ATA_STATIC_ID and AHCI bug or /dev/hands ?
gavin at FreeBSD.org
Tue May 26 13:46:01 UTC 2009
On Sun, 2009-05-24 at 21:41 +0300, triosoft at triosoft.com.ua wrote:
> I found some odd behavior of ata device numbering. It seems, that there
> is no effect of option ATA_STATIC_ID in CURRENT.
> What I have:
> supermicro server with 4 sata ports onto Intel ICH7 SATA300 controller
> recent CURRENT
> sata mode in BIOS is Enhanced
> ahci support in BIOS in Enabled
> option ATA_STATIC_ID in KERNCONF
> So, if I boot with 4 HDD connected to sata i found ad4,ad6,ad8,ad10 as
> disks. but when I detach disk, which "was" ad8 in last case, and then
> reboot - I see ad4,ad6,ad8! and not ad4,ad6,ad10! atacontrol list
> doesn't show an empty ata channel. Only channels with HDDs connected. I
> check 6.x 7.x on the same server with _the same configuration in bios_ -
> and all works as suspected - there are empty ata channels, there are
> really static device numbers. when I set AHCI to Disabled I have just
> two ata channels (with AHCI - I have four, one for each sata disk) and
> with or without HDDs connected I see empty channels and have really
> static dev nums in the same CURRENT.
> So my question is - is it my fault? does I miss something? Or it is a bug?
This sounds like a bug to me. I think you're probably best submitting a
PR, and please include (or link to) a verbose dmesg from both 7.x and
HEAD, and the output of "atacontrol list" from both.
More information about the freebsd-current