percpu: use raw_local_irq_* in _this_cpu op
PATCH 1/1 percpu: use raw_local_irq_* in _this_cpu op
It doesn't make sense to trace irq off or do irq flags lock proving inside 'this_cpu' operations, so replace local_irq_* with raw_local_irq_* in 'this_cpu' op.
Also the patch fixes one lockdep warning[1], which is caused by the added local_irq_save/restore(flags) in this_cpu_inc called by __debug_atomic_inc: kernel/lockdep.c
[1],
[ 0.162841] ------------[ cut here ]------------ [ 0.167694] WARNING: at kernel/lockdep.c:3493 check_flags+0xc0/0x1d0() [ 0.174468] Modules linked in: [ 0.177703] Backtrace: [ 0.180328] [<c00171f0>] (dump_backtrace+0x0/0x110) from [<c0412320>] (dump_stack+0x18/0x1c) [ 0.189086] r6:c051f778 r5:00000da5 r4:00000000 r3:60000093 [ 0.195007] [<c0412308>] (dump_stack+0x0/0x1c) from [<c00410e8>] (warn_slowpath_common+0x54/0x6c) [ 0.204223] [<c0041094>] (warn_slowpath_common+0x0/0x6c) from [<c0041124>] (warn_slowpath_null+0x24/0x2c) [ 0.214111] r8:00000000 r7:00000000 r6:ee069598 r5:60000013 r4:ee082000 [ 0.220825] r3:00000009 [ 0.223693] [<c0041100>] (warn_slowpath_null+0x0/0x2c) from [<c0088f38>] (check_flags+0xc0/0x1d0) [ 0.232910] [<c0088e78>] (check_flags+0x0/0x1d0) from [<c008d348>] (lock_acquire+0x4c/0x11c) [ 0.241668] [<c008d2fc>] (lock_acquire+0x0/0x11c) from [<c0415aa4>] (_raw_spin_lock+0x3c/0x74) [ 0.250610] [<c0415a68>] (_raw_spin_lock+0x0/0x74) from [<c010a844>] (set_task_comm+0x20/0xc0) [ 0.259521] r6:ee069588 r5:ee0691c0 r4:ee082000 [ 0.264404] [<c010a824>] (set_task_comm+0x0/0xc0) from [<c0060780>] (kthreadd+0x28/0x108) [ 0.272857] r8:00000000 r7:00000013 r6:c0044a08 r5:ee0691c0 r4:ee082000 [ 0.279571] r3:ee083fe0 [ 0.282470] [<c0060758>] (kthreadd+0x0/0x108) from [<c0044a08>] (do_exit+0x0/0x6dc) [ 0.290405] r5:c0060758 r4:00000000 [ 0.294189] ---[ end trace 1b75b31a2719ed1c ]--- [ 0.299041] possible reason: unannotated irqs-on. [ 0.303955] irq event stamp: 5 [ 0.307159] hardirqs last enabled at (4): [<c001331c>] no_work_pending+0x8/0x2c [ 0.314880] hardirqs last disabled at (5): [<c0089b08>] trace_hardirqs_on_caller+0x60/0x26c [ 0.323547] softirqs last enabled at (0): [<c003f754>] copy_process+0x33c/0xef4 [ 0.331207] softirqs last disabled at (0): [< (null)>] (null) [ 0.337585] CPU0: thread -1, cpu 0, socket 0, mpidr 80000000
--- include/linux/percpu.h | 20 ++++++++++---------- 1 files changed, 10 insertions(+), 10 deletions(-)
include/linux/percpu.h +10 -10
@@ -294,9 +294,9 @@ do { \#define _this_cpu_generic_to_op (pcp, val, op) \ do { \ unsigned long flags; \ - local_irq_save(flags); \ + raw_local_irq_save(flags); \ *__this_cpu_ptr(&(pcp)) op val; \ - local_irq_restore(flags); \ + raw_local_irq_restore(flags); \ } while (0)#ifndef this_cpu_write
@@ -395,10 +395,10 @@ do { \ ({ \typeof (pcp) ret__; \unsigned long flags; \ -local_irq_save (flags); \ +raw_local_irq_save (flags); \__this_cpu_add (pcp, val); \ ret__ =__this_cpu_read (pcp); \ -local_irq_restore (flags); \ +raw_local_irq_restore (flags); \ ret__; \ })
@@ -425,10 +425,10 @@ do { \#define _this_cpu_generic_xchg (pcp, nval) \ ({ typeof(pcp) ret__; \ unsigned long flags; \ - local_irq_save(flags); \ + raw_local_irq_save(flags); \ ret__ = __this_cpu_read(pcp); \ __this_cpu_write(pcp, nval); \ - local_irq_restore(flags); \ + raw_local_irq_restore(flags); \ ret__; \ })
@@ -453,11 +453,11 @@ do { \ ({ \typeof (pcp) ret__; \unsigned long flags; \ -local_irq_save (flags); \ +raw_local_irq_save (flags); \ ret__ =__this_cpu_read (pcp); \if (ret__ == (oval)) \__this_cpu_write (pcp, nval); \ -local_irq_restore (flags); \ +raw_local_irq_restore (flags); \ ret__; \ })
@@ -490,10 +490,10 @@ do { \ ({ \int ret__; \unsigned long flags; \ -local_irq_save (flags); \ +raw_local_irq_save (flags); \ ret__ =__this_cpu_generic_cmpxchg_double (pcp1, pcp2, \ oval1, oval2, nval1, nval2); \ -local_irq_restore (flags); \ +raw_local_irq_restore (flags); \ ret__; \ })
Signature · 1 line
1.7.9
Quoted 3 lines · Ming Lei
It doesn't make sense to trace irq off or do irq flags lock proving inside '
this_cpu' operations, so replace local_irq_* with raw_local_irq_* in 'this_cpu' op.
Tejun Heo
maintainerQuoted 7 lines · Ming Lei
It doesn't make sense to trace irq off or do irq flags lock proving inside '
this_cpu' operations, so replace local_irq_* with raw_local_irq_* in 'this_cpu' op.Also the patch fixes one lockdep warning[1], which is caused by the added local_irq_save/restore(flags) in
this_cpu_inccalled by__debug_atomic_inc:kernel/lockdep.c
I think this isn't gonna hurt anything but I don't understand why the lockdep warning is triggering when using traced version. Can you please explain that in a bit more detail in the patch description?
Thank you.
Signature · 1 line
tejun
Hi,
Quoted 5 lines · Tejun Heo
Quoted 4 lines · Ming Lei
It doesn't make sense to trace irq off or do irq flags lock proving inside '
this_cpu' operations, so replace local_irq_* with raw_local_irq_* in 'this_cpu' op.
Quoted 7 lines
Quoted 3 lines
Also the patch fixes one lockdep warning[1], which is caused by the added local_irq_save/restore(flags) in
this_cpu_inccalled by__debug_atomic_inc:kernel/lockdep.cI think this isn't gonna hurt anything but I don't understand why the lockdep warning is triggering when using traced version. Can you please explain that in a bit more detail in the patch description?
In trace_hardirqs_on_caller:kernel/lockdep.c, __debug_atomic_inc will be called to add on 'this_cpu' variable, so may introduce recursive trace_hardirqs_on|off_caller called.
For the lockdep warning, I reproduced it on ARM, see the path below:
kernel_thread_helper /*irq disabled*/
->entry: trace_hardirqs_on_caller /*hardirqs_enabled was set*/ ->trace_hardirqs_off_caller /*hardirqs_enabled cleared*/ __this_cpu_add(redundant_hardirqs_on) ->trace_hardirqs_off_caller /*irq disabled, so call here*/
so the 'unannotated irqs-on' warning will be triggered somewhere because irq will be enabled just after the irq trace inside kernel_thread_helper.
You can refer to log of commit ac78884e6d89714d18b32b5b7d574116ecfb7c88 (ARM: lockdep: fix unannotated irqs-on) about irq trace inside kernel_thread_helper.
thanks,
Signature · 1 line
Ming Lei
Quoted 12 lines · Ming Lei
Hi,
Quoted 9 lines · Tejun Heo
Quoted 7 lines · Ming Lei
It doesn't make sense to trace irq off or do irq flags lock proving inside '
this_cpu' operations, so replace local_irq_* with raw_local_irq_* in 'this_cpu' op.Also the patch fixes one lockdep warning[1], which is caused by the added local_irq_save/restore(flags) in
this_cpu_inccalled by__debug_atomic_inc:kernel/lockdep.c
Quoted 7 lines
Quoted 3 lines
I think this isn't gonna hurt anything but I don't understand why the lockdep warning is triggering when using traced version. ?Can you please explain that in a bit more detail in the patch description?
In trace_hardirqs_on_caller:kernel/lockdep.c,
__debug_atomic_incwill be called to add on 'this_cpu' variable, so may introduce recursive trace_hardirqs_on|off_caller called.
Don't we need to prevent this kind of recursion first?
UNTESTED patch, I guess it'll smooth your concern.
---
kernel/lockdep.c +8 -5
@@ -2561,6 +2561,8 @@ void trace_hardirqs_on_caller(unsigned long ip)if (unlikely (!debug_locks || current->lockdep_recursion))return ; + current->lockdep_recursion =1 ; +if (unlikely (current->hardirqs_enabled)) {/* * Neither irq nor preemption are disabled here
@@ -2568,7 +2570,7 @@ void trace_hardirqs_on_caller(unsigned long ip)* in a stat is not a big deal. */ __debug_atomic_inc (redundant_hardirqs_on); -return ; + goto out; }/*
@@ -2577,23 +2579,24 @@ void trace_hardirqs_on_caller(unsigned long ip)* enabled.. someone messed up their IRQ state tracing. */ if (DEBUG_LOCKS_WARN_ON (!irqs_disabled ())) -return ; + goto out;/* * See the fine text that goes along with this variable definition. */ if (DEBUG_LOCKS_WARN_ON (unlikely (early_boot_irqs_disabled))) -return ; + goto out;/* * Can't allow enabling interrupts while in an interrupt handler, * that's general bad form and such. Recursion, limited stack etc.. */ if (DEBUG_LOCKS_WARN_ON (current->hardirq_context)) -return ; + goto out; - current->lockdep_recursion =1 ;__trace_hardirqs_on_caller (ip); + +out: current->lockdep_recursion =0 ; }EXPORT_SYMBOL (trace_hardirqs_on_caller);
Quoted 17 lines · Yong Zhang
Quoted 16 lines · Ming Lei
Hi,
Quoted 12 lines · Tejun Heo
Quoted 7 lines · Ming Lei
It doesn't make sense to trace irq off or do irq flags lock proving inside '
this_cpu' operations, so replace local_irq_* with raw_local_irq_* in 'this_cpu' op.Also the patch fixes one lockdep warning[1], which is caused by the added local_irq_save/restore(flags) in
this_cpu_inccalled by__debug_atomic_inc:kernel/lockdep.cI think this isn't gonna hurt anything but I don't understand why the lockdep warning is triggering when using traced version. ?Can you please explain that in a bit more detail in the patch description?
Quoted 5 lines
Quoted 3 lines
In trace_hardirqs_on_caller:kernel/lockdep.c,
__debug_atomic_incwill be called to add on 'this_cpu' variable, so may introduce recursive trace_hardirqs_on|off_caller called.Don't we need to prevent this kind of recursion first?
IMO, lockdep is designed as not tracing or proving itself, and just avoiding to trace __debug_atomic_inc is enough to fix the warning, so it is not necessary to enlarge the protection range with current->lockdep_recursion.
Quoted 44 lines
UNTESTED patch, I guess it'll smooth your concern.
---kernel/lockdep.c +5
-3@@ -2561,6 +2561,8 @@ void trace_hardirqs_on_caller(unsigned long ip) if (unlikely(!debug_locks || current->lockdep_recursion)) return; + current->lockdep_recursion = 1; + if (unlikely(current->hardirqs_enabled)) { /** Neither irq nor preemption are disabled here @@ -2568,7 +2570,7 @@ void trace_hardirqs_on_caller(unsigned long ip)* in a stat is not a big deal. */ __debug_atomic_inc(redundant_hardirqs_on); - return; + goto out; } /*@@ -2577,23 +2579,24 @@ void trace_hardirqs_on_caller(unsigned long ip)* enabled.. someone messed up their IRQ state tracing. */ if (DEBUG_LOCKS_WARN_ON(!irqs_disabled())) - return; + goto out; /* * See the fine text that goes along with this variable definition. */ if (DEBUG_LOCKS_WARN_ON(unlikely(early_boot_irqs_disabled))) - return; + goto out; /* * Can't allow enabling interrupts while in an interrupt handler, * that's general bad form and such. Recursion, limited stack etc.. */ if (DEBUG_LOCKS_WARN_ON(current->hardirq_context))
kernel/lockdep.c void trace_hardirqs_on_caller(unsigned long ip) · line 2602
Quoted 10 lines
kernel/lockdep.c +3
-2- return; + goto out; - current->lockdep_recursion = 1; __trace_hardirqs_on_caller(ip); + +out: current->lockdep_recursion = 0; } EXPORT_SYMBOL(trace_hardirqs_on_caller);
thanks,
Signature · 1 line
Ming Lei
Tejun Heo
maintainerQuoted 10 lines · Ming Lei
Quoted 9 lines · Tejun Heo
Quoted 7 lines · Ming Lei
It doesn't make sense to trace irq off or do irq flags lock proving inside '
this_cpu' operations, so replace local_irq_* with raw_local_irq_* in 'this_cpu' op.Also the patch fixes one lockdep warning[1], which is caused by the added local_irq_save/restore(flags) in
this_cpu_inccalled by__debug_atomic_inc:kernel/lockdep.c
Quoted 7 lines
Quoted 3 lines
I think this isn't gonna hurt anything but I don't understand why the lockdep warning is triggering when using traced version. Can you please explain that in a bit more detail in the patch description?
In trace_hardirqs_on_caller:kernel/lockdep.c,
__debug_atomic_incwill be called to add on 'this_cpu' variable, so may introduce recursive trace_hardirqs_on|off_caller called.
Ah, okay, so lockdep itself is using this_cpu ops. Can you please repost the patch with the above info in the description?
Thank you.
Signature · 1 line
tejun