linux-kernel - Re: mpic IRQ_TYPE

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:   Thu, 31 Aug 2017 08:52:21 +1000
From:   Michael Ellerman <mpe@...erman.id.au>
To:     Gregory Fong <gregory.0xf0@...il.com>,
        linuxppc-dev@...ts.ozlabs.org
Cc:     "linux-kernel\@vger.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: mpic IRQ_TYPE_BOTH handling

Hi Gregory,

Gregory Fong <gregory.0xf0@...il.com> writes:
> Hi all,
>
> In arch/powerpc/sysdev/mpic.c , it looks like IRQ_TYPE_EDGE_BOTH is
> handled the same way as IRQ_TYPE_EDGE_FALLING:
>
> static unsigned int mpic_type_to_vecpri(struct mpic *mpic, unsigned int type)
> {
>         /* Now convert sense value */
>         switch(type & IRQ_TYPE_SENSE_MASK) {
>         case IRQ_TYPE_EDGE_RISING:
>                 return MPIC_INFO(VECPRI_SENSE_EDGE) |
>                        MPIC_INFO(VECPRI_POLARITY_POSITIVE);
>         case IRQ_TYPE_EDGE_FALLING:
>         case IRQ_TYPE_EDGE_BOTH:
>                 return MPIC_INFO(VECPRI_SENSE_EDGE) |
>                        MPIC_INFO(VECPRI_POLARITY_NEGATIVE);
>         case IRQ_TYPE_LEVEL_HIGH:
>                 return MPIC_INFO(VECPRI_SENSE_LEVEL) |
>                        MPIC_INFO(VECPRI_POLARITY_POSITIVE);
>         case IRQ_TYPE_LEVEL_LOW:
>         default:
>                 return MPIC_INFO(VECPRI_SENSE_LEVEL) |
>                        MPIC_INFO(VECPRI_POLARITY_NEGATIVE);
>         }
> }
>
> If IRQ_TYPE_EDGE_BOTH is unsupported, shouldn't we be returning an
> error, instead of silently setting to use IRQ_TYPE_EDGE_FALLING?
> Something like the following (sorry if the diff wraps weirdly, on
> webmail at the moment):

I don't know this code so I asked Ben and he said something like
"PowerMacs never use BOTH, so it hasn't mattered, but Freescale machines
might".

So if you want to send a proper signed-off patch, and confirm that all
the callers can handle the error properly, then we can merge it.

cheers

> ----8<----
> diff --git a/arch/powerpc/sysdev/mpic.c b/arch/powerpc/sysdev/mpic.c
> index b9aac95..5867ea2 100644
> --- a/arch/powerpc/sysdev/mpic.c
> +++ b/arch/powerpc/sysdev/mpic.c
> @@ -876,6 +876,9 @@ int mpic_set_irq_type(struct irq_data *d, unsigned
> int flow_type)
>         if (src >= mpic->num_sources)
>                 return -EINVAL;
>
> +       if (flow_type & IRQ_TYPE_SENSE_MASK == IRQ_TYPE_EDGE_BOTH)
> +               return -EINVAL;
> +
>         vold = mpic_irq_read(src, MPIC_INFO(IRQ_VECTOR_PRI));
>
>         /* We don't support "none" type */
> ---->8----
>
> Thanks,
> Gregory