linux-kernel - Re: [PATCH] genirq: Fix handle_nested_irq() for IRQF

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 14 Mar 2017 09:38:51 +0100 (CET)
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Tony Lindgren <tony@...mide.com>
cc:     linux-kernel@...r.kernel.org
Subject: Re: [PATCH] genirq: Fix handle_nested_irq() for IRQF_SHARED

On Mon, 13 Mar 2017, Tony Lindgren wrote:

> Shared interrupts can be registered with handle_nested_irq(), but
> currently only one of the registered handlers gets called.
> 
> The use of shared interrupts with handle_nested_irq() is probably rare,
> but at least a case of a shared VBUS interrupt between USB PHY and
> battery charger drivers for a PMIC makes sense.
> 
> Typically for_each_action_of_desc() gets called via handle_irq_event()
> at _handle_irq_event_percpu(), but handle_nested_irq() is different
> because of it's dummy irq_nested_primary_handler().
> 
> Let's fix the issue by calling the handler for all the entries in
> the action list.
>
> Fixes: 399b5da29b9f ("genirq: Support nested threaded irq handling")

It's not a fix, it's an extension. The nested facility was not meant to
handle shared interrupts in the first place. I really hope that hardware
folks finally understand that irq sharing is crap and broken...

> Signed-off-by: Tony Lindgren <tony@...mide.com>
> ---
>  kernel/irq/chip.c | 8 +++++---
>  1 file changed, 5 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c
> --- a/kernel/irq/chip.c
> +++ b/kernel/irq/chip.c
> @@ -348,9 +348,11 @@ void handle_nested_irq(unsigned int irq)
>  	irqd_set(&desc->irq_data, IRQD_IRQ_INPROGRESS);
>  	raw_spin_unlock_irq(&desc->lock);
>  
> -	action_ret = action->thread_fn(action->irq, action->dev_id);
> -	if (!noirqdebug)
> -		note_interrupt(desc, action_ret);
> +	for_each_action_of_desc(desc, action) {
> +		action_ret = action->thread_fn(action->irq, action->dev_id);
> +		if (!noirqdebug)
> +			note_interrupt(desc, action_ret);

This is wrong. See how __handle_irq_event_percpu() does this. But no need
to resend. See:

http://lkml.kernel.org/r/1488904098-5350-1-git-send-email-ckeepax@opensource.wolfsonmicro.com

Thanks,

	tglx