Skip to contentSkip to conversation

probe issue for Dice/OXFW drivers

DISCUSSION Clemens Ladisch 2 participants 2 messages Aug 12, 2015 – Aug 17, 2015 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw
Quoted 5 lines · Takashi Sakamoto

Recently, I received an issue from a user of Mackie Onyx-i series. According to the user, snd-dice is not applied to his Dice model. [...] According to the log, I cannot find the first transaction which snd-dice performs. Additionally, no error messages in syslog.

There are some checks performed before the first actual transaction.

The kernel loads the driver if an entry in dice_id_table[] matches, i.e., if version==1. This values indeed is in the unit directory:

Quoted 18 lines

ROM header and bus information block
-----------------------------------------------------------------

400  0404f507  bus_info_length 4, crc_length 4, crc 62727
404  31333934  bus_name "1394"
408  e0008102  irmc 1, cmc 1, isc 1, bmc 0, cyc_clk_acc 0, max_rec 8 (512)
40c  000ff210  company_id 000ff2     |
410  01800af7  device_id 1001800af7  | EUI-64 000ff21001800af7

root directory
-----------------------------------------------------------------

414  000639d3  directory_length 6, crc 14803
418  03000ff2  vendor
41c  8100000a  --> descriptor leaf at 444
420  17000006  model
424  8100000d  --> descriptor leaf at 458
428  0c0087c0  node capabilities per IEEE 1394
42c  d1000001  --> unit directory at 430
Quoted 7 lines

unit directory at 430
-----------------------------------------------------------------

430  00045246  directory_length 4, crc 21062
434  12000ff2  specifier id
438  13000001  version
43c  17000006  model
440  8100000b  --> descriptor leaf at 46c

The dice_interface_check() function then makes other checks:
/*
* Check that GUID and unit directory are constructed according to DICE
* rules, i.e., that the specifier ID is the GUID's OUI, and that the
* GUID chip ID consists of the 8-bit category ID, the 10-bit product
* ID, and a 22-bit serial number.
*/

These values are correct, except for the category ID.

The driver already has an category ID exception for Weiss devices. It would be possible to add another exception, but given that firmware writers apparently like to change this value without good reason, it might be a better idea to just remove the category ID check.

Regards,
Clemens

#2 lore raw

Hi Clemens,

Quoted 34 lines · Clemens Ladisch
Quoted 18 lines

ROM header and bus information block
-----------------------------------------------------------------

400  0404f507  bus_info_length 4, crc_length 4, crc 62727
404  31333934  bus_name "1394"
408  e0008102  irmc 1, cmc 1, isc 1, bmc 0, cyc_clk_acc 0, max_rec 8 (512)
40c  000ff210  company_id 000ff2     |
410  01800af7  device_id 1001800af7  | EUI-64 000ff21001800af7

root directory
-----------------------------------------------------------------

414  000639d3  directory_length 6, crc 14803
418  03000ff2  vendor
41c  8100000a  --> descriptor leaf at 444
420  17000006  model
424  8100000d  --> descriptor leaf at 458
428  0c0087c0  node capabilities per IEEE 1394
42c  d1000001  --> unit directory at 430
Quoted 7 lines

unit directory at 430
-----------------------------------------------------------------

430  00045246  directory_length 4, crc 21062
434  12000ff2  specifier id
438  13000001  version
43c  17000006  model
440  8100000b  --> descriptor leaf at 46c

The dice_interface_check() function then makes other checks:
/*
* Check that GUID and unit directory are constructed according to DICE
* rules, i.e., that the specifier ID is the GUID's OUI, and that the
* GUID chip ID consists of the 8-bit category ID, the 10-bit product
* ID, and a 22-bit serial number.
*/

Quoted 1 line

These values are correct, except for the category ID.

Aha. I've overlooked this case. I assumed that we can see 'No such device' in syslog, due to returning -ENODEV.

Quoted 4 lines

The driver already has an category ID exception for Weiss devices. It would be possible to add another exception, but given that firmware writers apparently like to change this value without good reason, it might be a better idea to just remove the category ID check.

Sounds to work, while I've never read such a specification about the Dice rule (I'm not so good at Dice itself). Is it possible that you post patches for this issue?

Regards

Takashi Sakamoto