Skip to contentSkip to conversation

x86/resctrl: fix an imbalance in domain_remove_cpu

PATCH Superseded Qian Cai 3 participants 8 messages Dec 8, 2019 – Dec 12, 2019

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Qian Cai

author
#0 lore raw

PATCH 1/1 x86/resctrl: fix an imbalance in domain_remove_cpu

domain_add_cpu() calls domain_setup_mon_state() only when r->mon_capable is true where it will initialize d->mbm_over. However, domain_remove_cpu() calls cancel_delayed_work(&d->mbm_over) without checking r->mon_capable. Hence, it triggers a debugobjects warning when offlining CPUs because those timer debugobjects are never initialized.

ODEBUG: assert_init not available (active state 0) object type:
timer_list hint: 0x0

WARNING: CPU: 143 PID: 789 at lib/debugobjects.c:484
debug_print_object+0xfe/0x140
Hardware name: HP Synergy 680 Gen9/Synergy 680 Gen9 Compute Module, BIOS
I40 05/23/2018
RIP: 0010:debug_print_object+0xfe/0x140
Call Trace:
debug_object_assert_init+0x1f5/0x240
del_timer+0x6f/0xf0
try_to_grab_pending+0x42/0x3c0
cancel_delayed_work+0x7d/0x150
resctrl_offline_cpu+0x3c0/0x520
cpuhp_invoke_callback+0x197/0x1120
cpuhp_thread_fun+0x252/0x2f0
smpboot_thread_fn+0x255/0x440
kthread+0x1e6/0x210
ret_from_fork+0x3a/0x50
Fixes e33026831bdb ("x86/intel_rdt/mbm: Handle counter overflow")
---
 arch/x86/kernel/cpu/resctrl/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kernel/cpu/resctrl/core.c +1 -1
index 03eb90d00af0..89049b343c7a 100644
@@ -618,7 +618,7 @@ static void domain_remove_cpu(int cpu, struct rdt_resource *r)
 		if (static_branch_unlikely(&rdt_mon_enable_key))
 			rmdir_mondata_subdir_allrdtgrp(r, d->id);
 		list_del(&d->list);
-		if (is_mbm_enabled())
+		if (r->mon_capable && is_mbm_enabled())
 			cancel_delayed_work(&d->mbm_over);
 		if (is_llc_occupancy_enabled() &&  has_busy_rmid(r, d)) {
 			/*
Signature · 1 line
2.21.0 (Apple Git-122.2)
#1 lore raw

Hi Qian,

Quoted 7 lines · Qian Cai

domain_add_cpu() calls domain_setup_mon_state() only when r->mon_capable is true where it will initialize d->mbm_over. However, domain_remove_cpu() calls cancel_delayed_work(&d->mbm_over) without checking r->mon_capable. Hence, it triggers a debugobjects warning when offlining CPUs because those timer debugobjects are never initialized.

Could you elaborate a little more on the failure symptom?
If I understand correctly, the error you described was due to r->mon_capable set to false while is_mbm_enabled() returns true? Which means on this platform rdt_mon_features is non zero?
And in get_rdt_mon_resources() it will invoke rdt_get_mon_l3_config(), however the only possible failure to do not set r->mon_capable is that it
failed in dom_data_init() due to kcalloc() failure? Then the logic in
get_rdt_resources() is that it will ignore the return error if rdt allocate feature is supported on this platform? If this is the case, the r->mon_capable is not an indicator for whether the overflow thread has been created, right? Can we simply remove the check of r->mon_capable in domain_add_cpu() and invoke domain_setup_mon_state() directly?

Quoted 30 lines

ODEBUG: assert_init not available (active state 0) object type:
timer_list hint: 0x0

WARNING: CPU: 143 PID: 789 at lib/debugobjects.c:484
debug_print_object+0xfe/0x140
Hardware name: HP Synergy 680 Gen9/Synergy 680 Gen9 Compute Module, BIOS
I40 05/23/2018
RIP: 0010:debug_print_object+0xfe/0x140
Call Trace:
debug_object_assert_init+0x1f5/0x240
del_timer+0x6f/0xf0
try_to_grab_pending+0x42/0x3c0
cancel_delayed_work+0x7d/0x150
resctrl_offline_cpu+0x3c0/0x520
cpuhp_invoke_callback+0x197/0x1120
cpuhp_thread_fun+0x252/0x2f0
smpboot_thread_fn+0x255/0x440
kthread+0x1e6/0x210
ret_from_fork+0x3a/0x50
Fixes e33026831bdb ("x86/intel_rdt/mbm: Handle counter overflow")
---
 arch/x86/kernel/cpu/resctrl/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kernel/cpu/resctrl/core.c +0 -0
index 03eb90d00af0..89049b343c7a 100644

arch/x86/kernel/cpu/resctrl/core.c static void domain_remove_cpu(int cpu, struct rdt_resource *r) · line 622

Quoted 6 lines
arch/x86/kernel/cpu/resctrl/core.c +1 -1
@@ -618,7 +618,7 @@ static void domain_remove_cpu(int cpu, struct rdt_resource *r)
                if (static_branch_unlikely(&rdt_mon_enable_key))
                        rmdir_mondata_subdir_allrdtgrp(r, d->id);
                list_del(&d->list);
-               if (is_mbm_enabled())
+               if (r->mon_capable && is_mbm_enabled())
                        cancel_delayed_work(&d->mbm_over);

Humm, it looks like there are two places within this function invoked cancel_delayed_work(&d->mbm_over),
why not adding the check for both of them?

thanks,
Y

author
#2 lore raw
Quoted 7 lines · Ryan Chen

Could you elaborate a little more on the failure symptom? If I understand correctly, the error you described was due to r->mon_capable set to false while is_mbm_enabled() returns true?

Yes.

Quoted 1 line

Which means on this platform rdt_mon_features is non zero?

No idea. I did add some debug code that indicated resctrl_online_cpu() found 3 resources in for_each_capable_rdt_resource(r). Only the first one set r->mon_capable.

Quoted 3 lines

And in get_rdt_mon_resources() it will invoke rdt_get_mon_l3_config(), however the only possible failure to do not set r->mon_capable is that it failed in dom_data_init() due to kcalloc() failure? Then the logic in

Very likely. Should be easy to confirm.

Quoted 2 lines

get_rdt_resources() is that it will ignore the return error if rdt allocate feature is supported on this platform? If

Yes.

Quoted 2 lines

this is the case, the r->mon_capable
is not an indicator for whether the overflow thread has been created, right?

I am not sure about that.

Quoted 2 lines

Can we simply remove the check of r->mon_capable in domain_add_cpu() and invoke domain_setup_mon_state() directly?

That should work too, but it is so perfect align with the r->alloc_capable check above, so I am not sure it is a good idea to break it.

Quoted 3 lines

Humm, it looks like there are two places within this function invoked cancel_delayed_work(&d->mbm_over),
why not adding the check for both of them?

Because I am not sure about the second one. It was never executed due to “cpu != d->mbm_work_cpu“ even after offlining all CPUs except cpu 0 and never cause anything wrong yet, so I could not test it yet, but I can see why it might need a similar check too if d->mbm_work_cpu is non-zero and could trigger the same imbalance.

author
#3 lore raw
In reply to Ryan Chen
Quoted 14 lines · Ryan Chen

Hi Qian,

Quoted 7 lines · Qian Cai

domain_add_cpu() calls domain_setup_mon_state() only when r->mon_capable is true where it will initialize d->mbm_over. However, domain_remove_cpu() calls cancel_delayed_work(&d->mbm_over) without checking r->mon_capable. Hence, it triggers a debugobjects warning when offlining CPUs because those timer debugobjects are never initialized.

Quoted 12 lines

Could you elaborate a little more on the failure symptom?
If I understand correctly, the error you described was due to r->mon_capable set to false while is_mbm_enabled() returns true? Which means on this platform rdt_mon_features is non zero?
And in get_rdt_mon_resources() it will invoke rdt_get_mon_l3_config(), however the only possible failure to do not set r->mon_capable is that it failed in dom_data_init() due to kcalloc() failure? Then the logic in get_rdt_resources() is that it will ignore the return error if rdt allocate feature is supported on this platform? If this is the case, the r->mon_capable is not an indicator for whether the overflow thread has been created, right? Can we simply remove the check of r->mon_capable in domain_add_cpu() and invoke domain_setup_mon_state() directly?

Actually,

domain_add_cpu r->name = L3, r->alloc_capable = 1, r->mon_capable = 1 domain_add_cpu r->name = L3DATA, r->alloc_capable = 1, r->mon_capable = 0 domain_add_cpu r->name = L3CODE, r->alloc_capable = 1, r->mon_capable = 0

rdt_get_mon_l3_config() will only set r->mon_capable = 1 for L3.

Quoted 25 lines
Quoted 25 lines

ODEBUG: assert_init not available (active state 0) object type:
timer_list hint: 0x0

WARNING: CPU: 143 PID: 789 at lib/debugobjects.c:484
debug_print_object+0xfe/0x140
Hardware name: HP Synergy 680 Gen9/Synergy 680 Gen9 Compute Module, BIOS
I40 05/23/2018
RIP: 0010:debug_print_object+0xfe/0x140
Call Trace:
debug_object_assert_init+0x1f5/0x240
del_timer+0x6f/0xf0
try_to_grab_pending+0x42/0x3c0
cancel_delayed_work+0x7d/0x150
resctrl_offline_cpu+0x3c0/0x520
cpuhp_invoke_callback+0x197/0x1120
cpuhp_thread_fun+0x252/0x2f0
smpboot_thread_fn+0x255/0x440
kthread+0x1e6/0x210
ret_from_fork+0x3a/0x50
Fixes e33026831bdb ("x86/intel_rdt/mbm: Handle counter overflow")

---
arch/x86/kernel/cpu/resctrl/core.c | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 14 lines

arch/x86/kernel/cpu/resctrl/core.c static void domain_remove_cpu(int cpu, struct rdt_resource *r) · line 622

Quoted 11 lines
arch/x86/kernel/cpu/resctrl/core.c +1 -1
index 03eb90d00af0..89049b343c7a 100644
@@ -618,7 +618,7 @@ static void domain_remove_cpu(int cpu, struct rdt_resource *r)
               if (static_branch_unlikely(&rdt_mon_enable_key))
                       rmdir_mondata_subdir_allrdtgrp(r, d->id);
               list_del(&d->list);
-               if (is_mbm_enabled())
+               if (r->mon_capable && is_mbm_enabled())
                       cancel_delayed_work(&d->mbm_over);

Humm, it looks like there are two places within this function invoked cancel_delayed_work(&d->mbm_over),
why not adding the check for both of them?

Here it only check L3, so it will skip correctly for L3DATA and L3CODE to not call cancel_delayed_work(). Recalled the above that only L3 will have r->capable set.

if (r == &rdt_resources_all[RDT_RESOURCE_L3]) {
	if (is_mbm_enabled() && cpu == d->mbm_work_cpu) {			
		cancel_delayed_work(&d->mbm_over);

Hence, r->mon_capable check seems redundant here.

maintainer
#4 lore raw

Hi Qian,

Quoted 77 lines · Qian Cai
Quoted 12 lines · Ryan Chen

Hi Qian,

Quoted 7 lines · Qian Cai

domain_add_cpu() calls domain_setup_mon_state() only when r->mon_capable is true where it will initialize d->mbm_over. However, domain_remove_cpu() calls cancel_delayed_work(&d->mbm_over) without checking r->mon_capable. Hence, it triggers a debugobjects warning when offlining CPUs because those timer debugobjects are never initialized.

Quoted 12 lines

Could you elaborate a little more on the failure symptom?
If I understand correctly, the error you described was due to r->mon_capable set to false while is_mbm_enabled() returns true? Which means on this platform rdt_mon_features is non zero?
And in get_rdt_mon_resources() it will invoke rdt_get_mon_l3_config(), however the only possible failure to do not set r->mon_capable is that it failed in dom_data_init() due to kcalloc() failure? Then the logic in get_rdt_resources() is that it will ignore the return error if rdt allocate feature is supported on this platform? If this is the case, the r->mon_capable is not an indicator for whether the overflow thread has been created, right? Can we simply remove the check of r->mon_capable in domain_add_cpu() and invoke domain_setup_mon_state() directly?

Actually,

domain_add_cpu r->name = L3, r->alloc_capable = 1, r->mon_capable = 1 domain_add_cpu r->name = L3DATA, r->alloc_capable = 1, r->mon_capable = 0 domain_add_cpu r->name = L3CODE, r->alloc_capable = 1, r->mon_capable = 0

rdt_get_mon_l3_config() will only set r->mon_capable = 1 for L3.

Quoted 25 lines
Quoted 25 lines

ODEBUG: assert_init not available (active state 0) object type:
timer_list hint: 0x0

WARNING: CPU: 143 PID: 789 at lib/debugobjects.c:484
debug_print_object+0xfe/0x140
Hardware name: HP Synergy 680 Gen9/Synergy 680 Gen9 Compute Module, BIOS
I40 05/23/2018
RIP: 0010:debug_print_object+0xfe/0x140
Call Trace:
debug_object_assert_init+0x1f5/0x240
del_timer+0x6f/0xf0
try_to_grab_pending+0x42/0x3c0
cancel_delayed_work+0x7d/0x150
resctrl_offline_cpu+0x3c0/0x520
cpuhp_invoke_callback+0x197/0x1120
cpuhp_thread_fun+0x252/0x2f0
smpboot_thread_fn+0x255/0x440
kthread+0x1e6/0x210
ret_from_fork+0x3a/0x50
Fixes e33026831bdb ("x86/intel_rdt/mbm: Handle counter overflow")

---
arch/x86/kernel/cpu/resctrl/core.c | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 14 lines

arch/x86/kernel/cpu/resctrl/core.c static void domain_remove_cpu(int cpu, struct rdt_resource *r) · line 622

Quoted 11 lines
arch/x86/kernel/cpu/resctrl/core.c +1 -1
index 03eb90d00af0..89049b343c7a 100644
@@ -618,7 +618,7 @@ static void domain_remove_cpu(int cpu, struct rdt_resource *r)
               if (static_branch_unlikely(&rdt_mon_enable_key))
                       rmdir_mondata_subdir_allrdtgrp(r, d->id);
               list_del(&d->list);
-               if (is_mbm_enabled())
+               if (r->mon_capable && is_mbm_enabled())
                       cancel_delayed_work(&d->mbm_over);

Humm, it looks like there are two places within this function invoked cancel_delayed_work(&d->mbm_over),
why not adding the check for both of them?

Here it only check L3, so it will skip correctly for L3DATA and L3CODE to not call cancel_delayed_work(). Recalled the above that only L3 will have r->capable set.

Quoted 6 lines
if (r == &rdt_resources_all[RDT_RESOURCE_L3]) {
	if (is_mbm_enabled() && cpu == d->mbm_work_cpu) {			
		cancel_delayed_work(&d->mbm_over);

Hence, r->mon_capable check seems redundant here.

Thank you very much for catching this. Your change looks like the right thing to do. As Ryan pointed out this is not obvious at first and looking back at this commit at a later time may benefit from some more details. For example, how about:

"A system that supports resource monitoring may have multiple resources while not all of these resources are capable of monitoring. Monitoring related state is initialized only for resources that are capable of monitoring and correspondingly this state should subsequently only be removed from these resources that are capable of monitoring.

domain_add_cpu() calls domain_setup_mon_state() only when r->mon_capable is true where it will initialize d->mbm_over. However, domain_remove_cpu() calls cancel_delayed_work(&d->mbm_over) without checking r->mon_capable resulting in an attempt to cancel d->mbm_over on all resources, even those that never initialized d->mbm_over because they are not capable of monitoring. Hence, it triggers a debugobjects warning when offlining CPUs because those timer debugobjects are never initialized.

ODEBUG:..."

Reinette

author
#5 lore raw
Quoted 20 lines · Reinette Chatre

"A system that supports resource monitoring may have multiple resources while not all of these resources are capable of monitoring. Monitoring related state is initialized only for resources that are capable of monitoring and correspondingly this state should subsequently only be removed from these resources that are capable of monitoring.

domain_add_cpu() calls domain_setup_mon_state() only when r->mon_capable is true where it will initialize d->mbm_over. However, domain_remove_cpu() calls cancel_delayed_work(&d->mbm_over) without checking r->mon_capable resulting in an attempt to cancel d->mbm_over on all resources, even those that never initialized d->mbm_over because they are not capable of monitoring. Hence, it triggers a debugobjects warning when offlining CPUs because those timer debugobjects are never initialized.

Quoted 1 line

ODEBUG:..."

Looks better to me. Do you want me to send a v2 for it or you could update it for merging?

maintainer
#6 lore raw

Hi Qian,

Quoted 20 lines · Qian Cai
Quoted 20 lines · Reinette Chatre

"A system that supports resource monitoring may have multiple resources while not all of these resources are capable of monitoring. Monitoring related state is initialized only for resources that are capable of monitoring and correspondingly this state should subsequently only be removed from these resources that are capable of monitoring.

domain_add_cpu() calls domain_setup_mon_state() only when r->mon_capable is true where it will initialize d->mbm_over. However, domain_remove_cpu() calls cancel_delayed_work(&d->mbm_over) without checking r->mon_capable resulting in an attempt to cancel d->mbm_over on all resources, even those that never initialized d->mbm_over because they are not capable of monitoring. Hence, it triggers a debugobjects warning when offlining CPUs because those timer debugobjects are never initialized.

Quoted 4 lines
Quoted 1 line

ODEBUG:..."

Looks better to me. Do you want me to send a v2 for it or you could update it for merging?

Could you please send v2? I am not the one that provides final approval for inclusion nor the one that will take care of merging afterwards.

Thank you very much

Reinette

#7 lore raw
In reply to Qian Cai
Quoted 11 lines · Qian Cai

Here it only check L3, so it will skip correctly for L3DATA and L3CODE to not call cancel_delayed_work(). Recalled the above that only L3 will have r->capable set.

if (r == &rdt_resources_all[RDT_RESOURCE_L3]) {
        if (is_mbm_enabled() && cpu == d->mbm_work_cpu) {
                cancel_delayed_work(&d->mbm_over);

Hence, r->mon_capable check seems redundant here.

I see. Thanks for explaining.

Signature · 2 lines
thanks,
Ryan