perf/hw_breakpoint: Optimize for thousands of tasks
Subsystems (7)
v4 cover perf/hw_breakpoint: Optimize for thousands of tasks
The hw_breakpoint subsystem's code has seen little change in over 10 years. In that time, systems with >100s of CPUs have become common, along with improvements to the perf subsystem: using breakpoints on thousands of concurrent tasks should be a supported usecase.
The breakpoint constraints accounting algorithm is the major bottleneck in doing so:
1. toggle_bp_slot() and fetch_bp_busy_slots() are O(#cpus * #tasks):
Both iterate through all CPUs and call task_bp_pinned(), which is O(#tasks).
2. Everything is serialized on a global mutex, 'nr_bp_mutex'.
The series progresses with the simpler optimizations and finishes with the more complex optimizations:
1. We first optimize task_bp_pinned() to only take O(1) on average.
2. Rework synchronization to allow concurrency when checking and updating breakpoint constraints for tasks.
3. Eliminate the O(#cpus) loops in the CPU-independent case.
Along the way, smaller micro-optimizations and cleanups are done as they seemed obvious when staring at the code (but likely insignificant).
The result is (on a system with 256 CPUs) that we go from:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64
[ ^ more aggressive benchmark parameters took too long ]
| # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 236.418 [sec] | | 123134.794271 usecs/op | 7880626.833333 usecs/op/cpu
... to the following with all optimizations:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.067 [sec] | | 35.292187 usecs/op | 2258.700000 usecs/op/cpu
On the used test system, that's an effective speedup of ~3490x per op.
Which is on par with the theoretical ideal performance through optimizations in hw_breakpoint.c (constraints accounting disabled), and only 12% slower than no breakpoints at all.
Changelog
---------
v4:
* Fix percpu_is_read_locked(): Due to spurious read_count increments in __percpu_down_read_trylock() if sem->block != 0, check that !sem->block (reported by Peter).
* Apply Reviewed/Acked-by.
v3: https://lkml.kernel.org/r/[email protected]
* Fix typos.
* Introduce hw_breakpoint_is_used() for the test.
* Add WARN_ON in bp_blots_histogram_add().
* Don't use raw_smp_processor_id() in test.
* Apply Acked-by/Reviewed-by given in v2 for mostly unchanged patches.
v2: https://lkml.kernel.org/r/[email protected]
* Add KUnit test suite.
* Remove struct bp_busy_slots and simplify functions.
* Add "powerpc/hw_breakpoint: Avoid relying on caller synchronization".
* Add "locking/percpu-rwsem: Add percpu_is_write_locked() and percpu_is_read_locked()".
* Use percpu-rwsem instead of rwlock.
* Use task_struct::perf_event_mutex instead of sharded mutex.
* Drop v1 "perf/hw_breakpoint: Optimize task_bp_pinned() if CPU-independent".
* Add "perf/hw_breakpoint: Introduce bp_slots_histogram".
* Add "perf/hw_breakpoint: Optimize max_bp_pinned_slots() for CPU-independent task targets".
* Add "perf/hw_breakpoint: Optimize toggle_bp_slot() for CPU-independent task targets".
* Apply Acked-by/Reviewed-by given in v1 for unchanged patches.
==> Speedup of ~3490x (vs. ~3315x in v1).
v1: https://lore.kernel.org/all/[email protected]/
Marco Elver (14):
perf/hw_breakpoint: Add KUnit test for constraints accounting perf/hw_breakpoint: Provide hw_breakpoint_is_used() and use in test perf/hw_breakpoint: Clean up headers
perf/hw_breakpoint: Optimize list of per-task breakpoints perf/hw_breakpoint: Mark data __ro_after_init perf/hw_breakpoint: Optimize constant number of breakpoint slots perf/hw_breakpoint: Make hw_breakpoint_weight() inlinable perf/hw_breakpoint: Remove useless code related to flexible
breakpoints
powerpc/hw_breakpoint: Avoid relying on caller synchronization locking/percpu-rwsem: Add percpu_is_write_locked() and
percpu_is_read_locked()
perf/hw_breakpoint: Reduce contention with large number of tasks perf/hw_breakpoint: Introduce bp_slots_histogram perf/hw_breakpoint: Optimize max_bp_pinned_slots() for CPU-independent
task targets
perf/hw_breakpoint: Optimize toggle_bp_slot() for CPU-independent task
targets
arch/powerpc/kernel/hw_breakpoint.c | 53 ++- arch/sh/include/asm/hw_breakpoint.h | 5 +- arch/x86/include/asm/hw_breakpoint.h | 5 +- include/linux/hw_breakpoint.h | 4 +- include/linux/percpu-rwsem.h | 6 + include/linux/perf_event.h | 3 +- kernel/events/Makefile | 1 + kernel/events/hw_breakpoint.c | 638 ++++++++++++++++++++------- kernel/events/hw_breakpoint_test.c | 333 ++++++++++++++ kernel/locking/percpu-rwsem.c | 6 + lib/Kconfig.debug | 10 + 11 files changed, 885 insertions(+), 179 deletions(-) create mode 100644 kernel/events/hw_breakpoint_test.c
Signature · 1 line
2.37.2.672.g94769d06f0-goog
PATCH v4 1/14 perf/hw_breakpoint: Add KUnit test for constraints accounting
Add KUnit test for hw_breakpoint constraints accounting, with various interesting mixes of breakpoint targets (some care was taken to catch interesting corner cases via bug-injection).
The test cannot be built as a module because it requires access to hw_breakpoint_slots(), which is not inlinable or exported on all architectures.
---
v3:
* Don't use raw_smp_processor_id().
v2:
* New patch.
--- kernel/events/Makefile | 1 + kernel/events/hw_breakpoint_test.c | 323 +++++++++++++++++++++++++++++ lib/Kconfig.debug | 10 + 3 files changed, 334 insertions(+) create mode 100644 kernel/events/hw_breakpoint_test.c
kernel/events/Makefile +1 -0
@@ -2,4 +2,5 @@ obj-y := core.o ring_buffer.o callchain.o obj-$(CONFIG_HAVE_HW_BREAKPOINT) += hw_breakpoint.o +obj-$(CONFIG_HW_BREAKPOINT_KUNIT_TEST) += hw_breakpoint_test.o obj-$(CONFIG_UPROBES) += uprobes.o
kernel/events/hw_breakpoint_test.c new file +323 -0
323 lines not on this page · raw message
lib/Kconfig.debug +10 -0
@@ -2533,6 +2533,16 @@ config STACKINIT_KUNIT_TEST CONFIG_GCC_PLUGIN_STRUCTLEAK, CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF, or CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF_ALL. +config HW_BREAKPOINT_KUNIT_TEST + bool "Test hw_breakpoint constraints accounting" if !KUNIT_ALL_TESTS + depends on HAVE_HW_BREAKPOINT + depends on KUNIT=y + default KUNIT_ALL_TESTS + help + Tests for hw_breakpoint constraints accounting. + + If unsure, say N. + config TEST_UDELAY tristate "udelay test driver" help
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Add KUnit test for constraints accounting
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 724c299c6a0e412b5679d7ebb9b3f4e00bd2aa78 Gitweb: https://git.kernel.org/tip/724c299c6a0e412b5679d7ebb9b3f4e00bd2aa78 Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:06 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:20 +02:00
perf/hw_breakpoint: Add KUnit test for constraints accounting
Add KUnit test for hw_breakpoint constraints accounting, with various interesting mixes of breakpoint targets (some care was taken to catch interesting corner cases via bug-injection).
The test cannot be built as a module because it requires access to hw_breakpoint_slots(), which is not inlinable or exported on all architectures.
--- kernel/events/Makefile | 1 +- kernel/events/hw_breakpoint_test.c | 323 ++++++++++++++++++++++++++++- lib/Kconfig.debug | 10 +- 3 files changed, 334 insertions(+) create mode 100644 kernel/events/hw_breakpoint_test.c
kernel/events/Makefile +1 -0
@@ -2,4 +2,5 @@ obj-y := core.o ring_buffer.o callchain.o obj-$(CONFIG_HAVE_HW_BREAKPOINT) += hw_breakpoint.o +obj-$(CONFIG_HW_BREAKPOINT_KUNIT_TEST) += hw_breakpoint_test.o obj-$(CONFIG_UPROBES) += uprobes.o
kernel/events/hw_breakpoint_test.c new file +323 -0
323 lines not on this page · raw message
lib/Kconfig.debug +10 -0
@@ -2530,6 +2530,16 @@ config STACKINIT_KUNIT_TEST CONFIG_GCC_PLUGIN_STRUCTLEAK, CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF, or CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF_ALL. +config HW_BREAKPOINT_KUNIT_TEST + bool "Test hw_breakpoint constraints accounting" if !KUNIT_ALL_TESTS + depends on HAVE_HW_BREAKPOINT + depends on KUNIT=y + default KUNIT_ALL_TESTS + help + Tests for hw_breakpoint constraints accounting. + + If unsure, say N. + config TEST_UDELAY tristate "udelay test driver" help
PATCH v4 2/14 perf/hw_breakpoint: Provide hw_breakpoint_is_used() and use in test
In reply to Marco ElverProvide hw_breakpoint_is_used() to check if breakpoints are in use on the system.
Use it in the KUnit test to verify the global state before and after a test case.
---
v3:
* New patch.
--- include/linux/hw_breakpoint.h | 3 +++ kernel/events/hw_breakpoint.c | 29 +++++++++++++++++++++++++++++ kernel/events/hw_breakpoint_test.c | 12 +++++++++++- 3 files changed, 43 insertions(+), 1 deletion(-)
include/linux/hw_breakpoint.h +3 -0
@@ -74,6 +74,7 @@ register_wide_hw_breakpoint(struct perf_event_attr *attr,extern int register_perf_hw_breakpoint (struct perf_event *bp);extern void unregister_hw_breakpoint (struct perf_event *bp);extern void unregister_wide_hw_breakpoint (struct perf_event *__percpu *cpu_events); +extern bool hw_breakpoint_is_used (void );extern int dbg_reserve_bp_slot (struct perf_event *bp);extern int dbg_release_bp_slot (struct perf_event *bp);
@@ -121,6 +122,8 @@ register_perf_hw_breakpoint(struct perf_event *bp) { return -ENOSYS; }static inline void unregister_hw_breakpoint (struct perf_event *bp) { }static inline void unregister_wide_hw_breakpoint (struct perf_event *__percpu *cpu_events) { } +static inline bool hw_breakpoint_is_used (void ) {return false; } +static inline int reserve_bp_slot (struct perf_event *bp) {return -ENOSYS ; }static inline void release_bp_slot (struct perf_event *bp) { }
kernel/events/hw_breakpoint.c +29 -0
@@ -604,6 +604,35 @@ void unregister_wide_hw_breakpoint(struct perf_event * __percpu *cpu_events) }EXPORT_SYMBOL_GPL (unregister_wide_hw_breakpoint); +/** +* hw_breakpoint_is_used - check if breakpoints are currently used +* +* Returns: true if breakpoints are used, false otherwise. +*/ +bool hw_breakpoint_is_used (void ) +{ +int cpu; + +if (!constraints_initialized) +return false; + +for_each_possible_cpu (cpu) { +for (int type =0 ; type <TYPE_MAX ; ++type) { +struct bp_cpuinfo *info =get_bp_info (cpu, type); + +if (info->cpu_pinned) +return true; + +for (int slot =0 ; slot < nr_slots[type]; ++slot) { +if (info->tsk_pinned[slot]) +return true; + } + } + } + +return false; +} +static struct notifier_block hw_breakpoint_exceptions_nb = { .notifier_call = hw_breakpoint_exceptions_notify,/* we need to be notified first */
kernel/events/hw_breakpoint_test.c +11 -1
@@ -294,7 +294,14 @@ static struct kunit_case hw_breakpoint_test_cases[] = {static int test_init (struct kunit *test) {/* Most test cases want 2 distinct CPUs. */ -return num_online_cpus () <2 ? -EINVAL :0 ; +if (num_online_cpus () <2 ) +return -EINVAL ; + +/* Want the system to not use breakpoints elsewhere. */ +if (hw_breakpoint_is_used ()) +return -EBUSY ; + +return 0 ; }static void test_exit (struct kunit *test)
@@ -308,6 +315,9 @@ static void test_exit(struct kunit *test)kthread_stop (__other_task); __other_task =NULL ; } + +/* Verify that internal state agrees that no breakpoints are in use. */ +KUNIT_EXPECT_FALSE (test,hw_breakpoint_is_used ()); }static struct kunit_suite hw_breakpoint_test_suite = {
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Provide hw_breakpoint_is_used() and use in test
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: c5b81449f915a28bb9c7725e53aebab3ba39b4a2 Gitweb: https://git.kernel.org/tip/c5b81449f915a28bb9c7725e53aebab3ba39b4a2 Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:07 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:20 +02:00
perf/hw_breakpoint: Provide hw_breakpoint_is_used() and use in test
Provide hw_breakpoint_is_used() to check if breakpoints are in use on the system.
Use it in the KUnit test to verify the global state before and after a test case.
--- include/linux/hw_breakpoint.h | 3 +++- kernel/events/hw_breakpoint.c | 29 +++++++++++++++++++++++++++++- kernel/events/hw_breakpoint_test.c | 12 +++++++++++- 3 files changed, 43 insertions(+), 1 deletion(-)
include/linux/hw_breakpoint.h +3 -0
@@ -74,6 +74,7 @@ register_wide_hw_breakpoint(struct perf_event_attr *attr,extern int register_perf_hw_breakpoint (struct perf_event *bp);extern void unregister_hw_breakpoint (struct perf_event *bp);extern void unregister_wide_hw_breakpoint (struct perf_event *__percpu *cpu_events); +extern bool hw_breakpoint_is_used (void );extern int dbg_reserve_bp_slot (struct perf_event *bp);extern int dbg_release_bp_slot (struct perf_event *bp);
@@ -121,6 +122,8 @@ register_perf_hw_breakpoint(struct perf_event *bp) { return -ENOSYS; }static inline void unregister_hw_breakpoint (struct perf_event *bp) { }static inline void unregister_wide_hw_breakpoint (struct perf_event *__percpu *cpu_events) { } +static inline bool hw_breakpoint_is_used (void ) {return false; } +static inline int reserve_bp_slot (struct perf_event *bp) {return -ENOSYS ; }static inline void release_bp_slot (struct perf_event *bp) { }
kernel/events/hw_breakpoint.c +29 -0
@@ -604,6 +604,35 @@ void unregister_wide_hw_breakpoint(struct perf_event * __percpu *cpu_events) }EXPORT_SYMBOL_GPL (unregister_wide_hw_breakpoint); +/** +* hw_breakpoint_is_used - check if breakpoints are currently used +* +* Returns: true if breakpoints are used, false otherwise. +*/ +bool hw_breakpoint_is_used (void ) +{ +int cpu; + +if (!constraints_initialized) +return false; + +for_each_possible_cpu (cpu) { +for (int type =0 ; type <TYPE_MAX ; ++type) { +struct bp_cpuinfo *info =get_bp_info (cpu, type); + +if (info->cpu_pinned) +return true; + +for (int slot =0 ; slot < nr_slots[type]; ++slot) { +if (info->tsk_pinned[slot]) +return true; + } + } + } + +return false; +} +static struct notifier_block hw_breakpoint_exceptions_nb = { .notifier_call = hw_breakpoint_exceptions_notify,/* we need to be notified first */
kernel/events/hw_breakpoint_test.c +11 -1
@@ -294,7 +294,14 @@ static struct kunit_case hw_breakpoint_test_cases[] = {static int test_init (struct kunit *test) {/* Most test cases want 2 distinct CPUs. */ -return num_online_cpus () <2 ? -EINVAL :0 ; +if (num_online_cpus () <2 ) +return -EINVAL ; + +/* Want the system to not use breakpoints elsewhere. */ +if (hw_breakpoint_is_used ()) +return -EBUSY ; + +return 0 ; }static void test_exit (struct kunit *test)
@@ -308,6 +315,9 @@ static void test_exit(struct kunit *test)kthread_stop (__other_task); __other_task =NULL ; } + +/* Verify that internal state agrees that no breakpoints are in use. */ +KUNIT_EXPECT_FALSE (test,hw_breakpoint_is_used ()); }static struct kunit_suite hw_breakpoint_test_suite = {
PATCH v4 3/14 perf/hw_breakpoint: Clean up headers
In reply to Marco ElverClean up headers:
- Remove unused <linux/kallsyms.h>
- Remove unused <linux/kprobes.h>
- Remove unused <linux/module.h>
- Remove unused <linux/smp.h>
- Add <linux/export.h> for EXPORT_SYMBOL_GPL().
- Add <linux/mutex.h> for mutex.
- Sort alphabetically.
- Move <linux/hw_breakpoint.h> to top to test it compiles on its own.
---
v2:
* Move to start of series.
--- kernel/events/hw_breakpoint.c | 19 +++++++++---------- 1 file changed, 9 insertions(+), 10 deletions(-)
kernel/events/hw_breakpoint.c +9 -10
@@ -17,23 +17,22 @@* This file contains the arch-independent routines. */ +#include <linux/hw_breakpoint.h> + +#include <linux/bug.h> +#include <linux/cpu.h> +#include <linux/export.h> +#include <linux/init.h> #include <linux/irqflags.h> -#include <linux/kallsyms.h> -#include <linux/notifier.h> -#include <linux/kprobes.h> #include <linux/kdebug.h> #include <linux/kernel.h> -#include <linux/module.h> +#include <linux/list.h> +#include <linux/mutex.h> +#include <linux/notifier.h> #include <linux/percpu.h> #include <linux/sched.h> -#include <linux/init.h> #include <linux/slab.h> -#include <linux/list.h> -#include <linux/cpu.h> -#include <linux/smp.h> -#include <linux/bug.h> -#include <linux/hw_breakpoint.h> /* * Constraints data */
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Clean up headers
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 089cdcb0cd1c25343fa56d3eabbe878df31a7c0e Gitweb: https://git.kernel.org/tip/089cdcb0cd1c25343fa56d3eabbe878df31a7c0e Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:08 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:21 +02:00
perf/hw_breakpoint: Clean up headers
Clean up headers:
- Remove unused <linux/kallsyms.h>
- Remove unused <linux/kprobes.h>
- Remove unused <linux/module.h>
- Remove unused <linux/smp.h>
- Add <linux/export.h> for EXPORT_SYMBOL_GPL().
- Add <linux/mutex.h> for mutex.
- Sort alphabetically.
- Move <linux/hw_breakpoint.h> to top to test it compiles on its own.
--- kernel/events/hw_breakpoint.c | 19 +++++++++---------- 1 file changed, 9 insertions(+), 10 deletions(-)
kernel/events/hw_breakpoint.c +9 -10
@@ -17,23 +17,22 @@* This file contains the arch-independent routines. */ +#include <linux/hw_breakpoint.h> + +#include <linux/bug.h> +#include <linux/cpu.h> +#include <linux/export.h> +#include <linux/init.h> #include <linux/irqflags.h> -#include <linux/kallsyms.h> -#include <linux/notifier.h> -#include <linux/kprobes.h> #include <linux/kdebug.h> #include <linux/kernel.h> -#include <linux/module.h> +#include <linux/list.h> +#include <linux/mutex.h> +#include <linux/notifier.h> #include <linux/percpu.h> #include <linux/sched.h> -#include <linux/init.h> #include <linux/slab.h> -#include <linux/list.h> -#include <linux/cpu.h> -#include <linux/smp.h> -#include <linux/bug.h> -#include <linux/hw_breakpoint.h> /* * Constraints data */
PATCH v4 4/14 perf/hw_breakpoint: Optimize list of per-task breakpoints
In reply to Marco ElverOn a machine with 256 CPUs, running the recently added perf breakpoint benchmark results in:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 236.418 [sec] | | 123134.794271 usecs/op | 7880626.833333 usecs/op/cpu
The benchmark tests inherited breakpoint perf events across many threads.
Looking at a perf profile, we can see that the majority of the time is spent in various hw_breakpoint.c functions, which execute within the 'nr_bp_mutex' critical sections which then results in contention on that mutex as well:
37.27% [kernel] [k] osq_lock
34.92% [kernel] [k] mutex_spin_on_owner
12.15% [kernel] [k] toggle_bp_slot
11.90% [kernel] [k] __reserve_bp_slotThe culprit here is task_bp_pinned(), which has a runtime complexity of O(#tasks) due to storing all task breakpoints in the same list and iterating through that list looking for a matching task. Clearly, this does not scale to thousands of tasks.
Instead, make use of the "rhashtable" variant "rhltable" which stores multiple items with the same key in a list. This results in average runtime complexity of O(1) for task_bp_pinned().
With the optimization, the benchmark shows:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.208 [sec] | | 108.422396 usecs/op | 6939.033333 usecs/op/cpu
On this particular setup that's a speedup of ~1135x.
While one option would be to make task_struct a breakpoint list node, this would only further bloat task_struct for infrequently used data. Furthermore, after all optimizations in this series, there's no evidence it would result in better performance: later optimizations make the time spent looking up entries in the hash table negligible (we'll reach the theoretical ideal performance i.e. no constraints).
---
v2:
* Commit message tweaks.
--- include/linux/perf_event.h | 3 +- kernel/events/hw_breakpoint.c | 56 ++++++++++++++++++++++------------- 2 files changed, 37 insertions(+), 22 deletions(-)
include/linux/perf_event.h +2 -1
@@ -36,6 +36,7 @@ struct perf_guest_info_callbacks { };#ifdef CONFIG_HAVE_HW_BREAKPOINT +#include <linux/rhashtable-types.h> #include <asm/hw_breakpoint.h> #endif
@@ -178,7 +179,7 @@ struct hw_perf_event {* creation and event initalization. */ struct arch_hw_breakpoint info; -struct list_head bp_list; +struct rhlist_head bp_list; };#endif struct {/* amd_iommu */
kernel/events/hw_breakpoint.c +35 -21
140 lines not on this page · raw message
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Optimize list of per-task breakpoints
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 0370dc314df35579b751d1b77c9169f071444962 Gitweb: https://git.kernel.org/tip/0370dc314df35579b751d1b77c9169f071444962 Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:09 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:21 +02:00
perf/hw_breakpoint: Optimize list of per-task breakpoints
On a machine with 256 CPUs, running the recently added perf breakpoint benchmark results in:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 236.418 [sec] | | 123134.794271 usecs/op | 7880626.833333 usecs/op/cpu
The benchmark tests inherited breakpoint perf events across many threads.
Looking at a perf profile, we can see that the majority of the time is spent in various hw_breakpoint.c functions, which execute within the 'nr_bp_mutex' critical sections which then results in contention on that mutex as well:
37.27% [kernel] [k] osq_lock
34.92% [kernel] [k] mutex_spin_on_owner
12.15% [kernel] [k] toggle_bp_slot
11.90% [kernel] [k] __reserve_bp_slotThe culprit here is task_bp_pinned(), which has a runtime complexity of O(#tasks) due to storing all task breakpoints in the same list and iterating through that list looking for a matching task. Clearly, this does not scale to thousands of tasks.
Instead, make use of the "rhashtable" variant "rhltable" which stores multiple items with the same key in a list. This results in average runtime complexity of O(1) for task_bp_pinned().
With the optimization, the benchmark shows:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.208 [sec] | | 108.422396 usecs/op | 6939.033333 usecs/op/cpu
On this particular setup that's a speedup of ~1135x.
While one option would be to make task_struct a breakpoint list node, this would only further bloat task_struct for infrequently used data. Furthermore, after all optimizations in this series, there's no evidence it would result in better performance: later optimizations make the time spent looking up entries in the hash table negligible (we'll reach the theoretical ideal performance i.e. no constraints).
--- include/linux/perf_event.h | 3 +- kernel/events/hw_breakpoint.c | 56 +++++++++++++++++++++------------- 2 files changed, 37 insertions(+), 22 deletions(-)
include/linux/perf_event.h +2 -1
@@ -36,6 +36,7 @@ struct perf_guest_info_callbacks { };#ifdef CONFIG_HAVE_HW_BREAKPOINT +#include <linux/rhashtable-types.h> #include <asm/hw_breakpoint.h> #endif
@@ -178,7 +179,7 @@ struct hw_perf_event {* creation and event initalization. */ struct arch_hw_breakpoint info; -struct list_head bp_list; +struct rhlist_head bp_list; };#endif struct {/* amd_iommu */
kernel/events/hw_breakpoint.c +35 -21
140 lines not on this page · raw message
PATCH v4 5/14 perf/hw_breakpoint: Mark data __ro_after_init
In reply to Marco ElverMark read-only data after initialization as __ro_after_init.
While we are here, turn 'constraints_initialized' into a bool.
--- kernel/events/hw_breakpoint.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
kernel/events/hw_breakpoint.c +3 -3
@@ -46,7 +46,7 @@ struct bp_cpuinfo { }; static DEFINE_PER_CPU(struct bp_cpuinfo, bp_cpuinfo[TYPE_MAX]); -static int nr_slots[TYPE_MAX ]; +static int nr_slots[TYPE_MAX ]__ro_after_init ;static struct bp_cpuinfo *get_bp_info (int cpu,enum bp_type_idx type) {
@@ -62,7 +62,7 @@ static const struct rhashtable_params task_bps_ht_params = { .automatic_shrinking = true, }; -static int constraints_initialized; +static bool constraints_initialized__ro_after_init ;/* Gather the number of total pinned and un-pinned bp in a cpuset */ struct bp_busy_slots {
@@ -739,7 +739,7 @@ int __init init_hw_breakpoint(void)if (ret) goto err; - constraints_initialized =1 ; + constraints_initialized = true;perf_pmu_register (&perf_breakpoint,"breakpoint" ,PERF_TYPE_BREAKPOINT );
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Mark data __ro_after_init
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: db5f6f853194c5e02d8551425b5e86b7e0b81806 Gitweb: https://git.kernel.org/tip/db5f6f853194c5e02d8551425b5e86b7e0b81806 Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:10 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:21 +02:00
perf/hw_breakpoint: Mark data __ro_after_init
Mark read-only data after initialization as __ro_after_init.
While we are here, turn 'constraints_initialized' into a bool.
--- kernel/events/hw_breakpoint.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
kernel/events/hw_breakpoint.c +3 -3
@@ -46,7 +46,7 @@ struct bp_cpuinfo { }; static DEFINE_PER_CPU(struct bp_cpuinfo, bp_cpuinfo[TYPE_MAX]); -static int nr_slots[TYPE_MAX ]; +static int nr_slots[TYPE_MAX ]__ro_after_init ;static struct bp_cpuinfo *get_bp_info (int cpu,enum bp_type_idx type) {
@@ -62,7 +62,7 @@ static const struct rhashtable_params task_bps_ht_params = { .automatic_shrinking = true, }; -static int constraints_initialized; +static bool constraints_initialized__ro_after_init ;/* Gather the number of total pinned and un-pinned bp in a cpuset */ struct bp_busy_slots {
@@ -739,7 +739,7 @@ int __init init_hw_breakpoint(void)if (ret) goto err; - constraints_initialized =1 ; + constraints_initialized = true;perf_pmu_register (&perf_breakpoint,"breakpoint" ,PERF_TYPE_BREAKPOINT );
PATCH v4 6/14 perf/hw_breakpoint: Optimize constant number of breakpoint slots
In reply to Marco ElverOptimize internal hw_breakpoint state if the architecture's number of breakpoint slots is constant. This avoids several kmalloc() calls and potentially unnecessary failures if the allocations fail, as well as subtly improves code generation and cache locality.
The protocol is that if an architecture defines hw_breakpoint_slots via the preprocessor, it must be constant and the same for all types.
--- arch/sh/include/asm/hw_breakpoint.h | 5 +- arch/x86/include/asm/hw_breakpoint.h | 5 +- kernel/events/hw_breakpoint.c | 94 ++++++++++++++++++---------- 3 files changed, 63 insertions(+), 41 deletions(-)
arch/sh/include/asm/hw_breakpoint.h +1 -4
@@ -48,10 +48,7 @@ struct pmu;/* Maximum number of UBC channels */ #define HBP_NUM 2 -static inline int hw_breakpoint_slots (int type) -{ -return HBP_NUM ; -} +#define hw_breakpoint_slots (type) (HBP_NUM)/* arch/sh/kernel/hw_breakpoint.c */ extern int arch_check_bp_in_kernelspace (struct arch_hw_breakpoint *hw);
arch/x86/include/asm/hw_breakpoint.h +1 -4
@@ -44,10 +44,7 @@ struct arch_hw_breakpoint {/* Total number of available HW breakpoint registers */ #define HBP_NUM 4 -static inline int hw_breakpoint_slots (int type) -{ -return HBP_NUM ; -} +#define hw_breakpoint_slots (type) (HBP_NUM)struct perf_event_attr ;struct perf_event ;
kernel/events/hw_breakpoint.c +61 -33
143 lines not on this page · raw message
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Optimize constant number of breakpoint slots
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: be3f152568cc7f5f573d21d5f86a2c4f3cc047ab Gitweb: https://git.kernel.org/tip/be3f152568cc7f5f573d21d5f86a2c4f3cc047ab Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:11 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:22 +02:00
perf/hw_breakpoint: Optimize constant number of breakpoint slots
Optimize internal hw_breakpoint state if the architecture's number of breakpoint slots is constant. This avoids several kmalloc() calls and potentially unnecessary failures if the allocations fail, as well as subtly improves code generation and cache locality.
The protocol is that if an architecture defines hw_breakpoint_slots via the preprocessor, it must be constant and the same for all types.
--- arch/sh/include/asm/hw_breakpoint.h | 5 +- arch/x86/include/asm/hw_breakpoint.h | 5 +- kernel/events/hw_breakpoint.c | 94 +++++++++++++++++---------- 3 files changed, 63 insertions(+), 41 deletions(-)
arch/sh/include/asm/hw_breakpoint.h +1 -4
@@ -48,10 +48,7 @@ struct pmu;/* Maximum number of UBC channels */ #define HBP_NUM 2 -static inline int hw_breakpoint_slots (int type) -{ -return HBP_NUM ; -} +#define hw_breakpoint_slots (type) (HBP_NUM)/* arch/sh/kernel/hw_breakpoint.c */ extern int arch_check_bp_in_kernelspace (struct arch_hw_breakpoint *hw);
arch/x86/include/asm/hw_breakpoint.h +1 -4
@@ -44,10 +44,7 @@ struct arch_hw_breakpoint {/* Total number of available HW breakpoint registers */ #define HBP_NUM 4 -static inline int hw_breakpoint_slots (int type) -{ -return HBP_NUM ; -} +#define hw_breakpoint_slots (type) (HBP_NUM)struct perf_event_attr ;struct perf_event ;
kernel/events/hw_breakpoint.c +61 -33
143 lines not on this page · raw message
PATCH v4 7/14 perf/hw_breakpoint: Make hw_breakpoint_weight() inlinable
In reply to Marco ElverDue to being a __weak function, hw_breakpoint_weight() will cause the compiler to always emit a call to it. This generates unnecessarily bad code (register spills etc.) for no good reason; in fact it appears in profiles of perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512:
...
0.70% [kernel] [k] hw_breakpoint_weight
...While a small percentage, no architecture defines its own hw_breakpoint_weight() nor are there users outside hw_breakpoint.c, which makes the fact it is currently __weak a poor choice.
Change hw_breakpoint_weight()'s definition to follow a similar protocol to hw_breakpoint_slots(), such that if <asm/hw_breakpoint.h> defines hw_breakpoint_weight(), we'll use it instead.
The result is that it is inlined and no longer shows up in profiles.
--- include/linux/hw_breakpoint.h | 1 - kernel/events/hw_breakpoint.c | 4 +++- 2 files changed, 3 insertions(+), 2 deletions(-)
include/linux/hw_breakpoint.h +0 -1
@@ -80,7 +80,6 @@ extern int dbg_reserve_bp_slot(struct perf_event *bp);extern int dbg_release_bp_slot (struct perf_event *bp);extern int reserve_bp_slot (struct perf_event *bp);extern void release_bp_slot (struct perf_event *bp); -int hw_breakpoint_weight (struct perf_event *bp);int arch_reserve_bp_slot (struct perf_event *bp);void arch_release_bp_slot (struct perf_event *bp);void arch_unregister_hw_breakpoint (struct perf_event *bp);
kernel/events/hw_breakpoint.c +3 -1
@@ -124,10 +124,12 @@ static __init int init_breakpoint_slots(void) }#endif -__weak int hw_breakpoint_weight (struct perf_event *bp) +#ifndef hw_breakpoint_weight +static inline int hw_breakpoint_weight (struct perf_event *bp) {return 1 ; } +#endif static inline enum bp_type_idx find_slot_idx (u64 bp_type) {
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Make hw_breakpoint_weight() inlinable
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 9caf87be118f4639537404eeb67dd444a3716e9a Gitweb: https://git.kernel.org/tip/9caf87be118f4639537404eeb67dd444a3716e9a Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:12 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:22 +02:00
perf/hw_breakpoint: Make hw_breakpoint_weight() inlinable
Due to being a __weak function, hw_breakpoint_weight() will cause the compiler to always emit a call to it. This generates unnecessarily bad code (register spills etc.) for no good reason; in fact it appears in profiles of perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512:
...
0.70% [kernel] [k] hw_breakpoint_weight
...While a small percentage, no architecture defines its own hw_breakpoint_weight() nor are there users outside hw_breakpoint.c, which makes the fact it is currently __weak a poor choice.
Change hw_breakpoint_weight()'s definition to follow a similar protocol to hw_breakpoint_slots(), such that if <asm/hw_breakpoint.h> defines hw_breakpoint_weight(), we'll use it instead.
The result is that it is inlined and no longer shows up in profiles.
--- include/linux/hw_breakpoint.h | 1 - kernel/events/hw_breakpoint.c | 4 +++- 2 files changed, 3 insertions(+), 2 deletions(-)
include/linux/hw_breakpoint.h +0 -1
@@ -80,7 +80,6 @@ extern int dbg_reserve_bp_slot(struct perf_event *bp);extern int dbg_release_bp_slot (struct perf_event *bp);extern int reserve_bp_slot (struct perf_event *bp);extern void release_bp_slot (struct perf_event *bp); -int hw_breakpoint_weight (struct perf_event *bp);int arch_reserve_bp_slot (struct perf_event *bp);void arch_release_bp_slot (struct perf_event *bp);void arch_unregister_hw_breakpoint (struct perf_event *bp);
kernel/events/hw_breakpoint.c +3 -1
@@ -124,10 +124,12 @@ err: }#endif -__weak int hw_breakpoint_weight (struct perf_event *bp) +#ifndef hw_breakpoint_weight +static inline int hw_breakpoint_weight (struct perf_event *bp) {return 1 ; } +#endif static inline enum bp_type_idx find_slot_idx (u64 bp_type) {
PATCH v4 8/14 perf/hw_breakpoint: Remove useless code related to flexible breakpoints
In reply to Marco ElverFlexible breakpoints have never been implemented, with bp_cpuinfo::flexible always being 0. Unfortunately, they still occupy 4 bytes in each bp_cpuinfo and bp_busy_slots, as well as computing the max flexible count in fetch_bp_busy_slots().
This again causes suboptimal code generation, when we always know that !!slots.flexible will be 0.
Just get rid of the flexible "placeholder" and remove all real code related to it. Make a note in the comment related to the constraints algorithm but don't remove them from the algorithm, so that if in future flexible breakpoints need supporting, it should be trivial to revive them (along with reverting this change).
---
v2:
* Also remove struct bp_busy_slots, and simplify functions.
--- kernel/events/hw_breakpoint.c | 57 +++++++++++------------------------ 1 file changed, 17 insertions(+), 40 deletions(-)
kernel/events/hw_breakpoint.c +17 -40
@@ -45,8 +45,6 @@ struct bp_cpuinfo {#else unsigned int *tsk_pinned;#endif -/* Number of non-pinned cpu/task breakpoints in a cpu */ -unsigned int flexible;/* XXX: placeholder, see fetch_this_slot() */ }; static DEFINE_PER_CPU(struct bp_cpuinfo, bp_cpuinfo[TYPE_MAX]);
@@ -67,12 +65,6 @@ static const struct rhashtable_params task_bps_ht_params = {static bool constraints_initialized__ro_after_init ; -/* Gather the number of total pinned and un-pinned bp in a cpuset */ -struct bp_busy_slots { -unsigned int pinned; -unsigned int flexible; -}; -/* Serialize accesses to the above constraints */ static DEFINE_MUTEX (nr_bp_mutex);
@@ -190,14 +182,14 @@ static const struct cpumask *cpumask_of_bp(struct perf_event *bp) }/* -* Report the number of pinned/un-pinned breakpoints we have in -* a given cpu (cpu > -1) or in all of them (cpu = -1). +* Returns the max pinned breakpoint slots in a given +* CPU (cpu > -1) or across all of them (cpu = -1). */ -static void -fetch_bp_busy_slots (struct bp_busy_slots *slots,struct perf_event *bp, -enum bp_type_idx type) +static int +max_bp_pinned_slots (struct perf_event *bp,enum bp_type_idx type) {const struct cpumask *cpumask =cpumask_of_bp (bp); +int pinned_slots =0 ;int cpu;for_each_cpu (cpu, cpumask) {
@@ -210,24 +202,10 @@ fetch_bp_busy_slots(struct bp_busy_slots *slots, struct perf_event *bp,else nr +=task_bp_pinned (cpu, bp, type); -if (nr > slots->pinned) - slots->pinned = nr; - - nr = info->flexible; -if (nr > slots->flexible) - slots->flexible = nr; + pinned_slots =max (nr, pinned_slots); } -} -/* -* For now, continue to consider flexible as pinned, until we can -* ensure no flexible event can ever be scheduled before a pinned event -* in a same cpu. -*/ -static void -fetch_this_slot (struct bp_busy_slots *slots,int weight) -{ - slots->pinned += weight; +return pinned_slots; }/*
@@ -298,7 +276,12 @@ __weak void arch_unregister_hw_breakpoint(struct perf_event *bp) }/* -* Constraints to check before allowing this new breakpoint counter: +* Constraints to check before allowing this new breakpoint counter. +* +* Note: Flexible breakpoints are currently unimplemented, but outlined in the +* below algorithm for completeness. The implementation treats flexible as +* pinned due to no guarantee that we currently always schedule flexible events +* before a pinned event in a same CPU. * * == Non-pinned counter == (Considered as pinned for now) *
@@ -340,8 +323,8 @@ __weak void arch_unregister_hw_breakpoint(struct perf_event *bp)*/ static int __reserve_bp_slot (struct perf_event *bp,u64 bp_type) { -struct bp_busy_slots slots = {0 };enum bp_type_idx type; +int max_pinned_slots;int weight;int ret;
@@ -357,15 +340,9 @@ static int __reserve_bp_slot(struct perf_event *bp, u64 bp_type) type =find_slot_idx (bp_type); weight =hw_breakpoint_weight (bp); -fetch_bp_busy_slots (&slots, bp, type); -/* -* Simulate the addition of this breakpoint to the constraints -* and see the result. -*/ -fetch_this_slot (&slots, weight); - -/* Flexible counters need to keep at least one slot */ -if (slots.pinned + (!!slots.flexible) >hw_breakpoint_slots_cached (type)) +/* Check if this new breakpoint can be satisfied across all CPUs. */ + max_pinned_slots =max_bp_pinned_slots (bp, type) + weight; +if (max_pinned_slots >hw_breakpoint_slots_cached (type))return -ENOSPC ; ret =arch_reserve_bp_slot (bp);
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Remove useless code related to flexible breakpoints
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 24198ad373ad1e30b638aa147142dc21ab5757e7 Gitweb: https://git.kernel.org/tip/24198ad373ad1e30b638aa147142dc21ab5757e7 Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:13 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:22 +02:00
perf/hw_breakpoint: Remove useless code related to flexible breakpoints
Flexible breakpoints have never been implemented, with bp_cpuinfo::flexible always being 0. Unfortunately, they still occupy 4 bytes in each bp_cpuinfo and bp_busy_slots, as well as computing the max flexible count in fetch_bp_busy_slots().
This again causes suboptimal code generation, when we always know that !!slots.flexible will be 0.
Just get rid of the flexible "placeholder" and remove all real code related to it. Make a note in the comment related to the constraints algorithm but don't remove them from the algorithm, so that if in future flexible breakpoints need supporting, it should be trivial to revive them (along with reverting this change).
--- kernel/events/hw_breakpoint.c | 57 ++++++++++------------------------ 1 file changed, 17 insertions(+), 40 deletions(-)
kernel/events/hw_breakpoint.c +17 -40
@@ -45,8 +45,6 @@ struct bp_cpuinfo {#else unsigned int *tsk_pinned;#endif -/* Number of non-pinned cpu/task breakpoints in a cpu */ -unsigned int flexible;/* XXX: placeholder, see fetch_this_slot() */ }; static DEFINE_PER_CPU(struct bp_cpuinfo, bp_cpuinfo[TYPE_MAX]);
@@ -67,12 +65,6 @@ static const struct rhashtable_params task_bps_ht_params = {static bool constraints_initialized__ro_after_init ; -/* Gather the number of total pinned and un-pinned bp in a cpuset */ -struct bp_busy_slots { -unsigned int pinned; -unsigned int flexible; -}; -/* Serialize accesses to the above constraints */ static DEFINE_MUTEX (nr_bp_mutex);
@@ -190,14 +182,14 @@ static const struct cpumask *cpumask_of_bp(struct perf_event *bp) }/* -* Report the number of pinned/un-pinned breakpoints we have in -* a given cpu (cpu > -1) or in all of them (cpu = -1). +* Returns the max pinned breakpoint slots in a given +* CPU (cpu > -1) or across all of them (cpu = -1). */ -static void -fetch_bp_busy_slots (struct bp_busy_slots *slots,struct perf_event *bp, -enum bp_type_idx type) +static int +max_bp_pinned_slots (struct perf_event *bp,enum bp_type_idx type) {const struct cpumask *cpumask =cpumask_of_bp (bp); +int pinned_slots =0 ;int cpu;for_each_cpu (cpu, cpumask) {
@@ -210,24 +202,10 @@ fetch_bp_busy_slots(struct bp_busy_slots *slots, struct perf_event *bp,else nr +=task_bp_pinned (cpu, bp, type); -if (nr > slots->pinned) - slots->pinned = nr; - - nr = info->flexible; -if (nr > slots->flexible) - slots->flexible = nr; + pinned_slots =max (nr, pinned_slots); } -} -/* -* For now, continue to consider flexible as pinned, until we can -* ensure no flexible event can ever be scheduled before a pinned event -* in a same cpu. -*/ -static void -fetch_this_slot (struct bp_busy_slots *slots,int weight) -{ - slots->pinned += weight; +return pinned_slots; }/*
@@ -298,7 +276,12 @@ __weak void arch_unregister_hw_breakpoint(struct perf_event *bp) }/* -* Constraints to check before allowing this new breakpoint counter: +* Constraints to check before allowing this new breakpoint counter. +* +* Note: Flexible breakpoints are currently unimplemented, but outlined in the +* below algorithm for completeness. The implementation treats flexible as +* pinned due to no guarantee that we currently always schedule flexible events +* before a pinned event in a same CPU. * * == Non-pinned counter == (Considered as pinned for now) *
@@ -340,8 +323,8 @@ __weak void arch_unregister_hw_breakpoint(struct perf_event *bp)*/ static int __reserve_bp_slot (struct perf_event *bp,u64 bp_type) { -struct bp_busy_slots slots = {0 };enum bp_type_idx type; +int max_pinned_slots;int weight;int ret;
@@ -357,15 +340,9 @@ static int __reserve_bp_slot(struct perf_event *bp, u64 bp_type) type =find_slot_idx (bp_type); weight =hw_breakpoint_weight (bp); -fetch_bp_busy_slots (&slots, bp, type); -/* -* Simulate the addition of this breakpoint to the constraints -* and see the result. -*/ -fetch_this_slot (&slots, weight); - -/* Flexible counters need to keep at least one slot */ -if (slots.pinned + (!!slots.flexible) >hw_breakpoint_slots_cached (type)) +/* Check if this new breakpoint can be satisfied across all CPUs. */ + max_pinned_slots =max_bp_pinned_slots (bp, type) + weight; +if (max_pinned_slots >hw_breakpoint_slots_cached (type))return -ENOSPC ; ret =arch_reserve_bp_slot (bp);
PATCH v4 9/14 powerpc/hw_breakpoint: Avoid relying on caller synchronization
In reply to Marco ElverInternal data structures (cpu_bps, task_bps) of powerpc's hw_breakpoint implementation have relied on nr_bp_mutex serializing access to them.
Before overhauling synchronization of kernel/events/hw_breakpoint.c, introduce 2 spinlocks to synchronize cpu_bps and task_bps respectively, thus avoiding reliance on callers synchronizing powerpc's hw_breakpoint.
---
v2:
* New patch.
--- arch/powerpc/kernel/hw_breakpoint.c | 53 ++++++++++++++++++++++------- 1 file changed, 40 insertions(+), 13 deletions(-)
arch/powerpc/kernel/hw_breakpoint.c +40 -13
@@ -15,6 +15,7 @@#include <linux/kernel.h> #include <linux/sched.h> #include <linux/smp.h> +#include <linux/spinlock.h> #include <linux/debugfs.h> #include <linux/init.h>
@@ -129,7 +130,14 @@ struct breakpoint {bool ptrace_bp; }; +/* +* While kernel/events/hw_breakpoint.c does its own synchronization, we cannot +* rely on it safely synchronizing internals here; however, we can rely on it +* not requesting more breakpoints than available. +*/ +static DEFINE_SPINLOCK (cpu_bps_lock); static DEFINE_PER_CPU(struct breakpoint *, cpu_bps[HBP_NUM_MAX]); +static DEFINE_SPINLOCK (task_bps_lock);static LIST_HEAD (task_bps);static struct breakpoint *alloc_breakpoint (struct perf_event *bp)
@@ -174,7 +182,9 @@ static int task_bps_add(struct perf_event *bp)if (IS_ERR (tmp))return PTR_ERR (tmp); +spin_lock (&task_bps_lock);list_add (&tmp->list, &task_bps); +spin_unlock (&task_bps_lock);return 0 ; }
@@ -182,6 +192,7 @@ static void task_bps_remove(struct perf_event *bp) {struct list_head *pos, *q; +spin_lock (&task_bps_lock);list_for_each_safe (pos, q, &task_bps) { struct breakpoint *tmp = list_entry(pos, struct breakpoint, list);
@@ -191,6 +202,7 @@ static void task_bps_remove(struct perf_event *bp)break ; } } +spin_unlock (&task_bps_lock); }/*
@@ -200,12 +212,17 @@ static void task_bps_remove(struct perf_event *bp)static bool all_task_bps_check (struct perf_event *bp) {struct breakpoint *tmp; +bool ret = false; +spin_lock (&task_bps_lock);list_for_each_entry (tmp, &task_bps, list) { -if (!can_co_exist (tmp, bp)) -return true; +if (!can_co_exist (tmp, bp)) { + ret = true; +break ; + } } -return false; +spin_unlock (&task_bps_lock); +return ret; }/*
@@ -215,13 +232,18 @@ static bool all_task_bps_check(struct perf_event *bp)static bool same_task_bps_check (struct perf_event *bp) {struct breakpoint *tmp; +bool ret = false; +spin_lock (&task_bps_lock);list_for_each_entry (tmp, &task_bps, list) {if (tmp->bp->hw.target == bp->hw.target && - !can_co_exist (tmp, bp)) -return true; + !can_co_exist (tmp, bp)) { + ret = true; +break ; + } } -return false; +spin_unlock (&task_bps_lock); +return ret; }static int cpu_bps_add (struct perf_event *bp)
@@ -234,6 +256,7 @@ static int cpu_bps_add(struct perf_event *bp)if (IS_ERR (tmp))return PTR_ERR (tmp); +spin_lock (&cpu_bps_lock); cpu_bp =per_cpu_ptr (cpu_bps, bp->cpu);for (i =0 ; i <nr_wp_slots (); i++) {if (!cpu_bp[i]) {
@@ -241,6 +264,7 @@ static int cpu_bps_add(struct perf_event *bp)break ; } } +spin_unlock (&cpu_bps_lock);return 0 ; }
@@ -249,6 +273,7 @@ static void cpu_bps_remove(struct perf_event *bp)struct breakpoint **cpu_bp;int i =0 ; +spin_lock (&cpu_bps_lock); cpu_bp =per_cpu_ptr (cpu_bps, bp->cpu);for (i =0 ; i <nr_wp_slots (); i++) {if (!cpu_bp[i])
@@ -260,19 +285,25 @@ static void cpu_bps_remove(struct perf_event *bp)break ; } } +spin_unlock (&cpu_bps_lock); }static bool cpu_bps_check (int cpu,struct perf_event *bp) {struct breakpoint **cpu_bp; +bool ret = false;int i; +spin_lock (&cpu_bps_lock); cpu_bp =per_cpu_ptr (cpu_bps, cpu);for (i =0 ; i <nr_wp_slots (); i++) { -if (cpu_bp[i] && !can_co_exist (cpu_bp[i], bp)) -return true; +if (cpu_bp[i] && !can_co_exist (cpu_bp[i], bp)) { + ret = true; +break ; + } } -return false; +spin_unlock (&cpu_bps_lock); +return ret; }static bool all_cpu_bps_check (struct perf_event *bp)
@@ -286,10 +317,6 @@ static bool all_cpu_bps_check(struct perf_event *bp)return false; } -/* -* We don't use any locks to serialize accesses to cpu_bps or task_bps -* because are already inside nr_bp_mutex. -*/ int arch_reserve_bp_slot (struct perf_event *bp) {int ret;
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] powerpc/hw_breakpoint: Avoid relying on caller synchronization
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: f95e5a3d59011eec1257d0e76de1e1f8969d426f Gitweb: https://git.kernel.org/tip/f95e5a3d59011eec1257d0e76de1e1f8969d426f Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:14 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:23 +02:00
powerpc/hw_breakpoint: Avoid relying on caller synchronization
Internal data structures (cpu_bps, task_bps) of powerpc's hw_breakpoint implementation have relied on nr_bp_mutex serializing access to them.
Before overhauling synchronization of kernel/events/hw_breakpoint.c, introduce 2 spinlocks to synchronize cpu_bps and task_bps respectively, thus avoiding reliance on callers synchronizing powerpc's hw_breakpoint.
--- arch/powerpc/kernel/hw_breakpoint.c | 53 +++++++++++++++++++++------- 1 file changed, 40 insertions(+), 13 deletions(-)
arch/powerpc/kernel/hw_breakpoint.c +40 -13
@@ -15,6 +15,7 @@#include <linux/kernel.h> #include <linux/sched.h> #include <linux/smp.h> +#include <linux/spinlock.h> #include <linux/debugfs.h> #include <linux/init.h>
@@ -129,7 +130,14 @@ struct breakpoint {bool ptrace_bp; }; +/* +* While kernel/events/hw_breakpoint.c does its own synchronization, we cannot +* rely on it safely synchronizing internals here; however, we can rely on it +* not requesting more breakpoints than available. +*/ +static DEFINE_SPINLOCK (cpu_bps_lock); static DEFINE_PER_CPU(struct breakpoint *, cpu_bps[HBP_NUM_MAX]); +static DEFINE_SPINLOCK (task_bps_lock);static LIST_HEAD (task_bps);static struct breakpoint *alloc_breakpoint (struct perf_event *bp)
@@ -174,7 +182,9 @@ static int task_bps_add(struct perf_event *bp)if (IS_ERR (tmp))return PTR_ERR (tmp); +spin_lock (&task_bps_lock);list_add (&tmp->list, &task_bps); +spin_unlock (&task_bps_lock);return 0 ; }
@@ -182,6 +192,7 @@ static void task_bps_remove(struct perf_event *bp) {struct list_head *pos, *q; +spin_lock (&task_bps_lock);list_for_each_safe (pos, q, &task_bps) { struct breakpoint *tmp = list_entry(pos, struct breakpoint, list);
@@ -191,6 +202,7 @@ static void task_bps_remove(struct perf_event *bp)break ; } } +spin_unlock (&task_bps_lock); }/*
@@ -200,12 +212,17 @@ static void task_bps_remove(struct perf_event *bp)static bool all_task_bps_check (struct perf_event *bp) {struct breakpoint *tmp; +bool ret = false; +spin_lock (&task_bps_lock);list_for_each_entry (tmp, &task_bps, list) { -if (!can_co_exist (tmp, bp)) -return true; +if (!can_co_exist (tmp, bp)) { + ret = true; +break ; + } } -return false; +spin_unlock (&task_bps_lock); +return ret; }/*
@@ -215,13 +232,18 @@ static bool all_task_bps_check(struct perf_event *bp)static bool same_task_bps_check (struct perf_event *bp) {struct breakpoint *tmp; +bool ret = false; +spin_lock (&task_bps_lock);list_for_each_entry (tmp, &task_bps, list) {if (tmp->bp->hw.target == bp->hw.target && - !can_co_exist (tmp, bp)) -return true; + !can_co_exist (tmp, bp)) { + ret = true; +break ; + } } -return false; +spin_unlock (&task_bps_lock); +return ret; }static int cpu_bps_add (struct perf_event *bp)
@@ -234,6 +256,7 @@ static int cpu_bps_add(struct perf_event *bp)if (IS_ERR (tmp))return PTR_ERR (tmp); +spin_lock (&cpu_bps_lock); cpu_bp =per_cpu_ptr (cpu_bps, bp->cpu);for (i =0 ; i <nr_wp_slots (); i++) {if (!cpu_bp[i]) {
@@ -241,6 +264,7 @@ static int cpu_bps_add(struct perf_event *bp)break ; } } +spin_unlock (&cpu_bps_lock);return 0 ; }
@@ -249,6 +273,7 @@ static void cpu_bps_remove(struct perf_event *bp)struct breakpoint **cpu_bp;int i =0 ; +spin_lock (&cpu_bps_lock); cpu_bp =per_cpu_ptr (cpu_bps, bp->cpu);for (i =0 ; i <nr_wp_slots (); i++) {if (!cpu_bp[i])
@@ -260,19 +285,25 @@ static void cpu_bps_remove(struct perf_event *bp)break ; } } +spin_unlock (&cpu_bps_lock); }static bool cpu_bps_check (int cpu,struct perf_event *bp) {struct breakpoint **cpu_bp; +bool ret = false;int i; +spin_lock (&cpu_bps_lock); cpu_bp =per_cpu_ptr (cpu_bps, cpu);for (i =0 ; i <nr_wp_slots (); i++) { -if (cpu_bp[i] && !can_co_exist (cpu_bp[i], bp)) -return true; +if (cpu_bp[i] && !can_co_exist (cpu_bp[i], bp)) { + ret = true; +break ; + } } -return false; +spin_unlock (&cpu_bps_lock); +return ret; }static bool all_cpu_bps_check (struct perf_event *bp)
@@ -286,10 +317,6 @@ static bool all_cpu_bps_check(struct perf_event *bp)return false; } -/* -* We don't use any locks to serialize accesses to cpu_bps or task_bps -* because are already inside nr_bp_mutex. -*/ int arch_reserve_bp_slot (struct perf_event *bp) {int ret;
PATCH v4 10/14 locking/percpu-rwsem: Add percpu_is_write_locked() and percpu_is_read_locked()
In reply to Marco ElverImplement simple accessors to probe percpu-rwsem's locked state:percpu_is_write_locked(), percpu_is_read_locked().
---
v4:
* Due to spurious read_count increments in __percpu_down_read_trylock() if sem->block != 0, check that !sem->block (reported by Peter).
v2:
* New patch.
--- include/linux/percpu-rwsem.h | 6 ++++++ kernel/locking/percpu-rwsem.c | 6 ++++++ 2 files changed, 12 insertions(+)
include/linux/percpu-rwsem.h +6 -0
@@ -121,9 +121,15 @@ static inline void percpu_up_read(struct percpu_rw_semaphore *sem)preempt_enable (); } +extern bool percpu_is_read_locked (struct percpu_rw_semaphore *);extern void percpu_down_write (struct percpu_rw_semaphore *);extern void percpu_up_write (struct percpu_rw_semaphore *); +static inline bool percpu_is_write_locked (struct percpu_rw_semaphore *sem) +{ +return atomic_read (&sem->block); +} +extern int __percpu_init_rwsem (struct percpu_rw_semaphore *,const char *,struct lock_class_key *);
kernel/locking/percpu-rwsem.c +6 -0
@@ -192,6 +192,12 @@ EXPORT_SYMBOL_GPL(__percpu_down_read); __sum; \ }) +bool percpu_is_read_locked (struct percpu_rw_semaphore *sem) +{ +return per_cpu_sum (*sem->read_count) !=0 && !atomic_read (&sem->block); +} +EXPORT_SYMBOL_GPL (percpu_is_read_locked); +/* * Return true if the modular sum of the sem->read_count per-CPU variable is * zero. If this sum is zero, then it is stable due to the fact that if any
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] locking/percpu-rwsem: Add percpu_is_write_locked() and percpu_is_read_locked()
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 01fe8a3f818e1074a9a95d624be4549ee7ea2b2b Gitweb: https://git.kernel.org/tip/01fe8a3f818e1074a9a95d624be4549ee7ea2b2b Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:15 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:23 +02:00
locking/percpu-rwsem: Add percpu_is_write_locked() and percpu_is_read_locked()
Implement simple accessors to probe percpu-rwsem's locked state:percpu_is_write_locked(), percpu_is_read_locked().
--- include/linux/percpu-rwsem.h | 6 ++++++ kernel/locking/percpu-rwsem.c | 6 ++++++ 2 files changed, 12 insertions(+)
include/linux/percpu-rwsem.h +6 -0
@@ -121,9 +121,15 @@ static inline void percpu_up_read(struct percpu_rw_semaphore *sem)preempt_enable (); } +extern bool percpu_is_read_locked (struct percpu_rw_semaphore *);extern void percpu_down_write (struct percpu_rw_semaphore *);extern void percpu_up_write (struct percpu_rw_semaphore *); +static inline bool percpu_is_write_locked (struct percpu_rw_semaphore *sem) +{ +return atomic_read (&sem->block); +} +extern int __percpu_init_rwsem (struct percpu_rw_semaphore *,const char *,struct lock_class_key *);
kernel/locking/percpu-rwsem.c +6 -0
@@ -192,6 +192,12 @@ EXPORT_SYMBOL_GPL(__percpu_down_read); __sum; \ }) +bool percpu_is_read_locked (struct percpu_rw_semaphore *sem) +{ +return per_cpu_sum (*sem->read_count) !=0 && !atomic_read (&sem->block); +} +EXPORT_SYMBOL_GPL (percpu_is_read_locked); +/* * Return true if the modular sum of the sem->read_count per-CPU variable is * zero. If this sum is zero, then it is stable due to the fact that if any
PATCH v4 11/14 perf/hw_breakpoint: Reduce contention with large number of tasks
In reply to Marco ElverWhile optimizing task_bp_pinned()'s runtime complexity to O(1) on average helps reduce time spent in the critical section, we still suffer due to serializing everything via 'nr_bp_mutex'. Indeed, a profile shows that now contention is the biggest issue:
95.93% [kernel] [k] osq_lock
0.70% [kernel] [k] mutex_spin_on_owner
0.22% [kernel] [k] smp_cfm_core_cond
0.18% [kernel] [k] task_bp_pinned
0.18% [kernel] [k] rhashtable_jhash2
0.15% [kernel] [k] queued_spin_lock_slowpathwhen running the breakpoint benchmark with (system with 256 CPUs):
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.207 [sec] | | 108.267188 usecs/op | 6929.100000 usecs/op/cpu
The main concern for synchronizing the breakpoint constraints data is that a consistent snapshot of the per-CPU and per-task data is observed.
The access pattern is as follows:
1. If the target is a task: the task's pinned breakpoints are counted, checked for space, and then appended to; only bp_cpuinfo::cpu_pinned is used to check for conflicts with CPU-only breakpoints; bp_cpuinfo::tsk_pinned are incremented/decremented, but otherwise unused.
2. If the target is a CPU: bp_cpuinfo::cpu_pinned are counted, along with bp_cpuinfo::tsk_pinned; after a successful check, cpu_pinned is incremented. No per-task breakpoints are checked.
Since rhltable safely synchronizes insertions/deletions, we can allow concurrency as follows:
1. If the target is a task: independent tasks may update and check the constraints concurrently, but same-task target calls need to be serialized; since bp_cpuinfo::tsk_pinned is only updated, but not checked, these modifications can happen concurrently by switching tsk_pinned to atomic_t.
2. If the target is a CPU: access to the per-CPU constraints needs to be serialized with other CPU-target and task-target callers (to stabilize the bp_cpuinfo::tsk_pinned snapshot).
We can allow the above concurrency by introducing a per-CPU constraints data reader-writer lock (bp_cpuinfo_sem), and per-task mutexes (reuses task_struct::perf_event_mutex):
1. If the target is a task: acquires perf_event_mutex, and acquires bp_cpuinfo_sem as a reader. The choice of percpu-rwsem minimizes contention in the presence of many read-lock but few write-lock
acquisitions: we assume many orders of magnitude more task target breakpoints creations/destructions than CPU target breakpoints.
2. If the target is a CPU: acquires bp_cpuinfo_sem as a writer.
With these changes, contention with thousands of tasks is reduced to the point where waiting on locking no longer dominates the profile:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.077 [sec] | | 40.201563 usecs/op | 2572.900000 usecs/op/cpu
21.54% [kernel] [k] task_bp_pinned
20.18% [kernel] [k] rhashtable_jhash2
6.81% [kernel] [k] toggle_bp_slot
5.47% [kernel] [k] queued_spin_lock_slowpath
3.75% [kernel] [k] smp_cfm_core_cond
3.48% [kernel] [k] bcmpOn this particular setup that's a speedup of 2.7x.
We're also getting closer to the theoretical ideal performance through optimizations in hw_breakpoint.c -- constraints accounting disabled:
| perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.067 [sec] | | 35.286458 usecs/op | 2258.333333 usecs/op/cpu
Which means the current implementation is ~12% slower than the theoretical ideal.
For reference, performance without any breakpoints:
| $> bench -r 30 breakpoint thread -b 0 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 0 breakpoints and 64 parallelism | Total time: 0.060 [sec] | | 31.365625 usecs/op | 2007.400000 usecs/op/cpu
On a system with 256 CPUs, the theoretical ideal is only ~12% slower than no breakpoints at all; the current implementation is ~28% slower.
---
v2:
* Use percpu-rwsem instead of rwlock.
* Use task_struct::perf_event_mutex. See code comment for reasoning. ==> Speedup of 2.7x (vs 2.5x in v1).
--- kernel/events/hw_breakpoint.c | 161 ++++++++++++++++++++++++++++------ 1 file changed, 133 insertions(+), 28 deletions(-)
kernel/events/hw_breakpoint.c +133 -28
@@ -19,6 +19,7 @@#include <linux/hw_breakpoint.h> +#include <linux/atomic.h> #include <linux/bug.h> #include <linux/cpu.h> #include <linux/export.h>
@@ -28,6 +29,7 @@#include <linux/kernel.h> #include <linux/mutex.h> #include <linux/notifier.h> +#include <linux/percpu-rwsem.h> #include <linux/percpu.h> #include <linux/rhashtable.h> #include <linux/sched.h>
@@ -41,9 +43,9 @@ struct bp_cpuinfo {unsigned int cpu_pinned;/* tsk_pinned[n] is the number of tasks having n+1 breakpoints */ #ifdef hw_breakpoint_slots -unsigned int tsk_pinned[hw_breakpoint_slots (0 )]; +atomic_t tsk_pinned[hw_breakpoint_slots (0 )];#else -unsigned int *tsk_pinned; +atomic_t *tsk_pinned;#endif };
@@ -65,8 +67,79 @@ static const struct rhashtable_params task_bps_ht_params = {static bool constraints_initialized__ro_after_init ; -/* Serialize accesses to the above constraints */ -static DEFINE_MUTEX (nr_bp_mutex); +/* +* Synchronizes accesses to the per-CPU constraints; the locking rules are: +* +* 1. Atomic updates to bp_cpuinfo::tsk_pinned only require a held read-lock +* (due to bp_slots_histogram::count being atomic, no update are lost). +* +* 2. Holding a write-lock is required for computations that require a +* stable snapshot of all bp_cpuinfo::tsk_pinned. +* +* 3. In all other cases, non-atomic accesses require the appropriately held +* lock (read-lock for read-only accesses; write-lock for reads/writes). +*/ +DEFINE_STATIC_PERCPU_RWSEM (bp_cpuinfo_sem); + +/* +* Return mutex to serialize accesses to per-task lists in task_bps_ht. Since +* rhltable synchronizes concurrent insertions/deletions, independent tasks may +* insert/delete concurrently; therefore, a mutex per task is sufficient. +* +* Uses task_struct::perf_event_mutex, to avoid extending task_struct with a +* hw_breakpoint-only mutex, which may be infrequently used. The caveat here is +* that hw_breakpoint may contend with per-task perf event list management. The +* assumption is that perf usecases involving hw_breakpoints are very unlikely +* to result in unnecessary contention. +*/ +static inline struct mutex *get_task_bps_mutex (struct perf_event *bp) +{ +struct task_struct *tsk = bp->hw.target; + +return tsk ? &tsk->perf_event_mutex :NULL ; +} + +static struct mutex *bp_constraints_lock (struct perf_event *bp) +{ +struct mutex *tsk_mtx =get_task_bps_mutex (bp); + +if (tsk_mtx) { +mutex_lock (tsk_mtx); +percpu_down_read (&bp_cpuinfo_sem); + }else { +percpu_down_write (&bp_cpuinfo_sem); + } + +return tsk_mtx; +} + +static void bp_constraints_unlock (struct mutex *tsk_mtx) +{ +if (tsk_mtx) { +percpu_up_read (&bp_cpuinfo_sem); +mutex_unlock (tsk_mtx); + }else { +percpu_up_write (&bp_cpuinfo_sem); + } +} + +static bool bp_constraints_is_locked (struct perf_event *bp) +{ +struct mutex *tsk_mtx =get_task_bps_mutex (bp); + +return percpu_is_write_locked (&bp_cpuinfo_sem) || + (tsk_mtx ?mutex_is_locked (tsk_mtx) : +percpu_is_read_locked (&bp_cpuinfo_sem)); +} + +static inline void assert_bp_constraints_lock_held (struct perf_event *bp) +{ +struct mutex *tsk_mtx =get_task_bps_mutex (bp); + +if (tsk_mtx) +lockdep_assert_held (tsk_mtx); +lockdep_assert_held (&bp_cpuinfo_sem); +}#ifdef hw_breakpoint_slots/*
@@ -97,7 +170,7 @@ static __init int init_breakpoint_slots(void)for (i =0 ; i <TYPE_MAX ; i++) {struct bp_cpuinfo *info =get_bp_info (cpu, i); - info->tsk_pinned =kcalloc (__nr_bp_slots[i],sizeof (int ),GFP_KERNEL ); + info->tsk_pinned =kcalloc (__nr_bp_slots[i],sizeof (atomic_t),GFP_KERNEL );if (!info->tsk_pinned) goto err; }
@@ -137,11 +210,19 @@ static inline enum bp_type_idx find_slot_idx(u64 bp_type)*/ static unsigned int max_task_bp_pinned (int cpu,enum bp_type_idx type) { -unsigned int *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned; +atomic_t *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned;int i; +/* +* At this point we want to have acquired the bp_cpuinfo_sem as a +* writer to ensure that there are no concurrent writers in +* toggle_bp_task_slot() to tsk_pinned, and we get a stable snapshot. +*/ +lockdep_assert_held_write (&bp_cpuinfo_sem); +for (i =hw_breakpoint_slots_cached (type) -1 ; i >=0 ; i--) { -if (tsk_pinned[i] >0 ) +ASSERT_EXCLUSIVE_WRITER (tsk_pinned[i]);/* Catch unexpected writers. */ +if (atomic_read (&tsk_pinned[i]) >0 )return i +1 ; }
@@ -158,6 +239,11 @@ static int task_bp_pinned(int cpu, struct perf_event *bp, enum bp_type_idx type)struct perf_event *iter;int count =0 ; +/* +* We need a stable snapshot of the per-task breakpoint list. +*/ +assert_bp_constraints_lock_held (bp); +rcu_read_lock (); head =rhltable_lookup (&task_bps_ht, &bp->hw.target, task_bps_ht_params);if (!head)
@@ -214,16 +300,25 @@ max_bp_pinned_slots(struct perf_event *bp, enum bp_type_idx type)static void toggle_bp_task_slot (struct perf_event *bp,int cpu,enum bp_type_idx type,int weight) { -unsigned int *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned; +atomic_t *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned;int old_idx, new_idx; +/* +* If bp->hw.target, tsk_pinned is only modified, but not used +* otherwise. We can permit concurrent updates as long as there are no +* other uses: having acquired bp_cpuinfo_sem as a reader allows +* concurrent updates here. Uses of tsk_pinned will require acquiring +* bp_cpuinfo_sem as a writer to stabilize tsk_pinned's value. +*/ +lockdep_assert_held_read (&bp_cpuinfo_sem); + old_idx =task_bp_pinned (cpu, bp, type) -1 ; new_idx = old_idx + weight;if (old_idx >=0 ) - tsk_pinned[old_idx]--; +atomic_dec (&tsk_pinned[old_idx]);if (new_idx >=0 ) - tsk_pinned[new_idx]++; +atomic_inc (&tsk_pinned[new_idx]); }/*
@@ -241,6 +336,7 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,/* Pinned counter cpu profiling */ if (!bp->hw.target) { +lockdep_assert_held_write (&bp_cpuinfo_sem);get_bp_info (bp->cpu, type)->cpu_pinned += weight;return 0 ; }
@@ -249,6 +345,11 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,for_each_cpu (cpu, cpumask)toggle_bp_task_slot (bp, cpu, type, weight); +/* +* Readers want a stable snapshot of the per-task breakpoint list. +*/ +assert_bp_constraints_lock_held (bp); +if (enable)return rhltable_insert (&task_bps_ht, &bp->hw.bp_list, task_bps_ht_params);else
@@ -354,14 +455,10 @@ static int __reserve_bp_slot(struct perf_event *bp, u64 bp_type)int reserve_bp_slot (struct perf_event *bp) { -int ret; - -mutex_lock (&nr_bp_mutex); - - ret =__reserve_bp_slot (bp, bp->attr.bp_type); - -mutex_unlock (&nr_bp_mutex); +struct mutex *mtx =bp_constraints_lock (bp); +int ret =__reserve_bp_slot (bp, bp->attr.bp_type); +bp_constraints_unlock (mtx);return ret; }
@@ -379,12 +476,11 @@ static void __release_bp_slot(struct perf_event *bp, u64 bp_type)void release_bp_slot (struct perf_event *bp) { -mutex_lock (&nr_bp_mutex); +struct mutex *mtx =bp_constraints_lock (bp);arch_unregister_hw_breakpoint (bp);__release_bp_slot (bp, bp->attr.bp_type); - -mutex_unlock (&nr_bp_mutex); +bp_constraints_unlock (mtx); }static int __modify_bp_slot (struct perf_event *bp,u64 old_type,u64 new_type)
@@ -411,11 +507,10 @@ static int __modify_bp_slot(struct perf_event *bp, u64 old_type, u64 new_type)static int modify_bp_slot (struct perf_event *bp,u64 old_type,u64 new_type) { -int ret; +struct mutex *mtx =bp_constraints_lock (bp); +int ret =__modify_bp_slot (bp, old_type, new_type); -mutex_lock (&nr_bp_mutex); - ret =__modify_bp_slot (bp, old_type, new_type); -mutex_unlock (&nr_bp_mutex); +bp_constraints_unlock (mtx);return ret; }
@@ -426,18 +521,28 @@ static int modify_bp_slot(struct perf_event *bp, u64 old_type, u64 new_type)*/ int dbg_reserve_bp_slot (struct perf_event *bp) { -if (mutex_is_locked (&nr_bp_mutex)) +int ret; + +if (bp_constraints_is_locked (bp))return -1 ; -return __reserve_bp_slot (bp, bp->attr.bp_type); +/* Locks aren't held; disable lockdep assert checking. */ +lockdep_off (); + ret =__reserve_bp_slot (bp, bp->attr.bp_type); +lockdep_on (); + +return ret; }int dbg_release_bp_slot (struct perf_event *bp) { -if (mutex_is_locked (&nr_bp_mutex)) +if (bp_constraints_is_locked (bp))return -1 ; +/* Locks aren't held; disable lockdep assert checking. */ +lockdep_off ();__release_bp_slot (bp, bp->attr.bp_type); +lockdep_on ();return 0 ; }
@@ -663,7 +768,7 @@ bool hw_breakpoint_is_used(void)return true;for (int slot =0 ; slot <hw_breakpoint_slots_cached (type); ++slot) { -if (info->tsk_pinned[slot]) +if (atomic_read (&info->tsk_pinned[slot]))return true; } }
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Reduce contention with large number of tasks
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 0912037fec1136d4e4796a3481f4a4ee09a2c325 Gitweb: https://git.kernel.org/tip/0912037fec1136d4e4796a3481f4a4ee09a2c325 Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:16 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:24 +02:00
perf/hw_breakpoint: Reduce contention with large number of tasks
While optimizing task_bp_pinned()'s runtime complexity to O(1) on average helps reduce time spent in the critical section, we still suffer due to serializing everything via 'nr_bp_mutex'. Indeed, a profile shows that now contention is the biggest issue:
95.93% [kernel] [k] osq_lock
0.70% [kernel] [k] mutex_spin_on_owner
0.22% [kernel] [k] smp_cfm_core_cond
0.18% [kernel] [k] task_bp_pinned
0.18% [kernel] [k] rhashtable_jhash2
0.15% [kernel] [k] queued_spin_lock_slowpathwhen running the breakpoint benchmark with (system with 256 CPUs):
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.207 [sec] | | 108.267188 usecs/op | 6929.100000 usecs/op/cpu
The main concern for synchronizing the breakpoint constraints data is that a consistent snapshot of the per-CPU and per-task data is observed.
The access pattern is as follows:
1. If the target is a task: the task's pinned breakpoints are counted, checked for space, and then appended to; only bp_cpuinfo::cpu_pinned is used to check for conflicts with CPU-only breakpoints; bp_cpuinfo::tsk_pinned are incremented/decremented, but otherwise unused.
2. If the target is a CPU: bp_cpuinfo::cpu_pinned are counted, along with bp_cpuinfo::tsk_pinned; after a successful check, cpu_pinned is incremented. No per-task breakpoints are checked.
Since rhltable safely synchronizes insertions/deletions, we can allow concurrency as follows:
1. If the target is a task: independent tasks may update and check the constraints concurrently, but same-task target calls need to be serialized; since bp_cpuinfo::tsk_pinned is only updated, but not checked, these modifications can happen concurrently by switching tsk_pinned to atomic_t.
2. If the target is a CPU: access to the per-CPU constraints needs to be serialized with other CPU-target and task-target callers (to stabilize the bp_cpuinfo::tsk_pinned snapshot).
We can allow the above concurrency by introducing a per-CPU constraints data reader-writer lock (bp_cpuinfo_sem), and per-task mutexes (reuses task_struct::perf_event_mutex):
1. If the target is a task: acquires perf_event_mutex, and acquires bp_cpuinfo_sem as a reader. The choice of percpu-rwsem minimizes contention in the presence of many read-lock but few write-lock
acquisitions: we assume many orders of magnitude more task target breakpoints creations/destructions than CPU target breakpoints.
2. If the target is a CPU: acquires bp_cpuinfo_sem as a writer.
With these changes, contention with thousands of tasks is reduced to the point where waiting on locking no longer dominates the profile:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.077 [sec] | | 40.201563 usecs/op | 2572.900000 usecs/op/cpu
21.54% [kernel] [k] task_bp_pinned
20.18% [kernel] [k] rhashtable_jhash2
6.81% [kernel] [k] toggle_bp_slot
5.47% [kernel] [k] queued_spin_lock_slowpath
3.75% [kernel] [k] smp_cfm_core_cond
3.48% [kernel] [k] bcmpOn this particular setup that's a speedup of 2.7x.
We're also getting closer to the theoretical ideal performance through optimizations in hw_breakpoint.c -- constraints accounting disabled:
| perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.067 [sec] | | 35.286458 usecs/op | 2258.333333 usecs/op/cpu
Which means the current implementation is ~12% slower than the theoretical ideal.
For reference, performance without any breakpoints:
| $> bench -r 30 breakpoint thread -b 0 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 0 breakpoints and 64 parallelism | Total time: 0.060 [sec] | | 31.365625 usecs/op | 2007.400000 usecs/op/cpu
On a system with 256 CPUs, the theoretical ideal is only ~12% slower than no breakpoints at all; the current implementation is ~28% slower.
--- kernel/events/hw_breakpoint.c | 161 +++++++++++++++++++++++++++------ 1 file changed, 133 insertions(+), 28 deletions(-)
kernel/events/hw_breakpoint.c +133 -28
@@ -19,6 +19,7 @@#include <linux/hw_breakpoint.h> +#include <linux/atomic.h> #include <linux/bug.h> #include <linux/cpu.h> #include <linux/export.h>
@@ -28,6 +29,7 @@#include <linux/kernel.h> #include <linux/mutex.h> #include <linux/notifier.h> +#include <linux/percpu-rwsem.h> #include <linux/percpu.h> #include <linux/rhashtable.h> #include <linux/sched.h>
@@ -41,9 +43,9 @@ struct bp_cpuinfo {unsigned int cpu_pinned;/* tsk_pinned[n] is the number of tasks having n+1 breakpoints */ #ifdef hw_breakpoint_slots -unsigned int tsk_pinned[hw_breakpoint_slots (0 )]; +atomic_t tsk_pinned[hw_breakpoint_slots (0 )];#else -unsigned int *tsk_pinned; +atomic_t *tsk_pinned;#endif };
@@ -65,8 +67,79 @@ static const struct rhashtable_params task_bps_ht_params = {static bool constraints_initialized__ro_after_init ; -/* Serialize accesses to the above constraints */ -static DEFINE_MUTEX (nr_bp_mutex); +/* +* Synchronizes accesses to the per-CPU constraints; the locking rules are: +* +* 1. Atomic updates to bp_cpuinfo::tsk_pinned only require a held read-lock +* (due to bp_slots_histogram::count being atomic, no update are lost). +* +* 2. Holding a write-lock is required for computations that require a +* stable snapshot of all bp_cpuinfo::tsk_pinned. +* +* 3. In all other cases, non-atomic accesses require the appropriately held +* lock (read-lock for read-only accesses; write-lock for reads/writes). +*/ +DEFINE_STATIC_PERCPU_RWSEM (bp_cpuinfo_sem); + +/* +* Return mutex to serialize accesses to per-task lists in task_bps_ht. Since +* rhltable synchronizes concurrent insertions/deletions, independent tasks may +* insert/delete concurrently; therefore, a mutex per task is sufficient. +* +* Uses task_struct::perf_event_mutex, to avoid extending task_struct with a +* hw_breakpoint-only mutex, which may be infrequently used. The caveat here is +* that hw_breakpoint may contend with per-task perf event list management. The +* assumption is that perf usecases involving hw_breakpoints are very unlikely +* to result in unnecessary contention. +*/ +static inline struct mutex *get_task_bps_mutex (struct perf_event *bp) +{ +struct task_struct *tsk = bp->hw.target; + +return tsk ? &tsk->perf_event_mutex :NULL ; +} + +static struct mutex *bp_constraints_lock (struct perf_event *bp) +{ +struct mutex *tsk_mtx =get_task_bps_mutex (bp); + +if (tsk_mtx) { +mutex_lock (tsk_mtx); +percpu_down_read (&bp_cpuinfo_sem); + }else { +percpu_down_write (&bp_cpuinfo_sem); + } + +return tsk_mtx; +} + +static void bp_constraints_unlock (struct mutex *tsk_mtx) +{ +if (tsk_mtx) { +percpu_up_read (&bp_cpuinfo_sem); +mutex_unlock (tsk_mtx); + }else { +percpu_up_write (&bp_cpuinfo_sem); + } +} + +static bool bp_constraints_is_locked (struct perf_event *bp) +{ +struct mutex *tsk_mtx =get_task_bps_mutex (bp); + +return percpu_is_write_locked (&bp_cpuinfo_sem) || + (tsk_mtx ?mutex_is_locked (tsk_mtx) : +percpu_is_read_locked (&bp_cpuinfo_sem)); +} + +static inline void assert_bp_constraints_lock_held (struct perf_event *bp) +{ +struct mutex *tsk_mtx =get_task_bps_mutex (bp); + +if (tsk_mtx) +lockdep_assert_held (tsk_mtx); +lockdep_assert_held (&bp_cpuinfo_sem); +}#ifdef hw_breakpoint_slots/*
@@ -97,7 +170,7 @@ static __init int init_breakpoint_slots(void)for (i =0 ; i <TYPE_MAX ; i++) {struct bp_cpuinfo *info =get_bp_info (cpu, i); - info->tsk_pinned =kcalloc (__nr_bp_slots[i],sizeof (int ),GFP_KERNEL ); + info->tsk_pinned =kcalloc (__nr_bp_slots[i],sizeof (atomic_t),GFP_KERNEL );if (!info->tsk_pinned) goto err; }
@@ -137,11 +210,19 @@ static inline enum bp_type_idx find_slot_idx(u64 bp_type)*/ static unsigned int max_task_bp_pinned (int cpu,enum bp_type_idx type) { -unsigned int *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned; +atomic_t *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned;int i; +/* +* At this point we want to have acquired the bp_cpuinfo_sem as a +* writer to ensure that there are no concurrent writers in +* toggle_bp_task_slot() to tsk_pinned, and we get a stable snapshot. +*/ +lockdep_assert_held_write (&bp_cpuinfo_sem); +for (i =hw_breakpoint_slots_cached (type) -1 ; i >=0 ; i--) { -if (tsk_pinned[i] >0 ) +ASSERT_EXCLUSIVE_WRITER (tsk_pinned[i]);/* Catch unexpected writers. */ +if (atomic_read (&tsk_pinned[i]) >0 )return i +1 ; }
@@ -158,6 +239,11 @@ static int task_bp_pinned(int cpu, struct perf_event *bp, enum bp_type_idx type)struct perf_event *iter;int count =0 ; +/* +* We need a stable snapshot of the per-task breakpoint list. +*/ +assert_bp_constraints_lock_held (bp); +rcu_read_lock (); head =rhltable_lookup (&task_bps_ht, &bp->hw.target, task_bps_ht_params);if (!head)
@@ -214,16 +300,25 @@ max_bp_pinned_slots(struct perf_event *bp, enum bp_type_idx type)static void toggle_bp_task_slot (struct perf_event *bp,int cpu,enum bp_type_idx type,int weight) { -unsigned int *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned; +atomic_t *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned;int old_idx, new_idx; +/* +* If bp->hw.target, tsk_pinned is only modified, but not used +* otherwise. We can permit concurrent updates as long as there are no +* other uses: having acquired bp_cpuinfo_sem as a reader allows +* concurrent updates here. Uses of tsk_pinned will require acquiring +* bp_cpuinfo_sem as a writer to stabilize tsk_pinned's value. +*/ +lockdep_assert_held_read (&bp_cpuinfo_sem); + old_idx =task_bp_pinned (cpu, bp, type) -1 ; new_idx = old_idx + weight;if (old_idx >=0 ) - tsk_pinned[old_idx]--; +atomic_dec (&tsk_pinned[old_idx]);if (new_idx >=0 ) - tsk_pinned[new_idx]++; +atomic_inc (&tsk_pinned[new_idx]); }/*
@@ -241,6 +336,7 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,/* Pinned counter cpu profiling */ if (!bp->hw.target) { +lockdep_assert_held_write (&bp_cpuinfo_sem);get_bp_info (bp->cpu, type)->cpu_pinned += weight;return 0 ; }
@@ -249,6 +345,11 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,for_each_cpu (cpu, cpumask)toggle_bp_task_slot (bp, cpu, type, weight); +/* +* Readers want a stable snapshot of the per-task breakpoint list. +*/ +assert_bp_constraints_lock_held (bp); +if (enable)return rhltable_insert (&task_bps_ht, &bp->hw.bp_list, task_bps_ht_params);else
@@ -354,14 +455,10 @@ static int __reserve_bp_slot(struct perf_event *bp, u64 bp_type)int reserve_bp_slot (struct perf_event *bp) { -int ret; - -mutex_lock (&nr_bp_mutex); - - ret =__reserve_bp_slot (bp, bp->attr.bp_type); - -mutex_unlock (&nr_bp_mutex); +struct mutex *mtx =bp_constraints_lock (bp); +int ret =__reserve_bp_slot (bp, bp->attr.bp_type); +bp_constraints_unlock (mtx);return ret; }
@@ -379,12 +476,11 @@ static void __release_bp_slot(struct perf_event *bp, u64 bp_type)void release_bp_slot (struct perf_event *bp) { -mutex_lock (&nr_bp_mutex); +struct mutex *mtx =bp_constraints_lock (bp);arch_unregister_hw_breakpoint (bp);__release_bp_slot (bp, bp->attr.bp_type); - -mutex_unlock (&nr_bp_mutex); +bp_constraints_unlock (mtx); }static int __modify_bp_slot (struct perf_event *bp,u64 old_type,u64 new_type)
@@ -411,11 +507,10 @@ static int __modify_bp_slot(struct perf_event *bp, u64 old_type, u64 new_type)static int modify_bp_slot (struct perf_event *bp,u64 old_type,u64 new_type) { -int ret; +struct mutex *mtx =bp_constraints_lock (bp); +int ret =__modify_bp_slot (bp, old_type, new_type); -mutex_lock (&nr_bp_mutex); - ret =__modify_bp_slot (bp, old_type, new_type); -mutex_unlock (&nr_bp_mutex); +bp_constraints_unlock (mtx);return ret; }
@@ -426,18 +521,28 @@ static int modify_bp_slot(struct perf_event *bp, u64 old_type, u64 new_type)*/ int dbg_reserve_bp_slot (struct perf_event *bp) { -if (mutex_is_locked (&nr_bp_mutex)) +int ret; + +if (bp_constraints_is_locked (bp))return -1 ; -return __reserve_bp_slot (bp, bp->attr.bp_type); +/* Locks aren't held; disable lockdep assert checking. */ +lockdep_off (); + ret =__reserve_bp_slot (bp, bp->attr.bp_type); +lockdep_on (); + +return ret; }int dbg_release_bp_slot (struct perf_event *bp) { -if (mutex_is_locked (&nr_bp_mutex)) +if (bp_constraints_is_locked (bp))return -1 ; +/* Locks aren't held; disable lockdep assert checking. */ +lockdep_off ();__release_bp_slot (bp, bp->attr.bp_type); +lockdep_on ();return 0 ; }
@@ -663,7 +768,7 @@ bool hw_breakpoint_is_used(void)return true;for (int slot =0 ; slot <hw_breakpoint_slots_cached (type); ++slot) { -if (info->tsk_pinned[slot]) +if (atomic_read (&info->tsk_pinned[slot]))return true; } }
PATCH v4 12/14 perf/hw_breakpoint: Introduce bp_slots_histogram
In reply to Marco ElverFactor out the existing atomic_t count[N] into its own struct called 'bp_slots_histogram', to generalize and make its intent clearer in preparation of reusing elsewhere. The basic idea of bucketing "total uses of N slots" resembles a histogram, so calling it such seems most intuitive.
No functional change.
---
v3:
* Also warn in bp_slots_histogram_add() if count goes below 0.
v2:
* New patch.
--- kernel/events/hw_breakpoint.c | 96 +++++++++++++++++++++++------------ 1 file changed, 63 insertions(+), 33 deletions(-)
kernel/events/hw_breakpoint.c +63 -33
@@ -36,19 +36,27 @@#include <linux/slab.h> /* -* Constraints data +* Datastructure to track the total uses of N slots across tasks or CPUs; +* bp_slots_histogram::count[N] is the number of assigned N+1 breakpoint slots. */ -struct bp_cpuinfo { -/* Number of pinned cpu breakpoints in a cpu */ -unsigned int cpu_pinned; -/* tsk_pinned[n] is the number of tasks having n+1 breakpoints */ +struct bp_slots_histogram {#ifdef hw_breakpoint_slots -atomic_t tsk_pinned[hw_breakpoint_slots (0 )]; +atomic_t count[hw_breakpoint_slots (0 )];#else -atomic_t *tsk_pinned; +atomic_t *count;#endif }; +/* +* Per-CPU constraints data. +*/ +struct bp_cpuinfo { +/* Number of pinned CPU breakpoints in a CPU. */ +unsigned int cpu_pinned; +/* Histogram of pinned task breakpoints in a CPU. */ +struct bp_slots_histogram tsk_pinned; +}; + static DEFINE_PER_CPU(struct bp_cpuinfo, bp_cpuinfo[TYPE_MAX]);static struct bp_cpuinfo *get_bp_info (int cpu,enum bp_type_idx type)
@@ -159,6 +167,18 @@ static inline int hw_breakpoint_slots_cached(int type)return __nr_bp_slots[type]; } +static __init bool +bp_slots_histogram_alloc (struct bp_slots_histogram *hist,enum bp_type_idx type) +{ + hist->count =kcalloc (hw_breakpoint_slots_cached (type),sizeof (*hist->count),GFP_KERNEL ); +return hist->count; +} + +static __init void bp_slots_histogram_free (struct bp_slots_histogram *hist) +{ +kfree (hist->count); +} +static __init int init_breakpoint_slots (void ) {int i, cpu, err_cpu;
@@ -170,8 +190,7 @@ static __init int init_breakpoint_slots(void)for (i =0 ; i <TYPE_MAX ; i++) {struct bp_cpuinfo *info =get_bp_info (cpu, i); - info->tsk_pinned =kcalloc (__nr_bp_slots[i],sizeof (atomic_t),GFP_KERNEL ); -if (!info->tsk_pinned) +if (!bp_slots_histogram_alloc (&info->tsk_pinned, i)) goto err; } }
@@ -180,7 +199,7 @@ static __init int init_breakpoint_slots(void) err:for_each_possible_cpu (err_cpu) {for (i =0 ; i <TYPE_MAX ; i++) -kfree (get_bp_info (err_cpu, i)->tsk_pinned); +bp_slots_histogram_free (&get_bp_info (err_cpu, i)->tsk_pinned);if (err_cpu == cpu)break ; }
@@ -189,6 +208,34 @@ static __init int init_breakpoint_slots(void) }#endif +static inline void +bp_slots_histogram_add (struct bp_slots_histogram *hist,int old,int val) +{ +const int old_idx = old -1 ; +const int new_idx = old_idx + val; + +if (old_idx >=0 ) +WARN_ON (atomic_dec_return_relaxed (&hist->count[old_idx]) <0 ); +if (new_idx >=0 ) +WARN_ON (atomic_inc_return_relaxed (&hist->count[new_idx]) <0 ); +} + +static int +bp_slots_histogram_max (struct bp_slots_histogram *hist,enum bp_type_idx type) +{ +for (int i =hw_breakpoint_slots_cached (type) -1 ; i >=0 ; i--) { +const int count =atomic_read (&hist->count[i]); + +/* Catch unexpected writers; we want a stable snapshot. */ +ASSERT_EXCLUSIVE_WRITER (hist->count[i]); +if (count >0 ) +return i +1 ; +WARN (count <0 ,"inconsistent breakpoint slots histogram" ); + } + +return 0 ; +} +#ifndef hw_breakpoint_weightstatic inline int hw_breakpoint_weight (struct perf_event *bp) {
@@ -205,13 +252,11 @@ static inline enum bp_type_idx find_slot_idx(u64 bp_type) }/* -* Report the maximum number of pinned breakpoints a task -* have in this cpu +* Return the maximum number of pinned breakpoints a task has in this CPU. */ static unsigned int max_task_bp_pinned (int cpu,enum bp_type_idx type) { -atomic_t *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned; -int i; +struct bp_slots_histogram *tsk_pinned = &get_bp_info (cpu, type)->tsk_pinned;/* * At this point we want to have acquired the bp_cpuinfo_sem as a
@@ -219,14 +264,7 @@ static unsigned int max_task_bp_pinned(int cpu, enum bp_type_idx type)* toggle_bp_task_slot() to tsk_pinned, and we get a stable snapshot. */ lockdep_assert_held_write (&bp_cpuinfo_sem); - -for (i =hw_breakpoint_slots_cached (type) -1 ; i >=0 ; i--) { -ASSERT_EXCLUSIVE_WRITER (tsk_pinned[i]);/* Catch unexpected writers. */ -if (atomic_read (&tsk_pinned[i]) >0 ) -return i +1 ; - } - -return 0 ; +return bp_slots_histogram_max (tsk_pinned, type); }/*
@@ -300,8 +338,7 @@ max_bp_pinned_slots(struct perf_event *bp, enum bp_type_idx type)static void toggle_bp_task_slot (struct perf_event *bp,int cpu,enum bp_type_idx type,int weight) { -atomic_t *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned; -int old_idx, new_idx; +struct bp_slots_histogram *tsk_pinned = &get_bp_info (cpu, type)->tsk_pinned;/* * If bp->hw.target, tsk_pinned is only modified, but not used
@@ -311,14 +348,7 @@ static void toggle_bp_task_slot(struct perf_event *bp, int cpu,* bp_cpuinfo_sem as a writer to stabilize tsk_pinned's value. */ lockdep_assert_held_read (&bp_cpuinfo_sem); - - old_idx =task_bp_pinned (cpu, bp, type) -1 ; - new_idx = old_idx + weight; - -if (old_idx >=0 ) -atomic_dec (&tsk_pinned[old_idx]); -if (new_idx >=0 ) -atomic_inc (&tsk_pinned[new_idx]); +bp_slots_histogram_add (tsk_pinned,task_bp_pinned (cpu, bp, type), weight); }/*
@@ -768,7 +798,7 @@ bool hw_breakpoint_is_used(void)return true;for (int slot =0 ; slot <hw_breakpoint_slots_cached (type); ++slot) { -if (atomic_read (&info->tsk_pinned[slot])) +if (atomic_read (&info->tsk_pinned.count[slot]))return true; } }
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Introduce bp_slots_histogram
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 16db2839a5a59c242df77308cf57342ce0c3768e Gitweb: https://git.kernel.org/tip/16db2839a5a59c242df77308cf57342ce0c3768e Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:17 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:24 +02:00
perf/hw_breakpoint: Introduce bp_slots_histogram
Factor out the existing atomic_t count[N] into its own struct called 'bp_slots_histogram', to generalize and make its intent clearer in preparation of reusing elsewhere. The basic idea of bucketing "total uses of N slots" resembles a histogram, so calling it such seems most intuitive.
No functional change.
--- kernel/events/hw_breakpoint.c | 96 ++++++++++++++++++++++------------ 1 file changed, 63 insertions(+), 33 deletions(-)
kernel/events/hw_breakpoint.c +63 -33
@@ -36,19 +36,27 @@#include <linux/slab.h> /* -* Constraints data +* Datastructure to track the total uses of N slots across tasks or CPUs; +* bp_slots_histogram::count[N] is the number of assigned N+1 breakpoint slots. */ -struct bp_cpuinfo { -/* Number of pinned cpu breakpoints in a cpu */ -unsigned int cpu_pinned; -/* tsk_pinned[n] is the number of tasks having n+1 breakpoints */ +struct bp_slots_histogram {#ifdef hw_breakpoint_slots -atomic_t tsk_pinned[hw_breakpoint_slots (0 )]; +atomic_t count[hw_breakpoint_slots (0 )];#else -atomic_t *tsk_pinned; +atomic_t *count;#endif }; +/* +* Per-CPU constraints data. +*/ +struct bp_cpuinfo { +/* Number of pinned CPU breakpoints in a CPU. */ +unsigned int cpu_pinned; +/* Histogram of pinned task breakpoints in a CPU. */ +struct bp_slots_histogram tsk_pinned; +}; + static DEFINE_PER_CPU(struct bp_cpuinfo, bp_cpuinfo[TYPE_MAX]);static struct bp_cpuinfo *get_bp_info (int cpu,enum bp_type_idx type)
@@ -159,6 +167,18 @@ static inline int hw_breakpoint_slots_cached(int type)return __nr_bp_slots[type]; } +static __init bool +bp_slots_histogram_alloc (struct bp_slots_histogram *hist,enum bp_type_idx type) +{ + hist->count =kcalloc (hw_breakpoint_slots_cached (type),sizeof (*hist->count),GFP_KERNEL ); +return hist->count; +} + +static __init void bp_slots_histogram_free (struct bp_slots_histogram *hist) +{ +kfree (hist->count); +} +static __init int init_breakpoint_slots (void ) {int i, cpu, err_cpu;
@@ -170,8 +190,7 @@ static __init int init_breakpoint_slots(void)for (i =0 ; i <TYPE_MAX ; i++) {struct bp_cpuinfo *info =get_bp_info (cpu, i); - info->tsk_pinned =kcalloc (__nr_bp_slots[i],sizeof (atomic_t),GFP_KERNEL ); -if (!info->tsk_pinned) +if (!bp_slots_histogram_alloc (&info->tsk_pinned, i)) goto err; } }
@@ -180,7 +199,7 @@ static __init int init_breakpoint_slots(void) err:for_each_possible_cpu (err_cpu) {for (i =0 ; i <TYPE_MAX ; i++) -kfree (get_bp_info (err_cpu, i)->tsk_pinned); +bp_slots_histogram_free (&get_bp_info (err_cpu, i)->tsk_pinned);if (err_cpu == cpu)break ; }
@@ -189,6 +208,34 @@ err: }#endif +static inline void +bp_slots_histogram_add (struct bp_slots_histogram *hist,int old,int val) +{ +const int old_idx = old -1 ; +const int new_idx = old_idx + val; + +if (old_idx >=0 ) +WARN_ON (atomic_dec_return_relaxed (&hist->count[old_idx]) <0 ); +if (new_idx >=0 ) +WARN_ON (atomic_inc_return_relaxed (&hist->count[new_idx]) <0 ); +} + +static int +bp_slots_histogram_max (struct bp_slots_histogram *hist,enum bp_type_idx type) +{ +for (int i =hw_breakpoint_slots_cached (type) -1 ; i >=0 ; i--) { +const int count =atomic_read (&hist->count[i]); + +/* Catch unexpected writers; we want a stable snapshot. */ +ASSERT_EXCLUSIVE_WRITER (hist->count[i]); +if (count >0 ) +return i +1 ; +WARN (count <0 ,"inconsistent breakpoint slots histogram" ); + } + +return 0 ; +} +#ifndef hw_breakpoint_weightstatic inline int hw_breakpoint_weight (struct perf_event *bp) {
@@ -205,13 +252,11 @@ static inline enum bp_type_idx find_slot_idx(u64 bp_type) }/* -* Report the maximum number of pinned breakpoints a task -* have in this cpu +* Return the maximum number of pinned breakpoints a task has in this CPU. */ static unsigned int max_task_bp_pinned (int cpu,enum bp_type_idx type) { -atomic_t *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned; -int i; +struct bp_slots_histogram *tsk_pinned = &get_bp_info (cpu, type)->tsk_pinned;/* * At this point we want to have acquired the bp_cpuinfo_sem as a
@@ -219,14 +264,7 @@ static unsigned int max_task_bp_pinned(int cpu, enum bp_type_idx type)* toggle_bp_task_slot() to tsk_pinned, and we get a stable snapshot. */ lockdep_assert_held_write (&bp_cpuinfo_sem); - -for (i =hw_breakpoint_slots_cached (type) -1 ; i >=0 ; i--) { -ASSERT_EXCLUSIVE_WRITER (tsk_pinned[i]);/* Catch unexpected writers. */ -if (atomic_read (&tsk_pinned[i]) >0 ) -return i +1 ; - } - -return 0 ; +return bp_slots_histogram_max (tsk_pinned, type); }/*
@@ -300,8 +338,7 @@ max_bp_pinned_slots(struct perf_event *bp, enum bp_type_idx type)static void toggle_bp_task_slot (struct perf_event *bp,int cpu,enum bp_type_idx type,int weight) { -atomic_t *tsk_pinned =get_bp_info (cpu, type)->tsk_pinned; -int old_idx, new_idx; +struct bp_slots_histogram *tsk_pinned = &get_bp_info (cpu, type)->tsk_pinned;/* * If bp->hw.target, tsk_pinned is only modified, but not used
@@ -311,14 +348,7 @@ static void toggle_bp_task_slot(struct perf_event *bp, int cpu,* bp_cpuinfo_sem as a writer to stabilize tsk_pinned's value. */ lockdep_assert_held_read (&bp_cpuinfo_sem); - - old_idx =task_bp_pinned (cpu, bp, type) -1 ; - new_idx = old_idx + weight; - -if (old_idx >=0 ) -atomic_dec (&tsk_pinned[old_idx]); -if (new_idx >=0 ) -atomic_inc (&tsk_pinned[new_idx]); +bp_slots_histogram_add (tsk_pinned,task_bp_pinned (cpu, bp, type), weight); }/*
@@ -768,7 +798,7 @@ bool hw_breakpoint_is_used(void)return true;for (int slot =0 ; slot <hw_breakpoint_slots_cached (type); ++slot) { -if (atomic_read (&info->tsk_pinned[slot])) +if (atomic_read (&info->tsk_pinned.count[slot]))return true; } }
PATCH v4 13/14 perf/hw_breakpoint: Optimize max_bp_pinned_slots() for CPU-independent task targets
In reply to Marco ElverRunning the perf benchmark with (note: more aggressive parameters vs. preceding changes, but same 256 CPUs host):
| $> perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512 | # Running 'breakpoint/thread' benchmark: | # Created/joined 100 threads with 4 breakpoints and 128 parallelism | Total time: 1.989 [sec] | | 38.854160 usecs/op | 4973.332500 usecs/op/cpu
20.43% [kernel] [k] queued_spin_lock_slowpath
18.75% [kernel] [k] osq_lock
16.98% [kernel] [k] rhashtable_jhash2
8.34% [kernel] [k] task_bp_pinned
4.23% [kernel] [k] smp_cfm_core_cond
3.65% [kernel] [k] bcmp
2.83% [kernel] [k] toggle_bp_slot
1.87% [kernel] [k] find_next_bit
1.49% [kernel] [k] __reserve_bp_slotWe can see that a majority of the time is now spent hashing task pointers to index into task_bps_ht in task_bp_pinned().
Obtaining the max_bp_pinned_slots() for CPU-independent task targets currently is O(#cpus), and calls task_bp_pinned() for each CPU, even if the result of task_bp_pinned() is CPU-independent.
The loop in max_bp_pinned_slots() wants to compute the maximum slots across all CPUs. If task_bp_pinned() is CPU-independent, we can do so by obtaining the max slots across all CPUs and adding task_bp_pinned().
To do so in O(1), use a bp_slots_histogram for CPU-pinned slots.
After this optimization:
| $> perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512 | # Running 'breakpoint/thread' benchmark: | # Created/joined 100 threads with 4 breakpoints and 128 parallelism | Total time: 1.930 [sec] | | 37.697832 usecs/op | 4825.322500 usecs/op/cpu
19.13% [kernel] [k] queued_spin_lock_slowpath
18.21% [kernel] [k] rhashtable_jhash2
15.46% [kernel] [k] osq_lock
6.27% [kernel] [k] toggle_bp_slot
5.91% [kernel] [k] task_bp_pinned
5.05% [kernel] [k] smp_cfm_core_cond
1.78% [kernel] [k] update_sg_lb_stats
1.36% [kernel] [k] llist_reverse_order
1.34% [kernel] [k] find_next_bit
1.19% [kernel] [k] bcmpSuggesting that time spent in task_bp_pinned() has been reduced. However, we're still hashing too much, which will be addressed in the subsequent change.
---
v3:
* Update hw_breakpoint_is_used() to include global cpu_pinned.
v2:
* New patch.
--- kernel/events/hw_breakpoint.c | 57 ++++++++++++++++++++++++++++++++--- 1 file changed, 53 insertions(+), 4 deletions(-)
kernel/events/hw_breakpoint.c +53 -4
@@ -64,6 +64,9 @@ static struct bp_cpuinfo *get_bp_info(int cpu, enum bp_type_idx type)return per_cpu_ptr (bp_cpuinfo + type, cpu); } +/* Number of pinned CPU breakpoints globally. */ +static struct bp_slots_histogram cpu_pinned[TYPE_MAX ]; +/* Keep track of the breakpoints attached to tasks */ static struct rhltable task_bps_ht;static const struct rhashtable_params task_bps_ht_params = {
@@ -194,6 +197,10 @@ static __init int init_breakpoint_slots(void) goto err; } } +for (i =0 ; i <TYPE_MAX ; i++) { +if (!bp_slots_histogram_alloc (&cpu_pinned[i], i)) + goto err; + }return 0 ; err:
@@ -203,6 +210,8 @@ static __init int init_breakpoint_slots(void)if (err_cpu == cpu)break ; } +for (i =0 ; i <TYPE_MAX ; i++) +bp_slots_histogram_free (&cpu_pinned[i]);return -ENOMEM ; }
@@ -270,6 +279,9 @@ static unsigned int max_task_bp_pinned(int cpu, enum bp_type_idx type)/* * Count the number of breakpoints of the same type and same task. * The given event must be not on the list. +* +* If @cpu is -1, but the result of task_bp_pinned() is not CPU-independent, +* returns a negative value. */ static int task_bp_pinned (int cpu,struct perf_event *bp,enum bp_type_idx type) {
@@ -288,9 +300,18 @@ static int task_bp_pinned(int cpu, struct perf_event *bp, enum bp_type_idx type) goto out;rhl_for_each_entry_rcu (iter, pos, head, hw.bp_list) { -if (find_slot_idx (iter->attr.bp_type) == type && - (iter->cpu <0 || cpu == iter->cpu)) - count +=hw_breakpoint_weight (iter); +if (find_slot_idx (iter->attr.bp_type) != type) +continue ; + +if (iter->cpu >=0 ) { +if (cpu ==-1 ) { + count =-1 ; + goto out; + }else if (cpu != iter->cpu) +continue ; + } + + count +=hw_breakpoint_weight (iter); } out:
@@ -316,6 +337,19 @@ max_bp_pinned_slots(struct perf_event *bp, enum bp_type_idx type)int pinned_slots =0 ;int cpu; +if (bp->hw.target && bp->cpu <0 ) { +int max_pinned =task_bp_pinned (-1 , bp, type); + +if (max_pinned >=0 ) { +/* +* Fast path: task_bp_pinned() is CPU-independent and +* returns the same value for any CPU. +*/ + max_pinned +=bp_slots_histogram_max (&cpu_pinned[type], type); +return max_pinned; + } + } +for_each_cpu (cpu, cpumask) {struct bp_cpuinfo *info =get_bp_info (cpu, type);int nr;
@@ -366,8 +400,11 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,/* Pinned counter cpu profiling */ if (!bp->hw.target) { +struct bp_cpuinfo *info =get_bp_info (bp->cpu, type); +lockdep_assert_held_write (&bp_cpuinfo_sem); -get_bp_info (bp->cpu, type)->cpu_pinned += weight; +bp_slots_histogram_add (&cpu_pinned[type], info->cpu_pinned, weight); + info->cpu_pinned += weight;return 0 ; }
@@ -804,6 +841,18 @@ bool hw_breakpoint_is_used(void) } } +for (int type =0 ; type <TYPE_MAX ; ++type) { +for (int slot =0 ; slot <hw_breakpoint_slots_cached (type); ++slot) { +/* +* Warn, because if there are CPU pinned counters, +* should never get here; bp_cpuinfo::cpu_pinned should +* be consistent with the global cpu_pinned histogram. +*/ +if (WARN_ON (atomic_read (&cpu_pinned[type].count[slot]))) +return true; + } + } +return false; }
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Optimize max_bp_pinned_slots() for CPU-independent task targets
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 9b1933b864a10e7f66b06d10c39217142baed28b Gitweb: https://git.kernel.org/tip/9b1933b864a10e7f66b06d10c39217142baed28b Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:18 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:24 +02:00
perf/hw_breakpoint: Optimize max_bp_pinned_slots() for CPU-independent task targets
Running the perf benchmark with (note: more aggressive parameters vs. preceding changes, but same 256 CPUs host):
| $> perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512 | # Running 'breakpoint/thread' benchmark: | # Created/joined 100 threads with 4 breakpoints and 128 parallelism | Total time: 1.989 [sec] | | 38.854160 usecs/op | 4973.332500 usecs/op/cpu
20.43% [kernel] [k] queued_spin_lock_slowpath
18.75% [kernel] [k] osq_lock
16.98% [kernel] [k] rhashtable_jhash2
8.34% [kernel] [k] task_bp_pinned
4.23% [kernel] [k] smp_cfm_core_cond
3.65% [kernel] [k] bcmp
2.83% [kernel] [k] toggle_bp_slot
1.87% [kernel] [k] find_next_bit
1.49% [kernel] [k] __reserve_bp_slotWe can see that a majority of the time is now spent hashing task pointers to index into task_bps_ht in task_bp_pinned().
Obtaining the max_bp_pinned_slots() for CPU-independent task targets currently is O(#cpus), and calls task_bp_pinned() for each CPU, even if the result of task_bp_pinned() is CPU-independent.
The loop in max_bp_pinned_slots() wants to compute the maximum slots across all CPUs. If task_bp_pinned() is CPU-independent, we can do so by obtaining the max slots across all CPUs and adding task_bp_pinned().
To do so in O(1), use a bp_slots_histogram for CPU-pinned slots.
After this optimization:
| $> perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512 | # Running 'breakpoint/thread' benchmark: | # Created/joined 100 threads with 4 breakpoints and 128 parallelism | Total time: 1.930 [sec] | | 37.697832 usecs/op | 4825.322500 usecs/op/cpu
19.13% [kernel] [k] queued_spin_lock_slowpath
18.21% [kernel] [k] rhashtable_jhash2
15.46% [kernel] [k] osq_lock
6.27% [kernel] [k] toggle_bp_slot
5.91% [kernel] [k] task_bp_pinned
5.05% [kernel] [k] smp_cfm_core_cond
1.78% [kernel] [k] update_sg_lb_stats
1.36% [kernel] [k] llist_reverse_order
1.34% [kernel] [k] find_next_bit
1.19% [kernel] [k] bcmpSuggesting that time spent in task_bp_pinned() has been reduced. However, we're still hashing too much, which will be addressed in the subsequent change.
--- kernel/events/hw_breakpoint.c | 57 +++++++++++++++++++++++++++++++--- 1 file changed, 53 insertions(+), 4 deletions(-)
kernel/events/hw_breakpoint.c +53 -4
@@ -64,6 +64,9 @@ static struct bp_cpuinfo *get_bp_info(int cpu, enum bp_type_idx type)return per_cpu_ptr (bp_cpuinfo + type, cpu); } +/* Number of pinned CPU breakpoints globally. */ +static struct bp_slots_histogram cpu_pinned[TYPE_MAX ]; +/* Keep track of the breakpoints attached to tasks */ static struct rhltable task_bps_ht;static const struct rhashtable_params task_bps_ht_params = {
@@ -194,6 +197,10 @@ static __init int init_breakpoint_slots(void) goto err; } } +for (i =0 ; i <TYPE_MAX ; i++) { +if (!bp_slots_histogram_alloc (&cpu_pinned[i], i)) + goto err; + }return 0 ; err:
@@ -203,6 +210,8 @@ err:if (err_cpu == cpu)break ; } +for (i =0 ; i <TYPE_MAX ; i++) +bp_slots_histogram_free (&cpu_pinned[i]);return -ENOMEM ; }
@@ -270,6 +279,9 @@ static unsigned int max_task_bp_pinned(int cpu, enum bp_type_idx type)/* * Count the number of breakpoints of the same type and same task. * The given event must be not on the list. +* +* If @cpu is -1, but the result of task_bp_pinned() is not CPU-independent, +* returns a negative value. */ static int task_bp_pinned (int cpu,struct perf_event *bp,enum bp_type_idx type) {
@@ -288,9 +300,18 @@ static int task_bp_pinned(int cpu, struct perf_event *bp, enum bp_type_idx type) goto out;rhl_for_each_entry_rcu (iter, pos, head, hw.bp_list) { -if (find_slot_idx (iter->attr.bp_type) == type && - (iter->cpu <0 || cpu == iter->cpu)) - count +=hw_breakpoint_weight (iter); +if (find_slot_idx (iter->attr.bp_type) != type) +continue ; + +if (iter->cpu >=0 ) { +if (cpu ==-1 ) { + count =-1 ; + goto out; + }else if (cpu != iter->cpu) +continue ; + } + + count +=hw_breakpoint_weight (iter); } out:
@@ -316,6 +337,19 @@ max_bp_pinned_slots(struct perf_event *bp, enum bp_type_idx type)int pinned_slots =0 ;int cpu; +if (bp->hw.target && bp->cpu <0 ) { +int max_pinned =task_bp_pinned (-1 , bp, type); + +if (max_pinned >=0 ) { +/* +* Fast path: task_bp_pinned() is CPU-independent and +* returns the same value for any CPU. +*/ + max_pinned +=bp_slots_histogram_max (&cpu_pinned[type], type); +return max_pinned; + } + } +for_each_cpu (cpu, cpumask) {struct bp_cpuinfo *info =get_bp_info (cpu, type);int nr;
@@ -366,8 +400,11 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,/* Pinned counter cpu profiling */ if (!bp->hw.target) { +struct bp_cpuinfo *info =get_bp_info (bp->cpu, type); +lockdep_assert_held_write (&bp_cpuinfo_sem); -get_bp_info (bp->cpu, type)->cpu_pinned += weight; +bp_slots_histogram_add (&cpu_pinned[type], info->cpu_pinned, weight); + info->cpu_pinned += weight;return 0 ; }
@@ -804,6 +841,18 @@ bool hw_breakpoint_is_used(void) } } +for (int type =0 ; type <TYPE_MAX ; ++type) { +for (int slot =0 ; slot <hw_breakpoint_slots_cached (type); ++slot) { +/* +* Warn, because if there are CPU pinned counters, +* should never get here; bp_cpuinfo::cpu_pinned should +* be consistent with the global cpu_pinned histogram. +*/ +if (WARN_ON (atomic_read (&cpu_pinned[type].count[slot]))) +return true; + } + } +return false; }
PATCH v4 14/14 perf/hw_breakpoint: Optimize toggle_bp_slot() for CPU-independent task targets
In reply to Marco ElverWe can still see that a majority of the time is spent hashing task pointers:
...
16.98% [kernel] [k] rhashtable_jhash2
...Doing the bookkeeping in toggle_bp_slots() is currently O(#cpus), calling task_bp_pinned() for each CPU, even if task_bp_pinned() is CPU-independent. The reason for this is to update the per-CPU 'tsk_pinned' histogram.
To optimize the CPU-independent case to O(1), keep a separate CPU-independent 'tsk_pinned_all' histogram.
The major source of complexity are transitions between "all CPU-independent task breakpoints" and "mixed CPU-independent and CPU-dependent task breakpoints". The code comments list all cases that require handling.
After this optimization:
| $> perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512 | # Running 'breakpoint/thread' benchmark: | # Created/joined 100 threads with 4 breakpoints and 128 parallelism | Total time: 1.758 [sec] | | 34.336621 usecs/op | 4395.087500 usecs/op/cpu
38.08% [kernel] [k] queued_spin_lock_slowpath
10.81% [kernel] [k] smp_cfm_core_cond
3.01% [kernel] [k] update_sg_lb_stats
2.58% [kernel] [k] osq_lock
2.57% [kernel] [k] llist_reverse_order
1.45% [kernel] [k] find_next_bit
1.21% [kernel] [k] flush_tlb_func_common
1.01% [kernel] [k] arch_install_hw_breakpointShowing that the time spent hashing keys has become insignificant.
With the given benchmark parameters, that's an improvement of 12% compared with the old O(#cpus) version.
And finally, using the less aggressive parameters from the preceding changes, we now observe:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.067 [sec] | | 35.292187 usecs/op | 2258.700000 usecs/op/cpu
Which is an improvement of 12% compared to without the histogram optimizations (baseline is 40 usecs/op). This is now on par with the theoretical ideal (constraints disabled), and only 12% slower than no breakpoints at all.
---
v3:
* Fix typo "5 cases" -> "4 cases".
* Update hw_breakpoint_is_used() to check tsk_pinned_all.
v2:
* New patch.
--- kernel/events/hw_breakpoint.c | 155 +++++++++++++++++++++++++++------- 1 file changed, 124 insertions(+), 31 deletions(-)
kernel/events/hw_breakpoint.c +124 -31
@@ -66,6 +66,8 @@ static struct bp_cpuinfo *get_bp_info(int cpu, enum bp_type_idx type)/* Number of pinned CPU breakpoints globally. */ static struct bp_slots_histogram cpu_pinned[TYPE_MAX ]; +/* Number of pinned CPU-independent task breakpoints. */ +static struct bp_slots_histogram tsk_pinned_all[TYPE_MAX ];/* Keep track of the breakpoints attached to tasks */ static struct rhltable task_bps_ht;
@@ -200,6 +202,8 @@ static __init int init_breakpoint_slots(void)for (i =0 ; i <TYPE_MAX ; i++) {if (!bp_slots_histogram_alloc (&cpu_pinned[i], i)) goto err; +if (!bp_slots_histogram_alloc (&tsk_pinned_all[i], i)) + goto err; }return 0 ;
@@ -210,8 +214,10 @@ static __init int init_breakpoint_slots(void)if (err_cpu == cpu)break ; } -for (i =0 ; i <TYPE_MAX ; i++) +for (i =0 ; i <TYPE_MAX ; i++) {bp_slots_histogram_free (&cpu_pinned[i]); +bp_slots_histogram_free (&tsk_pinned_all[i]); + }return -ENOMEM ; }
@@ -245,6 +251,26 @@ bp_slots_histogram_max(struct bp_slots_histogram *hist, enum bp_type_idx type)return 0 ; } +static int +bp_slots_histogram_max_merge (struct bp_slots_histogram *hist1,struct bp_slots_histogram *hist2, +enum bp_type_idx type) +{ +for (int i =hw_breakpoint_slots_cached (type) -1 ; i >=0 ; i--) { +const int count1 =atomic_read (&hist1->count[i]); +const int count2 =atomic_read (&hist2->count[i]); + +/* Catch unexpected writers; we want a stable snapshot. */ +ASSERT_EXCLUSIVE_WRITER (hist1->count[i]); +ASSERT_EXCLUSIVE_WRITER (hist2->count[i]); +if (count1 + count2 >0 ) +return i +1 ; +WARN (count1 <0 ,"inconsistent breakpoint slots histogram" ); +WARN (count2 <0 ,"inconsistent breakpoint slots histogram" ); + } + +return 0 ; +} +#ifndef hw_breakpoint_weightstatic inline int hw_breakpoint_weight (struct perf_event *bp) {
@@ -273,7 +299,7 @@ static unsigned int max_task_bp_pinned(int cpu, enum bp_type_idx type)* toggle_bp_task_slot() to tsk_pinned, and we get a stable snapshot. */ lockdep_assert_held_write (&bp_cpuinfo_sem); -return bp_slots_histogram_max (tsk_pinned, type); +return bp_slots_histogram_max_merge (tsk_pinned, &tsk_pinned_all[type], type); }/*
@@ -366,40 +392,22 @@ max_bp_pinned_slots(struct perf_event *bp, enum bp_type_idx type)return pinned_slots; } -/* -* Add a pinned breakpoint for the given task in our constraint table -*/ -static void toggle_bp_task_slot (struct perf_event *bp,int cpu, -enum bp_type_idx type,int weight) -{ -struct bp_slots_histogram *tsk_pinned = &get_bp_info (cpu, type)->tsk_pinned; - -/* -* If bp->hw.target, tsk_pinned is only modified, but not used -* otherwise. We can permit concurrent updates as long as there are no -* other uses: having acquired bp_cpuinfo_sem as a reader allows -* concurrent updates here. Uses of tsk_pinned will require acquiring -* bp_cpuinfo_sem as a writer to stabilize tsk_pinned's value. -*/ -lockdep_assert_held_read (&bp_cpuinfo_sem); -bp_slots_histogram_add (tsk_pinned,task_bp_pinned (cpu, bp, type), weight); -} -/* * Add/remove the given breakpoint in our constraint table */ static int -toggle_bp_slot (struct perf_event *bp,bool enable,enum bp_type_idx type, -int weight) +toggle_bp_slot (struct perf_event *bp,bool enable,enum bp_type_idx type,int weight) { -const struct cpumask *cpumask =cpumask_of_bp (bp); -int cpu; +int cpu, next_tsk_pinned;if (!enable) weight = -weight; -/* Pinned counter cpu profiling */ if (!bp->hw.target) { +/* +* Update the pinned CPU slots, in per-CPU bp_cpuinfo and in the +* global histogram. +*/ struct bp_cpuinfo *info =get_bp_info (bp->cpu, type);lockdep_assert_held_write (&bp_cpuinfo_sem);
@@ -408,9 +416,91 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,return 0 ; } -/* Pinned counter task profiling */ -for_each_cpu (cpu, cpumask) -toggle_bp_task_slot (bp, cpu, type, weight); +/* +* If bp->hw.target, tsk_pinned is only modified, but not used +* otherwise. We can permit concurrent updates as long as there are no +* other uses: having acquired bp_cpuinfo_sem as a reader allows +* concurrent updates here. Uses of tsk_pinned will require acquiring +* bp_cpuinfo_sem as a writer to stabilize tsk_pinned's value. +*/ +lockdep_assert_held_read (&bp_cpuinfo_sem); + +/* +* Update the pinned task slots, in per-CPU bp_cpuinfo and in the global +* histogram. We need to take care of 4 cases: +* +* 1. This breakpoint targets all CPUs (cpu < 0), and there may only +* exist other task breakpoints targeting all CPUs. In this case we +* can simply update the global slots histogram. +* +* 2. This breakpoint targets a specific CPU (cpu >= 0), but there may +* only exist other task breakpoints targeting all CPUs. +* +* a. On enable: remove the existing breakpoints from the global +* slots histogram and use the per-CPU histogram. +* +* b. On disable: re-insert the existing breakpoints into the global +* slots histogram and remove from per-CPU histogram. +* +* 3. Some other existing task breakpoints target specific CPUs. Only +* update the per-CPU slots histogram. +*/ + +if (!enable) { +/* +* Remove before updating histograms so we can determine if this +* was the last task breakpoint for a specific CPU. +*/ +int ret =rhltable_remove (&task_bps_ht, &bp->hw.bp_list, task_bps_ht_params); + +if (ret) +return ret; + } +/* +* Note: If !enable, next_tsk_pinned will not count the to-be-removed breakpoint. +*/ + next_tsk_pinned =task_bp_pinned (-1 , bp, type); + +if (next_tsk_pinned >=0 ) { +if (bp->cpu <0 ) {/* Case 1: fast path */ +if (!enable) + next_tsk_pinned +=hw_breakpoint_weight (bp); +bp_slots_histogram_add (&tsk_pinned_all[type], next_tsk_pinned, weight); + }else if (enable) {/* Case 2.a: slow path */ +/* Add existing to per-CPU histograms. */ +for_each_possible_cpu (cpu) { +bp_slots_histogram_add (&get_bp_info (cpu, type)->tsk_pinned, +0 , next_tsk_pinned); + } +/* Add this first CPU-pinned task breakpoint. */ +bp_slots_histogram_add (&get_bp_info (bp->cpu, type)->tsk_pinned, + next_tsk_pinned, weight); +/* Rebalance global task pinned histogram. */ +bp_slots_histogram_add (&tsk_pinned_all[type], next_tsk_pinned, + -next_tsk_pinned); + }else {/* Case 2.b: slow path */ +/* Remove this last CPU-pinned task breakpoint. */ +bp_slots_histogram_add (&get_bp_info (bp->cpu, type)->tsk_pinned, + next_tsk_pinned +hw_breakpoint_weight (bp), weight); +/* Remove all from per-CPU histograms. */ +for_each_possible_cpu (cpu) { +bp_slots_histogram_add (&get_bp_info (cpu, type)->tsk_pinned, + next_tsk_pinned, -next_tsk_pinned); + } +/* Rebalance global task pinned histogram. */ +bp_slots_histogram_add (&tsk_pinned_all[type],0 , next_tsk_pinned); + } + }else {/* Case 3: slow path */ +const struct cpumask *cpumask =cpumask_of_bp (bp); + +for_each_cpu (cpu, cpumask) { + next_tsk_pinned =task_bp_pinned (cpu, bp, type); +if (!enable) + next_tsk_pinned +=hw_breakpoint_weight (bp); +bp_slots_histogram_add (&get_bp_info (cpu, type)->tsk_pinned, + next_tsk_pinned, weight); + } + }/* * Readers want a stable snapshot of the per-task breakpoint list.
@@ -419,8 +509,8 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,if (enable)return rhltable_insert (&task_bps_ht, &bp->hw.bp_list, task_bps_ht_params); -else -return rhltable_remove (&task_bps_ht, &bp->hw.bp_list, task_bps_ht_params); + +return 0 ; }__weak int arch_reserve_bp_slot (struct perf_event *bp)
@@ -850,6 +940,9 @@ bool hw_breakpoint_is_used(void)*/ if (WARN_ON (atomic_read (&cpu_pinned[type].count[slot])))return true; + +if (atomic_read (&tsk_pinned_all[type].count[slot])) +return true; } }
Signature · 1 line
2.37.2.672.g94769d06f0-goog
[tip: perf/core] perf/hw_breakpoint: Optimize toggle_bp_slot() for CPU-independent task targets
The following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: ecdfb8896f2ad733097e6309d64f94db4cd1020c Gitweb: https://git.kernel.org/tip/ecdfb8896f2ad733097e6309d64f94db4cd1020c Author: Marco Elver <[email protected]> AuthorDate: Mon, 29 Aug 2022 14:47:19 +02:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Tue, 30 Aug 2022 10:56:24 +02:00
perf/hw_breakpoint: Optimize toggle_bp_slot() for CPU-independent task targets
We can still see that a majority of the time is spent hashing task pointers:
...
16.98% [kernel] [k] rhashtable_jhash2
...Doing the bookkeeping in toggle_bp_slots() is currently O(#cpus), calling task_bp_pinned() for each CPU, even if task_bp_pinned() is CPU-independent. The reason for this is to update the per-CPU 'tsk_pinned' histogram.
To optimize the CPU-independent case to O(1), keep a separate CPU-independent 'tsk_pinned_all' histogram.
The major source of complexity are transitions between "all CPU-independent task breakpoints" and "mixed CPU-independent and CPU-dependent task breakpoints". The code comments list all cases that require handling.
After this optimization:
| $> perf bench -r 100 breakpoint thread -b 4 -p 128 -t 512 | # Running 'breakpoint/thread' benchmark: | # Created/joined 100 threads with 4 breakpoints and 128 parallelism | Total time: 1.758 [sec] | | 34.336621 usecs/op | 4395.087500 usecs/op/cpu
38.08% [kernel] [k] queued_spin_lock_slowpath
10.81% [kernel] [k] smp_cfm_core_cond
3.01% [kernel] [k] update_sg_lb_stats
2.58% [kernel] [k] osq_lock
2.57% [kernel] [k] llist_reverse_order
1.45% [kernel] [k] find_next_bit
1.21% [kernel] [k] flush_tlb_func_common
1.01% [kernel] [k] arch_install_hw_breakpointShowing that the time spent hashing keys has become insignificant.
With the given benchmark parameters, that's an improvement of 12% compared with the old O(#cpus) version.
And finally, using the less aggressive parameters from the preceding changes, we now observe:
| $> perf bench -r 30 breakpoint thread -b 4 -p 64 -t 64 | # Running 'breakpoint/thread' benchmark: | # Created/joined 30 threads with 4 breakpoints and 64 parallelism | Total time: 0.067 [sec] | | 35.292187 usecs/op | 2258.700000 usecs/op/cpu
Which is an improvement of 12% compared to without the histogram optimizations (baseline is 40 usecs/op). This is now on par with the theoretical ideal (constraints disabled), and only 12% slower than no breakpoints at all.
--- kernel/events/hw_breakpoint.c | 155 ++++++++++++++++++++++++++------- 1 file changed, 124 insertions(+), 31 deletions(-)
kernel/events/hw_breakpoint.c +124 -31
@@ -66,6 +66,8 @@ static struct bp_cpuinfo *get_bp_info(int cpu, enum bp_type_idx type)/* Number of pinned CPU breakpoints globally. */ static struct bp_slots_histogram cpu_pinned[TYPE_MAX ]; +/* Number of pinned CPU-independent task breakpoints. */ +static struct bp_slots_histogram tsk_pinned_all[TYPE_MAX ];/* Keep track of the breakpoints attached to tasks */ static struct rhltable task_bps_ht;
@@ -200,6 +202,8 @@ static __init int init_breakpoint_slots(void)for (i =0 ; i <TYPE_MAX ; i++) {if (!bp_slots_histogram_alloc (&cpu_pinned[i], i)) goto err; +if (!bp_slots_histogram_alloc (&tsk_pinned_all[i], i)) + goto err; }return 0 ;
@@ -210,8 +214,10 @@ err:if (err_cpu == cpu)break ; } -for (i =0 ; i <TYPE_MAX ; i++) +for (i =0 ; i <TYPE_MAX ; i++) {bp_slots_histogram_free (&cpu_pinned[i]); +bp_slots_histogram_free (&tsk_pinned_all[i]); + }return -ENOMEM ; }
@@ -245,6 +251,26 @@ bp_slots_histogram_max(struct bp_slots_histogram *hist, enum bp_type_idx type)return 0 ; } +static int +bp_slots_histogram_max_merge (struct bp_slots_histogram *hist1,struct bp_slots_histogram *hist2, +enum bp_type_idx type) +{ +for (int i =hw_breakpoint_slots_cached (type) -1 ; i >=0 ; i--) { +const int count1 =atomic_read (&hist1->count[i]); +const int count2 =atomic_read (&hist2->count[i]); + +/* Catch unexpected writers; we want a stable snapshot. */ +ASSERT_EXCLUSIVE_WRITER (hist1->count[i]); +ASSERT_EXCLUSIVE_WRITER (hist2->count[i]); +if (count1 + count2 >0 ) +return i +1 ; +WARN (count1 <0 ,"inconsistent breakpoint slots histogram" ); +WARN (count2 <0 ,"inconsistent breakpoint slots histogram" ); + } + +return 0 ; +} +#ifndef hw_breakpoint_weightstatic inline int hw_breakpoint_weight (struct perf_event *bp) {
@@ -273,7 +299,7 @@ static unsigned int max_task_bp_pinned(int cpu, enum bp_type_idx type)* toggle_bp_task_slot() to tsk_pinned, and we get a stable snapshot. */ lockdep_assert_held_write (&bp_cpuinfo_sem); -return bp_slots_histogram_max (tsk_pinned, type); +return bp_slots_histogram_max_merge (tsk_pinned, &tsk_pinned_all[type], type); }/*
@@ -367,39 +393,21 @@ max_bp_pinned_slots(struct perf_event *bp, enum bp_type_idx type) }/* -* Add a pinned breakpoint for the given task in our constraint table -*/ -static void toggle_bp_task_slot (struct perf_event *bp,int cpu, -enum bp_type_idx type,int weight) -{ -struct bp_slots_histogram *tsk_pinned = &get_bp_info (cpu, type)->tsk_pinned; - -/* -* If bp->hw.target, tsk_pinned is only modified, but not used -* otherwise. We can permit concurrent updates as long as there are no -* other uses: having acquired bp_cpuinfo_sem as a reader allows -* concurrent updates here. Uses of tsk_pinned will require acquiring -* bp_cpuinfo_sem as a writer to stabilize tsk_pinned's value. -*/ -lockdep_assert_held_read (&bp_cpuinfo_sem); -bp_slots_histogram_add (tsk_pinned,task_bp_pinned (cpu, bp, type), weight); -} - -/* * Add/remove the given breakpoint in our constraint table */ static int -toggle_bp_slot (struct perf_event *bp,bool enable,enum bp_type_idx type, -int weight) +toggle_bp_slot (struct perf_event *bp,bool enable,enum bp_type_idx type,int weight) { -const struct cpumask *cpumask =cpumask_of_bp (bp); -int cpu; +int cpu, next_tsk_pinned;if (!enable) weight = -weight; -/* Pinned counter cpu profiling */ if (!bp->hw.target) { +/* +* Update the pinned CPU slots, in per-CPU bp_cpuinfo and in the +* global histogram. +*/ struct bp_cpuinfo *info =get_bp_info (bp->cpu, type);lockdep_assert_held_write (&bp_cpuinfo_sem);
@@ -408,9 +416,91 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,return 0 ; } -/* Pinned counter task profiling */ -for_each_cpu (cpu, cpumask) -toggle_bp_task_slot (bp, cpu, type, weight); +/* +* If bp->hw.target, tsk_pinned is only modified, but not used +* otherwise. We can permit concurrent updates as long as there are no +* other uses: having acquired bp_cpuinfo_sem as a reader allows +* concurrent updates here. Uses of tsk_pinned will require acquiring +* bp_cpuinfo_sem as a writer to stabilize tsk_pinned's value. +*/ +lockdep_assert_held_read (&bp_cpuinfo_sem); + +/* +* Update the pinned task slots, in per-CPU bp_cpuinfo and in the global +* histogram. We need to take care of 4 cases: +* +* 1. This breakpoint targets all CPUs (cpu < 0), and there may only +* exist other task breakpoints targeting all CPUs. In this case we +* can simply update the global slots histogram. +* +* 2. This breakpoint targets a specific CPU (cpu >= 0), but there may +* only exist other task breakpoints targeting all CPUs. +* +* a. On enable: remove the existing breakpoints from the global +* slots histogram and use the per-CPU histogram. +* +* b. On disable: re-insert the existing breakpoints into the global +* slots histogram and remove from per-CPU histogram. +* +* 3. Some other existing task breakpoints target specific CPUs. Only +* update the per-CPU slots histogram. +*/ + +if (!enable) { +/* +* Remove before updating histograms so we can determine if this +* was the last task breakpoint for a specific CPU. +*/ +int ret =rhltable_remove (&task_bps_ht, &bp->hw.bp_list, task_bps_ht_params); + +if (ret) +return ret; + } +/* +* Note: If !enable, next_tsk_pinned will not count the to-be-removed breakpoint. +*/ + next_tsk_pinned =task_bp_pinned (-1 , bp, type); + +if (next_tsk_pinned >=0 ) { +if (bp->cpu <0 ) {/* Case 1: fast path */ +if (!enable) + next_tsk_pinned +=hw_breakpoint_weight (bp); +bp_slots_histogram_add (&tsk_pinned_all[type], next_tsk_pinned, weight); + }else if (enable) {/* Case 2.a: slow path */ +/* Add existing to per-CPU histograms. */ +for_each_possible_cpu (cpu) { +bp_slots_histogram_add (&get_bp_info (cpu, type)->tsk_pinned, +0 , next_tsk_pinned); + } +/* Add this first CPU-pinned task breakpoint. */ +bp_slots_histogram_add (&get_bp_info (bp->cpu, type)->tsk_pinned, + next_tsk_pinned, weight); +/* Rebalance global task pinned histogram. */ +bp_slots_histogram_add (&tsk_pinned_all[type], next_tsk_pinned, + -next_tsk_pinned); + }else {/* Case 2.b: slow path */ +/* Remove this last CPU-pinned task breakpoint. */ +bp_slots_histogram_add (&get_bp_info (bp->cpu, type)->tsk_pinned, + next_tsk_pinned +hw_breakpoint_weight (bp), weight); +/* Remove all from per-CPU histograms. */ +for_each_possible_cpu (cpu) { +bp_slots_histogram_add (&get_bp_info (cpu, type)->tsk_pinned, + next_tsk_pinned, -next_tsk_pinned); + } +/* Rebalance global task pinned histogram. */ +bp_slots_histogram_add (&tsk_pinned_all[type],0 , next_tsk_pinned); + } + }else {/* Case 3: slow path */ +const struct cpumask *cpumask =cpumask_of_bp (bp); + +for_each_cpu (cpu, cpumask) { + next_tsk_pinned =task_bp_pinned (cpu, bp, type); +if (!enable) + next_tsk_pinned +=hw_breakpoint_weight (bp); +bp_slots_histogram_add (&get_bp_info (cpu, type)->tsk_pinned, + next_tsk_pinned, weight); + } + }/* * Readers want a stable snapshot of the per-task breakpoint list.
@@ -419,8 +509,8 @@ toggle_bp_slot(struct perf_event *bp, bool enable, enum bp_type_idx type,if (enable)return rhltable_insert (&task_bps_ht, &bp->hw.bp_list, task_bps_ht_params); -else -return rhltable_remove (&task_bps_ht, &bp->hw.bp_list, task_bps_ht_params); + +return 0 ; }__weak int arch_reserve_bp_slot (struct perf_event *bp)
@@ -850,6 +940,9 @@ bool hw_breakpoint_is_used(void)*/ if (WARN_ON (atomic_read (&cpu_pinned[type].count[slot])))return true; + +if (atomic_read (&tsk_pinned_all[type].count[slot])) +return true; } }