[ofa-general] Re: [PATCH] opensm/opensm/osm_subnet.c: add checks for HOQ and Leaf HOQ input values

[weiny2 at llnl.gov]

> On 08:38 Wed 09 Apr     , weiny2 at llnl.gov wrote:
>>
>> What if we set it to 0x13?  This would be the maximum value that will
>> not
>> "lock" up the fabric.  We could also add to the error message that the
>> admin needs to specify 0x14 if they specifically want "infinity" to be
>> set?
>
> I think in the case when parameter value provided by user is wrong it
> is not easy to guess correctly what original wishes was. Probably we
> just need to add something like:
>
>   ## valid values are <= 0x14
>
> in config file template and reject any invalid values (I mean set to
> defaults)?

The config file comments already mention this:

		"# The code of maximal time a packet can wait at the head of\n"
		"# transmission queue.\n"
		"# The actual time is 4.096usec * 2^<head_of_queue_lifetime>\n"
		"# The value 0x14 disables this mechanism\n"
		"head_of_queue_lifetime 0x%02x\n\n"

But I guess "disables" should be "infinity" to make this more clear.

I will leave it up to you as to which patch you want.  As Hal said I can
see either side.  Both patches warn the user that the value they submitted
was not valid; and subsequently what value OpenSM is using instead.

Whatever you want to do Sasha...  :-)

Ira