Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 27 Oct 2022 12:13:01 +0200
From:   Greg KH <gregkh@linuxfoundation.org>
To:     Rishabh Bhatnagar <risbhat@amazon.com>
Cc:     stable@vger.kernel.org, sashal@kernel.org, tglx@linutronix.de,
        mingo@redhat.com, linux-kernel@vger.kernel.org, benh@amazon.com,
        mbacco@amazon.com
Subject: Re: [PATCH 0/6] IRQ handling patches backport to 4.14 stable
Message-ID: <Y1pZrdySrzod3NCM@kroah.com>
References: <20220929210651.12308-1-risbhat@amazon.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220929210651.12308-1-risbhat@amazon.com>
Precedence: bulk

On Thu, Sep 29, 2022 at 09:06:45PM +0000, Rishabh Bhatnagar wrote:
> This patch series backports a bunch of patches related IRQ handling
> with respect to freeing the irq line while IRQ is in flight at CPU
> or at the hardware level.
> Recently we saw this issue in serial 8250 driver where the IRQ was being
> freed while the irq was in flight or not yet delivered to the CPU. As a
> result the irqchip was going into a wedged state and IRQ was not getting
> delivered to the cpu. These patches helped fixed the issue in 4.14
> kernel.
> Let us know if more patches need backporting.
> 
> Lukas Wunner (2):
>   genirq: Update code comments wrt recycled thread_mask
>   genirq: Synchronize only with single thread on free_irq()
> 
> Thomas Gleixner (4):
>   genirq: Delay deactivation in free_irq()
>   genirq: Fix misleading synchronize_irq() documentation
>   genirq: Add optional hardware synchronization for shutdown
>   x86/ioapic: Implement irq_get_irqchip_state() callback
> 
>  arch/x86/kernel/apic/io_apic.c |  46 ++++++++++++++
>  kernel/irq/autoprobe.c         |   6 +-
>  kernel/irq/chip.c              |   6 ++
>  kernel/irq/cpuhotplug.c        |   2 +-
>  kernel/irq/internals.h         |   5 ++
>  kernel/irq/manage.c            | 106 ++++++++++++++++++++++-----------
>  6 files changed, 133 insertions(+), 38 deletions(-)

A simple build test breaks with this series applied:

ld: kernel/irq/manage.o: in function `__synchronize_hardirq':
manage.c:(.text+0x149): undefined reference to `__irq_get_irqchip_state'
ld: kernel/irq/manage.o: in function `irq_get_irqchip_state':
manage.c:(.text+0x5a2): undefined reference to `__irq_get_irqchip_state'
make: *** [Makefile:1038: vmlinux] Error 1

How did you test this?

{sigh}

I'm dropping all of these from my queue.  I think you need to just keep
this in your device-specific tree as obviously this is not ready to be
backported anywhere.

greg k-h