linux-kernel - [PATCH 0/1] MIPS: Disable address swizzling on _

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]

Date:	Tue, 26 May 2009 16:57:34 -0700
From:	Kevin Cernekee <cernekee@...il.com>
To:	ralf@...ux-mips.org
Cc:	linux-mips@...ux-mips.org, linux-kernel@...r.kernel.org
Subject: [PATCH 0/1] MIPS: Disable address swizzling on __raw MMIO operations

I have a big-endian MIPS32-based ASIC configured as follows:

CONFIG_SWAP_IO_SPACE is not set

mangle-port.h says:

#define __swizzle_addr_b(port)	((port) ^ 3)
#define __swizzle_addr_w(port)	((port) ^ 2)
#define __swizzle_addr_l(port)	(port)
#define __swizzle_addr_q(port)	(port)

(copied from mach-ip32/mangle-port.h)

The PCI drivers use {read,write}[bwl].  PCI byte and word (16-bit)
accesses are address-swizzled but not endian-swapped.  Since 32-bit
accesses are the common case, this generates the most efficient code.

The MTD drivers (in my case, physmap) use __raw_{read,write}[bwl].
This is a problem because on MIPS, the __raw functions still enable
address swizzling.

I am submitting a patch to disable address swizzling for the __raw
operations.

There are currently three other MIPS platforms using address
swizzling:

txx9/jmr3927 only uses the swizzle facility to make the rtc-ds1742
driver work.  This driver uses standard (non-__raw) readb()
operations, which will continue to function normally with my patch in
place.

sgi-ip27 swizzles 16-bit PCI word addresses but not byte addresses.
It only registers a single platform_device (rtc-m48t35) which has no
__raw operations.

sgi-ip32 swizzles both 16-bit and 8-bit PCI addresses.  It registers
the following platform_device's:

8250 - uses standard read/write operations
meth - uses volatile struct accesses only
sgio2audio - uses readq/writeq
sgi_btns - uses readq/writeq
rtc_cmos - uses inb_p/outb_p

Based on this information, I do not believe that my change will have
an adverse impact on any other systems.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/