Problem with Promise Ultra66

jpc82 · Guru Joined: 09 Mar 2003 Posts: 326

I was wondering if there is some sort of known problem with Linux and the Promise Ultra66 PCI IDE card.

I have been running 2 80GB drives on this card with no problem for over a year, however when I added a 120GB drive to it I'm getting DMA errors, especially when I try to right file larger then a couple of gigs.
hdg: dma_intr: status=0x51 { DriveReady SeekComplete Error }
hdg: dma_intr: error=0x84 { DriveStatusError BadCRC }

However, when I just it in the built in controller I don't have any problems, or when I put it on an other computer.

Could this be a problem with the drivers for this card?

NOTE* The card is on its own chain, and the only drive doing any access at the times of the errors.

grepcomputers · Guru Joined: 16 Sep 2003 Posts: 375

Try updating the BIOS on the Promise IDE card.

Also, it wouldn't hurt to run drive diagnostic software from whoever manufactured the 120GB drive - there could be a hardware error.

cheers...

jpc82 · Guru Joined: 09 Mar 2003 Posts: 326

I ran the diagnotics and and even a low level format from the tools seagate provides and they both finished with no ptoblem.

I also did a destructive badblocks, with multiple passes, and I wrote one HUGE (120GB) file from urandom. Everythign while running at ATA100 (Other computer is ATA100 capable) with no error messages.

One my other computer ATA33 I wrote one huge 20GB file from urandom with no errors.

I will give the bios idea a try when I get a chance.

jpc82 · Guru Joined: 09 Mar 2003 Posts: 326

I tried updating the Bios for my mobo and the ultra66 card, however I'm still getting the errors.

grepcomputers · Guru Joined: 16 Sep 2003 Posts: 375

Hmmm.
You're using the drivers that came with the kernel? What kernel revision do you use?

What does lspci and dmesg tell you about the drives and the controller card?
Dmesg might also have more error messages that might be worth seeing.

Also, did you try a new IDE cable? A new power cable (if you have a spare one)?

cheers!

jpc82 · Guru Joined: 09 Mar 2003 Posts: 326

I have tried 3 different cables, and I'am currently using the one that was used to test in my other machine.

I'm currently using gentoo-dev-source-2.6.5

My lspci

jpc82 · Guru Joined: 09 Mar 2003 Posts: 326

I also noticed that when another drive on the controler is in use there is alot of errors

jpc82 · Guru Joined: 09 Mar 2003 Posts: 326

Well it seems the linux drivers for promise card don't like running with more then 2 cards attached to the controller.

When I put in another card, and keeped the two 80GB drives on the Ultra66, and put the 120GB on a Ultra100, I don't get any errors, even when all drives are under heavy load.

grepcomputers · Guru Joined: 16 Sep 2003 Posts: 375

I'm glad you figured it out. Sorry I have not been more helpful recently, but I've been madly trying to recover data from a dead drive.

good luck!

labrador · Guru Joined: 04 Oct 2003 Posts: 316

I just went through a similar experience. In my case a 20 GB
drive died in my software RAID array, and my 40 GB drive hot
spare kicked in. I bought a replacement drive to act as the next
hot spare which is 120 GB. Before comitting any data to it, I
ran badblocks. The badblocks write test didn't find any faults,
but it did trigger many of these dma_intr errors with error 0x84
BadCRC. I tried switching cables and it didn't help (as suggested
by many forum items on these dma_intr errors). I moved
the 120 GB drive over to the Ultra 133 card instead of using
the Ultra 66 card, and then tested again with badblocks. Now
no errors. There are 2 40 GB drives now on the Ultra 66 and
they are not causing any errors.

I have the latest BIOS for the Ultra 66 card - 2.0 build 18, and
that doesn't resolve the problem. The Ultra 66 has a problem
with either large drives or with drives having a large cache
(8K in my case).

labrador · Guru Joined: 04 Oct 2003 Posts: 316

I now have 2 40 GB drives on my Ultra66. The 120 GB seagate
was put on the Ultra 66 and generated dma_intr errors, and then
it was moved off to the Ultra133. So now I have a 40 GB Seagate
(fairly new) on the Ultra 66 and it also generates errors. There is
also a Quantum 40 GB drive on the Ultra 66 that has no problem.
The old Quantum 20 and 40 GB drives never had a problem on the Ultra 66.
So I have to conclude it isn't the size, but something related
to the firmware on seagate drives, or their faster data speed
(outperforms the older 20 and 40 GB Quantum 7200 RPM fireball)
according to hdparm -t

I have no choice but to buy a new Ultra 100, which will hopefully
avoid the problem as the Ultra 133 does.