media: staging/intel-ipu3: Fix race condition during set_fmt
Do not modify imgu_pipe->nodes[inode].vdev_fmt.fmt.pix_mp, until the
format has been correctly validated.
Otherwise, even if we use a backup variable, there is a period of time
where imgu_pipe->nodes[inode].vdev_fmt.fmt.pix_mp might have an invalid
value that can be used by other functions.
Arnd Bergmann [Mon, 22 Mar 2021 10:07:25 +0000 (11:07 +0100)]
media: dvb-usb: avoid -Wempty-body warnings
Building with 'make W=1' shows many warnings -Wempty-body warnings like
drivers/media/usb/dvb-usb/vp702x-fe.c: In function 'vp702x_fe_set_frontend':
drivers/media/usb/dvb-usb/vp702x-fe.c:190:46: error: suggest braces around empty body in an 'else' statement [-Werror=empty-body]
190 | deb_fe("tuning succeeded.\n");
drivers/media/usb/dvb-usb/dtt200u.c: In function 'dtt200u_rc_query':
drivers/media/usb/dvb-usb/dtt200u.c:124:58: error: suggest braces around empty body in an 'if' statement [-Werror=empty-body]
124 | deb_info("st->data: %*ph\n", 5, st->data);
drivers/media/usb/dvb-usb/m920x.c: In function 'm920x_rc_query':
drivers/media/usb/dvb-usb/m920x.c:207:58: error: suggest braces around empty body in an 'if' statement [-Werror=empty-body]
207 | deb("Unknown rc key %02x\n", rc_state[1]);
Change the empty dprintk() macros to no_printk(), which avoids this
warning and adds format string checking.
Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Sean Young <sean@mess.org> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Arnd Bergmann [Mon, 22 Mar 2021 10:07:24 +0000 (11:07 +0100)]
media: flexcop: avoid -Wempty-body warning
Building with 'make W=1' shows many warnings -Wempty-body warnings like
drivers/media/common/b2c2/flexcop-misc.c: In function 'flexcop_determine_revision':
drivers/media/common/b2c2/flexcop-misc.c:35:85: error: suggest braces around empty body in an 'else' statement [-Werror=empty-body]
35 | deb_info("this FlexCop has the 6 basic main hardware pid filter.\n");
drivers/media/usb/b2c2/flexcop-usb.c: In function 'flexcop_usb_process_frame':
drivers/media/usb/b2c2/flexcop-usb.c:357:79: error: suggest braces around empty body in an 'else' statement [-Werror=empty-body]
357 | deb_ts("not ts packet %*ph\n", 4, b+2);
| ^
drivers/media/common/b2c2/flexcop-misc.c: In function 'flexcop_determine_revision':
drivers/media/common/b2c2/flexcop-misc.c:35:85: error: suggest braces around empty body in an 'else' statement [-Werror=empty-body]
35 | deb_info("this FlexCop has the 6 basic main hardware pid filter.\n");
| ^
Change the empty dprintk() macros to no_printk(), which avoids this
warning and adds format string checking.
Signed-off-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Sean Young <sean@mess.org> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Andre Przywara [Wed, 27 Jan 2021 17:24:51 +0000 (18:24 +0100)]
media: dt-bindings: media: IR: Add H616 IR compatible string
Add the obvious compatible name to the existing IR binding, and pair
it with the existing A31 fallback compatible string, as the devices
are compatible.
On the way use enums to group all compatible devices together.
Signed-off-by: Andre Przywara <andre.przywara@arm.com> Acked-by: Rob Herring <robh@kernel.org> Signed-off-by: Sean Young <sean@mess.org> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
In case of error in dvb_usb_adapter_dvb_init() or
dvb_usb_adapter_dvb_init() d->num_adapters_initialized won't be
incremented, but dvb_usb_adapter_exit() relies on it:
for (n = 0; n < d->num_adapters_initialized; n++)
So, allocated objects won't be freed.
Signed-off-by: Pavel Skripkin <paskripkin@gmail.com> Reported-by: syzbot+3c2be7424cea3b932b0e@syzkaller.appspotmail.com Signed-off-by: Sean Young <sean@mess.org> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Arnd Bergmann [Mon, 22 Mar 2021 16:46:00 +0000 (17:46 +0100)]
media: dvb-usb-remote: fix dvb_usb_nec_rc_key_to_event type mismatch
gcc-11 warns about the prototype not exactly matching the function
definition:
drivers/media/usb/dvb-usb/dvb-usb-remote.c:363:20: error: argument 2 of type ‘u8[5]’ {aka ‘unsigned char[5]’} with mismatched bound [-Werror=array-parameter=]
363 | u8 keybuf[5], u32 *event, int *state)
| ~~~^~~~~~~~~
In file included from drivers/media/usb/dvb-usb/dvb-usb-common.h:13,
from drivers/media/usb/dvb-usb/dvb-usb-remote.c:9:
drivers/media/usb/dvb-usb/dvb-usb.h:490:65: note: previously declared as ‘u8[]’ {aka ‘unsigned char[]’}
490 | extern int dvb_usb_nec_rc_key_to_event(struct dvb_usb_device *, u8[], u32 *, int *);
| ^~~~
Colin Ian King [Tue, 23 Mar 2021 14:00:47 +0000 (15:00 +0100)]
media: dvb-frontends: Remove redundant error check on variable ret
An earlier commit removed a call to lgdt3306a_spectral_inversion and
omitted to remove the error return check. The check on ret is now
redundant and can be removed.
Addresses-Coverity: ("Logically dead code")
Fixes: d4a3fa6652e3 ("media: dvb-frontends: lgdt3306a.c: remove dead code") Signed-off-by: Colin Ian King <colin.king@canonical.com> Signed-off-by: Sean Young <sean@mess.org> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Dinghao Liu [Wed, 7 Apr 2021 05:46:06 +0000 (07:46 +0200)]
media: sun8i-di: Fix runtime PM imbalance in deinterlace_start_streaming
pm_runtime_get_sync() will increase the runtime PM counter
even it returns an error. Thus a pairing decrement is needed
to prevent refcount leak. Fix this by replacing this API with
pm_runtime_resume_and_get(), which will not change the runtime
PM counter on error.
Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Dinghao Liu [Wed, 7 Apr 2021 05:43:13 +0000 (07:43 +0200)]
media: platform: sti: Fix runtime PM imbalance in regs_show
pm_runtime_get_sync() will increase the runtime PM counter
even it returns an error. Thus a pairing decrement is needed
to prevent refcount leak. Fix this by replacing this API with
pm_runtime_resume_and_get(), which will not change the runtime
PM counter on error.
Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
media: imx-jpeg: Fix double free in mxc_jpeg_remove
The video_unregister_device already calls video_device_release,
so remove video_device_release, to avoid a double free, when removing
the module. This showed up in a repeated rmmod/insmod scenario.
Yang Yingliang [Tue, 6 Apr 2021 13:50:53 +0000 (15:50 +0200)]
media: i2c: adv7842: fix possible use-after-free in adv7842_remove()
This driver's remove path calls cancel_delayed_work(). However, that
function does not wait until the work function finishes. This means
that the callback function may still be running after the driver's
remove function has finished, which would result in a use-after-free.
Fix by calling cancel_delayed_work_sync(), which ensures that
the work is properly cancelled, no longer running, and unable
to re-schedule itself.
Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Yang Yingliang <yangyingliang@huawei.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Yang Yingliang [Tue, 6 Apr 2021 13:49:45 +0000 (15:49 +0200)]
media: i2c: tda1997: Fix possible use-after-free in tda1997x_remove()
This driver's remove path calls cancel_delayed_work(). However, that
function does not wait until the work function finishes. This means
that the callback function may still be running after the driver's
remove function has finished, which would result in a use-after-free.
Fix by calling cancel_delayed_work_sync(), which ensures that
the work is properly cancelled, no longer running, and unable
to re-schedule itself.
Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Yang Yingliang <yangyingliang@huawei.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Yang Yingliang [Tue, 6 Apr 2021 13:48:12 +0000 (15:48 +0200)]
media: i2c: adv7511-v4l2: fix possible use-after-free in adv7511_remove()
This driver's remove path calls cancel_delayed_work(). However, that
function does not wait until the work function finishes. This means
that the callback function may still be running after the driver's
remove function has finished, which would result in a use-after-free.
Fix by calling cancel_delayed_work_sync(), which ensures that
the work is properly cancelled, no longer running, and unable
to re-schedule itself.
Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Yang Yingliang <yangyingliang@huawei.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Yang Yingliang [Tue, 6 Apr 2021 13:42:46 +0000 (15:42 +0200)]
media: adv7604: fix possible use-after-free in adv76xx_remove()
This driver's remove path calls cancel_delayed_work(). However, that
function does not wait until the work function finishes. This means
that the callback function may still be running after the driver's
remove function has finished, which would result in a use-after-free.
Fix by calling cancel_delayed_work_sync(), which ensures that
the work is properly cancelled, no longer running, and unable
to re-schedule itself.
Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Yang Yingliang <yangyingliang@huawei.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Yang Yingliang [Tue, 6 Apr 2021 13:39:29 +0000 (15:39 +0200)]
media: tc358743: fix possible use-after-free in tc358743_remove()
This driver's remove path calls cancel_delayed_work(). However, that
function does not wait until the work function finishes. This means
that the callback function may still be running after the driver's
remove function has finished, which would result in a use-after-free.
Fix by calling cancel_delayed_work_sync(), which ensures that
the work is properly cancelled, no longer running, and unable
to re-schedule itself.
Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Yang Yingliang <yangyingliang@huawei.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
media: venus: don't de-reference NULL pointers at IRQ time
Smatch is warning that:
drivers/media/platform/qcom/venus/hfi_venus.c:1100 venus_isr() warn: variable dereferenced before check 'hdev' (see line 1097)
The logic basically does:
hdev = to_hfi_priv(core);
with is translated to:
hdev = core->priv;
If the IRQ code can receive a NULL pointer for hdev, there's
a bug there, as it will first try to de-reference the pointer,
and then check if it is null.
After looking at the code, it seems that this indeed can happen:
Basically, the venus IRQ thread is started with:
devm_request_threaded_irq()
So, it will only be freed after the driver unbinds.
In order to prevent the IRQ code to work with freed data,
the logic at venus_hfi_destroy() sets core->priv to NULL,
which would make the IRQ code to ignore any pending IRQs.
There is, however a race condition, as core->priv is set
to NULL only after being freed. So, we need also to move the
core->priv = NULL to happen earlier.
media: venus: use NULL instead of zero for pointers
As reported by sparse:
drivers/media/platform/qcom/venus/core.c:227:41: warning: Using plain integer as NULL pointer
drivers/media/platform/qcom/venus/core.c:228:34: warning: Using plain integer as NULL pointer
Two vars are using zero instead of NULL for pointers. Not really
an issue, but using NULL makes it clearer that the init data is
expecting a pointer.
The file name: Documentation/devicetree/bindings/media/i2c/ov2680.yaml
should be, instead: Documentation/devicetree/bindings/media/i2c/ovti,ov2680.yaml.
Update its cross-reference accordingly.
Fixes: 57226cd8c8bf ("media: dt-bindings: ov2680: convert bindings to yaml") Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The file name: Documentation/devicetree/bindings/media/i2c/rdacm2x-gmsl.yaml
should be, instead: Documentation/devicetree/bindings/media/i2c/imi,rdacm2x-gmsl.yaml.
media: s5p-mfc: Fix kernel-doc entries in s5p_mfc_common.h
Fixes scripts/kernel-doc warnings:
s5p_mfc_common.h:343: warning: Function parameter or member 'fw_buf' not described in 's5p_mfc_dev'
s5p_mfc_common.h:343: warning: Function parameter or member 'mem_size' not described in 's5p_mfc_dev'
s5p_mfc_common.h:343: warning: Function parameter or member 'mem_base' not described in 's5p_mfc_dev'
s5p_mfc_common.h:343: warning: Function parameter or member 'mem_bitmap' not described in 's5p_mfc_dev'
s5p_mfc_common.h:343: warning: Function parameter or member 'mem_virt' not described in 's5p_mfc_dev'
s5p_mfc_common.h:343: warning: Function parameter or member 'dma_base' not described in 's5p_mfc_dev'
s5p_mfc_common.h:343: warning: Function parameter or member 'watchdog_timer' not described in 's5p_mfc_dev'
s5p_mfc_common.h:343: warning: Function parameter or member 'fw_get_done' not described in 's5p_mfc_dev'
s5p_mfc_common.h:343: warning: Function parameter or member 'risc_on' not described in 's5p_mfc_dev'
s5p_mfc_common.h:714: warning: Function parameter or member 'src_bufs_cnt' not described in 's5p_mfc_ctx'
s5p_mfc_common.h:714: warning: Function parameter or member 'dst_bufs_cnt' not described in 's5p_mfc_ctx'
s5p_mfc_common.h:714: warning: Function parameter or member 'pb_count' not described in 's5p_mfc_ctx'
s5p_mfc_common.h:714: warning: Function parameter or member 'force_frame_type' not described in 's5p_mfc_ctx'
s5p_mfc_common.h:714: warning: Function parameter or member 'slice_mode' not described in 's5p_mfc_ctx'
s5p_mfc_common.h:714: warning: Function parameter or member 'slice_size' not described in 's5p_mfc_ctx'
s5p_mfc_common.h:714: warning: Function parameter or member 'frame_tag' not described in 's5p_mfc_ctx'
s5p_mfc_common.h:714: warning: Function parameter or member 'scratch_buf_size' not described in 's5p_mfc_ctx'
Reported-by: Hans verkuil <hverkuil@xs4all.nl> Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
media: exynos4-is: Fix kernel-doc entries in fimc-is.h
Fixes scripts/kernel-doc warnings:
fimc-is.h:286: warning: Function parameter or member 'fw' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'memory' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'f_w' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'isp' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'sensor' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'setfile' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'ctrl_handler' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'irq' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'sensor_index' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'i2h_cmd' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'h2i_cmd' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'fd_header' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'config' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'config_index' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'is_p_region' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'is_dma_p_region' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'is_shared_region' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'af' not described in 'fimc_is'
fimc-is.h:286: warning: Function parameter or member 'debugfs_entry' not described in 'fimc_is'
The f_w field is unused so remove it.
Reported-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Hans Verkuil [Fri, 26 Mar 2021 09:03:32 +0000 (10:03 +0100)]
media: adv7511-v4l2: correctly report EDIDs of 1 block
If the EDID has an odd number of blocks (usually just 1, but the
same problem occurs with 3 blocks), then VIDIOC_G_EDID reported 2
(or 4) blocks. Fix this.
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Hans Verkuil [Thu, 25 Mar 2021 10:39:37 +0000 (11:39 +0100)]
media: adv7604: support EDIDs up to 4 blocks
While the adv7604/11/12 hardware supported EDIDs up to 4 blocks, the
driver didn't. This patch adds support for this. It also improves support
for EDIDs that do not have a Source Physical Address: in that case the
spa location is set to the first byte of the second block, and the
'physical address' is just the two bytes at that location. This is per
the suggestion in the adv76xx documentation for such EDIDs.
Hans Verkuil [Wed, 24 Mar 2021 07:56:42 +0000 (08:56 +0100)]
media: adv7604: fix HPD support for adv7611/12
For the adv7604 the hotplug detect pin is controlled through gpio pins from
the SoC, but the adv7611 and adv7612 control the hotplug detect pin
themselves.
But the driver had no support for this, so the HPD was always high, even
when changing the EDID. Add proper support for this to the driver.
Daniel Almeida [Tue, 23 Mar 2021 18:57:09 +0000 (19:57 +0100)]
media: rkvdec: Do not require all controls to be present in every request
According to the v4l2 api, it is allowed to skip
setting a control if its contents haven't changed for performance
reasons: userspace should only update the controls that changed from
last frame rather then updating them all. Still some ancient code
that checks for mandatory controls has been left in this driver.
Remove it.
Fixes: cd33c830448b ("media: rkvdec: Add the rkvdec driver") Signed-off-by: Daniel Almeida <daniel.almeida@collabora.com> Reviewed-by: Ezequiel Garcia <ezequiel@collabora.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
media: VIDEO_IMX8_JPEG should depend on ARCH_MXC and not default to m
The i.MX8 QXP/QM integrated JPEG encoder/decoder is only present on
Freescale/NXP i.MX8 QXP and QM SoCs. Hence add a dependency on
ARCH_MXC, to prevent asking the user about this driver when configuring
a kernel without i.MX8 support.
Drop the "default m" (which means "default y" if CONFIG_MODULES is not
enabled), as merely enabling CONFIG_COMPILE_TEST should not enable
additional code.
The wrappers in include/linux/pci-dma-compat.h should go away.
The patch has been generated with the coccinelle script below and has been
hand modified to replace GFP_ with a correct flag.
It has been compile tested.
When memory is allocated in 'tw68_risc_buffer()' (tw68-risc.c) GFP_KERNEL
can be used because this function is only called from a vb2_ops buf_prepare
function.
The call chain is:
tw68_video_qops.buf_prepare (tw68-video.c)
--> tw68_buf_prepare (tw68-video.c)
--> tw68_risc_buffer
The wrappers in include/linux/pci-dma-compat.h should go away.
The patch has been generated with the coccinelle script below and has been
hand modified to replace GFP_ with a correct flag.
It has been compile tested.
When memory is allocated in 'tw686x_audio_dma_alloc()' (tw686x-audio.c)
GFP_KERNEL can be used because it is only called from a probe function
and no spinlock is taken in the between.
The call chain is:
tw686x_probe (tw686x-core.c)
--> tw686x_audio_init (tw686x-audio.c)
--> tw686x_audio_dma_alloc (tw686x-audio.c)
When memory is allocated in 'tw686x_memcpy_dma_alloc()' and in
'tw686x_sg_dma_alloc()' (tw686x-video.c) GFP_KERNEL can be used because
these functions are .alloc functions from a tw686x_dma_ops structure.
The wrappers in include/linux/pci-dma-compat.h should go away.
The patch has been generated with the coccinelle script below and has been
hand modified to replace GFP_ with a correct flag.
It has been compile tested.
When memory is allocated in 'cx23885_risc_buffer()' GFP_KERNEL can be used
because this function is only called from a vb2_ops buf_prepare function.
The call chain is:
cx23885_video_qops.buf_prepare (cx23885-video.c)
--> buffer_prepare (cx23885-video.c)
--> cx23885_risc_buffer
When memory is allocated in 'cx23885_risc_databuffer()' GFP_KERNEL can be
used because this function is only called from a function that already uses
GFP_KERNEL or from a vb2_ops buf_prepare
function.
The call chains are:
snd_cx23885_hw_params (cx23885-alsa.c) --> use GFP_KERNEL
--> cx23885_risc_databuffer
cx23885_qops.buffer_prepare (cx23885-417.c)
or
dvb_qops.buffer_prepare (cx23885-dvb.c)
--> cx23885_buf_prepare
--> cx23885_risc_databuffer
When memory is allocated in 'cx23885_risc_vbibuffer()' GFP_KERNEL can be
used because this function is only called from a vb2_ops buf_prepare
function.
The call chains are:
cx23885_vbi_qops.buffer_prepare (cx23885-vbi.c)
--> cx23885_risc_vbibuffer
Irui Wang [Thu, 25 Mar 2021 12:26:25 +0000 (13:26 +0100)]
media: mtk-vcodec: Separating mtk encoder driver
MTK H264 Encoder(VENC_SYS) and VP8 Encoder(VENC_LT_SYS) are two
independent hardware instance. They have their owner interrupt,
register mapping, and special clocks.
This patch separates them into two devices. This is a preparing
patch for adding device_link between the larbs and venc-device.
It's mainly for fixing the problem:
https://lkml.org/lkml/2019/9/3/316
Acked-by: Tiffany Lin <tiffany.lin@mediatek.com> Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org> Signed-off-by: Maoguang Meng <maoguang.meng@mediatek.com> Signed-off-by: Irui Wang <irui.wang@mediatek.com> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Updates binding document since the avc and vp8 hardware encoder in
MT8173 are now separated. Separate "mediatek,mt8173-vcodec-enc" to
"mediatek,mt8173-vcodec-enc-vp8" and "mediatek,mt8173-vcodec-enc".
This patch is not a compatible change, but we must do this modifaction
because MediaTek IOMMU add the device_link between the smi-larb
device and venc_device, if the venc device call the pm_runtime_get_sync,
the smi-larb's pm_runtime_get_sync also be called automatically.
There is a bit of backward compatibility for avc encoder, the avc
encoder device node still has compatible "mediatek,mt8173-vcodec-enc".
Acked-by: Tiffany Lin <tiffany.lin@mediatek.com> Signed-off-by: Hsin-Yi Wang <hsinyi@chromium.org> Signed-off-by: Maoguang Meng <maoguang.meng@mediatek.com> Signed-off-by: Irui Wang <irui.wang@mediatek.com> Reviewed-by: Rob Herring <robh@kernel.org> Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Hans Verkuil [Sat, 27 Mar 2021 11:27:40 +0000 (12:27 +0100)]
media: v4l2-ctrls.c: fix race condition in hdl->requests list
When a request is re-inited it will release all control handler
objects that are still in the request. It does that by unbinding
and putting all those objects. When the object is unbound the
obj->req pointer is set to NULL, and the object's unbind op is
called. When the object it put the object's release op is called
to free the memory.
For a request object that contains a control handler that means
that v4l2_ctrl_handler_free() is called in the release op.
A control handler used in a request has a pointer to the main
control handler that is created by the driver and contains the
current state of all controls. If the device is unbound (due to
rmmod or a forced unbind), then that main handler is freed, again
by calling v4l2_ctrl_handler_free(), and any outstanding request
objects that refer to that main handler have to be unbound and put
as well.
It does that by this test:
if (!hdl->req_obj.req && !list_empty(&hdl->requests)) {
I.e. the handler has no pointer to a request, so is the main
handler, and one or more request objects refer to this main
handler.
However, this test is wrong since hdl->req_obj.req is actually
NULL when re-initing a request (the object unbind will set req to
NULL), and the only reason this seemingly worked is that the
requests list is typically empty since the request's unbind op
will remove the handler from the requests list.
But if another thread is at the same time adding a new control
to a request, then there is a race condition where one thread
is removing a control handler object from the requests list and
another thread is adding one. The result is that hdl->requests
is no longer empty and the code thinks that a main handler is
being freed instead of a control handler that is part of a request.
There are two bugs here: first the test for hdl->req_obj.req: this
should be hdl->req_obj.ops since only the main control handler will
have a NULL pointer there.
The second is that adding or deleting request objects from the
requests list of the main handler isn't protected by taking the
main handler's lock.
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Reported-by: John Cox <jc@kynesim.co.uk> Fixes: 6fa6f831f095 ("media: v4l2-ctrls: add core request support") Tested-by: John Cox <jc@kynesim.co.uk> Reported-by: John Cox <jc@kynesim.co.uk> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Hans Verkuil [Tue, 18 Aug 2020 14:37:09 +0000 (16:37 +0200)]
media: vivid: add read-only int32 control
This read-only int32 control is used to test read-only controls in
combination with requests. It is set by the driver to the buffer sequence
counter module 256.
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Presently, minimum buffer count for output buffers for h264 &
hevc codec is kept as 8 which might not be sufficient for
few use cases so increasing the value to 18.
media: venus: pm: Toggle 6xx wrapper power in vcodec_control
When in vcodec_control_v4() on the 4xx path we select between one of two
vcodec cores to toggle power control bits. In 6xx there is only one core
which offsets the relevant control registers to a slightly different
address.
This patch accounts for the address offset for 6xx silicon.
This patch takes the downstream AXI halt routine and applies it when
IS_V6() is true.
bod: Converted to readl_poll_timeout()
Removed poll timeout for LPI register, testing showed the value
would always timeout and work, so the polling did nothing of value.
media: venus: core,pm: Vote for min clk freq during venus boot
Vote for min clk frequency for core clks during prepare and enable clocks
at boot sequence. Without this the controller clock runs at very low value
(9.6MHz) which is not sufficient to boot venus.
media: venus: hfi, vdec: v6 Add IS_V6() to existing IS_V4() if locations
In various places in the venus codebase we have if (IS_V4()) which takes
the code down paths for 4xx silicon. This logic is broadly applicable to
6xx silicon also. In this patch we add IS_V6() to various IS_V4() decision
locations.
media: venus: hfi,pm,firmware: Convert to block relative addressing
An upcoming silicon change places a number of existing blocks within the
Venus at different relative offsets to the base address of IO region.
In order to handle this difference this patch changes the address offsets
of the registers to function as offsets relative to the relevant sub-block
of registers within the IO region not the base address of the IO region.
As a result of this change venus_readl() and venus_writel() are deleted.
media: venus: core: Add io base variables for each block
New silicon means that the pre-determined offsets we have been using
in this driver no longer hold. Existing blocks of registers can exist at
different offsets relative to the IO base address.
This commit adds a routine to assign the IO base hooks a subsequent commit
will convert from absolute to relative addressing.
It is observed that on Venus v1 the default header-mode is producing
a bitstream which is not playble. Change the default header-mode to
joined with 1st frame.
media: venus: pm_helpers: Set opp clock name for v1
The rate of the core clock is set through devm_pm_opp_set_rate and
to avoid errors from it we have to set the name of the clock via
dev_pm_opp_set_clkname.
Fixes: 9a538b83612c ("media: venus: core: Add support for opp tables/perf voting") Cc: stable@vger.kernel.org # v5.10+ Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org> Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
media: venus: hfi_parser: Check for instance after hfi platform get
The inst function argument is != NULL only for Venus v1 and
we did not migrate v1 to a hfi_platform abstraction yet. So
check for instance != NULL only after hfi_platform_get returns
no error.
Fixes: e29929266be1 ("media: venus: Get codecs and capabilities from hfi platform") Cc: stable@vger.kernel.org # v5.12 Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org> Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
media: venus: hfi_parser: Don't initialize parser on v1
The Venus v1 behaves differently comparing with the other Venus
version in respect to capability parsing and when they are send
to the driver. So we don't need to initialize hfi parser for
multiple invocations like what we do for > v1 Venus versions.
Dikshita Agarwal [Wed, 24 Mar 2021 09:54:32 +0000 (10:54 +0100)]
media: venus: venc: Add support for Long Term Reference (LTR) controls
Add support for below LTR controls in encoder:
- V4L2_CID_MPEG_VIDEO_LTR_COUNT
- V4L2_CID_MPEG_VIDEO_FRAME_LTR_INDEX
- V4L2_CID_MPEG_VIDEO_USE_LTR_FRAMES
Dikshita Agarwal [Wed, 24 Mar 2021 09:54:31 +0000 (10:54 +0100)]
media: v4l2-ctrl: add controls for long term reference.
Long Term Reference (LTR) frames are the frames that are encoded
sometime in the past and stored in the DPB buffer list to be used
as reference to encode future frames.
This change adds controls to enable this feature.