Re: [Scst-devel] SCST WWN Authentication

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 454-5900

Vladislav Bolkhovitin wrote:
> Hm, my manual on Qlogic 2300 doesn't contain description of 
> atio->initiator_port_name field, it's called "RESERVED" there. Looks 
> like it's an addition of the newer firmware. So, I think, we should 
> use that field directly not doing any translation at all and get rid 
> of all current loopId->WWN code. Although, I'm not sure about 2200, 
> Nate's code has "/* on qla23xx */" comment. If 2200 doesn't have such 
> feature, seems we will have to use the current way.
>
> Could somebody confirm either atio->initiator_port_name works on 2200 
> or not?
I can confirm it does work.  All I have are 2200 based cards at the 
moment, with the 1.83 firmware.

With the release of Nate's enhancements, I think it becomes a non-issue, 
though it was a good learning experience and has helped me to get more 
familiar with the code (trying to shake off 5 years of C rust is an 
interesting experience :) ).

Calvin
>
> Vlad
>
> Calvin Morrow wrote:
>> I made some changes to try and address what I wrote earlier, namely 
>> in WWN authentication mode, comparing WWNs to WWNs, rather than 
>> trying to match up loop ids.  The changes address the issues where 
>> loop_ids didn't match up, but WWNs did.
>>
>> The one issue outstanding, is that there seem to be circumstances 
>> (first powering on an initiator) that SCST is looking for loop_ids 
>> and WWNs before the internal qlogic database has added the new nodes.
>>
>> q2t_send_cmd_to_scst:New command instance=0 loop_id=2 tag=49168 
>> wwpn=0021e000038be3fb (210000e08b03fbe3)
>> q2t_make_name:comparing port_name of 210000e08b03fbe3 to 
>> 210000e08b0573f1
>> q2t_make_name:comparing port_name of 210000e08b03fbe3 to 
>> 10000000c9231e0e
>> q2t_make_name:***ERROR*** qla2x00tgt(0): Unable to find wwn login for 
>> loop id 2, using loop id instead
>>
>> After a LIP reset, the database is refreshed and everything works.  
>> Any idea how we can force a qlogic database refresh without messing 
>> things up?
>>
>>
>> scsi(0): Waiting for LIP to complete...
>> scsi(0): GID_PT entry - nn 200000e08b04a27f pn 210000e08b04a27f 
>> portid=011400.
>> scsi(0): GID_PT entry - nn 200000e08b0573f1 pn 210000e08b0573f1 
>> portid=011600.
>> scsi(0): GID_PT entry - nn 20000000c9231e0e pn 10000000c9231e0e 
>> portid=011800.
>> scsi(0): device wrap (011800)
>> qla2x00_configure_loop: exiting normally
>> qla2x00: DPC handler sleeping
>> q2t_response_pkt:IMMED_NOTIFY
>> q2t_handle_imm_notify:Port logout (S 00000000 I 0x2 -> L 0x0)
>> q2t_unreg_sess:Unregistering session d7b78bc0 (loop_id=2, 
>> scst_sess=d3f719c4)
>> q2t_free_session_done:tgt 0 list_empty 1
>> q2t_response_pkt:NOTIFY_ACK seq 00000000 status 1
>> scsi(0): Asynchronous PORT UPDATE ignored 0004/ffff/1411.
>> scsi(0): ha state 5 init_done 1 oper_mode 1 topo 8
>> q2t_async_event:Async event 0x8014 occured
>> q2t_clear_tgt_db:Clearing targets DB
>> q2t_response_pkt:ACCEPT_TGT_IO instance 0 status 003d lun 0000 
>> read/write 0 data_length 0000 target_id 02 exchange_id c238
>> q2t_send_cmd_to_scst:New command instance=0 loop_id=2 tag=49720 
>> wwpn=0021e000038be3fb (210000e08b03fbe3)
>> q2t_make_name:comparing port_name of 210000e08b03fbe3 to 
>> 210000e08b0573f1
>> q2t_make_name:comparing port_name of 210000e08b03fbe3 to 
>> 10000000c9231e0e
>> q2t_make_name:comparing port_name of 210000e08b03fbe3 to 
>> 210000e08b03fbe3
>> q2t_make_name:found wwn: 21:00:00:e0:8b:03:fb:e3 for loop_id: 2
>>
>> Mark Buechler wrote:
>>
>>> I was the original one who put the WWN authentication in. I knew and 
>>> still know very little about the qlogic driver. All I did was poke 
>>> around and found the correct struct. Vlad then took the code and 
>>> integrated it better.
>>>
>>> - Mark.
>>>
>>> On 2/26/06, *Calvin Morrow* <cal...@co... 
>>> <mailto:cal...@co...>> wrote:
>>>
>>>     I started looking at the WWN auth code today in the Qla target
>>>     driver to
>>>     see if I could figure out why the WWN auth code doesn't work the 
>>> first
>>>     time an initiator connects, but does seem to recognize an
>>>     initiator the
>>>     second and subsequent times going forward.
>>>
>>>     I put in some debug code to see what was going on inside of the 
>>> make
>>>     name function.  What seems to be happening, is that the adapter
>>>     registers the WWN with a loop ID of > 129, but SCST recognizes
>>>     connection attempts with a different loop_id.  If you look at the
>>>     debug
>>>     below, you'll see that SCST thinks my Emulex adapter has loop_id 
>>> = 1,
>>>     but the qlogic adapter has it recorded as loop_id = 132.  When it
>>>     can't
>>>     compare the loop_ids, the q2t_make_name function fails.
>>>
>>>     q2t_send_cmd_to_scst:New command instance=0 loop_id=1 tag=49120
>>>     wwpn=0010000023c90e1e (10000000c9231e0e)
>>>     q2t_make_name:looking at loop_id 1 comparing to 130, recorded 
>>> wwn is
>>>     21:00:00:e0:8b:05:73:f1
>>>     q2t_make_name:looking at loop_id 1 comparing to 131, recorded 
>>> wwn is
>>>     21:00:00:e0:8b:03:fb:e3
>>>     q2t_make_name:looking at loop_id 1 comparing to 132, recorded 
>>> wwn is
>>>     10:00:00:00:c9:23:1e:0e
>>>     q2t_make_name:***ERROR*** qla2x00tgt(0): Unable to find wwn 
>>> login for
>>>     loop id 1, using loop id instead
>>>     q2t_do_send_cmd_to_scst:Context 2
>>>     [3120]: scst_init_session:scst: Name 1 not found, using default 
>>> group
>>>
>>>     After a reboot, my adapter will be recorded in the qlogic list as
>>>     loop_id of 1, and q2t_make_name will succeed as the loop_ids will
>>>     match.
>>>
>>>     My questions to you are twofold:  1)  Does anyone know why this
>>>     behavior
>>>     is being exhibited, and 2)  If we enable WWN Auth, is there any 
>>> reason
>>>     why we should be trying to match loop_id - loop_id to
>>>     WWN?  Wouldn't it
>>>     make more sense to try and match the WWNs in q2t_make_name
>>>     instead?  I'd
>>>     be happy to attempt to change the code, but would like to know
>>>     first if
>>>     there are any reasons why the system currently acts in this 
>>> fashion.
>>>
>>>     Calvin
>>>
>>>
>>>     -------------------------------------------------------
>>>     This SF.Net email is sponsored by xPML, a groundbreaking scripting
>>>     language
>>>     that extends applications into web and mobile media. Attend the
>>>     live webcast
>>>     and join the prime developer group breaking into this new coding
>>>     territory!
>>>     
>>> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642 
>>>
>>>     
>>> <http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642> 
>>>
>>>     _______________________________________________
>>>     Scst-devel mailing list
>>>     Scs...@li...
>>>     <mailto:Scs...@li...>
>>>     https://lists.sourceforge.net/lists/listinfo/scst-devel
>>>     <https://lists.sourceforge.net/lists/listinfo/scst-devel>
>>>
>>>
>>
>>
>> ------------------------------------------------------------------------
>>
>> --- qla2x00t.c    2006-02-26 21:54:27.000000000 -0700
>> +++ qla2x00t.c.wwn    2006-02-26 21:54:21.000000000 -0700
>> @@ -1149,12 +1153,17 @@ STATIC void q2t_alloc_session_done(struc
>>  }
>>  
>>  
>> -static char* q2t_make_name(scsi_qla_host_t *ha, int loop_id)
>> +static char* q2t_make_name(scsi_qla_host_t *ha, atio_entry_t *atio)
>>  #ifdef FC_SCST_WWN_AUTH
>>  {
>>      int wwn_found = 0;
>>      char *wwn_str;
>>      fc_port_t *fcl;
>> +    uint16_t *pn;
>> +    int loop_id;
>> +   
>> +    loop_id = GET_TARGET_ID(ha, atio);
>> +    pn = (uint16_t *)(((char *)atio)+0x2a);
>>  
>>      wwn_str = kmalloc(2*WWN_SIZE, GFP_ATOMIC);
>>      if (wwn_str == NULL) {
>> @@ -1162,9 +1171,16 @@ static char* q2t_make_name(scsi_qla_host
>>          goto out;
>>      }
>>  
>> -    /* Find the WWN in the port db given the loop_id */
>> +    /* Find the WWN in the port db that matches the initiator port 
>> name */
>>      list_for_each_entry(fcl, &ha->fcports, list) {
>> -        if (loop_id == (fcl->loop_id & 0xFF)) {
>> +        if ((pn[0] == (uint16_t)((fcl->port_name[0] << 8)
>> +        + fcl->port_name[1])) &&
>> +        (pn[1] == (uint16_t)((fcl->port_name[2] << 8)
>> +        + fcl->port_name[3])) &&
>> +        (pn[2] == (uint16_t)((fcl->port_name[4] << 8)
>> +        + fcl->port_name[5])) &&
>> +        (pn[3] == (uint16_t)((fcl->port_name[6] << 8)
>> +        + fcl->port_name[7]))) {          sprintf(wwn_str, 
>> "%02x:%02x:%02x:%02x:%02x:%02x:%02x:%02x",
>>              fcl->port_name[0], fcl->port_name[1],              
>> fcl->port_name[2], fcl->port_name[3], @@ -1188,6 +1204,9 @@ out:
>>  #else
>>  {
>>      char *s;
>> +    int loop_id;
>> +   
>> +    loop_id = GET_TARGET_ID(ha, atio);
>>      s = kmalloc(12, GFP_ATOMIC);
>>      if (s != NULL)
>>          snprintf(s, 12, "%d", loop_id);
>> @@ -1265,7 +1284,7 @@ STATIC int q2t_send_cmd_to_scst(scsi_qla
>>  
>>          /* register session (remote initiator) */
>>          {
>> -            char *name = q2t_make_name(ha, loop_id);
>> +            char *name = q2t_make_name(ha, atio);
>>              if (name == NULL)
>>                  goto out_free_sess;
>>              sess->scst_sess = scst_register_session(
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by xPML, a groundbreaking scripting 
> language
> that extends applications into web and mobile media. Attend the live 
> webcast
> and join the prime developer group breaking into this new coding 
> territory!
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
> _______________________________________________
> Scst-devel mailing list
> Scs...@li...
> https://lists.sourceforge.net/lists/listinfo/scst-devel
>