linux-toradex.git/drivers/media/platform/qcom/iris, branch master Linux kernel for Apalis and Colibri modules media: qcom: iris: increase H265D_MAX_SLICE to fix H.265 decoding on SC7280 2026-04-27T06:41:21+00:00 Dmitry Baryshkov dmitry.baryshkov@oss.qualcomm.com 2026-03-27T20:19:55+00:00 3e0b2053751657ed2924adfe3ff25b1450231e33 Follow the commit bfe1326573ff ("venus: Fix for H265 decoding failure.") and increase H265D_MAX_SLICE following firmware requirements on that platform. Otherwise decoding of the H.265 streams fails with the "insufficient scratch_1 buffer size" from the firmware. Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com> Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com> Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com> Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com> [bod: Fixed commit log withthe => with the] Fixes: e1f5d32608ec ("media: iris: Add internal buffer calculation for HEVC and VP9 decoders") Cc: stable@vger.kernel.org Signed-off-by: Bryan O'Donoghue <bod@kernel.org> Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org> 
Follow the commit bfe1326573ff ("venus: Fix for H265 decoding failure.")
and increase H265D_MAX_SLICE following firmware requirements on that
platform. Otherwise decoding of the H.265 streams fails with the
"insufficient scratch_1 buffer size" from the firmware.

Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
[bod: Fixed commit log withthe => with the]
Fixes: e1f5d32608ec ("media: iris: Add internal buffer calculation for HEVC and VP9 decoders")
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
media: iris: fix use-after-free of fmt_src during MBPF check 2026-04-27T06:41:21+00:00 Vishnu Reddy busanna.reddy@oss.qualcomm.com 2026-03-05T13:28:31+00:00 3d9593ad1a58c5acc3e5fa2a48222bb7632e6812 During concurrency testing, multiple instances can run in parallel, and each instance uses its own inst->lock while the core->lock protects the list of active instances. The race happens because these locks cover different scopes, inst->lock protects only the internals of a single instance, while the Macro Blocks Per Frame (MBPF) checker walks the core list under core->lock and reads fields like fmt_src->width and fmt_src->height. At the same time, iris_close() may free fmt_src and fmt_dst under inst->lock while the instance is still present in the core list. This allows a situation where the MBPF checker, still iterating through the core list, reaches an instance whose fmt_src was already freed by another thread and ends up dereferencing a dangling pointer, resulting in a use-after-free. This happens because the MBPF checker assumes that any instance in the core list is fully valid, but the freeing of fmt_src and fmt_dst without removing the instance from the core list is not correct. The correct ordering is to defer freeing fmt_src and fmt_dst until after the instance has been removed from the core list and all teardown under the core lock has completed, ensuring that no dangling pointers are ever exposed during MBPF checks. Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com> Signed-off-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com> Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com> Fixes: 5ad964ad5656 ("media: iris: Initialize and deinitialize encoder instance structure") Cc: stable@vger.kernel.org Signed-off-by: Bryan O'Donoghue <bod@kernel.org> Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org> 
During concurrency testing, multiple instances can run in parallel, and
each instance uses its own inst->lock while the core->lock protects the
list of active instances. The race happens because these locks cover
different scopes, inst->lock protects only the internals of a single
instance, while the Macro Blocks Per Frame (MBPF) checker walks the
core list under core->lock and reads fields like fmt_src->width and
fmt_src->height. At the same time, iris_close() may free fmt_src and
fmt_dst under inst->lock while the instance is still present in the core
list. This allows a situation where the MBPF checker, still iterating
through the core list, reaches an instance whose fmt_src was already
freed by another thread and ends up dereferencing a dangling pointer,
resulting in a use-after-free. This happens because the MBPF checker
assumes that any instance in the core list is fully valid, but the
freeing of fmt_src and fmt_dst without removing the instance from the
core list is not correct.

The correct ordering is to defer freeing fmt_src and fmt_dst until after
the instance has been removed from the core list and all teardown under
the core lock has completed, ensuring that no dangling pointers are ever
exposed during MBPF checks.

Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Signed-off-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com>
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Fixes: 5ad964ad5656 ("media: iris: Initialize and deinitialize encoder instance structure")
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
media: iris: switch to hardware mode after firmware boot 2026-04-27T06:41:21+00:00 Vikash Garodia vikash.garodia@oss.qualcomm.com 2026-03-13T13:19:36+00:00 95a337f92f0a602d4f935315bfbc8bf07f475e65 Currently the driver switches the vcodec GDSC to hardware (HW) mode before firmware load and boot sequence. GDSC can be powered off, keeping in hw mode, thereby the vcodec registers programmed in TrustZone (TZ) carry default (reset) values. Move the transition to HW mode after firmware load and boot sequence. The bug was exposed with driver configuring different stream ids to different devices via iommu-map. With registers carrying reset values, VPU would not generate desired stream-id, thereby leading to SMMU fault. For vpu4, when GDSC is switched to HW mode, there is a need to perform the reset operation. Without reset, there are occasional issues of register corruption observed. Hence the vpu GDSC switch also involves the reset. Co-developed-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com> Signed-off-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com> Signed-off-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com> Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com> Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com> [bod: occassional => occasional] Fixes: dde659d37036 ("media: iris: Introduce vpu ops for vpu4 with necessary hooks") Cc: stable@vger.kernel.org Signed-off-by: Bryan O'Donoghue <bod@kernel.org> Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org> 
Currently the driver switches the vcodec GDSC to hardware (HW) mode
before firmware load and boot sequence. GDSC can be powered off, keeping
in hw mode, thereby the vcodec registers programmed in TrustZone (TZ)
carry default (reset) values.
Move the transition to HW mode after firmware load and boot sequence.

The bug was exposed with driver configuring different stream ids to
different devices via iommu-map. With registers carrying reset values,
VPU would not generate desired stream-id, thereby leading to SMMU fault.

For vpu4, when GDSC is switched to HW mode, there is a need to perform
the reset operation. Without reset, there are occasional issues of
register corruption observed. Hence the vpu GDSC switch also involves
the reset.

Co-developed-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com>
Signed-off-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com>
Signed-off-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
[bod: occassional => occasional]
Fixes: dde659d37036 ("media: iris: Introduce vpu ops for vpu4 with necessary hooks")
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
media: iris: Fix dma_free_attrs() size in iris_hfi_queues_init() 2026-04-27T06:41:21+00:00 Thomas Fourier fourier.thomas@gmail.com 2026-02-13T09:13:27+00:00 4a49ae56b0e4268d48fd96babe0cc68596bc301a The core->iface_q_table_vaddr buffer is alloc'd with size queue_size but freed with sizeof(*q_tbl_hdr) which is different. Change the dma_free_attrs() size. Signed-off-by: Thomas Fourier <fourier.thomas@gmail.com> Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com> Fixes: d7378f84e94e ("media: iris: introduce iris core state management with shared queues") Cc: stable@vger.kernel.org Signed-off-by: Bryan O'Donoghue <bod@kernel.org> Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org> 
The core->iface_q_table_vaddr buffer is alloc'd with size queue_size
but freed with sizeof(*q_tbl_hdr) which is different.

Change the dma_free_attrs() size.

Signed-off-by: Thomas Fourier <fourier.thomas@gmail.com>
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Fixes: d7378f84e94e ("media: iris: introduce iris core state management with shared queues")
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
media: iris: Fix use-after-free in iris_release_internal_buffers() 2026-04-27T06:41:21+00:00 Dikshita Agarwal dikshita.agarwal@oss.qualcomm.com 2026-02-16T07:07:42+00:00 f27cfdcfc916bb59297825805f4c3499f89f9e76 The recent change in commit 1dabf00ee206 ("media: iris: gen1: Destroy internal buffers after FW releases") introduced a regression where session_release_buf() may free the buffer. The caller, iris_release_internal_buffers(), continued to access `buffer` after the call, leading to a potential use-after-free. Fix this by setting BUF_ATTR_PENDING_RELEASE before calling session_release_buf(), and reverting the flag if the call fails. This ensures no dereference occurs after potential freeing. Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Closes: https://lore.kernel.org/lkml/aYXvKAX3Pg3sL37P@stanley.mountain/#r Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com> Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com> Fixes: 1dabf00ee206 ("media: iris: gen1: Destroy internal buffers after FW releases") Cc: stable@vger.kernel.org Signed-off-by: Bryan O'Donoghue <bod@kernel.org> Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org> 
The recent change in commit 1dabf00ee206 ("media: iris: gen1: Destroy
internal buffers after FW releases") introduced a regression where
session_release_buf() may free the buffer. The caller,
iris_release_internal_buffers(), continued to access `buffer` after the
call, leading to a potential use-after-free.

Fix this by setting BUF_ATTR_PENDING_RELEASE before calling
session_release_buf(), and reverting the flag if the call fails. This
ensures no dereference occurs after potential freeing.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Closes: https://lore.kernel.org/lkml/aYXvKAX3Pg3sL37P@stanley.mountain/#r
Signed-off-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Fixes: 1dabf00ee206 ("media: iris: gen1: Destroy internal buffers after FW releases")
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
media: iris: fix QCOM_MDT_LOADER dependency 2026-04-27T06:41:21+00:00 Arnd Bergmann arnd@arndb.de 2026-02-05T14:56:19+00:00 a297c5165f91366cbc3490e630aabd1c0f70efb8 When build-testined with CONFIG_QCOM_MDT_LOADER=m and VIDEO_QCOM_IRIS=y, the kernel fails to link: x86_64-linux-ld: drivers/media/platform/qcom/iris/iris_firmware.o: in function `iris_fw_load': iris_firmware.c:(.text+0xb0): undefined reference to `qcom_mdt_get_size' iris_firmware.c:(.text+0xfd): undefined reference to `qcom_mdt_load' The problem is the conditional 'select' statement. Change this to make the driver built-in here regardless of CONFIG_ARCH_QCOM. Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com> Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com> Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> Fixes: d19b163356b8 ("media: iris: implement video firmware load/unload") Cc: stable@vger.kernel.org Signed-off-by: Bryan O'Donoghue <bod@kernel.org> Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org> 
When build-testined with CONFIG_QCOM_MDT_LOADER=m and VIDEO_QCOM_IRIS=y,
the kernel fails to link:

x86_64-linux-ld: drivers/media/platform/qcom/iris/iris_firmware.o: in function `iris_fw_load':
iris_firmware.c:(.text+0xb0): undefined reference to `qcom_mdt_get_size'
iris_firmware.c:(.text+0xfd): undefined reference to `qcom_mdt_load'

The problem is the conditional 'select' statement. Change this to
make the driver built-in here regardless of CONFIG_ARCH_QCOM.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Fixes: d19b163356b8 ("media: iris: implement video firmware load/unload")
Cc: stable@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-22T01:09:51+00:00 Linus Torvalds torvalds@linux-foundation.org 2026-02-22T00:37:42+00:00 bf4afc53b77aeaa48b5409da5c8da6bb4eff7f43 This was done entirely with mindless brute force, using git grep -l '\<k[vmz]*alloc_objs*(.*, GFP_KERNEL)' | xargs sed -i 's/\(alloc_objs*(.*\), GFP_KERNEL)/\1)/' to convert the new alloc_obj() users that had a simple GFP_KERNEL argument to just drop that argument. Note that due to the extreme simplicity of the scripting, any slightly more complex cases spread over multiple lines would not be triggered: they definitely exist, but this covers the vast bulk of the cases, and the resulting diff is also then easier to check automatically. For the same reason the 'flex' versions will be done as a separate conversion. Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> 
This was done entirely with mindless brute force, using

    git grep -l '\<k[vmz]*alloc_objs*(.*, GFP_KERNEL)' |
        xargs sed -i 's/\(alloc_objs*(.*\), GFP_KERNEL)/\1)/'

to convert the new alloc_obj() users that had a simple GFP_KERNEL
argument to just drop that argument.

Note that due to the extreme simplicity of the scripting, any slightly
more complex cases spread over multiple lines would not be triggered:
they definitely exist, but this covers the vast bulk of the cases, and
the resulting diff is also then easier to check automatically.

For the same reason the 'flex' versions will be done as a separate
conversion.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21T09:02:28+00:00 Kees Cook kees@kernel.org 2026-02-21T07:49:23+00:00 69050f8d6d075dc01af7a5f2f550a8067510366f This is the result of running the Coccinelle script from scripts/coccinelle/api/kmalloc_objs.cocci. The script is designed to avoid scalar types (which need careful case-by-case checking), and instead replace kmalloc-family calls that allocate struct or union object instances: Single allocations: kmalloc(sizeof(TYPE), ...) are replaced with: kmalloc_obj(TYPE, ...) Array allocations: kmalloc_array(COUNT, sizeof(TYPE), ...) are replaced with: kmalloc_objs(TYPE, COUNT, ...) Flex array allocations: kmalloc(struct_size(PTR, FAM, COUNT), ...) are replaced with: kmalloc_flex(*PTR, FAM, COUNT, ...) (where TYPE may also be *VAR) The resulting allocations no longer return "void *", instead returning "TYPE *". Signed-off-by: Kees Cook <kees@kernel.org> 
This is the result of running the Coccinelle script from
scripts/coccinelle/api/kmalloc_objs.cocci. The script is designed to
avoid scalar types (which need careful case-by-case checking), and
instead replace kmalloc-family calls that allocate struct or union
object instances:

Single allocations:	kmalloc(sizeof(TYPE), ...)
are replaced with:	kmalloc_obj(TYPE, ...)

Array allocations:	kmalloc_array(COUNT, sizeof(TYPE), ...)
are replaced with:	kmalloc_objs(TYPE, COUNT, ...)

Flex array allocations:	kmalloc(struct_size(PTR, FAM, COUNT), ...)
are replaced with:	kmalloc_flex(*PTR, FAM, COUNT, ...)

(where TYPE may also be *VAR)

The resulting allocations no longer return "void *", instead returning
"TYPE *".

Signed-off-by: Kees Cook <kees@kernel.org>
media: iris: Introduce vpu ops for vpu4 with necessary hooks 2026-01-20T15:22:24+00:00 Vikash Garodia vikash.garodia@oss.qualcomm.com 2025-12-10T12:36:04+00:00 dde659d37036c6edf5eeef81274b43fe12dfad53 Add power sequence for vpu4 by reusing from previous generation wherever possible. Hook up vpu4 op with vpu4 specific implemtation or resue from earlier generation wherever feasible, like clock calculation in this case. Co-developed-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com> Signed-off-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com> Signed-off-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com> Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com> Signed-off-by: Bryan O'Donoghue <bod@kernel.org> Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org> 
Add power sequence for vpu4 by reusing from previous generation wherever
possible. Hook up vpu4 op with vpu4 specific implemtation or resue from
earlier generation wherever feasible, like clock calculation in this
case.

Co-developed-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com>
Signed-off-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com>
Signed-off-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>
media: iris: Move vpu35 specific api to common to use for vpu4 2026-01-20T15:22:24+00:00 Vikash Garodia vikash.garodia@oss.qualcomm.com 2025-12-10T12:36:03+00:00 82b96a8d8d6ebedcda8ef2d051ddd0f43d8eee14 Some of the sequence and calculations for vpu4 is identical to vpu35, namely power sequence for vpu controller and the clock frequency calculation. Move those to common file that can be shared for both vpu35 and vpu4. This patch prepares for power sequence for vpu4 which is added in subsequent patch. Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> Co-developed-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com> Signed-off-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com> Signed-off-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com> Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com> Signed-off-by: Bryan O'Donoghue <bod@kernel.org> Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org> 
Some of the sequence and calculations for vpu4 is identical to vpu35,
namely power sequence for vpu controller and the clock frequency
calculation. Move those to common file that can be shared for both vpu35
and vpu4. This patch prepares for power sequence for vpu4 which is added
in subsequent patch.

Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Co-developed-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com>
Signed-off-by: Vishnu Reddy <busanna.reddy@oss.qualcomm.com>
Signed-off-by: Vikash Garodia <vikash.garodia@oss.qualcomm.com>
Reviewed-by: Dikshita Agarwal <dikshita.agarwal@oss.qualcomm.com>
Signed-off-by: Bryan O'Donoghue <bod@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil+cisco@kernel.org>