OSDN Git Service

android-x86/kernel.git
5 years agoHID: logitech-hidpp: make .probe usbhid capable
Benjamin Tissoires [Sat, 20 Apr 2019 11:22:05 +0000 (13:22 +0200)]
HID: logitech-hidpp: make .probe usbhid capable

The current custom solution for the G920 is not the best because
hid_hw_start() is not called at the end of the .probe().
It means that any configuration retrieved after the initial hid_hw_start
would not be exposed to user space without races.

We can simply force hid_hw_start to just enable the transport layer by
not using a connect_mask. This way, we can have a common path between
USB, Unifying and Bluetooth devices.

With this change, we can now support the non DJ receivers for low end
devices, which will allow us to fetch the actual names of the paired
device (instead of 'Logitech Wireless Receiver')

Tested with a M185 with the non unifying receiver, a T650 and many other
unifying devices, and the T651 over Bluetooth.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-hidpp: allow non HID++ devices to be handled by this module
Benjamin Tissoires [Sat, 20 Apr 2019 11:22:04 +0000 (13:22 +0200)]
HID: logitech-hidpp: allow non HID++ devices to be handled by this module

On the gaming mice, there are 2 interfaces, one for the mouse and one
for the macros. Better allow everybody to go through hid-logitech-hidpp
than trying to be smarter.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: add support for Logitech Bluetooth Mini-Receiver
Hans de Goede [Sat, 20 Apr 2019 11:22:03 +0000 (13:22 +0200)]
HID: logitech-dj: add support for Logitech Bluetooth Mini-Receiver

Add support for the Logitech Bluetooth Mini-Receiver in HID proxy mode

This requires some special handing in dj_find_receiver_dev because the
BT Mini-Receiver contains a built-in hub and has separate USB-devices
for the keyboard and mouse interfaces, rather then using 2 interfaces on
a single USB device. Otherwise this receiver works identical to the
standard non-unifying nano receivers.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: make appending of the HID++ descriptors conditional
Hans de Goede [Sat, 20 Apr 2019 11:22:02 +0000 (13:22 +0200)]
HID: logitech-dj: make appending of the HID++ descriptors conditional

Make the appending of the HID++ descriptors in logi_dj_ll_parse
conditional. This is a preparation patch for adding support for the
Logitech mini Bluetooth receiver in HID proxy mode (its default mode),
where some of the paired devices may not be Logitech devices and thus may
not be HID++ capable.

This uses a fake bit 63 in reports_supported, which is changed from an
u32 to an u64 for this. Bits <= 31 are not usable for this because that
would cause a behavioral change in logi_dj_recv_forward_null_report.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: remove false-positive error on double queueing of delayed-work
Hans de Goede [Sat, 20 Apr 2019 11:22:01 +0000 (13:22 +0200)]
HID: logitech-dj: remove false-positive error on double queueing of delayed-work

The various functions queueing work-items do not check there already is a
work-item queued before calling schedule_work(), as such they may race
with each-other and with the re-queuing done by the delayedwork_callback
itself.

This is fine as the delayedwork_callback simply is a nop if scheduled once
too much. I've actually seen the false-positive hid_err for this trigger
in practice, so lets remove it.

While at it also remove the somewhat overzealous debugging around the
schedule_work() calls.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: pick a better name for non-unifying receivers
Hans de Goede [Sat, 20 Apr 2019 11:22:00 +0000 (13:22 +0200)]
HID: logitech-dj: pick a better name for non-unifying receivers

hidpp_unifying_get_name() does not work for devices attached to
non-unifying receivers. Since we do get a device-type in the device-
connection report, we can pick a better name for these devices in
hid-logitech-dj.c .

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: deal with some KVMs adding an extra interface to the usbdev
Hans de Goede [Sat, 20 Apr 2019 11:21:59 +0000 (13:21 +0200)]
HID: logitech-dj: deal with some KVMs adding an extra interface to the usbdev

My Aten cs1764a KVM adds an extra interface to the receiver through which
it forwards mouse events, if a separate mouse is plugged in next to the
receiver dongle. This interface is present even if no extra mouse is
plugged in.

logitech-dj trying to handle this extra interface causes mouse events send
through the extra interface to not be properly handled.

This commit fixes this by treating any extra interfaces as hid-generic
interfaces.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: replace dev_err calls with hid_err calls
Hans de Goede [Sat, 20 Apr 2019 11:21:58 +0000 (13:21 +0200)]
HID: logitech-dj: replace dev_err calls with hid_err calls

Use hid_err consistently everywhere.

While at it also tweak some of the messages for clarity, to
consistently have a space after a ':' and in some cases to fit
within 80 chars.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: add support for 27 MHz mouse-only receivers
Hans de Goede [Sat, 20 Apr 2019 11:21:57 +0000 (13:21 +0200)]
HID: logitech-dj: add support for 27 MHz mouse-only receivers

27 MHz mouse-only receivers send an unnumbered input report with the mouse
data, add special handling for this and add the c51b product-id to the
logi_dj_receivers table.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: add support for 27 MHz receivers
Hans de Goede [Sat, 20 Apr 2019 11:21:56 +0000 (13:21 +0200)]
HID: logitech-dj: add support for 27 MHz receivers

Most Logitech wireless keyboard and mice using the 27 MHz are hidpp10
devices, add support to logitech-dj for their receivers.

Doing so leads to 2 improvements:

1) All these devices share the same USB product-id for their receiver,
making it impossible to properly map some special keys / buttons
which differ from device to device. Adding support to logitech-dj to
see these as hidpp10 devices allows us to get the actual device-id
from the keyboard / mouse.

2) It enables battery-monitoring of these devices

This patch uses a new HID group for 27Mhz devices, since the logitech-hidpp
code needs to be able to differentiate them from other devices instantiated
by the logitech-dj code.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: add support for the gaming unifying receiver
Benjamin Tissoires [Sat, 20 Apr 2019 11:21:55 +0000 (13:21 +0200)]
HID: logitech-dj: add support for the gaming unifying receiver

This receiver is almost identical to the normal unifying ones except:
- it is supposed to be paired to only one device (for performance reasons)
- the mice reports have a greater ranges in their values, so they are
  using a different report ID.

Tested on a G403 and a G900.

Co-authored-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: add support for non unifying receivers
Hans de Goede [Sat, 20 Apr 2019 11:21:54 +0000 (13:21 +0200)]
HID: logitech-dj: add support for non unifying receivers

We emulate the DJ functionality through the driver.

The receiver supports "fake device arrival" which behaves
like the probing of DJ devices.

A non-unifying receiver has 2 USB interfaces, the first one generates
standard keypresses and is compatible with the USB Keyboard Boot Subclass.
The second interface sends events for the mouse and special keys such as
the consumer-page keys. Events are split this way for BIOS / Windows /
generic-hid driver compatibility. This split does not actually match with
which device the event originate from, e.g. the consumer-page key events
originate from the keyboard but are delivered on the mouse interface.

To make sure the events are actually delivered to the dj_device
representing the originating device, we pick which dj_dev to forward
a "regular" input-report to based on the report-number, rather
then based on the originating interface.

Co-authored-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: add logi_dj_recv_queue_unknown_work helper
Hans de Goede [Sat, 20 Apr 2019 11:21:53 +0000 (13:21 +0200)]
HID: logitech-dj: add logi_dj_recv_queue_unknown_work helper

Add a logi_dj_recv_queue_unknown_work helper and implement query
rate-limiting inside this helper.

The motivations behind this are:

1) We need to queue workitems for reports with no place to forward them
from more places with the upcoming non-unifying receiver support, hence
the addition of the helper function.

2) When we've missed a pairing info report (or there is a race between
the report and input-events) and the input report is e.g. from a mouse
being moved, we will get a lot of these before we've finished (re-)
querying and enumerating the devices, hence the rate-limiting.

Note this also removes the:

if (!djrcv_dev->paired_dj_devices[hidpp_report->device_index])

check previously guarding the sending of an unknown workitem, the caller
of logi_dj_recv_queue_notification already does this check before calling
logi_dj_recv_queue_notification.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: support sharing struct dj_receiver_dev between USB-interfaces
Hans de Goede [Sat, 20 Apr 2019 11:21:52 +0000 (13:21 +0200)]
HID: logitech-dj: support sharing struct dj_receiver_dev between USB-interfaces

dj/HID++ receivers are really a single logical entity, but for BIOS/Windows
compatibility they have multiple USB interfaces. For the upcoming
non-unifying receiver support, we need to listen for events from / bind to
all USB-interfaces of the receiver.

This commit add support to the logitech-dj code for creating a single
dj_receiver_dev struct for all interfaces belonging to a single
USB-device / receiver, in preparation for adding non-unifying receiver
support.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: rename dj_receiver_dev.hdev to dj_receiver_dev.hidpp
Hans de Goede [Sat, 20 Apr 2019 11:21:51 +0000 (13:21 +0200)]
HID: logitech-dj: rename dj_receiver_dev.hdev to dj_receiver_dev.hidpp

For the upcoming non-unifying receiver support, we are going to bind to
all USB-interfaces of a receiver, sharing a single struct dj_receiver_dev
between the interfaces. This means that dj_receiver_dev will contain
multiple pointers to a struct hid_device. Rename the current hdev member
to hidpp to prepare for this.

While at it switch dev_err calls which we are touching anyways from
dev_err to hid_err.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: protect the paired_dj_devices access in add_djhid_dev with the...
Hans de Goede [Sat, 20 Apr 2019 11:21:50 +0000 (13:21 +0200)]
HID: logitech-dj: protect the paired_dj_devices access in add_djhid_dev with the lock

This protects against logi_dj_recv_add_djhid_device, adding a device to
paired_dj_devices from the delayedwork callback, racing versus
logi_dj_raw_event trying to access that device.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: remove unused querying_devices variable
Hans de Goede [Sat, 20 Apr 2019 11:21:49 +0000 (13:21 +0200)]
HID: logitech-dj: remove unused querying_devices variable

querying_devices is never set, so it can safely be removed.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: do not schedule the dj report itself
Benjamin Tissoires [Sat, 20 Apr 2019 11:21:48 +0000 (13:21 +0200)]
HID: logitech-dj: do not schedule the dj report itself

This is a preparatory patch for handling non DJ (HID++ only) receivers,
through this module. We can not use the dj_report in the delayed work
callback as the HID++ notifications are different both in size and meaning.

There should be no functional change.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: remove USB dependency
Benjamin Tissoires [Sat, 20 Apr 2019 11:21:47 +0000 (13:21 +0200)]
HID: logitech-dj: remove USB dependency

It is better to rely on the actual content of the report descriptors
to enable or not a HID interface.

While at it, remove the other USB dependency to have a fully USB
agnostic driver.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: declare and use a few HID++ 1.0 constants
Benjamin Tissoires [Sat, 20 Apr 2019 11:21:46 +0000 (13:21 +0200)]
HID: logitech-dj: declare and use a few HID++ 1.0 constants

For the non DJ receivers, we are going to need to re-use those constants,
better have them properly defined.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: use BIT() macro for RF Report types
Benjamin Tissoires [Sat, 20 Apr 2019 11:21:45 +0000 (13:21 +0200)]
HID: logitech-dj: use BIT() macro for RF Report types

Use BIT() macro for RF Report types.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: fix variable naming in logi_dj_hidpp_event
Benjamin Tissoires [Sat, 20 Apr 2019 11:21:44 +0000 (13:21 +0200)]
HID: logitech-dj: fix variable naming in logi_dj_hidpp_event

we are not dealing with a dj_report but a hidpp_event.
We don't need all of the struct description in this function, but having
the variable named `dj_report` feels weird.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech-dj: reshuffle logi_dj_recv_forward_*
Benjamin Tissoires [Sat, 20 Apr 2019 11:21:43 +0000 (13:21 +0200)]
HID: logitech-dj: reshuffle logi_dj_recv_forward_*

logi_dj_recv_forward_report() was only intended for DJ reports.
logi_dj_recv_forward_hidpp() is more generic at forwarding random HID
reports.

So rename logi_dj_recv_forward_report() into logi_dj_recv_forward_dj()
and logi_dj_recv_forward_hidpp() into logi_dj_recv_forward_report().

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: logitech: Stop setting drvdata to NULL on probe failure and remove
Benjamin Tissoires [Sat, 20 Apr 2019 11:21:42 +0000 (13:21 +0200)]
HID: logitech: Stop setting drvdata to NULL on probe failure and remove

There is no need to set drvdata to NULL on probe failure and remove,
the driver-core already does this for us.

[hdegoede@redhat.com: Isolate Logitech changes into a separate patch]
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: quirks: do not blacklist Logitech devices
Benjamin Tissoires [Sat, 20 Apr 2019 11:21:41 +0000 (13:21 +0200)]
HID: quirks: do not blacklist Logitech devices

I am actually suggesting people to not populate this list, and I should
probably start to apply my advices to myself.

The end result means that if your initrd is lacking hid-logitech-dj
or hid-logitech-hidpp, but still contains hid-generic, then your
keyboard will work during pre-init.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoMerge branch 'for-5.1/upstream-fixes' into for-5.2/logitech
Benjamin Tissoires [Tue, 23 Apr 2019 15:52:33 +0000 (17:52 +0200)]
Merge branch 'for-5.1/upstream-fixes' into for-5.2/logitech

5 years agoHID: intel-ish-hid: Add Comet Lake PCI device ID
Srinivas Pandruvada [Sat, 20 Apr 2019 02:00:31 +0000 (19:00 -0700)]
HID: intel-ish-hid: Add Comet Lake PCI device ID

Add Comet Lake PCI device ID to the supported device list.

Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoRevert "HID: i2c-hid: Disable runtime PM on Synaptics touchpad"
Hui Wang [Thu, 11 Apr 2019 09:02:03 +0000 (17:02 +0800)]
Revert "HID: i2c-hid: Disable runtime PM on Synaptics touchpad"

This reverts commit 74e7c6c877f620d65a8269692d089bbd066f626c.

It finally turns out the touchpad is an engineering sample and it is
not the Synaptics touchpad. Let us revert this patch otherwise it will
affect the real Synaptics touchpad.

Signed-off-by: Hui Wang <hui.wang@canonical.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: input: add mapping for Assistant key
Dmitry Torokhov [Tue, 2 Apr 2019 16:57:13 +0000 (09:57 -0700)]
HID: input: add mapping for Assistant key

According to HUTRR89 usage 0x1cb from the consumer page was assigned to
allow launching desktop-aware assistant application, so let's add the
mapping.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: i2c-hid: Disable runtime PM on Synaptics touchpad
Hui Wang [Fri, 29 Mar 2019 06:13:23 +0000 (14:13 +0800)]
HID: i2c-hid: Disable runtime PM on Synaptics touchpad

We have a new Dell laptop which has the synaptics I2C touchpad
(06cb:7e7e) on it. After booting up the Linux, the touchpad doesn't
work, there is no interrupt when touching the touchpad, after
disable the runtime PM, everything works well.

I also tried the quirk of I2C_HID_QUIRK_DELAY_AFTER_SLEEP, it is
better after applied this quirk, there are interrupts but data it
reports is invalid.

Signed-off-by: Hui Wang <hui.wang@canonical.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: logitech-hidpp: change low battery level threshold from 31 to 30 percent
Hans de Goede [Fri, 22 Mar 2019 07:41:40 +0000 (08:41 +0100)]
HID: logitech-hidpp: change low battery level threshold from 31 to 30 percent

According to hidpp20_batterylevel_get_battery_info my Logitech K270
keyboard reports only 2 battery levels. This matches with what I've seen
after testing with batteries at varying level of fullness, it always
reports either 5% or 30%.

Windows reports "battery good" for the 30% level. I've captured an USB
trace of Windows reading the battery and it is getting the same info
as the Linux hidpp code gets.

Now that Linux handles these devices as hidpp devices, it reports the
battery as being low as it treats anything under 31% as low, this leads
to the user constantly getting a "Keyboard battery is low" warning from
GNOME3, which is very annoying.

This commit fixes this by changing the low threshold to anything under
30%, which I assume is what Windows does.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: logitech-hidpp: remove hidpp_is_connected()
Hans de Goede [Fri, 22 Mar 2019 07:41:39 +0000 (08:41 +0100)]
HID: logitech-hidpp: remove hidpp_is_connected()

Remove the hidpp_is_connected() function wrapper, and have the callers
directly call hidpp_root_get_protocol_version() instead.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: logitech-hidpp: simplify printing of HID++ version
Hans de Goede [Fri, 22 Mar 2019 07:41:38 +0000 (08:41 +0100)]
HID: logitech-hidpp: simplify printing of HID++ version

Simply always print the HID++ version on hidpp_root_get_protocol_version
success.

This also fixes the version not being printed when a HID++ device
connected through a receiver is already connected when the hidpp driver
is loaded.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: quirks: Fix keyboard + touchpad on Lenovo Miix 630
Jeffrey Hugo [Tue, 26 Mar 2019 16:55:54 +0000 (09:55 -0700)]
HID: quirks: Fix keyboard + touchpad on Lenovo Miix 630

Similar to commit edfc3722cfef ("HID: quirks: Fix keyboard + touchpad on
Toshiba Click Mini not working"), the Lenovo Miix 630 has a combo
keyboard/touchpad device with vid:pid of 04F3:0400, which is shared with
Elan touchpads.  The combo on the Miix 630 has an ACPI id of QTEC0001,
which is not claimed by the elan_i2c driver, so key on that similar to
what was done for the Toshiba Click Mini.

Signed-off-by: Jeffrey Hugo <jeffrey.l.hugo@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: logitech: Handle 0 scroll events for the m560
Peter Hutterer [Tue, 19 Mar 2019 22:48:23 +0000 (08:48 +1000)]
HID: logitech: Handle 0 scroll events for the m560

hidpp_scroll_counter_handle_scroll() doesn't expect a 0-value scroll event, it
gets interpreted as a negative scroll direction event. This can cause scroll
direction resets and thus broken scrolling.

Fixes: 4435ff2f09a2fc ("HID: logitech: Enable high-resolution scrolling on Logitech mice")
Cc: stable@vger.kernel.org # v5.0
Reported-and-tested-by: Aimo Metsälä <aimetsal@outlook.com>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: debug: fix race condition with between rdesc_show() and device removal
He, Bo [Thu, 14 Mar 2019 02:28:21 +0000 (02:28 +0000)]
HID: debug: fix race condition with between rdesc_show() and device removal

There is a race condition that could happen if hid_debug_rdesc_show()
is running while hdev is in the process of going away (device removal,
system suspend, etc) which could result in NULL pointer dereference:

 BUG: unable to handle kernel paging request at 0000000783316040
 CPU: 1 PID: 1512 Comm: getevent Tainted: G     U     O 4.19.20-quilt-2e5dc0ac-00029-gc455a447dd55 #1
 RIP: 0010:hid_dump_device+0x9b/0x160
 Call Trace:
  hid_debug_rdesc_show+0x72/0x1d0
  seq_read+0xe0/0x410
  full_proxy_read+0x5f/0x90
  __vfs_read+0x3a/0x170
  vfs_read+0xa0/0x150
  ksys_read+0x58/0xc0
  __x64_sys_read+0x1a/0x20
  do_syscall_64+0x55/0x110
  entry_SYSCALL_64_after_hwframe+0x49/0xbe

Grab driver_input_lock to make sure the input device exists throughout the
whole process of dumping the rdesc.

[jkosina@suse.cz: update changelog a bit]
Signed-off-by: he, bo <bo.he@intel.com>
Signed-off-by: "Zhang, Jun" <jun.zhang@intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: logitech: check the return value of create_singlethread_workqueue
Kangjie Lu [Thu, 14 Mar 2019 05:24:02 +0000 (00:24 -0500)]
HID: logitech: check the return value of create_singlethread_workqueue

create_singlethread_workqueue may fail and return NULL. The fix checks if it is
NULL to avoid NULL pointer dereference.  Also, the fix moves the call of
create_singlethread_workqueue earlier to avoid resource-release issues.

Signed-off-by: Kangjie Lu <kjlu@umn.edu>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: Increase maximum report size allowed by hid_field_extract()
Kai-Heng Feng [Fri, 8 Mar 2019 05:11:17 +0000 (13:11 +0800)]
HID: Increase maximum report size allowed by hid_field_extract()

Commit 71f6fa90a353 ("HID: increase maximum global item tag report size
to 256") increases the max report size from 128 to 256.

We also need to update the report size in hid_field_extract() otherwise
it complains and truncates now valid report size:
[ 406.165461] hid-sensor-hub 001F:8086:22D8.0002: hid_field_extract() called with n (192) > 32! (kworker/5:1)

BugLink: https://bugs.launchpad.net/bugs/1818547
Fixes: 71f6fa90a353 ("HID: increase maximum global item tag report size to 256")
Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: steam: fix deadlock with input devices.
Rodrigo Rivas Costa [Fri, 15 Mar 2019 19:09:10 +0000 (20:09 +0100)]
HID: steam: fix deadlock with input devices.

When using this driver with the wireless dongle and some usermode
program that monitors every input device (acpid, for example), while
another usermode client opens and closes the low-level device
repeadedly, the system eventually deadlocks.

The reason is that steam_input_register_device() must not be called with
the mutex held, because the input subsystem has its own synchronization
that clashes with this one: it is possible that steam_input_open() is
called before input_register_device() returns, and since
steam_input_open() needs to lock the mutex, it deadlocks.

However we must hold the mutex when calling any function that sends
commands to the controller. If not, random commands end up falling fail.

Reported-by: Simon Gene Gottlieb <simon@gottliebtfreitag.de>
Signed-off-by: Rodrigo Rivas Costa <rodrigorivascosta@gmail.com>
Tested-by: Simon Gene Gottlieb <simon@gottliebtfreitag.de>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
5 years agoHID: uclogic: remove redudant duplicated null check on ver_ptr
Colin Ian King [Sat, 2 Mar 2019 22:23:38 +0000 (22:23 +0000)]
HID: uclogic: remove redudant duplicated null check on ver_ptr

Currently ver_ptr is being null checked twice, once before calling
usb_string and once afterwards.  The second null check is redundant
and can be removed, remove it.

Detected by CoverityScan, CID#1477308 ("Logically dead code")

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: quirks: Drop misused kernel-doc annotation
Andy Shevchenko [Tue, 5 Mar 2019 11:15:25 +0000 (14:15 +0300)]
HID: quirks: Drop misused kernel-doc annotation

The kernel-doc annotation is misused for hid_mouse_ignore_list. The script
complains about it:

drivers/hid/hid-quirks.c:894: warning: cannot understand function prototype: 'const struct hid_device_id hid_mouse_ignore_list[] = '

Drop the annotation to make script happy.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: hid-asus: select CONFIG_POWER_SUPPLY
Arnd Bergmann [Mon, 4 Mar 2019 19:54:43 +0000 (20:54 +0100)]
HID: hid-asus: select CONFIG_POWER_SUPPLY

The newly added power supply code fails to link when the power supply core
code is disabled:

drivers/hid/hid-asus.o: In function `asus_battery_get_property':
hid-asus.c:(.text+0x11de): undefined reference to `power_supply_get_drvdata'
drivers/hid/hid-asus.o: In function `asus_probe':
hid-asus.c:(.text+0x170c): undefined reference to `devm_power_supply_register'
hid-asus.c:(.text+0x1734): undefined reference to `power_supply_powers'
drivers/hid/hid-asus.o: In function `asus_raw_event':
hid-asus.c:(.text+0x1914): undefined reference to `power_supply_changed'

Select the subsystem from Kconfig as we do for other hid drivers already.

Fixes: 6311d329e12a ("HID: hid-asus: Add BT keyboard dock battery monitoring support")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoHID: quirks: use correct format chars in dbg_hid
Louis Taylor [Wed, 27 Feb 2019 11:07:20 +0000 (11:07 +0000)]
HID: quirks: use correct format chars in dbg_hid

When building with -Wformat, clang warns:

drivers/hid/hid-quirks.c:1075:27: warning: format specifies type
'unsigned short' but the argument has type '__u32' (aka 'unsigned int')
[-Wformat]
  bl_entry->driver_data, bl_entry->vendor,
 ^~~~~~~~~~~~~~~~
./include/linux/hid.h:1170:48: note: expanded from macro 'dbg_hid'
  printk(KERN_DEBUG "%s: " format, __FILE__, ##arg);      \
   ~~~~~~              ^~~
drivers/hid/hid-quirks.c:1076:4: warning: format specifies type
'unsigned short' but the argument has type '__u32' (aka 'unsigned int')
[-Wformat]
  bl_entry->product);
  ^~~~~~~~~~~~~~~~~
./include/linux/hid.h:1170:48: note: expanded from macro 'dbg_hid'
  printk(KERN_DEBUG "%s: " format, __FILE__, ##arg);      \
   ~~~~~~              ^~~
drivers/hid/hid-quirks.c:1242:12: warning: format specifies type
'unsigned short' but the argument has type '__u32' (aka 'unsigned int')
[-Wformat]
  quirks, hdev->vendor, hdev->product);
  ^~~~~~~~~~~~
./include/linux/hid.h:1170:48: note: expanded from macro 'dbg_hid'
  printk(KERN_DEBUG "%s: " format, __FILE__, ##arg);      \
   ~~~~~~              ^~~
drivers/hid/hid-quirks.c:1242:26: warning: format specifies type
'unsigned short' but the argument has type '__u32' (aka 'unsigned int')
[-Wformat]
  quirks, hdev->vendor, hdev->product);
^~~~~~~~~~~~~
./include/linux/hid.h:1170:48: note: expanded from macro 'dbg_hid'
  printk(KERN_DEBUG "%s: " format, __FILE__, ##arg);      \
   ~~~~~~              ^~~
4 warnings generated.

This patch fixes the format strings to use the correct format type for unsigned
ints.

Link: https://github.com/ClangBuiltLinux/linux/issues/378
Signed-off-by: Louis Taylor <louis@kragniz.eu>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
5 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid
Linus Torvalds [Fri, 8 Mar 2019 17:00:43 +0000 (09:00 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/hid/hid

Pull HID updates from Jiri Kosina:

 - support for Pro Pen slim, from Jason Gerecke

 - power management improvements to Intel-ISH driver, from Song Hongyan

 - UCLogic driver revamp in order to be able to support wider range of
   Huion tablets, from Nikolai Kondrashov

 - Asus Transbook support, from NOGUCHI Hiroshi

 - other assorted small bugfixes / cleanups and device ID additions

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid: (46 commits)
  HID: Remove Waltop tablets from hid_have_special_driver
  HID: Remove KYE tablets from hid_have_special_driver
  HID: Remove hid-uclogic entries from hid_have_special_driver
  HID: uclogic: Do not initialize non-USB devices
  HID: uclogic: Add support for Ugee G5
  HID: uclogic: Support Gray-coded rotary encoders
  HID: uclogic: Support faking Wacom pad device ID
  HID: uclogic: Add support for XP-Pen Deco 01
  HID: uclogic: Add support for XP-Pen Star G640
  HID: uclogic: Add support for XP-Pen Star G540
  HID: uclogic: Add support for Ugee EX07S frame controls
  HID: uclogic: Add support for Ugee M540
  HID: uclogic: Add support for Ugee 2150
  HID: uclogic: Support v2 protocol
  HID: uclogic: Support fragmented high-res reports
  HID: uclogic: Support in-range reporting emulation
  HID: uclogic: Designate current protocol v1
  HID: uclogic: Re-initialize tablets on resume
  HID: uclogic: Extract tablet parameter discovery into a module
  HID: uclogic: Extract report descriptors to a module
  ...

5 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/livepatchin...
Linus Torvalds [Fri, 8 Mar 2019 16:58:25 +0000 (08:58 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/livepatching/livepatching

Pull livepatching updates from Jiri Kosina:

 - support for something we call 'atomic replace', and allows for much
   better handling of cumulative patches (which is something very useful
   for distros), from Jason Baron with help of Petr Mladek and Joe
   Lawrence

 - improvement of handling of tasks blocking finalization, from Miroslav
   Benes

 - update of MAINTAINERS file to reflect move towards group
   maintainership

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/livepatching/livepatching: (22 commits)
  livepatch/selftests: use "$@" to preserve argument list
  livepatch: Module coming and going callbacks can proceed with all listed patches
  livepatch: Proper error handling in the shadow variables selftest
  livepatch: return -ENOMEM on ptr_id() allocation failure
  livepatch: Introduce klp_for_each_patch macro
  livepatch: core: Return EOPNOTSUPP instead of ENOSYS
  selftests/livepatch: add DYNAMIC_DEBUG config dependency
  livepatch: samples: non static warnings fix
  livepatch: update MAINTAINERS
  livepatch: Remove signal sysfs attribute
  livepatch: Send a fake signal periodically
  selftests/livepatch: introduce tests
  livepatch: Remove ordering (stacking) of the livepatches
  livepatch: Atomic replace and cumulative patches documentation
  livepatch: Remove Nop structures when unused
  livepatch: Add atomic replace
  livepatch: Use lists to manage patches, objects and functions
  livepatch: Simplify API by removing registration step
  livepatch: Don't block the removal of patches loaded after a forced transition
  livepatch: Consolidate klp_free functions
  ...

5 years agoMerge tag 'drm-next-2019-03-06' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 8 Mar 2019 16:23:15 +0000 (08:23 -0800)]
Merge tag 'drm-next-2019-03-06' of git://anongit.freedesktop.org/drm/drm

Pull drm updates from Dave Airlie:
 "This is the main drm pull request for the 5.1 merge window.

  The big changes I'd highlight are:
   - nouveau has HMM support now, there is finally an in-tree user so we
     can quieten down the rip it out people.
   - i915 now enables fastboot by default on Skylake+
   - Displayport Multistream support has been refactored and should
     hopefully be more reliable.

  Core:
   - header cleanups aiming towards removing drmP.h
   - dma-buf fence seqnos to 64-bits
   - common helper for DP mst hotplug for radeon,i915,amdgpu + new
     refcounting scheme
   - MST i2c improvements
   - drm_syncobj_cb removal
   - ARM FB compression fourcc
   - P010 + P016 fourcc
   - allwinner tiled format modifier
   - i2c over aux I2C_M_STOP support
   - DRM_AUTH handling fixes

  TTM:
   - ref/unref renaming

  New driver:
   - ARM komeda display driver

  scheduler:
   - refactor mirror list handling
   - rework hw fence processing
   - 0 run queue entity fix

  bridge:
   - TI DS90C185 LVDS bridge
   - thc631lvdm83d bridge improvements
   - cadence + allwinner DSI ported to generic phy

  panels:
   - Sitronix ST7701 panel
   - Kingdisplay KD097D04
   - LeMaker BL035-RGB-002
   - PDA 91-00156-A0
   - Innolux EE101IA-01D

  i915:
   - Enable fastboot by default on SKL+/VLV/CHV
   - Export RPCS configuration for ICL media driver
   - Coffelake PCI ID
   - CNL clocks setup fixes
   - ACPI/PMIC support for MIPI/DSI
   - Per-engine WA init for all engines
   - Shrinker locking fixes
   - Kerneldoc updates
   - Lots of ring improvements and reset fixes
   - Coffeelake GVT Support
   - VFIO GVT EDID Region support
   - runtime PM wakeref tracking
   - ILK->IVB primary plane enable delays
   - userptr mutex locking fixes
   - DSI fixes
   - LVDS/TV cleanups
   - HW readout fixes
   - LUT robustness fixes
   - ICL display and watermark fixes
   - gem mmap race fix

  amdgpu:
   - add scheduled dependencies interface
   - DCC on scanout surfaces
   - vega10/20 BACO support
   - Multiple IH rings on soc15
   - XGMI locking fixes
   - DC i2c/aux cleanups
   - runtime SMU debug interface
   - Kexec improvmeents
   - SR-IOV fixes
   - DC freesync + ABM fixes
   - GDS fixes
   - GPUVM fixes
   - vega20 PCIE DPM switching fixes
   - Context priority handling fixes

  radeon:
   - fix missing break in evergreen parser

  nouveau:
   - SVM support via HMM

  msm:
   - QCOM Compressed modifier support

  exynos:
   - s5pv210 rotator support

  imx:
   - zpos property support
   - pending update fixes

  v3d:
   - cache flush improvments

  vc4:
   - reflection support
   - HDMI overscan support

  tegra:
   - CEC refactoring
   - HDMI audio fixes
   - Tegra186 prep work
   - SOR crossbar device tree fixes

  sun4i:
   - implicit fencing support
   - YUV and scalar support improvements
   - A23 support
   - tiling fixes

  atmel-hlcdc:
   - clipping and rotation property fixes

  qxl:
   - BO and PRIME improvements
   - generic fbdev emulation

  dw-hdmi:
   - HDMI 2.0 2160p
   - YUV420 ouput

  rockchip:
   - implicit fencing support
   - reflection proerties

  virtio-gpu:
   - use generic fbdev emulation

  tilcdc:
   - cpufreq vs crtc init fix

  rcar-du:
   - R8A774C0 support
   - D3/E3 RGB output routing fixes and DPAD0 support
   - RA87744 LVDS support

  bochs:
   - atomic and generic fbdev emulation
   - ID mismatch error on bochs load

  meson:
   - remove firmware fbs"

* tag 'drm-next-2019-03-06' of git://anongit.freedesktop.org/drm/drm: (1130 commits)
  drm/amd/display: Use vrr friendly pageflip throttling in DC.
  drm/imx: only send commit done event when all state has been applied
  drm/imx: allow building under COMPILE_TEST
  drm/imx: imx-tve: depend on COMMON_CLK
  drm/imx: ipuv3-plane: add zpos property
  drm/imx: ipuv3-plane: add function to query atomic update status
  gpu: ipu-v3: prg: add function to get channel configure status
  gpu: ipu-v3: pre: add double buffer status readback
  drm/amdgpu: Bump amdgpu version for context priority override.
  drm/amdgpu/powerplay: fix typo in BACO header guards
  drm/amdgpu/powerplay: fix return codes in BACO code
  drm/amdgpu: add missing license on baco files
  drm/bochs: Fix the ID mismatch error
  drm/nouveau/dmem: use dma addresses during migration copies
  drm/nouveau/dmem: use physical vram addresses during migration copies
  drm/nouveau/dmem: extend copy function to allow direct use of physical addresses
  drm/nouveau/svm: new ioctl to migrate process memory to GPU memory
  drm/nouveau/dmem: device memory helpers for SVM
  drm/nouveau/svm: initial support for shared virtual memory
  drm/nouveau: prepare for enabling svm with existing userspace interfaces
  ...

5 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Fri, 8 Mar 2019 03:25:37 +0000 (19:25 -0800)]
Merge branch 'akpm' (patches from Andrew)

Merge more updates from Andrew Morton:

 - some of the rest of MM

 - various misc things

 - dynamic-debug updates

 - checkpatch

 - some epoll speedups

 - autofs

 - rapidio

 - lib/, lib/lzo/ updates

* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (83 commits)
  samples/mic/mpssd/mpssd.h: remove duplicate header
  kernel/fork.c: remove duplicated include
  include/linux/relay.h: fix percpu annotation in struct rchan
  arch/nios2/mm/fault.c: remove duplicate include
  unicore32: stop printing the virtual memory layout
  MAINTAINERS: fix GTA02 entry and mark as orphan
  mm: create the new vm_fault_t type
  arm, s390, unicore32: remove oneliner wrappers for memblock_alloc()
  arch: simplify several early memory allocations
  openrisc: simplify pte_alloc_one_kernel()
  sh: prefer memblock APIs returning virtual address
  microblaze: prefer memblock API returning virtual address
  powerpc: prefer memblock APIs returning virtual address
  lib/lzo: separate lzo-rle from lzo
  lib/lzo: implement run-length encoding
  lib/lzo: fast 8-byte copy on arm64
  lib/lzo: 64-bit CTZ on arm64
  lib/lzo: tidy-up ifdefs
  ipc/sem.c: replace kvmalloc/memset with kvzalloc and use struct_size
  ipc: annotate implicit fall through
  ...

5 years agosamples/mic/mpssd/mpssd.h: remove duplicate header
Brajeswar Ghosh [Fri, 8 Mar 2019 00:31:34 +0000 (16:31 -0800)]
samples/mic/mpssd/mpssd.h: remove duplicate header

Remove duplicate headers which are included more than once

Link: http://lkml.kernel.org/r/20190114170033.GA3674@hp-pavilion-15-notebook-pc-brajeswar
Signed-off-by: Brajeswar Ghosh <brajeswar.linux@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agokernel/fork.c: remove duplicated include
YueHaibing [Fri, 8 Mar 2019 00:31:31 +0000 (16:31 -0800)]
kernel/fork.c: remove duplicated include

Remove duplicated include.

Link: http://lkml.kernel.org/r/20181209062952.17736-1-yuehaibing@huawei.com
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Reviewed-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinclude/linux/relay.h: fix percpu annotation in struct rchan
Luc Van Oostenryck [Fri, 8 Mar 2019 00:31:28 +0000 (16:31 -0800)]
include/linux/relay.h: fix percpu annotation in struct rchan

The percpu member of this structure is declared as:
struct ... ** __percpu member;
So its type is:
__percpu pointer to pointer to struct ...

But looking at how it's used, its type should be:
pointer to __percpu pointer to struct ...
and it should thus be declared as:
struct ... * __percpu *member;

So fix the placement of '__percpu' in the definition of this
structures.

This silents a few Sparse's warnings like:
warning: incorrect type in initializer (different address spaces)
  expected void const [noderef] <asn:3> *__vpp_verify
  got struct sched_domain **

Link: http://lkml.kernel.org/r/20190118144902.79065-1-luc.vanoostenryck@gmail.com
Fixes: 017c59c042d01 ("relay: Use per CPU constructs for the relay channel buffer pointers")
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
Cc: Jens Axboe <axboe@suse.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoarch/nios2/mm/fault.c: remove duplicate include
Sabyasachi Gupta [Fri, 8 Mar 2019 00:31:24 +0000 (16:31 -0800)]
arch/nios2/mm/fault.c: remove duplicate include

Remove linux/ptrace.h which is included more than once

Link: http://lkml.kernel.org/r/5c45d345.1c69fb81.d90ed.8e05@mx.google.com
Signed-off-by: Sabyasachi Gupta <sabyasachi.linux@gmail.com>
Cc: Ley Foon Tan <lftan@altera.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agounicore32: stop printing the virtual memory layout
Geert Uytterhoeven [Fri, 8 Mar 2019 00:31:21 +0000 (16:31 -0800)]
unicore32: stop printing the virtual memory layout

Since commit ad67b74d2469 ("printk: hash addresses printed with %p"),
the virtual memory layout printed during boot up contains "ptrval"
instead of actual addresses.

Instead of changing the printing to "%px", and leaking virtual memory
layout information again, just remove the printing completely, cfr.
e.g.  commits 071929dbdd86 ("arm64: Stop printing the virtual memory
layout") and 31833332f798 ("m68k/mm: Stop printing the virtual memory
layout").

All interesting information (actual section sizes) is already printed by
mem_init_print_info() just above anyway.

Link: http://lkml.kernel.org/r/20190121152254.29079-1-geert+renesas@glider.be
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Kees Cook <keescook@chromium.org>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoMAINTAINERS: fix GTA02 entry and mark as orphan
Jann Horn [Fri, 8 Mar 2019 00:31:17 +0000 (16:31 -0800)]
MAINTAINERS: fix GTA02 entry and mark as orphan

The entry for GTA02 never had paths listed; fix that.  commit 9d76295ac608
("[ARM] GTA02/FreeRunner: Add machine definition"), which added the entry
for GTA02, created two new files named
arch/arm/mach-s3c2442/{include/mach/gta02.h,mach-gta02.c}, which were then
renamed in commit dd6f01b5ccba ("ARM: S3C2440: move mach-s3c2440/* into
mach-s3c24xx/") to
arch/arm/mach-s3c24xx/{include/mach/gta02.h,mach-gta02.c}.

Also, the GTA02 maintainer's email address is from a domain that doesn't
have an MX record anymore and appears to have expired.  Remove the
maintainer and mark the subsystem as orphan.

Link: http://lkml.kernel.org/r/20190215140444.37060-1-jannh@google.com
Signed-off-by: Jann Horn <jannh@google.com>
Cc: Nelson Castillo <arhuaco@freaks-unidos.net>
Cc: Nelson Castillo <nelsoneci@gmail.com>
Cc: Andy Green <andy@warmcat.com>
Cc: Ben Dooks <ben-linux@fluff.org>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: create the new vm_fault_t type
Souptick Joarder [Fri, 8 Mar 2019 00:31:14 +0000 (16:31 -0800)]
mm: create the new vm_fault_t type

Page fault handlers are supposed to return VM_FAULT codes, but some
drivers/file systems mistakenly return error numbers.  Now that all
drivers/file systems have been converted to use the vm_fault_t return
type, change the type definition to no longer be compatible with 'int'.
By making it an unsigned int, the function prototype becomes
incompatible with a function which returns int.  Sparse will detect any
attempts to return a value which is not a VM_FAULT code.

VM_FAULT_SET_HINDEX and VM_FAULT_GET_HINDEX values are changed to avoid
conflict with other VM_FAULT codes.

[jrdr.linux@gmail.com: fix warnings]
Link: http://lkml.kernel.org/r/20190109183742.GA24326@jordon-HP-15-Notebook-PC
Link: http://lkml.kernel.org/r/20190108183041.GA12137@jordon-HP-15-Notebook-PC
Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
Reviewed-by: William Kucharski <william.kucharski@oracle.com>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoarm, s390, unicore32: remove oneliner wrappers for memblock_alloc()
Mike Rapoport [Fri, 8 Mar 2019 00:31:10 +0000 (16:31 -0800)]
arm, s390, unicore32: remove oneliner wrappers for memblock_alloc()

arm, s390 and unicore32 use oneliner wrappers for memblock_alloc().
Replace their usage with direct call to memblock_alloc().

Link: http://lkml.kernel.org/r/1546248566-14910-7-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Suggested-by: Christoph Hellwig <hch@infradead.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Greentime Hu <green.hu@gmail.com>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Mark Salter <msalter@redhat.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Simek <michal.simek@xilinx.com>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Stafford Horne <shorne@gmail.com>
Cc: Stefan Kristiansson <stefan.kristiansson@saunalahti.fi>
Cc: Vincent Chen <deanbo422@gmail.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoarch: simplify several early memory allocations
Mike Rapoport [Fri, 8 Mar 2019 00:31:06 +0000 (16:31 -0800)]
arch: simplify several early memory allocations

There are several early memory allocations in arch/ code that use
memblock_phys_alloc() to allocate memory, convert the returned physical
address to the virtual address and then set the allocated memory to
zero.

Exactly the same behaviour can be achieved simply by calling
memblock_alloc(): it allocates the memory in the same way as
memblock_phys_alloc(), then it performs the phys_to_virt() conversion
and clears the allocated memory.

Replace the longer sequence with a simpler call to memblock_alloc().

Link: http://lkml.kernel.org/r/1546248566-14910-6-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Greentime Hu <green.hu@gmail.com>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Mark Salter <msalter@redhat.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Simek <michal.simek@xilinx.com>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Stafford Horne <shorne@gmail.com>
Cc: Stefan Kristiansson <stefan.kristiansson@saunalahti.fi>
Cc: Vincent Chen <deanbo422@gmail.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoopenrisc: simplify pte_alloc_one_kernel()
Mike Rapoport [Fri, 8 Mar 2019 00:31:01 +0000 (16:31 -0800)]
openrisc: simplify pte_alloc_one_kernel()

The pte_alloc_one_kernel() function allocates a page using
__get_free_page(GFP_KERNEL) when mm initialization is complete and
memblock_phys_alloc() on the earlier stages.  The physical address of
the page allocated with memblock_phys_alloc() is converted to the
virtual address and in the both cases the allocated page is cleared
using clear_page().

The code is simplified by replacing __get_free_page() with
get_zeroed_page() and by replacing memblock_phys_alloc() with
memblock_alloc().

Link: http://lkml.kernel.org/r/1546248566-14910-5-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Acked-by: Stafford Horne <shorne@gmail.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Greentime Hu <green.hu@gmail.com>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Mark Salter <msalter@redhat.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Simek <michal.simek@xilinx.com>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Stefan Kristiansson <stefan.kristiansson@saunalahti.fi>
Cc: Vincent Chen <deanbo422@gmail.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agosh: prefer memblock APIs returning virtual address
Mike Rapoport [Fri, 8 Mar 2019 00:30:57 +0000 (16:30 -0800)]
sh: prefer memblock APIs returning virtual address

Rather than use the memblock_alloc_base that returns a physical address
and then convert this address to the virtual one, use appropriate
memblock function that returns a virtual address.

There is a small functional change in the allocation of then
NODE_DATA().  Instead of panicing if the local allocation failed, the
non-local allocation attempt will be made.

Link: http://lkml.kernel.org/r/1546248566-14910-4-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Greentime Hu <green.hu@gmail.com>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Mark Salter <msalter@redhat.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Simek <michal.simek@xilinx.com>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Stafford Horne <shorne@gmail.com>
Cc: Stefan Kristiansson <stefan.kristiansson@saunalahti.fi>
Cc: Vincent Chen <deanbo422@gmail.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomicroblaze: prefer memblock API returning virtual address
Mike Rapoport [Fri, 8 Mar 2019 00:30:53 +0000 (16:30 -0800)]
microblaze: prefer memblock API returning virtual address

Rather than use the memblock_alloc_base that returns a physical address
and then convert this address to the virtual one, use appropriate
memblock function that returns a virtual address.

Link: http://lkml.kernel.org/r/1546248566-14910-3-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Tested-by: Michal Simek <michal.simek@xilinx.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Greentime Hu <green.hu@gmail.com>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Mark Salter <msalter@redhat.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Stafford Horne <shorne@gmail.com>
Cc: Stefan Kristiansson <stefan.kristiansson@saunalahti.fi>
Cc: Vincent Chen <deanbo422@gmail.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agopowerpc: prefer memblock APIs returning virtual address
Mike Rapoport [Fri, 8 Mar 2019 00:30:48 +0000 (16:30 -0800)]
powerpc: prefer memblock APIs returning virtual address

Patch series "memblock: simplify several early memory allocation", v4.

These patches simplify some of the early memory allocations by replacing
usage of older memblock APIs with newer and shinier ones.

Quite a few places in the arch/ code allocated memory using a memblock
API that returns a physical address of the allocated area, then
converted this physical address to a virtual one and then used memset(0)
to clear the allocated range.

More recent memblock APIs do all the three steps in one call and their
usage simplifies the code.

It's important to note that regardless of API used, the core allocation
is nearly identical for any set of memblock allocators: first it tries
to find a free memory with all the constraints specified by the caller
and then falls back to the allocation with some or all constraints
disabled.

The first three patches perform the conversion of call sites that have
exact requirements for the node and the possible memory range.

The fourth patch is a bit one-off as it simplifies openrisc's
implementation of pte_alloc_one_kernel(), and not only the memblock
usage.

The fifth patch takes care of simpler cases when the allocation can be
satisfied with a simple call to memblock_alloc().

The sixth patch removes one-liner wrappers for memblock_alloc on arm and
unicore32, as suggested by Christoph.

This patch (of 6):

There are a several places that allocate memory using memblock APIs that
return a physical address, convert the returned address to the virtual
address and frequently also memset(0) the allocated range.

Update these places to use memblock allocators already returning a
virtual address.  Use memblock functions that clear the allocated memory
instead of calling memset(0) where appropriate.

The calls to memblock_alloc_base() that were not followed by memset(0)
are replaced with memblock_alloc_try_nid_raw().  Since the latter does
not panic() when the allocation fails, the appropriate panic() calls are
added to the call sites.

Link: http://lkml.kernel.org/r/1546248566-14910-2-git-send-email-rppt@linux.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Guan Xuetao <gxt@pku.edu.cn>
Cc: Greentime Hu <green.hu@gmail.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Mark Salter <msalter@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Rich Felker <dalias@libc.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Stefan Kristiansson <stefan.kristiansson@saunalahti.fi>
Cc: Stafford Horne <shorne@gmail.com>
Cc: Vincent Chen <deanbo422@gmail.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolib/lzo: separate lzo-rle from lzo
Dave Rodgman [Fri, 8 Mar 2019 00:30:44 +0000 (16:30 -0800)]
lib/lzo: separate lzo-rle from lzo

To prevent any issues with persistent data, separate lzo-rle from lzo so
that it is treated as a separate algorithm, and lzo is still available.

Link: http://lkml.kernel.org/r/20190205155944.16007-3-dave.rodgman@arm.com
Signed-off-by: Dave Rodgman <dave.rodgman@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Markus F.X.J. Oberhumer <markus@oberhumer.com>
Cc: Matt Sealey <matt.sealey@arm.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nitin Gupta <nitingupta910@gmail.com>
Cc: Richard Purdie <rpurdie@openedhand.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
Cc: Sonny Rao <sonnyrao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolib/lzo: implement run-length encoding
Dave Rodgman [Fri, 8 Mar 2019 00:30:40 +0000 (16:30 -0800)]
lib/lzo: implement run-length encoding

Patch series "lib/lzo: run-length encoding support", v5.

Following on from the previous lzo-rle patchset:

  https://lkml.org/lkml/2018/11/30/972

This patchset contains only the RLE patches, and should be applied on
top of the non-RLE patches ( https://lkml.org/lkml/2019/2/5/366 ).

Previously, some questions were raised around the RLE patches.  I've
done some additional benchmarking to answer these questions.  In short:

 - RLE offers significant additional performance (data-dependent)

 - I didn't measure any regressions that were clearly outside the noise

One concern with this patchset was around performance - specifically,
measuring RLE impact separately from Matt Sealey's patches (CTZ & fast
copy).  I have done some additional benchmarking which I hope clarifies
the benefits of each part of the patchset.

Firstly, I've captured some memory via /dev/fmem from a Chromebook with
many tabs open which is starting to swap, and then split this into 4178
4k pages.  I've excluded the all-zero pages (as zram does), and also the
no-zero pages (which won't tell us anything about RLE performance).
This should give a realistic test dataset for zram.  What I found was
that the data is VERY bimodal: 44% of pages in this dataset contain 5%
or fewer zeros, and 44% contain over 90% zeros (30% if you include the
no-zero pages).  This supports the idea of special-casing zeros in zram.

Next, I've benchmarked four variants of lzo on these pages (on 64-bit
Arm at max frequency): baseline LZO; baseline + Matt Sealey's patches
(aka MS); baseline + RLE only; baseline + MS + RLE.  Numbers are for
weighted roundtrip throughput (the weighting reflects that zram does
more compression than decompression).

  https://drive.google.com/file/d/1VLtLjRVxgUNuWFOxaGPwJYhl_hMQXpHe/view?usp=sharing

Matt's patches help in all cases for Arm (and no effect on Intel), as
expected.

RLE also behaves as expected: with few zeros present, it makes no
difference; above ~75%, it gives a good improvement (50 - 300 MB/s on
top of the benefit from Matt's patches).

Best performance is seen with both MS and RLE patches.

Finally, I have benchmarked the same dataset on an x86-64 device.  Here,
the MS patches make no difference (as expected); RLE helps, similarly as
on Arm.  There were no definite regressions; allowing for observational
error, 0.1% (3/4178) of cases had a regression > 1 standard deviation,
of which the largest was 4.6% (1.2 standard deviations).  I think this
is probably within the noise.

  https://drive.google.com/file/d/1xCUVwmiGD0heEMx5gcVEmLBI4eLaageV/view?usp=sharing

One point to note is that the graphs show RLE appears to help very
slightly with no zeros present! This is because the extra code causes
the clang optimiser to change code layout in a way that happens to have
a significant benefit.  Taking baseline LZO and adding a do-nothing line
like "__builtin_prefetch(out_len);" immediately before the "goto next"
has the same effect.  So this is a real, but basically spurious effect -
it's small enough not to upset the overall findings.

This patch (of 3):

When using zram, we frequently encounter long runs of zero bytes.  This
adds a special case which identifies runs of zeros and encodes them
using run-length encoding.

This is faster for both compression and decompresion.  For high-entropy
data which doesn't hit this case, impact is minimal.

Compression ratio is within a few percent in all cases.

This modifies the bitstream in a way which is backwards compatible
(i.e., we can decompress old bitstreams, but old versions of lzo cannot
decompress new bitstreams).

Link: http://lkml.kernel.org/r/20190205155944.16007-2-dave.rodgman@arm.com
Signed-off-by: Dave Rodgman <dave.rodgman@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Markus F.X.J. Oberhumer <markus@oberhumer.com>
Cc: Matt Sealey <matt.sealey@arm.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nitin Gupta <nitingupta910@gmail.com>
Cc: Richard Purdie <rpurdie@openedhand.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
Cc: Sonny Rao <sonnyrao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolib/lzo: fast 8-byte copy on arm64
Matt Sealey [Fri, 8 Mar 2019 00:30:36 +0000 (16:30 -0800)]
lib/lzo: fast 8-byte copy on arm64

Enable faster 8-byte copies on arm64.

Link: http://lkml.kernel.org/r/20181127161913.23863-6-dave.rodgman@arm.com
Link: http://lkml.kernel.org/r/20190205141950.9058-4-dave.rodgman@arm.com
Signed-off-by: Matt Sealey <matt.sealey@arm.com>
Signed-off-by: Dave Rodgman <dave.rodgman@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Markus F.X.J. Oberhumer <markus@oberhumer.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nitin Gupta <nitingupta910@gmail.com>
Cc: Richard Purdie <rpurdie@openedhand.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
Cc: Sonny Rao <sonnyrao@google.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolib/lzo: 64-bit CTZ on arm64
Matt Sealey [Fri, 8 Mar 2019 00:30:33 +0000 (16:30 -0800)]
lib/lzo: 64-bit CTZ on arm64

LZO leaves some performance on the table by not realising that arm64 can
optimize count-trailing-zeros bit operations.

Add CONFIG_ARM64 to the checked definitions alongside CONFIG_X86_64 to
enable the use of rbit/clz instructions on full 64-bit quantities.

Link: http://lkml.kernel.org/r/20181127161913.23863-5-dave.rodgman@arm.com
Link: http://lkml.kernel.org/r/20190205141950.9058-3-dave.rodgman@arm.com
Signed-off-by: Matt Sealey <matt.sealey@arm.com>
Signed-off-by: Dave Rodgman <dave.rodgman@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Markus F.X.J. Oberhumer <markus@oberhumer.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nitin Gupta <nitingupta910@gmail.com>
Cc: Richard Purdie <rpurdie@openedhand.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
Cc: Sonny Rao <sonnyrao@google.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolib/lzo: tidy-up ifdefs
Dave Rodgman [Fri, 8 Mar 2019 00:30:29 +0000 (16:30 -0800)]
lib/lzo: tidy-up ifdefs

Patch series "lib/lzo: performance improvements", v5.

This patch (of 3):

Modify the ifdefs in lzodefs.h to be more consistent with normal kernel
macros (e.g., change __aarch64__ to CONFIG_ARM64).

Link: http://lkml.kernel.org/r/20190205141950.9058-2-dave.rodgman@arm.com
Signed-off-by: Dave Rodgman <dave.rodgman@arm.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: David S. Miller <davem@davemloft.net>
Cc: Nitin Gupta <nitingupta910@gmail.com>
Cc: Richard Purdie <rpurdie@openedhand.com>
Cc: Markus F.X.J. Oberhumer <markus@oberhumer.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
Cc: Sonny Rao <sonnyrao@google.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Matt Sealey <matt.sealey@arm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoipc/sem.c: replace kvmalloc/memset with kvzalloc and use struct_size
Gustavo A. R. Silva [Fri, 8 Mar 2019 00:30:26 +0000 (16:30 -0800)]
ipc/sem.c: replace kvmalloc/memset with kvzalloc and use struct_size

Use kvzalloc() instead of kvmalloc() and memset().

Also, make use of the struct_size() helper instead of the open-coded
version in order to avoid any potential type mistakes.

This code was detected with the help of Coccinelle.

Link: http://lkml.kernel.org/r/20190131214221.GA28930@embeddedor
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Manfred Spraul <manfred@colorfullife.com>
Cc: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoipc: annotate implicit fall through
Mathieu Malaterre [Fri, 8 Mar 2019 00:30:23 +0000 (16:30 -0800)]
ipc: annotate implicit fall through

There is a plan to build the kernel with -Wimplicit-fallthrough and this
place in the code produced a warning (W=1).

This commit remove the following warning:

  ipc/sem.c:1683:6: warning: this statement may fall through [-Wimplicit-fallthrough=]

Link: http://lkml.kernel.org/r/20190114203608.18218-1-malat@debian.org
Signed-off-by: Mathieu Malaterre <malat@debian.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinit/initramfs.c: provide more details in error messages
David Engraf [Fri, 8 Mar 2019 00:30:19 +0000 (16:30 -0800)]
init/initramfs.c: provide more details in error messages

Use distinct error messages when archive decompression failed.

Link: http://lkml.kernel.org/r/20190212075635.7373-1-david.engraf@sysgo.com
Signed-off-by: David Engraf <david.engraf@sysgo.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Tested-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Philippe Ombredanne <pombredanne@nexb.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolib/ubsan: default UBSAN_ALIGNMENT to not set
Anders Roxell [Fri, 8 Mar 2019 00:30:16 +0000 (16:30 -0800)]
lib/ubsan: default UBSAN_ALIGNMENT to not set

When booting an allmodconfig kernel, there are a lot of false-positives.
With a message like this 'UBSAN: Undefined behaviour in...' with a call
trace that follows.

UBSAN warnings are a result of enabling noisy CONFIG_UBSAN_ALIGNMENT
which is disabled by default if HAVE_EFFICIENT_UNALIGNED_ACCESS=y.

It's noisy even if don't have efficient unaligned access, e.g.  people
often add __cacheline_aligned_in_smp in structs, but forget to align
allocations of such struct (kmalloc() give 8-byte alignment in worst
case).

Rework so that when building a allmodconfig kernel that turns everything
into '=m' or '=y' will turn off UBSAN_ALIGNMENT.

[aryabinin@virtuozzo.com: changelog addition]
Link: http://lkml.kernel.org/r/20181217150326.30933-1-anders.roxell@linaro.org
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoscripts/gdb: replace flags (MS_xyz -> SB_xyz)
Jackie Liu [Fri, 8 Mar 2019 00:30:10 +0000 (16:30 -0800)]
scripts/gdb: replace flags (MS_xyz -> SB_xyz)

Since commit 1751e8a6cb93 ("Rename superblock flags (MS_xyz ->
SB_xyz)"), scripts/gdb should be updated to replace MS_xyz with SB_xyz.

This change didn't directly affect the running operation of scripts/gdb
until commit e262e32d6bde "vfs: Suppress MS_* flag defs within the
kernel unless explicitly enabled" removed the definitions used by
constants.py.

Update constants.py.in to utilise the new internal flags, matching the
implementation at fs/proc_namespace.c::show_sb_opts.

Note to stable, e262e32d6bde landed in v5.0-rc1 (which was just
released), so we'll want this picked back to 5.0 stable once this patch
hits mainline (akpm just picked it up).  Without this, debugging a
kernel a kernel via GDB+QEMU is broken in the 5.0 release.

[kieran.bingham@ideasonboard.com: add fixes tag, reword commit message]
Link: http://lkml.kernel.org/r/20190305103014.25847-1-kieran.bingham@ideasonboard.com
Fixes: e262e32d6bde "vfs: Suppress MS_* flag defs within the kernel unless explicitly enabled"
Signed-off-by: Jackie Liu <liuyun01@kylinos.cn>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Tested-by: Nick Desaulniers <ndesaulniers@google.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Cc: Felipe Balbi <felipe.balbi@linux.intel.com>
Cc: Dan Robertson <danlrobertson89@gmail.com>
Cc: Jan Kiszka <jan.kiszka@siemens.com>
Cc: David Howells <dhowells@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agokcov: convert kcov.refcount to refcount_t
Elena Reshetova [Fri, 8 Mar 2019 00:30:00 +0000 (16:30 -0800)]
kcov: convert kcov.refcount to refcount_t

atomic_t variables are currently used to implement reference
counters with the following properties:

 - counter is initialized to 1 using atomic_set()

 - a resource is freed upon counter reaching zero

 - once counter reaches zero, its further
   increments aren't allowed

 - counter schema uses basic atomic operations
   (set, inc, inc_not_zero, dec_and_test, etc.)

Such atomic variables should be converted to a newly provided refcount_t
type and API that prevents accidental counter overflows and underflows.
This is important since overflows and underflows can lead to
use-after-free situation and be exploitable.

The variable kcov.refcount is used as pure reference counter.  Convert
it to refcount_t and fix up the operations.

**Important note for maintainers:

Some functions from refcount_t API defined in lib/refcount.c have
different memory ordering guarantees than their atomic counterparts.

The full comparison can be seen in https://lkml.org/lkml/2017/11/15/57
and it is hopefully soon in state to be merged to the documentation
tree.  Normally the differences should not matter since refcount_t
provides enough guarantees to satisfy the refcounting use cases, but in
some rare cases it might matter.  Please double check that you don't
have some undocumented memory guarantees for this variable usage.

For the kcov.refcount it might make a difference
in following places:
 - kcov_put(): decrement in refcount_dec_and_test() only
   provides RELEASE ordering and control dependency on success
   vs. fully ordered atomic counterpart

Link: http://lkml.kernel.org/r/1547634429-772-1-git-send-email-elena.reshetova@intel.com
Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Suggested-by: Kees Cook <keescook@chromium.org>
Reviewed-by: David Windsor <dwindsor@gmail.com>
Reviewed-by: Hans Liljestrand <ishkamiel@gmail.com>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Andrea Parri <andrea.parri@amarulasolutions.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agokcov: no need to check return value of debugfs_create functions
Greg Kroah-Hartman [Fri, 8 Mar 2019 00:29:56 +0000 (16:29 -0800)]
kcov: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Link: http://lkml.kernel.org/r/20190122152151.16139-46-gregkh@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: "Steven Rostedt (VMware)" <rostedt@goodmis.org>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Anders Roxell <anders.roxell@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agokernel/configs: use .incbin directive to embed config_data.gz
Masahiro Yamada [Fri, 8 Mar 2019 00:29:53 +0000 (16:29 -0800)]
kernel/configs: use .incbin directive to embed config_data.gz

This slightly optimizes the kernel/configs.c build.

bin2c is not very efficient because it converts a data file into a huge
array to embed it into a *.c file.

Instead, we can use the .incbin directive.

Also, this simplifies the code; Makefile is cleaner, and the way to get
the offset/size of the config_data.gz is more straightforward.

I used the "asm" statement in *.c instead of splitting it into *.S
because MODULE_* tags are not supported in *.S files.

I also cleaned up kernel/.gitignore; "config_data.gz" is unneeded
because the top-level .gitignore takes care of the "*.gz" pattern.

[yamada.masahiro@socionext.com: v2]
Link: http://lkml.kernel.org/r/1550108893-21226-1-git-send-email-yamada.masahiro@socionext.com
Link: http://lkml.kernel.org/r/1549941160-8084-1-git-send-email-yamada.masahiro@socionext.com
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Alexander Popov <alex.popov@linux.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoconfigs: get rid of obsolete CONFIG_ENABLE_WARN_DEPRECATED
Alexey Brodkin [Fri, 8 Mar 2019 00:29:50 +0000 (16:29 -0800)]
configs: get rid of obsolete CONFIG_ENABLE_WARN_DEPRECATED

This Kconfig option was removed during v4.19 development in commit
771c035372a0 ("deprecate the '__deprecated' attribute warnings entirely
and for good") so there's no point to keep it in defconfigs any longer.

FWIW defconfigs were patched with:
--------------------------->8----------------------
find . -name *_defconfig -exec sed -i '/CONFIG_ENABLE_WARN_DEPRECATED/d' {} \;
--------------------------->8----------------------

Link: http://lkml.kernel.org/r/20190128152434.41969-1-abrodkin@synopsys.com
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agokernel/gcov/gcc_3_4.c: use struct_size() in kzalloc()
Gustavo A. R. Silva [Fri, 8 Mar 2019 00:29:47 +0000 (16:29 -0800)]
kernel/gcov/gcc_3_4.c: use struct_size() in kzalloc()

One of the more common cases of allocation size calculations is finding
the size of a structure that has a zero-sized array at the end, along
with memory for some number of elements for that array.  For example:

  struct foo {
      int stuff;
      void *entry[];
  };

  instance = kzalloc(sizeof(struct foo) + sizeof(void *) * count, GFP_KERNEL);

Instead of leaving these open-coded and prone to type mistakes, we can
now use the new struct_size() helper:

  instance = kzalloc(struct_size(instance, entry, count), GFP_KERNEL);

This code was detected with the help of Coccinelle.

Link: http://lkml.kernel.org/r/20190109172445.GA15908@embeddedor
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agosysctl: handle overflow for file-max
Christian Brauner [Fri, 8 Mar 2019 00:29:43 +0000 (16:29 -0800)]
sysctl: handle overflow for file-max

Currently, when writing

  echo 18446744073709551616 > /proc/sys/fs/file-max

/proc/sys/fs/file-max will overflow and be set to 0.  That quickly
crashes the system.

This commit sets the max and min value for file-max.  The max value is
set to long int.  Any higher value cannot currently be used as the
percpu counters are long ints and not unsigned integers.

Note that the file-max value is ultimately parsed via
__do_proc_doulongvec_minmax().  This function does not report error when
min or max are exceeded.  Which means if a value largen that long int is
written userspace will not receive an error instead the old value will be
kept.  There is an argument to be made that this should be changed and
__do_proc_doulongvec_minmax() should return an error when a dedicated min
or max value are exceeded.  However this has the potential to break
userspace so let's defer this to an RFC patch.

Link: http://lkml.kernel.org/r/20190107222700.15954-3-christian@brauner.io
Signed-off-by: Christian Brauner <christian@brauner.io>
Acked-by: Kees Cook <keescook@chromium.org>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Joe Lawrence <joe.lawrence@redhat.com>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Waiman Long <longman@redhat.com>
[christian@brauner.io: v4]
Link: http://lkml.kernel.org/r/20190210203943.8227-3-christian@brauner.io
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agosysctl: handle overflow in proc_get_long
Christian Brauner [Fri, 8 Mar 2019 00:29:40 +0000 (16:29 -0800)]
sysctl: handle overflow in proc_get_long

proc_get_long() is a funny function.  It uses simple_strtoul() and for a
good reason.  proc_get_long() wants to always succeed the parse and
return the maybe incorrect value and the trailing characters to check
against a pre-defined list of acceptable trailing values.  However,
simple_strtoul() explicitly ignores overflows which can cause funny
things like the following to happen:

  echo 18446744073709551616 > /proc/sys/fs/file-max
  cat /proc/sys/fs/file-max
  0

(Which will cause your system to silently die behind your back.)

On the other hand kstrtoul() does do overflow detection but does not
return the trailing characters, and also fails the parse when anything
other than '\n' is a trailing character whereas proc_get_long() wants to
be more lenient.

Now, before adding another kstrtoul() function let's simply add a static
parse strtoul_lenient() which:
 - fails on overflow with -ERANGE
 - returns the trailing characters to the caller

The reason why we should fail on ERANGE is that we already do a partial
fail on overflow right now.  Namely, when the TMPBUFLEN is exceeded.  So
we already reject values such as 184467440737095516160 (21 chars) but
accept values such as 18446744073709551616 (20 chars) but both are
overflows.  So we should just always reject 64bit overflows and not
special-case this based on the number of chars.

Link: http://lkml.kernel.org/r/20190107222700.15954-2-christian@brauner.io
Signed-off-by: Christian Brauner <christian@brauner.io>
Acked-by: Kees Cook <keescook@chromium.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Joe Lawrence <joe.lawrence@redhat.com>
Cc: Waiman Long <longman@redhat.com>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agorapidio/mport_cdev: mark expected switch fall-through
Gustavo A. R. Silva [Fri, 8 Mar 2019 00:29:36 +0000 (16:29 -0800)]
rapidio/mport_cdev: mark expected switch fall-through

In preparation for enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

This patch fixes the following warning:

  drivers/rapidio/devices/rio_mport_cdev.c: In function `mport_release_mapping':
  drivers/rapidio/devices/rio_mport_cdev.c:2151:3: warning: this statement may fall through [-Wimplicit-fallthrough=]
     rio_unmap_inb_region(mport, map->phys_addr);
     ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    CC      drivers/regulator/fixed-helper.o
    CC      drivers/pinctrl/stm32/pinctrl-stm32f429.o
  drivers/rapidio/devices/rio_mport_cdev.c:2152:2: note: here
    case MAP_DMA:
    ^~~~

Warning level 3 was used: -Wimplicit-fallthrough=3

This patch is part of the ongoing efforts to enable
-Wimplicit-fallthrough.

Link: http://lkml.kernel.org/r/20190212175014.GA14326@embeddedor
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: Alexandre Bounine <alex.bou9@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agodrivers/rapidio/rio_cm.c: fix potential oops in riocm_ch_listen()
Dan Carpenter [Fri, 8 Mar 2019 00:29:33 +0000 (16:29 -0800)]
drivers/rapidio/rio_cm.c: fix potential oops in riocm_ch_listen()

If riocm_get_channel() fails, then we should just return -EINVAL.
Calling riocm_put_channel() will trigger a NULL dereference and
generally we should call put() if the get() didn't succeed.

Link: http://lkml.kernel.org/r/20190110130230.GB27017@kadam
Fixes: b6e8d4aa1110 ("rapidio: add RapidIO channelized messaging driver")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Alexandre Bounine <alexandre.bounine@idt.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agokernel: workqueue: clarify wq_worker_last_func() caller requirements
Johannes Weiner [Fri, 8 Mar 2019 00:29:30 +0000 (16:29 -0800)]
kernel: workqueue: clarify wq_worker_last_func() caller requirements

This function can only be called safely from very specific scheduler
contexts.  Document those.

Link: http://lkml.kernel.org/r/20190206150528.31198-1-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Suggested-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoexec: increase BINPRM_BUF_SIZE to 256
Oleg Nesterov [Fri, 8 Mar 2019 00:29:26 +0000 (16:29 -0800)]
exec: increase BINPRM_BUF_SIZE to 256

Large enterprise clients often run applications out of networked file
systems where the IT mandated layout of project volumes can end up
leading to paths that are longer than 128 characters.  Bumping this up
to the next order of two solves this problem in all but the most
egregious case while still fitting into a 512b slab.

[oleg@redhat.com: update comment, per Kees]
Link: http://lkml.kernel.org/r/20181112160956.GA28472@redhat.com
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Ben Woodard <woodard@redhat.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Kees Cook <keescook@chromium.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agofs/exec.c: replace opencoded set_mask_bits()
Vineet Gupta [Fri, 8 Mar 2019 00:29:23 +0000 (16:29 -0800)]
fs/exec.c: replace opencoded set_mask_bits()

Link: http://lkml.kernel.org/r/1548275584-18096-2-git-send-email-vgupta@synopsys.com
Link: http://lkml.kernel.org/g/20150807115710.GA16897@redhat.com
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Reviewed-by: Anthony Yznaga <anthony.yznaga@oracle.com>
Acked-by: Oleg Nesterov <oleg@redhat.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Miklos Szeredi <mszeredi@redhat.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agofat: enable .splice_write to support splice on O_DIRECT file
Hou Tao [Fri, 8 Mar 2019 00:29:19 +0000 (16:29 -0800)]
fat: enable .splice_write to support splice on O_DIRECT file

Now splice() on O_DIRECT-opened fat file will return -EFAULT, that is
because the default .splice_write, namely default_file_splice_write(),
will construct an ITER_KVEC iov_iter and dio_refill_pages() in dio path
can not handle it.

Fix it by implementing .splice_write through iter_file_splice_write().

Spotted by xfs-tests generic/091.

Link: http://lkml.kernel.org/r/20190210094754.56355-1-houtao1@huawei.com
Signed-off-by: Hou Tao <houtao1@huawei.com>
Acked-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoautofs: clear O_NONBLOCK on the pipe
NeilBrown [Fri, 8 Mar 2019 00:29:16 +0000 (16:29 -0800)]
autofs: clear O_NONBLOCK on the pipe

autofs does not expect the pipe it is given to have O_NONBLOCK set -
specifically if __kernel_write() in autofs_write() returns -EAGAIN, this
is treated as a fatal error and the pipe is closed.

For safety autofs should, therefore, clear the O_NONBLOCK flag.

Releases of systemd prior to 8th February 2019 used
  pipe2(p, O_NONBLOCK|O_CLOEXEC)
and thus (inadvertently) set this flag.

Link: http://lkml.kernel.org/r/154993550902.3321.1183632970046073478.stgit@pluto-themaw-net
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agofs/autofs/inode.c: use seq_puts() for simple strings in autofs_show_options()
Ian Kent [Fri, 8 Mar 2019 00:29:12 +0000 (16:29 -0800)]
fs/autofs/inode.c: use seq_puts() for simple strings in autofs_show_options()

Fix checkpatch.sh WARNING about the use of seq_printf() to print simple
strings in autofs_show_options(), use seq_puts() in this case.

Link: http://lkml.kernel.org/r/154889012613.4863.12231175554744203482.stgit@pluto-themaw-net
Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoautofs: add ignore mount option
Ian Kent [Fri, 8 Mar 2019 00:29:09 +0000 (16:29 -0800)]
autofs: add ignore mount option

Add an autofs file system mount option that can be used to provide a
generic indicator to applications that the mount entry should be ignored
when displaying mount information.

In other OSes that provide autofs and that provide a mount list to user
space based on the kernel mount list a no-op mount option ("ignore" is
the one use on the most common OS) is allowed so that autofs file system
users can optionally use it.

The idea is that it be used by user space programs to exclude autofs
mounts from consideration when reading the mounts list.

Prior to the change to link /etc/mtab to /proc/self/mounts all I needed
to do to achieve this was to use mount(2) and not update the mtab but
now that no longer works.

I know the symlinking happened a long time ago and I considered doing
this then but, at the time I couldn't remember the commonly used option
name and thought persuading the various utility maintainers would be too
hard.

But now I have a RHEL request to do this for compatibility for a widely
used product so I want to go ahead with it and try and enlist the help
of some utility package maintainers.

Clearly, without the option nothing can be done so it's at least a
start.

Link: http://lkml.kernel.org/r/154725123970.11260.6113771566924907275.stgit@pluto-themaw-net
Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinit/calibrate.c: provide proper prototype
Valdis Kletnieks [Fri, 8 Mar 2019 00:29:06 +0000 (16:29 -0800)]
init/calibrate.c: provide proper prototype

Sparse issues a warning:

    CHECK   init/calibrate.c
  init/calibrate.c:271:28: warning: symbol 'calibration_delay_done' was not declared. Should it be static?

The actual issue is that it's a __weak symbol that archs can override
(in fact, ARM does so), but no prototype is provided.  Let's provide one
to prevent surprises.

Link: http://lkml.kernel.org/r/18827.1548750938@turing-police.cc.vt.edu
Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agofs/binfmt_elf.c: spread const a little
Alexey Dobriyan [Fri, 8 Mar 2019 00:29:03 +0000 (16:29 -0800)]
fs/binfmt_elf.c: spread const a little

Link: http://lkml.kernel.org/r/20190204202830.GC27482@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agofs/binfmt_elf.c: use list_for_each_entry()
Alexey Dobriyan [Fri, 8 Mar 2019 00:28:59 +0000 (16:28 -0800)]
fs/binfmt_elf.c: use list_for_each_entry()

[adobriyan@gmail.com: fixup compilation]
Link: http://lkml.kernel.org/r/20190205064334.GA2152@avx2
Link: http://lkml.kernel.org/r/20190204202800.GB27482@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agofs/binfmt_elf.c: don't be afraid of overflow
Alexey Dobriyan [Fri, 8 Mar 2019 00:28:56 +0000 (16:28 -0800)]
fs/binfmt_elf.c: don't be afraid of overflow

Number of ELF program headers is 16-bit by spec, so total size
comfortably fits into "unsigned int".

Space savings: 7 bytes!

add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-7 (-7)
Function                                     old     new   delta
load_elf_phdrs                               137     130      -7

Link: http://lkml.kernel.org/r/20190204202715.GA27482@avx2
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoepoll: use rwlock in order to reduce ep_poll_callback() contention
Roman Penyaev [Fri, 8 Mar 2019 00:28:53 +0000 (16:28 -0800)]
epoll: use rwlock in order to reduce ep_poll_callback() contention

The goal of this patch is to reduce contention of ep_poll_callback()
which can be called concurrently from different CPUs in case of high
events rates and many fds per epoll.  Problem can be very well
reproduced by generating events (write to pipe or eventfd) from many
threads, while consumer thread does polling.  In other words this patch
increases the bandwidth of events which can be delivered from sources to
the poller by adding poll items in a lockless way to the list.

The main change is in replacement of the spinlock with a rwlock, which
is taken on read in ep_poll_callback(), and then by adding poll items to
the tail of the list using xchg atomic instruction.  Write lock is taken
everywhere else in order to stop list modifications and guarantee that
list updates are fully completed (I assume that write side of a rwlock
does not starve, it seems qrwlock implementation has these guarantees).

The following are some microbenchmark results based on the test [1]
which starts threads which generate N events each.  The test ends when
all events are successfully fetched by the poller thread:

 spinlock
 ========

 threads  events/ms  run-time ms
       8       6402        12495
      16       7045        22709
      32       7395        43268

 rwlock + xchg
 =============

 threads  events/ms  run-time ms
       8      10038         7969
      16      12178        13138
      32      13223        24199

According to the results bandwidth of delivered events is significantly
increased, thus execution time is reduced.

This patch was tested with different sort of microbenchmarks and
artificial delays (e.g.  "udelay(get_random_int() & 0xff)") introduced
in kernel on paths where items are added to lists.

[1] https://github.com/rouming/test-tools/blob/master/stress-epoll.c

Link: http://lkml.kernel.org/r/20190103150104.17128-5-rpenyaev@suse.de
Signed-off-by: Roman Penyaev <rpenyaev@suse.de>
Cc: Davidlohr Bueso <dbueso@suse.de>
Cc: Jason Baron <jbaron@akamai.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoepoll: unify awaking of wakeup source on ep_poll_callback() path
Roman Penyaev [Fri, 8 Mar 2019 00:28:49 +0000 (16:28 -0800)]
epoll: unify awaking of wakeup source on ep_poll_callback() path

Original comment "Activate ep->ws since epi->ws may get deactivated at
any time" indeed sounds loud, but it is incorrect, because the path
where we check epi->ws is a path where insert to ovflist happens, i.e.
ep_scan_ready_list() has taken ep->mtx and waits for this callback to
finish, thus ep_modify() (which unregisters wakeup source) waits for
ep_scan_ready_list().

Here in this patch I simply call ep_pm_stay_awake_rcu(), which is a bit
extra for this path (indirectly protected by main ep->mtx, so even rcu
is not needed), but I do not want to create another naked
__ep_pm_stay_awake() variant only for this particular case, so rcu variant
is just better for all the cases.

Link: http://lkml.kernel.org/r/20190103150104.17128-4-rpenyaev@suse.de
Signed-off-by: Roman Penyaev <rpenyaev@suse.de>
Cc: Davidlohr Bueso <dbueso@suse.de>
Cc: Jason Baron <jbaron@akamai.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoepoll: make sure all elements in ready list are in FIFO order
Roman Penyaev [Fri, 8 Mar 2019 00:28:46 +0000 (16:28 -0800)]
epoll: make sure all elements in ready list are in FIFO order

Patch series "use rwlock in order to reduce ep_poll_callback()
contention", v3.

The last patch targets the contention problem in ep_poll_callback(),
which can be very well reproduced by generating events (write to pipe or
eventfd) from many threads, while consumer thread does polling.

The following are some microbenchmark results based on the test [1]
which starts threads which generate N events each.  The test ends when
all events are successfully fetched by the poller thread:

 spinlock
 ========

 threads  events/ms  run-time ms
       8       6402        12495
      16       7045        22709
      32       7395        43268

 rwlock + xchg
 =============

 threads  events/ms  run-time ms
       8      10038         7969
      16      12178        13138
      32      13223        24199

According to the results bandwidth of delivered events is significantly
increased, thus execution time is reduced.

This patch (of 4):

All coming events are stored in FIFO order and this is also should be
applicable to ->ovflist, which originally is stack, i.e.  LIFO.

Thus to keep correct FIFO order ->ovflist should reversed by adding
elements to the head of the read list but not to the tail.

Link: http://lkml.kernel.org/r/20190103150104.17128-2-rpenyaev@suse.de
Signed-off-by: Roman Penyaev <rpenyaev@suse.de>
Reviewed-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Jason Baron <jbaron@akamai.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agocheckpatch: add test for SPDX-License-Identifier on wrong line #
Joe Perches [Fri, 8 Mar 2019 00:28:42 +0000 (16:28 -0800)]
checkpatch: add test for SPDX-License-Identifier on wrong line #

Warn when any SPDX-License-Identifier: tag is not created on the proper
line number.

Link: http://lkml.kernel.org/r/9b74ee87f8c1b8fd310e213fcb4994d58610fcb6.camel@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: "Enrico Weigelt, metux IT consult" <lkml@metux.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agocheckpatch: allow reporting C99 style comments
Vadim Bendebury [Fri, 8 Mar 2019 00:28:38 +0000 (16:28 -0800)]
checkpatch: allow reporting C99 style comments

Presently C99 style comments are removed unconditionally before actual
patch validity check happens.  This is a problem for some third party
projects which use checkpatch.pl but do not allow C99 style comments.

This patch adds yet another variable, named C99_COMMENT_TOLERANCE.  If
it is included in the --ignore command line or config file options list,
C99 comments in the patch are reported as errors.

Tested by processing a patch with a C99 style comment, it passes the
check just fine unless '--ignore C99_COMMENT_TOLERANCE' is present in
.checkpatch.conf.

Link: http://lkml.kernel.org/r/20190110224957.25008-1-vbendeb@chromium.org
Signed-off-by: Vadim Bendebury <vbendeb@chromium.org>
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agocheckpatch: add some new alloc functions to various tests
Joe Perches [Fri, 8 Mar 2019 00:28:35 +0000 (16:28 -0800)]
checkpatch: add some new alloc functions to various tests

Many new generic allocation functions like the kvmalloc family have been
added recently to the kernel.

The allocation functions test now includes:

o kvmalloc and variants
o kstrdup_const
o kmemdup_nul
o dma_alloc_coherent
o alloc_skb and variants

Add a separate $allocFunctions variable to help make the allocation
functions test a bit more readable.

Miscellanea:

o Use $allocFunctions in the unnecessary OOM message test and
  add exclude uses with __GFP_NOWARN
o Use $allocFunctions in the unnecessary cast test
o Add the kvmalloc family to the preferred sizeof alloc style
  foo = kvmalloc(sizeof(*foo), ...)

Link: http://lkml.kernel.org/r/a5e60a2b93e10baf84af063f6c8e56402273105d.camel@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agocheckpatch: verify SPDX comment style
Joe Perches [Fri, 8 Mar 2019 00:28:32 +0000 (16:28 -0800)]
checkpatch: verify SPDX comment style

Using SPDX commenting style // or /* is specified for various file types
in Documentation/process/license-rules.rst so add an appropriate test for
.[chsS] files because many proposed file additions and patches do not use
the correct style.

Link: http://lkml.kernel.org/r/8b02899853247a2c67669561761f354dd3bd110e.camel@perches.com
Signed-off-by: Joe Perches <joe@perches.com>
Cc: Rob Herring <robh@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolib/test_firmware.c: remove some dead code
Dan Carpenter [Fri, 8 Mar 2019 00:28:28 +0000 (16:28 -0800)]
lib/test_firmware.c: remove some dead code

The test_fw_config->reqs allocation succeeded so these addresses can't
be NULL.

Also on the second error path, we forgot to set "rc = -ENOMEM;".

Link: http://lkml.kernel.org/r/20190221183700.GA1737@kadam
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: "Luis R. Rodriguez" <mcgrof@kernel.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolib/assoc_array.c: mark expected switch fall-through
Gustavo A. R. Silva [Fri, 8 Mar 2019 00:28:25 +0000 (16:28 -0800)]
lib/assoc_array.c: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

This patch fixes the following warning:

  lib/assoc_array.c: In function `assoc_array_delete':
  lib/assoc_array.c:1110:3: warning: this statement may fall through [-Wimplicit-fallthrough=]
     for (slot = 0; slot < ASSOC_ARRAY_FAN_OUT; slot++) {
     ^~~
  lib/assoc_array.c:1118:2: note: here
    case assoc_array_walk_tree_empty:
    ^~~~

Warning level 3 was used: -Wimplicit-fallthrough=3

This patch is part of the ongoing efforts to enable
-Wimplicit-fallthrough.

Link: http://lkml.kernel.org/r/20190212212206.GA16378@embeddedor
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Cc: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agolib/test_ubsan.c: VLA no longer used in kernel
Olof Johansson [Fri, 8 Mar 2019 00:28:21 +0000 (16:28 -0800)]
lib/test_ubsan.c: VLA no longer used in kernel

Since we now build with -Wvla, any use of VLA throws a warning.
Including this test, so...  maybe we should just remove the test?

  lib/test_ubsan.c: In function 'test_ubsan_vla_bound_not_positive':
  lib/test_ubsan.c:48:2: warning: ISO C90 forbids variable length array 'buf' [-Wvla]

For the out-of-bounds test, switch to non-VLA setup.

  lib/test_ubsan.c: In function 'test_ubsan_out_of_bounds':
  lib/test_ubsan.c:64:2: warning: ISO C90 forbids variable length array 'arr' [-Wvla]

Link: http://lkml.kernel.org/r/20190113183210.56154-1-olof@lixom.net
Signed-off-by: Olof Johansson <olof@lixom.net>
Acked-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Colin Ian King <colin.king@canonical.com>
Cc: Jinbum Park <jinb.park7@gmail.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>