Video Engine has a dedicated interrupt line so this driver doesn't
need to use IRQF_SHARED flag so remove it. Also, it'd be good for
following what Thomas recommended in the IRQF_ONESHOT support
patch like below:
"Note that for now IRQF_ONESHOT cannot be used with IRQF_SHARED to
avoid complex acco
VE_INTERRUPT_CAPTURE_COMPLETE and VE_INTERRUPT_COMP_COMPLETE are
not set at the same time but the current interrupt handling
mechanism of this driver doesn't clear the interrupt flag until
both two are set, and this behavior causes unnecessary interrupt
handler calls. In fact, this driver provides
AST2500 silicon revision A1 and A2 have a silicon bug which causes
extremly long capturing time on specific resolutions (1680 width).
To fix the bug, this commit adjusts the capturing window register
setting to 1728 if detected width is 1680. The compression window
register setting will be kept as
Differently from other Aspeed drivers, this driver calls clock
control APIs in interrupt context. Since ECLK is coupled with a
reset bit in clk-aspeed module, aspeed_clk_enable will make 10ms of
busy waiting delay for triggering the reset and it will eventually
disturb other drivers' interrupt hand
Mode detection doesn't require source buffer allocation so this
commit removes that.
Signed-off-by: Jae Hyun Yoo
Reviewed-by: Eddie James
---
v2 -> v3:
Corrected re-allocation logic of source buffers.
v1 -> v2:
New.
drivers/media/platform/aspeed-video.c | 37 ---
1 f
Video engine clock control can be double disabled and eventually
it causes a kernel warning with stack dump printing out like below:
[ 515.540498] [ cut here ]
[ 515.545174] WARNING: CPU: 0 PID: 1310 at drivers/clk/clk.c:684
clk_core_unprepare+0x13c/0x170
[ 515.553806]
Currently, this driver calls clk_prepare and clk_unprepare from
interrupt context too but these should be called from sleepable
context only. To fix this issue, this commit splits out
clk_enable/disable and clk_prepare/unprepare, and it places
clk_prepare/unprepare calls into the module probe/remov
This patch series improves stability of Aspeed video engine driver by fixing
clock control and irq handling logic in the driver. Also, it adds a couple of
bug fixes and a workaroud for a silicon bug.
Changes since v2:
- Corrected re-allocation logic of source buffer.
- Removed an incorrect patch f
Currently, this driver prints out too much log messages when a
mode change happens, video turned off by screen saver and etc.
Actually, all cases are reported to user space properly. Also,
these are not critical errors but recoverable things, so this
commit changes the log level of some noisy print
In case of watchdog timeout detected while doing mode detection,
it's better triggering video engine hardware reset immediately so
this commit fixes code for the case. Other than the case, it will
trigger video engine hardware reset after RESOLUTION_CHANGE_DELAY.
Signed-off-by: Jae Hyun Yoo
Revie
There are cases that interrupt bits are cleared by a 500ms delayed
work which causes unnecessary irq calls. Also, the current
interrupt handler returns IRQ_HANDLED always but it should return
IRQ_NONE if there is any unhandled interrupt. So this commit
refines the interrupt handling logic to fix th
Hi Robin,
On Thu, May 30, 2019 at 8:26 PM Robin Gong wrote:
> What's soc chip and board you used? Could you post log?
The mainline kernel we've been using does not have any sdma firmware, and
as a result we get this in the log:
# uname -a
Linux Chimera 5.2.0-rc2-00041-g22a1787e8f40 #66
Em Fri, 31 May 2019 16:40:21 -0300
Mauro Carvalho Chehab escreveu:
> Em Fri, 31 May 2019 15:01:06 -0400
> Nicolas Dufresne escreveu:
>
> > Actually I went a little into it, the black and white alternance for
> > lines clearly does not work.
>
> Black and white alternance? Here, the backgrou
Hi Brad,
On Fri, May 31, 2019 at 12:20:45PM -0500, Brad Love wrote:
> Hi Shawn,
>
> Just found time to get back to this. I fixed all the checkpatch strict
> warnings, no problem. Then I noticed a few comments of yours that I
> somehow missed first time around. I've replied inline to those.
>
>
Hello Nicholas and Fabio,
The sdma firmware turned out to be a non-issue: adding "qos=0" to kmssink
(as suggested by Nicholas) is what fixed kmssink performance, unrelated to sdma.
(Although I still see an sdma issue, unrelated to this thread - I will talk
to Robin about it outside of the thread)
Hi Ryan,
On 5/31/2019 9:32 AM, Ryan Chen wrote:
On 5/24/19 6:17 PM, Jae Hyun Yoo wrote:
AST2500 silicon revision A1 and A2 have a silicon bug which causes
extremly long capturing time on specific resolutions (1680 width).
To fix the bug, this commit adjusts the capturing window register
settin
If one tries to access a page that doesn't exists (e.g. [1]), the page
shows that Django is on debug mode. According to Django
documentation[2], is not a good idea running with debugging enabled on
production environment. To fix this, just change the `DEBUG = True` to
`DEBUG = False` on `settings.p
Akinobu Mita writes:
>> Can anyone test this patch and send a tested-by?
>
> In my devicetree, vdd-supply is not defined. So it falls back to the dummy
> regulator and works fine.
Would that work also in a non devicetree build, ie. in a platform_data based one
(as this is one of the mach-pxa tar
Em Fri, 31 May 2019 15:01:06 -0400
Nicolas Dufresne escreveu:
> Le vendredi 31 mai 2019 à 12:40 -0300, Mauro Carvalho Chehab a écrit :
> > Hi all,
> >
> > For a long time, we were running an old Patchwork version. The thing is that
> > we had applied some patches on the top of it, and the upgrad
Mauro Carvalho Chehab writes:
> Hi Sakari,
>
> Em Fri, 31 May 2019 14:27:24 +0300
> Sakari Ailus escreveu:
>
>> Hi Mauro,
>>
>> On Wed, May 29, 2019 at 03:25:18PM -0400, Mauro Carvalho Chehab wrote:
>> > From: Robert Jarzmik
>
> As stated here, this patch is not really mine. It is a rebased ve
Le vendredi 31 mai 2019 à 12:40 -0300, Mauro Carvalho Chehab a écrit :
> Hi all,
>
> For a long time, we were running an old Patchwork version. The thing is that
> we had applied some patches on the top of it, and the upgrade was not
> trivial.
>
> Today, we upgraded it to its latest stable versi
Hi,
Le vendredi 31 mai 2019 à 12:40 -0300, Mauro Carvalho Chehab a écrit :
> Hi all,
>
> For a long time, we were running an old Patchwork version. The thing is that
> we had applied some patches on the top of it, and the upgrade was not
> trivial.
>
> Today, we upgraded it to its latest stable
Hi Sakari,
On Friday, May 31, 2019 1:54:26 PM CEST Sakari Ailus wrote:
> Return an error for which == V4L2_SUBDEV_FORMAT_TRY if
> CONFIG_VIDEO_V4L2_SUBDEV_API is not enabled. This is because the try
> formats are not available in that case.
As far as I can see, there are 21 V4L2 subdevice drive
Hi Sean,
Discussed this with Mauro on irc. I went ahead and made the le16_to_cpu
fixes and set up v6. The series should be good to go now.
Cheers,
Brad
On 31/05/2019 12.20, Brad Love wrote:
> Hi Shawn,
>
> Just found time to get back to this. I fixed all the checkpatch strict
> warnings, no pr
Includes support to identify and use two Hauppauge device.
HVR-1955:
- LGDT3306a ATSC/QAM demod
- si2177 tuner
- cx25840 decoder for analog tv/composite/s-video/audio
HVR-1975 dual-frontend:
- LGDT3306a ATSC/QAM demod
- si2168 DVB-C/T/T2 demod
- si2177 tuner
- cx25840 decoder for analog tv/composi
Works in ATSC and QAM as is, DVB is completely untested.
Firmware required.
Signed-off-by: Brad Love
---
Changes since v5:
- No changes
Changes since v3:
- Fix firmware name to be consistent
drivers/media/tuners/si2157.c | 6 ++
drivers/media/tuners/si2157_priv.h | 3 ++-
2 files chan
i2c client device is the "new" method to attach to dvb modules, include
support for this functionality. Cleanup code has been added to init in
case of failure, as well as to frontend exit.
Required by Hauppauge HVR-1975
Signed-off-by: Brad Love
---
Since v5:
- no changes
Since v4:
- commit messa
All changes are equivalent and backwards compatible.
All current devices have been changed to use fe[0]
Code has been added to dvb init to support cleanup after failure.
Multiple frontends are required by Hauppauge HVR-1975, which is in a
later commit.
Signed-off-by: Brad Love
---
Since v5:
- no
Hauppauge device HVR1955 and HVR1975 are old Cypress based
devices. When originally produced the demods were lacking
upstream drivers and the tuner was unsupported. Well fast
forward to now and the only thing missing is the identification
of si2177 tuner in the si2157 driver, as well as extension
o
Checkpatch reports an extra blank line, so remove such unneeded
line.
Signed-off-by: Fabio Estevam
---
drivers/staging/media/imx/imx7-mipi-csis.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/drivers/staging/media/imx/imx7-mipi-csis.c
b/drivers/staging/media/imx/imx7-mipi-csis.c
index acc9
There is no need for initializing the 'ret' variable as it will
be assigned at:
ret = mipi_csis_parse_dt(pdev, state);
Remove the unneeded initialization.
Signed-off-by: Fabio Estevam
---
drivers/staging/media/imx/imx7-mipi-csis.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Currently the return value from clk_bulk_prepare_enable() is checked,
but it is not propagate it in the case of failure.
Fix it and also move the error message to the caller of
mipi_csis_clk_enable().
Signed-off-by: Fabio Estevam
---
drivers/staging/media/imx/imx7-mipi-csis.c | 14 +++--
Le vendredi 31 mai 2019 à 12:07 -0400, Sven Van Asbroeck a écrit :
> Hello Nicholas, thank you so much for investigating.
>
> On Fri, May 31, 2019 at 11:34 AM Nicolas Dufresne
> wrote:
> > Now, if that pipeline was live, this would be a problem.
>
> This is where my gstreamer knowledge gets rea
Includes support to identify and use two Hauppauge device.
HVR-1955:
- LGDT3306a ATSC/QAM demod
- si2177 tuner
- cx25840 decoder for analog tv/composite/s-video/audio
HVR-1975 dual-frontend:
- LGDT3306a ATSC/QAM demod
- si2168 DVB-C/T/T2 demod
- si2177 tuner
- cx25840 decoder for analog tv/composi
Hauppauge device HVR1955 and HVR1975 are old Cypress based
devices. When originally produced the demods were lacking
upstream drivers and the tuner was unsupported. Well fast
forward to now and the only thing missing is the identification
of si2177 tuner in the si2157 driver, as well as extension
o
i2c client device is the "new" method to attach to dvb modules, include
support for this functionality. Cleanup code has been added to init in
case of failure, as well as to frontend exit.
Required by Hauppauge HVR-19x5 with si2168/lgdt3306a/si2177
Signed-off-by: Brad Love
---
Since v4:
- commit
Works in ATSC and QAM as is, DVB is completely untested.
Firmware required.
Signed-off-by: Brad Love
---
Changes since v4:
- No changes
Changes since v3:
- Fix firmware name to be consistent
drivers/media/tuners/si2157.c | 6 ++
drivers/media/tuners/si2157_priv.h | 3 ++-
2 files chan
Multiple frontends are required by Hauppauge HVR-1975, which is added
in a later commit.
All changes are equivalent and backwards compatible.
All current devices have been changed to use fe[0]
Code has been added to dvb init to support cleanup after failure.
Signed-off-by: Brad Love
---
Since v4
Hi Shawn,
Just found time to get back to this. I fixed all the checkpatch strict
warnings, no problem. Then I noticed a few comments of yours that I
somehow missed first time around. I've replied inline to those.
On 05/04/2019 10.24, Sean Young wrote:
> On Wed, Feb 27, 2019 at 01:16:06PM -0600,
Hi Dongchun,
On Thu, May 23, 2019 at 06:22:02PM +0800, dongchun@mediatek.com wrote:
> From: Dongchun Zhu
>
> Add a V4L2 sub-device driver for the OmniVision image sensor.
> This is a camera sensor using the I2C bus for control and the
> CSI-2 bus for data.
>
> Signed-off-by: Dongchun Zhu
>
> On 5/24/19 6:17 PM, Jae Hyun Yoo wrote:
>> AST2500 silicon revision A1 and A2 have a silicon bug which causes
>> extremly long capturing time on specific resolutions (1680 width).
>> To fix the bug, this commit adjusts the capturing window register
>> setting to 1728 if detected width is 168
Hi Dongchun,
Thanks for the patch.
On Thu, May 23, 2019 at 06:22:04PM +0800, dongchun@mediatek.com wrote:
> From: Dongchun Zhu
>
> Add device tree binding documentation for the OV02A10 CMOS image sensor.
>
> Signed-off-by: Dongchun Zhu
> ---
> .../devicetree/bindings/media/i2c/ov02a10.tx
Hello Nicholas, thank you so much for investigating.
On Fri, May 31, 2019 at 11:34 AM Nicolas Dufresne wrote:
>
> Now, if that pipeline was live, this would be a problem.
This is where my gstreamer knowledge gets really hazy.
What does it mean for a pipeline to be 'live' ?
Would this be a proble
Hi all,
For a long time, we were running an old Patchwork version. The thing is that
we had applied some patches on the top of it, and the upgrade was not
trivial.
Today, we upgraded it to its latest stable version. Just like before,
you can access it via:
https://patchwork.linuxtv.org
Le mercredi 29 mai 2019 à 19:52 -0400, Nicolas Dufresne a écrit :
> Le mercredi 29 mai 2019 à 16:55 -0400, Sven Van Asbroeck a écrit :
> > Philipp and Fabio,
> >
> > On Wed, May 29, 2019 at 12:53 PM Fabio Estevam wrote:
> > > Does this patch from Philipp fix the problem?
> > > https://git.pengutr
The imx_media_pipeline_pad() function return NULL pointers on error, it
never returns error pointers.
Fixes: 3ef46bc97ca2 ("media: staging/imx: Improve pipeline searching")
Signed-off-by: Dan Carpenter
---
drivers/staging/media/imx/imx-media-csi.c | 4 ++--
drivers/staging/media/imx/imx7-media-
Hi Sakari,
Em Fri, 31 May 2019 14:27:24 +0300
Sakari Ailus escreveu:
> Hi Mauro,
>
> On Wed, May 29, 2019 at 03:25:18PM -0400, Mauro Carvalho Chehab wrote:
> > From: Robert Jarzmik
As stated here, this patch is not really mine. It is a rebased version of a
patch that was delegated to a sub-m
Return an error for which == V4L2_SUBDEV_FORMAT_TRY if
CONFIG_VIDEO_V4L2_SUBDEV_API is not enabled. This is because the try
formats are not available in that case. Do the check here so that drivers
don't need to.
Signed-off-by: Sakari Ailus
Depends-on: ("media: v4l2-subdev: Verify v4l2_subdev_cal
Hi Marco,
On Thu, Apr 04, 2019 at 09:39:56AM +0200, Marco Felsch wrote:
> Hi,
>
> during my work on [1] I prepared a patch to avoid driver interal ifdef
> dances for:
> - v4l2_subdev_get_try_format
> - v4l2_subdev_get_try_crop
> - v4l2_subdev_get_try_compose
> helper functions. Jacopo did some
Return an error for which == V4L2_SUBDEV_FORMAT_TRY if
CONFIG_VIDEO_V4L2_SUBDEV_API is not enabled. This is because the try
formats are not available in that case. Do the check here so that drivers
don't need to.
Signed-off-by: Sakari Ailus
Depends-on: ("media: v4l2-subdev: Verify v4l2_subdev_cal
Hi Mauro,
On Wed, May 29, 2019 at 03:25:18PM -0400, Mauro Carvalho Chehab wrote:
> From: Robert Jarzmik
>
> In the soc_camera removal, the board specific power callback was
> dropped. This at least will remove the power optimization from ezx and
> em-x270 pxa based boards.
>
> As to recreate th
On Friday, May 10, 2019 12:09:26 PM CEST Sakari Ailus wrote:
> Allow drivers and firmware tell ACPI that there's no need to power on a
> device for probe. This requires both a hint from the firmware as well as
> an indication from a driver to leave the device off.
>
> Signed-off-by: Sakari Ailus
There are several other SoCs that contain Hantro IP based VPUs, such as
NXP i.MX8MQ (Hantro G1 and G2) and i.MX8MM (Hantro G1, G2, and H1). To
maximize code sharing, add initial support for these SoCs to the
Rockchip VPU driver, after renaming it to Hantro VPU.
This series is based on the br-v5.3g
Rename the driver and all relevant identifiers from Rockchip to Hantro,
as other Hantro IP based VPU implementations can be supported by the
same driver.
The RK3288 decoder is Hantro G1 based, the encoder is Hantro H1.
This patch just renames, no functional changes.
Signed-off-by: Philipp Zabel
Add support for multiple register ranges with SoC specific names.
Signed-off-by: Philipp Zabel
---
Changes since v2 [1]:
- Move array of reg_names out of struct hantro_variant
- Use ARRAY_SIZE to set num_regs
- Dynamically allocate base address array in struct hantro_dev and
rename it to ba
On i.MX8MQ/MM a separate control block contains registers for per-core
resets, clock gating, and fuse register control.
Signed-off-by: Philipp Zabel
---
drivers/staging/media/hantro/hantro.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/staging/media/hantro/hantro.h
b/drivers/st
It can be helpful to know which video device was registered at which
device node.
Signed-off-by: Philipp Zabel
Reviewed-by: Boris Brezillon
---
drivers/staging/media/hantro/hantro_drv.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/staging/media/hantro/hantro_drv
For now this just enables MPEG-2 decoding on the Hantro G1 on i.MX8MQ.
Signed-off-by: Philipp Zabel
---
Changes since v2 [1]:
- Adapted to changes in patches 4 and 5
[1] https://patchwork.linuxtv.org/patch/56420/
---
drivers/staging/media/hantro/Kconfig| 8 +-
drivers/staging/media/h
This should enable MPEG-2 decoding on the Hantro G1 and JPEG encoding on
the Hantro H1 on i.MX8MM.
Signed-off-by: Philipp Zabel
Changes since v2 [1]:
- Adapted to changes in patches 4 and 5
[1] https://patchwork.linuxtv.org/patch/56425/
---
drivers/staging/media/hantro/hantro_drv.c | 1 +
The i.MX8MQ bindings will use different IRQ names ("g1" instead of
"vdpu", and "g2"), so make them configurable. This also allows to
register more than two IRQs, which will be required for i.MX8MM support
later (it will add "h1" instead of "vepu").
Signed-off-by: Philipp Zabel
---
Changes since v
It seems that on i.MX8MQ the power domain controller does not propagate
resets to the VPU cores on resume. Add a callback to allow implementing
manual reset of the VPU cores after ungating the power domain.
Signed-off-by: Philipp Zabel
Reviewed-by: Boris Brezillon
---
drivers/staging/media/hant
Dynamically allocate clocks and move clock names out of struct
hantro_variant. This lifts the four clock limit and allows to use
ARRAY_SIZE() to fill .num_clocks to reduce the risk of mismatches.
Signed-off-by: Philipp Zabel
---
New in v3.
---
drivers/staging/media/hantro/hantro.h| 6 ++
Add devicetree binding documentation for the Hantro G1/G2 VPU on i.MX8MQ
and for the Hantro G1/G2/H1 VPU on i.MX8MM.
Signed-off-by: Philipp Zabel
---
.../devicetree/bindings/media/imx8m-vpu.txt | 56 +++
1 file changed, 56 insertions(+)
create mode 100644 Documentation/devicet
63 matches
Mail list logo
