Re: Problem with SiI3114 SATA RAID using 5.3-BETA1-20040823

Harald Schmalzbauer wrote:
> Am Donnerstag, 9. September 2004 04:03 schrieb Doug White:
> 
>>On Tue, 31 Aug 2004, Kenneth Stailey wrote:
>>
>>>I'm trying to upgrade a TYAN Tiger K8WS S2875S from FreeBSD-5.2.1-p8 to
>>>5.3-BETA1-20040823.
>>>
>>>It fails because instead of seeing the SiI3114 as ar0 (RAID-1) it sees
>>>two separate disks:
>>
>>[...]
>>
>>Can you post the output of boot -v, or at least the part where its trying
>>to detect the metadata type?
> 
> 
> If you're interested in the result for the sil0649(0680) chipset you can find 
> them here (RELENG_5 from today compared to 5.2.1-RELEASE):
> 
> http://www.schmalzbauer.de/statics/sil0680-v-boot-5.2.1-live.txt
> http://www.schmalzbauer.de/statics/sil0680-v-boot-5.3BETA3-09.09.04.txt
> 
> 
> Note the "ar: LSI check1 failed" line
> It's broken since June the 25th when rev. 1.27 for ata-raid.h was commited. 
> Reverting to sources before that date at least atacontrol created arrays were 
> detected, even nor rebuild was possible, but that's another story. For now, 
> upgrading to 5.3 is not possible if you have a system installed on ar0 with 
> the sil chips.

Depends on what BIOS and metadata you have on there, simple as that.
Now, maybe we should make ata-raid go through all the types it knows on 
all controllers, but that could really mess up things when disks are 
moved around between controllers.
I'll think about a workable solution but its fairly long down the TODO 
list..

-Søren