Re: libata interface fatal error

[Florian Effenberger <florian@effenberger.org>]

Hi Tejun,

> Okay, ich8.  I don't think the chipset is at fault here and you have a
> lot of disks.  My primary suspect is power supply problem but things
> like this are hard to prove.  With the merged speed down fix, libata
> will do the right thing after a few errors, so ignoring the problem
> wouldn't be a too bad idea.  If you're curious, you can try to connect
> drives to different SATA ports and power lanes and see whether errors
> follow the disk, port or power lane.

exactly, should be four disks in the machine.

What power supply would you recommend for this type of disks? I think we 
got a 450W Enermax, IIRC.

What do you mean by "merged speed down fix"? Is your fix for the speed 
down logic implemented in the current kernel, so I don't have to patch 
anymore (except when I want to force 1.5Gbps right from the beginning)?

All SATA parts are used, so reconnecting is not an option. But I can try 
to switch the power supply (lanes).

Thanks for all your kind help, that is much appreciated!
Florian