On Mon, May 17, 2021 at 07:11:52PM +0200, Greg KH wrote:

On Mon, May 17, 2021 at 01:47:39PM -0300, Jason Gunthorpe wrote:

quoted

qib should not be creating a mess of kobjects to attach to the port
kobject - this is all attributes. The proper API is to create an
attribute_group list and create it against the port's kobject.

Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
 drivers/infiniband/hw/qib/qib.h       |   5 +-
 drivers/infiniband/hw/qib/qib_sysfs.c | 596 +++++++++++---------------
 2 files changed, 248 insertions(+), 353 deletions(-)

diff --git a/drivers/infiniband/hw/qib/qib.h b/drivers/infiniband/hw/qib/qib.h
index 88497739029e02..3decd6d0843172 100644
+++ b/drivers/infiniband/hw/qib/qib.h

@@ -521,10 +521,7 @@ struct qib_pportdata {
 
 	struct qib_devdata *dd;
 	struct qib_chippport_specific *cpspec; /* chip-specific per-port */
-	struct kobject pport_kobj;
-	struct kobject pport_cc_kobj;
-	struct kobject sl2vl_kobj;
-	struct kobject diagc_kobj;
+	const struct attribute_group *groups[5];

As you initialize these all at once, why not just make this:
	struct attribute_group **groups;

and then set the groups up at build time instead of runtime as part of a
larger structure like the ATTRIBUTE_GROUPS() macro does for "simple"
drivers?  That way you aren't fixed at the array size here and someone
has to go and check to verify you really have properly 0 terminated the
list and set up the pointers properly.

qib has a variable list of group memberships that can only be
determined at runtime:

        if (qib_cc_table_size && ppd->congestion_entries_shadow)
                *cur_group++ = &port_ccmgta_attribute_group;

So it can't be setup statically at compile time.

hfi1 was transformed as you describe.

Thanks,
Jason