Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 16 Aug 2005 18:57:31 -0400
From:      jason <jason@ec.rr.com>
To:        Mark Kane <mark@mkproductions.org>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: How to Force UDMA100 Mode on Boot?
Message-ID:  <43026F5B.4010705@ec.rr.com>
In-Reply-To: <43014383.1040400@mkproductions.org>
References:  <430128F1.9@mkproductions.org> <43013444.8080808@ec.rr.com> <43014383.1040400@mkproductions.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Mark Kane wrote:

> Hi, thanks for the response. The thread somehow got broken up due to 
> some subject formatting (there was a space inserted somehow). Here are 
> the threads:
>
> http://lists.freebsd.org/mailman/htdig/freebsd-questions/2005-August/095212.html 
>
> http://lists.freebsd.org/mailman/htdig/freebsd-questions/2005-August/095227.html 
>
> http://lists.freebsd.org/mailman/htdig/freebsd-questions/2005-August/095335.html 
>
>
> I have 5 hard drives, and when copying data between them in certain 
> configurations (such as drive placement) I get READ and WRITE wouldn 
> errors. All the cables are brand new, as are two of the hard drives.
>
> Similar errors happened on the last board I had. I had the same model 
> (Giga-Byte K8NS Pro) a couple months ago that had other issues in 
> addition to this. I sent it to the factory for a RMA, and a brand new 
> one came back. Before I sent it in, I was using Windows XP and it 
> would automatically downgrade it to 100 so I wouldn't see any errors. 
> When I switched over to FreeBSD and it tried to operate in 133 mode, I 
> got errors instead of the OS trying to hide it.
>
> Note that throughout this whole problem I never got a "FAILURE" 
> message until today, except that is only on one drive, and one that I 
> think is in fact going bad.
>
> It's gotta be something with the controller.
>
> I can't get you the dmesg info right now since I'm doing a scan on 
> that one hard drive that I think is failing. But it is an nForce 3 
> chipset on a Giga-Byte K8NS Pro motherboard.
>
> I would really like to solve the DMA problems, but if not I think the 
> easiest is trying to downgrade it to UDMA100 on boot, which is what 
> this post is about.
>
> Thanks
>
> -Mark
>
> jason wrote:
>
>> Mark Kane wrote:
>>
>>> Hi everyone. I've had a thread going here on the lists about DMA 
>>> problems in 133 mode. In a nutshell, some drives give DMA_WRITE and 
>>> DMA_READ errors when in 133 mode with certain configurations, 
>>> however don't have any problems in 100 or 66 mode. After looking in 
>>> to many solutions I think I'm just going to run it at 100, since 
>>> from my research the benefit isn't that noticeable.
>>>
>>> I know about atacontrol to set it manually, but I'd like to set 
>>> UDMA100 mode automatically on boot since I have 5 hard drives. I 
>>> also know the sysctl hw.ata.ata_dma, but that doesn't say anything 
>>> about using 100 vs 133.
>>>
>>> Thanks in advance.
>>>
>>> -Mark
>>>
>> Sounds like a cable issue, but could it be a buggy bios?  How about 
>> some information since I did not see your previous postings.
>>
>>
>>  > dmesg|grep DMA
>> atapci1: <nVidia nForce2 UDMA133 controller> port 
>> 0xf000-0xf00f,0x376,0x170-0x177,0x3f6,0x1f0-0x1f7 at device 9.0 on pci0
>> ad0: 38172MB <MAXTOR 6L040J2/A93.0500> [77557/16/63] at ata0-master 
>> UDMA133
>>
>>
>> Also if I don't have a cd in the drive I get "acd0: CDRW <LITE-ON 
>> LTR-40125S/ZS0K> at ata1-master PIO4" for acd0.  If there is a disc 
>> in the drive it is set to UDMA66 at boot up.   Or first use  if it 
>> was not in at boot.
>
>
I checked the links and saw no dmesg or drive models for all drives.  I
can tell you UDMA 133 was not an official spec, a least at first.  It
was a maxtor only thing, and not all chipsets handled it.  Basically
maxtor tightened the timings on the ide cable signals to squence extra
data.  If you have new and old drives, plus different brands I would not
expect to run at 133 speeds.  I don't care to look it up now, but you
may want to find out if any other drive manufactures support the 133
spec today.  Also there are load and signal degregation issues with to
think about with longer cables.  If you have a full tower case with the
longest cables you can buy you won't get max speeds even at the 133
setting.  If you know someone whos works in a pc rpair shop you could
ask them what cable length does to drive speeds.  I am told if you want
to copy drives for customers you want to get a good cable, but only with
1 device per cable and get it as short as possible.  A 2 inch cable will
dramatically shorten the time to copy whole drives compared to a 16 inch
cable.




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?43026F5B.4010705>