efikamx:linux-kernel.git
7 years agosmartbook: fix GPU resource allocation to match other boards
Matt Sealey [Sat, 13 Nov 2010 16:00:05 +0000 (10:00 -0600)]
smartbook: fix GPU resource allocation to match other boards

7 years agoreport energy and charge full design
Saleem Abdulrasool [Fri, 12 Nov 2010 21:14:54 +0000 (13:14 -0800)]
report energy and charge full design

7 years agoreport energy and charge full
Saleem Abdulrasool [Fri, 12 Nov 2010 21:11:21 +0000 (13:11 -0800)]
report energy and charge full

7 years agoreport energy and charge now
Saleem Abdulrasool [Fri, 12 Nov 2010 21:07:37 +0000 (13:07 -0800)]
report energy and charge now

7 years agoadd efikasb_batt_write
Saleem Abdulrasool [Fri, 12 Nov 2010 20:56:30 +0000 (12:56 -0800)]
add efikasb_batt_write

7 years agoremove unnecessary goto usage in efikasb_batt_read
Saleem Abdulrasool [Fri, 12 Nov 2010 20:53:09 +0000 (12:53 -0800)]
remove unnecessary goto usage in efikasb_batt_read

7 years agofix whitespace
Saleem Abdulrasool [Fri, 12 Nov 2010 20:47:39 +0000 (12:47 -0800)]
fix whitespace

7 years agorename hex_to_bin to get compat-wireless working
Marcin Juszkiewicz [Tue, 9 Nov 2010 14:39:32 +0000 (15:39 +0100)]
rename hex_to_bin to get compat-wireless working

7 years agosyslog: distinguish between /proc/kmsg and syscalls
Kees Cook [Tue, 2 Feb 2010 05:21:17 +0000 (21:21 -0800)]
syslog: distinguish between /proc/kmsg and syscalls

This allows the LSM to distinguish between syslog functions originating
from /proc/kmsg access and direct syscalls.  By default, the commoncaps
will now no longer require CAP_SYS_ADMIN to read an opened /proc/kmsg
file descriptor.  For example the kernel syslog reader can now drop
privileges after opening /proc/kmsg, instead of staying privileged with
CAP_SYS_ADMIN.  MAC systems that implement security_syslog have unchanged
behavior.

Signed-off-by: Kees Cook <kees.cook@canonical.com>
7 years agoFix display offset bug on the AUO panels. Might be better in the future to pull and...
Matt Sealey [Thu, 4 Nov 2010 16:47:08 +0000 (11:47 -0500)]
Fix display offset bug on the AUO panels. Might be better in the future to pull and patch EDID.

7 years agobump to .3
Matt Sealey [Thu, 28 Oct 2010 09:19:52 +0000 (04:19 -0500)]
bump to .3

7 years agoARM 5897/1: spinlock: don't use deprecated barriers on ARMv7
Matt Sealey [Thu, 28 Oct 2010 07:05:38 +0000 (02:05 -0500)]
ARM 5897/1: spinlock: don't use deprecated barriers on ARMv7

7 years agoARM 5849/1 ARMv7: fix Oprofile events count
Matt Sealey [Thu, 28 Oct 2010 06:44:19 +0000 (01:44 -0500)]
ARM 5849/1 ARMv7: fix Oprofile events count

7 years agoARM 5889/1: Add atomic64 routines for ARMv6k and above.
Matt Sealey [Thu, 28 Oct 2010 07:02:14 +0000 (02:02 -0500)]
ARM 5889/1: Add atomic64 routines for ARMv6k and above.

7 years agoARM 5997/1: Correct the VFPv3 detection
Matt Sealey [Thu, 28 Oct 2010 06:55:52 +0000 (01:55 -0500)]
ARM 5997/1: Correct the VFPv3 detection

7 years agoARM 5977/1: Enable backtrace printing on oops when PC is corrupted
Matt Sealey [Thu, 28 Oct 2010 06:53:57 +0000 (01:53 -0500)]
ARM 5977/1: Enable backtrace printing on oops when PC is corrupted

7 years agoARM 5868/1: ARM: fix "BUG: using smp_processor_id() in preemptible code"
Matt Sealey [Thu, 28 Oct 2010 06:48:00 +0000 (01:48 -0500)]
ARM 5868/1: ARM: fix "BUG: using smp_processor_id() in preemptible code"

7 years agoARM 5794/1 Flush the D-cache during copy_user_highpage()
Matt Sealey [Thu, 28 Oct 2010 06:42:05 +0000 (01:42 -0500)]
ARM 5794/1 Flush the D-cache during copy_user_highpage()

7 years agoARM 5784/1 fix early boot machine ID mismatch error display
Matt Sealey [Thu, 28 Oct 2010 06:40:22 +0000 (01:40 -0500)]
ARM 5784/1 fix early boot machine ID mismatch error display

7 years agoFix innocuous warning in ureadahead support
Matt Sealey [Thu, 28 Oct 2010 07:22:28 +0000 (02:22 -0500)]
Fix innocuous warning in ureadahead support

7 years agoipu: make disable/unitialized "error" into "info" since it doesn't mean anything...
Matt Sealey [Thu, 28 Oct 2010 07:43:38 +0000 (02:43 -0500)]
ipu: make disable/unitialized "error" into "info" since it doesn't mean anything went wrong

7 years agopata_fsl: make the dma rchan/wchan message in the kernel INFO so it doesn't get print...
Matt Sealey [Thu, 28 Oct 2010 07:41:27 +0000 (02:41 -0500)]
pata_fsl: make the dma rchan/wchan message in the kernel INFO so it doesn't get printed when we want to be quiet

7 years agopata_fsl: remove that annoying frame size error by allocating scatterlists with kzall...
Matt Sealey [Thu, 28 Oct 2010 07:54:49 +0000 (02:54 -0500)]
pata_fsl: remove that annoying frame size error by allocating scatterlists with kzalloc instead of defining tmp[64]

Results: no more warning
Impact: no performance hit (PATA is too slow on these devices for it to matter).

7 years agouio: fix config
Matt Sealey [Thu, 21 Oct 2010 21:26:44 +0000 (16:26 -0500)]
uio: fix config

7 years agoRevert "config: uio fix"
Matt Sealey [Thu, 21 Oct 2010 21:25:43 +0000 (16:25 -0500)]
Revert "config: uio fix"

This reverts commit 23a402db3fb0ce1020221ebdefd43ed90bc5c24c.

7 years agoconfig: uio fix
Matt Sealey [Thu, 21 Oct 2010 21:22:37 +0000 (16:22 -0500)]
config: uio fix

7 years agouio: remove from configs since Genesi's preferred method for VG GPU access is via...
Matt Sealey [Thu, 21 Oct 2010 20:21:00 +0000 (15:21 -0500)]
uio: remove from configs since Genesi's preferred method for VG GPU access is via the AMD GPU module

7 years agoversion bump to 14.2 encompassing stability fixes
Matt Sealey [Thu, 21 Oct 2010 18:22:44 +0000 (13:22 -0500)]
version bump to 14.2 encompassing stability fixes

7 years agomatching EfikaMX support for ENGR00126861-1 csi clock change
Matt Sealey [Thu, 21 Oct 2010 17:54:08 +0000 (12:54 -0500)]
matching EfikaMX support for ENGR00126861-1 csi clock change

7 years agoENGR00126156 ipuv3: check own clk count only when around clk rate
Jason Chen [Mon, 16 Aug 2010 08:25:41 +0000 (16:25 +0800)]
ENGR00126156 ipuv3: check own clk count only when around clk rate

If dual display enabled, the second one may not be able to set
correct clock rate if it's from ext clk.

Signed-off-by: Jason Chen <b02280@freescale.com>
7 years agoENGR00126861-2: ipu: pass csi clock in ipu platform data
Dinh Nguyen [Mon, 30 Aug 2010 22:27:14 +0000 (17:27 -0500)]
ENGR00126861-2: ipu: pass csi clock in ipu platform data

ipu: pass csi clock in platform data
ipu: use ipu hw rev instead of chip version

Acked-by: Rob Herring <r.herring@freescale.com>
Signed-off-by: Dinh Nguyen <Dinh.Nguyen@freescale.com>
7 years agoENGR00126861-1: mx51: pass csi clock in ipu platform data
Dinh Nguyen [Mon, 30 Aug 2010 22:27:14 +0000 (17:27 -0500)]
ENGR00126861-1: mx51: pass csi clock in ipu platform data

ipu: pass csi clock in platform data

Acked-by: Rob Herring <r.herring@freescale.com>
Signed-off-by: Dinh Nguyen <Dinh.Nguyen@freescale.com>
7 years agoENGR00125796 MX5x: Boot stop at "Starting kernel ..."
Jason Liu [Tue, 10 Aug 2010 02:44:00 +0000 (10:44 +0800)]
ENGR00125796 MX5x: Boot stop at "Starting kernel ..."

System sometime stop at "Starting kernel ...". After connect with
ICE, we found that system enter abort mode. The abort sometimes is
data abort and sometimes prefetch abort.This issue may caused by that
we config the PLL1 while CPU clock is runing on it.In fact, we don't
need do PLL1 enable operation if it already run up and which will
bring side-effect for the impact of cpu clock.

Signed-off-by:Jason Liu <r64343@freescale.com>

7 years agogenirq: Run irq handlers with interrupts disabled
Ingo Molnar [Fri, 26 Mar 2010 00:06:51 +0000 (00:06 +0000)]
genirq: Run irq handlers with interrupts disabled

Running interrupt handlers with interrupts enabled can cause stack
overflows. That has been observed with multiqueue NICs delivering all
their interrupts to a single core. We might band aid that somehow by
checking the interrupt stacks, but the real safe fix is to run the irq
handlers with interrupts disabled.

Drivers for whacky hardware still can reenable them in the handler
itself, if the need arises. (They do already due to lockdep)

The risk of doing this is rather low:

 - lockdep already enforces this
 - CONFIG_NOHZ has shaken out the drivers which relied on jiffies updates
 - time keeping is not longer sensitive to the timer interrupt being delayed

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Miller <davem@davemloft.net>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Linus Torvalds <torvalds@osdl.org>
LKML-Reference: <20100326000405.758579387@linutronix.de>

7 years agoENGR00125800 UIO: clean uioinfo when uninstall uio driver
Jie Zhou [Wed, 4 Aug 2010 03:26:39 +0000 (11:26 +0800)]
ENGR00125800 UIO: clean uioinfo when uninstall uio driver

The uioinfo should be cleaned up when uninstall, otherwise re-install
failure of uio_pdrv_genirq.ko will happen.

Signed-off-by: Jie Zhou <b30303@freescale.com>
7 years agotrace: ureadahead support
Matt Sealey [Wed, 20 Oct 2010 05:40:00 +0000 (00:40 -0500)]
trace: ureadahead support

7 years agoversion bump (2.6.31.14.1) so that package conflicts don't happen
Matt Sealey [Wed, 20 Oct 2010 00:48:30 +0000 (19:48 -0500)]
version bump (2.6.31.14.1) so that package conflicts don't happen

7 years agoconfigs: disable PREEMPT as it may be causing horrible problems
Matt Sealey [Tue, 19 Oct 2010 14:56:03 +0000 (09:56 -0500)]
configs: disable PREEMPT as it may be causing horrible problems

7 years agoFrom 148a2911a31f77f18778519bf300143f850fddaf Mon Sep 17 00:00:00 2001
Matt Sealey [Mon, 18 Oct 2010 14:05:37 +0000 (09:05 -0500)]
From 148a2911a31f77f18778519bf300143f850fddaf Mon Sep 17 00:00:00 2001
From: Steev Klimaszewski <steev@gentoo.org>
Date: Mon, 18 Oct 2010 08:40:53 -0500
Subject: [PATCH] CONFIG_PPP_MULTILINK and CONFIG_PPP_FILTER cannot be modular, since
 we want the support, build them in with y

7 years agoconfig: Add PPP modules
Matt Sealey [Sat, 16 Oct 2010 19:43:58 +0000 (14:43 -0500)]
config: Add PPP modules

7 years agoFixed logical fallacy re SD and WP pins for SDHC2 therefore finally fixing the SD...
Matt Sealey [Sat, 16 Oct 2010 18:18:19 +0000 (13:18 -0500)]
Fixed logical fallacy re SD and WP pins for SDHC2 therefore finally fixing the SD card bug.

7 years agousb: Increase timeout value for device reset
Dinh Nguyen [Tue, 13 Apr 2010 16:13:15 +0000 (11:13 -0500)]
usb: Increase timeout value for device reset

It seems that for USB IP on Freescale MX5x processors, it needs >750
usec for the reset to complete. This change should not hurt any other
EHCI hardware.

Signed-off-by: Dinh Nguyen <Dinh.Nguyen@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoMore moronic SD card stuff fixed. stuff CD into interrupt for TO2 instead of WP
Matt Sealey [Tue, 28 Sep 2010 19:00:41 +0000 (14:00 -0500)]
More moronic SD card stuff fixed. stuff CD into interrupt for TO2 instead of WP

7 years agoSmarttop: potential fix for TO2 SD card CD/WP handling. Neko is a moron.
Matt Sealey [Tue, 28 Sep 2010 18:46:00 +0000 (13:46 -0500)]
Smarttop: potential fix for TO2 SD card CD/WP handling. Neko is a moron.

7 years agorootwait by default
Matt Sealey [Tue, 28 Sep 2010 17:40:54 +0000 (12:40 -0500)]
rootwait by default

7 years agoSmarttop: remove modes which do not match IPU restrictions (vertical border needs... efikasb-10.07.11-20100928
Matt Sealey [Tue, 28 Sep 2010 17:15:55 +0000 (12:15 -0500)]
Smarttop: remove modes which do not match IPU restrictions (vertical border needs to be >2) in order not to have them be "reconfigured" later to be potentially incompatible with the monitor.

7 years agodefconfig update, IO scheduler NOOP, DEVTMPFS config for SB, CPU govs for MX
Matt Sealey [Wed, 15 Sep 2010 02:36:16 +0000 (21:36 -0500)]
defconfig update, IO scheduler NOOP, DEVTMPFS config for SB, CPU govs for MX

Why NOOP? Because CFQ is wasting CPU cycles for no performance gain. There should
be more left for the system now, and identical block device performance.

Devtmpfs in config enabled for Lucid/Maverick and other 2010 Linux support.

Added modules for CPU governors for the Smarttop board since power management is
useful.

7 years agoRemove custom Efika Smartbook backlight driver as it duplicates code
Matt Sealey [Fri, 10 Sep 2010 23:31:34 +0000 (18:31 -0500)]
Remove custom Efika Smartbook backlight driver as it duplicates code
and isn't nearly as good as the generic one now in use.

7 years agoUse the fine-grained, generic PWM platform driver for backlight.
Matt Sealey [Fri, 10 Sep 2010 23:29:25 +0000 (18:29 -0500)]
Use the fine-grained, generic PWM platform driver for backlight.

Update PWM driver to support the notifier from the MTL017 driver
such that when the LVDS is off, the backlight turns off too.

7 years agodefconfigs: add ramzswap module, enable oprofile module
Matt Sealey [Fri, 10 Sep 2010 18:10:08 +0000 (13:10 -0500)]
defconfigs: add ramzswap module, enable oprofile module

7 years agoramzswap: Issue #50, commit 6f52562b30, Parameters for module start
Matt Sealey [Fri, 10 Sep 2010 17:39:26 +0000 (12:39 -0500)]
ramzswap: Issue #50, commit 6f52562b30, Parameters for module start

7 years agoramzswap: merge fubar fixed.
Matt Sealey [Fri, 10 Sep 2010 16:55:35 +0000 (11:55 -0500)]
ramzswap: merge fubar fixed.

7 years agoStaging: xvmalloc memory allocator
Nitin Gupta [Tue, 22 Sep 2009 04:56:52 +0000 (10:26 +0530)]
Staging: xvmalloc memory allocator

* Features:
 - Low metadata overhead (just 4 bytes per object)
 - O(1) Alloc/Free - except when we have to call system page allocator to
   get additional memory.
 - Very low fragmentation: In all tests, xvmalloc memory usage is within 12%
   of "Ideal".
 - Pool based allocator: Each pool can grow and shrink.
 - It maps pages only when required. So, it does not hog vmalloc area which
   is very small on 32-bit systems.

SLUB allocator could not be used due to fragmentation issues:
http://code.google.com/p/compcache/wiki/AllocatorsComparison
Data here shows kmalloc using ~43% more memory than TLSF and xvMalloc
is showed ~2% more space efficiency than TLSF (due to smaller metadata).
Creating various kmem_caches can reduce space efficiency gap but still
problem of being limited to low memory exists. Also, it depends on
allocating higher order pages to reduce fragmentation - this is not
acceptable for ramzswap as it is used under memory crunch (its a swap
device!).

SLOB allocator could not be used do to reasons mentioned here:
http://lkml.org/lkml/2009/3/18/210

* Implementation:
It uses two-level bitmap search to find free list containing block of
correct size. This idea is taken from TLSF (Two-Level Segregate Fit)
allocator and is well explained in its paper (see [Links] below).

* Limitations:
 - Poor scalability: No per-cpu data structures (work in progress).

[Links]
1. Details and Performance data:
http://code.google.com/p/compcache/wiki/xvMalloc
http://code.google.com/p/compcache/wiki/xvMallocPerformance

2. TLSF memory allocator:
home: http://rtportal.upv.es/rtmalloc/
paper: http://rtportal.upv.es/rtmalloc/files/MRBC_2008.pdf

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: xvmalloc memory allocator
Nitin Gupta [Tue, 22 Sep 2009 04:56:52 +0000 (10:26 +0530)]
Staging: xvmalloc memory allocator

* Features:
 - Low metadata overhead (just 4 bytes per object)
 - O(1) Alloc/Free - except when we have to call system page allocator to
   get additional memory.
 - Very low fragmentation: In all tests, xvmalloc memory usage is within 12%
   of "Ideal".
 - Pool based allocator: Each pool can grow and shrink.
 - It maps pages only when required. So, it does not hog vmalloc area which
   is very small on 32-bit systems.

SLUB allocator could not be used due to fragmentation issues:
http://code.google.com/p/compcache/wiki/AllocatorsComparison
Data here shows kmalloc using ~43% more memory than TLSF and xvMalloc
is showed ~2% more space efficiency than TLSF (due to smaller metadata).
Creating various kmem_caches can reduce space efficiency gap but still
problem of being limited to low memory exists. Also, it depends on
allocating higher order pages to reduce fragmentation - this is not
acceptable for ramzswap as it is used under memory crunch (its a swap
device!).

SLOB allocator could not be used do to reasons mentioned here:
http://lkml.org/lkml/2009/3/18/210

* Implementation:
It uses two-level bitmap search to find free list containing block of
correct size. This idea is taken from TLSF (Two-Level Segregate Fit)
allocator and is well explained in its paper (see [Links] below).

* Limitations:
 - Poor scalability: No per-cpu data structures (work in progress).

[Links]
1. Details and Performance data:
http://code.google.com/p/compcache/wiki/xvMalloc
http://code.google.com/p/compcache/wiki/xvMallocPerformance

2. TLSF memory allocator:
home: http://rtportal.upv.es/rtmalloc/
paper: http://rtportal.upv.es/rtmalloc/files/MRBC_2008.pdf

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoauthor Tejun Heo <tj@kernel.org>
Matt Sealey [Fri, 10 Sep 2010 16:06:05 +0000 (11:06 -0500)]
author Tejun Heo <tj@kernel.org>
Wed, 24 Mar 2010 08:04:11 +0000 (17:04 +0900)

include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h

Picked out manually from 5a0e3ad6af8660be21ca98a971cd00f331318c05 so we don't break further ramzswap updates when cherry-picking

7 years agoStaging: ramzswap: Update copyright notice
Nitin Gupta [Thu, 28 Jan 2010 15:51:35 +0000 (21:21 +0530)]
Staging: ramzswap: Update copyright notice

Update copyright notice.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: Handle case for invalid backing swap
Nitin Gupta [Thu, 28 Jan 2010 15:49:59 +0000 (21:19 +0530)]
Staging: ramzswap: Handle case for invalid backing swap

Currently, we crash (issue BUG_ON) if backing swap
disk size is zero. This can happen is user specified
an extended partition or simply a bad disk as backing
swap. A crash is really an unpleasant surprise to user
for such trivial problems.

Now, we check for this condition and simply fail device
initialization if this is the case.

Additional cleanups:
 * use static for all functions
 * remove extra newline between functions
 * memset backing_swap_name to NULL on device reset

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: Remove redundant check for zero page
Nitin Gupta [Thu, 28 Jan 2010 15:43:41 +0000 (21:13 +0530)]
Staging: ramzswap: Remove redundant check for zero page

ramzswap_free_page() already handles the case for zero filled
pages. So, remove redundant logic for the same in ramzswap_write().

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: Return proper error code on device create failure
Nitin Gupta [Thu, 28 Jan 2010 15:43:40 +0000 (21:13 +0530)]
Staging: ramzswap: Return proper error code on device create failure

Currently, we return 0 if create_device() fails and 1 otherwise.
Now, proper error code is returned from create_device() and the
same is propagated as module error code from ramzswap_init().

Also added some cleanups for ramzswap_init(), improving function
structure.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: Set block size to PAGE_SIZE
Nitin Gupta [Thu, 28 Jan 2010 15:43:39 +0000 (21:13 +0530)]
Staging: ramzswap: Set block size to PAGE_SIZE

ramzswap block size needs to be set equal to PAGE_SIZE to
avoid receiving any unaligned block I/O requests (happens
due to readahead logic during swapon). These unaligned
accesses produce unnecessary I/O errors, scaring users.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: Flush block device before reset
Nitin Gupta [Thu, 28 Jan 2010 15:43:38 +0000 (21:13 +0530)]
Staging: ramzswap: Flush block device before reset

Make sure we flush block device before freeing all metadata
during reset ioctl.

Signed-off-by: Nitin Gupta <ngupta@vflar.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: Use lock for 64-bit stats
Nitin Gupta [Thu, 28 Jan 2010 15:43:37 +0000 (21:13 +0530)]
Staging: ramzswap: Use lock for 64-bit stats

64-bit stats corruption was observed when ramzswap was
used on SMP systems. To prevent this, use separate spinlock
to protect these stats.

Also, replace stat_*() with rzs_stat*() to avoid possible
conflict with core kernel code.

Eventually, these will be converted to per-cpu counters
if this driver finds use on large scale systems and this
locking is found to affect scalability.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: Free memory when create_device is failed
Minchan Kim [Mon, 11 Jan 2010 07:15:53 +0000 (16:15 +0900)]
Staging: ramzswap: Free memory when create_device is failed

If create_device is failed, it can't free gendisk and request_queue of
preceding devices. It cause memory leak.

This patch fixes it.

Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: Minor spelling fixes
C yp [Wed, 6 Jan 2010 12:42:00 +0000 (13:42 +0100)]
Staging: ramzswap: Minor spelling fixes

Also removed an extra semicolon.

Signed-off-by: Cyp <cyp561@gmail.com>
Cc: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: remove unused #include <linux/version.h>
Huang Weiyi [Sun, 13 Dec 2009 14:06:25 +0000 (22:06 +0800)]
Staging: ramzswap: remove unused #include <linux/version.h>

Remove unused #include <linux/version.h>('s) in
  drivers/staging/ramzswap/ramzswap_drv.c

Signed-off-by: Huang Weiyi <weiyi.huang@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: ramzswap: remove ARM specific d-cache hack
Nitin Gupta [Sat, 12 Dec 2009 06:14:46 +0000 (11:44 +0530)]
Staging: ramzswap: remove ARM specific d-cache hack

Remove d-cache hack in ramzswap driver that was needed
to workaround a bug in ARM version of update_mmu_cache()
which caused stale data in d-cache to be transferred to
userspace. This bug was fixed by git commit:
787b2faadc4356b6c2c71feb42fb944fece9a12f
This also brings down one entry in TODO file.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Acked-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoStaging: virtual block device driver (ramzswap)
Nitin Gupta [Tue, 22 Sep 2009 04:56:53 +0000 (10:26 +0530)]
Staging: virtual block device driver (ramzswap)

Creates RAM based block devices (/dev/ramzswapX) which can be
used (only) as swap disks. Pages swapped to these are compressed
and stored in memory itself.

The module is called ramzswap.ko. It depends on:
 - xvmalloc memory allocator (compiled with this driver)
 - lzo_compress.ko
 - lzo_decompress.ko

See ramzswap.txt for usage details.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoRevert "ramzswap driver"
Matt Sealey [Fri, 10 Sep 2010 15:50:50 +0000 (10:50 -0500)]
Revert "ramzswap driver"

This reverts commit b675d5e9dd1193538655257ab90fb458ca4de6b8.

7 years agoTake out TVE and W1 data since it's unused now
Matt Sealey [Fri, 10 Sep 2010 15:44:05 +0000 (10:44 -0500)]
Take out TVE and W1 data since it's unused now

7 years agoramzswap driver
Matt Sealey [Fri, 10 Sep 2010 15:41:51 +0000 (10:41 -0500)]
ramzswap driver

7 years agoRevert "swap: Add flag to identify block swap devices"
Matt Sealey [Fri, 10 Sep 2010 15:40:32 +0000 (10:40 -0500)]
Revert "swap: Add flag to identify block swap devices"

This reverts commit f2063595426101ca556e6ad9c711e2336557ce7d.

7 years agoswap: Add flag to identify block swap devices
Nitin Gupta [Mon, 17 May 2010 05:32:42 +0000 (11:02 +0530)]
swap: Add flag to identify block swap devices

Added SWP_BLKDEV flag to distinguish block and regular file backed
swap devices. We could also check if a swap is entire block device,
rather than a file, by:
S_ISBLK(swap_info_struct->swap_file->f_mapping->host->i_mode)
but, I think, simply checking this flag is more convenient.

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Nigel Cunningham <nigel@tuxonice.net>
Acked-by: Pekka Enberg <penberg@cs.helsinki.fi>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
7 years agoSmartbook: don't register TVE, W1 or FEC since they don't exist
Matt Sealey [Thu, 9 Sep 2010 21:35:28 +0000 (16:35 -0500)]
Smartbook: don't register TVE, W1 or FEC since they don't exist

7 years agoLid Switch: Name fix, whitespace fixes
Matt Sealey [Thu, 9 Sep 2010 21:28:25 +0000 (16:28 -0500)]
Lid Switch: Name fix, whitespace fixes

7 years agoVDDD regulator for audio to quit complaining
Matt Sealey [Thu, 9 Sep 2010 21:28:05 +0000 (16:28 -0500)]
VDDD regulator for audio to quit complaining

fixes further sgtl5000 complaints on boot

7 years agoUse i2c-imx instead of mxc_i2c for i2c bus traffic
Matt Sealey [Thu, 9 Sep 2010 21:27:00 +0000 (16:27 -0500)]
Use i2c-imx instead of mxc_i2c for i2c bus traffic

fixes mtl017 complaints on boot (and speeds boot up slightly)
fixes sgtl5000 complaints on boot
fixes battery i2c complaints (and oops) on boot

7 years agomtl017 update: edid dump and screen data
Matt Sealey [Thu, 9 Sep 2010 21:01:10 +0000 (16:01 -0500)]
mtl017 update: edid dump and screen data

Update EDID dump to match SII9022 (rows of 32 bytes instead of one per line)

DEBUG_MTL017 controls this dump data.

whitespace fixes abound.

7 years agoSmartbook support for MXCFB_DEFAULT_BPP although 32-bit still seems banded
Matt Sealey [Thu, 9 Sep 2010 19:14:52 +0000 (14:14 -0500)]
Smartbook support for MXCFB_DEFAULT_BPP although 32-bit still seems banded

7 years agoUnbreak defconfig (argh)
Matt Sealey [Thu, 9 Sep 2010 18:47:25 +0000 (13:47 -0500)]
Unbreak defconfig (argh)

7 years agoARM: Fix wrong register in proc-arm6_7.S data abort handler
Russell King [Wed, 3 Feb 2010 15:48:03 +0000 (15:48 +0000)]
ARM: Fix wrong register in proc-arm6_7.S data abort handler

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
7 years agoUpdate defconfig, seemed to have lost some stuff
Matt Sealey [Thu, 9 Sep 2010 18:02:20 +0000 (13:02 -0500)]
Update defconfig, seemed to have lost some stuff

7 years agoVPU allocation fails for some reason - actually the reason is that fbi->fix.line_leng...
Matt Sealey [Thu, 9 Sep 2010 17:41:12 +0000 (12:41 -0500)]
VPU allocation fails for some reason - actually the reason is that fbi->fix.line_length can be zero. In this situation, DON'T try and dma_alloc_writecombine with 0 length, or it'll OOPS.

As a side effect of not doing this, the V4L driver would happily go off and touch a framebuffer address that was invalid, and there may have been a slight SNAFU with a divide by zero on the VPU framebuffer.

Arnaud Patard from Mandriva noticed these a long while ago, but now here they are.

BUG: these are temporary hacks, let's try and work out why line_length is 0..

7 years agoRatonalize and cleanup the 16/32bit mode define. it's in mx51_efikamx.h now. 32bit...
Matt Sealey [Thu, 9 Sep 2010 17:18:24 +0000 (12:18 -0500)]
Ratonalize and cleanup the 16/32bit mode define. it's in mx51_efikamx.h now. 32bit works.

7 years agoMake RT2870 driver less chatty (# messages) and disable debug by default
Matt Sealey [Thu, 9 Sep 2010 15:55:25 +0000 (10:55 -0500)]
Make RT2870 driver less chatty (# messages) and disable debug by default

7 years agopixclk_limit now a hardcoded define (117MHz) and no more 1080p hack (dangerous anyway!)
Matt Sealey [Thu, 9 Sep 2010 15:54:35 +0000 (10:54 -0500)]
pixclk_limit now a hardcoded define (117MHz) and no more 1080p hack (dangerous anyway!)

7 years agoRoll back to 16-bit displays
Matt Sealey [Thu, 9 Sep 2010 15:13:19 +0000 (10:13 -0500)]
Roll back to 16-bit displays

7 years agodefault plat data mode needs to be 16bit otherwise the initial framebuffer cannot...
Matt Sealey [Wed, 8 Sep 2010 17:24:15 +0000 (12:24 -0500)]
default plat data mode needs to be 16bit otherwise the initial framebuffer cannot be allocated (this is very strange, but, it will be changed later anyway). 32-bit framebuffer now is set on boot, but colours are warped.

7 years agoroll back serial snafu to BSP version
Matt Sealey [Wed, 8 Sep 2010 16:50:27 +0000 (11:50 -0500)]
roll back serial snafu to BSP version

7 years agodefconfig update: CONFIG_PREEMPT & CONFIG_UACCESS_WITH_MEMCPY for best speed
Matt Sealey [Wed, 8 Sep 2010 15:06:35 +0000 (10:06 -0500)]
defconfig update: CONFIG_PREEMPT & CONFIG_UACCESS_WITH_MEMCPY for best speed

7 years ago32-bit display fixes to stay consistent
Matt Sealey [Wed, 8 Sep 2010 14:36:10 +0000 (09:36 -0500)]
32-bit display fixes to stay consistent

7 years agodefconfig: enable CONFIG_SECURITY_FILE_CAPABILITIES for ChromiumOS (sync BSP)
Matt Sealey [Wed, 8 Sep 2010 13:07:00 +0000 (08:07 -0500)]
defconfig: enable CONFIG_SECURITY_FILE_CAPABILITIES for ChromiumOS (sync BSP)

7 years agoMake default bits per pixel 32-bit
Matt Sealey [Wed, 8 Sep 2010 13:03:22 +0000 (08:03 -0500)]
Make default bits per pixel 32-bit

7 years agoSync defconfigs to BSP: DMA_ZONE=96, VMSPLIT 2G/2G
Matt Sealey [Wed, 8 Sep 2010 12:59:56 +0000 (07:59 -0500)]
Sync defconfigs to BSP: DMA_ZONE=96, VMSPLIT 2G/2G

7 years agoboard_is -> machine_is typo fixed
Matt Sealey [Tue, 7 Sep 2010 20:10:27 +0000 (15:10 -0500)]
board_is -> machine_is typo fixed

7 years agoBacklight: clamp to on or off because the PWM config causes flickering.
Matt Sealey [Tue, 7 Sep 2010 19:49:22 +0000 (14:49 -0500)]
Backlight: clamp to on or off because the PWM config causes flickering.

7 years agodefconfig: kernel uaccess with memcpy enable, preempt enable
Matt Sealey [Tue, 7 Sep 2010 19:46:39 +0000 (14:46 -0500)]
defconfig: kernel uaccess with memcpy enable, preempt enable

7 years agoPatch from Catalin Marinas @ ARM to fix ARM6K depending on CPU_V6 but not CPU_V7
Matt Sealey [Tue, 7 Sep 2010 15:58:59 +0000 (10:58 -0500)]
Patch from Catalin Marinas @ ARM to fix ARM6K depending on CPU_V6 but not CPU_V7

7 years agoUpdate defconfig for Smartbook to match Smarttop
Matt Sealey [Tue, 7 Sep 2010 15:54:55 +0000 (10:54 -0500)]
Update defconfig for Smartbook to match Smarttop

7 years agoEfikaMX Smartbook: LCD backlight naming fix. Don't do Smarttop things on Smartbook...
Matt Sealey [Tue, 7 Sep 2010 15:54:17 +0000 (10:54 -0500)]
EfikaMX Smartbook: LCD backlight naming fix. Don't do Smarttop things on Smartbook in FB.

7 years agoEfika MX Smartbook: Fix battery naming
Matt Sealey [Tue, 7 Sep 2010 15:53:56 +0000 (10:53 -0500)]
Efika MX Smartbook: Fix battery naming