LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Hardware (https://www.linuxquestions.org/questions/linux-hardware-18/)
-   -   Hard drives and DMA (https://www.linuxquestions.org/questions/linux-hardware-18/hard-drives-and-dma-222690/)

Hammett 08-26-2004 10:54 AM

Hard drives and DMA
 
Hi all!!

This is driving me nuts....really....
I have an ASUS P4S8X-X mobo with 2 Maxtor HDD (80 and 120GB) both in the same IDE channel (hda and hdb).

The IDE cable i had was a normal cable (ATA 66 or 100) and i decided to change it for a rounded one (ATA 133).

Now the problem is that if i use the rounded cable, i have a lot of problems with my hard drives (basically the dma is not working at all) with these errors (taken for /var/log/messages)

Aug 26 17:26:17 Hammett kernel: hda: dma_intr: status=0x51 { DriveReady SeekComplete Error }
Aug 26 17:26:17 Hammett kernel: hda: dma_intr: error=0x84 { DriveStatusError BadCRC }
Aug 26 17:27:09 Hammett kernel: hda: DMA disabled

Aug 26 17:27:08 Hammett kernel: hdb: dma_intr: status=0x51 { DriveReady SeekComplete Error }
Aug 26 17:27:08 Hammett kernel: hdb: dma_intr: error=0x84 { DriveStatusError BadCRC }

I have the correct IDE controler loaded in the kernel (SiS 5513, I made myself sure of that)

But...if i use the normal IDE cable, i have no problem at all!!! DMA is set correctly i have no errors on /var/log/messages. Well, the only problem is that i cannot burn CD at max speed cause cable is ATA 66 or 100 (don't remember) so the speed of burning is 12x instead of 52x.

I checked ASUS website, and my mobo supports IDE UDMA 133.

So, my question is....
What the hell is happening here?? It's mobo's fault? (maybe upgrada the bios) or it's the rounded cables fault?? What do IDE cables have on disabling DMA "on the fly" and those errors?? Should I try to buy new ATA 133 IDE cables and try again?

Any help is more than appretiated cause i don't understand anything of all this.

kilgoretrout 08-26-2004 12:13 PM

First what kind of hard drives do you have, either ATA 100 or 133 would be my guess. Hooking a 133 cable to an ATA 100 hard drive won't do anything to increase the speed; it will still run at ATA 100 speeds.
Second, rounded cables look nice and all but I've seen many people have problems with them. You have to suspect the cables since when you go back to the old ones they work fine. IIRC ATA 133 and 100(Ultra ATA) both use the same cable, an 80 wire cable instead of the standard 40 wire ide cable. There are only 40 wires used by ultra ata but the additional wires are necessary and placed between the conducting wires to prevent "cross talking" which is a problem at ultra ata speeds.
You can easily tell the difference between a standard 40 ide cable and an 80 wire ultra ata ide cable; on a flat ribbon cable the standard cable bumps look much coarser than the ultra ide cable bumps. CD drives normally have the standard cable since they can't operate past ata 66 speeds anyway. Hooking an ultra ata cable to a cd drive is a waste of money as they are usually more expensive. Hooking an ultra ata hard drive(100 or 133) to a standard 40 wire ide cable will cause it to operate at ata 66 speed and should be avoided.

I think you need to step back and figure out exactly what kind of hardware you have. Are your hard drives ATA 66, 100 or 133? Is your ribbon cable a standard 40 conductor or an ultra 80 conductor? If your drives are either ata 100 or 133 and you have the ultra 80 conductor ribbon cable, going to the round cables won't give you any performance increase and appear to be the source of your problems.

Hammett 08-26-2004 12:27 PM

kilgoretrout,

First of all thanks for the post.
Both HD are ATA 133 and the flat ribbons are 40 cables each. The rounded i bought are 80.
Actually, now i have one rounded for DVD and burner and flat for hard drives. When transfering from HD to HD i get transfer around 30-40MB/s, that i think is more than acceptable considering they share the IDE channel.
As I never have seen a flat 80 wire IDE ribbon, i cannot tell if it's 80 or 40. On the ribbon nothing is printed that can give me a clue (but i thing it's 40 due to the fact that's very thin).
If the transfers HD to HD do you think are correct (if they are too low, just tell me) i won't change anything, if they're too low, i'll buy a flat 80 wire ide ribbon.

Crashed_Again 08-26-2004 03:41 PM

Hammett, I had a similar issue like this a long time ago. The problem in my case was the jumper settings on my hard drives. Double check that they are correct and if they are set to cable select change them to master and slave.

P.S. If you are from California I didn't mean any offense by using the word slave.

Hammett 08-26-2004 05:23 PM

Crashed_Again

Don't worry about master and slave, I also call it like this. The drives have always been master and slave.
I guess (and kilgoretrout's post made me clear it) that the problem is the IDE cable. Now I just would like to know if transfer of around 30-40MB/s from HD to HD sharing the same IDE channel are acceptable (for ATA 133 drives) or if I need to buy a 80 wires flat IDE ribbon.

Anyways, many thanks for the post. ^_^

Electro 08-26-2004 06:20 PM

Maxtor drives are know to have DMA problems when they are sharing the IDE channel. Also using a motherboard based SIS chipset makes the problem even worst. You can disconnect the slave hard drive to see if that helps. If it does, put the slave hard drive on another seperate IDE channel. You can also include noapic or apci=off on the append line for lilo or on the kernel line in grub.

Copying files from one hard drive to that other on the same IDE channel will not be fast. It will be about a third of the transfer rate of the hard drive.

Not all round IDE cables are the same. You will have to go through a few round cables until the problems are minimized. Keep bending to a minimum. Bending the wire screws up the propertise and makes sending high frequencies through it harder.

Hammett 08-26-2004 06:34 PM

Electro,

With the old flat ribbon IDE cable, I have no problems, the transfers seem to be OK (based on the information you have given: 133/3 = 44Mb/s) that's more or less what i get.
The DMA is not disabled anymore and all the errors from /var/log/messages are now gone. I suppose it was rounded IDE cables's fault.
I'll try to keep the bending to the minumum, although with a Thermaltake chasis is a bit difficult, as the place for HD is *very* close to one of the doors....

Thanks again to all of you ^_^

Electro 08-27-2004 02:58 PM

Quote:

(based on the information you have given: 133/3 = 44Mb/s) that's more or less what i get.
Wrong calculation. Its the transfer rate that is given from hdparm. It will be about 10 to 20 megabytes to copy between one hard drive to the other hard drive on the same IDE channel. Hard drives never reach 100 or 133 megabytes per second. They can if the hard drive was directly connected to the motherboard and the filesystem you are using is raw.

Hammett 08-27-2004 06:01 PM

When i do the hdparm -Tt test on both HD i get around 56-58Mb/s and trnsfering between HDs (based on the info given by gkrellm) is around 30-40Mb/s.


All times are GMT -5. The time now is 06:58 PM.