Re: [PATCH RFC] RDMA/core: Handle ARPHRD_NONE devices

[Chuck Lever III <chuck.lever@oracle.com>]

> On Jun 6, 2023, at 7:17 PM, Jason Gunthorpe <jgg@nvidia.com> wrote:
> 
> On Tue, Jun 06, 2023 at 08:15:36PM +0000, Chuck Lever III wrote:
>> 
>> 
>>> On Jun 6, 2023, at 11:48 AM, Jason Gunthorpe <jgg@nvidia.com> wrote:
>>> 
>>> On Fri, Jun 02, 2023 at 03:24:30PM -0400, Chuck Lever wrote:
>>>> From: Chuck Lever <chuck.lever@oracle.com>
>>>> 
>>>> We would like to enable the use of siw on top of a VPN that is
>>>> constructed and managed via a tun device. That hasn't worked up
>>>> until now because ARPHRD_NONE devices (such as tun devices) have
>>>> no GID for the RDMA/core to look up.
>>>> 
>>>> But it turns out that the egress device has already been picked for
>>>> us. addr_handler() just has to do the right thing with it.
>>>> 
>>>> Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
>>>> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
>>>> ---
>>>> drivers/infiniband/core/cma.c |    4 ++++
>>>> 1 file changed, 4 insertions(+)
>>>> 
>>>> diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c
>>>> index 56e568fcd32b..3351dc5afa17 100644
>>>> --- a/drivers/infiniband/core/cma.c
>>>> +++ b/drivers/infiniband/core/cma.c
>>>> @@ -704,11 +704,15 @@ cma_validate_port(struct ib_device *device, u32 port,
>>>> ndev = dev_get_by_index(dev_addr->net, bound_if_index);
>>>> if (!ndev)
>>>> return ERR_PTR(-ENODEV);
>>>> + } else if (dev_type == ARPHRD_NONE) {
>>>> + sgid_attr = rdma_get_gid_attr(device, port, 0);
>>> 
>>> It seems believable, should it be locked to iwarp devices?
>>> 
>>> More broadly, should iwarp devices just always do this and skip all
>>> the rest of it?
>>> 
>>> I think it also has to check that the returned netdev in the sgid_attr
>>> matches the egress netdev selected?
>> 
>> Both @ndev and sgid_attr.ndev are NULL here. 
> 
> The nedev to check is the dev_addr->bound_dev_if, that represents the netdev
> 
> It is some iwarp mistake that the sgid attr's don't have proper
> netdevs, that is newer than iwarp so it never got updated.
> 
> So, maybe it is too hard to fix for this, and maybe we can just assume
> it all has to be right

sgid_attr.index is not set it looks like. I will send you an "official"
version of this patch and we can go from there.


--
Chuck Lever