24 Oct, 2022
1 commit
-
The wrong structure name causes memory overflow issue, this patch fixes
it.Signed-off-by: Alison Wang
19 Oct, 2022
1 commit
-
Free active_resource and eventfd while calling usdpaa_release.
Signed-off-by: Brick Yang
Signed-off-by: Rohit Raj
18 Oct, 2022
8 commits
-
Changing local variable "ret" from unsinged to signed. The check for the
return value for "ret" less than zero had no effect.Signed-off-by: Oliver F. Brown
Reviewed-by: Guoniu.zhou -
This patch adds input and output types, shapes and offset in network
info.Signed-off-by: Feng Guo
Acked-by: Peng Fan -
Send cancel inference messages to the ethosu subsystem to abort
inference execution there.Signed-off-by: Davide Grohmann
Signed-off-by: Alison Wang
Acked-by: Peng Fan -
Add core message that allows user space to fetch information about
network models built into the firmware.Support referencing a network model by index that has been built into
the firmwae binary.Signed-off-by: Kristofer Jonsson
Signed-off-by: Alison Wang
Acked-by: Peng Fan -
ETHOSU_IOCTL_CAPABILITIES_REQ is a read ioctl with a parameter.
Signed-off-by: Per Åstrand
Signed-off-by: Alison Wang
Acked-by: Peng Fan -
Avoid to use pointers to kernel memory as identify messages, prior this
change, if memory is reused that can lead to unexpected conflicts.Remove ref counting from capabilities since memory is freed in only one
place.Finally, extract the capabilities code in its own files.
Signed-off-by: Davide Grohmann
Signed-off-by: Kristofer Jonsson
Signed-off-by: Alison Wang
Acked-by: Peng Fan -
Avoid undefined behaviour from calling dma_alloc_coherent with
requested size=0 as reported by UBSAN.Signed-off-by: Per Åstrand
Signed-off-by: Alison Wang
Reviewed-by: Peng Fan -
Add message type ETHOSU_CORE_MSG_POWER_RSP for Ethos-U driver.
Signed-off-by: Alison Wang
Reviewed-by: Peng Fan
17 Oct, 2022
1 commit
-
Make sure interface name is null terminated by appending null
character at the end of the string.Signed-off-by: Rohit Raj
12 Oct, 2022
1 commit
-
Add parallel csi support for iMX93
Signed-off-by: Guoniu.zhou
Reviewed-by: Robby Cai
30 Sep, 2022
1 commit
-
This is the 5.15.71 stable release
* tag 'v5.15.71': (144 commits)
Linux 5.15.71
ext4: use locality group preallocation for small closed files
ext4: avoid unnecessary spreading of allocations among groups
...Signed-off-by: Jason Liu
Conflicts:
drivers/net/phy/aquantia_main.c
drivers/tty/serial/fsl_lpuart.c
28 Sep, 2022
2 commits
-
[ Upstream commit e01f5c8d6af231b3b09e23c1fe8a4057cdcc4e42 ]
This device is reported as using the RTL8188EUS chip.
It has the improbable USB ID of 0bda:ffef, which normally would belong
to Realtek, but this ID works for the reporter.Signed-off-by: Larry Finger
Cc: stable
Link: https://lore.kernel.org/r/20220814175027.2689-1-Larry.Finger@lwfinger.net
Signed-off-by: Greg Kroah-Hartman
Signed-off-by: Sasha Levin -
[ Upstream commit 6723b283c44a3fdf9f922ae9788aab38bd909211 ]
The new r8188eu driver doesn't actually support devices with vendor ID 0bda
and product ID f179[0][1][2], remove the ID so owners of these devices
don't have to blacklist the staging driver.[0] https://github.com/lwfinger/rtl8188eu/issues/366#issuecomment-888511731
[1] https://github.com/lwfinger/rtl8188eu/issues/385
[2] https://github.com/lwfinger/rtl8188eu/issues/385#issuecomment-973013539Cc: Larry Finger
CC: Phillip Potter
Signed-off-by: Candy Febriyanto
Link: https://lore.kernel.org/r/YZaBTq9vlMaJDFz2@mainframe.localdomain
Signed-off-by: Greg Kroah-Hartman
Stable-dep-of: e01f5c8d6af2 ("staging: r8188eu: Add Rosewill USB-N150 Nano to device tables")
Signed-off-by: Sasha Levin
27 Sep, 2022
1 commit
-
This is the 5.15.70 stable release
* tag 'v5.15.70': (2444 commits)
Linux 5.15.70
ALSA: hda/sigmatel: Fix unused variable warning for beep power change
cgroup: Add missing cpus_read_lock() to cgroup_attach_task_all()
...Signed-off-by: Jason Liu
Conflicts:
arch/arm/boot/dts/imx6ul.dtsi
arch/arm/mm/mmu.c
arch/arm64/boot/dts/freescale/imx8mp-evk.dts
drivers/gpu/drm/imx/dcss/dcss-kms.c
drivers/media/platform/nxp/imx-jpeg/mxc-jpeg.c
drivers/media/platform/nxp/imx-jpeg/mxc-jpeg.h
drivers/mtd/nand/raw/gpmi-nand/gpmi-nand.c
drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c
drivers/soc/fsl/Kconfig
drivers/soc/imx/gpcv2.c
drivers/usb/dwc3/host.c
net/dsa/slave.c
sound/soc/fsl/imx-card.c
08 Sep, 2022
2 commits
-
commit b2fa9e13bbf101c662c4cd974608242a0db98cfc upstream.
The old rtl8188eu module, removed in commit 55dfa29b43d2 ("staging:
rtl8188eu: remove rtl8188eu driver from staging dir") (Linux kernel
v5.15-rc1), required (through a MODULE_FIRMWARE call()) the
rtlwifi/rtl8188eufw.bin firmware file, which the new r8188eu driver no
longer requires.I have tested a few RTL8188EUS-based Wi-Fi cards and, while supported by
both drivers, they do not work when using the new one and the firmware
wasn't manually loaded. According to Larry Finger, the module
maintainer, all such cards need the firmware and the driver should
depend on it (see the linked mails).Add a proper MODULE_FIRMWARE() call, like it was done in the old driver.
Thanks to Greg Kroah-Hartman and Larry Finger for quick responses to my
questions.Cc: stable
Link: https://answers.launchpad.net/ubuntu/+source/linux-meta-hwe-5.15/+question/702611
Link: https://lore.kernel.org/lkml/YukkBu3TNODO3or9@nx64de-df6d00/
Signed-off-by: Grzegorz Szymaszek
Link: https://lore.kernel.org/r/YulcdKfhA8dPQ78s@nx64de-df6d00
Acked-by: Phillip Potter
Acked-by: Larry Finger
Signed-off-by: Greg Kroah-Hartman -
commit e230a4455ac3e9b112f0367d1b8e255e141afae0 upstream.
_Read/Write_MACREG callbacks are NULL so the read/write_macreg_hdl()
functions don't do anything except free the "pcmd" pointer. It
results in a use after free. Delete them.Fixes: 2865d42c78a9 ("staging: r8712u: Add the new driver to the mainline kernel")
Cc: stable
Reported-by: Zheng Wang
Signed-off-by: Dan Carpenter
Link: https://lore.kernel.org/r/Yw4ASqkYcUhUfoY2@kili
Signed-off-by: Greg Kroah-Hartman
06 Sep, 2022
2 commits
-
When a frame is dequeued by the CEETM, the effective packet size is a
function of the actual packet size, the configured Overhead Accounting
Length (OAL), and the configured Minimum Packet Size (MPS). This
effective packet size is used to update the CEETM shaper credits. This
MPS value should be applied to the actual packet size before the OAL
value is applied. However, the OAL value is applied to the actual packet
size first. Then, the MPS is applied. This results in an incorrect value
being subtracted from the CEETM credit for that particular Logical
Network Interface (LNI) shaper.To avoid an erroneous shaper credit count, do not configure both OAL and
MPS on an LNI shaper.Signed-off-by: Camelia Groza
-
The WQ number is stored in the last 3 bits of the DEST_WQ FQD field. Fix
the mask that is used to extract it from 0x3 to 0x7.Signed-off-by: Camelia Groza
01 Sep, 2022
1 commit
-
Pass interface name to kernel through IOCTL as string instead
of character pointer.Signed-off-by: Rohit Raj
LF-6573
LF-5837
17 Aug, 2022
8 commits
-
[ Upstream commit 6a0c054930d554ad8f8044ef1fc856d9da391c81 ]
There are sleep in atomic context bugs when dm_fsync_timer_callback is
executing. The root cause is that the memory allocation functions with
GFP_KERNEL or GFP_NOIO parameters are called in dm_fsync_timer_callback
which is a timer handler. The call paths that could trigger bugs are
shown below:(interrupt context)
dm_fsync_timer_callback
write_nic_byte
kzalloc(sizeof(data), GFP_KERNEL); //may sleep
usb_control_msg
kmalloc(.., GFP_NOIO); //may sleep
write_nic_dword
kzalloc(sizeof(data), GFP_KERNEL); //may sleep
usb_control_msg
kmalloc(.., GFP_NOIO); //may sleepThis patch uses delayed work to replace timer and moves the operations
that may sleep into the delayed work in order to mitigate bugs.Fixes: 8fc8598e61f6 ("Staging: Added Realtek rtl8192u driver to staging")
Signed-off-by: Duoming Zhou
Link: https://lore.kernel.org/r/20220710103002.63283-1-duoming@zju.edu.cn
Signed-off-by: Greg Kroah-Hartman
Signed-off-by: Sasha Levin -
[ Upstream commit 143201a6435bf65f0115435e9dc6d95c66b908e9 ]
Not all DPB entries will be used most of the time. Unused entries will
thus have invalid timestamps. They will produce negative buffer index
which is not specifically handled. This works just by chance in current
code. It will even produce bogus pointer, but since it's not used, it
won't do any harm.Let's fix that brittle design by skipping writing DPB entry altogether
if timestamp is invalid.Fixes: 86caab29da78 ("media: cedrus: Add HEVC/H.265 decoding support")
Signed-off-by: Jernej Skrabec
Reviewed-by: Ezequiel Garcia
Signed-off-by: Hans Verkuil
Signed-off-by: Mauro Carvalho Chehab
Signed-off-by: Sasha Levin -
[ Upstream commit d8f6f1c56d5469e22eeb7cc1f3580b29e2f0fef5 ]
Fix typos in comments within the Hantro driver.
Signed-off-by: Sebastian Fricke
Reviewed-by: Nicolas Dufresne
Signed-off-by: Hans Verkuil
Signed-off-by: Mauro Carvalho Chehab
Signed-off-by: Sasha Levin -
[ Upstream commit d95a63daca85f4bca3b70e622c75586b5bf0ea5c ]
Reference Picture Set lists provide indices of short and long term
reference in DBP array.
Fix Hantro to not do a look up in DBP entries.
Make documentation more clear about it.[hverkuil: fix typo in commit log]
Signed-off-by: Benjamin Gaignard
Reviewed-by: Ezequiel Garcia
Signed-off-by: Hans Verkuil
Signed-off-by: Mauro Carvalho Chehab
Signed-off-by: Sasha Levin -
[ Upstream commit 53a3e71095c572333ceea30762565dbedec951ca ]
Add a 'postprocessed' boolean property to struct hantro_fmt
to signal that a format is produced by the post-processor.
This will allow to introduce the G2 post-processor in a simple way.Signed-off-by: Ezequiel Garcia
Signed-off-by: Andrzej Pietrasiewicz
Signed-off-by: Hans Verkuil
Signed-off-by: Mauro Carvalho Chehab
Signed-off-by: Sasha Levin -
[ Upstream commit 9393761aec4c56b7f2f19d21f806d316731401c1 ]
When the post-processor hardware block is enabled, the driver
allocates an internal queue of buffers for the decoder enginer,
and uses the vb2 queue for the post-processor engine.For instance, on a G1 core, the decoder engine produces NV12 buffers
and the post-processor engine can produce YUY2 buffers. The decoder
engine expects motion vectors to be appended to the NV12 buffers,
but this is only required for CODECs that need motion vectors,
such as H.264.Fix the post-processor logic accordingly.
Signed-off-by: Ezequiel Garcia
Signed-off-by: Andrzej Pietrasiewicz
Signed-off-by: Hans Verkuil
Signed-off-by: Mauro Carvalho Chehab
Signed-off-by: Sasha Levin -
[ Upstream commit 104a70e1d0bcef28db13c4192b8729086089651c ]
Bit 21 in register 0x24 (slice header info 1) actually represents
negated version of low delay flag. This can be seen in vendor Cedar
library source code. While this flag is not part of the standard, it can
be found in reference HEVC implementation.Fix macro name and change it to flag.
Fixes: 86caab29da78 ("media: cedrus: Add HEVC/H.265 decoding support")
Signed-off-by: Jernej Skrabec
Signed-off-by: Hans Verkuil
Signed-off-by: Mauro Carvalho Chehab
Signed-off-by: Sasha Levin -
commit 09b204eb9de9fdf07d028c41c4331b5cfeb70dd7 upstream.
The three bugs are here:
__func__, s3a_buf->s3a_data->exp_id);
__func__, md_buf->metadata->exp_id);
__func__, dis_buf->dis_data->exp_id);The list iterator 's3a_buf/md_buf/dis_buf' will point to a bogus
position containing HEAD if the list is empty or no element is found.
This case must be checked before any use of the iterator, otherwise
it will lead to a invalid memory access.To fix this bug, add an check. Use a new variable '*_iter' as the
list iterator, while use the old variable '*_buf' as a dedicated
pointer to point to the found element.Link: https://lore.kernel.org/linux-media/20220414041415.3342-1-xiam0nd.tong@gmail.com
Cc: stable@vger.kernel.org
Fixes: ad85094b293e4 ("Revert "media: staging: atomisp: Remove driver"")
Signed-off-by: Xiaomeng Tong
Signed-off-by: Mauro Carvalho Chehab
Signed-off-by: Greg Kroah-Hartman
09 Aug, 2022
1 commit
-
M33 firmware will be loaded when ethosu device opens, and driver
shouldn't send msg to M33 before M33 boot up.Signed-off-by: Feng Guo
Reviewed-by: Peng Fan
03 Aug, 2022
1 commit
-
Fill the num_planes member of v4l2_buf.vb2_buf in discard buffer
since it will be used when fill ISI out buffer register, otherwise,
there are maybe have an issue when capture XRGB24 on iMX8MP platformSigned-off-by: Guoniu.zhou
Reviewed-by: Sandor Yu
02 Aug, 2022
1 commit
-
Reproduce steps:
grecorder-1.0 -u 2 -w 1280 -e 720 -v 1 -i3
gst-launch-1.0 videotestsrc num-buffers=100 ! video/x-raw,format=BGRx !
v4l2convert ! video/x-raw,format=RGB16 ! videoconvert ! waylandsinkKernel will dump when convert format using ISI memory to memory after run
ISI capture. The reason is that ISI m2m will use capture num_planes as plane
count. When the planes of format used in m2m are different with ISI capture,
invalid data will be used, so it leads to kernel dump. The resolution is to
use num_planes in v4l2 buffer which passed by videobuf2 core.Signed-off-by: Guoniu.zhou
Reviewed-by: Sandor Yu
01 Aug, 2022
1 commit
-
Kernel will dump randomly afer run about 650 times AP1302 resolution switching
test. The reason is that ISI HW is filling buffer and user space call streamoff
to stop streaming. After that, driver will do cleanup in stop_streaming. During
this time, ISI interrupt triggered and change the buffer v4l2 state, then back
to stop_streaming and continue to do cleanup, v4l2-core will do buffer state
checking every time when driver try to release buffer to v4l2 core, so it leads
to kernel dump as bellow. The solution is to add spin lock to make sure atomic
cleanup operation without ISI interrupt.Got EOS from element "pipeline0"[ 8640.544083] ------------[ cut here ]------------
.
Execution end[ 8640.549919] WARNING: CPU: 1 PID: 21556 at drivers/media/common/videobuf2/videobuf2-core.c:1960 __vb2_queue_cancel+0x1b0/0x21c
ed after 0:00:02[ 8640.562588] Modules linked in: imx8_media_dev(C)
.716028126
Setti[ 8640.568587] CPU: 1 PID: 21556 Comm: gst-launch-1.0 Tainted: G WC 5.15.52-03897-g7b0c32a97056-dirty #61
ng pipeline to NU[ 8640.580554] Hardware name: NXP i.MX93 11X11 EVK board (DT)
LL ...
[ 8640.587497] pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[ 8640.595131] pc : __vb2_queue_cancel+0x1b0/0x21c
[ 8640.599646] lr : __vb2_queue_cancel+0x34/0x21c
[ 8640.604075] sp : ffff80000b9c3b30
[ 8640.607377] x29: ffff80000b9c3b30 x28: ffff0000091b8700 x27: 0000000000000000
[ 8640.614501] x26: 0000000000000000 x25: ffff80000b9c3d28 x24: ffff00000494f828
[ 8640.621625] x23: ffff000008fda000 x22: 0000000000000009 x21: ffff00000494e7c0
[ 8640.628749] x20: 0000000000000009 x19: ffff00000494e7c0 x18: 0000000000000000
[ 8640.635873] x17: 0000000000000000 x16: 0000000000000000 x15: 0000aaaafbda3510
[ 8640.642997] x14: 0000000000000000 x13: 0000000000000000 x12: 0000000000000000
[ 8640.650121] x11: 0000000000000000 x10: ffff80000b9c3970 x9 : ffff00006fbbf140
[ 8640.657245] x8 : fffffc0001c6e508 x7 : 0000000000000016 x6 : 0000000000000056
[ 8640.664369] x5 : ffff00000409e370 x4 : ffffffffffffffff x3 : ffffffffffffffff
[ 8640.671493] x2 : 0000000000000000 x1 : ffff000008bdaa00 x0 : 00000000ffffffff
[ 8640.678618] Call trace:
[ 8640.681053] __vb2_queue_cancel+0x1b0/0x21c
[ 8640.685229] vb2_core_streamoff+0x28/0xc0
[ 8640.689234] vb2_ioctl_streamoff+0x60/0xb0
[ 8640.693316] mxc_isi_cap_streamoff+0x4c/0x11c
[ 8640.697666] v4l_streamoff+0x28/0x3c
[ 8640.701237] __video_do_ioctl+0x178/0x3dc
[ 8640.705232] video_usercopy+0x368/0x6ec
[ 8640.709054] video_ioctl2+0x1c/0x30
[ 8640.712529] v4l2_ioctl+0x44/0x64
[ 8640.715832] __arm64_sys_ioctl+0xac/0xf0
[ 8640.719749] invoke_syscall+0x48/0x114
[ 8640.723493] el0_svc_common.constprop.0+0xd4/0xfc
[ 8640.728181] do_el0_svc+0x2c/0x94
[ 8640.731484] el0_svc+0x28/0x80
[ 8640.734534] el0t_64_sync_handler+0xa8/0x130
[ 8640.738790] el0t_64_sync+0x1a0/0x1a4
[ 8640.742439] ---[ end trace d116e7f9ad624e8c ]---
[ 8640.748360] ------------[ cut here ]------------Signed-off-by: Guoniu.zhou
Reviewed-by: Robby Cai
29 Jul, 2022
2 commits
-
From V4L2 spec, V4L2_PIX_FMT_NV12 is a two planes which are contiguous
in memory, but for ISI, it can support contiguous and non-contiguous
memory, so add NV12 format support for ISI memory to memory.Signed-off-by: Guoniu.zhou
Reviewed-by: Robby Cai -
From V4L2 spec, V4L2_PIX_FMT_NV12 is a two planes which are contiguous
in memory, but for ISI, it can support contiguous and non-contiguous memory,
so add NV12 format support for ISI capture.Signed-off-by: Guoniu.zhou
Reviewed-by: Robby Cai
27 Jul, 2022
1 commit
-
Incorrect typo issue for DWC mipi csi driver
Signed-off-by: Guoniu.zhou
Reviewed-by: Robby Cai
26 Jul, 2022
1 commit
-
As power management depends on NPU device is opened, this patch adds the
switch for it.Signed-off-by: Alison Wang
25 Jul, 2022
2 commits
-
Add Synopsys DWC MIPI CSI2 driver support for iMX93 platform
Signed-off-by: Guoniu.zhou
Reviewed-by: Robby Cai -
Add ISI support for iMX93
Signed-off-by: Guoniu.zhou
Reviewed-by: Robby Cai
14 Jul, 2022
1 commit
-
Interface names are passed from userspace to the kernel through ioctl.
In order to make sure all strings are null-terminated, initialize all
variables to 0 before copying the data from userspace.Signed-off-by: Camelia Groza