Hi Garrett,

Thanks for replying so quick! :) Please see my inline comments.


----- Original Message -----
From: Garrett D'Amore <[EMAIL PROTECTED]>
Date: Monday, October 22, 2007 11:24 am
Subject: Re: [networking-discuss] design document review request for 
IPoIB driver GLDv3 porting.
To: Lizhi Hou <[EMAIL PROTECTED]>
Cc: [email protected]

 > Lizhi Hou wrote:
 > > This is the design document for porting IPoIB driver to GLDv3
 > > framework. I am looking for folks to review this.
 > > I have done a prototype code based on this design. It works fine. :)
 >
 > This looks great.
 >
 > I'd like some explanation of how the dependency on soft rings works
 > out... its it just a performance issue, or are soft rings required
 > for
 > *correct* functionality?
 >

it is just a performance issue. this driver also works fine without soft 
rings.

 > I'd also like to see any IB-specific mac statistics enumerated.
 >
The gldv3 porting will not introduce new statistics. So I did not talk 
about statistics variables in my document.
In GLDv2 driver, these statistics are supported:
        sp->glds_overflow = state->id_tx_short;    /* Tx overflow */
    sp->glds_speed = state->id_link_speed;
    sp->glds_media = GLDM_IB;
    sp->glds_errxmt = state->id_ah_error;    /* failed AH translation */
    sp->glds_norcvbuf = state->id_rx_short;    /* # times below water 
mark */
    sp->glds_intr = state->id_num_intrs;    /* number of intrs */

 > Do you anticipate having link-up/link-down messages ala ethernet
 > handled?
Yes.
The driver will register async handle to IB framework. When the link 
up/down, the IB framework will call this async handle. Then the driver 
async handle will call mac_link_update() to inform GLDv3.

 > Also you need to fix the reference to mac_ib.h, (I assume the file
 > is a
 > .h file, but you dropped the suffix at one point.)
You are right. This is my typo error. I will fix this.

 >
 >    -- Garrett
 >
 > >
 > > Thanks in advance!
 > > Lizhi
 > > ------------------------------------------------------------------
 > ------
 > >
 > > _______________________________________________
 > > networking-discuss mailing list
 > > [email protected]
 >
 > _______________________________________________
 > networking-discuss mailing list
 > [email protected]
 >


Thanks,
Lizhi
_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to