linux-kernel - Re: [PATCH v7 5/6] iommu/arm-smmu-v3: Add support for non-strict mode

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: <20180918171016.GL16498@arm.com>
Date:   Tue, 18 Sep 2018 18:10:17 +0100
From:   Will Deacon <will.deacon@....com>
To:     Robin Murphy <robin.murphy@....com>
Cc:     joro@...tes.org, thunder.leizhen@...wei.com,
        iommu@...ts.linux-foundation.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
        linuxarm@...wei.com, guohanjun@...wei.com, huawei.libin@...wei.com,
        john.garry@...wei.com
Subject: Re: [PATCH v7 5/6] iommu/arm-smmu-v3: Add support for non-strict mode

On Fri, Sep 14, 2018 at 03:30:23PM +0100, Robin Murphy wrote:
> From: Zhen Lei <thunder.leizhen@...wei.com>
> 
> Dynamically choose strict or non-strict mode for page table config based
> on the iommu domain type.

It's the domain type in conjunction with the attribute that determines
whether we use lazy or strict invalidation.

> Signed-off-by: Zhen Lei <thunder.leizhen@...wei.com>
> [rm: convert to domain attribute]
> Signed-off-by: Robin Murphy <robin.murphy@....com>
> ---
>  drivers/iommu/arm-smmu-v3.c | 30 ++++++++++++++++++++++++------
>  1 file changed, 24 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/iommu/arm-smmu-v3.c b/drivers/iommu/arm-smmu-v3.c
> index f10c852479fc..7bbfa5f7ce8e 100644
> --- a/drivers/iommu/arm-smmu-v3.c
> +++ b/drivers/iommu/arm-smmu-v3.c
> @@ -612,6 +612,7 @@ struct arm_smmu_domain {
>  	struct mutex			init_mutex; /* Protects smmu pointer */
>  
>  	struct io_pgtable_ops		*pgtbl_ops;
> +	bool				non_strict;
>  
>  	enum arm_smmu_domain_stage	stage;
>  	union {
> @@ -1633,6 +1634,9 @@ static int arm_smmu_domain_finalise(struct iommu_domain *domain)
>  	if (smmu->features & ARM_SMMU_FEAT_COHERENCY)
>  		pgtbl_cfg.quirks = IO_PGTABLE_QUIRK_NO_DMA;
>  
> +	if (smmu_domain->non_strict)
> +		pgtbl_cfg.quirks |= IO_PGTABLE_QUIRK_NON_STRICT;
> +
>  	pgtbl_ops = alloc_io_pgtable_ops(fmt, &pgtbl_cfg, smmu_domain);
>  	if (!pgtbl_ops)
>  		return -ENOMEM;
> @@ -1934,13 +1938,17 @@ static int arm_smmu_domain_get_attr(struct iommu_domain *domain,
>  {
>  	struct arm_smmu_domain *smmu_domain = to_smmu_domain(domain);
>  
> -	if (domain->type != IOMMU_DOMAIN_UNMANAGED)
> -		return -EINVAL;
> -
>  	switch (attr) {
>  	case DOMAIN_ATTR_NESTING:
> +		if (domain->type != IOMMU_DOMAIN_UNMANAGED)
> +			return -EINVAL;
>  		*(int *)data = (smmu_domain->stage == ARM_SMMU_DOMAIN_NESTED);
>  		return 0;
> +	case DOMAIN_ATTR_DMA_USE_FLUSH_QUEUE:
> +		if (domain->type != IOMMU_DOMAIN_DMA)
> +			return -EINVAL;
> +		*(int *)data = smmu_domain->non_strict;
> +		return 0;
>  	default:
>  		return -ENODEV;

Hmm, there's a change in behaviour here (and also in the set function)
which is that unknown attributes now return -ENODEV for managed domains
instead of -EINVAL. I don't know if that's a problem, but I'd be inclined
to switch on the domain type and then have a nested switch for the supported
attributes.

Will