volume: validate host-supplied config and fix init error path#10919
Open
lgirdwood wants to merge 3 commits into
Open
volume: validate host-supplied config and fix init error path#10919lgirdwood wants to merge 3 commits into
lgirdwood wants to merge 3 commits into
Conversation
lgirdwood
requested review from
dbaluta,
kv2019i,
lbetlej,
mmaka1 and
plbossart
as code owners
Contributor
There was a problem hiding this comment.
Pull request overview
Note
Copilot was unable to run its full agentic suite in this review.
Hardens the volume component against malformed host-supplied IPC payloads and fixes a use-after-free in an IPC3 init error path.
Changes:
- Add IPC4 init payload size validation before iterating per-channel config.
- Tighten attenuation control payload validation to require exactly 32 bits.
- Fix IPC3 init error cleanup order to avoid dereferencing freed memory.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
| src/audio/volume/volume_ipc4.c | Adds IPC4 init/config size validation and tightens attenuation payload size checks. |
| src/audio/volume/volume_ipc3.c | Adjusts error-path frees to avoid a use-after-free during init failure. |
Comment on lines
+133
to
+137
| if (cfg->size < sizeof(*vol) + channels_count * sizeof(vol->config[0])) { | ||
| comp_err(dev, "Invalid init payload size %zu for %u channels", | ||
| cfg->size, channels_count); | ||
| return -EINVAL; | ||
| } |
Member
Author
There was a problem hiding this comment.
No overflow here: channels_count is validated to be in [1, SOF_IPC_MAX_CHANNELS] (== 8) a few lines above, before this size check, so channels_count * sizeof(vol->config[0]) is bounded by 8 * a small struct and can't overflow size_t.
Comment on lines
283
to
286
| if (data_size < (int)sizeof(uint32_t) || data_size > sizeof(uint32_t)) { | ||
| comp_err(dev, "attenuation data size %d is incorrect", data_size); | ||
| return -EINVAL; | ||
| } |
Member
Author
There was a problem hiding this comment.
Addressed — the check is now a single data_size != (int)sizeof(uint32_t), with the cast keeping both sides signed (no signed/unsigned comparison).
singalsu
approved these changes
Jun 16, 2026
lyakh
reviewed
Jun 16, 2026
| mod_free(mod, cd); | ||
| /* free cd->vol before cd: freeing cd first would leave the | ||
| * following cd->vol read dereferencing freed memory | ||
| */ |
Collaborator
There was a problem hiding this comment.
please remove the obvious comment
| /* only support attenuation in format of 32bit; the payload is | ||
| * dereferenced as a uint32_t below so it must be exactly that size | ||
| */ | ||
| if (data_size < (int)sizeof(uint32_t) || data_size > sizeof(uint32_t)) { |
Collaborator
There was a problem hiding this comment.
ehm, data_size != sizeof(uint32_t)?
Member
Author
There was a problem hiding this comment.
Done — simplified to a single data_size != (int)sizeof(uint32_t) check. (Cast to int to keep it a clean signed comparison, since data_size is int — also addresses the signed/unsigned note.)
On an invalid ramp type the init error path freed the component data then read a field from it. Free the dependent allocation before the component data and stop dereferencing it afterwards. Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
lyakh
approved these changes
Jun 18, 2026
Member
Author
|
Some valgrind errors: |
Init read the per-channel config[] array using the stream channel count without checking the payload was large enough, reading past the mailbox. The payload comes in two forms: an all-channels entry (channel_id set to ALL_CHANNELS_MASK) that carries a single config applied to every channel, or one config entry per channel. The per-channel loop always indexed config[channel], so the common all-channels payload (one entry) was read out of bounds for every channel beyond the first; the result was then discarded, masking the over-read. Require at least one entry before dereferencing config[0] to detect the form, require one entry per channel only for the per-channel form, and index config[] by the selected entry so no read goes past the payload. Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
The attenuation setter only rejected oversized payloads, then dereferenced the data as a 32-bit value; a shorter payload read past the mailbox. Require exactly a 32-bit payload. Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Member
Author
|
Pushed a fix: the testbench gain CI (sof-hda-benchmark-gain16) was failing init with The previous size check required one config entry per channel, but the common all-channels payload (channel_id == ALL_CHANNELS_MASK) carries a single entry for all channels — a 64-byte payload = base_cfg(40) + 1 entry(24). The check wrongly rejected it. The init loop also indexed
This both fixes the CI regression and closes the original out-of-bounds read. Commit message updated to match. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Hardening of host-supplied data in the volume component, plus an error-path
fix:
freed and then dereferenced)
config array before iterating it
No functional change for valid configurations.