Tuesday, November 22, 2005

Martrix RAID.... Just what the doctor ordered!

From wikipedia: "Matrix RAID utilizes two physical disks. Part of each disk is assigned to a level 0 array, the other part to a level 1 array. This product provides a safe area (the level 1 section) for documents and other items that one wishes to store redundantly, and a faster area for operating system, applications, etc."


The only problem is that "Currently, most (all?) of the other cheap RAID BIOS products only allow one disk to participate in a single array." Which means that this is not possible in Hardware, though "Matrix RAID is a feature that first appeared in the Intel ICH6R RAID BIOS." Which means that it has initially been implemented in hardware. Now, only if this were available on my Hardware RAID controller, or even Software RAID on Linux.

Of DMA and southbridges....

After getting a new motherboard, Linux seemed to be running fine except that I kept getting this message on boot up:
VP_IDE: Unknown VIA SouthBridge, contact Vojtech Pavlik < vojtech@ucw.cz >

Later on, I realized that my CPU was being used every time I read/wrote anything to Disk. This CPU usage was very evident when I decided to burn a CD. It took up nearly 60% of my CPU time! Considering that this used to be about 5% on my older CPU,it is a 15 times increase over the previous value. This led me to believe that somehow the kernel has disabled the DMA, and all the reads/writes are being routed through the CPU.

I was wondering whether DMA is controlled by the north or south bridge of the computer, and I tried asking a few of my friends, but they were as clueless as me. I immediately opened my crappy text-book to see if they had any clues, but there were none :-( However, I did notice that the ATA/SATA connections go into the Southbridge(SB), and the SB is not directly connected to the CPU, but is to the NB. Wow! This made me believe that DMA is controlled by the SB. To confirm, a quick search on google revealed the following:


Which proved that I was right!

So much for that ;-)
Now it only remains to get a new kernel:d

Sunday, November 20, 2005

Xmms bug-fix.

I had been wanting to fix this xmms bug for quite some time now... Was feeling really lazy. Got down to doing it today though.

What's the bug? If you enqueue many songs, and then clear the playlist, and the load a new playlist, then the "queued list" isn't cleared, and contains pointer to the old(now non-existant list). This causes xmms to crash when the currently playing song gets over. It tries to play a non-existant entry! Made it clear the queued-list as well, when the playlist itself is cleared. owever, I'm not entirely sure this is a 100% corrective patch, because it may still be possible to make xmms crash by removing that song which has been queued. I haven't checked if xmms removes queued songs rom the queued-list when they are removed individually from the playlist.

If you want a patch, mail me. It's only 2 lines worth though ;-)