Re: [PATCH] SIW: Module initialization

bart.vanassche-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org wrote on 10/05/2010 12:57:21 PM:

Bart Van Assche [off-list ref] 
Sent by: bart.vanassche-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org

10/05/2010 12:57 PM 

To 

Bernard Metzler [off-list ref] 

cc 

netdev-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org 

Subject 

Re: [PATCH] SIW: Module initialization 

On Tue, Oct 5, 2010 at 8:54 AM, Bernard Metzler [off-list ref] 

wrote:

quoted

+static int loopback_enabled;
+module_param(loopback_enabled, int, 0644);
+MODULE_PARM_DESC(loopback_enabled, "enable_loopback");

A minor comment: since kernel 2.6.31 the type "bool" can be used for
boolean kernel module parameters.

oh, thanks. there are currently two more occurrences (MPA crc, 0copy 
transmit). 
it will be changed accordingly. 

quoted

+ * TODO: Dynamic device management (network device 

registration/removal).

The current implementation is such that one siw device is created for
each network device found at kernel module load time. That means that
you force the user to load the siw kernel module after all other
kernel modules that register a network device. I'm not sure that's a
good idea.

good point. do you have a suggestion here - would you like to see 
siw to be enabled more selectively? 
iwarp is a protocol on top of TCP, explicitly defining the 
semantics of data fetching and placement. end-to-end connectivity 
and efficient data shipping is provided by TCP/IP. not taking into 
account dedicated RDMA hardware, any TCP stream may carry iwarp 
traffic. from that point of view, binding a software based 
rdma stack to dedicated devices is a concession to the 
given environment, in particular to the given rdma 
connection management. therefore, we started with binding to all 
available network devices. 

quoted

+                       if (!siw_device) {
+                               siw_device = siw_p;
+                               siw_p->next = NULL;
+                       } else {
+                               siw_p->next = siw_device->next;
+                               siw_device->next = siw_p;
+                       }

Why a custom linked list implementation instead of using <linux/list.h> 

?

i agree. will be changed. 

Bernard.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html