The evolution of the hierarchical listing left warts like the following:
if (p->memdevs && !p->ports && !p->endpoints) {
jchilddevs = json_object_new_array();
...whereby it tried to avoid creating a container for child devices if
another container deeper in the hierarchy might supersede the upper-level
containers. I.e. if endpoints are included in the listing then there will
be nothing to report at the bus level. The protection is unnnecessary
because cond_add_put_array_suffix() already handles the case of dropping
empty containers when a lower level container subsumes all the objects.
Moreover, it's a broken check when adding objects at new levels of the
topology. CXL devices attached to an RCH cause memdevs to appear directly
beneath a bus object, and not an intervening port. So in preparation for
that change, delete all the unnecessary special casing for "jchildobj"
container creation.
Reported-by: Vishal Verma <vishal.l.ve...@intel.com>
Fixes: 41d6769393f4 ("cxl/list: Move enabled memdevs underneath their endpoint")
Signed-off-by: Dan Williams <dan.j.willi...@intel.com>
---
cxl/filter.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/cxl/filter.c b/cxl/filter.c
index 56c659965891..040e7deefb3e 100644
--- a/cxl/filter.c
+++ b/cxl/filter.c
@@ -971,7 +971,7 @@ walk_child_ports(struct cxl_port *parent_port, struct
cxl_filter_params *p,
continue;
}
- if (p->memdevs && !p->endpoints) {
+ if (p->memdevs) {
jchilddevs = json_object_new_array();
if (!jchilddevs) {
err(p,
@@ -1151,7 +1151,7 @@ int cxl_filter_walk(struct cxl_ctx *ctx, struct
cxl_filter_params *p)
}
}
- if (p->memdevs && !p->ports && !p->endpoints) {
+ if (p->memdevs) {
jchilddevs = json_object_new_array();
if (!jchilddevs) {
err(p,
@@ -1169,7 +1169,7 @@ int cxl_filter_walk(struct cxl_ctx *ctx, struct
cxl_filter_params *p)
continue;
}
}
- if (p->regions && !p->decoders) {
+ if (p->regions) {
jchildregions = json_object_new_array();
if (!jchildregions) {
err(p,
