kernel/drivers/media/platform/qcom/venus, branch linux-rolling-stable

media: venus: vdec: restrict EOS addr quirk to IRIS2 only

2026-03-04T12:21:17Z

[ Upstream commit 63c072e2937e6c9995df1b6a28523ed2ae68d364 ] On SM8250 (IRIS2) with firmware older than 1.0.087, the firmware could not handle a dummy device address for EOS buffers, so a NULL device address is sent instead. The existing check used IS_V6() alongside a firmware version gate: if (IS_V6(core) && is_fw_rev_or_older(core, 1, 0, 87)) fdata.device_addr = 0; else fdata.device_addr = 0xdeadb000; However, SC7280 which is also V6, uses a firmware string of the form "1.0.", which the version parser translates to 1.0.0. This unintentionally satisfies the `is_fw_rev_or_older(..., 1, 0, 87)` condition on SC7280. Combined with IS_V6() matching there as well, the quirk is incorrectly applied to SC7280, causing VP9 decode failures. Constrain the check to IRIS2 (SM8250) only, which is the only platform that needed this quirk, by replacing IS_V6() with IS_IRIS2(). This restores correct behavior on SC7280 (no forced NULL EOS buffer address). Fixes: 47f867cb1b63 ("media: venus: fix EOS handling in decoder stop command") Cc: stable@vger.kernel.org Reported-by: Mecid Closes: https://github.com/qualcomm-linux/kernel-topics/issues/222 Co-developed-by: Renjiang Han Signed-off-by: Renjiang Han Signed-off-by: Dikshita Agarwal Tested-by: Renjiang Han Signed-off-by: Bryan O'Donoghue Signed-off-by: Hans Verkuil Signed-off-by: Sasha Levin

media: venus: vdec: fix error state assignment for zero bytesused

2026-03-04T12:21:17Z

[ Upstream commit 93ecd6ee95c38cb533fa25f48d3c1c8cb69f410f ] When hfi_session_flush is issued, all queued buffers are returned to the V4L2 driver. Some of these buffers are not processed and have bytesused = 0. Currently, the driver marks such buffers as error even during drain operations, which can incorrectly flag EOS buffers. Only capture buffers with zero payload (and not EOS) should be marked with VB2_BUF_STATE_ERROR. The check is performed inside the non-EOS branch to ensure correct handling. Fixes: 51df3c81ba10b ("media: venus: vdec: Mark flushed buffers with error state") Signed-off-by: Renjiang Han Reviewed-by: Vikash Garodia Cc: stable@vger.kernel.org Signed-off-by: Bryan O'Donoghue Signed-off-by: Hans Verkuil Signed-off-by: Sasha Levin

media: venus: drop unused module aliases

2025-11-06T10:18:09Z

The driver has never supported anything but OF probing so drop the unused platform module aliases. Signed-off-by: Johan Hovold Reviewed-by: Dikshita Agarwal Signed-off-by: Bryan O'Donoghue Signed-off-by: Hans Verkuil

media: venus: drop bogus probe deferrals

2025-11-06T10:18:09Z

The encoder and decoder platform devices are registered by the venus driver as children of the venus device, but even if someone were to mess this up no amount of probe deferring is going to conjure up a parent. Relatedly, the venus driver sets its driver data before registering the child devices and if this ever breaks we want to learn about it by failing probe. Signed-off-by: Johan Hovold Reviewed-by: Vikash Garodia Reviewed-by: Dikshita Agarwal Reviewed-by: Bryan O'Donoghue Signed-off-by: Bryan O'Donoghue Signed-off-by: Hans Verkuil

media: Use of_reserved_mem_region_to_resource() for "memory-region"

2025-11-06T10:18:09Z

Use the newly added of_reserved_mem_region_to_resource() function to handle "memory-region" properties. Reviewed-by: Dikshita Agarwal Reviewed-by: Ming Qian Signed-off-by: Rob Herring (Arm) Reviewed-by: Bryan O'Donoghue Signed-off-by: Bryan O'Donoghue Signed-off-by: Hans Verkuil

media: qcom: venus: Drop unneeded v4l2_m2m_get_vq() NULL check

2025-10-17T09:31:15Z

The v4l2_m2m_get_vq() function never returns NULL. The check may have been intended to catch invalid format types, but that's not needed as the V4L2 core picks the appropriate VIDIOC_S_FMT ioctl handler based on the format type, so the type can't be incorrect. Drop the unneeded return value check. Signed-off-by: Laurent Pinchart Reviewed-by: Dikshita Agarwal Reviewed-by: Bryan O'Donoghue Signed-off-by: Hans Verkuil

Merge tag 'media/v6.18-1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media

2025-10-02T20:13:26Z

Pull media updates from Mauro Carvalho Chehab: - Added a new V4L2 clock helper - New camera sensor drivers - iris: Enable H.264/H.265 encoder support and fixes in iris driver common code - camss: add support for new SoC flavors - venus: add new SoC support - tc358743: support more infoframe types - Various fixes, driver improvements and cleanups * tag 'media/v6.18-1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media: (439 commits) media: venus: pm_helpers: add fallback for the opp-table media: qcom: camss: vfe: Fix BPL alignment for QCM2290 media: tuner: xc5000: Fix use-after-free in xc5000_release media: i2c: tc358743: Fix use-after-free bugs caused by orphan timer in probe media: b2c2: Fix use-after-free causing by irq_check_work in flexcop_pci_remove media: vsp1: Export missing vsp1_isp_free_buffer symbol media: renesas: vsp1: Convert to SYSTEM_SLEEP/RUNTIME_PM_OPS() media: renesas: ceu: Convert to RUNTIME_PM_OPS() media: renesas: fdp1: Convert to RUNTIME_PM_OPS() media: renesas: rcar-vin: Convert to DEFINE_SIMPLE_DEV_PM_OPS() media: renesas: rcar_drif: Convert to DEFINE_SIMPLE_DEV_PM_OPS() media: uvcvideo: Mark invalid entities with id UVC_INVALID_ENTITY_ID media: uvcvideo: Support UVC_CROSXU_CONTROL_IQ_PROFILE media: uvcvideo: Run uvc_ctrl_init_ctrl for all controls media: uvcvideo: Shorten the transfer size non compliance message media: uvcvideo: Do not re-reference dev->udev media: uvcvideo: Use intf instead of udev for printks media: uvcvideo: Move video_device under video_queue media: uvcvideo: Drop stream->mutex media: uvcvideo: Move MSXU_CONTROL_METADATA definition to header ...

media: venus: pm_helpers: add fallback for the opp-table

2025-09-24T10:52:09Z

Since the device trees for both HFI_VERSION_1XX and HFI_VERSION_3XX do not include an opp-table and have not configured opp-pmdomain, they still need to use the frequencies defined in the driver's freq_tbl. Both core_power_v1 and core_power_v4 functions require core_clks_enable function during POWER_ON. Therefore, in the core_clks_enable function, if calling dev_pm_opp_find_freq_ceil to obtain the frequency fails, it needs to fall back to the freq_tbl to retrieve the frequency. Fixes: b179234b5e59 ("media: venus: pm_helpers: use opp-table for the frequency") Cc: stable@vger.kernel.org Reviewed-by: Dmitry Baryshkov Reviewed-by: Bryan O'Donoghue Reviewed-by: Vikash Garodia Closes: https://lore.kernel.org/linux-media/CA+G9fYu5=3n84VY+vTbCAcfFKOq7Us5vgBZgpypY4MveM=eVwg@mail.gmail.com Signed-off-by: Renjiang Han Signed-off-by: Bryan O'Donoghue Signed-off-by: Hans Verkuil

media: venus: core: Add qcm2290 DT compatible and resource data

2025-09-09T13:59:16Z

Add a qcm2290 compatible binding to the Venus core. The maximum concurrency is video decode at 1920x1080 (FullHD) with video encode at 1280x720 (HD). The driver is not available to firmware versions below 6.0.55 due to an internal requirement for secure buffers. The bandwidth tables incorporate a conservative safety margin to ensure stability under peak DDR and interconnect load conditions. Co-developed-by: Loic Poulain Signed-off-by: Loic Poulain Signed-off-by: Jorge Ramirez-Ortiz Reviewed-by: Bryan O'Donoghue Reviewed-by: Dikshita Agarwal Signed-off-by: Bryan O'Donoghue Signed-off-by: Hans Verkuil

media: venus: core: Sort dt_match alphabetically.

2025-09-09T13:59:16Z

Correctly sort the array of venus_dt_match entries. Signed-off-by: Jorge Ramirez-Ortiz Reviewed-by: Bryan O'Donoghue Reviewed-by: Dikshita Agarwal Signed-off-by: Bryan O'Donoghue Signed-off-by: Hans Verkuil