Re: [PATCH v3 09/14] iommu/arm-smmu: Prevent forced unbinding of Arm SMMU drivers

From: John Garry <john.garry@huawei.com>
To: Saravana Kannan <saravanak@google.com>
Cc: iommu@lists.linuxfoundation.org,
	Jean-Philippe Brucker <jean-philippe@linaro.org>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Bjorn Helgaas <bhelgaas@google.com>,
	Will Deacon <will@kernel.org>,
	"Isaac J. Manjarres" <isaacm@codeaurora.org>,
	Robin Murphy <robin.murphy@arm.com>
Subject: Re: [PATCH v3 09/14] iommu/arm-smmu: Prevent forced unbinding of Arm SMMU drivers
Date: Wed, 27 Nov 2019 11:04:25 +0000	[thread overview]
Message-ID: <af1dc92a-ca98-fb22-835f-5ceb85e86b1b@huawei.com> (raw)
In-Reply-To: <CAGETcx8Hkta6scFdiG=eQypsQ--jrR1YisaOQATCbMiu+aG8sg@mail.gmail.com>

On 26/11/2019 20:27, Saravana Kannan wrote:
> On Tue, Nov 26, 2019 at 1:13 AM John Garry <john.garry@huawei.com> wrote:
>>
>> On 21/11/2019 11:49, Will Deacon wrote:
>>> Forcefully unbinding the Arm SMMU drivers is a pretty dangerous operation,
>>> since it will likely lead to catastrophic failure for any DMA devices
>>> mastering through the SMMU being unbound. When the driver then attempts
>>> to "handle" the fatal faults, it's very easy to trip over dead data
>>> structures, leading to use-after-free.
>>>
>>> On John's machine, he reports that the machine was "unusable" due to
>>> loss of the storage controller following a forced unbind of the SMMUv3
>>> driver:
>>>
>>>     | # cd ./bus/platform/drivers/arm-smmu-v3
>>>     | # echo arm-smmu-v3.0.auto > unbind
>>>     | hisi_sas_v2_hw HISI0162:01: CQE_AXI_W_ERR (0x800) found!
>>>     | platform arm-smmu-v3.0.auto: CMD_SYNC timeout at 0x00000146
>>>     | [hwprod 0x00000146, hwcons 0x00000000]
>>>
>>> Prevent this forced unbinding of the drivers by setting "suppress_bind_attrs"
>>> to true.
>>
>> This seems a reasonable approach for now.
>>
>> BTW, I'll give this series a spin this week, which again looks to be
>> your iommu/module branch, excluding the new IORT patch.
> 

Hi Saravana,

> Is this on a platform where of_devlink creates device links between
> the iommu device and its suppliers?I'm guessing no? Because device
> links should for unbinding of all the consumers before unbinding the
> supplier.

I'm only really interested in ACPI, TBH.

> 
> Looks like it'll still allow the supplier to unbind if the consumers
> don't allow unbinding. Is that the case here?

So just unbinding the driver from a device does not delete the device 
nor exit the device from it's IOMMU group - so we keep the reference to 
the SMMU ko. As such, I don't know how to realistically test unloading 
the SMMU ko when we have platform devices involved. Maybe someone can 
enlighten me...

Thanks,
John
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu