| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <dc9b20aecc8740896b2b3e7352b8e0d73d43fed2.camel@collabora.com>
Date: Tue, 29 Apr 2025 08:47:22 -0400
From: Nicolas Dufresne <nicolas.dufresne@...labora.com>
To: Dikshita Agarwal <quic_dikshita@...cinc.com>, Vikash Garodia
<quic_vgarodia@...cinc.com>, Abhinav Kumar <quic_abhinavk@...cinc.com>,
Mauro Carvalho Chehab <mchehab@...nel.org>, Stefan Schmidt
<stefan.schmidt@...aro.org>, Hans Verkuil <hverkuil@...all.nl>, Bjorn
Andersson <andersson@...nel.org>, Konrad Dybcio <konradybcio@...nel.org>,
Rob Herring <robh@...nel.org>, Krzysztof Kozlowski <krzk+dt@...nel.org>,
Conor Dooley <conor+dt@...nel.org>
Cc: Bryan O'Donoghue <bryan.odonoghue@...aro.org>, Dmitry Baryshkov
<dmitry.baryshkov@....qualcomm.com>, Neil Armstrong
<neil.armstrong@...aro.org>, linux-media@...r.kernel.org,
linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org,
devicetree@...r.kernel.org,
20250417-topic-sm8x50-iris-v10-v7-0-f020cb1d0e98@...aro.org,
20250424-qcs8300_iris-v5-0-f118f505c300@...cinc.com, stable@...r.kernel.org
Subject: Re: [PATCH v2 01/23] media: iris: Skip destroying internal buffer
if not dequeued
Not mine to review, but wanted to highlight some best practices,
comment below...
Le lundi 28 avril 2025 à 14:58 +0530, Dikshita Agarwal a écrit :
> Firmware might hold the DPB buffers for reference in case of sequence
> change, so skip destroying buffers for which QUEUED flag is not removed.
> Also, make sure that all buffers are released during streamoff.
>
> Cc: stable@...r.kernel.org
> Fixes: 73702f45db81 ("media: iris: allocate, initialize and queue internal buffers")
> Signed-off-by: Dikshita Agarwal <quic_dikshita@...cinc.com>
> ---
> drivers/media/platform/qcom/iris/iris_buffer.c | 37 +++++++++++++++++++++++++-
> drivers/media/platform/qcom/iris/iris_buffer.h | 3 ++-
> drivers/media/platform/qcom/iris/iris_vdec.c | 4 +--
> drivers/media/platform/qcom/iris/iris_vidc.c | 6 +++--
> 4 files changed, 44 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/media/platform/qcom/iris/iris_buffer.c b/drivers/media/platform/qcom/iris/iris_buffer.c
> index e5c5a564fcb8..606d76b10be2 100644
> --- a/drivers/media/platform/qcom/iris/iris_buffer.c
> +++ b/drivers/media/platform/qcom/iris/iris_buffer.c
> @@ -376,7 +376,7 @@ int iris_destroy_internal_buffer(struct iris_inst *inst, struct iris_buffer *buf
> return 0;
> }
>
> -int iris_destroy_internal_buffers(struct iris_inst *inst, u32 plane)
> +int iris_destroy_internal_buffers(struct iris_inst *inst, u32 plane, bool force)
Its always tempting to just glue a boolean at the end of a parameter
list. But this has huge downside in code readability, see below...
> {
> const struct iris_platform_data *platform_data = inst->core->iris_platform_data;
> struct iris_buffer *buf, *next;
> @@ -396,6 +396,14 @@ int iris_destroy_internal_buffers(struct iris_inst *inst, u32 plane)
> for (i = 0; i < len; i++) {
> buffers = &inst->buffers[internal_buf_type[i]];
> list_for_each_entry_safe(buf, next, &buffers->list, list) {
> + /*
> + * during stream on, skip destroying internal(DPB) buffer
> + * if firmware did not return it.
> + * during close, destroy all buffers irrespectively.
> + */
> + if (!force && buf->attr & BUF_ATTR_QUEUED)
> + continue;
> +
> ret = iris_destroy_internal_buffer(inst, buf);
> if (ret)
> return ret;
> @@ -446,6 +454,33 @@ static int iris_release_input_internal_buffers(struct iris_inst *inst)
> return 0;
> }
>
> +void iris_get_num_queued_internal_buffers(struct iris_inst *inst, u32 plane)
> +{
> + const struct iris_platform_data *platform_data = inst->core->iris_platform_data;
> + struct iris_buffer *buf, *next;
> + struct iris_buffers *buffers;
> + const u32 *internal_buf_type;
> + u32 internal_buffer_count, i;
> + u32 count = 0;
> +
> + if (V4L2_TYPE_IS_OUTPUT(plane)) {
> + internal_buf_type = platform_data->dec_ip_int_buf_tbl;
> + internal_buffer_count = platform_data->dec_ip_int_buf_tbl_size;
> + } else {
> + internal_buf_type = platform_data->dec_op_int_buf_tbl;
> + internal_buffer_count = platform_data->dec_op_int_buf_tbl_size;
> + }
> +
> + for (i = 0; i < internal_buffer_count; i++) {
> + buffers = &inst->buffers[internal_buf_type[i]];
> + list_for_each_entry_safe(buf, next, &buffers->list, list)
> + count++;
> + if (count)
> + dev_err(inst->core->dev, "%d buffer of type %d not released",
> + count, internal_buf_type[i]);
> + }
> +}
> +
> int iris_alloc_and_queue_persist_bufs(struct iris_inst *inst)
> {
> struct iris_buffers *buffers = &inst->buffers[BUF_PERSIST];
> diff --git a/drivers/media/platform/qcom/iris/iris_buffer.h b/drivers/media/platform/qcom/iris/iris_buffer.h
> index c36b6347b077..03a32b91cf21 100644
> --- a/drivers/media/platform/qcom/iris/iris_buffer.h
> +++ b/drivers/media/platform/qcom/iris/iris_buffer.h
> @@ -106,7 +106,8 @@ void iris_get_internal_buffers(struct iris_inst *inst, u32 plane);
> int iris_create_internal_buffers(struct iris_inst *inst, u32 plane);
> int iris_queue_internal_buffers(struct iris_inst *inst, u32 plane);
> int iris_destroy_internal_buffer(struct iris_inst *inst, struct iris_buffer *buffer);
> -int iris_destroy_internal_buffers(struct iris_inst *inst, u32 plane);
> +int iris_destroy_internal_buffers(struct iris_inst *inst, u32 plane, bool force);
> +void iris_get_num_queued_internal_buffers(struct iris_inst *inst, u32 plane);
> int iris_alloc_and_queue_persist_bufs(struct iris_inst *inst);
> int iris_alloc_and_queue_input_int_bufs(struct iris_inst *inst);
> int iris_queue_buffer(struct iris_inst *inst, struct iris_buffer *buf);
> diff --git a/drivers/media/platform/qcom/iris/iris_vdec.c b/drivers/media/platform/qcom/iris/iris_vdec.c
> index 4143acedfc57..2c1a7162d2da 100644
> --- a/drivers/media/platform/qcom/iris/iris_vdec.c
> +++ b/drivers/media/platform/qcom/iris/iris_vdec.c
> @@ -408,7 +408,7 @@ int iris_vdec_streamon_input(struct iris_inst *inst)
>
> iris_get_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE);
>
> - ret = iris_destroy_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE);
> + ret = iris_destroy_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE, false);
> if (ret)
> return ret;
>
> @@ -496,7 +496,7 @@ int iris_vdec_streamon_output(struct iris_inst *inst)
>
> iris_get_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
>
> - ret = iris_destroy_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
> + ret = iris_destroy_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE, false);
If I was reviewing some changes (or even debugging) this specific C
file, I would not be able to understanding what this "false" means. I
would have to spend extra time, opening the declaration, going back and
forth, and breaking the flow.
An alternative approach is to keep the boolean parameter in a static
function (c local), and then add two function wrappers that have
explicit names.
regards,
Nicolas
> if (ret)
> return ret;
>
> diff --git a/drivers/media/platform/qcom/iris/iris_vidc.c b/drivers/media/platform/qcom/iris/iris_vidc.c
> index ca0f4e310f77..56531a7f0dfe 100644
> --- a/drivers/media/platform/qcom/iris/iris_vidc.c
> +++ b/drivers/media/platform/qcom/iris/iris_vidc.c
> @@ -233,8 +233,10 @@ int iris_close(struct file *filp)
> iris_session_close(inst);
> iris_inst_change_state(inst, IRIS_INST_DEINIT);
> iris_v4l2_fh_deinit(inst);
> - iris_destroy_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE);
> - iris_destroy_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
> + iris_destroy_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE, true);
> + iris_destroy_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE, true);
> + iris_get_num_queued_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE);
> + iris_get_num_queued_internal_buffers(inst, V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
> iris_remove_session(inst);
> mutex_unlock(&inst->lock);
> mutex_destroy(&inst->ctx_q_lock);
--
Nicolas Dufresne
Principal Engineer at Collabora
Powered by blists - more mailing lists