Re: [PATCH v7 1/2] cxl/hdm: Allow zero sized HDM decoders

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] cxl_dpa_free() early-exits for zero-size DPA reservations, permanently breaking the LIFO teardown constraint and runtime reconfiguration.
--

--- Patch [1]: [PATCH v7 1/2] cxl/hdm: Allow zero sized HDM decoders ---
commit 6f85fee709a80e6ce5f72e57edb7b3f3de0aa965
Author: Richard Cheng <icheng@nvidia.com>
cxl/hdm: Allow zero sized HDM decoders

This commit treats empty CXL HDM decoders as first-class entities by backing
them with a zero-sized kmalloc'd resource. This allows the OS to correctly
enumerate ports on systems where the BIOS commits zero-sized decoders.

> diff --git a/drivers/cxl/core/hdm.c b/drivers/cxl/core/hdm.c
> index 0c80b76a5f9b4..ccbab2e21f060 100644
> --- a/drivers/cxl/core/hdm.c
> +++ b/drivers/cxl/core/hdm.c

[ ... ]

> @@ -545,7 +569,7 @@ int cxl_dpa_free(struct cxl_endpoint_decoder *cxled)
>  	struct device *dev = &cxled->cxld.dev;
>  
>  	guard(rwsem_write)(&cxl_rwsem.dpa);
> -	if (!cxled->dpa_res)
> +	if (cxled_empty(cxled))
>  		return 0;

[Severity: High]
Does this early return prevent zero-sized reservations from being correctly
released?

In __cxl_dpa_reserve(), a zero-sized decoder is backed by a dummy resource,
assigned to cxled->dpa_res, and port->hdm_end is incremented. However,
because cxled_empty() returns true when resource_size() is 0, cxl_dpa_free()
will return 0 here and skip the call to devm_cxl_dpa_release().

This appears to leave the dummy resource allocated and port->hdm_end stranded.
Since port->hdm_end is never decremented, will this permanently break the LIFO
teardown constraint for any subsequent attempts to free preceding decoders on
the port?

>  	if (cxled->cxld.region) {
>  		dev_dbg(dev, "decoder assigned to: %s\n",

[ ... ]

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260708104024.68029-1-icheng@nvidia.com?part=1