[PATCH] sched/fair: Fix CFS bandwidth hrtimer expiry type

[PATCH] sched/fair: Fix CFS bandwidth hrtimer expiry type

[Odin Ugedal]

The time remaining until expiry of the refresh_timer can be negative.
Casting the type to an unsigned 64-bit value will cause integer
underflow, making the runtime_refresh_within return false instead of
true. These situations are rare, but they do happen.

This does not cause user-facing issues or errors; other than
possibly unthrottling cfs_rq's using runtime from the previous period(s),
making the CFS bandwidth enforcement less strict in those (special)
situations.

Signed-off-by: Odin Ugedal <odin@uged.al>
---
 kernel/sched/fair.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 23663318fb81..62446c052efb 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -5108,7 +5108,7 @@ static const u64 cfs_bandwidth_slack_period = 5 * NSEC_PER_MSEC;
 static int runtime_refresh_within(struct cfs_bandwidth *cfs_b, u64 min_expire)
 {
 	struct hrtimer *refresh_timer = &cfs_b->period_timer;
-	u64 remaining;
+	s64 remaining;
 
 	/* if the call-back is running a quota refresh is already occurring */
 	if (hrtimer_callback_running(refresh_timer))
@@ -5116,7 +5116,7 @@ static int runtime_refresh_within(struct cfs_bandwidth *cfs_b, u64 min_expire)
 
 	/* is a quota refresh about to occur? */
 	remaining = ktime_to_ns(hrtimer_expires_remaining(refresh_timer));
-	if (remaining < min_expire)
+	if (remaining < (s64)min_expire)
 		return 1;
 
 	return 0;
-- 
2.32.0

Re: [PATCH] sched/fair: Fix CFS bandwidth hrtimer expiry type

[Benjamin Segall]

Odin Ugedal <odin@uged.al> writes:

> The time remaining until expiry of the refresh_timer can be negative.
> Casting the type to an unsigned 64-bit value will cause integer
> underflow, making the runtime_refresh_within return false instead of
> true. These situations are rare, but they do happen.
>
> This does not cause user-facing issues or errors; other than
> possibly unthrottling cfs_rq's using runtime from the previous period(s),
> making the CFS bandwidth enforcement less strict in those (special)
> situations.

Yeah, extremely rare, not any real sort of problem when it does happen,
but no reason not to fix it and get the slight win in precision.

Reviewed-by: Ben Segall <bsegall@google.com>

>
> Signed-off-by: Odin Ugedal <odin@uged.al>
> ---
>  kernel/sched/fair.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index 23663318fb81..62446c052efb 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -5108,7 +5108,7 @@ static const u64 cfs_bandwidth_slack_period = 5 * NSEC_PER_MSEC;
>  static int runtime_refresh_within(struct cfs_bandwidth *cfs_b, u64 min_expire)
>  {
>  	struct hrtimer *refresh_timer = &cfs_b->period_timer;
> -	u64 remaining;
> +	s64 remaining;
>  
>  	/* if the call-back is running a quota refresh is already occurring */
>  	if (hrtimer_callback_running(refresh_timer))
> @@ -5116,7 +5116,7 @@ static int runtime_refresh_within(struct cfs_bandwidth *cfs_b, u64 min_expire)
>  
>  	/* is a quota refresh about to occur? */
>  	remaining = ktime_to_ns(hrtimer_expires_remaining(refresh_timer));
> -	if (remaining < min_expire)
> +	if (remaining < (s64)min_expire)
>  		return 1;
>  
>  	return 0;

[tip: sched/urgent] sched/fair: Fix CFS bandwidth hrtimer expiry type

[tip-bot2 for Odin Ugedal]

The following commit has been merged into the sched/urgent branch of tip:

Commit-ID:     72d0ad7cb5bad265adb2014dbe46c4ccb11afaba
Gitweb:        https://git.kernel.org/tip/72d0ad7cb5bad265adb2014dbe46c4ccb11afaba
Author:        Odin Ugedal <odin@uged.al>
AuthorDate:    Tue, 29 Jun 2021 14:14:52 +02:00
Committer:     Peter Zijlstra <peterz@infradead.org>
CommitterDate: Fri, 02 Jul 2021 15:58:24 +02:00

sched/fair: Fix CFS bandwidth hrtimer expiry type

The time remaining until expiry of the refresh_timer can be negative.
Casting the type to an unsigned 64-bit value will cause integer
underflow, making the runtime_refresh_within return false instead of
true. These situations are rare, but they do happen.

This does not cause user-facing issues or errors; other than
possibly unthrottling cfs_rq's using runtime from the previous period(s),
making the CFS bandwidth enforcement less strict in those (special)
situations.

Signed-off-by: Odin Ugedal <odin@uged.al>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Ben Segall <bsegall@google.com>
Link: https://lore.kernel.org/r/20210629121452.18429-1-odin@uged.al
---
 kernel/sched/fair.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 1e263c9..1b15a19 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -5054,7 +5054,7 @@ static const u64 cfs_bandwidth_slack_period = 5 * NSEC_PER_MSEC;
 static int runtime_refresh_within(struct cfs_bandwidth *cfs_b, u64 min_expire)
 {
 	struct hrtimer *refresh_timer = &cfs_b->period_timer;
-	u64 remaining;
+	s64 remaining;
 
 	/* if the call-back is running a quota refresh is already occurring */
 	if (hrtimer_callback_running(refresh_timer))
@@ -5062,7 +5062,7 @@ static int runtime_refresh_within(struct cfs_bandwidth *cfs_b, u64 min_expire)
 
 	/* is a quota refresh about to occur? */
 	remaining = ktime_to_ns(hrtimer_expires_remaining(refresh_timer));
-	if (remaining < min_expire)
+	if (remaining < (s64)min_expire)
 		return 1;
 
 	return 0;