On Sun Aug 18 12:42:54 EDT 2013, [email protected] wrote:
> i had a small panic on Friday, when a venti+fossil fs (a supermicro
> 5015A-EHF-D525 box) had an unexplained reboot followed by what appeared to
> be a disk problem. on inspection -- by IL booting it from an un-retired
> kenfs -- i noticed that the sd controller position had changed from sdE0 to
> sdC0.
>
> suspecting the bios setup, i checked the sata settings in bios, and the
> first controller was set to ata rather than ahci.
i suspect that this is caused not by a bios change, but by some subtile
confusion.
there are two different ways a drive can show up as sdE0. first, as an ide
driver
that is showing up in pci space, and not the blindly-probed ide addresses and
second as an ahci device. the hardware supports both modes, so with the right
combination of bios settings and drivers, the same hardware could be used as
either ahci or ide without fiddling bios. this may have to do with which vids
and
dids are in your version of the ahci driver.
we have about 10 of these machines running around running standard 9atom nix
kernels with bios selected to be compatable/ahci. once booted, the
configuration
looks like the following:
consoled; aux/dmi -t 1
1: type sysinfo 1 len 27 handle 1
mfg Supermicro
product X7SPA-HF
version 1234567890
serial 1234567890
uuid 534D434900026490250064902500D660
waketype powersw (0x6)
sku To Be Filled By O.E.M.
family <nil>
consoled; pci|grep disk
0.31.2: disk 01.06.01 8086/2922 14 0:0000b481 16 1:0000c001 16 2:0000bc01 16
3:0000b881 16 4:0000b801 32 5:febfb000 2048
consoled; cat /dev/sdctl
sdE ahci ahci port 0xfffffe00febfb000: 64a ncq ntf ss alp led clo pmb slum
pslum coal ems xs alhd xonly smb elmt iss 2 ncs 31 np 6 ghc 80000002 isr 0 pi
3f 0-5 ver 10200
i hope that helps. it may be just a matter of commenting out a few lines in the
ide driver, though one could likely solve the issue with a bios change as well.
- erik
