| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <CAJuCfpEF+voHHiFd1vjdPaXRpDj2H6WEgRvevM2dBQqBdrosSg@mail.gmail.com>
Date: Mon, 20 Mar 2023 16:17:46 -0700
From: Suren Baghdasaryan <surenb@...gle.com>
To: Domenico Cerasuolo <cerasuolodomenico@...il.com>
Cc: linux-kernel@...r.kernel.org, peterz@...radead.org,
brauner@...nel.org, chris@...isdown.name, hannes@...xchg.org
Subject: Re: [PATCH 4/4] sched/psi: allow unprivileged polling of N*2s period
On Thu, Mar 9, 2023 at 9:08 AM Domenico Cerasuolo
<cerasuolodomenico@...il.com> wrote:
>
> PSI offers 2 mechanisms to get information about a specific resource
> pressure. One is reading from /proc/pressure/<resource>, which gives
> average pressures aggregated every 2s. The other is creating a pollable
> fd for a specific resource and cgroup.
>
> The trigger creation requires CAP_SYS_RESOURCE, and gives the
> possibility to pick specific time window and threshold, spawing an RT
> thread to aggregate the data.
>
> Systemd would like to provide containers the option to monitor pressure
> on their own cgroup and sub-cgroups. For example, if systemd launches a
> container that itself then launches services, the container should have
> the ability to poll() for pressure in individual services. But neither
> the container nor the services are privileged.
>
> This patch implements a mechanism to allow unprivileged users to create
> pressure triggers. The difference with privileged triggers creation is
> that unprivileged ones must have a time window that's a multiple of 2s.
> This is so that we can avoid unrestricted spawning of rt threads, and
> use instead the same aggregation mechanism done for the averages, which
> runs indipendently of any triggers.
s/indipendently/independently
>
> Suggested-by: Johannes Weiner <hannes@...xchg.org>
> Signed-off-by: Domenico Cerasuolo <cerasuolodomenico@...il.com>
> ---
> Documentation/accounting/psi.rst | 4 ++
> include/linux/psi.h | 2 +-
> include/linux/psi_types.h | 7 +++
> kernel/cgroup/cgroup.c | 2 +-
> kernel/sched/psi.c | 105 ++++++++++++++++++++-----------
> 5 files changed, 83 insertions(+), 37 deletions(-)
>
> diff --git a/Documentation/accounting/psi.rst b/Documentation/accounting/psi.rst
> index 5e40b3f437f9..df6062eb3abb 100644
> --- a/Documentation/accounting/psi.rst
> +++ b/Documentation/accounting/psi.rst
> @@ -105,6 +105,10 @@ prevent overly frequent polling. Max limit is chosen as a high enough number
> after which monitors are most likely not needed and psi averages can be used
> instead.
>
> +Unprivileged users can also create monitors, with the only limitation that the
> +window size must be a multiple of 2s, in order to prevent excessive resource
> +usage.
> +
> When activated, psi monitor stays active for at least the duration of one
> tracking window to avoid repeated activations/deactivations when system is
> bouncing in and out of the stall state.
> diff --git a/include/linux/psi.h b/include/linux/psi.h
> index b029a847def1..ab26200c2803 100644
> --- a/include/linux/psi.h
> +++ b/include/linux/psi.h
> @@ -24,7 +24,7 @@ void psi_memstall_leave(unsigned long *flags);
>
> int psi_show(struct seq_file *s, struct psi_group *group, enum psi_res res);
> struct psi_trigger *psi_trigger_create(struct psi_group *group,
> - char *buf, enum psi_res res);
> + char *buf, enum psi_res res, struct file *file);
> void psi_trigger_destroy(struct psi_trigger *t);
>
> __poll_t psi_trigger_poll(void **trigger_ptr, struct file *file,
> diff --git a/include/linux/psi_types.h b/include/linux/psi_types.h
> index 1819afa8b198..e439f411c23b 100644
> --- a/include/linux/psi_types.h
> +++ b/include/linux/psi_types.h
> @@ -151,6 +151,9 @@ struct psi_trigger {
>
> /* Deferred event(s) from previous ratelimit window */
> bool pending_event;
> +
> + /* Used to differentiate destruction action*/
> + enum psi_aggregators aggregator;
> };
>
> struct psi_group {
> @@ -171,6 +174,10 @@ struct psi_group {
> /* Aggregator work control */
> struct delayed_work avgs_work;
>
> + /* Unprivileged triggers against N*PSI_FREQ windows */
> + struct list_head triggers;
> + u32 nr_triggers[NR_PSI_STATES - 1];
> +
> /* Total stall times and sampled pressure averages */
> u64 total[NR_PSI_AGGREGATORS][NR_PSI_STATES - 1];
> unsigned long avg[NR_PSI_STATES - 1][3];
> diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
> index 935e8121b21e..dead36969bba 100644
> --- a/kernel/cgroup/cgroup.c
> +++ b/kernel/cgroup/cgroup.c
> @@ -3761,7 +3761,7 @@ static ssize_t pressure_write(struct kernfs_open_file *of, char *buf,
> }
>
> psi = cgroup_psi(cgrp);
> - new = psi_trigger_create(psi, buf, res);
> + new = psi_trigger_create(psi, buf, res, of->file);
> if (IS_ERR(new)) {
> cgroup_put(cgrp);
> return PTR_ERR(new);
> diff --git a/kernel/sched/psi.c b/kernel/sched/psi.c
> index 476941c1cbea..fde91aa4e55f 100644
> --- a/kernel/sched/psi.c
> +++ b/kernel/sched/psi.c
> @@ -186,9 +186,9 @@ static void group_init(struct psi_group *group)
> seqcount_init(&per_cpu_ptr(group->pcpu, cpu)->seq);
> group->avg_last_update = sched_clock();
> group->avg_next_update = group->avg_last_update + psi_period;
> - INIT_DELAYED_WORK(&group->avgs_work, psi_avgs_work);
> mutex_init(&group->avgs_lock);
> - /* Init trigger-related members */
> +
> + /* Init rtpoll trigger-related members */
> atomic_set(&group->rtpoll_scheduled, 0);
> mutex_init(&group->rtpoll_trigger_lock);
> INIT_LIST_HEAD(&group->rtpoll_triggers);
> @@ -197,6 +197,11 @@ static void group_init(struct psi_group *group)
> init_waitqueue_head(&group->rtpoll_wait);
> timer_setup(&group->rtpoll_timer, poll_timer_fn, 0);
> rcu_assign_pointer(group->rtpoll_task, NULL);
> +
> + /* Init avg trigger-related members */
> + INIT_LIST_HEAD(&group->triggers);
> + memset(group->nr_triggers, 0, sizeof(group->nr_triggers));
> + INIT_DELAYED_WORK(&group->avgs_work, psi_avgs_work);
> }
>
> void __init psi_init(void)
> @@ -430,20 +435,23 @@ static u64 window_update(struct psi_window *win, u64 now, u64 value)
> return growth;
> }
>
> -static u64 update_triggers(struct psi_group *group, u64 now)
> +static u64 update_triggers(struct psi_group *group, u64 now, enum psi_aggregators aggregator)
> {
> struct psi_trigger *t;
> - u64 *total = group->total[PSI_POLL];
> + u64 *total = group->total[aggregator];
> + struct list_head *triggers = aggregator == PSI_AVGS ? &group->triggers
> + : &group->rtpoll_triggers;
> + u64 *aggregator_total = aggregator == PSI_AVGS ? group->avg_total : group->rtpoll_total;
I think if you follow my suggestion in patch#2 you can do:
struct trigger_info* tinfo = group->trig_info[aggregator];
and use that instead of the above if/else conditions.
>
> /*
> * On subsequent updates, calculate growth deltas and let
> * watchers know when their specified thresholds are exceeded.
> */
> - list_for_each_entry(t, &group->rtpoll_triggers, node) {
> + list_for_each_entry(t, triggers, node) {
> u64 growth;
> bool new_stall;
>
> - new_stall = group->rtpoll_total[t->state] != total[t->state];
> + new_stall = aggregator_total[t->state] != total[t->state];
>
> /* Check for stall activity or a previous threshold breach */
> if (!new_stall && !t->pending_event)
> @@ -553,8 +561,10 @@ static void psi_avgs_work(struct work_struct *work)
> * Once restarted, we'll catch up the running averages in one
> * go - see calc_avgs() and missed_periods.
> */
> - if (now >= group->avg_next_update)
> + if (now >= group->avg_next_update) {
> + update_triggers(group, now, PSI_AVGS);
> group->avg_next_update = update_averages(group, now);
> + }
>
> if (changed_states & PSI_STATE_RESCHEDULE) {
> schedule_delayed_work(dwork, nsecs_to_jiffies(
> @@ -571,9 +581,17 @@ static void init_triggers(struct psi_group *group, u64 now)
> list_for_each_entry(t, &group->rtpoll_triggers, node)
> window_reset(&t->win, now,
> group->total[PSI_POLL][t->state], 0);
> +
> + list_for_each_entry(t, &group->triggers, node)
> + window_reset(&t->win, now,
> + group->total[PSI_AVGS][t->state], 0);
> +
> memcpy(group->rtpoll_total, group->total[PSI_POLL],
> sizeof(group->rtpoll_total));
> group->rtpoll_next_update = now + group->rtpoll_min_period;
> +
> + memcpy(group->avg_total, group->total[PSI_AVGS],
> + sizeof(group->avg_total));
> }
>
> /* Schedule polling if it's not already scheduled or forced. */
> @@ -673,7 +691,7 @@ static void psi_rtpoll_work(struct psi_group *group)
> }
>
> if (now >= group->rtpoll_next_update) {
> - group->rtpoll_next_update = update_triggers(group, now);
> + group->rtpoll_next_update = update_triggers(group, now, PSI_POLL);
> if (changed_states & group->rtpoll_states)
> memcpy(group->rtpoll_total, group->total[PSI_POLL],
> sizeof(group->rtpoll_total));
> @@ -1243,16 +1261,19 @@ int psi_show(struct seq_file *m, struct psi_group *group, enum psi_res res)
> }
>
> struct psi_trigger *psi_trigger_create(struct psi_group *group,
> - char *buf, enum psi_res res)
> + char *buf, enum psi_res res, struct file *file)
> {
> struct psi_trigger *t;
> enum psi_states state;
> u32 threshold_us;
> + bool privileged;
> u32 window_us;
>
> if (static_branch_likely(&psi_disabled))
> return ERR_PTR(-EOPNOTSUPP);
>
> + privileged = cap_raised(file->f_cred->cap_effective, CAP_SYS_RESOURCE);
> +
> if (sscanf(buf, "some %u %u", &threshold_us, &window_us) == 2)
> state = PSI_IO_SOME + res * 2;
> else if (sscanf(buf, "full %u %u", &threshold_us, &window_us) == 2)
> @@ -1272,6 +1293,13 @@ struct psi_trigger *psi_trigger_create(struct psi_group *group,
> window_us > WINDOW_MAX_US)
> return ERR_PTR(-EINVAL);
>
> + /*
> + * Unprivileged users can only use 2s windows so that averages aggregation
> + * work is used, and no RT threads need to be spawned.
> + */
> + if (!privileged && window_us % 2000000)
> + return ERR_PTR(-EINVAL);
> +
> /* Check threshold */
> if (threshold_us == 0 || threshold_us > window_us)
> return ERR_PTR(-EINVAL);
> @@ -1291,10 +1319,11 @@ struct psi_trigger *psi_trigger_create(struct psi_group *group,
> t->last_event_time = 0;
> init_waitqueue_head(&t->event_wait);
> t->pending_event = false;
> + t->aggregator = privileged ? PSI_POLL : PSI_AVGS;
>
> mutex_lock(&group->rtpoll_trigger_lock);
>
> - if (!rcu_access_pointer(group->rtpoll_task)) {
> + if (privileged && !rcu_access_pointer(group->rtpoll_task)) {
> struct task_struct *task;
>
> task = kthread_create(psi_rtpoll_worker, group, "psimon");
> @@ -1308,11 +1337,16 @@ struct psi_trigger *psi_trigger_create(struct psi_group *group,
> rcu_assign_pointer(group->rtpoll_task, task);
> }
>
> - list_add(&t->node, &group->rtpoll_triggers);
> - group->rtpoll_min_period = min(group->rtpoll_min_period,
> - div_u64(t->win.size, UPDATES_PER_WINDOW));
> - group->rtpoll_nr_triggers[t->state]++;
> - group->rtpoll_states |= (1 << t->state);
> + if (privileged) {
> + list_add(&t->node, &group->rtpoll_triggers);
> + group->rtpoll_min_period = min(group->rtpoll_min_period,
> + div_u64(t->win.size, UPDATES_PER_WINDOW));
> + group->rtpoll_nr_triggers[t->state]++;
> + group->rtpoll_states |= (1 << t->state);
> + } else {
> + list_add(&t->node, &group->triggers);
> + group->nr_triggers[t->state]++;
> + }
>
> mutex_unlock(&group->rtpoll_trigger_lock);
>
> @@ -1346,22 +1380,26 @@ void psi_trigger_destroy(struct psi_trigger *t)
> u64 period = ULLONG_MAX;
>
> list_del(&t->node);
> - group->rtpoll_nr_triggers[t->state]--;
> - if (!group->rtpoll_nr_triggers[t->state])
> - group->rtpoll_states &= ~(1 << t->state);
> - /* reset min update period for the remaining triggers */
> - list_for_each_entry(tmp, &group->rtpoll_triggers, node)
> - period = min(period, div_u64(tmp->win.size,
> - UPDATES_PER_WINDOW));
> - group->rtpoll_min_period = period;
> - /* Destroy rtpoll_task when the last trigger is destroyed */
> - if (group->rtpoll_states == 0) {
> - group->rtpoll_until = 0;
> - task_to_destroy = rcu_dereference_protected(
> - group->rtpoll_task,
> - lockdep_is_held(&group->rtpoll_trigger_lock));
> - rcu_assign_pointer(group->rtpoll_task, NULL);
> - del_timer(&group->rtpoll_timer);
> + if (t->aggregator == PSI_AVGS) {
> + group->nr_triggers[t->state]--;
> + } else {
> + group->rtpoll_nr_triggers[t->state]--;
Same here. This can be:
group->trig_info[t->aggregator].nr_triggers[t->state]--;
if (t->aggregator == PSI_AVGS)
return;
// original code for PSI_POLL triggers
> + if (!group->rtpoll_nr_triggers[t->state])
> + group->rtpoll_states &= ~(1 << t->state);
> + /* reset min update period for the remaining triggers */
> + list_for_each_entry(tmp, &group->rtpoll_triggers, node)
> + period = min(period, div_u64(tmp->win.size,
> + UPDATES_PER_WINDOW));
> + group->rtpoll_min_period = period;
> + /* Destroy rtpoll_task when the last trigger is destroyed */
> + if (group->rtpoll_states == 0) {
> + group->rtpoll_until = 0;
> + task_to_destroy = rcu_dereference_protected(
> + group->rtpoll_task,
> + lockdep_is_held(&group->rtpoll_trigger_lock));
> + rcu_assign_pointer(group->rtpoll_task, NULL);
> + del_timer(&group->rtpoll_timer);
> + }
> }
> }
>
> @@ -1428,9 +1466,6 @@ static int psi_cpu_show(struct seq_file *m, void *v)
>
> static int psi_open(struct file *file, int (*psi_show)(struct seq_file *, void *))
> {
> - if (file->f_mode & FMODE_WRITE && !capable(CAP_SYS_RESOURCE))
> - return -EPERM;
> -
> return single_open(file, psi_show, NULL);
> }
Here you can eliminate psi_open() completely and use single_open() directly.
>
> @@ -1480,7 +1515,7 @@ static ssize_t psi_write(struct file *file, const char __user *user_buf,
> return -EBUSY;
> }
>
> - new = psi_trigger_create(&psi_system, buf, res);
> + new = psi_trigger_create(&psi_system, buf, res, file);
> if (IS_ERR(new)) {
> mutex_unlock(&seq->lock);
> return PTR_ERR(new);
> --
> 2.34.1
>
Powered by blists - more mailing lists