[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250805161035.GW26511@ziepe.ca>
Date: Tue, 5 Aug 2025 13:10:35 -0300
From: Jason Gunthorpe <jgg@...pe.ca>
To: Xu Yilun <yilun.xu@...ux.intel.com>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@...nel.org>,
linux-coco@...ts.linux.dev, kvmarm@...ts.linux.dev,
linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
aik@....com, lukas@...ner.de, Samuel Ortiz <sameo@...osinc.com>,
Suzuki K Poulose <Suzuki.Poulose@....com>,
Steven Price <steven.price@....com>,
Catalin Marinas <catalin.marinas@....com>,
Marc Zyngier <maz@...nel.org>, Will Deacon <will@...nel.org>,
Oliver Upton <oliver.upton@...ux.dev>
Subject: Re: [RFC PATCH v1 06/38] iommufd: Add and option to request for bar
mapping with IORESOURCE_EXCLUSIVE
On Tue, Aug 05, 2025 at 10:26:06AM +0800, Xu Yilun wrote:
> > IMHO there is no use case for that, it should arguably be global to
> > the whole kernel.
>
> I think there are 2 topics here:
>
> 1. Prevent VFIO mmap
> 2. Prevent /sys/../resource, /dev/mem users
>
> I assume you are refering to the 2nd, then I agree.
Yes, the region stuff is only about #2.
> >
> > > The original purpose of this pci_region_request_*() is to prevent
> > > further mmap/read/write against a vfio_cdev FD which would be used
> >
> > No way, the VFIO internal mmap should be controled by VFIO not by
>
> I assume your point is never to use more than one request region in the
> same driver to achieve some mutual exclusion. I'm good to it. We could
> switch to some bound flag.
Yes and yes
> > The only thing request region should do is prevent /sys/../resource,
> > /dev/mem users and so on, which is why it can and should be
> > global. Arguably VFIO should always block those things but
> > historically hasn't..
>
> Agree. So seems no need a global option?
I'd ask Alex if he is OK with a global behavior change to make vfio
exclusive, after any required fixing to make vfio only request the
regions once at driver probe time..
Jason
Powered by blists - more mailing lists