- Kernel - mailweb.openeuler.org

[PATCH OLK-5.10] net: sched: fix NULL pointer dereference in mq_attach
by Zhengchao Shao 02 Jun '23

02 Jun '23

mainline inclusion from mainline-v6.4-rc5 commit 36eec020fab668719b541f34d97f44e232ffa165 category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I79V6E CVE: NA Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- When use the following command to test: 1)ip link add bond0 type bond 2)ip link set bond0 up 3)tc qdisc add dev bond0 root handle ffff: mq 4)tc qdisc replace dev bond0 parent ffff:fff1 handle ffff: mq The kernel reports NULL pointer dereference issue. The stack information is as follows: Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000 Internal error: Oops: 0000000096000006 [#1] SMP Modules linked in: pstate: 20000005 (nzCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : mq_attach+0x44/0xa0 lr : qdisc_graft+0x20c/0x5cc sp : ffff80000e2236a0 x29: ffff80000e2236a0 x28: ffff0000c0e59d80 x27: ffff0000c0be19c0 x26: ffff0000cae3e800 x25: 0000000000000010 x24: 00000000fffffff1 x23: 0000000000000000 x22: ffff0000cae3e800 x21: ffff0000c9df4000 x20: ffff0000c9df4000 x19: 0000000000000000 x18: ffff80000a934000 x17: ffff8000f5b56000 x16: ffff80000bb08000 x15: 0000000000000000 x14: 0000000000000000 x13: 6b6b6b6b6b6b6b6b x12: 6b6b6b6b00000001 x11: 0000000000000000 x10: 0000000000000000 x9 : 0000000000000000 x8 : ffff0000c0be0730 x7 : bbbbbbbbbbbbbbbb x6 : 0000000000000008 x5 : ffff0000cae3e864 x4 : 0000000000000000 x3 : 0000000000000001 x2 : 0000000000000001 x1 : ffff8000090bc23c x0 : 0000000000000000 Call trace: mq_attach+0x44/0xa0 qdisc_graft+0x20c/0x5cc tc_modify_qdisc+0x1c4/0x664 rtnetlink_rcv_msg+0x354/0x440 netlink_rcv_skb+0x64/0x144 rtnetlink_rcv+0x28/0x34 netlink_unicast+0x1e8/0x2a4 netlink_sendmsg+0x308/0x4a0 sock_sendmsg+0x64/0xac ____sys_sendmsg+0x29c/0x358 ___sys_sendmsg+0x90/0xd0 __sys_sendmsg+0x7c/0xd0 __arm64_sys_sendmsg+0x2c/0x38 invoke_syscall+0x54/0x114 el0_svc_common.constprop.1+0x90/0x174 do_el0_svc+0x3c/0xb0 el0_svc+0x24/0xec el0t_64_sync_handler+0x90/0xb4 el0t_64_sync+0x174/0x178 This is because when mq is added for the first time, qdiscs in mq is set to NULL in mq_attach(). Therefore, when replacing mq after adding mq, we need to initialize qdiscs in the mq before continuing to graft. Otherwise, it will couse NULL pointer dereference issue in mq_attach(). And the same issue will occur in the attach functions of mqprio, taprio and htb. ffff:fff1 means that the repalce qdisc is ingress. Ingress does not allow any qdisc to be attached. Therefore, ffff:fff1 is incorrectly used, and the command should be dropped. Fixes: 6ec1c69a8f64 ("net_sched: add classful multiqueue dummy scheduler") Signed-off-by: Zhengchao Shao <shaozhengchao(a)huawei.com> Tested-by: Peilin Ye <peilin.ye(a)bytedance.com> Acked-by: Jamal Hadi Salim <jhs(a)mojatatu.com> Link: https://lore.kernel.org/r/20230527093747.3583502-1-shaozhengchao@huawei.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> --- net/sched/sch_api.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c index b12a3502fca5..b8eecd6704cb 100644 --- a/net/sched/sch_api.c +++ b/net/sched/sch_api.c @@ -1590,6 +1590,10 @@ static int tc_modify_qdisc(struct sk_buff *skb, struct nlmsghdr *n, NL_SET_ERR_MSG(extack, "Qdisc parent/child loop detected"); return -ELOOP; } + if (clid == TC_H_INGRESS) { + NL_SET_ERR_MSG(extack, "Ingress cannot graft directly"); + return -EINVAL; + } qdisc_refcount_inc(q); goto graft; } else { -- 2.34.1

2 1

[openeuler/rdma-core v2 0/4] src-openeuler/rdma-core:bugfixes
by Chengchang Tang 02 Jun '23

02 Jun '23

From: Juan Zhou <zhoujuan51(a)h-partners.com> Chengchang Tang (1): libhns: Add a judgment to the congestion control algorithm Haoyue Xu (1): libhns: Fix arrangement in SL to adapt MPI APP Junxian Huang (1): libhns: Fix incorrect post-send with direct wqe of wr-list in user space Luoyouming (1): libhns: Fix the owner bit error of sq in new io libibverbs/man/ibv_poll_cq.3 | 5 +++++ libibverbs/verbs.h | 15 ++++++++++++++- providers/hns/hns_roce_u_hw_v2.c | 33 ++++++++++++++++++++++++++------ providers/hns/hns_roce_u_verbs.c | 3 +++ 4 files changed, 49 insertions(+), 7 deletions(-) -- 2.30.0

1 4

Re: [PATCH OLK-5.10] config: enable CONFIG_BPF_STREAM_PARSER=y on arm64
by Wei Yongjun 02 Jun '23

02 Jun '23

On 2023/6/1 23:22, Liu Jian wrote: > hulk inclusion > category: feature > bugzilla: https://gitee.com/openeuler/kernel/issues/I79J3X > CVE: N/A > > ---------------------------------------------------- > > It is already enabled by default on the x86 and loongarch platforms, so > let's enable it on the arm64 platform. > > Signed-off-by: Liu Jian <liujian56(a)huawei.com> > --- > arch/arm64/configs/openeuler_defconfig | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > Reviewed-by: Wei Yongjun <weiyongjun1(a)huawe.com>

1 0

[PATCH openEuler-1.0-LTS] power: supply: bq24190: Fix use after free bug in bq24190_remove due to race condition
by Yongqiang Liu 02 Jun '23

02 Jun '23

From: Zhao Wenhui <zhaowenhui8(a)huawei.com> mainline inclusion from mainline-v6.3-rc4 commit 47c29d69212911f50bdcdd0564b5999a559010d4 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I7781Q CVE: CVE-2023-33288 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… ---------------------------------------- In bq24190_probe, &bdi->input_current_limit_work is bound with bq24190_input_current_limit_work. When external power changed, it will call bq24190_charger_external_power_changed to start the work. If we remove the module which will call bq24190_remove to make cleanup, there may be a unfinished work. The possible sequence is as follows: CPU0 CPUc1 |bq24190_input_current_limit_work bq24190_remove | power_supply_unregister | device_unregister | power_supply_dev_release| kfree(psy) | | | power_supply_get_property_from_supplier | //use Fix it by finishing the work before cleanup in the bq24190_remove Fixes: 97774672573a ("power_supply: Initialize changed_work before calling device_add") Signed-off-by: Zheng Wang <zyytlz.wz(a)163.com> Signed-off-by: Sebastian Reichel <sebastian.reichel(a)collabora.com> Conflicts: drivers/power/supply/bq24190_charger.c Signed-off-by: Zhao Wenhui <zhaowenhui8(a)huawei.com> Reviewed-by: Zhang Qiao <zhangqiao22(a)huawei.com> Reviewed-by: Chen Hui <judy.chenhui(a)huawei.com> Reviewed-by: Wang Weiyang <wangweiyang2(a)huawei.com> Signed-off-by: Yongqiang Liu <liuyongqiang13(a)huawei.com> --- drivers/power/supply/bq24190_charger.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/power/supply/bq24190_charger.c b/drivers/power/supply/bq24190_charger.c index b58df04d03b3..760feee2cad9 100644 --- a/drivers/power/supply/bq24190_charger.c +++ b/drivers/power/supply/bq24190_charger.c @@ -1821,6 +1821,7 @@ static int bq24190_remove(struct i2c_client *client) struct bq24190_dev_info *bdi = i2c_get_clientdata(client); int error; + cancel_delayed_work_sync(&bdi->input_current_limit_work); error = pm_runtime_get_sync(bdi->dev); if (error < 0) { dev_warn(bdi->dev, "pm_runtime_get failed: %i\n", error); -- 2.25.1

1 0

[PATCH OLK-5.10] sched: fix performance degradation on lmbench
by Hui Tang 01 Jun '23

01 Jun '23

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I7A718 -------------------------------- There are worse performance with the 'Fixes' when running "./lat_ctx -P $SYNC_MAX -s 64 16". The 'Fixes' which allocates memory for p->prefer_cpus even if "prefer_cpus" not be set. Before the 'Fixes', only test "p->prefer_cpus", after, add test "!cpumask_empty(p->prefer_cpus)" which causing performance degradation. select_task_rq_fair ->set_task_select_cpus ->prefer_cpus_valid ---- test cpumask_empty(p->prefer_cpus) Fixes: ebeb84ad373d ("cpuset: Introduce new interface for scheduler ...") Signed-off-by: Hui Tang <tanghui20(a)huawei.com> --- fs/proc/base.c | 6 ++++++ include/linux/sched.h | 1 + kernel/cgroup/cpuset.c | 3 +++ kernel/sched/fair.c | 24 ++++++++++++++++++++++++ 4 files changed, 34 insertions(+) diff --git a/fs/proc/base.c b/fs/proc/base.c index 5836b4b25052..22c65289128e 100644 --- a/fs/proc/base.c +++ b/fs/proc/base.c @@ -3298,6 +3298,12 @@ static ssize_t preferred_cpuset_write(struct file *file, const char __user *buf, if (retval < 0) goto out_free_cpumask; + if (!cpumask_empty(new_mask)) { + cpus_read_lock(); + dynamic_affinity_enable(); + cpus_read_unlock(); + } + retval = count; out_free_cpumask: diff --git a/include/linux/sched.h b/include/linux/sched.h index 3b8f72bf0c6b..3aae225f98a7 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2227,6 +2227,7 @@ int set_prefer_cpus_ptr(struct task_struct *p, const struct cpumask *new_mask); int sched_prefer_cpus_fork(struct task_struct *p, struct cpumask *mask); void sched_prefer_cpus_free(struct task_struct *p); +void dynamic_affinity_enable(void); #endif #ifdef CONFIG_BPF_SCHED diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index 2529a807599b..90de01cc6827 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -734,6 +734,9 @@ static int update_prefer_cpumask(struct cpuset *cs, struct cpuset *trialcs, update_tasks_prefer_cpumask(trialcs); + if (!cpumask_empty(trialcs->prefer_cpus)) + dynamic_affinity_enable(); + spin_lock_irq(&callback_lock); cpumask_copy(cs->prefer_cpus, trialcs->prefer_cpus); spin_unlock_irq(&callback_lock); diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index c7b560ffb75e..ff209d25c21c 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -7067,6 +7067,27 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) } #ifdef CONFIG_QOS_SCHED_DYNAMIC_AFFINITY + +#ifdef CONFIG_JUMP_LABEL +static DEFINE_STATIC_KEY_FALSE(__dynamic_affinity_used); + +static inline bool dynamic_affinity_used(void) +{ + return static_branch_unlikely(&__dynamic_affinity_used); +} + +void dynamic_affinity_enable(void) +{ + static_branch_enable_cpuslocked(&__dynamic_affinity_used); +} + +#else /* CONFIG_JUMP_LABEL */ +static bool dynamic_affinity_used(void) +{ + return true; +} +#endif + /* * Low utilization threshold for CPU * @@ -7076,6 +7097,9 @@ 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) && -- 2.17.1

2 1

[OLK-5.10 v2 0/7] A group of bugfixes for openeuler
by Chengchang Tang 01 Jun '23

01 Jun '23

From: Juan Zhou <zhoujuan51(a)h-partners.com> Chengchang Tang (2): RDMA/hns: fix iommu_map_sg() failed when MR bigger than 4G RDMA/hns: Fix sleeping in atomic context during DCA unloading Junxian Huang (5): RDMA/hns: Remove VF extend configuration RDMA/hns: Fix a missing check of atomic wr length RDMA/hns: Fix incorrect post-send with direct wqe of wr-list in kernel space RDMA/hns: Fix inaccurate error label name in init instance RDMA/hns: Fix a missing kfree() to hr_dev->priv drivers/infiniband/hw/hns/hns_roce_dca.c | 2 - drivers/infiniband/hw/hns/hns_roce_device.h | 3 +- drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 115 ++++++-------------- drivers/infiniband/hw/hns/hns_roce_hw_v2.h | 10 -- drivers/infiniband/hw/hns/hns_roce_main.c | 2 +- 5 files changed, 34 insertions(+), 98 deletions(-) -- 2.30.0

1 7

[PATCH openEuler-1.0-LTS] net: sched: fix NULL pointer dereference in mq_attach
by Yongqiang Liu 01 Jun '23

01 Jun '23

From: Zhengchao Shao <shaozhengchao(a)huawei.com> mainline inclusion from mainline-v6.4-rc5 commit 36eec020fab668719b541f34d97f44e232ffa165 category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I79V6E CVE: NA Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- When use the following command to test: 1)ip link add bond0 type bond 2)ip link set bond0 up 3)tc qdisc add dev bond0 root handle ffff: mq 4)tc qdisc replace dev bond0 parent ffff:fff1 handle ffff: mq The kernel reports NULL pointer dereference issue. The stack information is as follows: Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000 Internal error: Oops: 0000000096000006 [#1] SMP Modules linked in: pstate: 20000005 (nzCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc : mq_attach+0x44/0xa0 lr : qdisc_graft+0x20c/0x5cc sp : ffff80000e2236a0 x29: ffff80000e2236a0 x28: ffff0000c0e59d80 x27: ffff0000c0be19c0 x26: ffff0000cae3e800 x25: 0000000000000010 x24: 00000000fffffff1 x23: 0000000000000000 x22: ffff0000cae3e800 x21: ffff0000c9df4000 x20: ffff0000c9df4000 x19: 0000000000000000 x18: ffff80000a934000 x17: ffff8000f5b56000 x16: ffff80000bb08000 x15: 0000000000000000 x14: 0000000000000000 x13: 6b6b6b6b6b6b6b6b x12: 6b6b6b6b00000001 x11: 0000000000000000 x10: 0000000000000000 x9 : 0000000000000000 x8 : ffff0000c0be0730 x7 : bbbbbbbbbbbbbbbb x6 : 0000000000000008 x5 : ffff0000cae3e864 x4 : 0000000000000000 x3 : 0000000000000001 x2 : 0000000000000001 x1 : ffff8000090bc23c x0 : 0000000000000000 Call trace: mq_attach+0x44/0xa0 qdisc_graft+0x20c/0x5cc tc_modify_qdisc+0x1c4/0x664 rtnetlink_rcv_msg+0x354/0x440 netlink_rcv_skb+0x64/0x144 rtnetlink_rcv+0x28/0x34 netlink_unicast+0x1e8/0x2a4 netlink_sendmsg+0x308/0x4a0 sock_sendmsg+0x64/0xac ____sys_sendmsg+0x29c/0x358 ___sys_sendmsg+0x90/0xd0 __sys_sendmsg+0x7c/0xd0 __arm64_sys_sendmsg+0x2c/0x38 invoke_syscall+0x54/0x114 el0_svc_common.constprop.1+0x90/0x174 do_el0_svc+0x3c/0xb0 el0_svc+0x24/0xec el0t_64_sync_handler+0x90/0xb4 el0t_64_sync+0x174/0x178 This is because when mq is added for the first time, qdiscs in mq is set to NULL in mq_attach(). Therefore, when replacing mq after adding mq, we need to initialize qdiscs in the mq before continuing to graft. Otherwise, it will couse NULL pointer dereference issue in mq_attach(). And the same issue will occur in the attach functions of mqprio, taprio and htb. ffff:fff1 means that the repalce qdisc is ingress. Ingress does not allow any qdisc to be attached. Therefore, ffff:fff1 is incorrectly used, and the command should be dropped. Fixes: 6ec1c69a8f64 ("net_sched: add classful multiqueue dummy scheduler") Signed-off-by: Zhengchao Shao <shaozhengchao(a)huawei.com> Tested-by: Peilin Ye <peilin.ye(a)bytedance.com> Acked-by: Jamal Hadi Salim <jhs(a)mojatatu.com> Link: https://lore.kernel.org/r/20230527093747.3583502-1-shaozhengchao@huawei.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> Reviewed-by: Liu Jian <liujian56(a)huawei.com> Signed-off-by: Yongqiang Liu <liuyongqiang13(a)huawei.com> --- net/sched/sch_api.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c index c927d511bb65..448bfcdeab25 100644 --- a/net/sched/sch_api.c +++ b/net/sched/sch_api.c @@ -1514,6 +1514,10 @@ static int tc_modify_qdisc(struct sk_buff *skb, struct nlmsghdr *n, NL_SET_ERR_MSG(extack, "Qdisc parent/child loop detected"); return -ELOOP; } + if (clid == TC_H_INGRESS) { + NL_SET_ERR_MSG(extack, "Ingress cannot graft directly"); + return -EINVAL; + } qdisc_refcount_inc(q); goto graft; } else { -- 2.25.1

1 0

[PATCH OLK-5.10] sched: fix performance degradation on lmbench
by Hui Tang 01 Jun '23

01 Jun '23

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I7A718 -------------------------------- There are worse performance with the 'Fixes' when running "./lat_ctx -P $SYNC_MAX -s 64 16". The 'Fixes' which allocates memory for p->prefer_cpus even if "prefer_cpus" not be set. Before the 'Fixes', only test "p->prefer_cpus", after, add test "!cpumask_empty(p->prefer_cpus)" which causing performance degradation. select_task_rq_fair ->set_task_select_cpus ->prefer_cpus_valid ---- test cpumask_empty(p->prefer_cpus) Fixes: b48594bf6206 ("[Huawei] sched: Fix hardlockup on 'set_prefer_cpus_ptr'") Signed-off-by: Hui Tang <tanghui20(a)huawei.com> --- fs/proc/base.c | 3 +++ include/linux/sched.h | 1 + kernel/cgroup/cpuset.c | 3 +++ kernel/sched/fair.c | 24 ++++++++++++++++++++++++ 4 files changed, 31 insertions(+) diff --git a/fs/proc/base.c b/fs/proc/base.c index 5836b4b25052..03320373207f 100644 --- a/fs/proc/base.c +++ b/fs/proc/base.c @@ -3298,6 +3298,9 @@ static ssize_t preferred_cpuset_write(struct file *file, const char __user *buf, if (retval < 0) goto out_free_cpumask; + if (!cpumask_empty(new_mask)) + dynamic_affinity_enable(); + retval = count; out_free_cpumask: diff --git a/include/linux/sched.h b/include/linux/sched.h index 3b8f72bf0c6b..3aae225f98a7 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2227,6 +2227,7 @@ int set_prefer_cpus_ptr(struct task_struct *p, const struct cpumask *new_mask); int sched_prefer_cpus_fork(struct task_struct *p, struct cpumask *mask); void sched_prefer_cpus_free(struct task_struct *p); +void dynamic_affinity_enable(void); #endif #ifdef CONFIG_BPF_SCHED diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index 2529a807599b..90de01cc6827 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -734,6 +734,9 @@ static int update_prefer_cpumask(struct cpuset *cs, struct cpuset *trialcs, update_tasks_prefer_cpumask(trialcs); + if (!cpumask_empty(trialcs->prefer_cpus)) + dynamic_affinity_enable(); + spin_lock_irq(&callback_lock); cpumask_copy(cs->prefer_cpus, trialcs->prefer_cpus); spin_unlock_irq(&callback_lock); diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index c7b560ffb75e..34b344557e38 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -7067,6 +7067,27 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) } #ifdef CONFIG_QOS_SCHED_DYNAMIC_AFFINITY + +#ifdef CONFIG_JUMP_LABEL +static DEFINE_STATIC_KEY_FALSE(__dynamic_affinity_used); + +static inline bool dynamic_affinity_used(void) +{ + return static_branch_unlikely(&__dynamic_affinity_used); +} + +void dynamic_affinity_enable(void) +{ + static_branch_enable(&__dynamic_affinity_used); +} + +#else /* CONFIG_JUMP_LABEL */ +static bool dynamic_affinity_used(void) +{ + return true; +} +#endif + /* * Low utilization threshold for CPU * @@ -7076,6 +7097,9 @@ 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) && -- 2.17.1

2 1

[PATCH OLK-5.10] sched: fix performance degradation on lmbench
by Hui Tang 01 Jun '23

01 Jun '23

Offering: HULK hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I7A718 -------------------------------- There are worse performance with the 'Fixes' when running "./lat_ctx -P $SYNC_MAX -s 64 16". The 'Fixes' which allocates memory for p->prefer_cpus even if "prefer_cpus" not be set. Before the 'Fixes', only test "p->prefer_cpus", after, add test "!cpumask_empty(p->prefer_cpus)" which causing performance degradation. select_task_rq_fair ->set_task_select_cpus ->prefer_cpus_valid ---- test cpumask_empty(p->prefer_cpus) Fixes: b48594bf6206 ("[Huawei] sched: Fix hardlockup on 'set_prefer_cpus_ptr'") Signed-off-by: Hui Tang <tanghui20(a)huawei.com> --- fs/proc/base.c | 3 +++ include/linux/sched.h | 1 + kernel/cgroup/cpuset.c | 3 +++ kernel/sched/fair.c | 24 ++++++++++++++++++++++++ 4 files changed, 31 insertions(+) diff --git a/fs/proc/base.c b/fs/proc/base.c index 5836b4b25052..03320373207f 100644 --- a/fs/proc/base.c +++ b/fs/proc/base.c @@ -3298,6 +3298,9 @@ static ssize_t preferred_cpuset_write(struct file *file, const char __user *buf, if (retval < 0) goto out_free_cpumask; + if (!cpumask_empty(new_mask)) + dynamic_affinity_enable(); + retval = count; out_free_cpumask: diff --git a/include/linux/sched.h b/include/linux/sched.h index 3b8f72bf0c6b..3aae225f98a7 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2227,6 +2227,7 @@ int set_prefer_cpus_ptr(struct task_struct *p, const struct cpumask *new_mask); int sched_prefer_cpus_fork(struct task_struct *p, struct cpumask *mask); void sched_prefer_cpus_free(struct task_struct *p); +void dynamic_affinity_enable(void); #endif #ifdef CONFIG_BPF_SCHED diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index 2529a807599b..90de01cc6827 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -734,6 +734,9 @@ static int update_prefer_cpumask(struct cpuset *cs, struct cpuset *trialcs, update_tasks_prefer_cpumask(trialcs); + if (!cpumask_empty(trialcs->prefer_cpus)) + dynamic_affinity_enable(); + spin_lock_irq(&callback_lock); cpumask_copy(cs->prefer_cpus, trialcs->prefer_cpus); spin_unlock_irq(&callback_lock); diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index c7b560ffb75e..34b344557e38 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -7067,6 +7067,27 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) } #ifdef CONFIG_QOS_SCHED_DYNAMIC_AFFINITY + +#ifdef CONFIG_JUMP_LABEL +static DEFINE_STATIC_KEY_FALSE(__dynamic_affinity_used); + +static inline bool dynamic_affinity_used(void) +{ + return static_branch_unlikely(&__dynamic_affinity_used); +} + +void dynamic_affinity_enable(void) +{ + static_branch_enable(&__dynamic_affinity_used); +} + +#else /* CONFIG_JUMP_LABEL */ +static bool dynamic_affinity_used(void) +{ + return true; +} +#endif + /* * Low utilization threshold for CPU * @@ -7076,6 +7097,9 @@ 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) && -- 2.17.1

2 1

[PATCH] sched: fix performance degradation on lmbench
by Hui Tang 01 Jun '23

01 Jun '23

Offering: HULK hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I7A718 -------------------------------- There are worse performance with the 'Fixes' when running "./lat_ctx -P $SYNC_MAX -s 64 16". The 'Fixes' which allocates memory for p->prefer_cpus even if "prefer_cpus" not be set. Before the 'Fixes', only test "p->prefer_cpus", after, add test "!cpumask_empty(p->prefer_cpus)" which causing performance degradation. select_task_rq_fair ->set_task_select_cpus ->prefer_cpus_valid ---- test cpumask_empty(p->prefer_cpus) Fixes: b48594bf6206 ("[Huawei] sched: Fix hardlockup on 'set_prefer_cpus_ptr'") Signed-off-by: Hui Tang <tanghui20(a)huawei.com> --- fs/proc/base.c | 3 +++ include/linux/sched.h | 1 + kernel/cgroup/cpuset.c | 3 +++ kernel/sched/fair.c | 24 ++++++++++++++++++++++++ 4 files changed, 31 insertions(+) diff --git a/fs/proc/base.c b/fs/proc/base.c index 5836b4b25052..03320373207f 100644 --- a/fs/proc/base.c +++ b/fs/proc/base.c @@ -3298,6 +3298,9 @@ static ssize_t preferred_cpuset_write(struct file *file, const char __user *buf, if (retval < 0) goto out_free_cpumask; + if (!cpumask_empty(new_mask)) + dynamic_affinity_enable(); + retval = count; out_free_cpumask: diff --git a/include/linux/sched.h b/include/linux/sched.h index 3b8f72bf0c6b..3aae225f98a7 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2227,6 +2227,7 @@ int set_prefer_cpus_ptr(struct task_struct *p, const struct cpumask *new_mask); int sched_prefer_cpus_fork(struct task_struct *p, struct cpumask *mask); void sched_prefer_cpus_free(struct task_struct *p); +void dynamic_affinity_enable(void); #endif #ifdef CONFIG_BPF_SCHED diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c index 2529a807599b..90de01cc6827 100644 --- a/kernel/cgroup/cpuset.c +++ b/kernel/cgroup/cpuset.c @@ -734,6 +734,9 @@ static int update_prefer_cpumask(struct cpuset *cs, struct cpuset *trialcs, update_tasks_prefer_cpumask(trialcs); + if (!cpumask_empty(trialcs->prefer_cpus)) + dynamic_affinity_enable(); + spin_lock_irq(&callback_lock); cpumask_copy(cs->prefer_cpus, trialcs->prefer_cpus); spin_unlock_irq(&callback_lock); diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index c7b560ffb75e..34b344557e38 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -7067,6 +7067,27 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) } #ifdef CONFIG_QOS_SCHED_DYNAMIC_AFFINITY + +#ifdef CONFIG_JUMP_LABEL +static DEFINE_STATIC_KEY_FALSE(__dynamic_affinity_used); + +static inline bool dynamic_affinity_used(void) +{ + return static_branch_unlikely(&__dynamic_affinity_used); +} + +void dynamic_affinity_enable(void) +{ + static_branch_enable(&__dynamic_affinity_used); +} + +#else /* CONFIG_JUMP_LABEL */ +static bool dynamic_affinity_used(void) +{ + return true; +} +#endif + /* * Low utilization threshold for CPU * @@ -7076,6 +7097,9 @@ 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) && -- 2.17.1

1 0

2024

2023

2022

2021

2020

2019

Kernel