June 2023 - Kernel - mailweb.openeuler.org

[PATCH openEuler-22.03-LTS] scsi: iscsi: iscsi_tcp: Fix null-ptr-deref while calling getpeername()
by Zhong Jinghua 14 Jun '23

14 Jun '23

From: Mike Christie <michael.christie(a)oracle.com> mainline inclusion from mainline-v6.1-rc1 commit 57569c37f0add1b6489e1a1563c71519daf732cf category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I6I8YD CVE: NA Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… ---------------------------------------- Fix a NULL pointer crash that occurs when we are freeing the socket at the same time we access it via sysfs. The problem is that: 1. iscsi_sw_tcp_conn_get_param() and iscsi_sw_tcp_host_get_param() take the frwd_lock and do sock_hold() then drop the frwd_lock. sock_hold() does a get on the "struct sock". 2. iscsi_sw_tcp_release_conn() does sockfd_put() which does the last put on the "struct socket" and that does __sock_release() which sets the sock->ops to NULL. 3. iscsi_sw_tcp_conn_get_param() and iscsi_sw_tcp_host_get_param() then call kernel_getpeername() which accesses the NULL sock->ops. Above we do a get on the "struct sock", but we needed a get on the "struct socket". Originally, we just held the frwd_lock the entire time but in commit bcf3a2953d36 ("scsi: iscsi: iscsi_tcp: Avoid holding spinlock while calling getpeername()") we switched to refcount based because the network layer changed and started taking a mutex in that path, so we could no longer hold the frwd_lock. Instead of trying to maintain multiple refcounts, this just has us use a mutex for accessing the socket in the interface code paths. Link: https://lore.kernel.org/r/20220907221700.10302-1-michael.christie@oracle.com Fixes: bcf3a2953d36 ("scsi: iscsi: iscsi_tcp: Avoid holding spinlock while calling getpeername()") Signed-off-by: Mike Christie <michael.christie(a)oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen(a)oracle.com> Signed-off-by: Zhong Jinghua <zhongjinghua(a)huawei.com> --- drivers/scsi/iscsi_tcp.c | 73 ++++++++++++++++++++++++++++------------ drivers/scsi/iscsi_tcp.h | 2 ++ 2 files changed, 54 insertions(+), 21 deletions(-) diff --git a/drivers/scsi/iscsi_tcp.c b/drivers/scsi/iscsi_tcp.c index 4575d0f0dd6c..e8b0ac0193ee 100644 --- a/drivers/scsi/iscsi_tcp.c +++ b/drivers/scsi/iscsi_tcp.c @@ -566,6 +566,8 @@ iscsi_sw_tcp_conn_create(struct iscsi_cls_session *cls_session, tcp_conn = conn->dd_data; tcp_sw_conn = tcp_conn->dd_data; + mutex_init(&tcp_sw_conn->sock_lock); + tfm = crypto_alloc_ahash("crc32c", 0, CRYPTO_ALG_ASYNC); if (IS_ERR(tfm)) goto free_conn; @@ -600,11 +602,15 @@ iscsi_sw_tcp_conn_create(struct iscsi_cls_session *cls_session, static void iscsi_sw_tcp_release_conn(struct iscsi_conn *conn) { - struct iscsi_session *session = conn->session; struct iscsi_tcp_conn *tcp_conn = conn->dd_data; struct iscsi_sw_tcp_conn *tcp_sw_conn = tcp_conn->dd_data; struct socket *sock = tcp_sw_conn->sock; + /* + * The iscsi transport class will make sure we are not called in + * parallel with start, stop, bind and destroys. However, this can be + * called twice if userspace does a stop then a destroy. + */ if (!sock) return; @@ -612,9 +618,9 @@ static void iscsi_sw_tcp_release_conn(struct iscsi_conn *conn) iscsi_sw_tcp_conn_restore_callbacks(conn); sock_put(sock->sk); - spin_lock_bh(&session->frwd_lock); + mutex_lock(&tcp_sw_conn->sock_lock); tcp_sw_conn->sock = NULL; - spin_unlock_bh(&session->frwd_lock); + mutex_unlock(&tcp_sw_conn->sock_lock); sockfd_put(sock); } @@ -666,7 +672,6 @@ iscsi_sw_tcp_conn_bind(struct iscsi_cls_session *cls_session, struct iscsi_cls_conn *cls_conn, uint64_t transport_eph, int is_leading) { - struct iscsi_session *session = cls_session->dd_data; struct iscsi_conn *conn = cls_conn->dd_data; struct iscsi_tcp_conn *tcp_conn = conn->dd_data; struct iscsi_sw_tcp_conn *tcp_sw_conn = tcp_conn->dd_data; @@ -686,10 +691,10 @@ iscsi_sw_tcp_conn_bind(struct iscsi_cls_session *cls_session, if (err) goto free_socket; - spin_lock_bh(&session->frwd_lock); + mutex_lock(&tcp_sw_conn->sock_lock); /* bind iSCSI connection and socket */ tcp_sw_conn->sock = sock; - spin_unlock_bh(&session->frwd_lock); + mutex_unlock(&tcp_sw_conn->sock_lock); /* setup Socket parameters */ sk = sock->sk; @@ -725,8 +730,15 @@ static int iscsi_sw_tcp_conn_set_param(struct iscsi_cls_conn *cls_conn, break; case ISCSI_PARAM_DATADGST_EN: iscsi_set_param(cls_conn, param, buf, buflen); + + mutex_lock(&tcp_sw_conn->sock_lock); + if (!tcp_sw_conn->sock) { + mutex_unlock(&tcp_sw_conn->sock_lock); + return -ENOTCONN; + } tcp_sw_conn->sendpage = conn->datadgst_en ? sock_no_sendpage : tcp_sw_conn->sock->ops->sendpage; + mutex_unlock(&tcp_sw_conn->sock_lock); break; case ISCSI_PARAM_MAX_R2T: return iscsi_tcp_set_max_r2t(conn, buf); @@ -741,8 +753,8 @@ static int iscsi_sw_tcp_conn_get_param(struct iscsi_cls_conn *cls_conn, enum iscsi_param param, char *buf) { struct iscsi_conn *conn = cls_conn->dd_data; - struct iscsi_tcp_conn *tcp_conn = conn->dd_data; - struct iscsi_sw_tcp_conn *tcp_sw_conn = tcp_conn->dd_data; + struct iscsi_sw_tcp_conn *tcp_sw_conn; + struct iscsi_tcp_conn *tcp_conn; struct sockaddr_in6 addr; struct socket *sock; int rc; @@ -752,21 +764,36 @@ static int iscsi_sw_tcp_conn_get_param(struct iscsi_cls_conn *cls_conn, case ISCSI_PARAM_CONN_ADDRESS: case ISCSI_PARAM_LOCAL_PORT: spin_lock_bh(&conn->session->frwd_lock); - if (!tcp_sw_conn || !tcp_sw_conn->sock) { + if (!conn->session->leadconn) { spin_unlock_bh(&conn->session->frwd_lock); return -ENOTCONN; } - sock = tcp_sw_conn->sock; - sock_hold(sock->sk); + /* + * The conn has been setup and bound, so just grab a ref + * incase a destroy runs while we are in the net layer. + */ + iscsi_get_conn(conn->cls_conn); spin_unlock_bh(&conn->session->frwd_lock); + tcp_conn = conn->dd_data; + tcp_sw_conn = tcp_conn->dd_data; + + mutex_lock(&tcp_sw_conn->sock_lock); + sock = tcp_sw_conn->sock; + if (!sock) { + rc = -ENOTCONN; + goto sock_unlock; + } + if (param == ISCSI_PARAM_LOCAL_PORT) rc = kernel_getsockname(sock, (struct sockaddr *)&addr); else rc = kernel_getpeername(sock, (struct sockaddr *)&addr); - sock_put(sock->sk); +sock_unlock: + mutex_unlock(&tcp_sw_conn->sock_lock); + iscsi_put_conn(conn->cls_conn); if (rc < 0) return rc; @@ -805,17 +832,21 @@ static int iscsi_sw_tcp_host_get_param(struct Scsi_Host *shost, } tcp_conn = conn->dd_data; tcp_sw_conn = tcp_conn->dd_data; - sock = tcp_sw_conn->sock; - if (!sock) { - spin_unlock_bh(&session->frwd_lock); - return -ENOTCONN; - } - sock_hold(sock->sk); + /* + * The conn has been setup and bound, so just grab a ref + * incase a destroy runs while we are in the net layer. + */ + iscsi_get_conn(conn->cls_conn); spin_unlock_bh(&session->frwd_lock); - rc = kernel_getsockname(sock, - (struct sockaddr *)&addr); - sock_put(sock->sk); + mutex_lock(&tcp_sw_conn->sock_lock); + sock = tcp_sw_conn->sock; + if (!sock) + rc = -ENOTCONN; + else + rc = kernel_getsockname(sock, (struct sockaddr *)&addr); + mutex_unlock(&tcp_sw_conn->sock_lock); + iscsi_put_conn(conn->cls_conn); if (rc < 0) return rc; diff --git a/drivers/scsi/iscsi_tcp.h b/drivers/scsi/iscsi_tcp.h index 791453195099..1731956326e2 100644 --- a/drivers/scsi/iscsi_tcp.h +++ b/drivers/scsi/iscsi_tcp.h @@ -28,6 +28,8 @@ struct iscsi_sw_tcp_send { struct iscsi_sw_tcp_conn { struct socket *sock; + /* Taken when accessing the sock from the netlink/sysfs interface */ + struct mutex sock_lock; struct iscsi_sw_tcp_send out; /* old values for socket callbacks */ -- 2.31.1

2 1

[PATCH OLK-5.10] sched: Add static key for qos sched to improve performance
by Hui Tang 14 Jun '23

14 Jun '23

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I7D76D -------------------------------- Add cmdline of 'noqossched' which is enable by default. Use the static key to reduce noise for multi level prio, smt expell and prio load balance. Signed-off-by: Hui Tang <tanghui20(a)huawei.com> --- include/linux/sched.h | 26 ++++++++++++++--- kernel/sched/core.c | 11 ++++++++ kernel/sched/fair.c | 66 ++++++++++++++++++++++++++++--------------- kernel/sched/sched.h | 6 +++- kernel/signal.c | 3 +- 5 files changed, 84 insertions(+), 28 deletions(-) diff --git a/include/linux/sched.h b/include/linux/sched.h index 3aae225f98a7..fdd3324cc858 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1869,15 +1869,19 @@ extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk); __get_task_comm(buf, sizeof(buf), tsk); \ }) +#ifdef CONFIG_SMP #ifdef CONFIG_QOS_SCHED_SMT_EXPELLER -void qos_smt_check_need_resched(void); +static __always_inline bool qos_sched_enabled(void); +extern struct static_key_true qos_smt_expell_switch; +extern void qos_smt_check_need_resched(void); #endif - -#ifdef CONFIG_SMP static __always_inline void scheduler_ipi(void) { #ifdef CONFIG_QOS_SCHED_SMT_EXPELLER - qos_smt_check_need_resched(); + + if (qos_sched_enabled() && + static_branch_likely(&qos_smt_expell_switch)) + qos_smt_check_need_resched(); #endif /* * Fold TIF_NEED_RESCHED into the preempt_count; anybody setting @@ -2215,11 +2219,25 @@ static inline void sched_core_fork(struct task_struct *p) { } void sched_move_offline_task(struct task_struct *p); void sched_qos_offline_wait(void); int sched_qos_cpu_overload(void); + +extern struct static_key_true __qos_sched_switch; +static __always_inline bool qos_sched_enabled(void) +{ + if (static_branch_likely(&__qos_sched_switch)) + return true; + + return false; +} #else static inline int sched_qos_cpu_overload(void) { return 0; } + +static __always_inline bool qos_sched_enabled(void) +{ + return false; +} #endif #ifdef CONFIG_QOS_SCHED_DYNAMIC_AFFINITY diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 454bca0c9c6b..71c69c8161ff 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -3617,6 +3617,17 @@ static void __init init_schedstats(void) set_schedstats(__sched_schedstats); } +#ifdef CONFIG_QOS_SCHED +DEFINE_STATIC_KEY_TRUE(__qos_sched_switch); + +static int __init no_qos_sched_switch_setup(char *__unused) +{ + static_branch_disable(&__qos_sched_switch); + return 1; +} +__setup("noqossched", no_qos_sched_switch_setup); +#endif + #ifdef CONFIG_PROC_SYSCTL int sysctl_schedstats(struct ctl_table *table, int write, void *buffer, size_t *lenp, loff_t *ppos) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index ff209d25c21c..003ebd244d31 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -2997,14 +2997,15 @@ static inline void update_scan_period(struct task_struct *p, int new_cpu) #endif /* CONFIG_NUMA_BALANCING */ #ifdef CONFIG_QOS_SCHED_PRIO_LB -static void +static __always_inline void adjust_rq_cfs_tasks(void (*list_op)(struct list_head *, struct list_head *), struct rq *rq, struct sched_entity *se) { struct task_group *tg = task_group(task_of(se)); - if (sysctl_sched_prio_load_balance_enabled && is_offline_level(tg->qos_level)) + if (qos_sched_enabled() && sysctl_sched_prio_load_balance_enabled && + is_offline_level(tg->qos_level)) (*list_op)(&se->group_node, &rq->cfs_offline_tasks); else (*list_op)(&se->group_node, &rq->cfs_tasks); @@ -3235,7 +3236,8 @@ static long calc_group_shares(struct cfs_rq *cfs_rq) tg_shares = READ_ONCE(tg->shares); #ifdef CONFIG_QOS_SCHED_MULTILEVEL - tg_shares = qos_reweight(tg_shares, tg); + if (qos_sched_enabled()) + tg_shares = qos_reweight(tg_shares, tg); #endif load = max(scale_load_down(cfs_rq->load.weight), cfs_rq->avg.load_avg); @@ -3286,7 +3288,8 @@ static void update_cfs_group(struct sched_entity *se) #ifndef CONFIG_SMP shares = READ_ONCE(gcfs_rq->tg->shares); #ifdef CONFIG_QOS_SCHED_MULTILEVEL - shares = qos_reweight(shares, gcfs_rq->tg); + if (qos_sched_enabled()) + shares = qos_reweight(shares, gcfs_rq->tg); #endif if (likely(se->load.weight == shares)) @@ -7071,7 +7074,7 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) #ifdef CONFIG_JUMP_LABEL static DEFINE_STATIC_KEY_FALSE(__dynamic_affinity_used); -static inline bool dynamic_affinity_used(void) +static __always_inline bool dynamic_affinity_used(void) { return static_branch_unlikely(&__dynamic_affinity_used); } @@ -7082,7 +7085,7 @@ void dynamic_affinity_enable(void) } #else /* CONFIG_JUMP_LABEL */ -static bool dynamic_affinity_used(void) +static __always_inline bool dynamic_affinity_used(void) { return true; } @@ -7097,9 +7100,6 @@ int sysctl_sched_util_low_pct = 85; static inline bool prefer_cpus_valid(struct task_struct *p) { - if (!dynamic_affinity_used()) - return false; - return p->prefer_cpus && !cpumask_empty(p->prefer_cpus) && !cpumask_equal(p->prefer_cpus, p->cpus_ptr) && @@ -7208,7 +7208,9 @@ select_task_rq_fair(struct task_struct *p, int prev_cpu, int sd_flag, int wake_f lockdep_assert_held(&p->pi_lock); #ifdef CONFIG_QOS_SCHED_DYNAMIC_AFFINITY - set_task_select_cpus(p, &idlest_cpu, sd_flag); + p->select_cpus = p->cpus_ptr; + if (dynamic_affinity_used()) + set_task_select_cpus(p, &idlest_cpu, sd_flag); #endif if (sd_flag & SD_BALANCE_WAKE) { @@ -7770,9 +7772,13 @@ static int __unthrottle_qos_cfs_rqs(int cpu) return res; } -static int unthrottle_qos_cfs_rqs(int cpu) +static __always_inline int unthrottle_qos_cfs_rqs(int cpu) { int res; + + if (!qos_sched_enabled()) + return 0; + res = __unthrottle_qos_cfs_rqs(cpu); if (qos_timer_is_activated(cpu) && !qos_smt_expelled(cpu)) @@ -7781,8 +7787,11 @@ static int unthrottle_qos_cfs_rqs(int cpu) return res; } -static bool check_qos_cfs_rq(struct cfs_rq *cfs_rq) +static __always_inline bool check_qos_cfs_rq(struct cfs_rq *cfs_rq) { + if (!qos_sched_enabled()) + return false; + if (unlikely(__this_cpu_read(qos_cpu_overload))) { return false; } @@ -7864,8 +7873,11 @@ void init_qos_hrtimer(int cpu) * we should schedule offline tasks to run so that they can leave kernel * critical sections, and throttle them before returning to user mode. */ -static void qos_schedule_throttle(struct task_struct *p) +static __always_inline void qos_schedule_throttle(struct task_struct *p) { + if (!qos_sched_enabled()) + return; + if (unlikely(current->flags & PF_KTHREAD)) return; @@ -7876,7 +7888,7 @@ static void qos_schedule_throttle(struct task_struct *p) } #ifndef CONFIG_QOS_SCHED_SMT_EXPELLER -static bool qos_smt_expelled(int this_cpu) +static __always_inline bool qos_smt_expelled(int this_cpu) { return false; } @@ -7952,8 +7964,11 @@ static bool qos_sched_idle_cpu(int this_cpu) rq->nr_running); } -static bool qos_smt_expelled(int this_cpu) +static __always_inline bool qos_smt_expelled(int this_cpu) { + if (!qos_sched_enabled()) + return false; + if (!static_branch_likely(&qos_smt_expell_switch)) return false; @@ -8011,8 +8026,11 @@ static void qos_smt_send_ipi(int this_cpu) } } -static void qos_smt_expel(int this_cpu, struct task_struct *p) +static __always_inline void qos_smt_expel(int this_cpu, struct task_struct *p) { + if (!qos_sched_enabled()) + return; + if (!static_branch_likely(&qos_smt_expell_switch)) return; @@ -8035,7 +8053,7 @@ static bool _qos_smt_check_need_resched(int this_cpu, struct rq *rq) { int cpu; - if (!qos_smt_enabled()) + if (!qos_sched_enabled() || !qos_smt_enabled()) return false; for_each_cpu(cpu, cpu_smt_mask(this_cpu)) { @@ -8145,7 +8163,8 @@ pick_next_task_fair(struct rq *rq, struct task_struct *prev, struct rq_flags *rf #ifdef CONFIG_FAIR_GROUP_SCHED if (!prev || prev->sched_class != &fair_sched_class) { #ifdef CONFIG_QOS_SCHED - if (cfs_rq->idle_h_nr_running != 0 && rq->online) + if (qos_sched_enabled() && + cfs_rq->idle_h_nr_running != 0 && rq->online) goto qos_simple; else #endif @@ -8661,7 +8680,7 @@ static int task_hot(struct task_struct *p, struct lb_env *env) #ifdef CONFIG_QOS_SCHED_PRIO_LB /* Preempt sched idle cpu do not consider migration cost */ - if (sysctl_sched_prio_load_balance_enabled && + if (qos_sched_enabled() && sysctl_sched_prio_load_balance_enabled && cpus_share_cache(env->src_cpu, env->dst_cpu) && sched_idle_cpu(env->dst_cpu)) return 0; @@ -8770,7 +8789,9 @@ int can_migrate_task(struct task_struct *p, struct lb_env *env) return 0; #ifdef CONFIG_QOS_SCHED_DYNAMIC_AFFINITY - set_task_select_cpus(p, NULL, 0); + p->select_cpus = p->cpus_ptr; + if (dynamic_affinity_used()) + set_task_select_cpus(p, NULL, 0); if (!cpumask_test_cpu(env->dst_cpu, p->select_cpus)) { #else if (!cpumask_test_cpu(env->dst_cpu, p->cpus_ptr)) { @@ -8916,7 +8937,7 @@ static struct task_struct *detach_one_task(struct lb_env *env) return p; } #ifdef CONFIG_QOS_SCHED_PRIO_LB - if (sysctl_sched_prio_load_balance_enabled) { + if (qos_sched_enabled() && sysctl_sched_prio_load_balance_enabled) { loop++; if (loop == 1) { tasks = &env->src_rq->cfs_offline_tasks; @@ -9055,7 +9076,8 @@ static int detach_tasks(struct lb_env *env) } #ifdef CONFIG_QOS_SCHED_PRIO_LB - if (sysctl_sched_prio_load_balance_enabled && env->imbalance > 0) { + if (qos_sched_enabled() && sysctl_sched_prio_load_balance_enabled && + env->imbalance > 0) { /* * Avoid offline tasks starve to death if env->loop exceed * env->loop_max, so we should set env->loop to 0 and detach diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 27fd1240ac85..95ae9ac553fc 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -3065,8 +3065,12 @@ static inline int is_offline_level(long qos_level) #endif #ifdef CONFIG_QOS_SCHED_SMT_EXPELLER -static inline int task_has_qos_idle_policy(struct task_struct *p) +static __always_inline int task_has_qos_idle_policy(struct task_struct *p) { + if (!qos_sched_enabled() || + !static_branch_likely(&qos_smt_expell_switch)) + return 0; + return qos_idle_policy(task_group(p)->qos_level) && p->policy == SCHED_IDLE; } #endif diff --git a/kernel/signal.c b/kernel/signal.c index cf498d949f2f..dd6076c6235f 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -1050,7 +1050,8 @@ static void complete_signal(int sig, struct task_struct *p, enum pid_type type) t = p; do { #ifdef CONFIG_QOS_SCHED - sched_move_offline_task(t); + if (qos_sched_enabled()) + sched_move_offline_task(t); #endif task_clear_jobctl_pending(t, JOBCTL_PENDING_MASK); sigaddset(&t->pending.signal, SIGKILL); -- 2.17.1

2 1

[PATCH v2 OLK-5.10] arm64: entry: fix ABBA deadlock when el1_dbg preempts rq->lock context
by Yu Liao 14 Jun '23

14 Jun '23

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I7DAV3 CVE: NA -------------------------------- When arm64 enters el1_dbg, it will hold rcu_node lock. If rq_lock critical section is preempted by el1_dbg, it may cause an ABBA deadlock. The scenario is as follows: CPU0 CPU1 ---- ---- process context rcu_gp_fqs() lock(&rq->lock); lock(rcu_node_0); lock(&rq->lock); //el1_dbg() lock(rcu_node_0); rcu_nmi_enter() do not hold rcu_node in the nmi context, so fix this issue by keeping the same logical for el1_dbg without marking el1_dbg as nmi. Fixes: d8bb6718c4db ("arm64: Make debug exception handlers visible from RCU") Signed-off-by: Yu Liao <liaoyu15(a)huawei.com> --- arch/arm64/kernel/entry-common.c | 8 ++++++++ include/linux/preempt.h | 2 ++ kernel/panic.c | 5 +++++ kernel/rcu/tree.c | 2 +- 4 files changed, 16 insertions(+), 1 deletion(-) diff --git a/arch/arm64/kernel/entry-common.c b/arch/arm64/kernel/entry-common.c index 387c755b79f6..b8a527ac6781 100644 --- a/arch/arm64/kernel/entry-common.c +++ b/arch/arm64/kernel/entry-common.c @@ -17,6 +17,7 @@ #include <asm/mmu.h> #include <asm/sysreg.h> +static DEFINE_PER_CPU(int, dbg_count); /* * This is intended to match the logic in irqentry_enter(), handling the kernel * mode transitions only. @@ -150,10 +151,16 @@ static void noinstr el1_inv(struct pt_regs *regs, unsigned long esr) exit_to_kernel_mode(regs); } +int in_dbg(void) +{ + return this_cpu_read(dbg_count) > 0; +} + static void noinstr arm64_enter_el1_dbg(struct pt_regs *regs) { regs->lockdep_hardirqs = lockdep_hardirqs_enabled(); + this_cpu_inc(dbg_count); lockdep_hardirqs_off(CALLER_ADDR0); rcu_nmi_enter(); @@ -172,6 +179,7 @@ static void noinstr arm64_exit_el1_dbg(struct pt_regs *regs) rcu_nmi_exit(); if (restore) lockdep_hardirqs_on(CALLER_ADDR0); + this_cpu_dec(dbg_count); } static void noinstr el1_dbg(struct pt_regs *regs, unsigned long esr) diff --git a/include/linux/preempt.h b/include/linux/preempt.h index 7d9c1c0e149c..89d5281e060a 100644 --- a/include/linux/preempt.h +++ b/include/linux/preempt.h @@ -352,4 +352,6 @@ static __always_inline void migrate_enable(void) preempt_enable(); } +int in_dbg(void); + #endif /* __LINUX_PREEMPT_H */ diff --git a/kernel/panic.c b/kernel/panic.c index d991c3b1b559..9dadd3c5df57 100644 --- a/kernel/panic.c +++ b/kernel/panic.c @@ -761,3 +761,8 @@ static int __init panic_on_taint_setup(char *s) return 0; } early_param("panic_on_taint", panic_on_taint_setup); + +int __weak in_dbg(void) +{ + return 0; +} diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 0142855482f8..18b9c74474ec 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -957,7 +957,7 @@ void __rcu_irq_enter_check_tick(void) struct rcu_data *rdp = this_cpu_ptr(&rcu_data); // If we're here from NMI there's nothing to do. - if (in_nmi()) + if (in_nmi() || in_dbg()) return; RCU_LOCKDEP_WARN(rcu_dynticks_curr_cpu_in_eqs(), -- 2.25.1

2 1

[OLK-5.10 PATCH] arm64: entry: fix ABBA deadlock when el1_dbg preempts rq->lock context
by Yu Liao 14 Jun '23

14 Jun '23

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I7DAV3 CVE: NA -------------------------------- When arm64 enters el1_dbg, it will hold rcu_node lock. If rq_lock critical section is preempted by el1_dbg, it may cause an ABBA deadlock. The scenario is as follows: CPU0 CPU1 ---- ---- process context rcu_gp_fqs() lock(&rq->lock); lock(rcu_node_0); lock(&rq->lock); //el1_dbg() lock(rcu_node_0); rcu_nmi_enter() do not hold rcu_node in the nmi context, so fix this issue by keeping the same logical for el1_dbg without marking el1_dbg as nmi. Fixes: d8bb6718c4db ("arm64: Make debug exception handlers visible from RCU") Signed-off-by: Yu Liao <liaoyu15(a)huawei.com> --- arch/arm64/kernel/entry-common.c | 8 ++++++++ include/linux/preempt.h | 2 ++ kernel/panic.c | 5 +++++ kernel/rcu/tree.c | 2 +- 4 files changed, 16 insertions(+), 1 deletion(-) diff --git a/arch/arm64/kernel/entry-common.c b/arch/arm64/kernel/entry-common.c index 387c755b79f6..b8a527ac6781 100644 --- a/arch/arm64/kernel/entry-common.c +++ b/arch/arm64/kernel/entry-common.c @@ -17,6 +17,7 @@ #include <asm/mmu.h> #include <asm/sysreg.h> +static DEFINE_PER_CPU(int, dbg_count); /* * This is intended to match the logic in irqentry_enter(), handling the kernel * mode transitions only. @@ -150,10 +151,16 @@ static void noinstr el1_inv(struct pt_regs *regs, unsigned long esr) exit_to_kernel_mode(regs); } +int in_dbg(void) +{ + return this_cpu_read(dbg_count) > 0; +} + static void noinstr arm64_enter_el1_dbg(struct pt_regs *regs) { regs->lockdep_hardirqs = lockdep_hardirqs_enabled(); + this_cpu_inc(dbg_count); lockdep_hardirqs_off(CALLER_ADDR0); rcu_nmi_enter(); @@ -172,6 +179,7 @@ static void noinstr arm64_exit_el1_dbg(struct pt_regs *regs) rcu_nmi_exit(); if (restore) lockdep_hardirqs_on(CALLER_ADDR0); + this_cpu_dec(dbg_count); } static void noinstr el1_dbg(struct pt_regs *regs, unsigned long esr) diff --git a/include/linux/preempt.h b/include/linux/preempt.h index 7d9c1c0e149c..89d5281e060a 100644 --- a/include/linux/preempt.h +++ b/include/linux/preempt.h @@ -352,4 +352,6 @@ static __always_inline void migrate_enable(void) preempt_enable(); } +int in_dbg(void); + #endif /* __LINUX_PREEMPT_H */ diff --git a/kernel/panic.c b/kernel/panic.c index d991c3b1b559..9dadd3c5df57 100644 --- a/kernel/panic.c +++ b/kernel/panic.c @@ -761,3 +761,8 @@ static int __init panic_on_taint_setup(char *s) return 0; } early_param("panic_on_taint", panic_on_taint_setup); + +int __weak in_dbg(void) +{ + return 0; +} diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c index 0142855482f8..18b9c74474ec 100644 --- a/kernel/rcu/tree.c +++ b/kernel/rcu/tree.c @@ -957,7 +957,7 @@ void __rcu_irq_enter_check_tick(void) struct rcu_data *rdp = this_cpu_ptr(&rcu_data); // If we're here from NMI there's nothing to do. - if (in_nmi()) + if (in_nmi() || in_dbg()) return; RCU_LOCKDEP_WARN(rcu_dynticks_curr_cpu_in_eqs(), -- 2.25.1

1 0

openEuler Kernel SIG双周例会
by openEuler conference 14 Jun '23

14 Jun '23

您好！ Kernel SIG 邀请您参加 2023-06-16 14:00 召开的Zoom会议(自动录制) 会议主题：openEuler Kernel SIG双周例会会议内容： 1. 进展update 2. 议题征集中欢迎大家积极申报议题（新增议题可以直接回复邮件，或录入会议看板）会议链接：https://us06web.zoom.us/j/88482572951?pwd=dXV5bFh4Nyt2Z1VpN0NOL3hGSWVHUT09 会议纪要：https://etherpad.openeuler.org/p/Kernel-meetings 温馨提醒：建议接入会议后修改参会人的姓名，也可以使用您在gitee.com的ID 更多资讯尽在：https://openeuler.org/zh/ Hello! openEuler Kernel SIG invites you to attend the Zoom conference(auto recording) will be held at 2023-06-16 14:00, The subject of the conference is openEuler Kernel SIG双周例会, Summary: 1. 进展update 2. 议题征集中欢迎大家积极申报议题（新增议题可以直接回复邮件，或录入会议看板） You can join the meeting at https://us06web.zoom.us/j/88482572951?pwd=dXV5bFh4Nyt2Z1VpN0NOL3hGSWVHUT09. Add topics at https://etherpad.openeuler.org/p/Kernel-meetings. Note: You are advised to change the participant name after joining the conference or use your ID at gitee.com. More information: https://openeuler.org/en/

1 0

[PATCH openEuler-22.03-LTS-SP2 v4 0/2] add one bpf helper function for redissockmap
by Liu Jian 13 Jun '23

13 Jun '23

add one bpf helper function for redissockmap Liu Jian (2): net: add bpf_is_local_ipaddr bpf helper function net: let sockops can use bpf_get_current_comm() arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + drivers/net/Kconfig | 8 ++ drivers/net/Makefile | 1 + drivers/net/localip/Makefile | 8 ++ drivers/net/localip/localip.c | 157 +++++++++++++++++++++++++ include/trace/events/net.h | 4 + include/uapi/linux/bpf.h | 7 ++ net/core/filter.c | 20 ++++ net/core/net-traces.c | 1 + tools/include/uapi/linux/bpf.h | 7 ++ 11 files changed, 215 insertions(+) create mode 100644 drivers/net/localip/Makefile create mode 100644 drivers/net/localip/localip.c -- 2.34.1

1 2

[PATCH openEuler-22.03-LTS-SP2 v3 0/2] add two bpf helper function for redissockmap
by Liu Jian 13 Jun '23

13 Jun '23

v2->v3: change function pointer to tracepoint. add two bpf helper function for redissockmap Liu Jian (2): net: add bpf_is_local_ipaddr bpf helper function net: let sockops can use bpf_get_current_comm() arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + drivers/net/Kconfig | 8 ++ drivers/net/Makefile | 1 + drivers/net/localip/Makefile | 8 ++ drivers/net/localip/localip.c | 157 +++++++++++++++++++++++++ include/trace/events/net.h | 17 +++ include/uapi/linux/bpf.h | 7 ++ net/core/filter.c | 21 ++++ tools/include/uapi/linux/bpf.h | 7 ++ 10 files changed, 228 insertions(+) create mode 100644 drivers/net/localip/Makefile create mode 100644 drivers/net/localip/localip.c -- 2.34.1

1 2

[PATCH openEuler-22.03-LTS-SP2 v6 0/2] Add vhost net polling for high frequency TX mode
by Xu Kuohai 13 Jun '23

13 Jun '23

During high-frequency packet transmission, if vhost immediately notifies virtio driver after reading the virtio_queue empty, virtio driver will perform a kick after sending the next packet. However, if vhost waits a little longer, it could get the next packet sent by virtio driver without the need for virtio to perform a kick. This patch optimizes for this issue. If the TX interval recently recorded by vhost-net is within 50us, it is considered a high-frequency packet sending. At this time, after detecting the virtio_queue is empty, vhost-net will wait for new packets to arrive before notifing the virtio driver. Xu Kuohai (2): vhost_net: Suppress kick ratio when high frequency TX detected openeuler: vhost_net: Enable vhost net polling for openeuler arm64 and x86 arch/arm64/configs/openeuler_defconfig | 2 + arch/x86/configs/openeuler_defconfig | 2 + drivers/vhost/Kconfig | 22 +++++++++ drivers/vhost/net.c | 66 ++++++++++++++++++++++++++ 4 files changed, 92 insertions(+) -- 2.30.2

2 3

[PATCH openEuler-22.03-LTS-SP2 0/1] ext4: delete redundant uptodate
by Zheng Zengkai 13 Jun '23

13 Jun '23

The buffer uptodate state has been checked in function set_buffer_uptodate, there is no need use buffer_uptodate before calling set_buffer_uptodate and delete it. Yang Guo (1): ext4: delete redundant uptodate check for buffer fs/ext4/inode.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) -- 2.20.1

2 2

[PATCH openEuler-22.03-LTS-SP2 0/1] ext4: delete redundant uptodate
by Zheng Zengkai 13 Jun '23

13 Jun '23

The buffer uptodate state has been checked in function set_buffer_uptodate, there is no need use buffer_uptodate before calling set_buffer_uptodate and delete it. Yang Guo (1): ext4: delete redundant uptodate check for buffer fs/ext4/inode.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) -- 2.20.1

2 2

2024

2023

2022

2021

2020

2019

Kernel June 2023