Re: [PATCH v2 3/3] timers: Disable memory pre-allocation of timer debug objects

[Thomas Gleixner]

On Thu, Jun 12 2025 at 12:55, kernel test robot wrote:
> kernel test robot noticed
> "WARNING:possible_circular_locking_dependency_detected" on:

The lockdep issue comes from patch 2/3 which prints with the hash bucket
lock held.

> commit: 89fd87e046372ce70f8146357324cacd365369fe ("[PATCH v2 3/3] timers: Disable memory pre-allocation of timer debug objects")

This one triggers it because the preallocation disable makes debug
object go out of memory.

> [ 92.564990][ T1] _printk (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:2475) 
> [ 92.565340][ T1] lookup_object_or_alloc (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:700) 

That's the:

       debug_objects_disable("out of memory");

introduced by patch 2/3. The printk() in debug_objects_disable() needs
to be deferred.

> [ 92.565838][ T1] __debug_object_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:750) 
> [ 92.566274][ T1] debug_object_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:785) 

Which means that the changelog saying:

  "So the chance that debug_objects gets disabled because it is running out
   of free debug_object should be minimal."

is slightly off ....

Thanks,

        tglx