lockdep: Only trace IRQ edges
authorNicholas Piggin <npiggin@gmail.com>
Thu, 23 Jul 2020 10:56:14 +0000 (20:56 +1000)
committerPeter Zijlstra <peterz@infradead.org>
Wed, 26 Aug 2020 10:41:56 +0000 (12:41 +0200)
commit044d0d6de9f50192f9697583504a382347ee95ca
tree4ffacb721b2da9a252b54749ccc64e6acb2ae83d
parent99dc56feb7932020502d40107a712fa302b32082
lockdep: Only trace IRQ edges

Problem:

  raw_local_irq_save(); // software state on
  local_irq_save(); // software state off
  ...
  local_irq_restore(); // software state still off, because we don't enable IRQs
  raw_local_irq_restore(); // software state still off, *whoopsie*

existing instances:

 - lock_acquire()
     raw_local_irq_save()
     __lock_acquire()
       arch_spin_lock(&graph_lock)
         pv_wait() := kvm_wait() (same or worse for Xen/HyperV)
           local_irq_save()

 - trace_clock_global()
     raw_local_irq_save()
     arch_spin_lock()
       pv_wait() := kvm_wait()
 local_irq_save()

 - apic_retrigger_irq()
     raw_local_irq_save()
     apic->send_IPI() := default_send_IPI_single_phys()
       local_irq_save()

Possible solutions:

 A) make it work by enabling the tracing inside raw_*()
 B) make it work by keeping tracing disabled inside raw_*()
 C) call it broken and clean it up now

Now, given that the only reason to use the raw_* variant is because you don't
want tracing. Therefore A) seems like a weird option (although it can be done).
C) is tempting, but OTOH it ends up converting a _lot_ of code to raw just
because there is one raw user, this strips the validation/tracing off for all
the other users.

So we pick B) and declare any code that ends up doing:

raw_local_irq_save()
local_irq_save()
lockdep_assert_irqs_disabled();

broken. AFAICT this problem has existed forever, the only reason it came
up is because commit: 859d069ee1dd ("lockdep: Prepare for NMI IRQ
state tracking") changed IRQ tracing vs lockdep recursion and the
first instance is fairly common, the other cases hardly ever happen.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[rewrote changelog]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Tested-by: Marco Elver <elver@google.com>
Link: https://lkml.kernel.org/r/20200723105615.1268126-1-npiggin@gmail.com
arch/powerpc/include/asm/hw_irq.h
include/linux/irqflags.h