[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1517729157.378701.1612886684264.JavaMail.zimbra@nod.at>
Date: Tue, 9 Feb 2021 17:04:44 +0100 (CET)
From: Richard Weinberger <richard@....at>
To: Luca Risolia <luca.risolia@...dio.unibo.it>
Cc: Miklos Szeredi <miklos@...redi.hu>,
Vignesh Raghavendra <vigneshr@...com>,
fuse-devel <fuse-devel@...ts.sourceforge.net>,
Miquel Raynal <miquel.raynal@...tlin.com>,
linux-kernel <linux-kernel@...r.kernel.org>,
Boris Brezillon <boris.brezillon@...labora.com>,
linux-mtd <linux-mtd@...ts.infradead.org>,
Ron Minnich <rminnich@...gle.com>, sven <sven@...fation.org>
Subject: Re: [fuse-devel] [PATCH 0/8] MUSE: Userspace backed MTD v3
----- Ursprüngliche Mail -----
> for example, many existing video applications use mmap() to map the
> device memory to userspace memory. Adding support for mmap() to CUSE
> would allow these apps to work without any modifications with CUSE-based
> device drivers other than kernel drivers.
So you want to access device memory via CUSE?
We have plenty of mechanisms in Linux to allow userspace accessing device memory.
E.g. /dev/mem, UIO, VFIO.
A simple (but ugly!) approach would be redirecting mmap() requests on CUSE devices to /dev/mem.
hmm?
Thanks,
//richard
Powered by blists - more mailing lists