Re: kern/151608: [ata] FreeBSD doesn't detect Sandforce-based SSD

Thomas E. Spanjaard wrote:
> On 10/24/2010 21:18, Alexander Motin wrote:
>> Thomas E. Spanjaard wrote:
>>> In an attempt to get more information about this issue, I'd like to ask
>>> people on freebsd-current if they're using any Sandforce-based SSDs with
>>> FreeBSD. So far, it appears not a lot of people do, making it hard to
>>> debug this issue (http://www.freebsd.org/cgi/query-pr.cgi?pr=151608).
>> MCP5x SATA are not the best controllers for compatibility testing. They
>> have enough problems on their own, even without Sandforce.
>>
>> When you tested Marvell, have you tried to use mvs(4) driver?
> 
> No, I only tried to use ata(4). I tried using mvs(4) now, and that works!
> 
> mvs0: <Marvell 88SX6081 SATA controller> port 0x4000-0x40ff mem
> 0xb0100000-0xb01fffff irq 11 at device 2.0 on pci1
> mvs0: Gen-II, 8 3Gbps ports, Port Multiplier supported
> mvsch0: <Marvell SATA channel> at channel 0 on mvs0
> mvsch1: <Marvell SATA channel> at channel 1 on mvs0
> mvsch2: <Marvell SATA channel> at channel 2 on mvs0
> mvsch3: <Marvell SATA channel> at channel 3 on mvs0
> mvsch4: <Marvell SATA channel> at channel 4 on mvs0
> mvsch5: <Marvell SATA channel> at channel 5 on mvs0
> mvsch6: <Marvell SATA channel> at channel 6 on mvs0
> mvsch7: <Marvell SATA channel> at channel 7 on mvs0
> 
> ada0 at mvsch0 bus 0 scbus0 target 0 lun 0
> ada0: <OCZ-AGILITY2 3.5 1.11> ATA-8 SATA 2.x device
> ada0: 300.000MB/s transfers (SATA 2.x, UDMA6, PIO 2048bytes)
> ada0: Command Queueing enabled
> ada0: 85857MB (175836528 512 byte sectors: 16H 63S/T 16383C)
> 
> In any case, it seems this is not squarely to blame on the nVidia SATA
> controller, because ata(4) couldn't detect it either when on the Marvell
> one. So the bug in ata(4) remains.

I would agree if it was the first case.

I am not saying that ata(4) perfect. ata(4) is a big thing and Marvell
part is definitely not the best part of it. That's why there is several
new drivers: ahci(4), siis(4), mvs(4), which together with ATA_CAM
options should substitute most of it.

-- 
Alexander Motin