[ofa-general] Re: [PATCH 1/2] opensm: Handle trap repress on trap 144 generation

[Hal Rosenstock]

On Fri, Mar 6, 2009 at 4:05 PM, Hal Rosenstock <hal.rosenstock at gmail.com> wrote:
> - Show quoted text -
> On Fri, Mar 6, 2009 at 3:51 PM, Hal Rosenstock <hal.rosenstock at gmail.com> wrote:
>> On Fri, Mar 6, 2009 at 3:41 PM, Sasha Khapyorsky <sashak at voltaire.com> wrote:
>>> On 12:55 Fri 06 Mar     , Hal Rosenstock wrote:
>>>>
>>>> It could be subsequent patch but thought it best to include it. Do you
>>>> want this separate ?
>>>
>>> Without clear usage case I would prefer to not have it at all - finally
>>> it is on a fast path.
>>
>> It's the same as what is done in the kernel for MAD response but I
>> removed it because I could see you were against this.
>>
>>>>
>>>> > BR is not used anywhere in OpenSM.
>>>>
>>>> No, but someone might use ib_types.h to build BM.
>>>
>>> We cannot know what will be needed then - this someone will need to care
>>> anyway.
>>
>> You could say that about a lot of things accepted which aren't fully integrated.
>>
>>>> > And this function which should process TrapRepress method does nothing,
>>>> > right?
>>>>
>>>> Just some validation; It doesn't need to do anything (just retire the
>>>> transaction).
>>>
>>> Then I don't understand - how is trap 144 repress handled? And why
>>> those changes in trap_rcv_process_response() were needed?
>>
>> Perhaps it's being overly pendantic. I can revise the patch to not
>> have the repress get this far.
>
> I misread your comment.

Maybe I didn't misread.

> trap_rcv needs to cause the repress for the trap to be generated.

This is already done.

I think what you were saying was why does the incoming repress need to
get this far into the code, right ?

-- Hal

>
> -- Hal
>
>> -- Hal
>>
>>> Sasha
>>>
>>
>