audio: igo_nr: bound host active_channel_idx before adopting config

[abonislawski]

Adopt the blob only when igo_nr_check_config_validity() passes so a host index past process_enable[] cannot reach prepare/process.

[Copilot]

Pull request overview

This PR aims to prevent unsafe adoption of a host-provided igo_nr configuration blob by only applying it when igo_nr_check_config_validity() succeeds, so an out-of-range active_channel_idx cannot propagate into runtime processing.

Changes:

• Gate cd->config = *p_config; behind a config-validity check in igo_nr_set_igo_params().
• Avoid applying new runtime config when the blob fails validation.

Comments suppressed due to low confidence (1)

src/audio/igo_nr/igo_nr.c:728

• igo_nr_check_config_validity() calls comp_get_data_blob() internally. In this function we already called comp_get_data_blob() to get p_config, then call the validator which may read/consume/swap the blob again. If another host update lands between these two calls, the validity check could apply to a different blob than the one copied into cd->config (TOCTOU), reintroducing the out-of-bounds risk this change is trying to prevent. Validate the already-fetched p_config directly (or refactor the validator to take a config pointer) to avoid the double-blob access.

	/* Adopt the host blob only when new config is valid */
	if (p_config && igo_nr_check_config_validity(dev, cd) == 0) {
		comp_info(dev, "New config detected.");
		cd->config = *p_config;
		igo_nr_print_config(mod);
	}

[lgirdwood]

@lrudyX good to merge ?

[lrudyX]

@lrudyX good to merge ?
There is one test failing (test error), caused by QB environment (too long path). This can be ignored.