hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I9PVOZ

--------------------------------

Found a deadlock issue in mpam which is already fixed in x86 resctrl by commit 334b0f4e9b1b ("x86/resctrl: Fix a deadlock due to inaccurate reference").

There is a race condition which results in a deadlock when rmdir and mkdir concurrently:

Thread 1: rmdir /sys/fs/resctrl/p1 Thread 2: mkdir /sys/fs/resctrl/p1/mon_groups/m1

Thread 1 is deleting control group "p1". Holding rdtgroup_mutex, kernfs_remove() removes all kernfs nodes under directory "p1" recursively, then waits for sub kernfs node "mon_groups" to drop active reference.

Thread 2 is trying to create a subdirectory "m1" in the "mon_groups" directory. The wrapper kernfs_iop_mkdir() takes an active reference to the "mon_groups" directory but the code drops the active reference to the parent directory "p1" instead.

As a result, Thread 1 is blocked on waiting for active reference to drop and never release rdtgroup_mutex, while Thread 2 is also blocked on trying to get rdtgroup_mutex.

Thread 1 (rdtgroup_rmdir) Thread 2 (rdtgroup_mkdir) (rmdir /sys/fs/resctrl/p1) (mkdir /sys/fs/resctrl/p1/mon_groups/m1) ------------------------- ------------------------- kernfs_iop_mkdir /* * kn: "m1", parent_kn: "mon_groups", * prgrp_kn: parent_kn->parent: "p1", * * "mon_groups", parent_kn->active++: 1 */ kernfs_get_active(parent_kn) kernfs_iop_rmdir /* "p1", kn->active++ */ kernfs_get_active(kn)

rdtgroup_kn_lock_live atomic_inc(&rdtgrp->waitcount) /* "p1", kn->active-- */ kernfs_break_active_protection(kn) mutex_lock

rdtgroup_rmdir_ctrl free_all_child_rdtgrp sentry->flags = RDT_DELETED

rdtgroup_ctrl_remove rdtgrp->flags = RDT_DELETED kernfs_get(kn) kernfs_remove(rdtgrp->kn) __kernfs_remove /* "mon_groups", sub_kn */ atomic_add(KN_DEACTIVATED_BIAS, &sub_kn->active) kernfs_drain(sub_kn) /* * sub_kn->active == KN_DEACTIVATED_BIAS + 1, * waiting on sub_kn->active to drop, but it * never drops in Thread 2 which is blocked * on getting rdtgroup_mutex. */ Thread 1 hangs here ----> wait_event(sub_kn->active == KN_DEACTIVATED_BIAS) ... rdtgroup_mkdir rdtgroup_mkdir_mon(parent_kn, prgrp_kn) mkdir_rdt_prepare(parent_kn, prgrp_kn) rdtgroup_kn_lock_live(prgrp_kn) atomic_inc(&rdtgrp->waitcount) /* * "p1", prgrp_kn->active-- * * The active reference on "p1" is * dropped, but not matching the * actual active reference taken * on "mon_groups", thus causing * Thread 1 to wait forever while * holding rdtgroup_mutex. */ kernfs_break_active_protection( prgrp_kn) /* * Trying to get rdtgroup_mutex * which is held by Thread 1. */ Thread 2 hangs here ----> mutex_lock ...

The problem is that the creation of a subdirectory in the "mon_groups" directory incorrectly releases the active protection of its parent directory instead of itself before it starts waiting for rdtgroup_mutex. This is triggered by the rdtgroup_mkdir() flow calling rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() with kernfs node of the parent control group ("p1") as argument. It should be called with kernfs node "mon_groups" instead. What is currently missing is that the kn->priv of "mon_groups" is NULL instead of pointing to the rdtgrp.

Fix it by pointing kn->priv to rdtgrp when "mon_groups" is created. Then it could be passed to rdtgroup_kn_lock_live()/rdtgroup_kn_unlock() instead. And then it operates on the same rdtgroup structure but handles the active reference of kernfs node "mon_groups" to prevent deadlock. The same changes are also made to the "mon_data" directories.

This patch involved associating the private data of the "mon_groups" and "mon_data" directories to the resource group to which they belong instead of NULL as before.

Improving the directory removal safeguard to ensure that subdirectories of the resctrl root directory can only be removed if they are a child of the resctrl filesystem's root _and_ not associated with the default resource group.

This patch results in some unused function parameters that will be cleaned up in follow-up patch, this patch focuses on the fix only in support of backport efforts.

Fixes: 6065f5513f78 ("resctrlfs: init support resctrlfs") Signed-off-by: Yu Liao liaoyu15@huawei.com --- fs/resctrlfs.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-)

diff --git a/fs/resctrlfs.c b/fs/resctrlfs.c index 1bb1485d4440..4cf0a4aad386 100644 --- a/fs/resctrlfs.c +++ b/fs/resctrlfs.c @@ -387,7 +387,7 @@ static int resctrl_get_tree(struct fs_context *fc)

if (resctrl_mon_capable) { ret = mongroup_create_dir(resctrl_group_default.kn, - NULL, "mon_groups", + &resctrl_group_default, "mon_groups", &kn_mongrp); if (ret) goto out_info; @@ -718,7 +718,7 @@ static int mkdir_resctrl_prepare(struct kernfs_node *parent_kn, uint files = 0; int ret;

- prdtgrp = resctrl_group_kn_lock_live(prgrp_kn); + prdtgrp = resctrl_group_kn_lock_live(parent_kn); rdt_last_cmd_clear(); if (!prdtgrp) { ret = -ENODEV; @@ -807,7 +807,7 @@ static int mkdir_resctrl_prepare(struct kernfs_node *parent_kn, kernfs_activate(kn);

/* - * The caller unlocks the prgrp_kn upon success. + * The caller unlocks the parent_kn upon success. */ return 0;

@@ -824,7 +824,7 @@ static int mkdir_resctrl_prepare(struct kernfs_node *parent_kn, out_free_rdtgrp: kfree(rdtgrp); out_unlock: - resctrl_group_kn_unlock(prgrp_kn); + resctrl_group_kn_unlock(parent_kn); return ret; }

@@ -865,7 +865,7 @@ static int resctrl_group_mkdir_mon(struct kernfs_node *parent_kn, */ ret = resctrl_update_groups_config(prgrp);

- resctrl_group_kn_unlock(prgrp_kn); + resctrl_group_kn_unlock(parent_kn); return ret; }

@@ -899,7 +899,7 @@ static int resctrl_group_mkdir_ctrl_mon(struct kernfs_node *parent_kn, * Create an empty mon_groups directory to hold the subset * of tasks and cpus to monitor. */ - ret = mongroup_create_dir(kn, NULL, "mon_groups", NULL); + ret = mongroup_create_dir(kn, rdtgrp, "mon_groups", NULL); if (ret) { rdt_last_cmd_puts("kernfs subdir error\n"); goto out_list_del; @@ -913,7 +913,7 @@ static int resctrl_group_mkdir_ctrl_mon(struct kernfs_node *parent_kn, out_common_fail: mkdir_resctrl_prepare_clean(rdtgrp); out_unlock: - resctrl_group_kn_unlock(prgrp_kn); + resctrl_group_kn_unlock(parent_kn); return ret; }

@@ -1072,7 +1072,8 @@ static int resctrl_group_rmdir(struct kernfs_node *kn) * If the resctrl_group is a mon group and parent directory * is a valid "mon_groups" directory, remove the mon group. */ - if (rdtgrp->type == RDTCTRL_GROUP && parent_kn == resctrl_group_default.kn) + if (rdtgrp->type == RDTCTRL_GROUP && parent_kn == resctrl_group_default.kn && + rdtgrp != &resctrl_group_default) ret = resctrl_group_rmdir_ctrl(kn, rdtgrp, tmpmask); else if (rdtgrp->type == RDTMON_GROUP && is_mon_groups(parent_kn, kn->name))

hulk inclusion category: cleanup bugzilla: https://gitee.com/openeuler/kernel/issues/I9NZ3E

--------------------------------

resctrl_group_rm_ctrl() is only called in resctrl_group_rmdir_ctrl(), so merge resctrl_group_rm_ctrl() and resctrl_group_rmdir_ctrl() for the following fix patch. Also do the same on resctrl_group_rm_mon() and resctrl_group_rmdir_mon().

No functional change.

Signed-off-by: Yu Liao liaoyu15@huawei.com --- fs/resctrlfs.c | 17 +++-------------- 1 file changed, 3 insertions(+), 14 deletions(-)

diff --git a/fs/resctrlfs.c b/fs/resctrlfs.c index a3354cd0e54a..dc084a03c1b2 100644 --- a/fs/resctrlfs.c +++ b/fs/resctrlfs.c @@ -954,7 +954,7 @@ static int resctrl_group_mkdir(struct kernfs_node *parent_kn, const char *name, return -EPERM; }

-static void resctrl_group_rm_mon(struct resctrl_group *rdtgrp, +static int resctrl_group_rmdir_mon(struct kernfs_node *kn, struct resctrl_group *rdtgrp, cpumask_var_t tmpmask) { struct resctrl_group *prdtgrp = rdtgrp->mon.parent; @@ -985,19 +985,14 @@ static void resctrl_group_rm_mon(struct resctrl_group *rdtgrp, */ WARN_ON(list_empty(&prdtgrp->mon.crdtgrp_list)); list_del(&rdtgrp->mon.crdtgrp_list); -} - -static int resctrl_group_rmdir_mon(struct kernfs_node *kn, struct resctrl_group *rdtgrp, - cpumask_var_t tmpmask) -{ - resctrl_group_rm_mon(rdtgrp, tmpmask);

kernfs_remove(rdtgrp->kn);

return 0; }

-static void resctrl_group_rm_ctrl(struct resctrl_group *rdtgrp, cpumask_var_t tmpmask) +static int resctrl_group_rmdir_ctrl(struct kernfs_node *kn, struct resctrl_group *rdtgrp, + cpumask_var_t tmpmask) { int cpu;

@@ -1033,12 +1028,6 @@ static void resctrl_group_rm_ctrl(struct resctrl_group *rdtgrp, cpumask_var_t tm free_all_child_rdtgrp(rdtgrp);

list_del(&rdtgrp->resctrl_group_list); -} - -static int resctrl_group_rmdir_ctrl(struct kernfs_node *kn, struct resctrl_group *rdtgrp, - cpumask_var_t tmpmask) -{ - resctrl_group_rm_ctrl(rdtgrp, tmpmask);

kernfs_remove(rdtgrp->kn);

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I9RXHK

--------------------------------

We observed the following stack: #0 [ffff80054c9e3af0] __switch_to at ffffa05c1ad69fbc #1 [ffff80054c9e3b40] __schedule at ffffa05c1baf4014 #2 [ffff80054c9e3ba0] __cond_resched at ffffa05c1baf4a70 #3 [ffff80054c9e3bc0] kmem_cache_alloc_trace at ffffa05c1b130584 #4 [ffff80054c9e3c20] __resctrl_group_move_task at ffffa05c1ad9a674 #5 [ffff80054c9e3c70] resctrl_group_rmid_write at ffffa05c1ad9bc40 #6 [ffff80054c9e3cd0] resctrl_group_file_write at ffffa05c1ad986e4

__resctrl_group_move_task() is called by resctrl_group_rmid_write() in atomic context where sleep is prohibited.

Ensure calling context won't sleep by changing allocation mode from GFP_KERNEL to GFP_NOWAIT.

Fixes: e37caef15c18 ("resctrlfs: mpam: Build basic framework for mpam") Signed-off-by: Yu Liao liaoyu15@huawei.com --- arch/arm64/kernel/mpam/mpam_resctrl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/mpam/mpam_resctrl.c b/arch/arm64/kernel/mpam/mpam_resctrl.c index e2b8e9757cdd..ade7a2724d1c 100644 --- a/arch/arm64/kernel/mpam/mpam_resctrl.c +++ b/arch/arm64/kernel/mpam/mpam_resctrl.c @@ -1365,7 +1365,7 @@ int __resctrl_group_move_task(struct task_struct *tsk, struct task_move_callback *callback; int ret;

- callback = kzalloc(sizeof(*callback), GFP_KERNEL); + callback = kzalloc(sizeof(*callback), GFP_NOWAIT); if (!callback) return -ENOMEM; callback->work.func = move_myself;

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I9SXQ8

--------------------------------

resctrl_dom_mon_data() call do_device_sync() wait for other CPU to respond IPI, which is expensive (up to a few ms). resctrl_dom_mon_data() may be called hundreds of times by resctrl_group_mondata_show(), causing soft lockup.

Fix this by adding a cond_resched between each loop.

Do the same for resctrl_group_update_domain_ctrls().

Fixes: 8d7469cd8d3e ("arm64/mpam: Integrate monitor data for Memory Bandwidth if cdp enabled") Signed-off-by: Yu Liao liaoyu15@huawei.com --- arch/arm64/kernel/mpam/mpam_ctrlmon.c | 3 +++ 1 file changed, 3 insertions(+)

diff --git a/arch/arm64/kernel/mpam/mpam_ctrlmon.c b/arch/arm64/kernel/mpam/mpam_ctrlmon.c index 0f8bcf9f4d99..d1a7b07114ab 100644 --- a/arch/arm64/kernel/mpam/mpam_ctrlmon.c +++ b/arch/arm64/kernel/mpam/mpam_ctrlmon.c @@ -285,6 +285,7 @@ static void resctrl_group_update_domain_ctrls(struct rdtgroup *rdtgrp, resctrl_cdp_mpamid_map_val(entry->closid.reqpartid, cfg[i].conf_type, closid.reqpartid); resctrl_dom_ctrl_config(cdp_both_ctrl, r, dom, &para); + cond_resched(); } } } @@ -658,6 +659,8 @@ int resctrl_group_mondata_show(struct seq_file *m, void *arg) r->rid), type, md.u.mon);

usage += resctrl_dom_mon_data(r, d, md.priv); + + cond_resched(); } }