* [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
@ 2017-11-28 19:45 Yang Shi
2017-11-28 20:24 ` Waiman Long
0 siblings, 1 reply; 12+ messages in thread
From: Yang Shi @ 2017-11-28 19:45 UTC (permalink / raw)
To: longman, tglx; +Cc: Yang Shi, linux-kernel
There are nested loops on debug objects free path, sometimes it may take
over hundred thousands of loops, then cause soft lockup with !CONFIG_PREEMPT
occasionally, like below:
NMI watchdog: BUG: soft lockup - CPU#15 stuck for 22s! [stress-ng-getde:110342]
Modules linked in: binfmt_misc(E) tcp_diag(E)
inet_diag(E) bonding(E) intel_rapl(E) iosf_mbi(E)
x86_pkg_temp_thermal(E) coretemp(E) iTCO_wdt(E) iTCO_vendor_support(E)
kvm_intel(E) kvm(E) irqbypass(E) crct10dif_pclmul(E) crc32_pclmul(E)
dcdbas(E) ghash_clmulni_intel(E) aesni_intel(E) lrw(E) gf128mul(E)
glue_helper(E) ablk_helper(E) ipmi_devintf(E) sg(E) cryptd(E) pcspkr(E)
mei_me(E) lpc_ich(E) ipmi_si(E) mfd_core(E) mei(E) shpchp(E) wmi(E)
ipmi_msghandler(E) acpi_power_meter(E) nfsd(E) auth_rpcgss(E) nfs_acl(E)
lockd(E) grace(E) sunrpc(E) ip_tables(E) ext4(E) jbd2(E) mbcache(E)
sd_mod(E) mgag200(E) igb(E) drm_kms_helper(E) ixgbe(E) syscopyarea(E)
mdio(E) sysfillrect(E) sysimgblt(E) ptp(E) fb_sys_fops(E) pps_core(E)
ttm(E) drm(E) crc32c_intel(E) i2c_algo_bit(E) i2c_core(E)
megaraid_sas(E)
dca(E)
irq event stamp: 4340444
hardirqs last enabled at (4340443): [<ffffffff817d6476>]
_raw_spin_unlock_irqrestore+0x36/0x60
hardirqs last disabled at (4340444): [<ffffffff817d8c81>]
apic_timer_interrupt+0x91/0xa0
softirqs last enabled at (4340398): [<ffffffff817da179>]
__do_softirq+0x349/0x50e
softirqs last disabled at (4340391): [<ffffffff810a5255>]
irq_exit+0xf5/0x110
CPU: 15 PID: 110342 Comm: stress-ng-getde Tainted: G
E 4.9.44-003.ali3000.alios7.x86_64.debug #1
Hardware name: Dell Inc. PowerEdge R720xd/0X6FFV, BIOS
1.6.0 03/07/2013
task: ffff884cbb0d0000 task.stack: ffff884cabc70000
RIP: 0010:[<ffffffff817d647b>] [<ffffffff817d647b>]
_raw_spin_unlock_irqrestore+0x3b/0x60
RSP: 0018:ffff884cabc77b78 EFLAGS: 00000292
RAX: ffff884cbb0d0000 RBX: 0000000000000292 RCX: 0000000000000000
RDX: ffff884cbb0d0000 RSI: 0000000000000001 RDI: 0000000000000292
RBP: ffff884cabc77b88 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000001 R11: 0000000000000001 R12: ffffffff8357a0d8
R13: ffff884cabc77bc8 R14: ffffffff8357a0d0 R15: 00000000000000fc
FS: 00002aee845fd2c0(0000) GS:ffff8852bd400000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000002991808 CR3: 0000005123abf000 CR4: 00000000000406e0
Stack:
ffff884ff4fe0000 ffff884ff4fd8000 ffff884cabc77c00 ffffffff8141177e
0000000000000202 ffff884cbb0d0000 ffff884cabc77bc8 0000000000000006
ffff884ff4fda000 ffffffff8357a0d8 0000000000000000 91f5d976f6020b6c
Call Trace:
[<ffffffff8141177e>] debug_check_no_obj_freed+0x13e/0x220
[<ffffffff811f8751>] __free_pages_ok+0x1f1/0x5c0
[<ffffffff811fa785>] __free_pages+0x25/0x40
[<ffffffff812638db>] __free_slab+0x19b/0x270
[<ffffffff812639e9>] discard_slab+0x39/0x50
[<ffffffff812679f7>] __slab_free+0x207/0x270
[<ffffffff81269966>] ___cache_free+0xa6/0xb0
[<ffffffff8126c267>] qlist_free_all+0x47/0x80
[<ffffffff8126c5a9>] quarantine_reduce+0x159/0x190
[<ffffffff8126b3bf>] kasan_kmalloc+0xaf/0xc0
[<ffffffff8126b8a2>] kasan_slab_alloc+0x12/0x20
[<ffffffff81265e8a>] kmem_cache_alloc+0xfa/0x360
[<ffffffff812abc8f>] ? getname_flags+0x4f/0x1f0
[<ffffffff812abc8f>] getname_flags+0x4f/0x1f0
[<ffffffff812abe42>] getname+0x12/0x20
[<ffffffff81298da9>] do_sys_open+0xf9/0x210
[<ffffffff81298ede>] SyS_open+0x1e/0x20
[<ffffffff817d6e01>] entry_SYSCALL_64_fastpath+0x1f/0xc2
Code: 7f 18 53 48 8b 55 08 48 89 f3 be 01 00 00 00 e8 3c
cd 92 ff 4c 89 e7 e8 f4 0e 93 ff f6 c7 02 74 1b e8 3a ac 92 ff 48 89 df
57 9d <66> 66 90 66 90 65 ff 0d d1 ff 83 7e 5b 41 5c 5d c3 48 89 df 57
The code path might be called in either atomic or non-atomic context,
so touching softlockup watchdog instead of calling cond_resched() which
might fall asleep. However, it is unnecessary to touch the watchdog
every loop, so just touch the watchdog at every 10000 (best estimate) loops.
And, introduce a new knob: /sys/kernel/debug/debug_objects/suppress_lockup.
0 value means not suppress the softlockup message by touching the
watchdog, non-zero value means suppress the softlockup message.
The default value is zero.
Signed-off-by: Yang Shi <yang.s@alibaba-inc.com>
CC: Waiman Long <longman@redhat.com>
CC: Thomas Gleixner <tglx@linutronix.de>
---
v2 --> v3:
* Use debugfs_create_u32() helper API per Waiman's suggestion
v1 --> v2:
* Added suppress_lockup knob in debugfs per Waiman's suggestion
lib/debugobjects.c | 16 ++++++++++++++--
1 file changed, 14 insertions(+), 2 deletions(-)
diff --git a/lib/debugobjects.c b/lib/debugobjects.c
index 166488d..c960221 100644
--- a/lib/debugobjects.c
+++ b/lib/debugobjects.c
@@ -19,6 +19,7 @@
#include <linux/slab.h>
#include <linux/hash.h>
#include <linux/kmemleak.h>
+#include <linux/nmi.h>
#define ODEBUG_HASH_BITS 14
#define ODEBUG_HASH_SIZE (1 << ODEBUG_HASH_BITS)
@@ -67,6 +68,8 @@ struct debug_bucket {
static int debug_objects_allocated;
static int debug_objects_freed;
+static int suppress_lockup;
+
static void free_obj_work(struct work_struct *work);
static DECLARE_WORK(debug_obj_work, free_obj_work);
@@ -768,6 +771,10 @@ static void __debug_check_no_obj_freed(const void *address, unsigned long size)
debug_objects_maxchain = cnt;
max_loops += cnt;
+
+ if (max_loops > 10000 && ((max_loops % 10000) == 0)
+ && suppress_lockup != 0)
+ touch_softlockup_watchdog();
}
if (max_loops > debug_objects_maxloops)
@@ -812,7 +819,7 @@ static int debug_stats_open(struct inode *inode, struct file *filp)
static int __init debug_objects_init_debugfs(void)
{
- struct dentry *dbgdir, *dbgstats;
+ struct dentry *dbgdir, *dbgstats, *dbglockup;
if (!debug_objects_enabled)
return 0;
@@ -826,10 +833,15 @@ static int __init debug_objects_init_debugfs(void)
if (!dbgstats)
goto err;
+ dbglockup = debugfs_create_u32("suppress_lockup", 0644, dbgdir,
+ &suppress_lockup);
+ if (!dbglockup)
+ goto err;
+
return 0;
err:
- debugfs_remove(dbgdir);
+ debugfs_remove_recursive(dbgdir);
return -ENOMEM;
}
--
1.8.3.1
^ permalink raw reply related [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2017-11-28 19:45 [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT Yang Shi
@ 2017-11-28 20:24 ` Waiman Long
2017-12-07 19:16 ` Yang Shi
0 siblings, 1 reply; 12+ messages in thread
From: Waiman Long @ 2017-11-28 20:24 UTC (permalink / raw)
To: Yang Shi, tglx; +Cc: linux-kernel
On 11/28/2017 02:45 PM, Yang Shi wrote:
> There are nested loops on debug objects free path, sometimes it may take
> over hundred thousands of loops, then cause soft lockup with !CONFIG_PREEMPT
> occasionally, like below:
>
> NMI watchdog: BUG: soft lockup - CPU#15 stuck for 22s! [stress-ng-getde:110342]
> Modules linked in: binfmt_misc(E) tcp_diag(E)
> inet_diag(E) bonding(E) intel_rapl(E) iosf_mbi(E)
> x86_pkg_temp_thermal(E) coretemp(E) iTCO_wdt(E) iTCO_vendor_support(E)
> kvm_intel(E) kvm(E) irqbypass(E) crct10dif_pclmul(E) crc32_pclmul(E)
> dcdbas(E) ghash_clmulni_intel(E) aesni_intel(E) lrw(E) gf128mul(E)
> glue_helper(E) ablk_helper(E) ipmi_devintf(E) sg(E) cryptd(E) pcspkr(E)
> mei_me(E) lpc_ich(E) ipmi_si(E) mfd_core(E) mei(E) shpchp(E) wmi(E)
> ipmi_msghandler(E) acpi_power_meter(E) nfsd(E) auth_rpcgss(E) nfs_acl(E)
> lockd(E) grace(E) sunrpc(E) ip_tables(E) ext4(E) jbd2(E) mbcache(E)
> sd_mod(E) mgag200(E) igb(E) drm_kms_helper(E) ixgbe(E) syscopyarea(E)
> mdio(E) sysfillrect(E) sysimgblt(E) ptp(E) fb_sys_fops(E) pps_core(E)
> ttm(E) drm(E) crc32c_intel(E) i2c_algo_bit(E) i2c_core(E)
> megaraid_sas(E)
> dca(E)
> irq event stamp: 4340444
> hardirqs last enabled at (4340443): [<ffffffff817d6476>]
> _raw_spin_unlock_irqrestore+0x36/0x60
> hardirqs last disabled at (4340444): [<ffffffff817d8c81>]
> apic_timer_interrupt+0x91/0xa0
> softirqs last enabled at (4340398): [<ffffffff817da179>]
> __do_softirq+0x349/0x50e
> softirqs last disabled at (4340391): [<ffffffff810a5255>]
> irq_exit+0xf5/0x110
> CPU: 15 PID: 110342 Comm: stress-ng-getde Tainted: G
> E 4.9.44-003.ali3000.alios7.x86_64.debug #1
> Hardware name: Dell Inc. PowerEdge R720xd/0X6FFV, BIOS
> 1.6.0 03/07/2013
> task: ffff884cbb0d0000 task.stack: ffff884cabc70000
> RIP: 0010:[<ffffffff817d647b>] [<ffffffff817d647b>]
> _raw_spin_unlock_irqrestore+0x3b/0x60
> RSP: 0018:ffff884cabc77b78 EFLAGS: 00000292
> RAX: ffff884cbb0d0000 RBX: 0000000000000292 RCX: 0000000000000000
> RDX: ffff884cbb0d0000 RSI: 0000000000000001 RDI: 0000000000000292
> RBP: ffff884cabc77b88 R08: 0000000000000000 R09: 0000000000000000
> R10: 0000000000000001 R11: 0000000000000001 R12: ffffffff8357a0d8
> R13: ffff884cabc77bc8 R14: ffffffff8357a0d0 R15: 00000000000000fc
> FS: 00002aee845fd2c0(0000) GS:ffff8852bd400000(0000) knlGS:0000000000000000
> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 0000000002991808 CR3: 0000005123abf000 CR4: 00000000000406e0
> Stack:
> ffff884ff4fe0000 ffff884ff4fd8000 ffff884cabc77c00 ffffffff8141177e
> 0000000000000202 ffff884cbb0d0000 ffff884cabc77bc8 0000000000000006
> ffff884ff4fda000 ffffffff8357a0d8 0000000000000000 91f5d976f6020b6c
> Call Trace:
> [<ffffffff8141177e>] debug_check_no_obj_freed+0x13e/0x220
> [<ffffffff811f8751>] __free_pages_ok+0x1f1/0x5c0
> [<ffffffff811fa785>] __free_pages+0x25/0x40
> [<ffffffff812638db>] __free_slab+0x19b/0x270
> [<ffffffff812639e9>] discard_slab+0x39/0x50
> [<ffffffff812679f7>] __slab_free+0x207/0x270
> [<ffffffff81269966>] ___cache_free+0xa6/0xb0
> [<ffffffff8126c267>] qlist_free_all+0x47/0x80
> [<ffffffff8126c5a9>] quarantine_reduce+0x159/0x190
> [<ffffffff8126b3bf>] kasan_kmalloc+0xaf/0xc0
> [<ffffffff8126b8a2>] kasan_slab_alloc+0x12/0x20
> [<ffffffff81265e8a>] kmem_cache_alloc+0xfa/0x360
> [<ffffffff812abc8f>] ? getname_flags+0x4f/0x1f0
> [<ffffffff812abc8f>] getname_flags+0x4f/0x1f0
> [<ffffffff812abe42>] getname+0x12/0x20
> [<ffffffff81298da9>] do_sys_open+0xf9/0x210
> [<ffffffff81298ede>] SyS_open+0x1e/0x20
> [<ffffffff817d6e01>] entry_SYSCALL_64_fastpath+0x1f/0xc2
> Code: 7f 18 53 48 8b 55 08 48 89 f3 be 01 00 00 00 e8 3c
> cd 92 ff 4c 89 e7 e8 f4 0e 93 ff f6 c7 02 74 1b e8 3a ac 92 ff 48 89 df
> 57 9d <66> 66 90 66 90 65 ff 0d d1 ff 83 7e 5b 41 5c 5d c3 48 89 df 57
>
> The code path might be called in either atomic or non-atomic context,
> so touching softlockup watchdog instead of calling cond_resched() which
> might fall asleep. However, it is unnecessary to touch the watchdog
> every loop, so just touch the watchdog at every 10000 (best estimate) loops.
>
> And, introduce a new knob: /sys/kernel/debug/debug_objects/suppress_lockup.
> 0 value means not suppress the softlockup message by touching the
> watchdog, non-zero value means suppress the softlockup message.
> The default value is zero.
>
> Signed-off-by: Yang Shi <yang.s@alibaba-inc.com>
> CC: Waiman Long <longman@redhat.com>
> CC: Thomas Gleixner <tglx@linutronix.de>
> ---
> v2 --> v3:
> * Use debugfs_create_u32() helper API per Waiman's suggestion
> v1 --> v2:
> * Added suppress_lockup knob in debugfs per Waiman's suggestion
>
> lib/debugobjects.c | 16 ++++++++++++++--
> 1 file changed, 14 insertions(+), 2 deletions(-)
>
> diff --git a/lib/debugobjects.c b/lib/debugobjects.c
> index 166488d..c960221 100644
> --- a/lib/debugobjects.c
> +++ b/lib/debugobjects.c
> @@ -19,6 +19,7 @@
> #include <linux/slab.h>
> #include <linux/hash.h>
> #include <linux/kmemleak.h>
> +#include <linux/nmi.h>
>
> #define ODEBUG_HASH_BITS 14
> #define ODEBUG_HASH_SIZE (1 << ODEBUG_HASH_BITS)
> @@ -67,6 +68,8 @@ struct debug_bucket {
> static int debug_objects_allocated;
> static int debug_objects_freed;
>
> +static int suppress_lockup;
> +
> static void free_obj_work(struct work_struct *work);
> static DECLARE_WORK(debug_obj_work, free_obj_work);
>
> @@ -768,6 +771,10 @@ static void __debug_check_no_obj_freed(const void *address, unsigned long size)
> debug_objects_maxchain = cnt;
>
> max_loops += cnt;
> +
> + if (max_loops > 10000 && ((max_loops % 10000) == 0)
> + && suppress_lockup != 0)
> + touch_softlockup_watchdog();
> }
>
> if (max_loops > debug_objects_maxloops)
> @@ -812,7 +819,7 @@ static int debug_stats_open(struct inode *inode, struct file *filp)
>
> static int __init debug_objects_init_debugfs(void)
> {
> - struct dentry *dbgdir, *dbgstats;
> + struct dentry *dbgdir, *dbgstats, *dbglockup;
>
> if (!debug_objects_enabled)
> return 0;
> @@ -826,10 +833,15 @@ static int __init debug_objects_init_debugfs(void)
> if (!dbgstats)
> goto err;
>
> + dbglockup = debugfs_create_u32("suppress_lockup", 0644, dbgdir,
> + &suppress_lockup);
> + if (!dbglockup)
> + goto err;
> +
> return 0;
>
> err:
> - debugfs_remove(dbgdir);
> + debugfs_remove_recursive(dbgdir);
>
> return -ENOMEM;
> }
I am OK with your patch. Now it is up to Thomas to decide if he will
pick up your patch.
Cheers,
Longman
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2017-11-28 20:24 ` Waiman Long
@ 2017-12-07 19:16 ` Yang Shi
2017-12-28 15:36 ` Thomas Gleixner
0 siblings, 1 reply; 12+ messages in thread
From: Yang Shi @ 2017-12-07 19:16 UTC (permalink / raw)
To: Waiman Long, tglx; +Cc: linux-kernel
Hi Thomas,
Waiman has reviewed the patches, any comment from you is appreciated.
Thanks,
Yang
On 11/28/17 12:24 PM, Waiman Long wrote:
> On 11/28/2017 02:45 PM, Yang Shi wrote:
>> There are nested loops on debug objects free path, sometimes it may take
>> over hundred thousands of loops, then cause soft lockup with !CONFIG_PREEMPT
>> occasionally, like below:
>>
>> NMI watchdog: BUG: soft lockup - CPU#15 stuck for 22s! [stress-ng-getde:110342]
>> Modules linked in: binfmt_misc(E) tcp_diag(E)
>> inet_diag(E) bonding(E) intel_rapl(E) iosf_mbi(E)
>> x86_pkg_temp_thermal(E) coretemp(E) iTCO_wdt(E) iTCO_vendor_support(E)
>> kvm_intel(E) kvm(E) irqbypass(E) crct10dif_pclmul(E) crc32_pclmul(E)
>> dcdbas(E) ghash_clmulni_intel(E) aesni_intel(E) lrw(E) gf128mul(E)
>> glue_helper(E) ablk_helper(E) ipmi_devintf(E) sg(E) cryptd(E) pcspkr(E)
>> mei_me(E) lpc_ich(E) ipmi_si(E) mfd_core(E) mei(E) shpchp(E) wmi(E)
>> ipmi_msghandler(E) acpi_power_meter(E) nfsd(E) auth_rpcgss(E) nfs_acl(E)
>> lockd(E) grace(E) sunrpc(E) ip_tables(E) ext4(E) jbd2(E) mbcache(E)
>> sd_mod(E) mgag200(E) igb(E) drm_kms_helper(E) ixgbe(E) syscopyarea(E)
>> mdio(E) sysfillrect(E) sysimgblt(E) ptp(E) fb_sys_fops(E) pps_core(E)
>> ttm(E) drm(E) crc32c_intel(E) i2c_algo_bit(E) i2c_core(E)
>> megaraid_sas(E)
>> dca(E)
>> irq event stamp: 4340444
>> hardirqs last enabled at (4340443): [<ffffffff817d6476>]
>> _raw_spin_unlock_irqrestore+0x36/0x60
>> hardirqs last disabled at (4340444): [<ffffffff817d8c81>]
>> apic_timer_interrupt+0x91/0xa0
>> softirqs last enabled at (4340398): [<ffffffff817da179>]
>> __do_softirq+0x349/0x50e
>> softirqs last disabled at (4340391): [<ffffffff810a5255>]
>> irq_exit+0xf5/0x110
>> CPU: 15 PID: 110342 Comm: stress-ng-getde Tainted: G
>> E 4.9.44-003.ali3000.alios7.x86_64.debug #1
>> Hardware name: Dell Inc. PowerEdge R720xd/0X6FFV, BIOS
>> 1.6.0 03/07/2013
>> task: ffff884cbb0d0000 task.stack: ffff884cabc70000
>> RIP: 0010:[<ffffffff817d647b>] [<ffffffff817d647b>]
>> _raw_spin_unlock_irqrestore+0x3b/0x60
>> RSP: 0018:ffff884cabc77b78 EFLAGS: 00000292
>> RAX: ffff884cbb0d0000 RBX: 0000000000000292 RCX: 0000000000000000
>> RDX: ffff884cbb0d0000 RSI: 0000000000000001 RDI: 0000000000000292
>> RBP: ffff884cabc77b88 R08: 0000000000000000 R09: 0000000000000000
>> R10: 0000000000000001 R11: 0000000000000001 R12: ffffffff8357a0d8
>> R13: ffff884cabc77bc8 R14: ffffffff8357a0d0 R15: 00000000000000fc
>> FS: 00002aee845fd2c0(0000) GS:ffff8852bd400000(0000) knlGS:0000000000000000
>> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> CR2: 0000000002991808 CR3: 0000005123abf000 CR4: 00000000000406e0
>> Stack:
>> ffff884ff4fe0000 ffff884ff4fd8000 ffff884cabc77c00 ffffffff8141177e
>> 0000000000000202 ffff884cbb0d0000 ffff884cabc77bc8 0000000000000006
>> ffff884ff4fda000 ffffffff8357a0d8 0000000000000000 91f5d976f6020b6c
>> Call Trace:
>> [<ffffffff8141177e>] debug_check_no_obj_freed+0x13e/0x220
>> [<ffffffff811f8751>] __free_pages_ok+0x1f1/0x5c0
>> [<ffffffff811fa785>] __free_pages+0x25/0x40
>> [<ffffffff812638db>] __free_slab+0x19b/0x270
>> [<ffffffff812639e9>] discard_slab+0x39/0x50
>> [<ffffffff812679f7>] __slab_free+0x207/0x270
>> [<ffffffff81269966>] ___cache_free+0xa6/0xb0
>> [<ffffffff8126c267>] qlist_free_all+0x47/0x80
>> [<ffffffff8126c5a9>] quarantine_reduce+0x159/0x190
>> [<ffffffff8126b3bf>] kasan_kmalloc+0xaf/0xc0
>> [<ffffffff8126b8a2>] kasan_slab_alloc+0x12/0x20
>> [<ffffffff81265e8a>] kmem_cache_alloc+0xfa/0x360
>> [<ffffffff812abc8f>] ? getname_flags+0x4f/0x1f0
>> [<ffffffff812abc8f>] getname_flags+0x4f/0x1f0
>> [<ffffffff812abe42>] getname+0x12/0x20
>> [<ffffffff81298da9>] do_sys_open+0xf9/0x210
>> [<ffffffff81298ede>] SyS_open+0x1e/0x20
>> [<ffffffff817d6e01>] entry_SYSCALL_64_fastpath+0x1f/0xc2
>> Code: 7f 18 53 48 8b 55 08 48 89 f3 be 01 00 00 00 e8 3c
>> cd 92 ff 4c 89 e7 e8 f4 0e 93 ff f6 c7 02 74 1b e8 3a ac 92 ff 48 89 df
>> 57 9d <66> 66 90 66 90 65 ff 0d d1 ff 83 7e 5b 41 5c 5d c3 48 89 df 57
>>
>> The code path might be called in either atomic or non-atomic context,
>> so touching softlockup watchdog instead of calling cond_resched() which
>> might fall asleep. However, it is unnecessary to touch the watchdog
>> every loop, so just touch the watchdog at every 10000 (best estimate) loops.
>>
>> And, introduce a new knob: /sys/kernel/debug/debug_objects/suppress_lockup.
>> 0 value means not suppress the softlockup message by touching the
>> watchdog, non-zero value means suppress the softlockup message.
>> The default value is zero.
>>
>> Signed-off-by: Yang Shi <yang.s@alibaba-inc.com>
>> CC: Waiman Long <longman@redhat.com>
>> CC: Thomas Gleixner <tglx@linutronix.de>
>> ---
>> v2 --> v3:
>> * Use debugfs_create_u32() helper API per Waiman's suggestion
>> v1 --> v2:
>> * Added suppress_lockup knob in debugfs per Waiman's suggestion
>>
>> lib/debugobjects.c | 16 ++++++++++++++--
>> 1 file changed, 14 insertions(+), 2 deletions(-)
>>
>> diff --git a/lib/debugobjects.c b/lib/debugobjects.c
>> index 166488d..c960221 100644
>> --- a/lib/debugobjects.c
>> +++ b/lib/debugobjects.c
>> @@ -19,6 +19,7 @@
>> #include <linux/slab.h>
>> #include <linux/hash.h>
>> #include <linux/kmemleak.h>
>> +#include <linux/nmi.h>
>>
>> #define ODEBUG_HASH_BITS 14
>> #define ODEBUG_HASH_SIZE (1 << ODEBUG_HASH_BITS)
>> @@ -67,6 +68,8 @@ struct debug_bucket {
>> static int debug_objects_allocated;
>> static int debug_objects_freed;
>>
>> +static int suppress_lockup;
>> +
>> static void free_obj_work(struct work_struct *work);
>> static DECLARE_WORK(debug_obj_work, free_obj_work);
>>
>> @@ -768,6 +771,10 @@ static void __debug_check_no_obj_freed(const void *address, unsigned long size)
>> debug_objects_maxchain = cnt;
>>
>> max_loops += cnt;
>> +
>> + if (max_loops > 10000 && ((max_loops % 10000) == 0)
>> + && suppress_lockup != 0)
>> + touch_softlockup_watchdog();
>> }
>>
>> if (max_loops > debug_objects_maxloops)
>> @@ -812,7 +819,7 @@ static int debug_stats_open(struct inode *inode, struct file *filp)
>>
>> static int __init debug_objects_init_debugfs(void)
>> {
>> - struct dentry *dbgdir, *dbgstats;
>> + struct dentry *dbgdir, *dbgstats, *dbglockup;
>>
>> if (!debug_objects_enabled)
>> return 0;
>> @@ -826,10 +833,15 @@ static int __init debug_objects_init_debugfs(void)
>> if (!dbgstats)
>> goto err;
>>
>> + dbglockup = debugfs_create_u32("suppress_lockup", 0644, dbgdir,
>> + &suppress_lockup);
>> + if (!dbglockup)
>> + goto err;
>> +
>> return 0;
>>
>> err:
>> - debugfs_remove(dbgdir);
>> + debugfs_remove_recursive(dbgdir);
>>
>> return -ENOMEM;
>> }
>
> I am OK with your patch. Now it is up to Thomas to decide if he will
> pick up your patch.
>
> Cheers,
> Longman
>
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2017-12-07 19:16 ` Yang Shi
@ 2017-12-28 15:36 ` Thomas Gleixner
2018-01-03 23:54 ` Yang Shi
0 siblings, 1 reply; 12+ messages in thread
From: Thomas Gleixner @ 2017-12-28 15:36 UTC (permalink / raw)
To: Yang Shi; +Cc: Waiman Long, linux-kernel
On Fri, 8 Dec 2017, Yang Shi wrote:
> Hi Thomas,
>
> Waiman has reviewed the patches, any comment from you is appreciated.
I'm confused there is a V3 and a V8 of the 2/2 patch in my inbox. Can you
please resend both patches in their final version?
Thanks,
tglx
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2017-12-28 15:36 ` Thomas Gleixner
@ 2018-01-03 23:54 ` Yang Shi
0 siblings, 0 replies; 12+ messages in thread
From: Yang Shi @ 2018-01-03 23:54 UTC (permalink / raw)
To: Thomas Gleixner; +Cc: Waiman Long, linux-kernel
On 12/28/17 7:36 AM, Thomas Gleixner wrote:
> On Fri, 8 Dec 2017, Yang Shi wrote:
>> Hi Thomas,
>>
>> Waiman has reviewed the patches, any comment from you is appreciated.
>
> I'm confused there is a V3 and a V8 of the 2/2 patch in my inbox. Can you
> please resend both patches in their final version?
That "V8" is a typo, it should be "V2". I'm going to resend the whole
series soon.
Thanks,
Yang
>
> Thanks,
>
> tglx
>
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2018-01-18 9:02 ` Thomas Gleixner
2018-01-18 15:05 ` Peter Zijlstra
@ 2018-01-20 0:06 ` Yang Shi
1 sibling, 0 replies; 12+ messages in thread
From: Yang Shi @ 2018-01-20 0:06 UTC (permalink / raw)
To: Thomas Gleixner; +Cc: longman, LKML, Peter Zijlstra
On 1/18/18 1:02 AM, Thomas Gleixner wrote:
> On Thu, 18 Jan 2018, Yang Shi wrote:
>> On 1/17/18 4:21 AM, Thomas Gleixner wrote:
>>> There are two things which can be done here:
>>>
>>> 1) The collected objects can be put on a global free list and work
>>> scheduled to free them piecewise.
>>
>> I don't get your point here. objects free has already been done in a work.
>> free_object() -> schedule_work()
>
> But it's doing it in a loop for each freed object.
>
>> Do you mean free all of them out of the for loop in a batch? Then don't call
>> free_object() in the for loop?
>
> Somehing like this:
>
> diff --git a/lib/debugobjects.c b/lib/debugobjects.c
> index 2f5349c6e81a..d36940cdc658 100644
> --- a/lib/debugobjects.c
> +++ b/lib/debugobjects.c
> @@ -42,6 +42,7 @@ static struct debug_obj obj_static_pool[ODEBUG_POOL_SIZE] __initdata;
> static DEFINE_RAW_SPINLOCK(pool_lock);
>
> static HLIST_HEAD(obj_pool);
> +static HLIST_HEAD(obj_to_free);
>
> static int obj_pool_min_free = ODEBUG_POOL_SIZE;
> static int obj_pool_free = ODEBUG_POOL_SIZE;
> @@ -714,13 +715,12 @@ EXPORT_SYMBOL_GPL(debug_object_active_state);
> static void __debug_check_no_obj_freed(const void *address, unsigned long size)
> {
> unsigned long flags, oaddr, saddr, eaddr, paddr, chunks;
> - struct hlist_node *tmp;
> - HLIST_HEAD(freelist);
> struct debug_obj_descr *descr;
> enum debug_obj_state state;
> struct debug_bucket *db;
> + struct hlist_node *tmp;
> struct debug_obj *obj;
> - int cnt;
> + int cnt, freed = 0;
>
> saddr = (unsigned long) address;
> eaddr = saddr + size;
> @@ -751,21 +751,22 @@ static void __debug_check_no_obj_freed(const void *address, unsigned long size)
> goto repeat;
> default:
> hlist_del(&obj->node);
> - hlist_add_head(&obj->node, &freelist);
> + /* Put them on the freelist */
> + raw_spin_lock_irqsave(&pool_lock, flags);
> + hlist_add_head(&obj->node, &obj_to_free);
> + raw_spin_lock_irqrestore(&pool_lock, flags);
> + freed++;
> break;
> }
> }
> raw_spin_unlock_irqrestore(&db->lock, flags);
>
> - /* Now free them */
> - hlist_for_each_entry_safe(obj, tmp, &freelist, node) {
> - hlist_del(&obj->node);
> - free_object(obj);
> - }
> -
> if (cnt > debug_objects_maxchain)
> debug_objects_maxchain = cnt;
> }
> +
> + if (freed)
> + schedule_work(.....);
>
> The allocation side can look at the free list as well and grab objects from
> there if the pool level is low if that happens before the work can do that.
Thanks a lot for the example. I'm working on a preliminary patch, will
send out for review once I finish my tests.
>
>
>>>
>>> 2) We can do a cond_resched() if not in atomic context and interrupts are
>>> enabled.
>>
>> I did try this before I went with touching softlockup watchdog approach. The
>> problem is in_atomic() can't tell if it is in atomic context on non-preempt
>> kernel. For preempt kernel, it is easy.
>
> Peter, can we do anything about that?
As Peter confirmed, in_atomic() is not functional on !PREEMPT kernel,
shall we still use touching softlockup watchdog approach?
Thanks,
Yang
>
> Thanks,
>
> tglx
>
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2018-01-18 15:05 ` Peter Zijlstra
@ 2018-01-18 15:15 ` Thomas Gleixner
0 siblings, 0 replies; 12+ messages in thread
From: Thomas Gleixner @ 2018-01-18 15:15 UTC (permalink / raw)
To: Peter Zijlstra; +Cc: Yang Shi, longman, LKML
On Thu, 18 Jan 2018, Peter Zijlstra wrote:
> On Thu, Jan 18, 2018 at 10:02:25AM +0100, Thomas Gleixner wrote:
> > > > 2) We can do a cond_resched() if not in atomic context and interrupts are
> > > > enabled.
> > >
> > > I did try this before I went with touching softlockup watchdog approach. The
> > > problem is in_atomic() can't tell if it is in atomic context on non-preempt
> > > kernel. For preempt kernel, it is easy.
> >
> > Peter, can we do anything about that?
>
> Not really :/ Take for instance a spinlock on !PREEMPT, only the lock
> variable is touched, preempt_count isn't. So we cannot know.
Fair enough.
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2018-01-18 9:02 ` Thomas Gleixner
@ 2018-01-18 15:05 ` Peter Zijlstra
2018-01-18 15:15 ` Thomas Gleixner
2018-01-20 0:06 ` Yang Shi
1 sibling, 1 reply; 12+ messages in thread
From: Peter Zijlstra @ 2018-01-18 15:05 UTC (permalink / raw)
To: Thomas Gleixner; +Cc: Yang Shi, longman, LKML
On Thu, Jan 18, 2018 at 10:02:25AM +0100, Thomas Gleixner wrote:
> > > 2) We can do a cond_resched() if not in atomic context and interrupts are
> > > enabled.
> >
> > I did try this before I went with touching softlockup watchdog approach. The
> > problem is in_atomic() can't tell if it is in atomic context on non-preempt
> > kernel. For preempt kernel, it is easy.
>
> Peter, can we do anything about that?
Not really :/ Take for instance a spinlock on !PREEMPT, only the lock
variable is touched, preempt_count isn't. So we cannot know.
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2018-01-18 1:41 ` Yang Shi
@ 2018-01-18 9:02 ` Thomas Gleixner
2018-01-18 15:05 ` Peter Zijlstra
2018-01-20 0:06 ` Yang Shi
0 siblings, 2 replies; 12+ messages in thread
From: Thomas Gleixner @ 2018-01-18 9:02 UTC (permalink / raw)
To: Yang Shi; +Cc: longman, LKML, Peter Zijlstra
On Thu, 18 Jan 2018, Yang Shi wrote:
> On 1/17/18 4:21 AM, Thomas Gleixner wrote:
> > There are two things which can be done here:
> >
> > 1) The collected objects can be put on a global free list and work
> > scheduled to free them piecewise.
>
> I don't get your point here. objects free has already been done in a work.
> free_object() -> schedule_work()
But it's doing it in a loop for each freed object.
> Do you mean free all of them out of the for loop in a batch? Then don't call
> free_object() in the for loop?
Somehing like this:
diff --git a/lib/debugobjects.c b/lib/debugobjects.c
index 2f5349c6e81a..d36940cdc658 100644
--- a/lib/debugobjects.c
+++ b/lib/debugobjects.c
@@ -42,6 +42,7 @@ static struct debug_obj obj_static_pool[ODEBUG_POOL_SIZE] __initdata;
static DEFINE_RAW_SPINLOCK(pool_lock);
static HLIST_HEAD(obj_pool);
+static HLIST_HEAD(obj_to_free);
static int obj_pool_min_free = ODEBUG_POOL_SIZE;
static int obj_pool_free = ODEBUG_POOL_SIZE;
@@ -714,13 +715,12 @@ EXPORT_SYMBOL_GPL(debug_object_active_state);
static void __debug_check_no_obj_freed(const void *address, unsigned long size)
{
unsigned long flags, oaddr, saddr, eaddr, paddr, chunks;
- struct hlist_node *tmp;
- HLIST_HEAD(freelist);
struct debug_obj_descr *descr;
enum debug_obj_state state;
struct debug_bucket *db;
+ struct hlist_node *tmp;
struct debug_obj *obj;
- int cnt;
+ int cnt, freed = 0;
saddr = (unsigned long) address;
eaddr = saddr + size;
@@ -751,21 +751,22 @@ static void __debug_check_no_obj_freed(const void *address, unsigned long size)
goto repeat;
default:
hlist_del(&obj->node);
- hlist_add_head(&obj->node, &freelist);
+ /* Put them on the freelist */
+ raw_spin_lock_irqsave(&pool_lock, flags);
+ hlist_add_head(&obj->node, &obj_to_free);
+ raw_spin_lock_irqrestore(&pool_lock, flags);
+ freed++;
break;
}
}
raw_spin_unlock_irqrestore(&db->lock, flags);
- /* Now free them */
- hlist_for_each_entry_safe(obj, tmp, &freelist, node) {
- hlist_del(&obj->node);
- free_object(obj);
- }
-
if (cnt > debug_objects_maxchain)
debug_objects_maxchain = cnt;
}
+
+ if (freed)
+ schedule_work(.....);
The allocation side can look at the free list as well and grab objects from
there if the pool level is low if that happens before the work can do that.
> >
> > 2) We can do a cond_resched() if not in atomic context and interrupts are
> > enabled.
>
> I did try this before I went with touching softlockup watchdog approach. The
> problem is in_atomic() can't tell if it is in atomic context on non-preempt
> kernel. For preempt kernel, it is easy.
Peter, can we do anything about that?
Thanks,
tglx
^ permalink raw reply related [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2018-01-17 12:21 ` Thomas Gleixner
@ 2018-01-18 1:41 ` Yang Shi
2018-01-18 9:02 ` Thomas Gleixner
0 siblings, 1 reply; 12+ messages in thread
From: Yang Shi @ 2018-01-18 1:41 UTC (permalink / raw)
To: Thomas Gleixner; +Cc: longman, linux-kernel
On 1/17/18 4:21 AM, Thomas Gleixner wrote:
> On Thu, 4 Jan 2018, Yang Shi wrote:
>
>> There are nested loops on debug objects free path, sometimes it may take
>> over hundred thousands of loops, then cause soft lockup with !CONFIG_PREEMPT
>> occasionally, like below:
>
> Please trim back traces. The whole module info and whatever is completely
> irrelevant.
Yes, sure.
>
>> @@ -768,6 +771,10 @@ static void __debug_check_no_obj_freed(const void *address, unsigned long size)
>> debug_objects_maxchain = cnt;
>>
>> max_loops += cnt;
>> +
>> + if (max_loops > 10000 && ((max_loops % 10000) == 0)
>> + && suppress_lockup != 0)
>> + touch_softlockup_watchdog();
>
> This is voodoo programming.
>
> There are two things which can be done here:
>
> 1) The collected objects can be put on a global free list and work
> scheduled to free them piecewise.
I don't get your point here. objects free has already been done in a
work. free_object() -> schedule_work()
Do you mean free all of them out of the for loop in a batch? Then don't
call free_object() in the for loop?
>
> 2) We can do a cond_resched() if not in atomic context and interrupts are
> enabled.
I did try this before I went with touching softlockup watchdog approach.
The problem is in_atomic() can't tell if it is in atomic context on
non-preempt kernel. For preempt kernel, it is easy.
Thanks,
Yang
>
> I rather prefer to make that stuff robust than having crystal ball
> constants and magic debugfs knobs.
>
> Thanks,
>
> tglx
>
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2018-01-03 23:56 ` [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT Yang Shi
@ 2018-01-17 12:21 ` Thomas Gleixner
2018-01-18 1:41 ` Yang Shi
0 siblings, 1 reply; 12+ messages in thread
From: Thomas Gleixner @ 2018-01-17 12:21 UTC (permalink / raw)
To: Yang Shi; +Cc: longman, linux-kernel
On Thu, 4 Jan 2018, Yang Shi wrote:
> There are nested loops on debug objects free path, sometimes it may take
> over hundred thousands of loops, then cause soft lockup with !CONFIG_PREEMPT
> occasionally, like below:
Please trim back traces. The whole module info and whatever is completely
irrelevant.
> @@ -768,6 +771,10 @@ static void __debug_check_no_obj_freed(const void *address, unsigned long size)
> debug_objects_maxchain = cnt;
>
> max_loops += cnt;
> +
> + if (max_loops > 10000 && ((max_loops % 10000) == 0)
> + && suppress_lockup != 0)
> + touch_softlockup_watchdog();
This is voodoo programming.
There are two things which can be done here:
1) The collected objects can be put on a global free list and work
scheduled to free them piecewise.
2) We can do a cond_resched() if not in atomic context and interrupts are
enabled.
I rather prefer to make that stuff robust than having crystal ball
constants and magic debugfs knobs.
Thanks,
tglx
^ permalink raw reply [flat|nested] 12+ messages in thread
* [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT
2018-01-03 23:56 [PATCH 1/2 v3] lib: debugobjects: export max loops counter Yang Shi
@ 2018-01-03 23:56 ` Yang Shi
2018-01-17 12:21 ` Thomas Gleixner
0 siblings, 1 reply; 12+ messages in thread
From: Yang Shi @ 2018-01-03 23:56 UTC (permalink / raw)
To: longman, tglx; +Cc: Yang Shi, linux-kernel
There are nested loops on debug objects free path, sometimes it may take
over hundred thousands of loops, then cause soft lockup with !CONFIG_PREEMPT
occasionally, like below:
NMI watchdog: BUG: soft lockup - CPU#15 stuck for 22s! [stress-ng-getde:110342]
Modules linked in: binfmt_misc(E) tcp_diag(E)
inet_diag(E) bonding(E) intel_rapl(E) iosf_mbi(E)
x86_pkg_temp_thermal(E) coretemp(E) iTCO_wdt(E) iTCO_vendor_support(E)
kvm_intel(E) kvm(E) irqbypass(E) crct10dif_pclmul(E) crc32_pclmul(E)
dcdbas(E) ghash_clmulni_intel(E) aesni_intel(E) lrw(E) gf128mul(E)
glue_helper(E) ablk_helper(E) ipmi_devintf(E) sg(E) cryptd(E) pcspkr(E)
mei_me(E) lpc_ich(E) ipmi_si(E) mfd_core(E) mei(E) shpchp(E) wmi(E)
ipmi_msghandler(E) acpi_power_meter(E) nfsd(E) auth_rpcgss(E) nfs_acl(E)
lockd(E) grace(E) sunrpc(E) ip_tables(E) ext4(E) jbd2(E) mbcache(E)
sd_mod(E) mgag200(E) igb(E) drm_kms_helper(E) ixgbe(E) syscopyarea(E)
mdio(E) sysfillrect(E) sysimgblt(E) ptp(E) fb_sys_fops(E) pps_core(E)
ttm(E) drm(E) crc32c_intel(E) i2c_algo_bit(E) i2c_core(E)
megaraid_sas(E)
dca(E)
irq event stamp: 4340444
hardirqs last enabled at (4340443): [<ffffffff817d6476>]
_raw_spin_unlock_irqrestore+0x36/0x60
hardirqs last disabled at (4340444): [<ffffffff817d8c81>]
apic_timer_interrupt+0x91/0xa0
softirqs last enabled at (4340398): [<ffffffff817da179>]
__do_softirq+0x349/0x50e
softirqs last disabled at (4340391): [<ffffffff810a5255>]
irq_exit+0xf5/0x110
CPU: 15 PID: 110342 Comm: stress-ng-getde Tainted: G
E 4.9.44-003.ali3000.alios7.x86_64.debug #1
Hardware name: Dell Inc. PowerEdge R720xd/0X6FFV, BIOS
1.6.0 03/07/2013
task: ffff884cbb0d0000 task.stack: ffff884cabc70000
RIP: 0010:[<ffffffff817d647b>] [<ffffffff817d647b>]
_raw_spin_unlock_irqrestore+0x3b/0x60
RSP: 0018:ffff884cabc77b78 EFLAGS: 00000292
RAX: ffff884cbb0d0000 RBX: 0000000000000292 RCX: 0000000000000000
RDX: ffff884cbb0d0000 RSI: 0000000000000001 RDI: 0000000000000292
RBP: ffff884cabc77b88 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000001 R11: 0000000000000001 R12: ffffffff8357a0d8
R13: ffff884cabc77bc8 R14: ffffffff8357a0d0 R15: 00000000000000fc
FS: 00002aee845fd2c0(0000) GS:ffff8852bd400000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000002991808 CR3: 0000005123abf000 CR4: 00000000000406e0
Stack:
ffff884ff4fe0000 ffff884ff4fd8000 ffff884cabc77c00 ffffffff8141177e
0000000000000202 ffff884cbb0d0000 ffff884cabc77bc8 0000000000000006
ffff884ff4fda000 ffffffff8357a0d8 0000000000000000 91f5d976f6020b6c
Call Trace:
[<ffffffff8141177e>] debug_check_no_obj_freed+0x13e/0x220
[<ffffffff811f8751>] __free_pages_ok+0x1f1/0x5c0
[<ffffffff811fa785>] __free_pages+0x25/0x40
[<ffffffff812638db>] __free_slab+0x19b/0x270
[<ffffffff812639e9>] discard_slab+0x39/0x50
[<ffffffff812679f7>] __slab_free+0x207/0x270
[<ffffffff81269966>] ___cache_free+0xa6/0xb0
[<ffffffff8126c267>] qlist_free_all+0x47/0x80
[<ffffffff8126c5a9>] quarantine_reduce+0x159/0x190
[<ffffffff8126b3bf>] kasan_kmalloc+0xaf/0xc0
[<ffffffff8126b8a2>] kasan_slab_alloc+0x12/0x20
[<ffffffff81265e8a>] kmem_cache_alloc+0xfa/0x360
[<ffffffff812abc8f>] ? getname_flags+0x4f/0x1f0
[<ffffffff812abc8f>] getname_flags+0x4f/0x1f0
[<ffffffff812abe42>] getname+0x12/0x20
[<ffffffff81298da9>] do_sys_open+0xf9/0x210
[<ffffffff81298ede>] SyS_open+0x1e/0x20
[<ffffffff817d6e01>] entry_SYSCALL_64_fastpath+0x1f/0xc2
Code: 7f 18 53 48 8b 55 08 48 89 f3 be 01 00 00 00 e8 3c
cd 92 ff 4c 89 e7 e8 f4 0e 93 ff f6 c7 02 74 1b e8 3a ac 92 ff 48 89 df
57 9d <66> 66 90 66 90 65 ff 0d d1 ff 83 7e 5b 41 5c 5d c3 48 89 df 57
The code path might be called in either atomic or non-atomic context,
so touching softlockup watchdog instead of calling cond_resched() which
might fall asleep. However, it is unnecessary to touch the watchdog
every loop, so just touch the watchdog at every 10000 (best estimate) loops.
And, introduce a new knob: /sys/kernel/debug/debug_objects/suppress_lockup.
0 value means not suppress the softlockup message by touching the
watchdog, non-zero value means suppress the softlockup message.
The default value is zero.
Signed-off-by: Yang Shi <yang.s@alibaba-inc.com>
CC: Waiman Long <longman@redhat.com>
CC: Thomas Gleixner <tglx@linutronix.de>
---
v2 --> v3:
* Use debugfs_create_u32() helper API per Waiman's suggestion
v1 --> v2:
* Added suppress_lockup knob in debugfs per Waiman's suggestion
lib/debugobjects.c | 16 ++++++++++++++--
1 file changed, 14 insertions(+), 2 deletions(-)
diff --git a/lib/debugobjects.c b/lib/debugobjects.c
index 166488d..c960221 100644
--- a/lib/debugobjects.c
+++ b/lib/debugobjects.c
@@ -19,6 +19,7 @@
#include <linux/slab.h>
#include <linux/hash.h>
#include <linux/kmemleak.h>
+#include <linux/nmi.h>
#define ODEBUG_HASH_BITS 14
#define ODEBUG_HASH_SIZE (1 << ODEBUG_HASH_BITS)
@@ -67,6 +68,8 @@ struct debug_bucket {
static int debug_objects_allocated;
static int debug_objects_freed;
+static int suppress_lockup;
+
static void free_obj_work(struct work_struct *work);
static DECLARE_WORK(debug_obj_work, free_obj_work);
@@ -768,6 +771,10 @@ static void __debug_check_no_obj_freed(const void *address, unsigned long size)
debug_objects_maxchain = cnt;
max_loops += cnt;
+
+ if (max_loops > 10000 && ((max_loops % 10000) == 0)
+ && suppress_lockup != 0)
+ touch_softlockup_watchdog();
}
if (max_loops > debug_objects_maxloops)
@@ -812,7 +819,7 @@ static int debug_stats_open(struct inode *inode, struct file *filp)
static int __init debug_objects_init_debugfs(void)
{
- struct dentry *dbgdir, *dbgstats;
+ struct dentry *dbgdir, *dbgstats, *dbglockup;
if (!debug_objects_enabled)
return 0;
@@ -826,10 +833,15 @@ static int __init debug_objects_init_debugfs(void)
if (!dbgstats)
goto err;
+ dbglockup = debugfs_create_u32("suppress_lockup", 0644, dbgdir,
+ &suppress_lockup);
+ if (!dbglockup)
+ goto err;
+
return 0;
err:
- debugfs_remove(dbgdir);
+ debugfs_remove_recursive(dbgdir);
return -ENOMEM;
}
--
1.8.3.1
^ permalink raw reply related [flat|nested] 12+ messages in thread
end of thread, other threads:[~2018-01-20 0:07 UTC | newest]
Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-11-28 19:45 [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT Yang Shi
2017-11-28 20:24 ` Waiman Long
2017-12-07 19:16 ` Yang Shi
2017-12-28 15:36 ` Thomas Gleixner
2018-01-03 23:54 ` Yang Shi
2018-01-03 23:56 [PATCH 1/2 v3] lib: debugobjects: export max loops counter Yang Shi
2018-01-03 23:56 ` [PATCH 2/2 v3] lib: debugobjects: touch watchdog to avoid softlockup when !CONFIG_PREEMPT Yang Shi
2018-01-17 12:21 ` Thomas Gleixner
2018-01-18 1:41 ` Yang Shi
2018-01-18 9:02 ` Thomas Gleixner
2018-01-18 15:05 ` Peter Zijlstra
2018-01-18 15:15 ` Thomas Gleixner
2018-01-20 0:06 ` Yang Shi
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).