Re: [RFC PATCH 2/5] cxl/region: Factor decoder poison-list retrieval

From: Shaikh Kamaluddin

Date: Mon Nov 16 2026 - 20:03:32 EST


On Fri, Oct 16, 2026 at 11:21:03PM +0100, Jonathan Cameron wrote:
> On Sat, 10 Oct 2026 21:50:10 +0530
> Shaikh Kamaluddin <shaikhkamal2012@xxxxxxxxx> wrote:
>
> > poison_by_decoder() performs two different operations: it queries an
> > unmapped DPA gap preceding an endpoint decoder, then queries the DPA
> > resource assigned to that decoder and its region.
> >
> > Factor the second operation into cxl_get_poison_by_decoder(). Give the
> > helper an optional snapshot so a later region-probe caller can retain
> > records from exactly one mapped decoder. Keep the existing endpoint
> > walk, skipped-DPA queries, partition-tail queries, -EFAULT handling, and
> > trace behavior unchanged.
> >
> > Signed-off-by: Shaikh Kamaluddin <shaikhkamal2012@xxxxxxxxx>
> > ---
> > drivers/cxl/core/region.c | 41 ++++++++++++++++++++++++++++++++++-----
> > 1 file changed, 36 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/cxl/core/region.c b/drivers/cxl/core/region.c
> > index 75b8092e6dc2..c72524923b38 100644
> > --- a/drivers/cxl/core/region.c
> > +++ b/drivers/cxl/core/region.c
> > @@ -2940,6 +2940,40 @@ static int cxl_get_poison_unmapped(struct cxl_memdev *cxlmd,
> > return rc;
> > }
> >

Hi Jonathan,

> > +enum cxl_poison_query_policy {
> > + CXL_POISON_QUERY_DIAGNOSTIC,
> > + CXL_POISON_QUERY_PREONLINE,
>
> For now, perhaps focus on flags defining what you want done, not when the
> query is happening.
>

Make sense. I will drop the enum. The snapshot argument already tells
the helper what to do: NULL for trace path, a snapshot for region probe.
The helpler will return the error unchanged and each caller will handle
-EFAULT itself, as poison_by_decoder() does today.

> > +};
> > +
> > +static int cxl_get_poison_by_decoder(struct cxl_endpoint_decoder *cxled,
> > + struct cxl_poison_snapshot *snapshot,
> > + enum cxl_poison_query_policy policy)
> > +{
> > + struct cxl_memdev *cxlmd = cxled_to_memdev(cxled);
> > + struct cxl_dev_state *cxlds = cxlmd->cxlds;
> > + enum cxl_partition_mode mode;
> > + u64 offset, length;
> > + int rc;
> > +
> > + if (!cxled->dpa_res)
> > + return 0;
> > +
> > + mode = cxlds->part[cxled->part].mode;
> > + offset = cxled->dpa_res->start;
> > + length = resource_size(cxled->dpa_res);
> > + if (snapshot)
> > + rc = cxl_mem_get_poison_snapshot(cxlmd, offset, length,
> > + cxled->cxld.region, snapshot);
> > + else
> > + rc = cxl_mem_get_poison(cxlmd, offset, length,
> > + cxled->cxld.region);
> > +
> > + if (rc == -EFAULT && mode == CXL_PARTMODE_RAM &&
> > + policy == CXL_POISON_QUERY_DIAGNOSTIC)
>
> This is becoming odd enough I'd add some comments saying why
> we are eating the error return.
>

Will add one. A device that keeps no poison list for volatile ranges
returns Invalid Physical address for them (CXL r3.2 8.2.9.8.4.1) which
comes back as -EFAULT. The trace path treats that as "no records"
instead of failing the whole walk.

> > + rc = 0;
> > + return rc;
> > +}
> > +
> > static int poison_by_decoder(struct device *dev, void *arg)
> > {
> > struct cxl_poison_context *ctx = arg;
> > @@ -2971,11 +3005,8 @@ static int poison_by_decoder(struct device *dev, void *arg)
> > return rc;
> > }
> >
> > - offset = cxled->dpa_res->start;
> > - length = cxled->dpa_res->end - offset + 1;
> > - rc = cxl_mem_get_poison(cxlmd, offset, length, cxled->cxld.region);
> > - if (rc == -EFAULT && mode == CXL_PARTMODE_RAM)
> > - rc = 0;
> > + rc = cxl_get_poison_by_decoder(cxled, NULL,
> > + CXL_POISON_QUERY_DIAGNOSTIC);
> > if (rc)
> > return rc;
> >
>