Re: regarding ata_msg_*()

[Tejun Heo <htejun@gmail.com>]

Borislav Petkov wrote:
>> My suggestion is to keep the current (pre-msg_enable) model for the 
>> first conversion and categorize debug messages further after that.  So, 
>> message categories will be...
>>
>> ATA_MSG_ERR
>> ATA_MSG_WARN
>> ATA_MSG_INFO
>> -------------> all above are enabled by default
>> ATA_MSG_DEBUG
>> ATA_MSG_VDEBUG
>>
>> Then, you have 1-to-1 mapping w/ the existing messages.  You can simply 
>> incorporate message enabled tests into ata_*_printk() functions and the 
>> conversion would be trivial.
>>
>> After that's complete, we can diversify ATA_MSG_DEBUG and ATA_MSG_VDEBUG 
>> by separating out chatty ones out.  e.g. you can separate out SG 
>> mapping/unmapping (including padding) debug messages, which produce 
>> massive amount of logs when enabled, into ATA_MSG_SG or something. 
>> After several such separations, debug messages should be quite 
>> manageable && the categories wouldn't be too elaborate.
> 
> Yes, this sounds logical, I like it, it really fits libata more but at the time
> I made that msg classification I kinda wanted to stick closely to Becker's
> initial implementation :). But you're perfectly right, this is the way to go so
> I'll do that after reading what the others have to say. However, as I said
> before, IMHO it would be of higher priority now to convert to this new scheme
> and changing the dbg levels later would be trivial once we got all of libata
> done.

Jeff, what do you think?

Borislav, can you restore configuration messages in ata_dev_configure()? 
  Those messages need to be printed.

Thanks.

-- 
tejun