cpuhp: fix some st->target issues
v3 cover cpuhp: fix some st->target issues
Several small fixes that clean up some cpuhp inconsistencies. The first prevents target_store() from calling cpu_down() when target == state which prevents the cpu being incorrectly marked as dying. The second just makes the boot cpu have a valid cpuhp target rather than 0 (CPU_OFFLINE) while being in state CPU_ONLINE.
A further issue which these two patches don't address is that the cpuX/online file looks at the device->offline state and can thus get out of sync with the actual cpuhp state if the cpuhp target is used to change state.
v3: Added code to make sure st->target == target in the nop case.
Phil Auld (2):
cpuhp: make target_store() a nop when target == state
cpuhp: Set cpuhp target for boot cpu
kernel/cpu.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
Signature · 1 line
2.31.1
PATCH v3 1/2 cpuhp: make target_store() a nop when target == state
writing the current state back in hotplug/target calls cpu_down() which will set cpu dying even when it isn't and then nothing will ever clear it. A stress test that reads values and writes them back for all cpu device files in sysfs will trigger the BUG() in select_fallback_rq once all cpus are marked as dying.
kernel/cpu.c::target_store()
...
if (st->state < target)
ret = cpu_up(dev->id, target);
else
ret = cpu_down(dev->id, target);cpu_down() -> cpu_set_state()
bool bringup = st->state < target; ... if (cpu_dying(cpu) != !bringup) set_cpu_dying(cpu, !bringup);
Fix this by letting state==target fall through in the target_store() conditional. Also make sure st->target == target in that case.
--- kernel/cpu.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
kernel/cpu.c +3 -1
@@ -2326,8 +2326,10 @@ static ssize_t target_store(struct device *dev, struct device_attribute *attr,if (st->state < target) ret =cpu_up (dev->id, target); -else +else if (st->state > target) ret =cpu_down (dev->id, target); +else if (st->target != target) + st->target = target; out:unlock_device_hotplug ();return ret ? ret : count;
Signature · 1 line
2.31.1
Re: [PATCH v3 1/2] cpuhp: make target_store() a nop when target == state
Quoted 19 lines · Phil Auld
writing the current state back in hotplug/target calls
cpu_down()which will set cpu dying even when it isn't and then nothing will ever clear it. A stress test that reads values and writes them back for all cpu device files in sysfs will trigger theBUG()inselect_fallback_rqonce all cpus are marked as dying.kernel/cpu.c::target_store()
... if (st->state < target) ret = cpu_up(dev->id, target); else ret = cpu_down(dev->id, target);
cpu_down()->cpu_set_state()bool bringup = st->state < target; ... if (cpu_dying(cpu) != !bringup) set_cpu_dying(cpu, !bringup);
Quoted 4 lines
Fix this by letting state==target fall through in the
target_store()conditional. Also make surest->target== target in that case.Signed-off-by Phil Auld <[email protected]>
One nit below, otherwise:
Quoted 9 lines
--- kernel/cpu.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)kernel/cpu.c +0
-0
kernel/cpu.c static ssize_t target_store(struct device *dev, struct device_attribute *attr, · line 2331
Quoted 7 lines
kernel/cpu.c +2
-1@@ -2326,8 +2326,10 @@ static ssize_t target_store(struct device *dev, struct device_attribute *attr,if (st->state < target) ret =cpu_up (dev->id, target); -else +else if (st->state > target) ret =cpu_down (dev->id, target); +else if (st->target != target)
Should we make this:
else if (WARN(st->target != target))
Quoted 6 lines
kernel/cpu.c +1
-1+ st->target = target; out:unlock_device_hotplug ();return ret ? ret : count; --2.31.1
Re: [PATCH v3 1/2] cpuhp: make target_store() a nop when target == state
Quoted 20 lines · On Tue, Jul 19, 2022 at 03:34:03PM +0100 Valentin Schneider wrote:
Quoted 19 lines · Phil Auld
writing the current state back in hotplug/target calls
cpu_down()which will set cpu dying even when it isn't and then nothing will ever clear it. A stress test that reads values and writes them back for all cpu device files in sysfs will trigger theBUG()inselect_fallback_rqonce all cpus are marked as dying.kernel/cpu.c::target_store()
... if (st->state < target) ret = cpu_up(dev->id, target); else ret = cpu_down(dev->id, target);
cpu_down()->cpu_set_state()bool bringup = st->state < target; ... if (cpu_dying(cpu) != !bringup) set_cpu_dying(cpu, !bringup);
Quoted 8 lines
Quoted 4 lines
Fix this by letting state==target fall through in the
target_store()conditional. Also make surest->target== target in that case.Signed-off-by Phil Auld <[email protected]>One nit below, otherwise:
Reviewed-by Valentin Schneider <[email protected]>
Thanks!
Quoted 19 lines
Quoted 9 lines
--- kernel/cpu.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)kernel/cpu.c +0
-0kernel/cpu.c static ssize_t target_store(struct device *dev, struct device_attribute *attr, · line 2331
Quoted 7 lines
kernel/cpu.c +2
-1@@ -2326,8 +2326,10 @@ static ssize_t target_store(struct device *dev, struct device_attribute *attr,if (st->state < target) ret =cpu_up (dev->id, target); -else +else if (st->state > target) ret =cpu_down (dev->id, target); +else if (st->target != target)Should we make this:
Quoted 2 lines
else if (WARN(st->target != target))
If you think that's important I can make it a WARN, sure.
I'll try to remember to keep your Reviewed-bys this time if that's okay.
Cheers,
Phil
Quoted 7 lines
Quoted 6 lines
kernel/cpu.c +1
-1+ st->target = target; out:unlock_device_hotplug ();return ret ? ret : count; --2.31.1
PATCH v3 2/2 cpuhp: Set cpuhp target for boot cpu
In reply to Phil AuldSince the boot cpu does not go through the hotplug process it ends up with state == CPUHP_ONLINE but target == CPUHP_OFFLINE. Set the target to match in boot_cpu_hotplug_init().
--- kernel/cpu.c | 1 + 1 file changed, 1 insertion(+)
kernel/cpu.c +1 -0
@@ -2690,6 +2690,7 @@ void __init boot_cpu_hotplug_init(void)cpumask_set_cpu (smp_processor_id (), &cpus_booted_once_mask);#endif this_cpu_write (cpuhp_state.state,CPUHP_ONLINE ); +this_cpu_write (cpuhp_state.target,CPUHP_ONLINE ); }/*
Signature · 1 line
2.31.1
Re: [PATCH v3 2/2] cpuhp: Set cpuhp target for boot cpu
Quoted 5 lines · Phil Auld
Since the boot cpu does not go through the hotplug process it ends up with state ==
CPUHP_ONLINEbut target ==CPUHP_OFFLINE. Set the target to match inboot_cpu_hotplug_init().Signed-off-by Phil Auld <[email protected]>
Quoted 12 lines · On Mon, Jul 11, 2022 at 05:16:17PM -0400 Phil Auld wrote:
Several small fixes that clean up some cpuhp inconsistencies. The first prevents
target_store()from callingcpu_down()when target == state which prevents the cpu being incorrectly marked as dying. The second just makes the boot cpu have a valid cpuhp target rather than 0 (CPU_OFFLINE) while being in stateCPU_ONLINE.A further issue which these two patches don't address is that the cpuX/online file looks at the
device->offlinestate and can thus get out of sync with the actual cpuhp state if the cpuhp target is used to change state.
Quoted 12 lines
v3: Added code to make sure
st->target== target in the nop case.Phil Auld (2):
cpuhp: maketarget_store()a nop when target == state
cpuhp: Set cpuhp target for boot cpukernel/cpu.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)Signature · 1 line
2.31.1
Ping?
Any thoughts on this? I'm not sure what tree it needs to go in but getting a cpu shutdown by writing the current target/state back in here seems wrong.
Thanks,
Phil