- 10 Aug, 2023 40 commits
-
-
Konrad Dybcio authored
Add vpu_version where I was able to retrieve the information to allow for more precise hardware-specific code path matching. Reviewed-by:
Dikshita Agarwal <quic_dikshita@quicinc.com> Reviewed-by:
Vikash Garodia <quic_vgarodia@quicinc.com> Signed-off-by:
Konrad Dybcio <konrad.dybcio@linaro.org> Signed-off-by:
Stanimir Varbanov <stanimir.k.varbanov@gmail.com> Signed-off-by:
Hans Verkuil <hverkuil-cisco@xs4all.nl>
-
Konrad Dybcio authored
The Video Processing Unit hardware version is the differentiator, based on which the video driver should decide which code paths to take. Up until now, we've relied on HFI versions instead, but that was just a happy accident between recent SoCs. Add a field in the res struct and add correlated definitions that will be used to account for the aforementioned differences. Reviewed-by:
-
Konrad Dybcio authored
The startup procedure shouldn't be started with interrupts masked, as that may entail silent failures. Kick off initialization only after the interrupts are unmasked. Cc: stable@vger.kernel.org # v4.12+ Fixes: d96d3f30 ("[media] media: venus: hfi: add Venus HFI files") Signed-off-by:
-
Konrad Dybcio authored
As per information from Qualcomm [1], this property is not really supported beyond msm8916 (HFI V1) and some newer HFI versions really dislike receiving it, going as far as crashing the device. Only consider toggling it (via the module option) on HFIV1. While at it, get rid of the global static variable (which defaulted to zero) which was never explicitly assigned to for V1. Note: [1] is a reply to the actual message in question, as lore did not properly receive some of the emails.. [1] https://lore.kernel.org/lkml/955cd520-3881-0c22-d818-13fe9a47e124@linaro.org/ Fixes: 7ed9e0b3 ("media: venus: hfi, vdec: v6 Add IS_V6() to existing IS_V4() if locations") Signed-off-by:
-
Yunfei Dong authored
The vpu maybe null pointer or unreasonable value when scp crash, need to validate that the vpu pointer and the vpu instance within this context is valid in case of leading to kernel reboot. Fixes: 27a274db ("[media] vcodec: mediatek: Add Mediatek VP8 Video Encoder Driver") Signed-off-by:
Yunfei Dong <yunfei.dong@mediatek.com> Reported-by:
Steve Cho <stevecho@google.com> Signed-off-by:
-
Yunfei Dong authored
The vpu maybe null pointer or unreasonable value when scp crash, need to validate that the vpu pointer and the vpu instance within this context is valid in case of leading to kernel reboot. Fixes: 590577a4 ("[media] vcodec: mediatek: Add Mediatek V4L2 Video Decoder Driver") Signed-off-by:
Nicolas Dufresne <nicolas.dufresne@collabora.com> Signed-off-by:
-
Mingjia Zhang authored
Adding to support capture formats V4L2_PIX_FMT_MT2110T and V4L2_PIX_FMT_MT2110R for 10bit playback. Need to get the size of each plane again when user space setting syntax to get 10bit information. V4L2_PIX_FMT_MT2110T for AV1/VP9/HEVC. V4L2_PIX_FMT_MT2110R for H264. Signed-off-by:
Mingjia Zhang <mingjia.zhang@mediatek.com> Co-developed-by:
-
Mingjia Zhang authored
Define one uncompressed capture format V4L2_PIX_FMT_MT2110R in order to support 10bit for H264 in mt8195. Signed-off-by:
-
Mingjia Zhang authored
Define one uncompressed capture format V4L2_PIX_FMT_MT2110T in order to support 10bit for AV1/VP9/HEVC in mt8195. Signed-off-by:
-
Yunfei Dong authored
Move all decoder files to folder decoder. Move all encoder files to folder encoder. Move common files which shared for encoder and decoder to folder common. Change include header files and Makefile to fix build error. Signed-off-by:
AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by:
-
Yunfei Dong authored
remove unused include header for .c files Signed-off-by:
-
Yunfei Dong authored
Fix unreasonable coding style. Fix unreasonable parameter definition. Signed-off-by:
-
Yunfei Dong authored
Adding different dev struct for encoder and decoder, remove 'struct mtk_vcodec_dev'. Signed-off-by:
-
Yunfei Dong authored
Adding different context struct for encoder and decoder and removing struct 'mtk_vcodec_ctx'. Signed-off-by:
-
Yunfei Dong authored
Adding different macro mtk_v4l2_vdec_dbg and mtk_v4l2_venc_dbg for encoder and decoder. Then calling the common macro mtk_v4l2_debug to print debug message. Replace pr_err with dev_err for 'mtk_v4l2_err' debug message. Replace pr_debug with dev_dbg for 'mtk_v4l2_debug' debug message. Signed-off-by:
-
Yunfei Dong authored
'mtk_vcodec_debug' and 'mtk_vcodec_err' depends on 'mtk_vcodec_ctx' to get the index of each instance. Define two different macro mtk_vdec_debug and mtk_venc_debug for decoder and encoder, and re-write macro mtk_vcodec_debug as the common interface which is called by mtk_vdec_debug and mtk_venc_debug. The vcodec debug log can be separeated by encoder and decoder. Signed-off-by:
-
Yunfei Dong authored
Removing unresonable and useless debug log enter and leave in order to simply the log message. Signed-off-by:
-
Yunfei Dong authored
The shared struct 'mtk_vcodec_ctx/dev' will be changed to 'mtk_vcodec_enc_ctx/dev' and 'mtk_vcodec_dec_ctx/dev' in order to separate encoder and decoder. Removing common struct 'mtk_vcodec_ctx/dev' for shared interface which encoder and decoder used at the same time. Then encoder and decoder can call the same interface independently. Signed-off-by:
-
Yunfei Dong authored
Align scp and vpu firmware interface, remove the depedency for 'struct mtk_vcodec_dev' and 'struct mtk_vcodec_ctx'. It will be much easier to separate video encoder and decoder. Signed-off-by:
-
Yunfei Dong authored
remove unused parameter in struct mtk_vcodec_dev. Signed-off-by:
-
Sakari Ailus authored
Initialise timing struct in cio2_hw_init() to zero in order to avoid a compiler warning. The warning was a false positive. Reported-by:
Sakari Ailus <sakari.ailus@linux.intel.com> Signed-off-by:
-
Sakari Ailus authored
Initialise best_div local variable to avoid a compiler warning. The warning was harmless though. Reported-by:
-
Hans de Goede authored
The current error-checking of me->stages in sh_css_sp_init_pipeline() has some issues / weirdness: 1. It is checked at the top of the function, but only using the atomisp custom assert() macro which e.g. smatch does not recognize 2. It is first dereferenced in "first_binary = me->stages->binary", but outside of the assert it is checked much later, triggering the following smatch warning: drivers/staging/media/atomisp/pci/sh_css_sp.c:1255 sh_css_sp_init_pipeline() warn: variable dereferenced before check 'me->stages' (see line 1224) Drop the custom assert() calls (note 'me' is never NULL) and instead add a regular check for me->stages not being set. Reported-by:
Hans de Goede <hdegoede@redhat.com> Signed-off-by:
-
Hans de Goede authored
The atomisp code base has a custom assert() macro, a couple of functions use this in a construction like the following: assert(pipe); assert(pipe->stream); if ((!pipe) || (!pipe->stream)) { ia_css_debug_dtrace(IA_CSS_DEBUG_TRACE_PRIVATE, "allocate_mipi_frames(%p) exit: ...\n", pipe); return -EINVAL; } The second assert is seen by smatch as dereferencing "pipe" in the above example (and dereferencing "dvs_6axis_config" in the other case). Following by the dereferenced variable being checked (a second time) in the following if () statement. This triggers the following smatch warnings: drivers/staging/media/atomisp/pci/sh_css_mipi.c:356 allocate_mipi_frames() warn: variable dereferenced before check 'pipe' (see line 355) drivers/staging/media/atomisp/pci/sh_css_mipi.c:562 send_mipi_frames() warn: variable dereferenced before check 'pipe' (see line 561) drivers/staging/media/atomisp/pci/sh_css_param_dvs.c:208 free_dvs_6axis_table() warn: variable dereferenced before check 'dvs_6axis_config' (see line 206) The custom assert() macro actually expands to a BUG() call and BUG() calls should not be used in the kernel. Remove the assert() calls to fix the smatch warnings and in case of [allocate|send]_mipi_frames() also remove the if () return -EINVAL block since these functions are never called with a NULL pipe. Reported-by: -
Hans de Goede authored
The asd is a sub-structure of the main driver data struct, so it is never NULL. Drop the unnecessary NULL checks in a couple of places. Reported-by:
Dan Carpenter <dan.carpenter@linaro.org> Closes: https://lore.kernel.org/linux-media/533f6930-434a-45f3-afff-127003fa64c9@moroto.mountain/Signed-off-by:
-
Colin Ian King authored
The if statement that compares msgs[i].len != 3 is always false because it is in a code block where msg[i].len is equal to 3. The check is redundant and can be removed. As detected by cppcheck static analysis: drivers/media/usb/go7007/go7007-i2c.c:168:20: warning: Opposite inner 'if' condition leads to a dead code block. [oppositeInnerCondition] Link: https://lore.kernel.org/linux-media/20230727174007.635572-1-colin.i.king@gmail.com Fixes: 866b8695 ("Staging: add the go7007 video driver") Signed-off-by:
Colin Ian King <colin.i.king@gmail.com> Signed-off-by:
Mauro Carvalho Chehab <mchehab@kernel.org>
-
Arnd BergmannArnd Bergmann authored
This code was previously part of the VIDEO_IPU3_CIO2 driver, which could be built-in or a loadable module, but after the move it turned into a builtin-only driver. This fails to link when the I2C subsystem is a module: x86_64-linux-ld: drivers/media/pci/intel/ipu-bridge.o: in function `ipu_bridge_unregister_sensors': ipu-bridge.c:(.text+0x50): undefined reference to `i2c_unregister_device' x86_64-linux-ld: drivers/media/pci/intel/ipu-bridge.o: in function `ipu_bridge_init': ipu-bridge.c:(.text+0x9c9): undefined reference to `i2c_acpi_new_device_by_fwnode' In general, drivers should not have to be built-in, so change the option to a tristate with the corresponding dependency. This in turn opens a new problem with the dependency, as the IPU bridge can be a loadable module while the ipu3 driver itself is built-in, producing a new link failure: 86_64-linux-ld: drivers/media/pci/intel/ipu3/ipu3-cio2.o: in function `cio2_pci_probe': ipu3-cio2.c:(.text+0x197e): undefined reference to `ipu_bridge_init' In order to fix this, restore the old Kconfig option that controlled the ipu bridge driver before it was split out, but make it select a hidden symbol that now corresponds to the bridge driver. When other drivers get added that share ipu-bridge, this should cover all corner cases, and allow any combination of them to be built-in or modular. Link: https://lore.kernel.org/linux-media/20230727122331.2421453-1-arnd@kernel.org Fixes: 881ca259 ("media: ipu3-cio2: rename cio2 bridge to ipu bridge and move out of ipu3")' Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Reviewed-by:
Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by:
-
Hans Verkuil authored
The CEC interrupt is only needed if userspace wants to monitor the CEC pin for an unconfigured CEC device. That gives it the most precise CEC pin debugging results. This avoids a corner case where the interrupt is enabled for a short period when the adapter is about to be configured. Signed-off-by:
-
Hans Verkuil authored
Since the CEC pin framework now keeps track of the interrupt and calls disable_irq when the kthread stops, there is no longer any need for the cec-gpio driver to do this in the free callback. So drop this code. Signed-off-by:
-
Hans Verkuil authored
The CEC pin framework needs a bit more control over the interrupt handling: make sure that the disable_irq op is called even if the device node is unregistered, log the state of the interrupt in debugfs, and disable the interrupt when the kernel thread is stopped. Signed-off-by:
-
Hans Verkuil authored
Use IRQF_NO_AUTOEN rather than manually disabling the requested interrupt. Signed-off-by:
-
Hans Verkuil authored
Describe the new callbacks and clarify when the adap->lock mutex is held or not. Signed-off-by:
-
Hans Verkuil authored
The adap_configured() callback was called with the adap->lock mutex held if the 'configured' argument was false, and without the adap->lock mutex held if that argument was true. That was very confusing, and so split this up in a adap_unconfigured() callback and a high-level configured() callback. This also makes it easier to understand when the mutex is held: all low-level adap_* callbacks are called with the mutex held. All other callbacks are called without that mutex held. Signed-off-by:
-
Hans Verkuil authored
A potential deadlock was found by Zheng Zhang with a local syzkaller instance. The problem is that when a non-blocking CEC transmit is canceled by calling cec_data_cancel, that in turn can call the high-level received() driver callback, which can call cec_transmit_msg() to transmit a new message. The cec_data_cancel() function is called with the adap->lock mutex held, and cec_transmit_msg() tries to take that same lock. The root cause is that the received() callback can either be used to pass on a received message (and then adap->lock is not held), or to report a canceled transmit (and then adap->lock is held). This is confusing, so create a new low-level adap_nb_transmit_canceled callback that reports back that a non-blocking transmit was canceled. And the received() callback is only called when a message is received, as was the case before commit f9d0ecbf ("media: cec: correctly pass on reply results") complicated matters. Reported-by:
Zheng Zhang <zheng.zhang@email.ucr.edu> Signed-off-by:
-
Sakari Ailus authored
Document that sub-device notifiers are now registered using v4l2_async_subdev_nf_init(). No documentation is changed as it seems that sub-device notifiers were not documented apart from kernel-doc comments. Signed-off-by:
-
Sakari Ailus authored
Set the v4l2_device already in async notifier init, so struct device related to it will be available before the notifier is registered. This requires separating notifier initialisation into two functions, one that takes v4l2_device as its argument, v4l2_async_nf_init and v4l2_async_subdev_nf_init, for sub-device notifiers. Registering the notifier will use a single function, v4l2_async_nf_register. This is done in order to make struct device available earlier, during construction of the async connections, for sensible debug prints. Signed-off-by:
-
Sakari Ailus authored
Initialise V4L2 async notifier and parse DT for async sub-devices later, just before registering the notifier. This way the device can be made available to the V4L2 async framework from the notifier init time onwards. A subsequent patch will add struct v4l2_device as an argument to v4l2_async_nf_init(). Signed-off-by:
-
Sakari Ailus authored
Initialise the V4L2 async notifier after registering the V4L2 device, just before parsing DT for async sub-devices. This way the device can be made available to the V4L2 async framework from the notifier init time onwards. A subsequent patch will add struct v4l2_device as an argument to v4l2_async_nf_init(). Signed-off-by:
-
Sakari Ailus authored
Initialise the V4L2 async notifier after registering the V4L2 device, just before parsing DT for async sub-devices. This way the device can be made available to the V4L2 async framework from the notifier init time onwards. A subsequent patch will add struct v4l2_device as an argument to v4l2_async_nf_init(). Signed-off-by:
Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Philipp Zabel <p.zabel@pengutronix.de> # imx6qp Tested-by: Niklas Söderlund <niklas.soderlund@ragnatech.se> # rcar + adv746x Tested-by: Aishwarya Kothari <aishwarya.kothari@toradex.com> # Apalis i.MX6Q with TC358743 Tested-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com> # Renesas RZ/G2L SMARC Signed-off-by:
-
Sakari Ailus authored
Initialise V4L2 async notifier and parse DT for async sub-devices later, just before registering the notifier. This way the device can be made available to the V4L2 async framework from the notifier init time onwards. A subsequent patch will add struct v4l2_device as an argument to v4l2_async_nf_init(). Signed-off-by:
-
