1. Patchew
  2. linux
  3. [PATCH 0/5] Improve WARN_ON_ONCE() output by adding the condition string
  4. View patch

[PATCH 5/5] bugs/core: Do not print CPU and PID values in__warn() output

Ingo Molnar posted 5 patches 8 months, 4 weeks ago
There is a newer version of this series
[PATCH 5/5] bugs/core: Do not print CPU and PID values in__warn() output
Posted by Ingo Molnar 8 months, 4 weeks ago 
For most warnings it's repeated anyway a few lines later, as
part of the register dump:

  WARNING: CPU: 0 PID: 0 at kernel/sched/core.c:8511 sched_init+0x20/0x410
           ^^^^^^^^^^^^^
  Modules linked in:
  CPU: 0 UID: 0 PID: 0 Comm: swapper Not tainted 6.14.0-01616-g94d7af2844aa #4 PREEMPT(undef)
  ^^^^^^^^^^^^^^^^^^

In fact the later dump is richer, as it includes the 'comm' line as well.

As a side effect, this makes some space for the 'file' field to be augmented
with extra information (the condition string in particular).

Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 kernel/panic.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/kernel/panic.c b/kernel/panic.c
index d8635d5cecb2..3101c21ca39f 100644
--- a/kernel/panic.c
+++ b/kernel/panic.c
@@ -725,12 +725,9 @@ void __warn(const char *file, int line, void *caller, unsigned taint,
 	disable_trace_on_warning();
 
 	if (file)
-		pr_warn("WARNING: CPU: %d PID: %d at %s:%d %pS\n",
-			raw_smp_processor_id(), current->pid, file, line,
-			caller);
+		pr_warn("WARNING: %s:%d at %pS\n", file, line, caller);
 	else
-		pr_warn("WARNING: CPU: %d PID: %d at %pS\n",
-			raw_smp_processor_id(), current->pid, caller);
+		pr_warn("WARNING: at %pS\n", caller);
 
 #pragma GCC diagnostic push
 #ifndef __clang__
-- 
2.45.2
Re: [PATCH 5/5] bugs/core: Do not print CPU and PID values in__warn() output
Posted by Peter Zijlstra 8 months, 4 weeks ago 
On Wed, Mar 26, 2025 at 09:47:50AM +0100, Ingo Molnar wrote:
> For most warnings it's repeated anyway a few lines later, as
> part of the register dump:
> 
>   WARNING: CPU: 0 PID: 0 at kernel/sched/core.c:8511 sched_init+0x20/0x410
>            ^^^^^^^^^^^^^
>   Modules linked in:
>   CPU: 0 UID: 0 PID: 0 Comm: swapper Not tainted 6.14.0-01616-g94d7af2844aa #4 PREEMPT(undef)
>   ^^^^^^^^^^^^^^^^^^
> 
> In fact the later dump is richer, as it includes the 'comm' line as well.
> 
> As a side effect, this makes some space for the 'file' field to be augmented
> with extra information (the condition string in particular).
> 
> Signed-off-by: Ingo Molnar <mingo@kernel.org>
> ---
>  kernel/panic.c | 7 ++-----
>  1 file changed, 2 insertions(+), 5 deletions(-)
> 
> diff --git a/kernel/panic.c b/kernel/panic.c
> index d8635d5cecb2..3101c21ca39f 100644
> --- a/kernel/panic.c
> +++ b/kernel/panic.c
> @@ -725,12 +725,9 @@ void __warn(const char *file, int line, void *caller, unsigned taint,
>  	disable_trace_on_warning();
>  
>  	if (file)
> -		pr_warn("WARNING: CPU: %d PID: %d at %s:%d %pS\n",
> -			raw_smp_processor_id(), current->pid, file, line,
> -			caller);
> +		pr_warn("WARNING: %s:%d at %pS\n", file, line, caller);
>  	else
> -		pr_warn("WARNING: CPU: %d PID: %d at %pS\n",
> -			raw_smp_processor_id(), current->pid, caller);
> +		pr_warn("WARNING: at %pS\n", caller);

Over the years I've had concurrent WARNs interleaved on the console.
This duplicate information would allow untangling some of that.
Re: [PATCH 5/5] bugs/core: Do not print CPU and PID values in__warn() output
Posted by Ingo Molnar 8 months, 3 weeks ago 
* Peter Zijlstra <peterz@infradead.org> wrote:

> On Wed, Mar 26, 2025 at 09:47:50AM +0100, Ingo Molnar wrote:
> > For most warnings it's repeated anyway a few lines later, as
> > part of the register dump:
> > 
> >   WARNING: CPU: 0 PID: 0 at kernel/sched/core.c:8511 sched_init+0x20/0x410
> >            ^^^^^^^^^^^^^
> >   Modules linked in:
> >   CPU: 0 UID: 0 PID: 0 Comm: swapper Not tainted 6.14.0-01616-g94d7af2844aa #4 PREEMPT(undef)
> >   ^^^^^^^^^^^^^^^^^^
> > 
> > In fact the later dump is richer, as it includes the 'comm' line as well.
> > 
> > As a side effect, this makes some space for the 'file' field to be augmented
> > with extra information (the condition string in particular).
> > 
> > Signed-off-by: Ingo Molnar <mingo@kernel.org>
> > ---
> >  kernel/panic.c | 7 ++-----
> >  1 file changed, 2 insertions(+), 5 deletions(-)
> > 
> > diff --git a/kernel/panic.c b/kernel/panic.c
> > index d8635d5cecb2..3101c21ca39f 100644
> > --- a/kernel/panic.c
> > +++ b/kernel/panic.c
> > @@ -725,12 +725,9 @@ void __warn(const char *file, int line, void *caller, unsigned taint,
> >  	disable_trace_on_warning();
> >  
> >  	if (file)
> > -		pr_warn("WARNING: CPU: %d PID: %d at %s:%d %pS\n",
> > -			raw_smp_processor_id(), current->pid, file, line,
> > -			caller);
> > +		pr_warn("WARNING: %s:%d at %pS\n", file, line, caller);
> >  	else
> > -		pr_warn("WARNING: CPU: %d PID: %d at %pS\n",
> > -			raw_smp_processor_id(), current->pid, caller);
> > +		pr_warn("WARNING: at %pS\n", caller);
> 
> Over the years I've had concurrent WARNs interleaved on the console.
> This duplicate information would allow untangling some of that.

Fair enough.

With patch #1 below:

 - Reintroduced CPU and PID

 - Added a '->comm[]' to the warning line: in practice it's often more 
   valuable than a PID...

 - Re-ordered the fields of information by value: higher value to lower 
   value.

Here's an overview of the output variants (-v3 is the latestest):

 -vanilla: WARNING: CPU: 0 PID: 0 at kernel/sched/core.c:8511 sched_init+0x20/0x410
 -v1:      WARNING: CPU: 0 PID: 0 at [ptr == 0 && 1] kernel/sched/core.c:8511 sched_init+0x20/0x410
 -v2:      WARNING: [ptr == 0 && 1] kernel/sched/core.c:8511 at sched_init+0x20/0x410
 -v3:      WARNING: [ptr == 0 && 1] kernel/sched/core.c:8511 at sched_init+0x20/0x410 CPU#0: swapper/0

I think I like -v3: a poor guy's single line trace entry in essence.

I suppose we'll have some worst-case situations with really complex 
conditions bloating the output when they trigger, but a quick git grep 
suggests that most WARN_ON_ONCE() instances are sane, in kernel/ at 
least.

Side note, found some WARN_ON_ONCE() weirdnesses:

  kernel/sched/syscalls.c:        WARN_ON_ONCE(sched_setscheduler_nocheck(p, SCHED_FIFO, &sp) != 0);
  kernel/sched/syscalls.c:        WARN_ON_ONCE(sched_setscheduler_nocheck(p, SCHED_FIFO, &sp) != 0);
  kernel/sched/syscalls.c:        WARN_ON_ONCE(sched_setattr_nocheck(p, &attr) != 0);

I don't think WARN_ON()s with complex side effects are considered 
kernel META these days, right? The pattern in patch #2 below is
longer, but less surprising, IMHO.

Thanks,

	Ingo

---
 kernel/panic.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/kernel/panic.c b/kernel/panic.c
index 01746e89a7a1..6478dd70ec6d 100644
--- a/kernel/panic.c
+++ b/kernel/panic.c
@@ -724,10 +724,15 @@ void __warn(const char *file, int line, void *caller, unsigned taint,
 
 	disable_trace_on_warning();
 
-	if (file)
-		pr_warn("WARNING: %s:%d at %pS\n", file, line, caller);
-	else
-		pr_warn("WARNING: at %pS\n", caller);
+	if (file) {
+		pr_warn("WARNING: %s:%d at %pS CPU#%d: %s/%d\n",
+			file, line, caller,
+			raw_smp_processor_id(), current->comm, current->pid);
+	} else {
+		pr_warn("WARNING: at %pS CPU#%d: %s/%d\n",
+			caller,
+			raw_smp_processor_id(), current->comm, current->pid);
+	}
 
 #pragma GCC diagnostic push
 #ifndef __clang__

 kernel/sched/syscalls.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/kernel/sched/syscalls.c b/kernel/sched/syscalls.c
index c326de1344fb..1a91231d3ad8 100644
--- a/kernel/sched/syscalls.c
+++ b/kernel/sched/syscalls.c
@@ -846,7 +846,10 @@ int sched_setscheduler_nocheck(struct task_struct *p, int policy,
 void sched_set_fifo(struct task_struct *p)
 {
 	struct sched_param sp = { .sched_priority = MAX_RT_PRIO / 2 };
-	WARN_ON_ONCE(sched_setscheduler_nocheck(p, SCHED_FIFO, &sp) != 0);
+	int err;
+
+	err = sched_setscheduler_nocheck(p, SCHED_FIFO, &sp);
+	WARN_ON_ONCE(err);
 }
 EXPORT_SYMBOL_GPL(sched_set_fifo);

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.