| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Free vpdma buffers in vpe_release. Otherwise it was generating random
backtrace.
Signed-off-by: Harinarayan Bhatta <harinarayan@ti.com>
Signed-off-by: Somnath Mukherjee <somnath@ti.com>
Signed-off-by: Nikhil Devshatwar <nikhil.nd@ti.com>
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
get_q_data can be called with different values for type
e.g. vpe_try_crop calls it with the buffer type which gets passed
from user space
Framework doesn't check wheather its correct type or not
If user space passes wrong type, kernel should not crash.
Return NULL when the passed type is invalid.
Signed-off-by: Nikhil Devshatwar <nikhil.nd@ti.com>
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Increasing max buffer height and width to allow for padded buffers.
Signed-off-by: Harinarayan Bhatta <harinarayan@ti.com>
Signed-off-by: Nikhil Devshatwar <nikhil.nd@ti.com>
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
The video source can generate the data in the SEQ_TB buffer format.
In the case of TI SoC, the IVA_HD can generate the interlaced content in
the SEQ_TB buffer format. This is the format where the top and bottom field
data can be contained in a single buffer. For example, for NV12, interlaced
format, the data in Y buffer will be arranged as Y-top followed by
Y-bottom. And likewise for UV plane.
Also, queuing one buffer of SEQ_TB is equivalent to queuing two different
buffers for top and bottom fields. Driver needs to take care of this when
handling source buffer lists.
Signed-off-by: Nikhil Devshatwar <nikhil.nd@ti.com>
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Current VPE driver does not start the job until all the buffers for
a transaction are queued. When running in multiple context, this might
increase the processing latency.
Alternate solution would be to try to continue the same context as long as
buffers for the transaction are ready; else switch the context. This may
increase number of context switches but it reduces latency significantly.
In this approach, the job_ready always succeeds as long as there are
buffers on the CAPTURE and OUTPUT stream. Processing may start immediately
as the first 2 iterations don't need extra source buffers. Shift all the
source buffers after each iteration and remove the oldest buffer.
Also, with this removes the constraint of pre buffering 3 buffers before
call to STREAMON in case of de-interlacing.
Signed-off-by: Nikhil Devshatwar <nikhil.nd@ti.com>
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
The motion detection block requires 3 fields to create the motion vector
data. This means that using the default method the first progressive
frame is only generated after 3rd field is consumed.
Hence by default for N input field we would generate N - 2 progressive
frames.
In order to generate N progressive frames from N fields we use the
line averaging mode of the de-interlacer for the first 2 fields and then
revert back to the preferred Edge Directed Interpolation method (using
the motion vector).
Thus creating 2 line averaged frames + N - 2 motion based frames for a
total of N frames.
Signed-off-by: Archit Taneja <archit@ti.com>
Signed-off-by: Nikhil Devshatwar <nikhil.nd@ti.com>
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
On DRA7 since l3_noc event are being reported it was found that
when the write descriptor was being written it was consistently
causing bus error events.
The write address was improperly programmed.
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Add a helper to set the background color during vpdma transfer.
This is needed when VPDMA is generating 32 bits RGB format
to have the Alpha channel set to an appropriate value.
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
The VPDMA (Video Port DMA) as found in devices such as DRA7xx is
used for both the Video Processing Engine (VPE) and the Video Input
Port (VIP). Some devices may have multiple VIP instances each with
its own VPDMA engine. Within VIP two slices can use a single VPDMA
engine simultaneously. So support for multi instances and multiple
clients has been added to VPDMA. Needed modification to the existing
helper functions were then reflected to VPE.
Multi-clients registers offset have also been added in preparation.
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
The VPDMA (Video Port DMA) as found in devices such as DRA7xx is
used for both the Video Processing Engine (VPE) and the Video Input
Port (VIP).
In preparation for this we need to turn vpdma into its own
kernel module.
Signed-off-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
While checking why we need i386 checking, I noticed that
the serial code referred at the driver was moved to another
place. Update it to make clear from where such code came from.
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Signed-off-by: Sean Young <sean@mess.org>
|
|
Some CEC adapters will receive messages that they initiated. Add a
check that will ignore such messages.
Most hardware behaves correctly in this respect, but I have seen
adapters that don't, so just filter this out in the framework.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
It can fix known problems with embedded video_device structs.
Signed-off-by: Minghsiu Tsai <minghsiu.tsai@mediatek.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Trivial fix to spelling mistake "unnsupported" to "unsupported"
in debug message.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
The maximum 'Max TMDS Rate' in the HDMI VSDB block is 340 MHz, not 600.
Higher rates are advertised in the HDMI Forum VSDB block.
So lower the Max TMDS rate in the HDMI VSDB block that the vivid driver
uses to 300 MHz, which is typical of most HDMI 1.4b devices.
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Fix to return error code -ENODEV from the error handling
case instead of 0, as done elsewhere in this function.
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Kernel source files need not include <linux/kconfig.h> explicitly
because the top Makefile forces to include it with:
-include $(srctree)/include/linux/kconfig.h
Remove another reduntdant include, that managed to sneak by commit
97139d4a6f26 ("treewide: remove redundant #include <linux/kconfig.h>").
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Many lirc drivers have their own receive buffers which are freed on
unplug (e.g. ir_lirc_unregister). This means that ir->buf->wait_poll
will be freed directly after unplug so do not remove yourself from the
wait queue.
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
If you unplug an lirc device while reading from it, you will get an
use after free as the cdev is freed while still in use.
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
[ 101.457944] ------------[ cut here ]------------
[ 101.457954] WARNING: CPU: 3 PID: 1819 at kernel/sched/core.c:7708 __might_sleep+0x7e/0x80
[ 101.457960] do not call blocking ops when !TASK_RUNNING; state=1 set at [<ffffffffc0364bc2>] lirc_dev_fop_read+0x292/0x4e0 [lirc_dev]
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
In learning mode, you can get much longer messages which can run out
of lengths. The usb message will slightly larger.
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
The wideband receiver is a little awkward on the redrat3. Data arrives
on a different endpoint, and the learning command must be reissued
every time data is learned.
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
If redrat3_delete() is called, ensure ep_in and udev members are set
up so we don't dereference null in the error path. Also ensure that
rc dev device exists before we enable the receiver and that the
led urb exists before we create the led device.
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Cleanup the error logic, removing checks for things that
should be always initialized when the routines are called,
and remove some bogus messages.
[mchehab@s-opensource.com: fix some merge conflicts]
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
No need to duplicate these in the rc name.
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Rename the pnp driver in sysfs from /sys/bus/pnp/drivers/Winbond CIR
to /sys/bus/pnp/drivers/winbond-cir
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Fix to return error code -ENODEV from the error handling
case instead of 0, as done elsewhere in this function.
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Acked-by: Patrice Chotard <patrice.chotard@st.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
'buf' is malloced in dibusb_rc_query() and should be freed before
leaving from the error handling cases, otherwise it will cause
memory leak.
Fixes: ff1c123545d7 ("[media] dibusb: handle error code on RC query")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
A previous patch tried to fix a build error, but introduced another
warning:
drivers/media/platform/mtk-mdp/mtk_mdp_core.c:71:13: error: ‘mtk_mdp_clock_off’ defined but not used [-Werror=unused-function]
drivers/media/platform/mtk-mdp/mtk_mdp_core.c:62:13: error: ‘mtk_mdp_clock_on’ defined but not used [-Werror=unused-function]
This marks all the PM functions as __maybe_unused and removes
the #ifdef around them, as that will always do the right thing.
Fixes: 1b06fcf56aa6 ("[media] media: mtk-mdp: fix build error")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
While this patch sounded a good idea, unfortunately, it causes
bad dependencies, as drivers that would otherwise work without
the DVB core will now break:
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/tea5767.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/tea5761.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/tda827x.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/tda18218.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/qt1010.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/mt2266.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/mt20xx.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/mt2060.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/mc44s803.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/fc0013.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/fc0012.ko] undefined!
ERROR: "dvb_tuner_simple_release" [drivers/media/tuners/fc0011.ko] undefined!
So, we have to revert it.
Note: as the argument for the release ops changed from "int"
to "void", we needed to change it at the revert patch, to
avoid compilation issues like:
drivers/media/tuners/tea5767.c:437:23: error: initialization from incompatible pointer type [-Werror=incompatible-pointer-types]
.release = tea5767_release,
^~~~~~~~~~~~~~~
This reverts commit 22a613e89825ea7a3984a968463cc6d425bd8856.
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
On some weird randconfigs, it is possible to select DVB
drivers, without having the DVB_CORE:
CONFIG_DVB_AU8522=m
CONFIG_DVB_AU8522_V4L=m
CONFIG_DVB_TUNER_DIB0090=m
This was never supposed to work, but changeset 22a613e89825
("[media] dvb_frontend: merge duplicate dvb_tuner_ops.release implementations")
caused it to be exposed:
drivers/built-in.o: In function `fc0011_attach':
(.text+0x1598fb): undefined reference to `dvb_tuner_simple_release'
drivers/built-in.o:(.rodata+0x55e58): undefined reference to `dvb_tuner_simple_release'
drivers/built-in.o:(.rodata+0x57398): undefined reference to `dvb_tuner_simple_release'
Fixes: 22a613e89825 ("[media] dvb_frontend: merge duplicate dvb_tuner_ops.release implementations")
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Kernel takes care that interrupts from one source are serialized.
So there's no need to use spinlock_irq_save.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Spinlock nvt_lock is a member of struct nvt_dev and there's no need
to prefix it with nvt_. So remove this prefix.
[mchehab@s-opensource.org: change the prefix also at the open function,
as the patch removing it were not applied (yet?)]
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Using a separate spinlock to protect access to substruct tx of struct
nvt_dev doesn't provide any actual benefit. We can use spinlock
nvt_lock to protect all access to struct nvt_dev and get rid of
nvt->tx.lock.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Member pdev of struct nvt_dev is needed only to access &pdev->dev.
We can get rid of this it by using rdev->dev.parent instead
(both point to the same struct device).
Setting rdev->dev.parent can be removed from the probe function
as this is done by devm_rc_allocate_device now.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
While removing all interfaces in media_device_unregister(), all
media_interface pointers are freed. This is illegal and results in
double kfree() if any media_interface is still linked at this point;
maybe because a userspace process still has a file handle. Once the
process closes the file handle, dvb_media_device_free() gets called,
which frees the dvb_device.intf_devnode again.
This patch removes the unnecessary kfree() call, and documents who's
responsible for really freeing it.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
media_gobj_destroy() may be called twice on one instance - once by
media_device_unregister() and again by dvb_media_device_free(). The
function media_remove_intf_links() establishes and documents the
convention that mdev==NULL means that the object is not registered,
but nobody ever NULLs this variable. So this patch really implements
this behavior, and adds another mdev==NULL check to
media_gobj_destroy() to protect against double removal.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
This commit amends my old commit fe35637b0a9f ("[media] dvb_frontend:
eliminate blocking wait in dvb_unregister_frontend()"), which added
kref to struct dvb_frontend_private. It turned out that there are
several use-after-free bugs left, which affect the struct
dvb_frontend. Protecting it with kref also protects struct
dvb_frontend_private, so we can simply move it.
This is how the use-after-free looks like in KASAN:
BUG: KASAN: use-after-free in string+0x60/0xb1 at addr ffff880033bd9fc0
Read of size 1 by task kworker/0:2/617
CPU: 0 PID: 617 Comm: kworker/0:2 Not tainted 4.8.0-rc1-hosting+ #60
Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
Workqueue: usb_hub_wq hub_event
0000000000000000 ffff880033757218 ffffffff81394e50 ffff880033bd9fd0
ffff880035c03b00 ffff880033757240 ffffffff811f271d ffff880033bd9fc0
1ffff1000677b3f8 ffffed000677b3f8 ffff8800337572b8 ffffffff811f2afe
Call Trace:
[...]
[<ffffffff813a2d2f>] vsnprintf+0x39d/0x7e9
[<ffffffff813993f9>] add_uevent_var+0x10f/0x1dc
[<ffffffff814fe5ca>] rc_dev_uevent+0x55/0x6f
[<ffffffff814438f8>] dev_uevent+0x2e1/0x316
[<ffffffff81399744>] kobject_uevent_env+0x27e/0x701
[<ffffffff81399bd2>] kobject_uevent+0xb/0xd
[<ffffffff81443445>] device_del+0x322/0x383
[<ffffffff81500c0c>] rc_unregister_device+0x98/0xc3
[<ffffffff81508fb4>] dvb_usb_remote_exit+0x7a/0x90
[<ffffffff81506157>] dvb_usb_exit+0x1d/0xe5
[<ffffffff81506e90>] dvb_usb_device_exit+0x69/0x7d
[<ffffffff8150a181>] pctv452e_usb_disconnect+0x7b/0x80
[...]
Object at ffff880033bd9fc0, in cache kmalloc-16 size: 16
Allocated:
[...]
Freed:
PID = 617
[...]
[<ffffffff811f034c>] kfree+0xd9/0x166
[<ffffffff814fe513>] ir_free_table+0x2f/0x51
[<ffffffff81500bc1>] rc_unregister_device+0x4d/0xc3
[<ffffffff81508fb4>] dvb_usb_remote_exit+0x7a/0x90
[<ffffffff81506157>] dvb_usb_exit+0x1d/0xe5
[<ffffffff81506e90>] dvb_usb_device_exit+0x69/0x7d
[<ffffffff8150a181>] pctv452e_usb_disconnect+0x7b/0x80
Another one:
BUG: KASAN: use-after-free in do_sys_poll+0x336/0x6b8 at addr ffff88003563fcc0
Read of size 8 by task tuner on fronte/1042
CPU: 1 PID: 1042 Comm: tuner on fronte Tainted: G B 4.8.0-rc1-hosting+ #60
Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
0000000000000000 ffff88003353f910 ffffffff81394e50 ffff88003563fd80
ffff880035c03200 ffff88003353f938 ffffffff811f271d ffff88003563fc80
1ffff10006ac7f98 ffffed0006ac7f98 ffff88003353f9b0 ffffffff811f2afe
Call Trace:
[...]
[<ffffffff812289b3>] do_sys_poll+0x336/0x6b8
[...]
[<ffffffff81228ed9>] SyS_poll+0xa9/0x194
[...]
Object at ffff88003563fc80, in cache kmalloc-256 size: 256
Allocated:
[...]
Freed:
PID = 617
[...]
[<ffffffff811f034c>] kfree+0xd9/0x166
[<ffffffff814eb60d>] dvb_unregister_device+0xd6/0xe5
[<ffffffff814fa4ed>] dvb_unregister_frontend+0x4b/0x66
[<ffffffff8150810b>] dvb_usb_adapter_frontend_exit+0x69/0xac
[<ffffffff8150617d>] dvb_usb_exit+0x43/0xe5
[<ffffffff81506e90>] dvb_usb_device_exit+0x69/0x7d
[<ffffffff8150a181>] pctv452e_usb_disconnect+0x7b/0x80
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Ensure that STB0899_POSTPROC_GPIO_POWER is set synchronously.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Prepare for making "release" asynchronous (via kref). Some operations
may need to be run synchronously in dvb_frontend_detach(), and that's
why we need a "detach" callback.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
This code duplication is confusing and error prone. Let's merge them
by moving the release/dvb_detach call into one function with one
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
It is not clear what this return value means. All implemenations
return 0, and the one caller ignores the value. Let's remove this
useless return value completely.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Most release callback functions are identical: free the "tuner_priv"
and clear it. Let's eliminate some bloat by providing this simple
implementation in the dvb_frontend library.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
Commit da677fe14364 ("[media] dvb-core/en50221: use kref to manage
struct dvb_ca_private") moved the dvb_unregister_device() call to the
kref callback, but that left lots of stale device state visible to
userspace (e.g. in sysfs). By using dvb_remove_device() and
dvb_free_device() instead of dvb_unregister_device(), we can avoid
that.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
dvb_unregister_device() has a major problem: it combines unregistering
with memory disposal. Sometimes, it is necessary to unregister a
device, but no memory can be freed yet, because a process still has a
(stale) file handle. Therefore, we need to split
dvb_unregister_device(). This will allow sanitizing a few callers.
With my new design, dvb_unregister_device() appears misnamed, but to
reduce patch noise, I'm not renaming it just yet.
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
These are immutable. Making them "const" allows the compiler to move
them to the "rodata" section.
Note that cxd2841er_t_c_ops cannot be made "const", because
cxd2841er_attach() modifies it. Ouch!
[mchehab@s-opensource.com: fix merge conflicts]
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
rc_unregister_device() will first call ir_free_table(), and later
device_del(); however, the latter causes a call to rc_dev_uevent(),
which prints rc_map.name, which at this point has already bee freed.
This fixes a use-after-free bug found with KASAN.
As reported by Shuah:
"I am seeing the following when I do rmmod on au0828
BUG: KASAN: use-after-free in string+0x170/0x1f0 at addr ffff8801bd513000
Read of size 1 by task rmmod/1831
CPU: 1 PID: 1831 Comm: rmmod Tainted: G W 4.9.0-rc5 #5
Hardware name: Hewlett-Packard HP ProBook 6475b/180F, BIOS 68TTU Ver. F.04 08/03/2012
ffff8801aea2f680 ffffffff81b37ad3 ffff8801fa403b80 ffff8801bd513000
ffff8801aea2f6a8 ffffffff8156c301 ffff8801aea2f738 ffff8801bd513000
ffff8801fa403b80 ffff8801aea2f728 ffffffff8156c59a ffff8801aea2f770
Call Trace:
dump_stack+0x67/0x94
[<ffffffff8156c301>] kasan_object_err+0x21/0x70
[<ffffffff8156c59a>] kasan_report_error+0x1fa/0x4d0
[<ffffffffa116f05f>] ? au0828_exit+0x10/0x21 [au0828]
[<ffffffff8156c8b3>] __asan_report_load1_noabort+0x43/0x50
[<ffffffff81b58b20>] ? string+0x170/0x1f0
[<ffffffff81b58b20>] string+0x170/0x1f0
[<ffffffff81b621c4>] vsnprintf+0x374/0x1c50
[<ffffffff81b61e50>] ? pointer+0xa80/0xa80
[<ffffffff8156b676>] ? save_stack+0x46/0xd0
[<ffffffff81566faa>] ? __kmalloc+0x14a/0x2a0
[<ffffffff81b3d70a>] ? kobject_get_path+0x9a/0x200
[<ffffffff81b408c2>] ? kobject_uevent_env+0x282/0xca0
[<ffffffff81b412eb>] ? kobject_uevent+0xb/0x10
[<ffffffff81f10104>] ? device_del+0x434/0x6d0
[<ffffffffa0fea717>] ? rc_unregister_device+0x177/0x240 [rc_core]
[<ffffffffa116eeb0>] ? au0828_rc_unregister+0x60/0xb0 [au0828]
The problem is fixed with this patch on Linux 4.9-rc4"
Signed-off-by: Max Kellermann <max.kellermann@gmail.com>
Tested-by: Shuah Khan <shuahkh@osg.samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
It should call of_node_put on non-null poiner.
Cc: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Peter Chen <peter.chen@nxp.com>
Acked-by: Benoit Parrot <bparrot@ti.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
|
devm_pinctrl_get() can fail so we should check for that.
Fixes: 0a6824bc10de ('[media] v4l2: blackfin: select proper pinctrl state in ppi_set_params if CONFIG_PINCTRL is enabled')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
|
Harinarayan Bhatta
Nikhil Devshatwar
Archit Taneja
Benoit Parrot
Mauro Carvalho Chehab
Sean Young
Hans Verkuil
Minghsiu Tsai
Colin Ian King
Hans Verkuil
Wei Yongjun
Paul Bolle
Arnd Bergmann
Heiner Kallweit
Max Kellermann
Peter Chen
Dan Carpenter