comment. Refer Documentation/doc-guide/kernel-doc.rst * Copy the current buses, Kconfig warnings: (for reference only) WARNING: unmet direct dependencies detected for ACPI_HOTPLUG_IGNORE_OSC Depends on [n]: ACPI [=y] && ACPI_HOTPLUG_CPU [=n] Selected by [y]: - X86 [=y] && ACPI [=y] && HOTPLUG_CPU [=y] vim +719 arch/x86/kvm/../../../virt/kvm/eventfd.c 713 714 /** 715 * eventfd_shadow 716 * Save the eventfd pointer and operation. 717 * To protect Kabi, these members are placed here 718 */ > 719 struct eventfd_shadow { 720 struct list_head node; 721 struct _ioeventfd *eventfd; 722 /* Indicate the operations required for this eventfd */ 723 u8 flag; 724 }; 725 -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0

[openeuler:OLK-6.6 2411/2411] mm/mempolicy.c:1115:32: warning: variable 'vma' set but not used
by kernel test robot 29 Jun '25

29 Jun '25

tree: https://gitee.com/openeuler/kernel.git OLK-6.6 head: c02455c4cce8052f74fc11c09e9073c01c7f33a4 commit: dba69db9a5fc7c2c9c0cda0f32bb3b5f3d73ef93 [2411/2411] mm/damon/vaddr: Add demotion interface for migrating cold pages to target nodemask config: x86_64-buildonly-randconfig-2001-20250626 (https://download.01.org/0day-ci/archive/20250629/202506291741.f76J7DJD-lkp@…) compiler: gcc-12 (Debian 12.2.0-14) 12.2.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250629/202506291741.f76J7DJD-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/202506291741.f76J7DJD-lkp@intel.com/ All warnings (new ones prefixed by >>): mm/mempolicy.c: In function 'migrate_area_to_node': >> mm/mempolicy.c:1115:32: warning: variable 'vma' set but not used [-Wunused-but-set-variable] 1115 | struct vm_area_struct *vma; | ^~~ Kconfig warnings: (for reference only) WARNING: unmet direct dependencies detected for ACPI_HOTPLUG_IGNORE_OSC Depends on [n]: ACPI [=y] && ACPI_HOTPLUG_CPU [=n] Selected by [y]: - X86 [=y] && ACPI [=y] && HOTPLUG_CPU [=y] vim +/vma +1115 mm/mempolicy.c 1106 1107 /* 1108 * Migrate area pages from one node to a target node. 1109 * Returns error or the number of pages not migrated. 1110 */ 1111 static int migrate_area_to_node(struct mm_struct *mm, int source, int dest, 1112 unsigned long start, unsigned long end, int flags) 1113 { 1114 nodemask_t nmask; > 1115 struct vm_area_struct *vma; 1116 LIST_HEAD(pagelist); 1117 int err = 0; 1118 struct migration_target_control mtc = { 1119 .nid = dest, 1120 .gfp_mask = GFP_HIGHUSER_MOVABLE | __GFP_THISNODE, 1121 }; 1122 1123 nodes_clear(nmask); 1124 node_set(source, nmask); 1125 1126 /* 1127 * This does not "check" the range but isolates all pages that 1128 * need migration. Between passing in the full user address 1129 * space range and MPOL_MF_DISCONTIG_OK, this call can not fail. 1130 */ 1131 vma = find_vma(mm, 0); 1132 VM_BUG_ON(!(flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))); 1133 queue_pages_range(mm, start, end, &nmask, 1134 flags | MPOL_MF_DISCONTIG_OK, &pagelist, false); 1135 1136 if (!list_empty(&pagelist)) { 1137 err = migrate_pages(&pagelist, alloc_migration_target, NULL, 1138 (unsigned long)&mtc, MIGRATE_SYNC, MR_DAMON_DEMOTION, NULL); 1139 if (err) 1140 putback_movable_pages(&pagelist); 1141 } 1142 1143 return err; 1144 } 1145 -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki

1 0