linux-kernel - Re: [RFC] iommu: arm-smmu: stall support

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20170927143534.GQ8398@8bytes.org>
Date:   Wed, 27 Sep 2017 16:35:34 +0200
From:   Joerg Roedel <joro@...tes.org>
To:     Jean-Philippe Brucker <jean-philippe.brucker@....com>
Cc:     Rob Clark <robdclark@...il.com>,
        linux-arm-msm <linux-arm-msm@...r.kernel.org>,
        Will Deacon <Will.Deacon@....com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        "iommu@...ts.linux-foundation.org" <iommu@...ts.linux-foundation.org>,
        "linux-arm-kernel@...ts.infradead.org" 
        <linux-arm-kernel@...ts.infradead.org>
Subject: Re: [RFC] iommu: arm-smmu: stall support

Hi Jean,

On Wed, Sep 27, 2017 at 02:49:00PM +0100, Jean-Philippe Brucker wrote:
> I like this approach. When the device driver registers a fault handler,
> it also tells when it would like to be called (either in atomic context,
> blocking context, or both).

Is there a use-case for calling the same handler from both contexts?

> enum iommu_fault_status {
>         IOMMU_FAULT_STATUS_NONE = 0,
>         IOMMU_FAULT_STATUS_FAILURE,
>         IOMMU_FAULT_STATUS_INVALID,
>         IOMMU_FAULT_STATUS_HANDLED,
>         IOMMU_FAULT_STATUS_IGNORE,
> };

This all certainly makes sense for the PRI/PASID case, but I don't think
that it makes sense yet to extend the existing report_iommu_fault()
interface to also handle PASID/PPR faults.

The later needs a lot more parameters to successfully handle a fault. In
the AMD driver these are all in 'struct fault', the relevant members
are:

        u64 address;
        u16 devid;
        u16 pasid;
        u16 tag;
        u16 finish;
        u16 flags;

And passing all this through the existing interface which also handles
non-pasid faults is cumbersome. So I'd like to keep the PASID/PPR
interface separate from the old one for now.

Regards,

	Joerg