Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Sun, 4 Nov 2018 11:19:49 +0100 (CET)
From:   Thomas Gleixner <tglx@linutronix.de>
To:     Holger Schurig <holgerschurig@gmail.com>
cc:     linux-kernel@vger.kernel.org, Ingo Molnar <mingo@redhat.com>,
        x86@kernel.org
Subject: Re: BUG: "do_IRQ: 0.39 No irq handler for vector" from a 16550
 port
In-Reply-To: <87h8gzfr2p.fsf@gmail.com>
Message-ID: <alpine.DEB.2.21.1811041110380.3160@nanos.tec.linutronix.de>
References: <87k1lvg2z4.fsf@gmail.com> <87h8gzfr2p.fsf@gmail.com>
User-Agent: Alpine 2.21 (DEB 202 2017-01-01)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, 2 Nov 2018, Holger Schurig wrote:

> I said that kernel 4.9 doesn't show the issue. The same was for later
> kernels up to 4.13.
> 
> I had a compilation issue with 4.14 (which I later solved, something
> unrelated with tools/objcopy when compiling for a different
> architecture), so I did a git bisect between v4.13 and v4.15. This is
> the outcome:
>
> git bisect bad 464d12309e1b5829597793db551ae8ecaecf4036
> # first bad commit: [464d12309e1b5829597793db551ae8ecaecf4036] x86/vector: Switch IOAPIC to global reservation mode

Which is not surprising because the old model just hid the issue of
interrupts which came in after the interrupt had been torn down.

I have no idea what causes this, but it seems to be related to the
particular hardware/BIOS combination.

Thanks,

	tglx