linux-kernel - Re: [PATCH] iommu/vt-d: Fix to flush cache of PASID directory table

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]

Message-ID: <5044217c-9380-f50a-2809-0e3dc80f7bd2@intel.com>
Date:   Fri, 16 Jun 2023 11:06:24 +0800
From:   Yanfei Xu <yanfei.xu@...el.com>
To:     Baolu Lu <baolu.lu@...ux.intel.com>, <dwmw2@...radead.org>,
        <joro@...tes.org>, <will@...nel.org>, <robin.murphy@....com>
CC:     <iommu@...ts.linux.dev>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] iommu/vt-d: Fix to flush cache of PASID directory table


On 6/16/2023 10:01 AM, Baolu Lu wrote:
> On 6/15/23 3:16 PM, Yanfei Xu wrote:
>> Even the PCI devices don't support pasid capability, PASID
>> table is mandatory for a PCI device in scalable mode. However
>> flushing cache of pasid directory table for these devices are
>> not taken after pasid table is allocated as the "size" of
>> table is zero. Fix to assign it with a page size.
>
> Documentation/process/submitting-patches.rst
>
> Please add more information about
>
> - Describe your problem.
> - Any background of the problem?
> - How your change fixes the problem.
> ...
>
Got it! Will improve these in commit message.

Just noticed this when reading the iommu code, no actual problem 
encountered.

>>
>> Fixes: 194b3348bdbb ("iommu/vt-d: Fix PASID directory pointer 
>> coherency")
>
> Do you need a Cc stable?

Yes, will add Cc: <stable@...r.kernel.org>

>
>> Signed-off-by: Yanfei Xu <yanfei.xu@...el.com>
>> ---
>>   drivers/iommu/intel/pasid.c | 4 +++-
>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/iommu/intel/pasid.c b/drivers/iommu/intel/pasid.c
>> index c5d479770e12..bde7df055865 100644
>> --- a/drivers/iommu/intel/pasid.c
>> +++ b/drivers/iommu/intel/pasid.c
>> @@ -115,7 +115,9 @@ int intel_pasid_alloc_table(struct device *dev)
>>                     intel_pasid_max_id);
>>         size = max_pasid >> (PASID_PDE_SHIFT - 3);
>> -    order = size ? get_order(size) : 0;
>> +    if (!size)
>> +        size = PAGE_SIZE;
>> +    order = get_order(size);
>>       pages = alloc_pages_node(info->iommu->node,
>>                    GFP_KERNEL | __GFP_ZERO, order);
>>       if (!pages) {
>
> Is it similar to
A little difference that real size may less than memory size calculated by
order. But it is no effect. I think this is simpler.

Thanks,
Yanfei

>
> diff --git a/drivers/iommu/intel/pasid.c b/drivers/iommu/intel/pasid.c
> index c5d479770e12..49fc5a038a14 100644
> --- a/drivers/iommu/intel/pasid.c
> +++ b/drivers/iommu/intel/pasid.c
> @@ -129,7 +129,7 @@ int intel_pasid_alloc_table(struct device *dev)
>         info->pasid_table = pasid_table;
>
>         if (!ecap_coherent(info->iommu->ecap))
> -               clflush_cache_range(pasid_table->table, size);
> +               clflush_cache_range(pasid_table->table, (1 << order) * 
> PAGE_SIZE);
>
>         return 0;
>  }
>
> ?
>
> Best regards,
> baolu