[ofa-general] RE: [PATCH] libibmad: added support for handling of BM (Baseboard management) MADs - FIXED without rmpp

[Itai Baz]

As far as I know and checked, there is no need to implement a special
transactions mechanism, and the code is fine
There is transaction mechanism defined for IB-ML and IB-MME in the spec,
not for BM MADs.

Itai 

-----Original Message-----
From: Hal Rosenstock [mailto:hal.rosenstock at gmail.com] 
Sent: Thursday, March 19, 2009 5:49 PM
To: Itai Baz
Cc: Sasha Khapyorsky; OpenIB
Subject: Re: [PATCH] libibmad: added support for handling of BM
(Baseboard management) MADs - FIXED without rmpp

On Thu, Mar 19, 2009 at 11:00 AM, Itai Baz <itaib at mellanox.co.il> wrote:
> As far as I'm aware, the response_expected handling is sufficient.
> BMGet and BMSet always have response.
> BMSend may be used both as a request from client to server, and as a 
> response from server to client

Yes, but is transaction handling (request/response) used or needed on
some BMSends ?

-- Hal

> -----Original Message-----
> From: Hal Rosenstock [mailto:hal.rosenstock at gmail.com]
> Sent: Thursday, March 19, 2009 4:50 PM
> To: Itai Baz
> Cc: Sasha Khapyorsky; OpenIB
> Subject: Re: [PATCH] libibmad: added support for handling of BM 
> (Baseboard management) MADs - FIXED without rmpp
>
> On Thu, Mar 19, 2009 at 10:25 AM, Itai Baz <itaib at mellanox.com> wrote:
>> This patch adds support for handling of BM (Baseboard management)
> MADs.
>>
>> I checked Hal's comment regarding RMPP, indeed there is no need for 
>> it
>
>> for BM, so I have removed rmpp, and i'm using now mad_rpc
>
> Thanks.
>
> Also, is the response_expected handling sufficient ? For any BMSend, 
> it will say no response expected. Is that what is really desired ? I'm

> not sure on what basis the BMA sets the response bit in the AM on a 
> "response" to a BMSend but that is used in the kernel (to determine an

> incoming BM response). Should this mimic that idea ?
>
> -- Hal
>
> <snip...>
>