Andi I recently upgraded to 4GB of ram and I don't seem to have the corruption problems. I'm running hardware raid 0 accross 2 sata drives. I copied practicaly the entire root filesystem onto the sata array as a test. This was about 34G of stuff. My root filesystem is a normal IDE drive. I ran rpm -V -a before and after the test. I have an arima hdama motherboard and I have IOMMU disabled in the bios. Any idea if this makes a difference? Mark Andi Kleen wrote:

On Tue, 06 Jan 2004 16:08:13 -0500 Mark Horton wrote:

...

SATA raid does work under suse you just need compatible hardware. I have a 4 channel 3ware sata raid controller card (8506-4LP) and it works great under suse 9.0 pro x86_64.

This controller card works better (for me anyway) than on-board sata because on-board sata is usually run through the standard 133 mhz pci bus. 4 sata drives in raid 0 would flood this bus most certainly. 2 drives would actually. The 3ware card is pci-x and has about 4-5 times the bandwidth.

Please note that 3ware only works out of the box with < ~3.5GB of memory (= no IOMMU needed because no memory is above the 4GB boundary) To avoid corruption with more you need an patched kernel with a workaround that is not officially released yet. Inofficial version in ftp://ftp.suse.com/pub/people/ak/test12/

-Andi

...

I recently upgraded to 4GB of ram and I don't seem to have the corruption problems. I'm running hardware raid 0 accross 2 sata drives. I copied practicaly the entire root filesystem onto the sata array as a test. This was about 34G of stuff. My root filesystem is a normal IDE drive. I ran rpm -V -a before and after the test.

What kernel do you run? The old 2.4.19 kernel didn't have this problem. It was only a side effect of some other changes in the 2.4.21 kernel.

I'm running: uname -a Linux suse 2.4.21-171-smp #1 SMP Wed Dec 17 17:52:34 UTC 2003 x86_64 x86_64 x86_64 GNU/Linux I just reran the test by copying the majority of my root filesystem to the sata array and still no changes reported with rpm -V -a. I also read through the previous threads on this subject and someone had the suspicion that it could also be related to a tyan issue (bios?). Since I dont have tyan I wonder if this is correct. I guess I should be glad its not effecting my system in any case. Mark

...

I have an arima hdama motherboard and I have IOMMU disabled in the bios. Any idea if this makes a difference?

It doesn't make a difference. Linux can use the IOMMU even when it is disabled in the BIOS.

-Andi