Hacker's ramblings

I've had my run of bad things with Intel soft-RAID earlier. The constant RAID-verify -runs made me want stop using it. As its my Windows-box, I just wanted something that is hardware-based, reasonably fast and affordable. My choice is HighPoint RocketRAID 620.

For a switch-over -project I had a simple plan:

1. Clonezilla the existing RAID into an another drive which I could plug into a motherboard for the duration of the move
2. Un-configure the Intel soft-RAID at the motherboard
3. Plug in the RocketRAID-card
4. Change the hard-drive cables from motherboard RAID-connectors into RocketRAID-card
5. Configure a new RAID-1 mirror wit RocketRAID
6. Clonezilla the data back to the newly created RAID-1 volume
7. Be happy and continue computing

Guess what. Things fell trough at point #6. I was using an USB-bootable Clonezilla live on my first data move and obviously was planning to use it for the second one too. Whichever Linux-distro they use as the base for Clonezilla, they don't have the driver for the RAID-card. Darn!

The next best thing is a commercial distro for Clonezilla, Parted Magic. They used to be free (as in beer and speech), but they went commercial. The price is $9 USD for a single download, so I got it. And guess what again! They don't support Highpoint RocketRAID either.

I did ask about it in their support forums (closed to registered users only, sorry). And they replied:

We do not do "random" out-of-tree drivers because commonly these are supported by their vendors in a haphazard way. E.g. in the HighPoint case the latest driver is 3 versions behind our kernel version.

Luckily the vendor is providing the partial source code for the driver. There is a binary-part of in it and it is kind-of open-source. The biggest problem seems to be, that it doesn't build on any reasonably modern Linux.

By googling, I found that somebody else had the same process of thought and there was a Github project for the upgraded driver. Unfortunately that too was 3 years old and wouldn't build. Also it was for the vendor driver 1.1, and they already had 1.2 out.

In this imperfect world everything that you need to be done properly, you need to do by yourself. So, here it is: https://github.com/HQJaTu/rr62x
You can help yourselves with that one.

This is how it looks on my dmesg:

[ 85.518732] rr62x: module license 'Proprietary' taints kernel.
[ 85.518737] Disabling lock debugging due to kernel taint
[ 85.519709] rr62x:RocketRAID 62x SATA controller driver v1.2 (Jul 1 2012)
[ 85.735773] rr62x:adapter at PCI 3:0:0, IRQ 16
[ 85.950487] rr62x:[0 0 ] start port.
[ 85.950488] rr62x:[0 0 ] start port hard reset (probe 1).
[ 86.150712] rr62x:[0 1 ] start port.
[ 86.150712] rr62x:[0 1 ] start port hard reset (probe 1).
[ 89.093649] rr62x:[0 0 ] start port soft reset (probe 1).
[ 89.841048] rr62x:[0 1 ] start port soft reset (probe 1).
[ 90.501075] rr62x:[0 0 ] port started successfully.
[ 90.501078] rr62x:[0 0 0] device probed successfully.
[ 90.791364] rr62x:[0 1 ] port started successfully.
[ 90.791369] rr62x:[0 1 0] device probed successfully.
[ 90.806570] scsi host13: rr62x
[ 90.806870] scsi 13:0:0:0: Direct-Access HPT DISK_13_0 4.00 PQ: 0 ANSI: 5
[ 90.809711] sd 13:0:0:0: [sdd] 2930114560 512-byte logical blocks: (1.50 TB/1.36 TiB) [ 90.809847] sd 13:0:0:0: [sdd] Write Protect is off
[ 90.809852] sd 13:0:0:0: [sdd] Mode Sense: 2f 00 00 00
[ 90.809909] sd 13:0:0:0: [sdd] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
[ 90.832339] sdd: unknown partition table
[ 90.832903] sd 13:0:0:0: [sdd] Attached SCSI disk

I've tested that with Linux 4.0.4 and 3.19.3. It builds and works on both. Any comments, Github forks, pull requests, etc. are welcome. I will get back to the actual disk cloning project later, the driver won't  help unless it is used properly in an operating system.