CVE-2023-52792

What it actually does

The flaw leaves endpoint and switch decoder resources allocated after cxl_region_setup_targets() fails, without setting the decoder's region to NULL or freeing the decoder, which blocks future dpa_size_store() calls with -EBUSY and causes upstream switch commit errors.

Why it matters to the business

Organizations using CXL memory expansion in multi-switch topologies may encounter repeated region creation failures after an initial invalid attempt, reducing the reliability of dynamic memory scaling in Linux-based servers and appliances.

How an attack unfolds

An attacker able to trigger repeated CXL region attach operations with invalid configurations can induce the error path, leaving decoder resources locked and preventing legitimate region setup or reuse in affected kernels.

How to fix it

Update the Linux kernel to a version containing commit 90db4c1d5ebaf574d3c3065c055977982c378a83 which returns immediately on cxl_region_setup_targets() failure and adds an overflow check for the targets array.

How to confirm the fix worked

Verify the presence of commit 90db4c1d5ebaf574d3c3065c055977982c378a83 in the kernel source tree; test by attempting an invalid region creation followed by a valid one in a 2-switch by 2-endpoint topology and confirm both succeed without -EBUSY.

Typical attack path

1. Identify a system exposing CXL region configuration interfaces
2. Submit an invalid region creation request that exceeds target limits or uses an unsupported topology
3. Trigger the error path in cxl_region_attach that performs incomplete cleanup
4. Observe subsequent valid region creation attempts fail with resource busy or commit errors

How to detect it

Inspect kernel logs and dmesg output for repeated cxl region attach failures, -EBUSY returns from dpa_size_store, or commit errors on upstream switch decoders following region configuration attempts.

What your security team should do

• Prioritize kernel updates containing the referenced commit on all systems using CXL regions
• Review CXL region configuration scripts and automation for error handling around attach failures
• Monitor kernel logs for cxl region decoder errors until patches are deployed