Re: ASUS M2A-VM (SB600): AHCI setting in BIOS; AHCI and UDMA and `softreset failed (device not ready)`

[Robert Hancock <hancockrwd@gmail.com>]

On 01/16/2012 04:03 AM, Paul Menzel wrote:
> Dear Linux folks,
>
>
> please keep my address in CC list since I am not subscribed.
>
> I am sorry for asking again for clarification for messages from Linux in
> `/var/log/kern.log`.
>
>          Jan 13 02:13:15 joe kernel: [    1.477687] ahci 0000:00:12.0: version 3.0
>
> Whose version is this indication? Looking at the Wikipedia article [1]
> and the AHCI link therein [2], specification numbers seem to 1.x.

That's the driver version (which is of questionable usefulness), not the 
AHCI version.

>
> Additionally having a sentence there, like `chipset support AHCI version
> 3.0` would be “better” in my opinion. If you agree, I can send a patch.
>
>          Jan 13 02:13:15 joe kernel: [    1.477718] ahci 0000:00:12.0: PCI INT A ->  GSI 22 (level, low) ->  IRQ 22
>          Jan 13 02:13:15 joe kernel: [    1.477748] ahci 0000:00:12.0: ASUS M2A-VM: enabling 64bit DMA
>          Jan 13 02:13:15 joe kernel: [    1.477981] ahci 0000:00:12.0: AHCI 0001.0100 32 slots 4 ports 3 Gbps 0xf impl SATA mode

That line indicates AHCI 1.1.

>          Jan 13 02:13:15 joe kernel: [    1.477986] ahci 0000:00:12.0: flags: 64bit ncq sntf ilck pm led clo pmp pio slum part ccc
>          Jan 13 02:13:15 joe kernel: [    1.479052] scsi2 : ahci
>          Jan 13 02:13:15 joe kernel: [    1.479142] scsi3 : ahci
>          Jan 13 02:13:15 joe kernel: [    1.479220] scsi4 : ahci
>          Jan 13 02:13:15 joe kernel: [    1.479296] scsi5 : ahci
>          Jan 13 02:13:15 joe kernel: [    1.479434] ata3: SATA max UDMA/133 abar m1024@0xfe02f000 port 0xfe02f100 irq 22
>          Jan 13 02:13:15 joe kernel: [    1.479438] ata4: SATA max UDMA/133 abar m1024@0xfe02f000 port 0xfe02f180 irq 22
>          Jan 13 02:13:15 joe kernel: [    1.479442] ata5: SATA max UDMA/133 abar m1024@0xfe02f000 port 0xfe02f200 irq 22
>          Jan 13 02:13:15 joe kernel: [    1.479447] ata6: SATA max UDMA/133 abar m1024@0xfe02f000 port 0xfe02f280 irq 22
>
> So, AHCI and UDMA are unrelated? (Just to be sure once and for all. ;-))

Basically, yes

>
>          Jan 13 02:13:15 joe kernel: [    1.640626] ata1.00: ATAPI: TOSHIBA DVD-ROM SD-M1712, 1004, max UDMA/33
>          Jan 13 02:13:15 joe kernel: [    1.656502] ata1.00: configured for UDMA/33
>          Jan 13 02:13:15 joe kernel: [    1.796038] ata6: SATA link down (SStatus 0 SControl 300)
>          Jan 13 02:13:15 joe kernel: [    1.796076] ata5: SATA link down (SStatus 0 SControl 300)
>          Jan 13 02:13:15 joe kernel: [    1.796132] ata4: SATA link down (SStatus 0 SControl 300)
>          Jan 13 02:13:15 joe kernel: [    1.808018] usb 2-1: new low speed USB device number 2 using ohci_hcd
>          Jan 13 02:13:15 joe kernel: [    1.968017] ata3: softreset failed (device not ready)
>          Jan 13 02:13:15 joe kernel: [    1.968063] ata3: applying PMP SRST workaround and retrying
>
> This seems to be related to a hardware bug in the SB600 chipset [3]. The
> discussion [4] indicates that setting `CONFIG_SATA_PMP=n` fixes this
> issue. Although reading the option description [5] I do not understand
> what it does and if it is advisable to disable it.

Turning that off would prevent you from using any SATA port multipliers 
(usually found in things like multi-drive external enclosures). The 
kernel already worked around the issue so I don't think you need to 
worry about this.

>
>          […]
>
>          Jan 13 02:13:15 joe kernel: [    2.140032] ata3: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
>          Jan 13 02:13:15 joe kernel: [    2.146085] ata3.00: ATA-8: WDC WD20EARS-60MVWB0, 51.0AB51, max UDMA/100
>          Jan 13 02:13:15 joe kernel: [    2.146088] ata3.00: 3907029168 sectors, multi 1: LBA48 NCQ (depth 31/32), AA
>          Jan 13 02:13:15 joe kernel: [    2.146094] ata3.00: SB600 AHCI: limiting to 255 sectors per cmd
>          Jan 13 02:13:15 joe kernel: [    2.152134] ata3.00: SB600 AHCI: limiting to 255 sectors per cmd
>          Jan 13 02:13:15 joe kernel: [    2.152138] ata3.00: configured for UDMA/100
>
> This seems to be related to [6], although the chipset SB600 seems to
> support 3.0 Gbps instead of only 1.5 Gbps with the VIA K8T890. The drive
> only seems to propagate UDMA/100 which is fine.

The UDMA speed setting really doesn't matter with SATA devices. The mode 
selection is really just a vestige of PATA and doesn't actually affect 
anything unless there is actually a physical PATA link somewhere in the 
chain (which there almost certainly is not).

>
> My main question is, that in the BIOS »Onboard SATA Type« is set to »IDE
> Controller« by default. Is that true? But looking at the output AHCI
> seems to be used nevertheless so the BIOS setting seems to be ignored or
> does not make any difference to setting it to »AHCI«. So would changing
> the option to »AHCI« give any benefit? Would it get rid of the
> `softreset failed (device not ready)` too?

I'm not sure what that mode selection would actually affect on this 
board. On Intel chipsets you can only use AHCI if the BIOS is in the 
correct mode since the AHCI device isn't exposed otherwise, but on most 
other chipsets, but most others expose the AHCI capabilities all the 
time and so it doesn't really matter. Maybe it changes the device ID and 
therefore affects which driver Windows will load or let you use for the 
controller but I doubt it really makes much difference in Linux. (I 
doubt it would affect that SRST message either, that seems to just be a 
bug in that chipset.)

>
>
> Thanks,
>
> Paul
>
>
> [1] https://en.wikipedia.org/wiki/Advanced_Host_Controller_Interface
> [2] http://www.intel.com/technology/serialata/ahci.htm
> [3] https://bugzilla.redhat.com/show_bug.cgi?id=468800
> [4] http://www.linuxquestions.org/questions/linux-kernel-70/ata4-softreset-failed-device-not-ready-865155/
> [5] http://cateee.net/lkddb/web-lkddb/SATA_PMP.html
> [6] http://www.spinics.net/lists/linux-ide/msg41224.html