Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 16 Nov 2011 11:14:52 +0100
From:      Andrea Venturoli <ml@netfence.it>
To:        freebsd-questions@freebsd.org
Subject:   ATA trouble again
Message-ID:  <4EC38D1C.30205@netfence.it>

next in thread | raw e-mail | index | archive | help
Hello.

This is a follow-up of a previous thread, which can be found here:

> http://lists.freebsd.org/pipermail/freebsd-questions/2011-July/232257.html



I finally have two working WD Green drives: they have run severel times 
for more than a few days continuosly and I've run their full diags a few 
times over a span of some weeks, so I think I can be sure of that.
Still I experience lot of troubles; however, trying different boxes, I 
found out that these depend heavily on the controller I use.



A) A-Bit KV7 (featuring a VIA KT600) on 7.3 and 8.2/i386
With only one disk, I get some NID_NOT_FOUND errors, but I could almost 
say they can be ignored.
With two disks, the NID_NOT_FOUND errors come frequently; rsync will 
work, but a "diff" will sooner or later fail.

I tried Linux on this box: it frequently complains about errors and 
resets the ATA channels; this possibly will result in performance 
penalties, but it seems to work.

Since then this box is dead, so I cannot do other tests.



B) BioStar A785G3 with 8.1/i386:
> # pciconf -lv
> atapci0@pci0:0:17:0:    class=0x010601 card=0x43911002 chip=0x43911002 rev=0x00 hdr=0x00
>     vendor     = 'ATI Technologies Inc. / Advanced Micro Devices, Inc.'
>     device     = 'SB700 SATA Controller [AHCI mode]'
>     class      = mass storage
>     subclass   = SATA

This works flawlessly.



C) Intel SC5400BRP with 7.3/i386:
> atapci2@pci0:0:31:2:    class=0x01018f card=0x34768086 chip=0x26808086 rev=0x09 hdr=0x00
>     vendor     = 'Intel Corporation'
>     device     = '631xESB/632xESB/3100 Serial ATA Storage Controller'
>     class      = mass storage
>     subclass   = ATA

No way this will work: using two disk gives gobs of errors, until the 
whole box crashes.
With only one disk it appears to work for a while, but will eventually 
panic over a few day.
This same controller used to work with a single Hitachi drive.
With two Seagates drives, I get READ_DMA48/WRITE_DMA48 timeouts, the 
disks will hang for minutes, but then they will start again; this does 
not happen as soon as I start using them, but will happen every some 
minutes after a few hours of usage.



D) Intel SC5600BRP with 7.3/amd64:
> atapci0@pci0:0:31:2:    class=0x01018f card=0x34dc8086 chip=0x3a208086 rev=0x00 hdr=0x00
>     vendor     = 'Intel Corporation'
>     device     = 'SATA2(4Port2) (ICH10 Family)'
>     class      = mass storage
>     subclass   = ATA
> atapci1@pci0:0:31:5:    class=0x010185 card=0x34dc8086 chip=0x3a268086 rev=0x00 hdr=0x00
>     vendor     = 'Intel Corporation'
>     device     = 'SATA2(2Port2) (ICH10 Family)'
>     class      = mass storage
>     subclass   = ATA

This works flawlessly, with every HD I could try (one or two).



E) Promise PCI controller with 7.3/i386 and /amd64:
> atapci0@pci0:7:2:0:     class=0x018000 card=0x3d17105a chip=0x3d17105a rev=0x02 hdr=0x00
>     vendor     = 'Promise Technology Inc'
>     device     = 'SATA 300 TX4 Controller (PDC40718-GP)'
>     class      = mass storage

This will give few errors if not none at all; however the box will 
sometimes reboot or just hang (phisical reset required).
This happens on two different box with both WD Green and Seagate drives.




So: altough a bit flaky, WD drives are not the main problem.
I'll still appreciate some help WRT the third box (Intel SC5400BRP with 
7.3/i386).
Different BIOS settings did not help with WD drives, but I still have to 
test them with Seagate's.
Possibly an OS upgrade would help? 8.x also has an AHCI driver that 7.3 
is missing.
I don't care that much for top of the chart performance, I just want 
this thing to work decently.


  bye & Thanks
	av.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4EC38D1C.30205>