Re: [PATCH v6 2/2] tools/testing/cxl: Enable zero sized decoders under hb0

[Richard Cheng <icheng@nvidia.com>]

On Thu, Jun 25, 2026 at 09:46:01AM +0800, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [Medium] Incorrect HPA base for zero-sized mock decoders causes them to overlap the start of decoder[0].
> --
> 
> commit e4836660bfdfee0da74cc6f18fe1e3c9826c4c04
> Author: Richard Cheng <icheng@nvidia.com>
> 
> tools/testing/cxl: Enable zero sized decoders under hb0
> 
> This commit introduces a mock_zero_size_decoders module parameter to cxl_test.
> It is intended to exercise zero-size DPA reservations by mocking committed,
> empty decoders above the auto-region under host-bridge0.
> 
> > diff --git a/tools/testing/cxl/test/cxl.c b/tools/testing/cxl/test/cxl.c
> > index 418669927fb00..117691b3c62f5 100644
> > --- a/tools/testing/cxl/test/cxl.c
> > +++ b/tools/testing/cxl/test/cxl.c
> [ ... ]
> > @@ -1145,6 +1178,20 @@ static bool mock_init_hdm_decoder(struct cxl_decoder *cxld)
> >  	base = window->base_hpa;
> >  	if (extended_linear_cache)
> >  		base += mock_auto_region_size;
> > +
> > +	/*
> > +	 * With mock_zero_size_decoders, decoders 1 and 2 of the special
> > +	 * endpoints mock committed, locked, empty decoders above the
> > +	 * decoder[0] auto-region (CXL r4.0 8.2.4.20.12). commit_end then
> > +	 * points at a zero-size decoder, exercising the zero-size
> > +	 * reservation and poison-by-endpoint code paths.
> > +	 */
> > +	if (cxld->id == 1 || cxld->id == 2) {
> > +		size_zero_mock_decoder_ep(cxld, base);
> 
> [Severity: Medium]
> Since the comment indicates these decoders are mocked above the decoder[0]
> auto-region, should base + mock_auto_region_size be passed here instead
> of base?
> 
> Passing base appears to set the zero-sized decoder's HPA to the start of the
> region, causing it to overlap with decoder[0].
> 
> > +		port->commit_end = cxld->id;
> > +		WARN_ON_ONCE(!cxld_registry_new(cxld));
> > +		return false;
> > +	}
> [ ... ]
> > @@ -1214,6 +1263,26 @@ static bool mock_init_hdm_decoder(struct cxl_decoder *cxld)
> >  
> >  		cxld_registry_update(cxld);
> >  		put_device(dev);
> > +
> > +		if (!mock_zero_size_decoders)
> > +			continue;
> > +
> > +		/*
> > +		 * Mirror the endpoint: commit the next two switch decoders
> > +		 * as zero-size + locked so the empty-decoder layout extends
> > +		 * end-to-end through the switch and host bridge.
> > +		 */
> > +		for (id = 1; id <= 2; id++) {
> > +			dev = device_find_child(&iter->dev, &id,
> > +						match_decoder_by_index);
> > +			if (WARN_ON(!dev))
> > +				continue;
> > +			cxld = to_cxl_decoder(dev);
> > +			size_zero_mock_decoder_sw(cxld, base, i);
> 
> [Severity: Medium]
> Similarly, does this incorrectly place the switch zero-sized mock decoders
> at the start of decoder[0]'s HPA range instead of above it?
> 
> > +			iter->commit_end = id;
> > +			cxld_registry_update(cxld);
> > +			put_device(dev);
> > +		}
> >  	}
> 
> -- 
> Sashiko AI review · https://sashiko.dev/#/patchset/20260625092857.41139-1-icheng@nvidia.com?part=2


This doesn't stand neither, [base, base-1] is an empty range, it decodes nothing and can't
overlap decoder[0]. Region assembly never sees it, endpoint empty decoder are skipped via cxled_empty(), and
the switch matching predicates compare range ends, which an empty range can never satisfy.

--Richard