workqueue, writeback: better worker information in task dumps
Subsystems (4)
Tejun Heo
maintainerPATCH 1/1 workqueue, writeback: better worker information in task dumps
Hello,
One of the problems that arise when converting dedicated custom threadpool to workqueue is that the shared worker pool used by workqueue anonimizes each worker making it more difficult to identify what the worker was doing on which target from the output of sysrq-t or debug dump from oops, BUG() and friends.
For example, after writeback is converted to use workqueue instead of priviate thread pool, there's no easy to tell which backing device a writeback work item was working on at the time of task dump, which, according to our writeback brethren, is important in tracking down issues with a lot of mounted file systems on a lot of different devices.
This patchset implements a way for a work function to mark its execution instance so that task dump of the worker task includes information to indicate what the work item was doing.
An example WARN dump would look like the following.
WARNING: at /work/os/work/fs/fs-writeback.c:1015 bdi_writeback_workfn+0x2b4/0x3c0() Modules linked in: Pid: 28, comm: kworker/u18:0 Not tainted 3.9.0-rc1-work+ #24 empty empty/S3992 Workqueue: writeback bdi_writeback_workfn (flush-8:16) ffffffff820a3a98 ffff88015b927cb8 ffffffff81c61855 ffff88015b927cf8 ffffffff8108f500 0000000000000000 ffff88007a171948 ffff88007a1716b0 ffff88015b49df00 ffff88015b8d3940 0000000000000000 ffff88015b927d08 Call Trace: [<ffffffff81c61855>] dump_stack+0x19/0x1b [<ffffffff8108f500>] warn_slowpath_common+0x70/0xa0 ...
This patchset contains the following three patches.
0001-kthread-implement-probe_kthread_data.patch 0002-workqueue-include-workqueue-info-when-printing-debug.patch 0003-writeback-set-worker-desc-to-identify-writeback-work.patch
0001 adds a speculative accessor for kthread_data.
0002 implements worker description.
0003 updates writeback to set its flusher name as worker description.
This patchset is on top of
[1] workqueue: NUMA affinity for unbound workqueues
+ [2] writeback: convert writeback to unbound workqueue
+ [3] arch: unify task dump debug info
and available in the following git branch.
git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq.git review-better-dbg
Due to the dependencies, routing this set would be a bit complicated. I think it would be best to first pull wq/for-3.10 to block tree and apply workqueue conversion patchset. The patches in this series then can go through -mm on top of all three pieces - workqueue, block and archs.
diffstat follows. Thanks.
fs/fs-writeback.c | 1 include/linux/kthread.h | 1 include/linux/workqueue.h | 5 ++ kernel/kthread.c | 19 ++++++++++ kernel/sched/core.c | 1 kernel/workqueue.c | 79 ++++++++++++++++++++++++++++++++++++++++++++ kernel/workqueue_internal.h | 12 ++++++ lib/dump_stack.c | 2 + 8 files changed, 119 insertions(+), 1 deletion(-)
--
tejun
[1] http://thread.gmane.org/gmane.linux.kernel/1465626
[2] http://thread.gmane.org/gmane.linux.kernel/1453006
[3] https://lkml.org/lkml/2013/3/29/354
Tejun Heo
maintainerPATCH 1/3 kthread: implement probe_kthread_data()
Implement probe_kthread_data() which returns kthread_data if accessible. The function is equivalent to kthread_data() except that the specified @task may not be a kthread or its vfork_done is already cleared rendering struct kthread inaccessible. In the former case, probe_kthread_data() may return any value. In the latter, NULL.
This will be used to safely print debug information without affecting synchronization in the normal paths. Workqueue debug info printing on dump_stack() and friends will make use of it.
--- include/linux/kthread.h | 1 + kernel/kthread.c | 19 +++++++++++++++++++ 2 files changed, 20 insertions(+)
include/linux/kthread.h +1 -0
@@ -43,6 +43,7 @@ bool kthread_should_stop(void);bool kthread_should_park (void );bool kthread_freezable_should_stop (bool *was_frozen);void *kthread_data (struct task_struct *k); +void *probe_kthread_data (struct task_struct *k);int kthread_park (struct task_struct *k);void kthread_unpark (struct task_struct *k);void kthread_parkme (void );
kernel/kthread.c +19 -0
@@ -17,6 +17,7 @@#include <linux/slab.h> #include <linux/freezer.h> #include <linux/ptrace.h> +#include <linux/uaccess.h> #include <trace/events/sched.h> static DEFINE_SPINLOCK (kthread_create_lock);
@@ -122,6 +123,24 @@ void *kthread_data(struct task_struct *task)return to_kthread (task)->data; } +/** +* probe_kthread_data - speculatively version of kthread_data() +* @task: possible kthread task in question +* +* @task could be a kthread task. Return the data value specified when it +* was created if accessible. If @task isn't a kthread task or its data is +* inaccessible for any reason, %NULL is returned. This function requires +* that @task itself is safe to dereference. +*/ +void *probe_kthread_data (struct task_struct *task) +{ +struct kthread *kthread =to_kthread (task); +void *data =NULL ; + +probe_kernel_read (&data, &kthread->data,sizeof (data)); +return data; +} +static void __kthread_parkme (struct kthread *self) {__set_current_state (TASK_INTERRUPTIBLE );
Signature · 1 line
1.8.1.4
Re: [PATCH 1/3] kthread: implement probe_kthread_data()
Quoted 4 lines · Tejun Heo
Implement
probe_kthread_data()which returnskthread_dataif accessible. The function is equivalent tokthread_data()except that the specified @task may not be a kthread
^^^^^^^^^^^^^^^^^^^^^^^^^
This doesn't necessarily mean its ->vfork_done != NULL... but I guess the "false positive" is fine for your purpose.
Quoted 11 lines
or its
vfork_doneis already
cleared rendering struct kthread inaccessible.
...kernel/kthread.c +7
-0+/** +* probe_kthread_data - speculatively version of kthread_data() +* @task: possible kthread task in question +* +* @task could be a kthread task. Return the data value specified when it +* was created if accessible. If @task isn't a kthread task or its data is +* inaccessible for any reason, %NULL is returned. This function requires
kernel/kthread.c void *kthread_data(struct task_struct *task) · line 142
Quoted 10 lines
kernel/kthread.c +10
-0+* that @task itself is safe to dereference. +*/ +void *probe_kthread_data (struct task_struct *task) +{ +struct kthread *kthread =to_kthread (task); +void *data =NULL ; + +probe_kernel_read (&data, &kthread->data,sizeof (data)); +return data; +}
OK, but we can simply check ->vfork_done != NULL ?
We do not care if we race with the exiting thread which can clear its ->vfork_done. If it is safe to dereference this @task then kthread is also safe.
kthread-introduce-to_live_kthread.patch in -mm adds the trivial helper,
static inline struct kthread *to_kthread(struct task_struct *k)
{
return __to_kthread(k->vfork_done);
}
static struct kthread *to_live_kthread(struct task_struct *k)
{
struct completion *vfork = ACCESS_ONCE(k->vfork_done);
if (likely(vfork))
return __to_kthread(vfork);
return NULL;
}So, perhaps,
voif *kthread_data_safe(struct task_struct *task)
{
struct kthread *kthread = to_live_kthread(task);
return kthread ? kthread->data : NULL;
}?
Oleg.
Tejun Heo
maintainerRe: [PATCH 1/3] kthread: implement probe_kthread_data()
Hello, Oleg.
Quoted 10 lines · Oleg Nesterov
kernel/kthread.c void *kthread_data(struct task_struct *task) · line 142
Quoted 8 lines
kernel/kthread.c +8
-0+void *probe_kthread_data (struct task_struct *task) +{ +struct kthread *kthread =to_kthread (task); +void *data =NULL ; + +probe_kernel_read (&data, &kthread->data,sizeof (data)); +return data; +}OK, but we can simply check ->vfork_done != NULL ?
Hmm... what if ->vfork_done is pointing to some weird place?
Thanks.
--
tejun
Re: [PATCH 1/3] kthread: implement probe_kthread_data()
Hi Tejun,
Quoted 13 lines · Tejun Heo
Quoted 10 lines · Oleg Nesterov
kernel/kthread.c void *kthread_data(struct task_struct *task) · line 142
Quoted 8 lines
kernel/kthread.c +8
-0+void *probe_kthread_data (struct task_struct *task) +{ +struct kthread *kthread =to_kthread (task); +void *data =NULL ; + +probe_kernel_read (&data, &kthread->data,sizeof (data)); +return data; +}OK, but we can simply check ->vfork_done != NULL ?
Quoted 1 line
Hmm... what if ->vfork_done is pointing to some weird place?
Aah... "weird place" is not possible if we know that @task is kthread, it is either NULL or it points into tsk->stack which can only go away along with task. Note that kthread_stop() already relies on this.
But I guess I missed the fact that this helper should be safe even this @task can be the a vfork'ed user-space process, yes?
In this case, yes, ->vfork_done can point to task->parent->stack, not good...
Perhaps,
voif *kthread_data_safe(struct task_struct *task)
{
if (task->parent == kthreadd_task) {
struct kthread *kthread = to_live_kthread(task);
if (kthread)
return kthread->data;
}
return NULL;
}?
Or we can add to_live_kthread_safe() which checks "parent == kthreadd_task" instead.
Oleg.
A message that is not in the archive (sent off-list or not yet received): CAOS58YNMbTY2fZzUXeyLwuA+nowW2AhgLvQwKf_0jWXDen6HzQ@mail.gmail.com
Re: [PATCH 1/3] kthread: implement probe_kthread_data()
Quoted 8 lines · Tejun Heo
Quoted 3 lines · Oleg Nesterov
But I guess I missed the fact that this helper should be safe even this @task can be the a vfork'ed user-space process, yes?
Yeap, we really don't know what to expect.
OK,
Quoted 5 lines
Quoted 2 lines
Or we can add
to_live_kthread_safe()which checks "parent ==kthreadd_task" instead.But that would be more fragile than
probe_read. I mean, this thing will be called on BUGs and oopses.
OK, I agree.
Sorry for noise and thanks for your explanations, I really missed the intent.
Oleg.
Jan Kara
reviewerRe: [PATCH 1/3] kthread: implement probe_kthread_data()
In reply to Tejun HeoQuoted 9 lines · Tejun Heo
Implement
probe_kthread_data()which returnskthread_dataif accessible. The function is equivalent tokthread_data()except that the specified @task may not be a kthread or itsvfork_doneis already cleared rendering struct kthread inaccessible. In the former case,probe_kthread_data()may return any value. In the latter, NULL.This will be used to safely print debug information without affecting synchronization in the normal paths. Workqueue debug info printing on
dump_stack()and friends will make use of it.
Thanks for looking into this. Just one typo correction below:
kernel/kthread.c void *kthread_data(struct task_struct *task) · line 127
Quoted 6 lines
kernel/kthread.c +2
-0@@ -122,6 +123,24 @@ void *kthread_data(struct task_struct *task)return to_kthread (task)->data; } +/** +* probe_kthread_data - speculatively version of kthread_data()
^^^^ speculative
Honza
Signature · 2 lines
Jan Kara <[email protected]> SUSE Labs, CR
Tejun Heo
maintainerPATCH 2/3 workqueue: include workqueue info when printing debug dump of a worker task
In reply to Tejun HeoOne of the problems that arise when converting dedicated custom threadpool to workqueue is that the shared worker pool used by workqueue anonimizes each worker making it more difficult to identify what the worker was doing on which target from the output of sysrq-t or debug dump from oops, BUG() and friends.
This patch implements set_worker_desc() which can be called from any workqueue work function to set its description. When the worker task is dumped for whatever reason - sysrq-t, WARN, BUG, oops, lockdep assertion and so on - the description will be printed out together with the workqueue name and the worker function pointer.
The printing side is implemented by print_worker_info() which is called from functions in task dump paths - sched_show_task() and dump_stack_print_info(). print_worker_info() can be safely called on any task in any state as long as the task struct itself is accessible. It uses probe_*() functions to access worker fields. It may print garbage if something went very wrong, but it wouldn't cause (another) oops.
The description is currently limited to 24bytes including the terminating \0. worker->desc_valid and workder->desc[] are added and the 64 bytes marker which was already incorrect before adding the new fields is moved to the correct position.
Here's an example dump with writeback updated to set the bdi name as worker desc.
Hardware name: Bochs Modules linked in: Pid: 7, comm: kworker/u9:0 Not tainted 3.9.0-rc1-work+ #1 Workqueue: writeback bdi_writeback_workfn (flush-8:0) ffffffff820a3ab0 ffff88000f6e9cb8 ffffffff81c61845 ffff88000f6e9cf8 ffffffff8108f50f 0000000000000000 0000000000000000 ffff88000cde16b0 ffff88000cde1aa8 ffff88001ee19240 ffff88000f6e9fd8 ffff88000f6e9d08 Call Trace: [<ffffffff81c61845>] dump_stack+0x19/0x1b [<ffffffff8108f50f>] warn_slowpath_common+0x7f/0xc0 [<ffffffff8108f56a>] warn_slowpath_null+0x1a/0x20 [<ffffffff81200150>] bdi_writeback_workfn+0x2a0/0x3b0 ...
--- include/linux/workqueue.h | 5 +++ kernel/sched/core.c | 1 + kernel/workqueue.c | 79 +++++++++++++++++++++++++++++++++++++++++++++ kernel/workqueue_internal.h | 12 ++++++- lib/dump_stack.c | 2 ++ 5 files changed, 98 insertions(+), 1 deletion(-)
include/linux/workqueue.h +5 -0
@@ -92,6 +92,9 @@ enum {/* bit mask for work_busy() return values */ WORK_BUSY_PENDING =1 <<0 ,WORK_BUSY_RUNNING =1 <<1 , + +/* maximum string length for set_worker_desc() */ + WORKER_DESC_LEN = 24, }; struct work_struct {
@@ -447,6 +450,8 @@ extern void workqueue_set_max_active(struct workqueue_struct *wq,extern bool current_is_workqueue_rescuer (void );extern bool workqueue_congested (int cpu,struct workqueue_struct *wq);extern unsigned int work_busy (struct work_struct *work); +extern __printf(1, 2) void set_worker_desc(const char *fmt, ...); +extern void print_worker_info(const char *log_lvl, struct task_struct *task); /** * queue_work - queue work on a workqueue
kernel/sched/core.c +1 -0
@@ -4630,6 +4630,7 @@ void sched_show_task(struct task_struct *p)task_pid_nr (p), ppid, (unsigned long)task_thread_info(p)->flags); +print_worker_info (KERN_INFO , p);show_stack (p,NULL ); }
kernel/workqueue.c +79 -0
97 lines not on this page · raw message
kernel/workqueue_internal.h +11 -1
@@ -29,15 +29,25 @@ struct worker {struct work_struct *current_work;/* L: work being processed */ work_func_t current_func;/* L: current_work's fn */ struct pool_workqueue *current_pwq;/* L: current_work's pwq */ +bool desc_valid;/* ->desc is valid */ struct list_head scheduled;/* L: scheduled works */ + +/* 64 bytes boundary on 64bit, 32 on 32bit */ +struct task_struct *task;/* I: worker task */ struct worker_pool *pool;/* I: the associated pool */ /* L: for rescuers */ -/* 64 bytes boundary on 64bit, 32 on 32bit */ +unsigned long last_active;/* L: last active timestamp */ unsigned int flags;/* X: flags */ int id;/* I: worker id */ +/* +* Opaque string set with work_set_desc(). Printed out with task +* dump for debugging - WARN, BUG, panic or sysrq. +*/ +char desc[WORKER_DESC_LEN ]; +/* used only by rescuers to point to the target workqueue */ struct workqueue_struct *rescue_wq;/* I: the workqueue to rescue */ };
lib/dump_stack.c +2 -0
@@ -44,6 +44,8 @@ void dump_stack_print_info(const char *log_lvl)init_utsname ()->release, (int )strcspn (init_utsname ()->version," " ), init_utsname()->version, dump_stack_arch_desc_str); + +print_worker_info (log_lvl, current); }/**
Signature · 1 line
1.8.1.4
Tejun Heo
maintainerPATCH 3/3 writeback: set worker desc to identify writeback workers in task dumps
In reply to Tejun HeoWriteback has been recently converted to use workqueue instead of its private thread pool implementation. One negative side effect of this conversion is that there's no easy to tell which backing device a writeback work item was working on at the time of task dump, be it sysrq-t, BUG, WARN or whatever, which, according to our writeback brethren, is important in tracking down issues with a lot of mounted file systems on a lot of different devices.
This patch restores that information using the new worker description facility. bdi_writeback_workfn() calls set_work_desc() to identify which bdi it's working on. The description is printed out together with the worqueue name and worker function as in the following example dump.
WARNING: at /work/os/work/fs/fs-writeback.c:1015 bdi_writeback_workfn+0x2b4/0x3c0() Modules linked in: Pid: 28, comm: kworker/u18:0 Not tainted 3.9.0-rc1-work+ #24 empty empty/S3992 Workqueue: writeback bdi_writeback_workfn (flush-8:16) ffffffff820a3a98 ffff88015b927cb8 ffffffff81c61855 ffff88015b927cf8 ffffffff8108f500 0000000000000000 ffff88007a171948 ffff88007a1716b0 ffff88015b49df00 ffff88015b8d3940 0000000000000000 ffff88015b927d08 Call Trace: [<ffffffff81c61855>] dump_stack+0x19/0x1b [<ffffffff8108f500>] warn_slowpath_common+0x70/0xa0 [<ffffffff8108f54a>] warn_slowpath_null+0x1a/0x20 [<ffffffff81200144>] bdi_writeback_workfn+0x2b4/0x3c0 [<ffffffff810b4c87>] process_one_work+0x1d7/0x660 [<ffffffff810b5c72>] worker_thread+0x122/0x380 [<ffffffff810bdfea>] kthread+0xea/0xf0 [<ffffffff81c6cedc>] ret_from_fork+0x7c/0xb0
--- fs/fs-writeback.c | 1 + 1 file changed, 1 insertion(+)
fs/fs-writeback.c +1 -0
@@ -1009,6 +1009,7 @@ void bdi_writeback_workfn(struct work_struct *work)struct backing_dev_info *bdi = wb->bdi;long pages_written; +set_worker_desc ("flush-%s" ,dev_name (bdi->dev)); current->flags |=PF_SWAPWRITE ; if (likely(!current_is_workqueue_rescuer() ||
Signature · 1 line
1.8.1.4