linux-kernel - Re: [PATCH v2] irqchip/mips-gic: Add missing shared interrupt handler

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] [day] [month] [year] [list]

Date:   Thu, 6 Jul 2017 09:36:27 -0700
From:   Paul Burton <paul.burton@...tec.com>
To:     Thomas Gleixner <tglx@...utronix.de>, <jrjang@...il.com>
CC:     Jason Cooper <jason@...edaemon.net>,
        Marc Zyngier <marc.zyngier@....com>,
        LKML <linux-kernel@...r.kernel.org>,
        Matt Redfearn <matt.redfearn@...tec.com>,
        Ralf Baechle <ralf@...ux-mips.org>
Subject: Re: [PATCH v2] irqchip/mips-gic: Add missing shared interrupt handler

Hello,

On Thursday, 6 July 2017 05:03:59 PDT Thomas Gleixner wrote:
> On Thu, 6 Jul 2017, jrjang@...il.com wrote:
> 
> CC+ MIPS folks. There is a reason WHY I added them to my previous reply.

Thanks Thomas.

> > From: Jun-Ru Chang <jrjang@...il.com>
> > 
> > Commit b87281e7f205 ("irqchip/mips-gic: Remove device IRQ domain")
> > removes the device IRQ domain and uses gic_irq_domain_alloc() to
> > allocate the shared/local domain. However, the shared interrupt handler
> > is not set after allocating. It causes that the system hangs with
> > "unexpected IRQ" messages disaply.

I don't see the driver doing irq_set_handler(..., handle_level_irq) for device 
interrupts prior to b87281e7f205 ("irqchip/mips-gic: Remove device IRQ 
domain"), so I'm confused about 1) why this is a problem given that we don't 
see it on our systems and 2) why you point the finger at that commit.

What system are you using?

I presume it's a device interrupt you're seeing issues with? How are you using 
it? Via device tree or otherwise?

Can you provide directions to reproduce this problem?

> > And commit 8ada00a650ec ("irqchip/mips-gic: Replace static map with
> > dynamic") renames gic_irq_domain_alloc() to gic_irq_domain_map() to set
> > up the handler and chip. Fix this by setting the handle_level_irq
> > handler for shared interrupts.

I also don't understand from this why that commit is relevant either. That 
commit didn't remove any calls to irq_set_handler, which is what your patch 
adds. You outline above some of what 8ada00a650ec ("irqchip/mips-gic: Replace 
static map with dynamic") does, but not why it's a problem. It's difficult to 
see why you think it needs fixing if you don't describe why you think it's 
broken.

Thanks,
    Paul

> > 
> > Fixes: b87281e7f205 ("irqchip/mips-gic: Remove device IRQ domain")
> > Signed-off-by: Jun-Ru Chang <jrjang@...il.com>
> > ---
> > Change in v2:
> > - changed commit message
> > 
> >  drivers/irqchip/irq-mips-gic.c | 1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/drivers/irqchip/irq-mips-gic.c
> > b/drivers/irqchip/irq-mips-gic.c index 929f8558bf1c..0a5fb30deb17 100644
> > --- a/drivers/irqchip/irq-mips-gic.c
> > +++ b/drivers/irqchip/irq-mips-gic.c
> > @@ -716,6 +716,7 @@ static int gic_irq_domain_map(struct irq_domain *d,
> > unsigned int virq,> 
> >  		if (err)
> >  		
> >  			return err;
> > 
> > +		irq_set_handler(virq, handle_level_irq);
> > 
> >  		return gic_shared_irq_domain_map(d, virq, hwirq, 0);
> >  	
> >  	}


Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)