uref->usage_index can be indirectly controlled by userspace, hence leading
to a potential exploitation of the Spectre variant 1 vulnerability.
This field is used as an array index by the hiddev_ioctl_usage() function,
when 'cmd' is either HIDIOCGCOLLECTIONINDEX, HIDIOCGUSAGES or
HIDIOCSUSAGES.
For cmd == HIDIOCGCOLLECTIONINDEX case, uref->usage_index is compared to
field->maxusage and then used as an index to dereference field->usage
array. The same thing happens to the cmd == HIDIOC{G,S}USAGES cases, where
uref->usage_index is checked against an array maximum value and then it is
used as an index in an array.
This is a summary of the HIDIOCGCOLLECTIONINDEX case, which matches the
traditional Spectre V1 first load:
copy_from_user(uref, user_arg, sizeof(*uref))
if (uref->usage_index >= field->maxusage)
goto inval;
i = field->usage[uref->usage_index].collection_index;
return i;
This patch fixes this by sanitizing field uref->usage_index before using it
to index field->usage (HIDIOCGCOLLECTIONINDEX) or field->value in
HIDIOC{G,S}USAGES arrays, thus, avoiding speculation in the first load.
Cc: <stable@vger.kernel.org>
Signed-off-by: Breno Leitao <leitao@debian.org>
--
v2: Contemplate cmd == HIDIOC{G,S}USAGES case
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
uref->field_index, uref->usage_index, finfo.field_index and cinfo.index can be
indirectly controlled by user-space, hence leading to a potential exploitation
of the Spectre variant 1 vulnerability.
This issue was detected with the help of Smatch:
drivers/hid/usbhid/hiddev.c:473 hiddev_ioctl_usage() warn: potential spectre issue 'report->field' (local cap)
drivers/hid/usbhid/hiddev.c:477 hiddev_ioctl_usage() warn: potential spectre issue 'field->usage' (local cap)
drivers/hid/usbhid/hiddev.c:757 hiddev_ioctl() warn: potential spectre issue 'report->field' (local cap)
drivers/hid/usbhid/hiddev.c:801 hiddev_ioctl() warn: potential spectre issue 'hid->collection' (local cap)
Fix this by sanitizing such structure fields before using them to index
report->field, field->usage and hid->collection
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
[1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2
Cc: stable@vger.kernel.org
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
This is the mindless scripted replacement of kernel use of POLL*
variables as described by Al, done by this script:
for V in IN OUT PRI ERR RDNORM RDBAND WRNORM WRBAND HUP RDHUP NVAL MSG; do
L=`git grep -l -w POLL$V | grep -v '^t' | grep -v /um/ | grep -v '^sa' | grep -v '/poll.h$'|grep -v '^D'`
for f in $L; do sed -i "-es/^\([^\"]*\)\(\<POLL$V\>\)/\\1E\\2/" $f; done
done
with de-mangling cleanups yet to come.
NOTE! On almost all architectures, the EPOLL* constants have the same
values as the POLL* constants do. But they keyword here is "almost".
For various bad reasons they aren't the same, and epoll() doesn't
actually work quite correctly in some cases due to this on Sparc et al.
The next patch from Al will sort out the final differences, and we
should be all done.
Scripted-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Instead of calling into usbhid code directly, let's use the standard
accessors for the transport HID drivers, and stop clobbering their error
codes with -EIO.
This also allows us to remove usbhid_get/put_power(), leaving only
usbhid_power().
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Instead of calling into usbhid code directly, let's use the standard
accessors for the transport HID drivers, and stop clobbering their errors
with -EIO.
This also allows us make usbhid_open and close static.
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
We need to store the minor number each drivers. In case of hidraw, the
minor number is stored stores in struct hidraw. But hiddev's minor is
located in struct hid_device.
The hid-core driver announces a kernel message which driver is loaded when
HID device connected, but hiddev's minor number is always zero. To proper
display hiddev's minor number, we need to store the minor number asked from
usb core and do some refactoring work (move from hiddev.c to hiddev.h) to
access hiddev in hid-core.
[jkosina@suse.cz: rebase on top of newer codebase]
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jaejoong Kim <climbbb.kim@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
It looks like a bunch of devices do not like to be polled
for their reports at init time. When you look into the details,
it seems that for those that are requiring the quirk
HID_QUIRK_NO_INIT_REPORTS, the driver fails to retrieve part
of the features/inputs while others (more generic) work.
IMO, it should be acceptable to remove the need for the quirk
in the general case. On the small amount of cases where
we actually need to read the current values, the driver
in charge (hid-mt or wacom) already retrieves the features
manually.
There are 2 cases where we might need to retrieve the reports at
init:
1. hiddev devices with specific use-space tool
2. a device that would require the driver to fetch a specific
feature/input at plug
For case 2, I have seen this a few time on hid-multitouch. It
is solved in hid-multitouch directly by fetching the feature.
I hope it won't be too common and this can be solved on a per-case
basis (crossing fingers).
For case 1, we moved the implementation of HID_QUIRK_NO_INIT_REPORTS
in hiddev. When somebody starts calling ioctls that needs an initial
update, the hiddev device will fetch the initial state of the reports
to mimic the current behavior. This adds a small amount of time during
the first HIDIOCGUSAGE(S), but it should be acceptable in
most cases. To keep the currently known broken devices, we have to
keep around HID_QUIRK_NO_INIT_REPORTS, but the scope will only be
for hiddev.
Note that I don't think hidraw would be affected and I checked that
the FF drivers that need to interact with the report fields are all
using output reports, which are not initialized by
usbhid_init_reports().
NO_INIT_INPUT_REPORTS is then replaced by HID_QUIRK_NO_INIT_REPORTS:
there is no point keeping it for just one device.
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Fix up affected files that include this signal functionality via sched.h.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This patch validates the num_values parameter from userland during the
HIDIOCGUSAGES and HIDIOCSUSAGES commands. Previously, if the report id was set
to HID_REPORT_ID_UNKNOWN, we would fail to validate the num_values parameter
leading to a heap overflow.
Cc: stable@vger.kernel.org
Signed-off-by: Scott Bauer <sbauer@plzdonthack.me>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
This removes most of the dependencies between hid drivers and usbhid.
The patch was constructed by replacing all occurences of
usbhid_wait_io() by its hid_hw_wait() counterpart.
Then, drivers not requiring USB_HID anymore have their USB_HID
dependency cleaned in the Kconfig file.
As of today, few drivers are still requiring an explicit USB layer
dependency:
* ntrig (a patch is on its way)
* multitouch (one patch following and another on its way)
* lenovo tpkbd
* roccat
* sony
The last three are two deeply using direct calls to the usb subsystem
to be able to be cleaned right now.
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Reviewed-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
This allows the hid drivers to be independent from the transport layer.
The patch was constructed by replacing all occurences of
usbhid_submit_report() by its hid_hw_request() counterpart.
Then, drivers not requiring USB_HID anymore have their USB_HID
dependency cleaned in the Kconfig file.
Finally, few drivers still depends on USB_HID. Many of them
are requiring the io wait callback. They are found in the next patch.
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Reviewed-by: Mika Westerberg <mika.westerberg@linux.intel.com>
For the sensor-hub part:
Tested-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
When the file has been open in non-blocking mode, EIO or ERESTARTSYS
would never be returned even if they should (for example when device
has been unplugged, you want EIO and not EAGAIN to be returned).
Move the O_NONBLOCK check after other checks have been performed.
Base on similar patch done to hidraw by
Founder Fang <founder.fang@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
No need to check whether unsigned variable is less than 0.
CC: Jiri Kosina <jkosina@suse.cz>
CC: linux-usb@vger.kernel.org
CC: linux-input@vger.kernel.org
Signed-off-by: Tushar Behera <tushar.behera@linaro.org>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Everytime a HID device is opened, a new hiddev_list is allocated with
kzalloc. This requires 64KB of physically contiguous memory, which could
easily push a heavily loaded system over the edge.
Allocating the same amount of memory with vmalloc shouldn't be nearly as
demanding, so let's do that instead. The memory isn't used for DMA and
doesn't look particularly performance sensitive, so this should be safe.
Signed-off-by: Havard Skinnemoen <hskinnemoen@google.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
both callers of device_get_devnode() are only interested in lower 16bits
and nobody tries to return anything wider than 16bit anyway.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Smatch has a new check for Rosenberg type information leaks where
structs are copied to the user with uninitialized stack data in them.
In this case, the hiddev_devinfo struct has a two byte hole.
struct hiddev_devinfo {
__u32 bustype; /* 0 4 */
__u32 busnum; /* 4 4 */
__u32 devnum; /* 8 4 */
__u32 ifnum; /* 12 4 */
__s16 vendor; /* 16 2 */
__s16 product; /* 18 2 */
__s16 version; /* 20 2 */
/* XXX 2 bytes hole, try to pack */
__u32 num_applications; /* 24 4 */
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
There are a couple use after free bugs here.
Signed-off-by: Dan Carpenter <error27@gmail.com>
[jkosina@suse.cz: removed already fixed hunk]
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Commit 6cb4b04079 ("HID: hiddev: fix race between hiddev_disconnect
and hiddev_release") made it possible to access hiddev (for unlocking
the existance mutex) once hiddev has been kfreed.
Change the order so that this can not happen (always unlock the mutex first,
it is needed only to protect access to ->exist and ->open).
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
When hiddev_disconnect() runs with chardev open, it will proceed with
usbhid_close(). When userspace in parallel runs the hiddev_release(),
it sees !hiddev->exists (as it has been already set so by
hiddev_disconnect()) and kfrees hiddev while hiddev_disconnect() hasn't
finished yet.
Serialize the access to hiddev->exists and hiddev->open by existancelock.
Reported-by: mike-@cinci.rr.com
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
As they are static members of fix size, there is no need to NULL-check them.
Signed-off-by: Daniel Mack <zonque@gmail.com>
Cc: Dmitry Torokhov <dtor@mail.ru>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
hiddev_read: in case mutex_lock_interruptible will be interrupted
remove the task from the wait queue.
Signed-off-by: Peter Waechtler <pwaechtler@mac.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
There was an extra tab so the close curly brace didn't match up with
the right if statement.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Neaten current uses of dev_<level> by adding and using
hid specific hid_<level> macros.
Convert existing uses of dev_<level> uses to hid_<level>.
Convert hid-pidff printk uses to hid_<level>.
Remove err_hid and use hid_err instead.
Add missing newlines to logging messages where necessary.
Coalesce format strings.
Add and use pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Other miscellaneous changes:
Add const struct hid_device * argument to hid-core functions
extract() and implement() so hid_<level> can be used by them.
Fix bad indentation in hid-core hid_input_field function
that calls extract() function above.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Currently, if the device has been removed before hiddev_ioctl(),
the -EIO is returned. If it's removed while hiddev_ioctl() is in
progress, some commands are still processed fine, others
return -ENODEV. This change takes the "existancelock" before
processing ioctl commands and releases it at the end.
If the device has been removed, always returns -ENODEV.
Signed-off-by: Valentine Barshak <vbarshak@mvista.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
A USB HID device can be disconnected at any time.
If this happens right before or while hiddev_ioctl is in progress,
the hiddev_ioctl tries to access invalid hiddev->hid pointer.
When the hid device is disconnected, the hiddev_disconnect()
ends up with a call to hid_device_release() which frees
hid_device, but doesn't set the hiddev->hid pointer to NULL.
If the deallocated memory region has been re-used by the kernel,
this can cause a crash or memory corruption.
Since disconnect can happen at any time, we can't initialize
struct hid_device *hid = hiddev->hid at the beginning of ioctl
and then use it.
This change checks hiddev->exist flag while holding
the existancelock and uses hid_device only if it exists.
Signed-off-by: Valentine Barshak <vbarshak@mvista.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
The big kernel lock has been removed from all these files at some point,
leaving only the #include.
Remove this too as a cleanup.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'llseek' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/bkl:
vfs: make no_llseek the default
vfs: don't use BKL in default_llseek
llseek: automatically add .llseek fop
libfs: use generic_file_llseek for simple_attr
mac80211: disallow seeks in minstrel debug code
lirc: make chardev nonseekable
viotape: use noop_llseek
raw: use explicit llseek file operations
ibmasmfs: use generic_file_llseek
spufs: use llseek in all file operations
arm/omap: use generic_file_llseek in iommu_debug
lkdtm: use generic_file_llseek in debugfs
net/wireless: use generic_file_llseek in debugfs
drm: use noop_llseek
All file_operations should get a .llseek operation so we can make
nonseekable_open the default for future file operations without a
.llseek pointer.
The three cases that we can automatically detect are no_llseek, seq_lseek
and default_llseek. For cases where we can we can automatically prove that
the file offset is always ignored, we use noop_llseek, which maintains
the current behavior of not returning an error from a seek.
New drivers should normally not use noop_llseek but instead use no_llseek
and call nonseekable_open at open time. Existing drivers can be converted
to do the same when the maintainer knows for certain that no user code
relies on calling seek on the device file.
The generated code is often incorrectly indented and right now contains
comments that clarify for each added line why a specific variant was
chosen. In the version that gets submitted upstream, the comments will
be gone and I will manually fix the indentation, because there does not
seem to be a way to do that using coccinelle.
Some amount of new code is currently sitting in linux-next that should get
the same modifications, which I will do at the end of the merge window.
Many thanks to Julia Lawall for helping me learn to write a semantic
patch that does all this.
===== begin semantic patch =====
// This adds an llseek= method to all file operations,
// as a preparation for making no_llseek the default.
//
// The rules are
// - use no_llseek explicitly if we do nonseekable_open
// - use seq_lseek for sequential files
// - use default_llseek if we know we access f_pos
// - use noop_llseek if we know we don't access f_pos,
// but we still want to allow users to call lseek
//
@ open1 exists @
identifier nested_open;
@@
nested_open(...)
{
<+...
nonseekable_open(...)
...+>
}
@ open exists@
identifier open_f;
identifier i, f;
identifier open1.nested_open;
@@
int open_f(struct inode *i, struct file *f)
{
<+...
(
nonseekable_open(...)
|
nested_open(...)
)
...+>
}
@ read disable optional_qualifier exists @
identifier read_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
expression E;
identifier func;
@@
ssize_t read_f(struct file *f, char *p, size_t s, loff_t *off)
{
<+...
(
*off = E
|
*off += E
|
func(..., off, ...)
|
E = *off
)
...+>
}
@ read_no_fpos disable optional_qualifier exists @
identifier read_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
@@
ssize_t read_f(struct file *f, char *p, size_t s, loff_t *off)
{
... when != off
}
@ write @
identifier write_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
expression E;
identifier func;
@@
ssize_t write_f(struct file *f, const char *p, size_t s, loff_t *off)
{
<+...
(
*off = E
|
*off += E
|
func(..., off, ...)
|
E = *off
)
...+>
}
@ write_no_fpos @
identifier write_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
@@
ssize_t write_f(struct file *f, const char *p, size_t s, loff_t *off)
{
... when != off
}
@ fops0 @
identifier fops;
@@
struct file_operations fops = {
...
};
@ has_llseek depends on fops0 @
identifier fops0.fops;
identifier llseek_f;
@@
struct file_operations fops = {
...
.llseek = llseek_f,
...
};
@ has_read depends on fops0 @
identifier fops0.fops;
identifier read_f;
@@
struct file_operations fops = {
...
.read = read_f,
...
};
@ has_write depends on fops0 @
identifier fops0.fops;
identifier write_f;
@@
struct file_operations fops = {
...
.write = write_f,
...
};
@ has_open depends on fops0 @
identifier fops0.fops;
identifier open_f;
@@
struct file_operations fops = {
...
.open = open_f,
...
};
// use no_llseek if we call nonseekable_open
////////////////////////////////////////////
@ nonseekable1 depends on !has_llseek && has_open @
identifier fops0.fops;
identifier nso ~= "nonseekable_open";
@@
struct file_operations fops = {
... .open = nso, ...
+.llseek = no_llseek, /* nonseekable */
};
@ nonseekable2 depends on !has_llseek @
identifier fops0.fops;
identifier open.open_f;
@@
struct file_operations fops = {
... .open = open_f, ...
+.llseek = no_llseek, /* open uses nonseekable */
};
// use seq_lseek for sequential files
/////////////////////////////////////
@ seq depends on !has_llseek @
identifier fops0.fops;
identifier sr ~= "seq_read";
@@
struct file_operations fops = {
... .read = sr, ...
+.llseek = seq_lseek, /* we have seq_read */
};
// use default_llseek if there is a readdir
///////////////////////////////////////////
@ fops1 depends on !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier readdir_e;
@@
// any other fop is used that changes pos
struct file_operations fops = {
... .readdir = readdir_e, ...
+.llseek = default_llseek, /* readdir is present */
};
// use default_llseek if at least one of read/write touches f_pos
/////////////////////////////////////////////////////////////////
@ fops2 depends on !fops1 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier read.read_f;
@@
// read fops use offset
struct file_operations fops = {
... .read = read_f, ...
+.llseek = default_llseek, /* read accesses f_pos */
};
@ fops3 depends on !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier write.write_f;
@@
// write fops use offset
struct file_operations fops = {
... .write = write_f, ...
+ .llseek = default_llseek, /* write accesses f_pos */
};
// Use noop_llseek if neither read nor write accesses f_pos
///////////////////////////////////////////////////////////
@ fops4 depends on !fops1 && !fops2 && !fops3 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier read_no_fpos.read_f;
identifier write_no_fpos.write_f;
@@
// write fops use offset
struct file_operations fops = {
...
.write = write_f,
.read = read_f,
...
+.llseek = noop_llseek, /* read and write both use no f_pos */
};
@ depends on has_write && !has_read && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier write_no_fpos.write_f;
@@
struct file_operations fops = {
... .write = write_f, ...
+.llseek = noop_llseek, /* write uses no f_pos */
};
@ depends on has_read && !has_write && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier read_no_fpos.read_f;
@@
struct file_operations fops = {
... .read = read_f, ...
+.llseek = noop_llseek, /* read uses no f_pos */
};
@ depends on !has_read && !has_write && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
@@
struct file_operations fops = {
...
+.llseek = noop_llseek, /* no read or write fn */
};
===== End semantic patch =====
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Julia Lawall <julia@diku.dk>
Cc: Christoph Hellwig <hch@infradead.org>
Now that hiddev_driver isn't being used for anything, there's no
reason to keep it around. This patch (as1419) gets rid of it
entirely.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
My macbook infrared remote control was broken by commit
bd25f4dd69 ("HID: hiddev: use
usb_find_interface, get rid of BKL").
This device appears in dmesg as:
apple 0003:05AC:8242.0001: hiddev0,hidraw0: USB HID v1.11 Device
[Apple Computer, Inc. IR Receiver] on usb-0000:00:1d.2-1/input0
It stopped working as lircd was getting ENODEV when opening /dev/usb/hiddev0.
AFAICS hiddev_driver is a dummy driver so usb_find_interface(&hiddev_driver)
does not find anything.
The device is associated with the usbhid driver, so let's do
usb_find_interface(&hid_driver) instead.
$ ls -l /sys/devices/pci0000:00/0000:00:1d.2/usb7/7-1/7-1:1.0/usb/hiddev0/device/driver
lrwxrwxrwx 1 root root 0 2010-09-12 16:28 /sys/devices/pci0000:00/0000:00:1d.2/usb7/7-1/7-1:1.0/usb/hiddev0/device/driver -> ../../../../../../bus/usb/drivers/usbhid
Signed-off-by: Guillaume Chazarain <guichaz@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Commit bd25f4dd69 ("HID: hiddev: use usb_find_interface,
get rid of BKL") introduced using of private intfdata in hiddev for
purpose of storing hiddev pointer.
This is a problem, because intf pointer is already being set to struct
hid_device pointer by HID core. This obviously lead to memory corruptions
at device disconnect time, such as
WARNING: at lib/kobject.c:595 kobject_put+0x37/0x4b()
kobject: '(null)' (ffff88011e9cd898): is not initialized, yet kobject_put() is being called.
Convert hiddev into accessing hiddev through struct hid_device which is
in intfdata already.
Reported-and-tested-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Reported-and-tested-by: Heinz Diehl <htd@fritha.org>
Reported-and-tested-by: Alan Ott <alan@signal11.us>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
One of our users reports consistently hitting a NULL dereference that
resolves to the "hid_to_usb_dev(hid);" call in hiddev_ioctl(), when
disconnecting a Lego WeDo USB HID device from an OLPC XO running
Scratch software. There's a FIXME comment and a guard against the
dereference, but that happens farther down the function than the
initial dereference does.
This patch moves the call to be below the guard, and the user reports
that it fixes the problem for him. OLPC bug report:
http://dev.laptop.org/ticket/10174
Signed-off-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
This removes the private hiddev_table in the usbhid
driver and changes it to use usb_find_interface
instead.
The advantage is that we can avoid the race between
usb_register_dev and usb_open and no longer need the
big kernel lock.
This doesn't introduce race condition -- the intf pointer could be
invalidated only in hiddev_disconnect() through usb_deregister_dev(),
but that will block on minor_rwsem and not actually remove the device
until usb_open().
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: "Greg Kroah-Hartman" <gregkh@suse.de>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
The race between ioctl and disconnect is guarded by low level
hiddev device mutex (existancelock) since the commit
07903407 ("HID: hiddev cleanup -- handle all error conditions
properly"), therefore we can remove the lock_kernel() from
hiddev_ioctl_usage().
Acked-by: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
This allows subsytems to provide devtmpfs with non-default permissions
for the device node. Instead of the default mode of 0600, null, zero,
random, urandom, full, tty, ptmx now have a mode of 0666, which allows
non-privileged processes to access standard device nodes in case no
other userspace process applies the expected permissions.
This also fixes a wrong assignment in pktcdvd and a checkpatch.pl complain.
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
hiddev userspace driver uses a rignbuffer to store the parsed usages
that should be returned through read(). This buffer is 64 bytes long,
which is sufficient for queueing single USB 1.0 low-speed report, which
is of maximum size 48 bytes.
There are however USB HID devices which are full-speed USB devices, and
therefore they are free to produce reports 64 bytes long. This is correctly
handled by HID core, but read() on hiddev node gets stuck forever, because
the ring buffer loops infinitely (as it is exactly 64 bytes long as well),
never advancing the buffer pointer.
Plus, the core driver is ready to handle highspeed devices, so we should be
able to handle reports from such devices in the hiddev driver as well, which
means we need larger ringbuffer.
Reported-by: Michael Zeisel <michael.zeisel@philips.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
This adds support for USB drivers to report their requested nodename to
userspace. It also updates a number of USB drivers to provide the
needed subdirectory and device name to be used for them.
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Jan Blunck <jblunck@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
upon further thought this code is still racy.
retval = usb_register_dev(usbhid->intf, &hiddev_class);
here you open a window during which open can happen
if (retval) {
err_hid("Not able to get a minor for this device.");
hid->hiddev = NULL;
kfree(hiddev);
return -1;
} else {
hid->minor = usbhid->intf->minor;
hiddev_table[usbhid->intf->minor - HIDDEV_MINOR_BASE] = hiddev;
and will fail because hiddev_table hasn't been updated
The obvious fix of using a mutex to guard hiddev_table doesn't work because
usb_open() and usb_register_dev() take minor_rwsem and we'd have an AB-BA
deadlock. We need a lock usb_open() also takes in the right order and that leaves
only one option, BKL. I don't like it but I see no alternative.
Once the usb_open() implements something better than lock_kernel(), we could also
do so.
Signed-off-by: Oliver Neukum <oneukum@suse.de>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
* 'bkl-removal' of git://git.lwn.net/linux-2.6:
Rationalize fasync return values
Move FASYNC bit handling to f_op->fasync()
Use f_lock to protect f_flags
Rename struct file->f_ep_lock
