linux-kernel - [PATCH 0/4] Fix and extend uio_dmem

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-next>] [day] [month] [year] [list]

Message-Id: <1463476940-26791-1-git-send-email-viktorin@rehivetech.com>
Date:	Tue, 17 May 2016 11:22:16 +0200
From:	Jan Viktorin <viktorin@...ivetech.com>
To:	devicetree@...r.kernel.org, linux-kernel@...r.kernel.org
Cc:	Jan Viktorin <viktorin@...ivetech.com>,
	Rob Herring <robh+dt@...nel.org>,
	Pawel Moll <pawel.moll@....com>,
	Mark Rutland <mark.rutland@....com>,
	Ian Campbell <ijc+devicetree@...lion.org.uk>,
	Kumar Gala <galak@...eaurora.org>,
	"Hans J. Koch" <hjk@...sjkoch.de>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Subject: [PATCH 0/4] Fix and extend uio_dmem_genirq

Hello,

as a part of my work related usually to FPGAs, I was playing with the
uio_dmem_genirq. First, I've discovered a (possible) bug in the
uio_dmem_genirq in exposing the dynamic regions to userspace (patch 0001).
When specifying a number of dynamic regions, one mapping is (due to
counting IORESOURCE_IRQ) skipped:

.../maps/map0/addr: 0xe000b000  <<< IORESOURCE_MEM from reg = <...>
.../maps/map0/size: 0x00001000
.../maps/map1/addr: 0xffffffff  <<< no mapping (skipped due to IORESOURCE_IRQ)
.../maps/map1/size: 0x00008000
.../maps/map2/addr: 0x1e210000  <<< the first dynamic mapping
.../maps/map2/size: 0x00008000
.../maps/map3/addr: 0x1e218000
.../maps/map3/size: 0x00008000
.../maps/map4/addr: 0x1e220000
.../maps/map4/size: 0x00008000

Next, I needed to do some DMA from userspace and I didn't like to change
the kernel at all. The uio_pdrv_genirq has a feature to specify of_id
when calling insmod to bind to a specific device. I've implemented this
(patches 0002-0004) for uio_dmem_genirq and added two OF properties to
specify memory allocations. The work is inspired by those commits (but
reworked):

* https://github.com/analogdevicesinc/linux/commit/757e2c54a613abed6145fcd572cf6c5d8b3fb7fc
* https://github.com/analogdevicesinc/linux/commit/45ec4148b290302c7de8e6a1de0d9f7b2833339b

Regards
Jan


Jan Viktorin (4):
  uio: fix dmem_region_start computation
  uio: UIO_IRQ_NONE is a valid option for uioinfo->irq
  uio: introduce devicetree bindings for uio_dmem_genirq
  uio: allow binding uio_pdrv_genirq.c to devices using command line
    option

 .../devicetree/bindings/uio/uio_dmem_genirq.txt    | 16 ++++
 drivers/uio/uio_dmem_genirq.c                      | 93 +++++++++++++++-------
 2 files changed, 81 insertions(+), 28 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/uio/uio_dmem_genirq.txt

-- 
2.8.0