platform/kernel/linux-rpi.git
16 years agoarm: bus_id -> dev_name() and dev_set_name() conversions
Kay Sievers [Fri, 30 May 2008 15:42:11 +0000 (17:42 +0200)]
arm: bus_id -> dev_name() and dev_set_name() conversions

Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agosparc64: fix up bus_id changes in sparc core code
Greg Kroah-Hartman [Fri, 2 May 2008 04:02:41 +0000 (06:02 +0200)]
sparc64: fix up bus_id changes in sparc core code

This converts all instances of bus_id in the sparc core kernel to use
either dev_set_name(), or dev_name() depending on the need.

This is done in anticipation of removing the bus_id field from struct
driver.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years ago3c59x: handle pci_name() being const
Greg Kroah-Hartman [Wed, 2 Jul 2008 19:46:22 +0000 (12:46 -0700)]
3c59x: handle pci_name() being const

This changes vortex_prob1() to handle pci_name() now returning a
constant string.

Cc: Steffen Klassert <klassert@mathematik.tu-chemnitz.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoMTD: handle pci_name() being const
Greg Kroah-Hartman [Wed, 2 Jul 2008 19:46:22 +0000 (12:46 -0700)]
MTD: handle pci_name() being const

This changes the MTD core to handle pci_name() now returning a constant
string.

Cc: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoHP iLO driver
David Altobelli [Wed, 2 Jul 2008 15:38:53 +0000 (09:38 -0600)]
HP iLO driver

A driver for the HP iLO/iLO2 management processor, which allows userspace
programs to query the management processor.  Programs can open a channel
to the device (/dev/hpilo/dXccbN), and use this to send/receive queries.
The O_EXCL open flag is used to indicate that a particular channel cannot
be shared between processes.  This driver will replace various packages
HP has shipped, including hprsm and hp-ilo.

Signed-off-by: David Altobelli <david.altobelli@hp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agosysdev: Convert the x86 mce tolerant sysdev attribute to generic attribute
Andi Kleen [Tue, 1 Jul 2008 16:48:43 +0000 (18:48 +0200)]
sysdev: Convert the x86 mce tolerant sysdev attribute to generic attribute

Use the new generic int attribute accessors for the x86 mce tolerant
attribute. Simple example to illustrate the new macros.

There are much more places all over the tree that could be converted
like this.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agosysdev: Add utility functions for simple int/ulong variable sysdev attributes
Andi Kleen [Tue, 1 Jul 2008 16:48:42 +0000 (18:48 +0200)]
sysdev: Add utility functions for simple int/ulong variable sysdev attributes

This adds a new sysdev_ext_attribute that stores a pointer to the variable
it manages and some utility functions/macro to easily use them.

Previously all users wrote custom macros to generate show/store
functions for each variable, with this it is possible to avoid
that in many cases.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agosysdev: Pass the attribute to the low level sysdev show/store function
Andi Kleen [Tue, 1 Jul 2008 16:48:41 +0000 (18:48 +0200)]
sysdev: Pass the attribute to the low level sysdev show/store function

This allow to dynamically generate attributes and share show/store
functions between attributes. Right now most attributes are generated
by special macros and lots of duplicated code. With the attribute
passed it's instead possible to attach some data to the attribute
and then use that in shared low level functions to do different things.

I need this for the dynamically generated bank attributes in the x86
machine check code, but it'll allow some further cleanups.

I converted all users in tree to the new show/store prototype. It's a single
huge patch to avoid unbisectable sections.

Runtime tested: x86-32, x86-64
Compiled only: ia64, powerpc
Not compile tested/only grep converted: sh, arm, avr32

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodriver core: Suppress sysfs warnings for device_rename().
Cornelia Huck [Tue, 10 Jun 2008 09:09:08 +0000 (11:09 +0200)]
driver core: Suppress sysfs warnings for device_rename().

driver core: Suppress sysfs warnings for device_rename().

Renaming network devices to an already existing name is not
something we want sysfs to print a scary warning for, since the
callers can deal with this correctly. So let's introduce
sysfs_create_link_nowarn() which gets rid of the common warning.

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agokobject: Transmit return value of call_usermodehelper() to caller
Wang Chen [Tue, 24 Jun 2008 08:59:02 +0000 (16:59 +0800)]
kobject: Transmit return value of call_usermodehelper() to caller

kobject_uevent_env() drops the return value of call_usermodehelper().
It will make upper caller, such as dm_send_uevents(), to lose error
information.

BTW, Previously kobject_uevent_env() transmitted return of
call_usermodehelper() to callers, but
commit 5f123fbd80f4f788554636f02bf73e40f914e0d6
"[PATCH] merge kobject_uevent and kobject_hotplug" removed it.

Signed-off-by: Wang Chen <wangchen@cn.fujitsu.com>
Cc: Kay Sievers <kay.sievers@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agosysfs-rules.txt: reword API stability statement
Nathan Lynch [Wed, 2 Jul 2008 17:21:30 +0000 (10:21 -0700)]
sysfs-rules.txt: reword API stability statement

The first paragraph of this document implies that user space developers
shouldn't use sysfs at all, but then it goes on to describe rules that
developers should follow when accessing sysfs.  Not only is this somewhat
self-contradictory, it has been shown to discourage developers from using
established sysfs interfaces.

A note of caution is more appropriate than a blanket "sysfs will never
be stable" assertion.

Signed-off-by: Nathan Lynch <ntl@pobox.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodebugfs: Implement debugfs_remove_recursive()
Haavard Skinnemoen [Tue, 1 Jul 2008 13:14:51 +0000 (15:14 +0200)]
debugfs: Implement debugfs_remove_recursive()

debugfs_remove_recursive() will remove a dentry and all its children.
Drivers can use this to zap their whole debugfs tree so that they don't
need to keep track of every single debugfs dentry they created.

It may fail to remove the whole tree in certain cases:

sh-3.2# rmmod atmel-mci < /sys/kernel/debug/mmc0/ios/clock
mmc0: card b368 removed
atmel_mci atmel_mci.0: Lost dma0chan1, falling back to PIO
sh-3.2# ls /sys/kernel/debug/mmc0/
ios

But I'm not sure if that case can be handled in any sane manner.

Signed-off-by: Haavard Skinnemoen <haavard.skinnemoen@atmel.com>
Cc: Pierre Ossman <drzeus-list@drzeus.cx>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoHOWTO: change email addresses of James in HOWTO
Tsugikazu Shibata [Fri, 20 Jun 2008 01:59:52 +0000 (10:59 +0900)]
HOWTO: change email addresses of James in HOWTO

Signed-off-by: Tsugikazu Shibata <tshibata@ab.jp.nec.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoalways enable FW_LOADER unless EMBEDDED=y
Adrian Bunk [Tue, 10 Jun 2008 16:04:08 +0000 (19:04 +0300)]
always enable FW_LOADER unless EMBEDDED=y

James Bottomley recently discovered that we have
{request,release}_firmware() dummies for the case of the actual
functions not being available and has a fix for the bug that was
actually causing build errors for built-in users with
CONFIG_FW_LOADER=m.

But now missing selects on FW_LOADER are no longer visible at
compile-time at all and can become runtime problems.

FW_LOADER is infrastructure with relatively small codesize we can safely
enable for everyone, and only for people who really need small kernels
(and can be expected to know what they are doing) it matters being able
to disable it.

This patch therefore always sets FW_LOADER=y and allows users only to
disable it with EMBEDDED=y.

As a bonus, we can then get rid of all "select FW_LOADER" plus the due
to it required "depends on HOTPLUG" which removes some complexity from
our Kconfig files.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agouio-howto.tmpl: use unique output names
Mike Frysinger [Tue, 24 Jun 2008 21:25:00 +0000 (14:25 -0700)]
uio-howto.tmpl: use unique output names

The Userspace I/O HOWTO template sets two different sections with the same
html output name (about.html).  This clearly won't work, so change the
first one to a unique "aboutthis.html" to prevent clobbering.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Acked-by: Hans J. Koch <hjk@linutronix.de>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agouio-howto.tmpl: use standard copyright/legal markings
Mike Frysinger [Tue, 24 Jun 2008 21:24:57 +0000 (14:24 -0700)]
uio-howto.tmpl: use standard copyright/legal markings

The Userspace I/O HOWTO document uses straight <sect1> tags and plain text
to describe copyright/legal information.  It should instead use the
<copyright> and <legalnotice> tags like all other documents in the kernel.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Acked-by: Hans J. Koch <hjk@linutronix.de>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agosysfs: don't call notify_change
Miklos Szeredi [Mon, 16 Jun 2008 11:46:47 +0000 (13:46 +0200)]
sysfs: don't call notify_change

sysfs_chmod_file() calls notify_change() to change the permission bits
on a sysfs file.  Replace with explicit call to sysfs_setattr() and
fsnotify_change().

This is equivalent, except that security_inode_setattr() is not
called.  This function is called by drivers, so the security checks do
not make any sense.

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agosysdev: fix debugging statements in registration code.
Ben Dooks [Thu, 12 Jun 2008 18:00:34 +0000 (19:00 +0100)]
sysdev: fix debugging statements in registration code.

The systdev_class_register() and sysdev_register() functions have
pr_debug() statements which are enabled when the user selects the
driver core debug. Both of these routines do not produce the
correct output, as they make assumptions about data which has not
been initialised.

In sysdev_class_register() the code uses the kobject_name(&cls->kset.kobj)
at the start of the function, but this is not setup until later in the
same call. Change this to use cls->name which is passed in from the caller.

The sysdev_register() function tries to get the name of the sysdev by
kobject_name(&sysdev->kobj), but that isn't setup until later in the same
function so change this message to use the name of the sysdev's class and
add another message once the name is initialised.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agokobject: should use kobject_put() in kset-example
Li Zefan [Fri, 13 Jun 2008 03:09:16 +0000 (11:09 +0800)]
kobject: should use kobject_put() in kset-example

We should call kobject_put() instead of kfree() if kobject_init_and_add()
returns an error, shouldn't we? Don't set up a bad example ;)

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agokobject: reorder kobject to save space on 64 bit builds
Richard Kennedy [Mon, 2 Jun 2008 10:07:25 +0000 (11:07 +0100)]
kobject: reorder kobject to save space on 64 bit builds

reorder kobject to save space on 64 bit builds.
shrinks from 72 to 64 bytes & moves allocated kobject to a smaller
slab.

Signed-off-by: Richard Kennedy <richard@rsk.demon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUIO: minor style and comment fixes
Uwe Kleine-König [Tue, 10 Jun 2008 07:14:48 +0000 (09:14 +0200)]
UIO: minor style and comment fixes

Signed-off-by: Uwe Kleine-König <Uwe.Kleine-Koenig@digi.com>
Signed-off-by: Hans J. Koch <hjk@linutronix.de>
16 years agoUIO: add generic UIO platform driver
Uwe Kleine-König [Sat, 31 May 2008 09:37:27 +0000 (11:37 +0200)]
UIO: add generic UIO platform driver

This patch adds a generic UIO platform driver. It eliminates the need for a
dedicated kernel module for simple platform devices. Users only need to
implement their irq handler in platform code and fill a struct uio_info
there. This helps avoiding code duplication as UIO platform drivers often
share a lot of common code.

Signed-off-by: Uwe Kleine-König <Uwe.Kleine-Koenig@digi.com>
Signed-off-by: Hans J. Koch <hjk@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUIO: Add write function to allow irq masking
Hans J. Koch [Fri, 23 May 2008 11:50:14 +0000 (13:50 +0200)]
UIO: Add write function to allow irq masking

Sometimes it is necessary to enable/disable the interrupt of a UIO device
from the userspace part of the driver. With this patch, the UIO kernel driver
can implement an "irqcontrol()" function that does this. Userspace can write
an s32 value to /dev/uioX (usually 0 or 1 to turn the irq off or on). The
UIO core will then call the driver's irqcontrol function.

Signed-off-by: Hans J. Koch <hjk@linutronix.de>
Acked-by: Uwe Kleine-König <Uwe.Kleine-Koenig@digi.com>
Acked-by: Magnus Damm <damm@igel.co.jp>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUIO: fix UIO Kconfig dependencies
Uwe Kleine-König [Sat, 31 May 2008 09:37:27 +0000 (11:37 +0200)]
UIO: fix UIO Kconfig dependencies

ae210f188614bb3d1ee3f19c64e28e3cdd44877c introduced a big "if UIO"/"endif"
where all uio drivers are defined.  So know there is no need for them to
depend explicitly on UIO.

Signed-off-by: Uwe Kleine-König <Uwe.Kleine-Koenig@digi.com>
Signed-off-by: Hans J. Koch <hjk@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agopnp: add acpi:* modalias entries
Kay Sievers [Wed, 28 May 2008 21:06:47 +0000 (23:06 +0200)]
pnp: add acpi:* modalias entries

Along with the non-modalias conformant "pnp:*" aliases, we add "acpi:*"
entries to PNP drivers, to allow module autoloading by ACPI PNP device
entries, which export proper modalias information, without any specific
userspace modprobe mangling.

Cc: Adam Belay <ambx1@neo.rr.com>
Cc: Thomas Renninger <trenn@suse.de>
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodriver core: fix a lot of printk usages of bus_id
Greg Kroah-Hartman [Fri, 2 May 2008 04:02:41 +0000 (06:02 +0200)]
driver core: fix a lot of printk usages of bus_id

We have the dev_printk() variants for this kind of thing, use them
instead of directly trying to access the bus_id field of struct device.

This is done in order to remove bus_id entirely.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodriver core: remove DEVICE_ID_SIZE define
Greg Kroah-Hartman [Fri, 2 May 2008 04:02:41 +0000 (06:02 +0200)]
driver core: remove DEVICE_ID_SIZE define

There is no such thing as a "device id size" in the driver core, so
remove the define and fix up any users of this odd define in the rest of
the kernel.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodriver core: remove DEVICE_NAME_SIZE define
Kay Sievers [Fri, 2 May 2008 04:02:41 +0000 (06:02 +0200)]
driver core: remove DEVICE_NAME_SIZE define

There is no such thing as a "device name size" in the driver core, so
remove the define and fix up any users of this odd define in the rest of
the kernel.

Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodriver core: remove KOBJ_NAME_LEN define
Kay Sievers [Fri, 2 May 2008 04:02:41 +0000 (06:02 +0200)]
driver core: remove KOBJ_NAME_LEN define

Kobjects do not have a limit in name size since a while, so stop
pretending that they do.

Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoclass: change internal semaphore to a mutex
Dave Young [Wed, 28 May 2008 16:28:39 +0000 (09:28 -0700)]
class: change internal semaphore to a mutex

Now that the lockdep infrastructure in the class core is in place, we
should be able to properly change the internal class semaphore to be a
mutex.

David wrote the original patch, and Greg fixed it up to apply properly
due to all of the recent changes in this area.

From: Dave Young <hidave.darkstar@gmail.com>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoclass: add lockdep infrastructure
Matthew Wilcox [Wed, 28 May 2008 16:28:39 +0000 (09:28 -0700)]
class: add lockdep infrastructure

This adds the infrastructure to properly handle lockdep issues when the
internal class semaphore is changed to a mutex.

Matthew wrote the original patch, and Greg fixed it up to work properly
with the class_create() function.

From: Matthew Wilcox <matthew@wil.cx>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Dave Young <hidave.darkstar@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoclass: fix docbook comments for class_private structure
Greg Kroah-Hartman [Wed, 28 May 2008 16:28:39 +0000 (09:28 -0700)]
class: fix docbook comments for class_private structure

Removes a field that has been deleted, and adds a description fo the
class_dirs field which was previously undocumented.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoclass: rename "sem" to "class_sem" in internal class structure
Greg Kroah-Hartman [Wed, 28 May 2008 16:28:39 +0000 (09:28 -0700)]
class: rename "sem" to "class_sem" in internal class structure

This renames the struct class "sem" field to be "class_sem" to make
things easier when struct bus_type and struct class merge in the future.
It also makes grepping for fields easier as well.

Based on an idea from Kay.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoclass: rename "subsys" to "class_subsys" in internal class structure
Greg Kroah-Hartman [Wed, 28 May 2008 16:28:39 +0000 (09:28 -0700)]
class: rename "subsys" to "class_subsys" in internal class structure

This renames the struct class "subsys" field to be "class_subsys" to
make things easier when struct bus_type and struct class merge in the
future.  It also makes grepping for fields easier as well.

Based on an idea from Kay.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoclass: rename "interfaces" to "class_interfaces" in internal class structure
Greg Kroah-Hartman [Wed, 28 May 2008 16:28:39 +0000 (09:28 -0700)]
class: rename "interfaces" to "class_interfaces" in internal class structure

This renames the struct class "interfaces" field to be
"class_interfaces" to make things easier when struct bus_type and struct
class merge in the future.  It also makes grepping for fields easier as
well.

Based on an idea from Kay.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoclass: rename "devices" to "class_devices" in internal class structure
Greg Kroah-Hartman [Wed, 28 May 2008 16:28:39 +0000 (09:28 -0700)]
class: rename "devices" to "class_devices" in internal class structure

This renames the struct class "devices" field to be "class_devices" to
make things easier when struct bus_type and struct class merge in the
future.  It also makes grepping for fields easier as well.

Based on an idea from Kay.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoclass: move driver core specific parts to a private structure
Greg Kroah-Hartman [Tue, 22 Jan 2008 23:17:41 +0000 (18:17 -0500)]
class: move driver core specific parts to a private structure

This moves the portions of struct class that are dynamic (kobject and
lock and lists) out of the main structure and into a dynamic, private,
structure.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agobluetooth: remove improper bluetooth class symlinks.
Greg Kroah-Hartman [Tue, 27 May 2008 18:50:16 +0000 (11:50 -0700)]
bluetooth: remove improper bluetooth class symlinks.

Don't create symlinks in a class to a device that is not owned by the
class.  If the bluetooth subsystem really wants to point to all of the
devices it controls, it needs to create real devices, not fake symlinks.

Cc: Maxim Krasnyansky <maxk@qualcomm.com>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoinfiniband: make cm_device use a struct device and not a kobject.
Greg Kroah-Hartman [Tue, 27 May 2008 07:17:53 +0000 (10:17 +0300)]
infiniband: make cm_device use a struct device and not a kobject.

This object really should be a struct device, or at least contain a
pointer to a struct device, as it is trying to create a separate device
tree outside of the main device tree.  This patch fixes this problem.

It is needed for the class core rework that is being done in the driver
core.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Roland Dreier <rolandd@cisco.com>
Cc: Sean Hefty <sean.hefty@intel.com>
Cc: Hal Rosenstock <hal.rosenstock@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoinfiniband: rename "device" to "ib_device" in cm_device
Greg Kroah-Hartman [Tue, 27 May 2008 07:17:53 +0000 (10:17 +0300)]
infiniband: rename "device" to "ib_device" in cm_device

This pointer really is a struct ib_device, not a struct device, so name
it properly to help prevent confusion.

This makes the followon patch in this series much smaller and easier to
understand as well.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Roland Dreier <rolandd@cisco.com>
Cc: Hal Rosenstock <hal.rosenstock@gmail.com>
Acked-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoblock: make /proc/partitions and /proc/diskstats use class_find_device()
Greg Kroah-Hartman [Thu, 22 May 2008 21:21:08 +0000 (17:21 -0400)]
block: make /proc/partitions and /proc/diskstats use class_find_device()

Use the proper class iterator function instead of mucking around in the
internals of the class structures.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoblock: move header for /proc/partitions to seq_start
Greg Kroah-Hartman [Thu, 22 May 2008 21:21:08 +0000 (17:21 -0400)]
block: move header for /proc/partitions to seq_start

The seq_start call is the better place for the header for the file, that
way we don't have to be mucking in the class structure to try to figure
out if this is the first partition or not.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoblock: make proc files seq_start use the class_find_device()
Greg Kroah-Hartman [Thu, 22 May 2008 21:21:08 +0000 (17:21 -0400)]
block: make proc files seq_start use the class_find_device()

Use the proper class iterator function instead of mucking around in the
internals of the class structures.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoblock: make /proc/diskstats only build if CONFIG_PROC_FS is enabled
Randy Dunlap [Fri, 23 May 2008 16:44:11 +0000 (09:44 -0700)]
block: make /proc/diskstats only build if CONFIG_PROC_FS is enabled

These functions are only needed if CONFIG_PROC_FS is enabled, so save
the space when it is not.

This also makes it easier for a patch later in this series to work
properly if CONFIG_PROC_FS is not enabled.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoblock: make blk_lookup_devt use the class iterator function
Greg Kroah-Hartman [Thu, 22 May 2008 21:21:08 +0000 (17:21 -0400)]
block: make blk_lookup_devt use the class iterator function

Use the proper class iterator function instead of mucking around in the
internals of the class structures.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoblock: make printk_partition use the class iterator function
Greg Kroah-Hartman [Thu, 22 May 2008 21:21:08 +0000 (17:21 -0400)]
block: make printk_partition use the class iterator function

Use the proper class iterator function instead of mucking around in the
internals of the class structures.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoblock: fix compiler warning in genhd.c
Greg Kroah-Hartman [Thu, 22 May 2008 21:21:08 +0000 (17:21 -0400)]
block: fix compiler warning in genhd.c

Warn if something really bad happens if we can't create this link.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoDriver Core: add ability for class_find_device to start in middle of list
Greg Kroah-Hartman [Thu, 22 May 2008 21:21:08 +0000 (17:21 -0400)]
Driver Core: add ability for class_find_device to start in middle of list

This mirrors the functionality that driver_find_device has as well.

We add a start variable, and all callers of the function are fixed up at
the same time.

The block layer will be using this new functionality in a follow-on
patch.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoDriver Core: add ability for class_for_each_device to start in middle of list
Greg Kroah-Hartman [Thu, 22 May 2008 21:21:08 +0000 (17:21 -0400)]
Driver Core: add ability for class_for_each_device to start in middle of list

This mirrors the functionality that driver_for_each_device has as well.

We add a start variable, and all callers of the function are fixed up at
the same time.

The block layer will be using this new functionality in a follow-on
patch.

Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: convert device_create_drvdata to device_create
Greg Kroah-Hartman [Tue, 22 Jul 2008 03:03:34 +0000 (20:03 -0700)]
device create: convert device_create_drvdata to device_create

Now that device_create() has been audited, rename things back to the
original call to be sane.

Keep the device_create_drvdata macro around to make merges easier.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodriver core: remove device_create()
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
driver core: remove device_create()

There are no more users of this, and it is racy.  Use
device_create_drvdata() or device_create_vargs() instead.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: x86: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: x86: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: usb: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: usb: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: spi: convert device_create to device_create_drvdata
David Brownell [Fri, 6 Jun 2008 20:48:07 +0000 (15:48 -0500)]
device create: spi: convert device_create to device_create_drvdata

Switch over to use the shiny new device_create_drvdata() call
instead of the original device_create() calls, so this continues
to work after device_create() is  removed.

Note that this driver never had the race which motivated removing
the original call; it locked correctly.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: sound: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: sound: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Jaroslav Kysela <perex@perex.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: scsi: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: scsi: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: s390: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: s390: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: net: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: net: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: mtd: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: mtd: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: misc: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: misc: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: mips: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: mips: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: macintosh: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: macintosh: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: isdn: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: isdn: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Karsten Keil <kkeil@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: infiniband: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: infiniband: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Roland Dreier <rolandd@cisco.com>
Cc: Sean Hefty <sean.hefty@intel.com>
Cc: Hal Rosenstock <hal.rosenstock@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: ieee1394: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: ieee1394: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Ben Collins <ben.collins@ubuntu.com>
Acked-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: ide: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: ide: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: i2c: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: i2c: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: hwmon: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: hwmon: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Mark M. Hoffman <mhoffman@lightlink.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: hid: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: hid: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: framebuffer: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: framebuffer: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: dvb: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: dvb: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: dca: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: dca: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: coda: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: coda: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Cc: Jan Harkes <jaharkes@cs.cmu.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: char: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: char: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodevice create: block: convert device_create to device_create_drvdata
Greg Kroah-Hartman [Wed, 21 May 2008 19:52:33 +0000 (12:52 -0700)]
device create: block: convert device_create to device_create_drvdata

device_create() is race-prone, so use the race-free
device_create_drvdata() instead as device_create() is going away.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoFirmware: fix typo in example code
Greg Kroah-Hartman [Wed, 21 May 2008 22:21:38 +0000 (18:21 -0400)]
Firmware: fix typo in example code

It's not like this stuff even builds right now, but this will make
Randy's scripts happy :)

Cc: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agodebugfs: Add a reference to the debugfs API documentation.
Robert P. J. Day [Mon, 19 May 2008 22:06:00 +0000 (00:06 +0200)]
debugfs: Add a reference to the debugfs API documentation.

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Cc: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agokobject: replace '/' with '!' in name
Kay Sievers [Tue, 6 May 2008 20:24:04 +0000 (22:24 +0200)]
kobject: replace '/' with '!' in name

Some (block) devices have a '/' in the name, and need special
handling. Let's have that rule to the core, so we can remove it
from the block class.

Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agosysfs: add /sys/dev/{char,block} to lookup sysfs path by major:minor
Dan Williams [Mon, 21 Apr 2008 17:51:07 +0000 (10:51 -0700)]
sysfs: add /sys/dev/{char,block} to lookup sysfs path by major:minor

Why?:
There are occasions where userspace would like to access sysfs
attributes for a device but it may not know how sysfs has named the
device or the path.  For example what is the sysfs path for
/dev/disk/by-id/ata-ST3160827AS_5MT004CK?  With this change a call to
stat(2) returns the major:minor then userspace can see that
/sys/dev/block/8:32 links to /sys/block/sdc.

What are the alternatives?:
1/ Add an ioctl to return the path: Doable, but sysfs is meant to reduce
   the need to proliferate ioctl interfaces into the kernel, so this
   seems counter productive.

2/ Use udev to create these symlinks: Also doable, but it adds a
   udev dependency to utilities that might be running in a limited
   environment like an initramfs.

3/ Do a full-tree search of sysfs.

[kay.sievers@vrfy.org: fix duplicate registrations]
[kay.sievers@vrfy.org: cleanup suggestions]

Cc: Neil Brown <neilb@suse.de>
Cc: Tejun Heo <htejun@gmail.com>
Acked-by: Kay Sievers <kay.sievers@vrfy.org>
Reviewed-by: SL Baur <steve@xemacs.org>
Acked-by: Kay Sievers <kay.sievers@vrfy.org>
Acked-by: Mark Lord <lkml@rtr.ca>
Acked-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6
Linus Torvalds [Mon, 21 Jul 2008 22:42:20 +0000 (15:42 -0700)]
Merge git://git./linux/kernel/git/gregkh/usb-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6: (100 commits)
  usb-storage: revert DMA-alignment change for Wireless USB
  USB: use reset_resume when normal resume fails
  usb_gadget: composite cdc gadget fault handling
  usb gadget: minor USBCV fix for composite framework
  USB: Fix bug with byte order in isp116x-hcd.c fio write/read
  USB: fix double kfree in ipaq in error case
  USB: fix build error in cdc-acm for CONFIG_PM=n
  USB: remove board-specific UP2OCR configuration from pxa27x-udc
  USB: EHCI: Reconciling USB register differences on MPC85xx vs MPC83xx
  USB: Fix pointer/int cast in USB devio code
  usb gadget: g_cdc dependso on NET
  USB: Au1xxx-usb: suspend/resume support.
  USB: Au1xxx-usb: clean up ohci/ehci bus glue sources.
  usbfs: don't store bad pointers in registration
  usbfs: fix race between open and unregister
  usbfs: simplify the lookup-by-minor routines
  usbfs: send disconnect signals when device is unregistered
  USB: Force unbinding of drivers lacking reset_resume or other methods
  USB: ohci-pnx4008: I2C cleanups and fixes
  USB: debug port converter does not accept more than 8 byte packets
  ...

16 years agousb-storage: revert DMA-alignment change for Wireless USB
Alan Stern [Mon, 30 Jun 2008 17:39:59 +0000 (13:39 -0400)]
usb-storage: revert DMA-alignment change for Wireless USB

This patch (as1110) reverts an earlier patch meant to help with
Wireless USB host controllers.  These controllers can have bulk
maxpacket values larger than 512, which puts unusual constraints on
the sizes of scatter-gather list elements.  However it turns out that
the block layer does not provide the support we need to enforce these
constraints; merely changing the DMA alignment mask doesn't help.
Hence there's no reason to keep the original patch.  The Wireless USB
problem will have to be solved a different way.

In addition, there is a reason to get rid of the earlier patch.  By
dereferencing a pointer stored in the ep_in array of struct
usb_device, the current code risks an invalid memory access when it
runs concurrently with device removal.  The members of that array are
cleared before the driver's disconnect method is called, so it should
not try to use them.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: use reset_resume when normal resume fails
Alan Stern [Mon, 30 Jun 2008 15:14:43 +0000 (11:14 -0400)]
USB: use reset_resume when normal resume fails

This patch (as1109b) makes USB-Persist more resilient to errors.  With
the current code, if a normal resume fails, it's an unrecoverable
error.  With the patch, if a normal resume fails (and if the device is
enabled for USB-Persist) then a reset-resume is tried.

This fixes the problem reported in Bugzilla #10977.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agousb_gadget: composite cdc gadget fault handling
David Brownell [Tue, 1 Jul 2008 20:18:20 +0000 (13:18 -0700)]
usb_gadget: composite cdc gadget fault handling

These two fixes ensure the new "CDC Composite Device" gadget
fails cleanly when it's loaded on hardware that can't support
this particular gadget driver.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agousb gadget: minor USBCV fix for composite framework
David Lopo [Tue, 1 Jul 2008 20:14:17 +0000 (13:14 -0700)]
usb gadget: minor USBCV fix for composite framework

Fill in a reserved/unused device qualifier field to ensure that
the USBCV tests will always pass.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: Fix bug with byte order in isp116x-hcd.c fio write/read
Julien May [Fri, 27 Jun 2008 13:01:18 +0000 (15:01 +0200)]
USB: Fix bug with byte order in isp116x-hcd.c fio write/read

URB payload data are transfered in wrong byte order on a big endinan
architecture (AVR32).

Signed-off-by: Julien May <mailinglist@miromico.ch>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: fix double kfree in ipaq in error case
Oliver Neukum [Mon, 30 Jun 2008 12:33:57 +0000 (14:33 +0200)]
USB: fix double kfree in ipaq in error case

in the error case the ipaq driver leaves a dangling pointer to already
freed memory that will be freed again.

Signed-off-by: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: fix build error in cdc-acm for CONFIG_PM=n
Oliver Neukum [Tue, 1 Jul 2008 17:10:08 +0000 (19:10 +0200)]
USB: fix build error in cdc-acm for CONFIG_PM=n

Here's the fix. cdc-wdm has the same problem. The fix is the same.

Signed-off-by: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: remove board-specific UP2OCR configuration from pxa27x-udc
Daniel Ribeiro [Wed, 2 Jul 2008 02:09:27 +0000 (23:09 -0300)]
USB: remove board-specific UP2OCR configuration from pxa27x-udc

Remove the board-specific UP2OCR configuration from the
pxa27x-udc driver.

Signed-off-by: Daniel Ribeiro <drwyrm@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: EHCI: Reconciling USB register differences on MPC85xx vs MPC83xx
Srikanth Srinivasan [Wed, 2 Jul 2008 07:14:33 +0000 (02:14 -0500)]
USB: EHCI: Reconciling USB register differences on MPC85xx vs MPC83xx

A couple of USB register initializations had to be changed on MPC85xx
platforms.  This is due to the internal SoC buses being different on
MPC83xx SoCs vs MPC85xx SoCs.

We currently handle this via an ifdef since 83xx and 85xx are mutually
exclusive kernel builds.

Signed-off-by: Srikanth Srinivasan <srikanth.srinivasan@freescale.com>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: Fix pointer/int cast in USB devio code
David Howells [Wed, 2 Jul 2008 11:28:55 +0000 (12:28 +0100)]
USB: Fix pointer/int cast in USB devio code

Fix pointer/int cast in USB devio code, and thus avoid a compiler warning.

A void* data argument passed to bus_find_device() and thence to match_devt()
is used to carry a 32-bit datum.  However, casting directly between a u32 and
a pointer is not permitted - there must be an intermediate cast via (unsigned)
long.

This was introduced by the following patch:

commit 94b1c9fa060ece2c8f080583beb6cc6008e41413
Author: Alan Stern <stern@rowland.harvard.edu>
Date:   Tue Jun 24 14:47:12 2008 -0400

    usbfs: simplify the lookup-by-minor routines

    This patch (as1105) simplifies the lookup-by-minor-number code in
    usbfs.  Instead of passing the minor number to the callback, which
    must then reconstruct the entire dev_t value, the patch passes the
    dev_t value directly.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: David Howells <dhowells@redhat.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agousb gadget: g_cdc dependso on NET
Randy Dunlap [Thu, 3 Jul 2008 21:44:59 +0000 (14:44 -0700)]
usb gadget: g_cdc dependso on NET

g_cdc needs to depend on NET, otherwise net-related build errors happen:

ERROR: "netif_carrier_on" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "netif_carrier_off" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "netif_rx" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "alloc_etherdev_mq" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "free_netdev" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "register_netdev" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "print_mac" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "dev_kfree_skb_any" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "__alloc_skb" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "eth_type_trans" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "unregister_netdev" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "__netif_schedule" [drivers/usb/gadget/g_cdc.ko] undefined!
ERROR: "skb_put" [drivers/usb/gadget/g_cdc.ko] undefined!

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: Au1xxx-usb: suspend/resume support.
Manuel Lauss [Mon, 23 Jun 2008 07:09:37 +0000 (09:09 +0200)]
USB: Au1xxx-usb: suspend/resume support.

Copy the OHCI/EHCI PM callbacks of the PCI implementation since
they work equally well on Au1xxx hardware.

Tested on Au1200.

Signed-off-by: Manuel Lauss <mano@roarinelk.homelinux.net>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: Au1xxx-usb: clean up ohci/ehci bus glue sources.
Manuel Lauss [Mon, 23 Jun 2008 07:08:29 +0000 (09:08 +0200)]
USB: Au1xxx-usb: clean up ohci/ehci bus glue sources.

- Fold multiple probe/remove callbacks into one function;
- minor style fixes, no functional changes.

Tested on Au1200.

Signed-off-by: Manuel Lauss <mano@roarinelk.homelinux.net>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agousbfs: don't store bad pointers in registration
Alan Stern [Tue, 24 Jun 2008 18:47:29 +0000 (14:47 -0400)]
usbfs: don't store bad pointers in registration

This patch (as1107) fixes a small bug in the usbfs registration and
unregistration code.  It avoids leaving an error value stored in the
device's usb_classdev field and it avoids trying to unregister a NULL
pointer.  (It also fixes a rather extreme overuse of whitespace.)

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agousbfs: fix race between open and unregister
Alan Stern [Tue, 24 Jun 2008 18:47:19 +0000 (14:47 -0400)]
usbfs: fix race between open and unregister

This patch (as1106) fixes a race between opening and unregistering
device files in usbfs.  The current code drops its reference to the
device and then reacquires it, ignoring the possibility that the
device structure might have been removed in the meantime.  It also
doesn't check whether the device is already in the NOTATTACHED state
when the file is opened.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agousbfs: simplify the lookup-by-minor routines
Alan Stern [Tue, 24 Jun 2008 18:47:12 +0000 (14:47 -0400)]
usbfs: simplify the lookup-by-minor routines

This patch (as1105) simplifies the lookup-by-minor-number code in
usbfs.  Instead of passing the minor number to the callback, which
must then reconstruct the entire dev_t value, the patch passes the
dev_t value directly.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agousbfs: send disconnect signals when device is unregistered
Alan Stern [Tue, 24 Jun 2008 18:47:04 +0000 (14:47 -0400)]
usbfs: send disconnect signals when device is unregistered

USB device files are accessible in two ways: as files in usbfs and as
character device nodes.  The two paths are supposed to behave
identically, but they don't.  When the underlying USB device is
unplugged, disconnect signals are sent to processes with open usbfs
files (if they requested these signals) but not to processes with open
device node files.

This patch (as1104) fixes the bug by moving the disconnect-signalling
code into a common subroutine which is called from both paths.
Putting this subroutine in devio.c removes the only out-of-file
reference to struct dev_state, and so the structure's declaration can
be moved from usb.h into devio.c.

Finally, the new subroutine performs one extra action: It kills all
the outstanding async URBs.  (I'd kill the outstanding synchronous
URBs too, if there was any way to do it.)  In the past this hasn't
mattered much, because devices were unregistered from usbfs only
when they were disconnected.  But now the unregistration can also
occur whenever devices are unbound from the usb_generic driver.  At
any rate, killing URBs when a device is unregistered from usbfs seems
like a good thing to do.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: Force unbinding of drivers lacking reset_resume or other methods
Alan Stern [Mon, 23 Jun 2008 20:00:40 +0000 (16:00 -0400)]
USB: Force unbinding of drivers lacking reset_resume or other methods

This patch (as1024) takes care of a FIXME issue: Drivers that don't
have the necessary suspend, resume, reset_resume, pre_reset, or
post_reset methods will be unbound and their interface reprobed when
one of the unsupported events occurs.

This is made slightly more difficult by the fact that bind operations
won't work during a system sleep transition.  So instead the code has
to defer the operation until the transition ends.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: ohci-pnx4008: I2C cleanups and fixes
Jean Delvare [Wed, 18 Jun 2008 12:46:27 +0000 (14:46 +0200)]
USB: ohci-pnx4008: I2C cleanups and fixes

Various cleanups and fixes to the i2c code in ohci-pnx4008:
* Delete empty isp1301_command. The i2c driver command implementation
  is optional, so there's no point in providing an empty
  implementation.
* Give a name to isp1301_driver. I'm surprised that i2c-core accepted
  to register this driver at all. I've chosen "isp1301_pnx" as the
  name, because it's not a generic ISP1301 driver (much like the
  isp1301_omap driver.) We might want to make the name even more
  specific (but "isp1301_ohci_pnx4008" doesn't fit.)
* The ISP1301 is definitely not a hardware monitoring device.
* Fix a memory leak on failure in isp1301_attach. If
  i2c_attach_client fails, the client is not registered so
  isp1301_detach is never called and the i2c_client memory is lost.
* Use strlcpy instead of strcpy.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Vitaly Wool <vitalywool@gmail.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
16 years agoUSB: debug port converter does not accept more than 8 byte packets
Aleksey Gorelov [Thu, 19 Jun 2008 22:22:17 +0000 (15:22 -0700)]
USB: debug port converter does not accept more than 8 byte packets

USB debug port only supports 8 byte rx/tx packets. Although spec implies that
"if a packet larger than eight bytes is received from the remote computer, the
device must break the larger packet into eight-byte packets before sending the
data to the Debug Port", the real PLX NET20DC device does not handle it right -
data is corrupted on debug port end if serial interface sends >8 byte urbs.
Patch below fixes the issue by limiting tx urb to 8 byte.

Signed off by: Aleks Gorelov <dared1st@yahoo.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>