i386/104719: Seagate ST3802110A errors/delays when using PIO4 or
UDMA33
Matt Mills
matt_mills at btopenworld.com
Mon Oct 23 12:10:12 PDT 2006
>Number: 104719
>Category: i386
>Synopsis: Seagate ST3802110A errors/delays when using PIO4 or UDMA33
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: freebsd-i386
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Mon Oct 23 19:10:10 GMT 2006
>Closed-Date:
>Last-Modified:
>Originator: Matt Mills
>Release: 6.2-PRERELEASE
>Organization:
N/A
>Environment:
FreeBSD titan 6.2-PRERELEASE FreeBSD 6.2-PRERELEASE #20: Fri Oct 20 18:34:23 BST 2006 matt at titan:/usr/src/sys/i386/compile/titan i386
>Description:
This is tested on both 6.1-STABLE and 6.2-PRERELEASE.
I installed a brand new ST3802110A hard drive and moved an existing installation onto it. The motherboard is an HP Asus CUR-DLS with dual P3 933 and SMP enabled:
acpi0: <HP HWPC20F>
atapci1: <ServerWorks ROSB4 UDMA33 controller>
ad0: 76319MB <Seagate ST3802110A 3.AAJ>
When the drive was running at UDMA33, almost every time I entered a command and hit enter there was a delay and a distinctive humming sound before the command was processed. It sounded almost like the drive kept powering down until it was accessed and then powered up again. I tried a few disk intensive commands like pkgversion, portsclean and kernel compilation which worked fine, but again, shortly after completion you would hear the drive hum quieten. I also saw a couple of READ/WRITE_DMA errors but I do not have these available any longer.
I tried disabling DMA which forced back to PIO4 mode. This resulted in a very occasional:
kernel: ad0: TIMEOUT - WRITE_MUL retrying (1 retry left) LBA=4615355
kernel: ad0: TIMEOUT - READ_MUL retrying (1 retry left) LBA=83786751
Even during times of relative inactivity. At one stage (when running at PIO4) I returned to the computer and it had completely locked up, so I had to physically power down.
I tried disabling ACPI and even APIC to no avail.
>How-To-Repeat:
Presumably you could repeat this by running 6.1-STABLE or 6.2-PRERELEASE on a Seagate ST3802110A connected to an onboard IDE connector of an HP Asus CUR-DLS motherboard.
>Fix:
I installed a PCI IDE ATA100 controller and connected the problematic drive:
atapci0: <CMD 649 UDMA100 controller>
The hums and associated delays have now disappeared and 48+ hours have passed without a problem. Furthermore, the drive is now running at the intended UDMA100 and performing well.
More details can be provided if necessary, though I would rather perform as little reproduction as possible since uptime of the machine is quite critical.
>Release-Note:
>Audit-Trail:
>Unformatted:
More information about the freebsd-i386
mailing list