Hi, Thank you for your reply. On Mon, Jul 14, 2014 at 5:36 PM, Tejun Heo <tj@xxxxxxxxxx> wrote: > So, the device initially reports that it doesn't support LBA48 and > then later changes its mind later on for some reason. That's weird. > I wonder whether the controller is meddling with the IDENTIFY data. > AFAICS, there isn't anything in ATA spec which would make a device > change whether it reports 48 bit support or not. > > Can you please try the followings? This doesn't happen every-time i boot, only sometimes (about 50/50). Once i booted and the disk was disabled, i created dmesg1.nodisk > 1. After boot is complete, re-trigger device attach by doing > > echo - - - > /sys/class/scsi_host/hostN/scan Partial dmesg2.reset (i just wanted to see what this output to dmesg, i didn't reboot) > 2. Boot with libata.force=dump_id and report full dmesg after boot. dmesg3.dump_id > 3. Boot with libata.force=nohrst and see whether anything changes. dmesg4.nohrst > Thanks. Thank you for your time. Both dump_id and nohrst detected the drive the first time i used them. Should i keep trying one or the other until the disk is dropped again (or not)? Should i specify both boot options simultaneously? Kind regards, Nuno Magalhães
Attachment:
dmesg1.nodisk
Description: Binary data
Attachment:
dmesg2.reset
Description: Binary data
Attachment:
dmesg3.dump_id
Description: Binary data
Attachment:
dmesg4.nohrst
Description: Binary data
