HID: wacom: ask for a in-prox report when it was missed
If noone listens to the input device when a tool comes in proximity,
the tablet does not send the in-prox event when a client becomes available.
That means that no events will be sent until the tool is taken out of
proximity.
In this situation, ask for the report WACOM_REPORT_INTUOSREAD which will
read the corresponding feature and generate an in-prox event.
To make some generation of hardware working, we need to unset the
quirk NO_GET set by hid-core because the interfaces are seen as "boot
mouse".
We don't schedule this read in a worker while we are in an IO interrupt.
We know that usbhid will do it asynchronously. If this is triggered by
uhid, then this is obviously a client side bug :)
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com> Acked-by: Jason Gerecke <killertofu@gmail.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
HID: hid-sensor-hub: fix attribute read for logical usage id
For defining enumeration values like report or power status events, the
enumeration usage ids are enclosed in a logical collection. In this case we
need to match logical usage id for pending read on this usage id. For example,
in the below field, when read is requested for 0319, the report will
contain one of the status usages like 850, 851 etc. In this case the raw
event will not match 0319. So when logical collection matches, then wake
up the pending thread.
Jim Keir [Thu, 26 Feb 2015 09:14:03 +0000 (09:14 +0000)]
HID: pidff: support more than one concurrent effect
The PID driver (usbhid/hid-pidff.c) does not set the effect ID when
uploading an effect. The result is that the initial upload works but
subsequent uploads to modify effect parameters are all directed at the
last-created effect.
The targeted effect ID must be passed back to the device when effect
parameters are changed. This is done at the start of
"pidff_set_condition_report", "pidff_set_periodic_report" etc. based on
the value of "pidff->block_load[PID_EFFECT_
BLOCK_INDEX].value[0]".
This value is only ever set during pidff_request_effect_upload.
The result is stored in "pidff->pid_id[effect->id]" at the end of
pid_upload_effect, for later use. However, if an effect is modified and
re-sent then this identifier is not being copied back from
pidff->pid_id[effect->id] before sending the command to the device. The
fix is to do this at the start of pidff_upload_effect.
Signed-off-by: Jim Keir <jimkeir@oracledbadirect.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Jean Delvare [Fri, 20 Feb 2015 16:33:33 +0000 (17:33 +0100)]
HID: Stop hiding options with !EXPERT
Many HID driver options are hidden unless EXPERT is set. While I
understand the idea of simplifying the kernel configuration for most
users, in practice I believe it adds more confusion than it helps.
One thing that worries me is that, in non-EXPERT mode, these drivers
will be either built-in or modular based on apparent magic. For
example, switching INPUT and HID from m to y will cause all these
drivers to be built into the kernel when they were previously built
as modules. Short of enabling EXPERT mode altogether, the user has no
control over that.
Generally I do not think tristate options should depend on !EXPERT.
Of these, 11 of 15 are currently in the hid subsystem.
The HID_LOGITECH option is even worse than the others. Sub-options
depend on it, and this causes menuconfig and friends to display the
option even though the user can't change its value. The help page for
HID_LOGITECH will not explain why the value can't be changed. It only
says, for example:
Depends on: INPUT [=y] && HID [=y]
and that leaves the user puzzled about why the option is forced to y.
You might argue that this is a Kconfig bug, but that doesn't make it
less annoying for the user.
Even worse is that some of the sub-options of HID_LOGITECH select
INPUT_FF_MEMLESS, which in turn gets out of control for the user. So,
if you set INPUT=y and HID=y (something most general purpose
distributions would do these days, as both modules would get loaded on
a vast majority of systems otherwise), and you want support for
force-feedback game controllers, you can't have that as a module, it
has to be built-in, regardless of how rare these devices are.
Of course, all this madness goes away once EXPERT is enabled, but then
the rest of the kernel configuration becomes more complex, which
totally voids the original point.
For this reason, I would like all HID device tristate options to be
displayed regardless of EXPERT being set or not. We can let the
default settings still depend on EXPERT, that's not intrusive.
Signed-off-by: Jean Delvare <jdelvare@suse.de> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
HID: uclogic: make input_mapping independent of usb
No need to retrieve the USB handle in input_mapping() when we already
do that in probe. It also allows to use the quirk without having to
add the product ID matching.
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Forest Wilkinson [Fri, 13 Mar 2015 06:58:16 +0000 (23:58 -0700)]
HID: tivo: enable all buttons on the TiVo Slide Pro remote
The linux kernel has supported the TiVo Slide remote control for some time, but
does not recognize the USB ID of the newer Slide Pro. This patch adds the
missing data structures so the newer remote will be recognized by the driver,
thereby allowing the TiVo, LiveTV, and Thumbs Up/Down buttons to be
mapped with a hwdb file.
Seth Forshee [Wed, 11 Mar 2015 22:26:41 +0000 (17:26 -0500)]
HID: multitouch: Add support for button type usage
According to [1], Windows Precision Touchpad devices must supply
a button type usage in the device capabilities feature report. A
value of 0 indicates that the device contains a depressible
button (i.e. it's a click-pad) whereas a value of 1 indicates
a non-depressible button. Add support for this usage and set
INPUT_PROP_BUTTONPAD on the touchpad input device whenever a
depressible button is present.
Jason Gerecke [Wed, 11 Mar 2015 17:25:41 +0000 (10:25 -0700)]
HID: wacom: Add battery presence indicator to wireless tablets
Declare the POWER_SUPPLY_PROP_PRESENT property to provide userspace
with a way to determine if the battery on a wireless tablet is plugged
in. Although current wireless tablets do not explicitly report this
information, it can be inferred from other state information. In
particular, a battery is assumed to be present if any of the following
are true: a non-zero battery level reported, the battery is reported as
charging, or the tablet is operating wirelessly.
Note: The last condition above may not strictly hold for the Graphire
Wireless (it charges from a DC barrel jack instead of a USB port), but I
do not know what is reported in the no-battery condition.
Signed-off-by: Jason Gerecke <killertofu@gmail.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Jason Gerecke [Fri, 6 Mar 2015 19:47:43 +0000 (11:47 -0800)]
HID: wacom: Status packet provides 'charging', not 'powered' bit
The status packet for tablets which can use a wireless module contains a
bit that is set if the battery is charging. This bit will be 0 if either
a battery is not present or if the battery has reached full charge. Note
that the charging circuit may continue to charge the battery for a short
time after reaching "100%".
Jason Gerecke [Fri, 6 Mar 2015 19:47:42 +0000 (11:47 -0800)]
HID: wacom: Report battery status for Intuos Pro and Intuos5
Calls the wacom_status_irq function to report battery status for the
Intuos Pro and Intuos5 (in addition to the already-reporting Intuos
and last-generation Bamboo).
Jason Gerecke [Fri, 6 Mar 2015 19:47:41 +0000 (11:47 -0800)]
HID: wacom: Provide battery charge state to system over USB if available
If a wireless adapter (which contains the charging circuitry) is
detected as being attached to the tablet then create a new battery
interface and update its status as data is reported. Also destroy the
battery if the adapter goes away.
Tablets like the Intuos, Intuos Pro, and Bamboo have a connector for an
optional wireless module that can be connected on the fly. The presence
(or absence) of this module is indicated in a status report recieved
from the tablet. This patch adds a workqueue function that will create
or destroy a power_supply object at runtime to match the current state
of the WACOM_QUIRK_BATTERY flag.
Jason Gerecke [Fri, 6 Mar 2015 19:47:39 +0000 (11:47 -0800)]
HID: wacom: Centralize updating of wacom_wac battery status
Has the 'wacom_notify_battery' function take on the job of detecting if
updating the power supply is necessary to remove multiple
nearly-identical 'if' blocks.
Jason Gerecke [Fri, 6 Mar 2015 19:47:38 +0000 (11:47 -0800)]
HID: wacom: Move handling of Intuos status packets to seperate function
In addition to the touchswitch state for "Intuos", these packets are
also sent by the Intuos Pro, Intuos5, and last-generation Bamboo
tablets when using a wired connection. They contain, among other
things, information about the optional wireless module and battery
charge state (to be supported in subsuquent patches).
Dmitry Torokhov [Wed, 4 Mar 2015 22:10:26 +0000 (14:10 -0800)]
HID: map telephony usage page
Currently HID code maps usages from telephony page into BTN_0, BTN_1, etc
keys which get interpreted by mousedev and userspace as left/right/middle
button clicks, which is not really helpful.
This change adds mappings for usages that have corresponding input event
definitions, and leaves the rest unmapped. This can be changed when
there are userspace consumers for more telephony usages.
In commit 08177f4 (HID: uclogic: merge hid-huion driver in hid-uclogic)
HID_HUION depends explicitely on USB_HID because it contained quite some
USB-isms.
Now that hid-uclogic is the new home of Huion tablets, we need to also
add the dependency on USB_HID to this driver.
Reported-by: Jiri Kosina <jkosina@suse.cz> Reported-by: kbuild test robot <fengguang.wu@intel.com> Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
HID: uclogic: actually invert the in-range bit for huion tablets only
This hack is only needed for Huion tablets. It does not seem to have
any effect on the other tablets handled by this device right now, but
it's better to check for the product id sooner than discovering that
we have messed up one tablet later.
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com> Reviewed-by: Nikolai Kondrashov <spbnick@gmail.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
NO_EMPTY_INPUT is useful when MULTI_INPUT is set. It prevents to create
empty input nodes that user space does not know what to do with.
It does not seem to be required at the moment, this is just a preventive
patch. This check is only made during the plug of the device, so it does
not hurt to have it.
Reviewed-by: Nikolai Kondrashov <spbnick@gmail.com> Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
HID: uclogic: name the input nodes based on their tool
We append "Pen", "Pad", "Mouse", "Keyboard", "Consumer Control" or
"System Control" suffix to the appropriate input node to match what
the Wacom driver does and be more convenient for the user to know
which one is which.
Reviewed-by: Nikolai Kondrashov <spbnick@gmail.com> Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Based on a patch from: Nikolai Kondrashov <Nikolai.Kondrashov@redhat.com>
Most of the tablets handled by hid-uclogic already use MULTI_INPUT.
For the ones which are not quirked in usbhid/hidquirks, they have a
custom report descriptor which contains only one report per HID
interface. For those tablets HID_QUIRK_MULTI_INPUT is transparent.
According to https://github.com/DIGImend/tablets, the only problematic
tablet currently handled by hid-uclogic is the TWHA60 v3. This tablet
presents different report descriptors from the ones currently quirked.
This is not a problem per se, given that this tablet is not supported
currently in this version (it needs the same command as a Huion to
start forwarding events).
Reviewed-by: Nikolai Kondrashov <spbnick@gmail.com> Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Ping Cheng [Fri, 20 Feb 2015 22:27:30 +0000 (14:27 -0800)]
HID: wacom: rely on actual touch down count to decide touch_down
touch_down is a flag to indicate if there are touches on tablet
or not. Since one set of touch events may be posted over more
than one data packet/touch frame, and pen may come in proximity
while touch events are partially sent, counting all touch events
for the set reflects the actual status of touch_down.
Ping Cheng [Fri, 20 Feb 2015 22:25:58 +0000 (14:25 -0800)]
HID: wacom: do not send pen events before touch is up/forced out
If pen comes in proximity while touch is down, we force touch up
before sending pen events. Otherwise, there can be unfinished
touch events compete with pen events. This idea has been fully
implemented for Tablet PCs. But other tablets that support both
pen and touch are not fully considered.
HID: wacom: add full support of the Wacom Bamboo PAD
The stylus of this device works just fine out of the box.
The touch is seen by default as a mouse with relative events and some
gestures.
The wireless and the wired version have slightly different firmwares, but
the debug mode 2 on the feature 2 is common to the 2 devices. In this mode,
all the reports are emitted through the debug interface (pen, raw touch
and mouse emulation), so we have to re-route manually the events.
We keep the Pen interface as a HID_GENERIC one because it works, and only
parse the raw touches while discarding the mouse emulation & gestures.
Switching the default in raw mode allows us to have a consistent user
experience accross all the multitouch touchpads (and enable the touch part
of the devices).
Note that the buttons of this devices are reported through the touch
interface. There is no 'Pad' interface. It seemed more natural to have
the BTN_LEFT and BTN_RIGHT reported with the touch because they are
placed under the touch interface and it looks like they belong to the
touch part.
HID: wacom: store the hid_device pointers of the sibling devices
The Bamboo PAD in debug mode needs to re-route events from the debug
interface to the Pen interface. This can be easily done with
hid_input_report(), but that means that we need to keep a reference to
the various hid_devices.
There should be only one touch and one pen interface per physical tablet,
so there is no need to keep a list of hid-devices, plain pointers are
sufficient.
Andrew Duggan [Wed, 25 Feb 2015 01:36:50 +0000 (17:36 -0800)]
HID: rmi: disable palm detect gesture when present
A touchpad may have firmware based palm detection code enabled which
suppresses 2D data from being reported when the firmware believes a palm is
on the touchpad. This functionality is meant to be used in mouse mode without
a driver. When a driver is present, the driver can do a better job of
determining if a contact is a palm. If this gesture is enabled on a touchpad
operating in rmi mode then the firmware will not properly clear the palm detect
interrupt, causing the touchpad to interrupt indefinately. This patch disables
the palm detect gesture when the touchpad is operating in rmi mode.
Andrew Duggan [Wed, 25 Feb 2015 01:36:49 +0000 (17:36 -0800)]
HID: rmi: disable dribble packets on Synaptics touchpads
When a finger is lifted from a Synaptics touchpad the firmware will continue
to interrupts for up to a second. These additional interrupts are know and
dribble interrupts. Since the data read from the touchpad does not change
the input subsystem only reports a single event. This makes the servicing of
dribble interrupts on Linux unnecessary. This patch simply disables dribble
interupts when configuring the touchpad.
Mika Westerberg [Mon, 23 Feb 2015 13:52:45 +0000 (15:52 +0200)]
HID: i2c-hid: Add support for ACPI GPIO interrupts
The HID over I2C specification allows to have the interrupt for a HID
device to be GPIO instead of directly connected to the IO-APIC.
Add support for this so that when the driver does not find proper interrupt
number from the I2C client structure we check if it has ACPI GpioInt()
resource listed in _CRS. If it is found we convert it to an interrupt
number and use it instead.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Olivier Gay [Tue, 17 Feb 2015 13:00:03 +0000 (14:00 +0100)]
HID: expose country code in sysfs
This commit exposes in sysfs the HID country code that is stored in the country
member of hid_device structure. It identifies the country code of localized
hardware.
For example some keyboards use it to exhibit the language of the key layout. It
helps the upper layer to identify the localized hardware and setup the correct
language to use.
For USB HID devices the country code comes for the HID descriptor and for
Bluetooth HID devices it is the HIDCountryCode attribute from the SDP database.
Jakub Sitnicki [Sat, 21 Feb 2015 19:51:08 +0000 (20:51 +0100)]
HID: microsoft: Add ID for NE7K wireless keyboard
Microsoft Natural Wireless Ergonomic Keyboard 7000 has special My
Favorites 1..5 keys which are handled through a vendor-defined usage
page (0xff05).
Apply MS_ERGONOMY quirks handling to USB PID 0x071d (Microsoft Microsoft
2.4GHz Transceiver V1.0) so that the My Favorites 1..5 keys are reported
as KEY_F14..18 events.
HID: hid-sensor-hub: Enhance feature report set API
Current API only allows setting one offset in the field. This API
is extended to set multiple offsets in the field report.
Also update parameters in the users of this API.
HID: hid-sensor-hub: Enhance get feature report API
Some hid sensor feature report can contain more than one reports.
This API can now support receiving multiple values from the feature
report.
Also update the parameters in the users of this API.
Add additional flag to read in async mode. In this mode the caller will get
reply via registered callback for capture_sample. Callbacks can be registered
using sensor_hub_register_callback function. The usage id parameter of the
capture_sample can be matched with the usage id of the requested attribute.
HID: hid-sensor-hub: Add support for application collection
Section 4.2.5 of HID Sensor hub specification allows two methods
defining sensor devices.
- Each sensor device by its own collection
- A top level application collection object, including multiple
sensors.
In the first method, each sensor can be in its own sensor application
collection without a physical collection.
In the second method there is a usage id for collection type, which
is defined as an application collection, with multiple physical
collections in it. It is possible to define fusion sensor with this
and may have its own handler. If there is a callback registered
for the collection type, then forward all reports for sensors in
its collection to this handler.
Current implementation only allows one outstanding synchronous read.
This is a performance hit when user mode is requesting raw reads
of sensor attributes on multiple sensors together.
This change changes the mutex lock to per hid sensor hub device instead
of global lock. Although request to hid sensor hub is serialized, there
can be multiple outstanding read requests pending for responses via
hid reports.
Seth Forshee [Fri, 20 Feb 2015 17:45:11 +0000 (11:45 -0600)]
HID: i2c-hid: Limit reads to wMaxInputLength bytes for input events
d1c7e29e8d27 (HID: i2c-hid: prevent buffer overflow in early IRQ)
changed hid_get_input() to read ihid->bufsize bytes, which can be
more than wMaxInputLength. This is the case with the Dell XPS 13
9343, and it is causing events to be missed. In some cases the
missed events are releases, which can cause the cursor to jump or
freeze, among other problems. Limit the number of bytes read to
min(wMaxInputLength, ihid->bufsize) to prevent such problems.
Fixes: d1c7e29e8d27 "HID: i2c-hid: prevent buffer overflow in early IRQ" Signed-off-by: Seth Forshee <seth.forshee@canonical.com> Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Per-controller spinlock needs to be properly initialized during device probe.
[jkosina@suse.cz: massage changelog]
[jkosina@suse.cz: drop hunk that has already been applied by previous
patch] Signed-off-by: Frank Praznik <frank.praznik@oh.rr.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Jiri Kosina [Mon, 23 Feb 2015 10:15:44 +0000 (11:15 +0100)]
HID: sony: initialize sony_dev_list_lock properly
sony_dev_list_lock spinlock (which was introduced in d2d782fccee ("HID: sony:
Prevent duplicate controller connections") is not being initialized properly.
Fix that.
Reported-by: Pavel Machek <pavel@ucw.cz> Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Linus Torvalds [Mon, 23 Feb 2015 02:21:14 +0000 (18:21 -0800)]
Linux 4.0-rc1
.. after extensive statistical analysis of my G+ polling, I've come to
the inescapable conclusion that internet polls are bad.
Big surprise.
But "Hurr durr I'ma sheep" trounced "I like online polls" by a 62-to-38%
margin, in a poll that people weren't even supposed to participate in.
Who can argue with solid numbers like that? 5,796 votes from people who
can't even follow the most basic directions?
In contrast, "v4.0" beat out "v3.20" by a slimmer margin of 56-to-44%,
but with a total of 29,110 votes right now.
Now, arguably, that vote spread is only about 3,200 votes, which is less
than the almost six thousand votes that the "please ignore" poll got, so
it could be considered noise.
Linus Torvalds [Mon, 23 Feb 2015 02:05:13 +0000 (18:05 -0800)]
Merge tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Pull ext4 fixes from Ted Ts'o:
"Ext4 bug fixes.
We also reserved code points for encryption and read-only images (for
which the implementation is mostly just the reserved code point for a
read-only feature :-)"
* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
ext4: fix indirect punch hole corruption
ext4: ignore journal checksum on remount; don't fail
ext4: remove duplicate remount check for JOURNAL_CHECKSUM change
ext4: fix mmap data corruption in nodelalloc mode when blocksize < pagesize
ext4: support read-only images
ext4: change to use setup_timer() instead of init_timer()
ext4: reserve codepoints used by the ext4 encryption feature
jbd2: complain about descriptor block checksum errors
Linus Torvalds [Mon, 23 Feb 2015 01:42:14 +0000 (17:42 -0800)]
Merge branch 'for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull more vfs updates from Al Viro:
"Assorted stuff from this cycle. The big ones here are multilayer
overlayfs from Miklos and beginning of sorting ->d_inode accesses out
from David"
* 'for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (51 commits)
autofs4 copy_dev_ioctl(): keep the value of ->size we'd used for allocation
procfs: fix race between symlink removals and traversals
debugfs: leave freeing a symlink body until inode eviction
Documentation/filesystems/Locking: ->get_sb() is long gone
trylock_super(): replacement for grab_super_passive()
fanotify: Fix up scripted S_ISDIR/S_ISREG/S_ISLNK conversions
Cachefiles: Fix up scripted S_ISDIR/S_ISREG/S_ISLNK conversions
VFS: (Scripted) Convert S_ISLNK/DIR/REG(dentry->d_inode) to d_is_*(dentry)
SELinux: Use d_is_positive() rather than testing dentry->d_inode
Smack: Use d_is_positive() rather than testing dentry->d_inode
TOMOYO: Use d_is_dir() rather than d_inode and S_ISDIR()
Apparmor: Use d_is_positive/negative() rather than testing dentry->d_inode
Apparmor: mediated_filesystem() should use dentry->d_sb not inode->i_sb
VFS: Split DCACHE_FILE_TYPE into regular and special types
VFS: Add a fallthrough flag for marking virtual dentries
VFS: Add a whiteout dentry type
VFS: Introduce inode-getting helpers for layered/unioned fs environments
Infiniband: Fix potential NULL d_inode dereference
posix_acl: fix reference leaks in posix_acl_create
autofs4: Wrong format for printing dentry
...
Linus Torvalds [Sun, 22 Feb 2015 17:57:16 +0000 (09:57 -0800)]
Merge branch 'fixes' of git://ftp.arm.linux.org.uk/~rmk/linux-arm
Pull ARM fix from Russell King:
"Just one fix this time around. __iommu_alloc_buffer() can cause a
BUG() if dma_alloc_coherent() is called with either __GFP_DMA32 or
__GFP_HIGHMEM set. The patch from Alexandre addresses this"
* 'fixes' of git://ftp.arm.linux.org.uk/~rmk/linux-arm:
ARM: 8305/1: DMA: Fix kzalloc flags in __iommu_alloc_buffer()
Al Viro [Sun, 22 Feb 2015 03:05:11 +0000 (22:05 -0500)]
debugfs: leave freeing a symlink body until inode eviction
As it is, we have debugfs_remove() racing with symlink traversals.
Supply ->evict_inode() and do freeing there - inode will remain
pinned until we are done with the symlink body.
And rip the idiocy with checking if dentry is positive right after
we'd verified debugfs_positive(), which is a stronger check...
Cc: stable@vger.kernel.org Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
trylock_super(): replacement for grab_super_passive()
I've noticed significant locking contention in memory reclaimer around
sb_lock inside grab_super_passive(). Grab_super_passive() is called from
two places: in icache/dcache shrinkers (function super_cache_scan) and
from writeback (function __writeback_inodes_wb). Both are required for
progress in memory allocator.
Grab_super_passive() acquires sb_lock to increment sb->s_count and check
sb->s_instances. It seems sb->s_umount locked for read is enough here:
super-block deactivation always runs under sb->s_umount locked for write.
Protecting super-block itself isn't a problem: in super_cache_scan() sb
is protected by shrinker_rwsem: it cannot be freed if its slab shrinkers
are still active. Inside writeback super-block comes from inode from bdi
writeback list under wb->list_lock.
This patch removes locking sb_lock and checks s_instances under s_umount:
generic_shutdown_super() unlinks it under sb->s_umount locked for write.
New variant is called trylock_super() and since it only locks semaphore,
callers must call up_read(&sb->s_umount) instead of drop_super(sb) when
they're done.
Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
David Howells [Thu, 29 Jan 2015 12:02:36 +0000 (12:02 +0000)]
fanotify: Fix up scripted S_ISDIR/S_ISREG/S_ISLNK conversions
Fanotify probably doesn't want to watch autodirs so make it use d_can_lookup()
rather than d_is_dir() when checking a dir watch and give an error on fake
directories.
Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
David Howells [Thu, 29 Jan 2015 12:02:35 +0000 (12:02 +0000)]
VFS: (Scripted) Convert S_ISLNK/DIR/REG(dentry->d_inode) to d_is_*(dentry)
Convert the following where appropriate:
(1) S_ISLNK(dentry->d_inode) to d_is_symlink(dentry).
(2) S_ISREG(dentry->d_inode) to d_is_reg(dentry).
(3) S_ISDIR(dentry->d_inode) to d_is_dir(dentry). This is actually more
complicated than it appears as some calls should be converted to
d_can_lookup() instead. The difference is whether the directory in
question is a real dir with a ->lookup op or whether it's a fake dir with
a ->d_automount op.
In some circumstances, we can subsume checks for dentry->d_inode not being
NULL into this, provided we the code isn't in a filesystem that expects
d_inode to be NULL if the dirent really *is* negative (ie. if we're going to
use d_inode() rather than d_backing_inode() to get the inode pointer).
Note that the dentry type field may be set to something other than
DCACHE_MISS_TYPE when d_inode is NULL in the case of unionmount, where the VFS
manages the fall-through from a negative dentry to a lower layer. In such a
case, the dentry type of the negative union dentry is set to the same as the
type of the lower dentry.
However, if you know d_inode is not NULL at the call site, then you can use
the d_is_xxx() functions even in a filesystem.
There is one further complication: a 0,0 chardev dentry may be labelled
DCACHE_WHITEOUT_TYPE rather than DCACHE_SPECIAL_TYPE. Strictly, this was
intended for special directory entry types that don't have attached inodes.
The following perl+coccinelle script was used:
use strict;
my @callers;
open($fd, 'git grep -l \'S_IS[A-Z].*->d_inode\' |') ||
die "Can't grep for S_ISDIR and co. callers";
@callers = <$fd>;
close($fd);
unless (@callers) {
print "No matches\n";
exit(0);
}
David Howells [Thu, 29 Jan 2015 12:02:31 +0000 (12:02 +0000)]
Apparmor: Use d_is_positive/negative() rather than testing dentry->d_inode
Use d_is_positive(dentry) or d_is_negative(dentry) rather than testing
dentry->d_inode as the dentry may cover another layer that has an inode when
the top layer doesn't or may hold a 0,0 chardev that's actually a whiteout.
Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
David Howells [Thu, 29 Jan 2015 12:02:28 +0000 (12:02 +0000)]
VFS: Add a fallthrough flag for marking virtual dentries
Add a DCACHE_FALLTHRU flag to indicate that, in a layered filesystem, this is
a virtual dentry that covers another one in a lower layer that should be used
instead. This may be recorded on medium if directory integration is stored
there.
The flag can be set with d_set_fallthru() and tested with d_is_fallthru().
Original-author: Valerie Aurora <vaurora@redhat.com> Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
David Howells [Thu, 29 Jan 2015 12:02:27 +0000 (12:02 +0000)]
VFS: Add a whiteout dentry type
Add DCACHE_WHITEOUT_TYPE and provide a d_is_whiteout() accessor function. A
d_is_miss() accessor is also added for ordinary cache misses and
d_is_negative() is modified to indicate either an ordinary miss or an enforced
miss (whiteout).
Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
David Howells [Thu, 29 Jan 2015 12:02:27 +0000 (12:02 +0000)]
VFS: Introduce inode-getting helpers for layered/unioned fs environments
Introduce some function for getting the inode (and also the dentry) in an
environment where layered/unioned filesystems are in operation.
The problem is that we have places where we need *both* the union dentry and
the lower source or workspace inode or dentry available, but we can only have
a handle on one of them. Therefore we need to derive the handle to the other
from that.
The idea is to introduce an extra field in struct dentry that allows the union
dentry to refer to and pin the lower dentry.
Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Linus Torvalds [Sun, 22 Feb 2015 03:41:38 +0000 (19:41 -0800)]
Merge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus
Pull MIPS updates from Ralf Baechle:
"This is the main pull request for MIPS:
- a number of fixes that didn't make the 3.19 release.
- a number of cleanups.
- preliminary support for Cavium's Octeon 3 SOCs which feature up to
48 MIPS64 R3 cores with FPU and hardware virtualization.
- support for MIPS R6 processors.
Revision 6 of the MIPS architecture is a major revision of the MIPS
architecture which does away with many of original sins of the
architecture such as branch delay slots. This and other changes in
R6 require major changes throughout the entire MIPS core
architecture code and make up for the lion share of this pull
request.
- finally some preparatory work for eXtendend Physical Address
support, which allows support of up to 40 bit of physical address
space on 32 bit processors"
[ Ahh, MIPS can't leave the PAE brain damage alone. It's like
every CPU architect has to make that mistake, but pee in the snow
by changing the TLA. But whether it's called PAE, LPAE or XPA,
it's horrid crud - Linus ]
* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus: (114 commits)
MIPS: sead3: Corrected get_c0_perfcount_int
MIPS: mm: Remove dead macro definitions
MIPS: OCTEON: irq: add CIB and other fixes
MIPS: OCTEON: Don't do acknowledge operations for level triggered irqs.
MIPS: OCTEON: More OCTEONIII support
MIPS: OCTEON: Remove setting of processor specific CVMCTL icache bits.
MIPS: OCTEON: Core-15169 Workaround and general CVMSEG cleanup.
MIPS: OCTEON: Update octeon-model.h code for new SoCs.
MIPS: OCTEON: Implement DCache errata workaround for all CN6XXX
MIPS: OCTEON: Add little-endian support to asm/octeon/octeon.h
MIPS: OCTEON: Implement the core-16057 workaround
MIPS: OCTEON: Delete unused COP2 saving code
MIPS: OCTEON: Use correct instruction to read 64-bit COP0 register
MIPS: OCTEON: Save and restore CP2 SHA3 state
MIPS: OCTEON: Fix FP context save.
MIPS: OCTEON: Save/Restore wider multiply registers in OCTEON III CPUs
MIPS: boot: Provide more uImage options
MIPS: Remove unneeded #ifdef __KERNEL__ from asm/processor.h
MIPS: ip22-gio: Remove legacy suspend/resume support
mips: pci: Add ifdef around pci_proc_domain
...
Linus Torvalds [Sun, 22 Feb 2015 03:21:54 +0000 (19:21 -0800)]
Merge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Pull ARM SoC fixes from Arnd Bergmann:
"A few fixes that came in too late to make it into the first set of
pull requests but would still be nice to have in -rc1.
The majority of these are trivial build fixes for bugs that I found
myself using randconfig testing, and a set of two patches from Uwe to
mark DT strings as 'const' where appropriate, to resolve inconsistent
section attributes"
* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
ARM: make of_device_ids const
ARM: make arrays containing machine compatible strings const
ARM: mm: Remove Kconfig symbol CACHE_PL310
ARM: rockchip: force built-in regulator support for PM
ARM: mvebu: build armada375-smp code conditionally
ARM: sti: always enable RESET_CONTROLLER
ARM: rockchip: make rockchip_suspend_init conditional
ARM: ixp4xx: fix {in,out}s{bwl} data types
ARM: prima2: do not select SMP_ON_UP
ARM: at91: fix pm declarations
ARM: davinci: multi-soc kernels require AUTO_ZRELADDR
ARM: davinci: davinci_cfg_reg cannot be init
ARM: BCM: put back ARCH_MULTI_V7 dependency for mobile
ARM: vexpress: use ARM_CPU_SUSPEND if needed
ARM: dts: add I2C device nodes for Broadcom Cygnus
ARM: dts: BCM63xx: fix L2 cache properties