On Mon, Oct 25, 2021 at 09:24:41AM +0000, Prabhakar Kushwaha wrote:

Dear Kamal,

Hi Prabhakar,

quoted

-----Original Message-----
From: Kamal Heib <redacted>
Sent: Monday, October 25, 2021 9:27 AM
To: linux-rdma@vger.kernel.org
Cc: Michal Kalderon <mkalderon@marvell.com>; Ariel Elior
[off-list ref]; Doug Ledford [off-list ref]; Jason
Gunthorpe [off-list ref]; Kamal Heib [off-list ref]
Subject:  [PATCH for-next] RDMA/qedr: Remove unsupported
qedr_resize_cq callback

External Email

----------------------------------------------------------------------
There is no need to return always zero for function which is not supported.

Fixes: a7efd7773e31 ("qedr: Add support for PD,PKEY and CQ verbs")
Signed-off-by: Kamal Heib <redacted>
---
 drivers/infiniband/hw/qedr/main.c  |  1 -  drivers/infiniband/hw/qedr/verbs.c |
10 ----------  drivers/infiniband/hw/qedr/verbs.h |  1 -
 3 files changed, 12 deletions(-)

Have you tested this patch? I afraid, there may be a crash because of  this 

I do not think that we will face a crash, because the libqedr in the
rdma-core package dose not implement the resize_cq() callback.

Furthermore, if there is a bug in the kernel rdma core this doesn't mean
that the qedr driver need to fake supporting resize_cq() to avoid a crash!.

Anyway, To be in the safe side we I'll prepare another patch that checks
for NULL in the core and return -EOPNOTSUPP if resize_cq() is not set
by the driver.

static int ib_uverbs_resize_cq(struct uverbs_attr_bundle *attrs)
{
<snip>

        cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, attrs);
        if (!cq)
                return -EINVAL;

        ret = cq->device->ops.resize_cq(cq, cmd.cqe, &attrs->driver_udata);    <<<< No check for NULL.


--pk

Thanks,
Kamal