bsa-acs: B_MEM_03, B_MEM_06, B_PER_09, B_PER_10, PCI_MM_05, PCI_MM_07 Linux tests should report WARNING instead of SKIPPED when DMA-capable devices lack drivers

[sunnywang-arm]

Currently, the following Linux-based tests are reported as SKIPPED when ACS Linux does not have drivers for DMA-capable devices.

• B_MEM_03 : Check Addressability for 64-bit-DMA dev
• B_MEM_06 : Check Addressability for non-64bit-DMA dev
• B_PER_09 — Check Memory Attributes of DMA
• B_PER_10 — Memory Attribute of I/O coherent DMA
• PCI_MM_05 — PCIe & PE common physical memory view
• PCI_MM_07 — No extra address translation

However, if the platform includes DMA-capable devices, the corresponding architectural rules are still applicable and should be reviewed, regardless of driver availability.

Reporting these tests as SKIPPED may hide potential compliance issues and give the impression that the checks are not relevant to the system.

Proposed change:

• Report WARNING instead of SKIPPED in such cases
• Print a clear message indicating:
  • DMA-capable devices are present
  • Required drivers are missing
  • Manual validation is required to ensure compliance

This approach would better reflect the test intent and improve visibility for potential compliance gaps.

[chetan-rathore]

Hi @sunnywang-arm,

If the required driver for a specific DMA-capable device is not present, ACS may not be able to determine whether that DMA-capable device exists on the system.

To meet the intent of the rule and avoid hiding a potentially applicable compliance check, the result should be reported as WARNING instead of SKIPPED. This covers both cases: either no DMA-capable device is present, or a DMA-capable device is present but ACS cannot detect it because the required Linux driver is missing or not bound.

The warning message should state that ACS could not detect any DMA-capable device through the available Linux driver path, and that if the system contains DMA-capable devices, manual validation is required to ensure compliance.

Let me know your thoughts.

Thanks,
Chetan

[sunnywang-arm]

@chetan-rathore The solution you proposed looks good to me.