I know that the 4/670MP isn't officially supported by Redhat, so please don't take this as a personal attack, I just thought someone might like to know, or someone may already know a work-around/fix for this problem... I recently took all of the hardware out of an SS-10, CPU's, disk controller, etc, and put it back in a 4/670MP that I had retired, and fired it up to play news server for a while. It has run stable under 2.2.x. It had 384MB of RAM in it. But I had other 4/670MP's around and I had this idea that I could take memory from them and put more in this machine allowing it to cache more and maybe speed up news processing. With 384MB the machine boots and runs Linux like a top. I tried to increase this to 512MB but when I put the additional memory in the boot does not complete. After doing the partition check, it prints something like: dma device 29 unknown dma device 29 unknown dma device 29 unknown dma device 29 unknown It prints that four times, then complains that it can't open the root partition and waits for L1-A. Take the last 128MB of memory back out and everything is OK again. No changes to fstab or disk configuration or anything other than memory. The memory test tested all 512MB without any problem and there wasn't any other indication of bad memory. Is there something about the 4/670MP DMA hardware that restricts it to 384MB? I tried putting more memory in one of these running SunOS once and it didn't complain but didn't run stable.