Dave Hylands reports:
| We've observed a problem with dma_alloc_writecombine when the system
| is under heavy load (heavy bus traffic). We've managed to reduce the
| problem to the following snippet, which is run from a kthread in a
| continuous loop:
|
| void *virtAddr;
| dma_addr_t physAddr;
| unsigned int numBytes = 256;
|
| for (;;) {
| virtAddr = dma_alloc_writecombine(NULL,
| numBytes, &physAddr, GFP_KERNEL);
| if (virtAddr == NULL) {
| printk(KERN_ERR "Running out of memory\n");
| break;
| }
|
| /* access DMA memory allocated */
| tmp = virtAddr;
| *tmp = 0x77;
|
| /* free DMA memory */
| dma_free_writecombine(NULL,
| numBytes, virtAddr, physAddr);
|
| ...sleep here...
| }
|
| By itself, the code will run forever with no issues. However, as we
| increase our bus traffic (typically using DMA) then the *tmp = 0x77
| line will eventually cause a page fault. If we add a small delay (a
| few microseconds) before the *tmp = 0x77, then we don't see a page
| fault, even under heavy load.
A dsb() is required after modifying the PTE entries to ensure that they
will always be visible. Add this dsb().
Reported-by: Dave Hylands <dhylands@gmail.com>
Tested-by: Dave Hylands <dhylands@gmail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
CPU_32v6K is selected by CPU_V7 but it only depends on CPU_V6.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
kunmap_atomic() is currently at level -4 on Rusty's "Hard To Misuse"
list[1] ("Follow common convention and you'll get it wrong"), except in
some architectures when CONFIG_DEBUG_HIGHMEM is set[2][3].
kunmap() takes a pointer to a struct page; kunmap_atomic(), however, takes
takes a pointer to within the page itself. This seems to once in a while
trip people up (the convention they are following is the one from
kunmap()).
Make it much harder to misuse, by moving it to level 9 on Rusty's list[4]
("The compiler/linker won't let you get it wrong"). This is done by
refusing to build if the type of its first argument is a pointer to a
struct page.
The real kunmap_atomic() is renamed to kunmap_atomic_notypecheck()
(which is what you would call in case for some strange reason calling it
with a pointer to a struct page is not incorrect in your code).
The previous version of this patch was compile tested on x86-64.
[1] http://ozlabs.org/~rusty/index.cgi/tech/2008-04-01.html
[2] In these cases, it is at level 5, "Do it right or it will always
break at runtime."
[3] At least mips and powerpc look very similar, and sparc also seems to
share a common ancestor with both; there seems to be quite some
degree of copy-and-paste coding here. The include/asm/highmem.h file
for these three archs mention x86 CPUs at its top.
[4] http://ozlabs.org/~rusty/index.cgi/tech/2008-03-30.html
[5] As an aside, could someone tell me why mn10300 uses unsigned long as
the first parameter of kunmap_atomic() instead of void *?
Signed-off-by: Cesar Eduardo Barros <cesarb@cesarb.net>
Cc: Russell King <linux@arm.linux.org.uk> (arch/arm)
Cc: Ralf Baechle <ralf@linux-mips.org> (arch/mips)
Cc: David Howells <dhowells@redhat.com> (arch/frv, arch/mn10300)
Cc: Koichi Yasutake <yasutake.koichi@jp.panasonic.com> (arch/mn10300)
Cc: Kyle McMartin <kyle@mcmartin.ca> (arch/parisc)
Cc: Helge Deller <deller@gmx.de> (arch/parisc)
Cc: "James E.J. Bottomley" <jejb@parisc-linux.org> (arch/parisc)
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> (arch/powerpc)
Cc: Paul Mackerras <paulus@samba.org> (arch/powerpc)
Cc: "David S. Miller" <davem@davemloft.net> (arch/sparc)
Cc: Thomas Gleixner <tglx@linutronix.de> (arch/x86)
Cc: Ingo Molnar <mingo@redhat.com> (arch/x86)
Cc: "H. Peter Anvin" <hpa@zytor.com> (arch/x86)
Cc: Arnd Bergmann <arnd@arndb.de> (include/asm-generic)
Cc: Rusty Russell <rusty@rustcorp.com.au> ("Hard To Misuse" list)
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
v2: Fixes from Mike Rapoport
- remove unused header files (mach/dma.h and mach/nand.h)
- remove tegra 1 references from Makefile.boot
v2: fixes from Russell King
- remove mach/io.h include from mach/iomap.h
- fix whitespace in Kconfig
v2: from Colin Cross
- fix invalid immediate in debug-macro.S
v3:
- allow selection of multiple boards
Signed-off-by: Colin Cross <ccross@android.com>
Signed-off-by: Erik Gilling <konkers@android.com>
This patch adds the Kconfig and Makefile for the new S5PV310 SoC.
It also updates arch/arm Kconfig, Makefile and arch/arm/mm/Kconfig
to include support for the new S5PV310.
Signed-off-by: Changhwan Youn <chaos.youn@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
* 'devel' of master.kernel.org:/home/rmk/linux-2.6-arm: (291 commits)
ARM: AMBA: Add pclk support to AMBA bus infrastructure
ARM: 6278/2: fix regression in RealView after the introduction of pclk
ARM: 6277/1: mach-shmobile: Allow users to select HZ, default to 128
ARM: 6276/1: mach-shmobile: remove duplicate NR_IRQS_LEGACY
ARM: 6246/1: mmci: support larger MMCIDATALENGTH register
ARM: 6245/1: mmci: enable hardware flow control on Ux500 variants
ARM: 6244/1: mmci: add variant data and default MCICLOCK support
ARM: 6243/1: mmci: pass power_mode to the translate_vdd callback
ARM: 6274/1: add global control registers definition header file for nuc900
mx2_camera: fix type of dma buffer virtual address pointer
mx2_camera: Add soc_camera support for i.MX25/i.MX27
arm/imx/gpio: add spinlock protection
ARM: Add support for the LPC32XX arch
ARM: LPC32XX: Arch config menu supoport and makefiles
ARM: LPC32XX: Phytec 3250 platform support
ARM: LPC32XX: Misc support functions
ARM: LPC32XX: Serial support code
ARM: LPC32XX: System suspend support
ARM: LPC32XX: GPIO, timer, and IRQ drivers
ARM: LPC32XX: Clock driver
...
smp_processor_id() must not be called from a preemptible context (this
is checked by CONFIG_DEBUG_PREEMPT). kmap_high_l1_vipt() was doing so.
This lead to a problem where the wrong per_cpu kmap_high_l1_vipt_depth
could be incremented, causing a BUG_ON(*depth <= 0); in
kunmap_high_l1_vipt().
The solution is to move the call to smp_processor_id() after the call
to preempt_disable().
Originally by: Andrew Howe <ahowe@nvidia.com>
Signed-off-by: Gary King <gking@nvidia.com>
Acked-by: Nicolas Pitre <nico.as.pitre@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This patch is in preparation for a subsequent patch which adds barriers
to the I/O accessors. Since the mandatory barriers may do an L2 cache
sync, this patch avoids a recursive call into l2x0_cache_sync() via the
write*() accessors and wmb() and a call into l2x0_cache_sync() with the
l2x0_lock held.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
All implementations of cpu_proc_fin() start by disabling interrupts
and then flush caches. Rather than have every processors proc_fin()
implementation do this, move it out into generic code - and move the
cache flush past setup_mm_for_reboot() (so it can benefit from having
caches still enabled.)
This allows cpu_proc_fin() to become independent of the L1/L2 cache
types, and eventually move the L2 cache flushing into the L2 support
code.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Statuses 3 (0b00011) and 6 (0x00110) of DFSR are Access Flags faults on
ARMv6K and ARMv7. Let's patch fsr_info[] at runtime if we are on ARMv7
or later.
Unfortunately, we don't have runtime check for 'K' extension, so we
can't check for it.
Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
On ARM one Linux PGD entry contains two hardware entries (see page
tables layout in pgtable.h). We normally guarantee that we always
fill both L1 entries. But create_mapping() doesn't follow the rule.
It can create inidividual L1 entries, so here we have to call
pmd_none() check in do_translation_fault() for the entry really
corresponded to address, not for the first of pair.
Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Add one more parameter to hook_fault_code() to be able to set 'code'
field of struct fsr_info.
Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
POSIX specify to use signal SIGBUS with code BUS_ADRALN for invalid
address alignment.
Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The DMA coherent remap area is used to provide an uncached mapping
of memory for coherency with DMA engines. Currently, we look for
any free hole which our allocation will fit in with page alignment.
However, this can lead to fragmentation of the area, and allows small
allocations to cross L1 entry boundaries. This is undesirable as we
want to move towards allocating sections of memory.
Align allocations according to the size, limiting the alignment between
the page and section sizes.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
We don't need our own implementation of this, use the generic
library implementation instead.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This changes the TCM handling so that a fixed area is reserved at
0xfffe0000-0xfffeffff for TCM. This areas is used by XScale but
XScale does not have TCM so the mechanisms are mutually exclusive.
This change is needed to make TCM detection more dynamic while
still being able to compile code into it, and is a must for the
unified ARM goals: the current TCM allocation at different places
in memory for each machine would be a nightmare if you want to
compile a single image for more than one machine with TCM so it
has to be nailed down in one place.
Signed-off-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
If TCM is in use, we should display it in the virtual memory
layout along with everything else.
Signed-off-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The earlier TCM memory regions were mapped as MT_MEMORY_UNCACHED
which doesn't really work on platforms supporting the new v6
features like the NX bit. Add unique MT_MEMORY_[I|D]TCM types
instead.
Cc: Nicolas Pitre <nico@fluxnic.net>
Signed-off-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Add a common early allocator function, in preparation for switching
over to LMB. When we do, this function will need to do a little more
than just allocating memory; we need it zero initialized too.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Move the platform specific bootmem memory reservations out of
arch/arm/mm/mmu.c into their respective platform files.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Everything should now be using sparsemem rather than discontigmem, so
remove the code supporting discontigmem from ARM.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Rather than storing the minimum size of the vmalloc area, store the
maximum permitted address of the vmalloc area instead.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The TLS register is only available on ARM1136 r1p0 and later.
Set HWCAP_TLS flags if hardware TLS is available and test for
it if CONFIG_CPU_32v6K is not set for V6.
Note that we set the TLS instruction in __kuser_get_tls
dynamically as suggested by Jamie Lokier <jamie@shareable.org>.
Also the __switch_to code is optimized out in most cases as
suggested by Nicolas Pitre <nico@fluxnic.net>.
Reviewed-by: Nicolas Pitre <nicolas.pitre@linaro.org>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
On i.MX35 the L2X0_AUX_CTRL register does not have sensible reset
default values. Allow them to be overwritten with the aux_val/aux_mask
arguments passed to l2x0_init().
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
RealView boards with certain revisions of the L210/L220 cache controller
may have issues (hardware deadlock) with the mandatory barriers (DSB
followed by an L2 cache sync) when ARM_DMA_MEM_BUFFERABLE is enabled.
The patch disables ARM_DMA_MEM_BUFFERABLE for these boards.
Tested-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Commit f4d6477f introduced a workaround for the lack of hardware
broadcasting of the cache maintenance operations on ARM11MPCore.
However, the workaround is only valid on CPUs that do not do speculative
loads into the D-cache.
This patch adds a Kconfig option with the corresponding help to make the
above clear. When the DMA_CACHE_RWFO option is disabled, the kernel
behaviour is that prior to the f4d6477f commit. This also allows ARMv6
UP processors with speculative loads to work correctly.
For other processors, a different workaround may be needed.
Cc: Ronen Shitrit <rshitrit@marvell.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
A recent patch for DMA cache maintenance on ARM11MPCore added a write
for ownership trick to the v6_dma_inv_range() function. Such operation
destroys data already present in the buffer. However, this function is
used with with dma_sync_single_for_device() which is supposed to
preserve the existing data transfered into the buffer. This patch adds a
combination of read/write for ownership to preserve the original data.
Reported-by: Ronen Shitrit <rshitrit@marvell.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This macro is not defined when !CONFIG_MMU so this patch moves the
CONSISTENT_* definitions to the CONFIG_MMU section.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
ARMv6 and above have a restriction whereby aliasing virtual:physical
mappings must not have differing memory type and sharability
attributes. Strictly, this covers the memory type (strongly ordered,
device, memory), cache attributes (uncached, write combine, write
through, write back read alloc, write back write alloc) and the
shared bit.
However, using ioremap() and its variants on system RAM results in
mappings which differ in these attributes from the main system RAM
mapping. Other architectures which similar restrictions approch this
problem in the same way - they do not permit ioremap on main system
RAM.
Make ARM behave in the same way, with a WARN_ON() such that users can
be traced and an alternative approach found.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The VM subsystem assumes that there are valid memmap entries to
the bank end aligned to MAX_ORDER_NR_PAGES. It will try and read
these page structs, and so we cannot free any memmap entries that
it may inspect.
Signed-off-by: Michael Bohan <mbohan@codeaurora.org>
Signed-off-by: Daniel Walker <dwalker@codeaurora.org>
When functions incoming parameters are not in input operands list gcc
4.5 does not load the parameters into registers before calling this
function but the inline assembly assumes valid addresses inside this
function. This breaks the code because r0 and r1 are invalid when
execution enters v4wb_copy_user_page ()
Also the constant needs to be used as third input operand so account
for that as well.
Tested on qemu arm.
CC: <stable@kernel.org>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Instruction faults on pre-ARMv6 CPUs are interpreted as
a 'translation fault', but do_translation_fault doesn't
handle well if user mode trying to run instruction above
TASK_SIZE, and result in the infinite retry of that
instruction.
CC: <stable@kernel.org>
Signed-off-by: Anfei Zhou <anfei.zhou@gmail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
When CONFIG_DEBUG_HIGHMEM is used, the fixmap entry used for a highmem page
by kmap_atomic() is always cleared by kunmap_atomic(). This helps find
bad usages such as dereferences after the unmap, or overflow into the
adjacent fixmap areas.
But this debugging aid is completely bypassed when a kmap for the same
page already exists as the kmap is reused instead. ON VIVT systems we
have no choice but to reuse that kmap due to cache coherency issues,
but on non VIVT systems we should always force the fixmap usage when
debugging is active.
Signed-off-by: Nicolas Pitre <nicolas.pitre@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
