Re: ata2: EH in ADMA mode, notifier 0x0 notifier_error 0x0 gen_ctl

[ROBERT HANCOCK <hancockr@shaw.ca>]

Christian wrote:
> 
> FYI:
> My system config is one 400GB disk at sda, and two 250GB disks on a dmraid 
> nvidia-fakeraid set of sdb and sdc.
> 
> My kernel message buffer gets quickly overrun by a flood of these error 
> messages:
> 
> Nov 17 22:48:12 ubuntu kernel: [  119.566540] attempt to access beyond end of 
> device
> Nov 17 22:48:12 ubuntu kernel: [  119.566602] sdb: rw=0, want=976784000, 
> limit=488397168

This seems like some other issue. For some reason the kernel is trying to access something way out at about 465GB on the /dev/sdb device..

> p.s:
> Why does the kernel report a queue depth of 31/32, but hdparm says its 32? Is 
> this correct?
> 
> ata1.00: ATA-7, max UDMA7, 781422768 sectors: LBA48 NCQ (depth 31/32)

The drive supports 32 but libata reserves one queue entry for its own use, therefore the actual queue depth in use is 31.