multiple issues with devstat_*(9)

[Alexander Motin]

On 11.04.2011 12:34, Alexander Best wrote:
> On Mon Apr 11 11, Alexander Motin wrote:
>> Alexander Best wrote:
>>> On Mon Apr 11 11, Alexander Motin wrote:
>>>> Alexander Best wrote:
>>>>> my cd0 is a pata atapi dvdrom drive and i have the following in my kernel conf:
>>>>>
>>>>> device          atacore
>>>>> device          ahci
>>>>> device          atajmicron
>>>>> device          atapci
>>>>> #device         atapicd
>>>>> device          atapicam
>>>>>
>>>>> device          umass
>>>>> device          scbus
>>>>> device          cd
>>>>> device          pass
>>>>> device          da
>>>>> device          random
>>>>> device          pty
>>>>> device          md
>>>>>
>>>>> the dmesg entries for cd0 are:
>>>>>
>>>>> cam_periph_alloc: attempt to re-allocate valid device cd0 rejected
>>>>> cdasync: Unable to attach new device due to status 0x6
>>>>> cd0 at ata2 bus 0 scbus8 target 0 lun 0
>>>>> cd0:<HL-DT-ST DVDRAM GSA-H10N JL12>  Removable CD-ROM SCSI-0 device
>>>>> cd0: 3.300MB/s transfers
>>>>> cd0: cd present [2149024 x 2048 byte records]
>>>> Aha, that's it. It is atapicam's lie. atapicam is one of a things that
>>>> will go away during migration to CAM ATA. If you remove `device
>>>> atapicam`, but add `options ATA_CAM` instead, CAM will manage that bus
>>>> directly and report it as ATA.
>>>
>>> thanks for the hint. i also read the following in the ahci(4) man page:
>>>
>>> "     Driver features include support for Serial ATA and ATAPI devices, ..."
>>>
>>> ...does that mean that my DVD drive can also attach to the ahci driver?
>>
>> If it is SATA and you connect it to AHCI controller -- yes, it will work
>> with ahci(4) controller driver.
>
> no it's PATA. thanks for the help. i'll remove atapicam from my kernel conf
> and add ATA_CAM.
>
> maybe we could have a list of supported controllers in the ahci(4) man page.

We could do it (you can find list of known inside ahci.c), but unlike 
others, AHCI is an open specification, so that list will never be 
complete. Also on some chipsets AHCI support also depends on BIOS, so 
specification of just a chipset revisions can be inaccurate.

> will the ataahci driver also dissapear soon?

Yes, at least disabled. Same as most part of the atasiliconimage and the 
atamarvell.

-- 
Alexander Motin