Project

General

Profile

Actions

Bug #3556

open

Problem w/ 3 TB 4k sector USB disk on oi_151a7

Added by Reginald Beardsley over 10 years ago. Updated about 6 years ago.

Status:
New
Priority:
Normal
Assignee:
Category:
-
Target version:
-
Start date:
2013-02-13
Due date:
% Done:

0%

Estimated time:
Difficulty:
Medium
Tags:
needs-triage

Description

format(1m) reports disk (Toshiba Canvio) as 4kB sector and creates EFI label per EFI spec at 0x1000. prtvtoc & ftmhard correctly access the EFI label at the 4kB boundaries.

However, dd(1) is unable to access the s0 slice unless a copy of the first 512 bytes of LBA 1 is placed at absolute offset 0x200 w/ the first 512 bytes of LBA 2 at 0x400. (note: failed to test just copying the slice data)

Adding an entry in sd-config-list configured the device as:

un_sys_blocksize = 0x200
un_tgt_blocksize = 0x1000
un_phy_blocksize = 0x1000

but

dd if=/dev/zero of=/dev/rdsk/c5t0d0s0 bs=64k count=1000

produces ~1.6 MB/s which is slower than seen w/ the default configuration. The s0 write produces no messages. Attempting to create a pool on c5t0d0s0 fails w/ a message in messages "Request rejected: too large for CDB".

W/ default configuration:

un_sys_blocksize = 0x1000
un_tgt_blocksize = 0x1000
un_phy_blocksize = 0x200

dd reports 27 MB/s and pool creation succeeds, but the system logs read-modify-write messages. One message for writing 100000 64kB blocks to s0 and lots of them for writing to the pool. Write to pool was at 5.6 MB/s.

System (HP N40L) reports 199 MB/s writing to 3 disk RAIDZ on SATA disks
(ST2000DM001) for which sd(7) has configured:

un_sys_blocksize = 0x200
un_tgt_blocksize = 0x200
un_phy_blocksize = 0x1000

but w/o requiring an sd-config-list entry in sd.conf.

Actions #1

Updated by Ken Mays over 10 years ago

  • Assignee set to OI illumos
Actions #2

Updated by Ken Mays about 6 years ago

We can close this one?

Actions

Also available in: Atom PDF