Re: [PATCH v2] ARM, ARM64, LONGARCH: Delay HW BP notification to task_work()

[Sebastian Andrzej Siewior]

On 2026-09-07 15:46:09 [+0100], Russell King wrote:
> On Fri, Sep 04, 2026 at 05:54:04PM +0200, Sebastian Andrzej Siewior wrote:
> > On 2026-09-03 13:31:46 [-0400], Steven Rostedt wrote:
> > > On Thu, 3 Sep 2026 17:47:24 +0100
> > > Russell King <linux@xxxxxxxxxxxxxxx> wrote:
> > > 
> > > > What do you expect to happen with this? Which arch maintainer should
> > > > apply this patch that affects three different independently maintained
> > > > architectures?
> > 
> > Well, if you as the ARM person are fine with ARM part, an ACK would be
> > appreciated. If you have something to object, it would be nice to let me
> > know so I can adjust accordingly.
> 
> First, I only just looked at the email - you may have noticed I'm not
> very active with the kernel anymore, as I'm a full-time carer at the
> moment.
> 
> Second, I'm not fine with the patch as-is, because it's not clear how
> it could be merged in its current form. It would be inappropriate to
> give an Ack, because an Ack means I'm fine with it. I'm not.
> 
> Third, I haven't actually reviewed the patch itself (see first point).
> 
> Looking deeper now, I see nothing in the linked report, nor in your
> write-up that identifies what is responsible for the context being
> inappropriate to deliver the SIGTRAP. See the first point - I'm not

The linked report shows a backtrace by "BUG: sleeping function"
originating from el0_breakpt() -> perf_bp_event() following by sending a
signal force_sig_fault() followed by rt_spin_lock() and
__might_resched(). It also contains a reproducer.

The first sentence has the part
| The hardware event is
| delivered with disabled interrupts and perf intrastrucure expects
| disabled interrupts while the overflow callback is invoked.

This is what happens, followed by 

| The callback then sends a SIGTRAP signal for which it acquires
| sighand_struct::siglock, a spinlock_t which becomes a sleeping lock and
| must not be acquired in atomic context.

This is not really specific to PREEMPT_RT as we enforce to not allow
raw_spinlock_t -> spinlock_t on !PREEMPT_RT via lockdep, too.

> able to read lots of code to work that out, and presumably you've
> already done that ground work to come up with a fix. This is *vital*
> information when someone looks back at why this change was made as
> it helps to understand how the problem occurs. This is, after all,
> the whole point of commit messages - to help people look back and
> understand why a change was made. The "what the changes are" part
> is already covered by the patch showing what the changes were. It's
> the stuff that *isn't* in the patch that needs to be in the commit
> message.
> 
> So... don't describe the code changes in a commit message.

I think both parts (describing the issue and the change) are part of the
commit message.

> So no, an Ack is not appropriate.

Sebastian