Re: [PATCH] sched/core: add taint on "BUG: sleeping function called from invalid context"

[Thomas Gleixner]

On Fri, 29 Jul 2016, Rusty Russell wrote:
> Vegard Nossum  writes:
> > Seeing this, it occurs to me that we should probably add a taint here:
> 
> Taint has traditionally meant "the user did something unsupported, take
> the bug report with a grain of salt".  Such as force removing a module.

We have changed the meaning of taint quite some time ago. It also denotes
e.g. that there was a bug, warning etc. So it's not necessarily 'user did
something wrong'.

It's very helpful to see that there was already a warnon/bug before the
backtrace you get from the reporter.

Thanks,

tglx

Re: [PATCH] sched/core: add taint on "BUG: sleeping function called from invalid context"

[Thomas Gleixner]

On Fri, 29 Jul 2016, Rusty Russell wrote:
> Vegard Nossum  writes:
> > Seeing this, it occurs to me that we should probably add a taint here:
> 
> Taint has traditionally meant "the user did something unsupported, take
> the bug report with a grain of salt".  Such as force removing a module.

We have changed the meaning of taint quite some time ago. It also denotes
e.g. that there was a bug, warning etc. So it's not necessarily 'user did
something wrong'.

It's very helpful to see that there was already a warnon/bug before the
backtrace you get from the reporter.

Thanks,

tglx

Re: [PATCH] sched/core: add taint on "BUG: sleeping function called from invalid context"

[Vegard Nossum]

On 07/29/2016 03:22 AM, Rusty Russell wrote:

Vegard Nossum  writes:

Seeing this, it occurs to me that we should probably add a taint here:


Taint has traditionally meant "the user did something unsupported, take
the bug report with a grain of salt".  Such as force removing a module.

So this seems wrong...


There are many types of taints. This is the one you mentioned:

   2 - A module was force loaded by insmod -f.

This is the one I'd like to set:

 512 - A kernel warning has occurred.

(from Documentation/sysctl/kernel.txt).

The flag I want to set is also set on any kind of WARN().


 BUG: sleeping function called from invalid context at mm/slab.h:388
 in_atomic(): 0, irqs_disabled(): 0, pid: 32211, name: trinity-c3
 Preemption disabled at:[] console_unlock+0x2f7/0x930


I am not proposing this patch on a whim: it was actually hiding real
bugs because I only saw an oops that happened later when in fact the
real bug happened where the "sleeping function called from invalid
context" was called.


Vegard

Re: [PATCH] sched/core: add taint on "BUG: sleeping function called from invalid context"

[Vegard Nossum]

On 07/29/2016 03:22 AM, Rusty Russell wrote:

Vegard Nossum  writes:

Seeing this, it occurs to me that we should probably add a taint here:


Taint has traditionally meant "the user did something unsupported, take
the bug report with a grain of salt".  Such as force removing a module.

So this seems wrong...


There are many types of taints. This is the one you mentioned:

   2 - A module was force loaded by insmod -f.

This is the one I'd like to set:

 512 - A kernel warning has occurred.

(from Documentation/sysctl/kernel.txt).

The flag I want to set is also set on any kind of WARN().


 BUG: sleeping function called from invalid context at mm/slab.h:388
 in_atomic(): 0, irqs_disabled(): 0, pid: 32211, name: trinity-c3
 Preemption disabled at:[] console_unlock+0x2f7/0x930


I am not proposing this patch on a whim: it was actually hiding real
bugs because I only saw an oops that happened later when in fact the
real bug happened where the "sleeping function called from invalid
context" was called.


Vegard

Re: [PATCH] sched/core: add taint on "BUG: sleeping function called from invalid context"

[Rusty Russell]

Vegard Nossum  writes:
> Seeing this, it occurs to me that we should probably add a taint here:

Taint has traditionally meant "the user did something unsupported, take
the bug report with a grain of salt".  Such as force removing a module.

So this seems wrong...

Cheers,
Rusty.


>
> BUG: sleeping function called from invalid context at mm/slab.h:388
> in_atomic(): 0, irqs_disabled(): 0, pid: 32211, name: trinity-c3
> Preemption disabled at:[] console_unlock+0x2f7/0x930
>
> CPU: 3 PID: 32211 Comm: trinity-c3 Not tainted 4.7.0-rc7+ #19
>^^^
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 
> Ubuntu-1.8.2-1ubuntu1 04/01/2014
>   8800b8a17160 81971441 88011a3c4c80
>  88011a3c4c80 8800b8a17198 81158067 0de6
>  88011a3c4c80 8390e07c 0184 
> Call Trace:
> [...]
>
> BUG: sleeping function called from invalid context at 
> arch/x86/mm/fault.c:1309
> in_atomic(): 0, irqs_disabled(): 0, pid: 32211, name: trinity-c3
> Preemption disabled at:[] down_trylock+0x13/0x80
>
> CPU: 3 PID: 32211 Comm: trinity-c3 Not tainted 4.7.0-rc7+ #19
>^^^
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 
> Ubuntu-1.8.2-1ubuntu1 04/01/2014
>   8800b8a17e08 81971441 88011a3c4c80
>  88011a3c4c80 8800b8a17e40 81158067 
>  88011a3c4c80 83437b20 051d 
> Call Trace:
> [...]
>
> Cc: Peter Zijlstra 
> Cc: Paul E. McKenney 
> Cc: Thomas Gleixner 
> Cc: Ingo Molnar 
> Cc: Rusty Russel 
> Signed-off-by: Vegard Nossum 
> ---
>  kernel/sched/core.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 97ee9ac..7171cf9 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -7573,6 +7573,7 @@ void ___might_sleep(const char *file, int line, int 
> preempt_offset)
>   }
>  #endif
>   dump_stack();
> + add_taint(TAINT_WARN, LOCKDEP_STILL_OK);
>  }
>  EXPORT_SYMBOL(___might_sleep);
>  #endif
> -- 
> 1.9.1

Re: [PATCH] sched/core: add taint on "BUG: sleeping function called from invalid context"

[Rusty Russell]

Vegard Nossum  writes:
> Seeing this, it occurs to me that we should probably add a taint here:

Taint has traditionally meant "the user did something unsupported, take
the bug report with a grain of salt".  Such as force removing a module.

So this seems wrong...

Cheers,
Rusty.


>
> BUG: sleeping function called from invalid context at mm/slab.h:388
> in_atomic(): 0, irqs_disabled(): 0, pid: 32211, name: trinity-c3
> Preemption disabled at:[] console_unlock+0x2f7/0x930
>
> CPU: 3 PID: 32211 Comm: trinity-c3 Not tainted 4.7.0-rc7+ #19
>^^^
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 
> Ubuntu-1.8.2-1ubuntu1 04/01/2014
>   8800b8a17160 81971441 88011a3c4c80
>  88011a3c4c80 8800b8a17198 81158067 0de6
>  88011a3c4c80 8390e07c 0184 
> Call Trace:
> [...]
>
> BUG: sleeping function called from invalid context at 
> arch/x86/mm/fault.c:1309
> in_atomic(): 0, irqs_disabled(): 0, pid: 32211, name: trinity-c3
> Preemption disabled at:[] down_trylock+0x13/0x80
>
> CPU: 3 PID: 32211 Comm: trinity-c3 Not tainted 4.7.0-rc7+ #19
>^^^
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 
> Ubuntu-1.8.2-1ubuntu1 04/01/2014
>   8800b8a17e08 81971441 88011a3c4c80
>  88011a3c4c80 8800b8a17e40 81158067 
>  88011a3c4c80 83437b20 051d 
> Call Trace:
> [...]
>
> Cc: Peter Zijlstra 
> Cc: Paul E. McKenney 
> Cc: Thomas Gleixner 
> Cc: Ingo Molnar 
> Cc: Rusty Russel 
> Signed-off-by: Vegard Nossum 
> ---
>  kernel/sched/core.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 97ee9ac..7171cf9 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -7573,6 +7573,7 @@ void ___might_sleep(const char *file, int line, int 
> preempt_offset)
>   }
>  #endif
>   dump_stack();
> + add_taint(TAINT_WARN, LOCKDEP_STILL_OK);
>  }
>  EXPORT_SYMBOL(___might_sleep);
>  #endif
> -- 
> 1.9.1