[PATCH v3] RDMA/siw: Fix handling of zero-sized Read and Receive Queues.

Bernard Metzler BMT at zurich.ibm.com
Fri Jan 8 05:51:23 EST 2021


-----"Jason Gunthorpe" <jgg at nvidia.com> wrote: -----

>To: "Bernard Metzler" <bmt at zurich.ibm.com>
>From: "Jason Gunthorpe" <jgg at nvidia.com>
>Date: 01/07/2021 08:40PM
>Cc: <linux-rdma at vger.kernel.org>, <linux-nvme at lists.infradead.org>,
><leon at kernel.org>, "Kamal Heib" <kamalheib1 at gmail.com>, "Yi Zhang"
><yi.zhang at redhat.com>, "kernel test robot" <lkp at intel.com>
>Subject: [EXTERNAL] Re: [PATCH v3] RDMA/siw: Fix handling of
>zero-sized Read and Receive Queues.
>
>On Wed, Dec 16, 2020 at 12:00:00PM +0100, Bernard Metzler wrote:
>> @@ -933,6 +937,7 @@ int siw_activate_tx(struct siw_qp *qp)
>>  
>>  		goto out;
>>  	}
>> +no_irq:
>>  	sqe = sq_get_next(qp);
>>  	if (sqe) {
>
>Can you please arrange this without the spaghetti goto's? goto is ok
>for error unwind at the tail of the function, but should not be used
>willy nilly. Move some of this into functions, use normal if
>statements, etc.
>
Okay. I tried to minimize changes but I now think
you are right - it's time to cleanup before it
gets too burlesque.

Will send a v4.

Thanks
Bernard.




More information about the Linux-nvme mailing list