Patch 1 addresses a smatch warning reported by Dan Carpenter.
Patch 2 fixes a couple of issues reported by Dietmar in capacity inversion
logic.
Patch 3 which was an RFC patch was dropped. The discussion has settled into
this patch which is now treated separately from this series
https://lore.kernel.org/lkml/[email protected]/
Changes in v3:
* Fix commit message in patch 2.
* Drop Patch 3
Changes in v2:
* Patch1: Improve indentation as suggested by Dietmar
* Patch2: Make sure to hold rcu_read_lock() as we need it's not
held in all paths.
LINK v1:
* https://lore.kernel.org/lkml/[email protected]/
LINK v2:
Sent in-reply-to v1
* https://lore.kernel.org/lkml/[email protected]/
* https://lore.kernel.org/lkml/[email protected]/
Qais Yousef (2):
sched/uclamp: Fix a uninitialized variable warnings
sched/fair: Fixes for capacity inversion detection
kernel/sched/fair.c | 48 +++++++++++++++++++++++++--------------------
1 file changed, 27 insertions(+), 21 deletions(-)
--
2.25.1
Addresses the following warnings:
> config: riscv-randconfig-m031-20221111
> compiler: riscv64-linux-gcc (GCC) 12.1.0
>
> smatch warnings:
> kernel/sched/fair.c:7263 find_energy_efficient_cpu() error: uninitialized symbol 'util_min'.
> kernel/sched/fair.c:7263 find_energy_efficient_cpu() error: uninitialized symbol 'util_max'.
Fixes: 244226035a1f ("sched/uclamp: Fix fits_capacity() check in feec()")
Reported-by: kernel test robot <[email protected]>
Reported-by: Dan Carpenter <[email protected]>
Signed-off-by: Qais Yousef (Google) <[email protected]>
---
kernel/sched/fair.c | 35 ++++++++++++++++-------------------
1 file changed, 16 insertions(+), 19 deletions(-)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index e9d906a9bba9..5a8e75d4a17b 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -7353,10 +7353,10 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
eenv_task_busy_time(&eenv, p, prev_cpu);
for (; pd; pd = pd->next) {
+ unsigned long util_min = p_util_min, util_max = p_util_max;
unsigned long cpu_cap, cpu_thermal_cap, util;
unsigned long cur_delta, max_spare_cap = 0;
unsigned long rq_util_min, rq_util_max;
- unsigned long util_min, util_max;
unsigned long prev_spare_cap = 0;
int max_spare_cap_cpu = -1;
unsigned long base_energy;
@@ -7375,6 +7375,8 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
eenv.pd_cap = 0;
for_each_cpu(cpu, cpus) {
+ struct rq *rq = cpu_rq(cpu);
+
eenv.pd_cap += cpu_thermal_cap;
if (!cpumask_test_cpu(cpu, sched_domain_span(sd)))
@@ -7393,24 +7395,19 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
* much capacity we can get out of the CPU; this is
* aligned with sched_cpu_util().
*/
- if (uclamp_is_used()) {
- if (uclamp_rq_is_idle(cpu_rq(cpu))) {
- util_min = p_util_min;
- util_max = p_util_max;
- } else {
- /*
- * Open code uclamp_rq_util_with() except for
- * the clamp() part. Ie: apply max aggregation
- * only. util_fits_cpu() logic requires to
- * operate on non clamped util but must use the
- * max-aggregated uclamp_{min, max}.
- */
- rq_util_min = uclamp_rq_get(cpu_rq(cpu), UCLAMP_MIN);
- rq_util_max = uclamp_rq_get(cpu_rq(cpu), UCLAMP_MAX);
-
- util_min = max(rq_util_min, p_util_min);
- util_max = max(rq_util_max, p_util_max);
- }
+ if (uclamp_is_used() && !uclamp_rq_is_idle(rq)) {
+ /*
+ * Open code uclamp_rq_util_with() except for
+ * the clamp() part. Ie: apply max aggregation
+ * only. util_fits_cpu() logic requires to
+ * operate on non clamped util but must use the
+ * max-aggregated uclamp_{min, max}.
+ */
+ rq_util_min = uclamp_rq_get(rq, UCLAMP_MIN);
+ rq_util_max = uclamp_rq_get(rq, UCLAMP_MAX);
+
+ util_min = max(rq_util_min, p_util_min);
+ util_max = max(rq_util_max, p_util_max);
}
if (!util_fits_cpu(util, util_min, util_max, cpu))
continue;
--
2.25.1
Traversing the Perf Domains requires rcu_read_lock() to be held and is
conditional on sched_energy_enabled(). Ensure right protections applied.
Also skip capacity inversion detection for our own pd; which was an
error.
Fixes: 44c7b80bffc3 ("sched/fair: Detect capacity inversion")
Reported-by: Dietmar Eggemann <[email protected]>
Signed-off-by: Qais Yousef (Google) <[email protected]>
---
kernel/sched/fair.c | 13 +++++++++++--
1 file changed, 11 insertions(+), 2 deletions(-)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 5a8e75d4a17b..34239d3118f0 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -8992,16 +8992,23 @@ static void update_cpu_capacity(struct sched_domain *sd, int cpu)
* * Thermal pressure will impact all cpus in this perf domain
* equally.
*/
- if (static_branch_unlikely(&sched_asym_cpucapacity)) {
+ if (sched_energy_enabled()) {
unsigned long inv_cap = capacity_orig - thermal_load_avg(rq);
- struct perf_domain *pd = rcu_dereference(rq->rd->pd);
+ struct perf_domain *pd;
+
+ rcu_read_lock();
+ pd = rcu_dereference(rq->rd->pd);
rq->cpu_capacity_inverted = 0;
for (; pd; pd = pd->next) {
struct cpumask *pd_span = perf_domain_span(pd);
unsigned long pd_cap_orig, pd_cap;
+ /* We can't be inverted against our own pd */
+ if (cpumask_test_cpu(cpu_of(rq), pd_span))
+ continue;
+
cpu = cpumask_any(pd_span);
pd_cap_orig = arch_scale_cpu_capacity(cpu);
@@ -9026,6 +9033,8 @@ static void update_cpu_capacity(struct sched_domain *sd, int cpu)
break;
}
}
+
+ rcu_read_unlock();
}
trace_sched_cpu_capacity_tp(rq);
--
2.25.1
On Thu, 12 Jan 2023 at 13:27, Qais Yousef <[email protected]> wrote:
>
> Traversing the Perf Domains requires rcu_read_lock() to be held and is
> conditional on sched_energy_enabled(). Ensure right protections applied.
>
> Also skip capacity inversion detection for our own pd; which was an
> error.
>
> Fixes: 44c7b80bffc3 ("sched/fair: Detect capacity inversion")
> Reported-by: Dietmar Eggemann <[email protected]>
> Signed-off-by: Qais Yousef (Google) <[email protected]>
Reviewed-by: Vincent Guittot <[email protected]>
> ---
> kernel/sched/fair.c | 13 +++++++++++--
> 1 file changed, 11 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index 5a8e75d4a17b..34239d3118f0 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -8992,16 +8992,23 @@ static void update_cpu_capacity(struct sched_domain *sd, int cpu)
> * * Thermal pressure will impact all cpus in this perf domain
> * equally.
> */
> - if (static_branch_unlikely(&sched_asym_cpucapacity)) {
> + if (sched_energy_enabled()) {
> unsigned long inv_cap = capacity_orig - thermal_load_avg(rq);
> - struct perf_domain *pd = rcu_dereference(rq->rd->pd);
> + struct perf_domain *pd;
> +
> + rcu_read_lock();
>
> + pd = rcu_dereference(rq->rd->pd);
> rq->cpu_capacity_inverted = 0;
>
> for (; pd; pd = pd->next) {
> struct cpumask *pd_span = perf_domain_span(pd);
> unsigned long pd_cap_orig, pd_cap;
>
> + /* We can't be inverted against our own pd */
> + if (cpumask_test_cpu(cpu_of(rq), pd_span))
> + continue;
> +
> cpu = cpumask_any(pd_span);
> pd_cap_orig = arch_scale_cpu_capacity(cpu);
>
> @@ -9026,6 +9033,8 @@ static void update_cpu_capacity(struct sched_domain *sd, int cpu)
> break;
> }
> }
> +
> + rcu_read_unlock();
> }
>
> trace_sched_cpu_capacity_tp(rq);
> --
> 2.25.1
>
On Thu, 12 Jan 2023 at 13:27, Qais Yousef <[email protected]> wrote:
>
> Addresses the following warnings:
>
> > config: riscv-randconfig-m031-20221111
> > compiler: riscv64-linux-gcc (GCC) 12.1.0
> >
> > smatch warnings:
> > kernel/sched/fair.c:7263 find_energy_efficient_cpu() error: uninitialized symbol 'util_min'.
> > kernel/sched/fair.c:7263 find_energy_efficient_cpu() error: uninitialized symbol 'util_max'.
>
> Fixes: 244226035a1f ("sched/uclamp: Fix fits_capacity() check in feec()")
> Reported-by: kernel test robot <[email protected]>
> Reported-by: Dan Carpenter <[email protected]>
> Signed-off-by: Qais Yousef (Google) <[email protected]>
Reviewed-by: Vincent Guittot <[email protected]>
> ---
> kernel/sched/fair.c | 35 ++++++++++++++++-------------------
> 1 file changed, 16 insertions(+), 19 deletions(-)
>
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index e9d906a9bba9..5a8e75d4a17b 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -7353,10 +7353,10 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
> eenv_task_busy_time(&eenv, p, prev_cpu);
>
> for (; pd; pd = pd->next) {
> + unsigned long util_min = p_util_min, util_max = p_util_max;
> unsigned long cpu_cap, cpu_thermal_cap, util;
> unsigned long cur_delta, max_spare_cap = 0;
> unsigned long rq_util_min, rq_util_max;
> - unsigned long util_min, util_max;
> unsigned long prev_spare_cap = 0;
> int max_spare_cap_cpu = -1;
> unsigned long base_energy;
> @@ -7375,6 +7375,8 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
> eenv.pd_cap = 0;
>
> for_each_cpu(cpu, cpus) {
> + struct rq *rq = cpu_rq(cpu);
> +
> eenv.pd_cap += cpu_thermal_cap;
>
> if (!cpumask_test_cpu(cpu, sched_domain_span(sd)))
> @@ -7393,24 +7395,19 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
> * much capacity we can get out of the CPU; this is
> * aligned with sched_cpu_util().
> */
> - if (uclamp_is_used()) {
> - if (uclamp_rq_is_idle(cpu_rq(cpu))) {
> - util_min = p_util_min;
> - util_max = p_util_max;
> - } else {
> - /*
> - * Open code uclamp_rq_util_with() except for
> - * the clamp() part. Ie: apply max aggregation
> - * only. util_fits_cpu() logic requires to
> - * operate on non clamped util but must use the
> - * max-aggregated uclamp_{min, max}.
> - */
> - rq_util_min = uclamp_rq_get(cpu_rq(cpu), UCLAMP_MIN);
> - rq_util_max = uclamp_rq_get(cpu_rq(cpu), UCLAMP_MAX);
> -
> - util_min = max(rq_util_min, p_util_min);
> - util_max = max(rq_util_max, p_util_max);
> - }
> + if (uclamp_is_used() && !uclamp_rq_is_idle(rq)) {
> + /*
> + * Open code uclamp_rq_util_with() except for
> + * the clamp() part. Ie: apply max aggregation
> + * only. util_fits_cpu() logic requires to
> + * operate on non clamped util but must use the
> + * max-aggregated uclamp_{min, max}.
> + */
> + rq_util_min = uclamp_rq_get(rq, UCLAMP_MIN);
> + rq_util_max = uclamp_rq_get(rq, UCLAMP_MAX);
> +
> + util_min = max(rq_util_min, p_util_min);
> + util_max = max(rq_util_max, p_util_max);
> }
> if (!util_fits_cpu(util, util_min, util_max, cpu))
> continue;
> --
> 2.25.1
>
The following commit has been merged into the sched/urgent branch of tip:
Commit-ID: da07d2f9c153e457e845d4dcfdd13568d71d18a4
Gitweb: https://git.kernel.org/tip/da07d2f9c153e457e845d4dcfdd13568d71d18a4
Author: Qais Yousef <[email protected]>
AuthorDate: Thu, 12 Jan 2023 12:27:08
Committer: Peter Zijlstra <[email protected]>
CommitterDate: Fri, 13 Jan 2023 11:40:21 +01:00
sched/fair: Fixes for capacity inversion detection
Traversing the Perf Domains requires rcu_read_lock() to be held and is
conditional on sched_energy_enabled(). Ensure right protections applied.
Also skip capacity inversion detection for our own pd; which was an
error.
Fixes: 44c7b80bffc3 ("sched/fair: Detect capacity inversion")
Reported-by: Dietmar Eggemann <[email protected]>
Signed-off-by: Qais Yousef (Google) <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Vincent Guittot <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
---
kernel/sched/fair.c | 13 +++++++++++--
1 file changed, 11 insertions(+), 2 deletions(-)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index be43731..0f87369 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -8868,16 +8868,23 @@ static void update_cpu_capacity(struct sched_domain *sd, int cpu)
* * Thermal pressure will impact all cpus in this perf domain
* equally.
*/
- if (static_branch_unlikely(&sched_asym_cpucapacity)) {
+ if (sched_energy_enabled()) {
unsigned long inv_cap = capacity_orig - thermal_load_avg(rq);
- struct perf_domain *pd = rcu_dereference(rq->rd->pd);
+ struct perf_domain *pd;
+
+ rcu_read_lock();
+ pd = rcu_dereference(rq->rd->pd);
rq->cpu_capacity_inverted = 0;
for (; pd; pd = pd->next) {
struct cpumask *pd_span = perf_domain_span(pd);
unsigned long pd_cap_orig, pd_cap;
+ /* We can't be inverted against our own pd */
+ if (cpumask_test_cpu(cpu_of(rq), pd_span))
+ continue;
+
cpu = cpumask_any(pd_span);
pd_cap_orig = arch_scale_cpu_capacity(cpu);
@@ -8902,6 +8909,8 @@ static void update_cpu_capacity(struct sched_domain *sd, int cpu)
break;
}
}
+
+ rcu_read_unlock();
}
trace_sched_cpu_capacity_tp(rq);
The following commit has been merged into the sched/urgent branch of tip:
Commit-ID: e26fd28db82899be71b4b949527373d0a6be1e65
Gitweb: https://git.kernel.org/tip/e26fd28db82899be71b4b949527373d0a6be1e65
Author: Qais Yousef <[email protected]>
AuthorDate: Thu, 12 Jan 2023 12:27:07
Committer: Peter Zijlstra <[email protected]>
CommitterDate: Fri, 13 Jan 2023 11:40:21 +01:00
sched/uclamp: Fix a uninitialized variable warnings
Addresses the following warnings:
> config: riscv-randconfig-m031-20221111
> compiler: riscv64-linux-gcc (GCC) 12.1.0
>
> smatch warnings:
> kernel/sched/fair.c:7263 find_energy_efficient_cpu() error: uninitialized symbol 'util_min'.
> kernel/sched/fair.c:7263 find_energy_efficient_cpu() error: uninitialized symbol 'util_max'.
Fixes: 244226035a1f ("sched/uclamp: Fix fits_capacity() check in feec()")
Reported-by: kernel test robot <[email protected]>
Reported-by: Dan Carpenter <[email protected]>
Signed-off-by: Qais Yousef (Google) <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Vincent Guittot <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
---
kernel/sched/fair.c | 35 ++++++++++++++++-------------------
1 file changed, 16 insertions(+), 19 deletions(-)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index c36aa54..be43731 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -7229,10 +7229,10 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
eenv_task_busy_time(&eenv, p, prev_cpu);
for (; pd; pd = pd->next) {
+ unsigned long util_min = p_util_min, util_max = p_util_max;
unsigned long cpu_cap, cpu_thermal_cap, util;
unsigned long cur_delta, max_spare_cap = 0;
unsigned long rq_util_min, rq_util_max;
- unsigned long util_min, util_max;
unsigned long prev_spare_cap = 0;
int max_spare_cap_cpu = -1;
unsigned long base_energy;
@@ -7251,6 +7251,8 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
eenv.pd_cap = 0;
for_each_cpu(cpu, cpus) {
+ struct rq *rq = cpu_rq(cpu);
+
eenv.pd_cap += cpu_thermal_cap;
if (!cpumask_test_cpu(cpu, sched_domain_span(sd)))
@@ -7269,24 +7271,19 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
* much capacity we can get out of the CPU; this is
* aligned with sched_cpu_util().
*/
- if (uclamp_is_used()) {
- if (uclamp_rq_is_idle(cpu_rq(cpu))) {
- util_min = p_util_min;
- util_max = p_util_max;
- } else {
- /*
- * Open code uclamp_rq_util_with() except for
- * the clamp() part. Ie: apply max aggregation
- * only. util_fits_cpu() logic requires to
- * operate on non clamped util but must use the
- * max-aggregated uclamp_{min, max}.
- */
- rq_util_min = uclamp_rq_get(cpu_rq(cpu), UCLAMP_MIN);
- rq_util_max = uclamp_rq_get(cpu_rq(cpu), UCLAMP_MAX);
-
- util_min = max(rq_util_min, p_util_min);
- util_max = max(rq_util_max, p_util_max);
- }
+ if (uclamp_is_used() && !uclamp_rq_is_idle(rq)) {
+ /*
+ * Open code uclamp_rq_util_with() except for
+ * the clamp() part. Ie: apply max aggregation
+ * only. util_fits_cpu() logic requires to
+ * operate on non clamped util but must use the
+ * max-aggregated uclamp_{min, max}.
+ */
+ rq_util_min = uclamp_rq_get(rq, UCLAMP_MIN);
+ rq_util_max = uclamp_rq_get(rq, UCLAMP_MAX);
+
+ util_min = max(rq_util_min, p_util_min);
+ util_max = max(rq_util_max, p_util_max);
}
if (!util_fits_cpu(util, util_min, util_max, cpu))
continue;