netdev - Re: [PATCH 1/2] uio: introduce UIO_MEM_DMA

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: <20240201044416.GA14176@lst.de>
Date: Thu, 1 Feb 2024 05:44:16 +0100
From: Christoph Hellwig <hch@....de>
To: Chris Leech <cleech@...hat.com>
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Nilesh Javali <njavali@...vell.com>, Christoph Hellwig <hch@....de>,
	John Meneghini <jmeneghi@...hat.com>, Lee Duncan <lduncan@...e.com>,
	Mike Christie <michael.christie@...cle.com>,
	Hannes Reinecke <hare@...nel.org>, netdev@...r.kernel.org,
	linux-kernel@...r.kernel.org, linux-scsi@...r.kernel.org,
	GR-QLogic-Storage-Upstream@...vell.com
Subject: Re: [PATCH 1/2] uio: introduce UIO_MEM_DMA_COHERENT type

On Wed, Jan 31, 2024 at 01:44:50PM -0800, Chris Leech wrote:
> On Wed, Jan 31, 2024 at 1:29 PM Greg Kroah-Hartman
> <gregkh@...uxfoundation.org> wrote:
> >
> > On Wed, Jan 31, 2024 at 11:17:31AM -0800, Chris Leech wrote:
> > > Add a UIO memtype specifically for sharing dma_alloc_coherent
> > > memory with userspace, backed by dma_mmap_coherent.
> > >
> > > This is mainly for the bnx2/bnx2x/bnx2i "cnic" interface, although there
> > > are a few other uio drivers which map dma_alloc_coherent memory and
> > > could be converted to use dma_mmap_coherent as well.
> >
> > What other drivers could use this?  Patches doing the conversion would
> > be welcome, otherwise, again, I am very loath to take this
> > one-off-change for just a single driver that shouldn't be doing this in
> > the first place :)
> 
> uio_pruss and uio_dmem_genirq both appear to mmap dma_alloc_coherent
> memory as UIO_MEM_PHYS.  It might not be an issue on that platforms
> where those are used, but I'd be happy to include untested patches to
> convert them for better adherence to the DMA APIs.

Yes, they do need fixing.