I also tried to use "libata.force=1:1.5G" to artificially slow down the wire speed and this seems to make those errors appear less often, while also slowing down the disk noticeably. ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300) ata1.00: failed command: WRITE FPDMA QUEUED ata1.00: exception Emask 0x0 SAct 0x7fffffff SErr 0x0 action 0圆 frozen ata1.00: NCQ disabled due to excessive errors ata1.00: device reported invalid CHS sector 0 I now compiled linux-drm-nightly as of today (git 89271faca1 from ) and the situation has improved a little: when writing to the disk, I get the below output about once a minute, and a pause and (probably) a bus reset, but no more crashes and unusable disks until reboot. gvfs-udisks2-vo: segfault at 8 ip 000000000040fb14 sp 00007fffed821a70 error 4 in gvfs-udisks2-volume-monitorĪfter this the disk is totally inaccessible. show_signal_msg: 168 callbacks suppressed blk_update_request: I/O error, dev sda, sector 0 sd 0:0:0:0: FAILED Result: hostbyte=DID_BAD_TARGET driverbyte=DRIVER_OK I use this instead of the stock Ubuntu kernel due to buggy Haswell PM support in everything ] dump_stack+0x45/0x57 I formatted this with LUKS + ext4 to use as an archive disk under Ubuntu 14.04 with an updated mainline kernel (3.19.0-031900-generic #201502091451). I have a brand new Seagate 8TB archive disk: Log of drives interfering with other disks on same controller Logs of the session a day after Comment #53ĭmesg and smartctl after letting the drive 'settle'ĭmesg from boot to drive failure after rsyncing ~250 GB of data Bisect to locate error for Seagate ST8000AS0002 archive driveīisect 4.1 to 4.2 to locate 2nd Seagate SMR bugĭmesg of the session where I'm experiencing problems on 3.14, ncq disabled
0 kommentar(er)
