[lkml]   [2001]   [Apr]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: Help with Fasttrack/100 Raid on Linux
On Wed, 18 Apr 2001, Wilfried Weissmann wrote:

> > off to the head or tail of the drive and get me that raid-voodoo-bios-os
> > communication transport layer, and do it ins DMA modes, NOW!"
> "voodoo" would be the sector where the raid configuration is stored
> then, right!? Maybe I did not understand you properly...

We are on the same page, good.

> I am really trying to understand what you want to do. From what I have
> seen this would go into kernel 2.5 first. For now there appears to be no
> information or this-would-be-new-in-2.5 list available of new
> development kernel so far, so I have to use my fantasy even more.
> So you want to introduce a new layer between the MD driver and the ATA
> HDs of IDE-RAIDs. This code would be placed in the ide-chipset driver.
> Configuration happens at system startup after the chipset is in service.
> There you read the config-sector (e.g. sector 9 on HPTs) of the disks
> and datafill your configuration tables. Plus you start one (or more) MD
> device with the according raid level(s). The MD then accesses your layer
> instead of talking directly to the disk driver, and the requests are
> mapped (buffer_head->rsector+10 with the HPTs) as the bios would do it.
> But the raid is still handled by MD. So you can completely reuse the
> raid code for all that.
> Pardon me, but please tell me that I am wrong, because this sounds odd
> to me! <:(
> Talking about the API... These should be the basic steps that we need to
> do (unordered, this is just brainstorming):
> *) device_size_calculation() should use a callback of the raid level to
> get the device size. Or the code should be completly moved over to the
> <raid level>_run()'s.
> *) Hide/unhide disks from the userland (this is just a cosmetic issue).

Disks must be visable, always, but blocked from direct mounting.

> *) Shift sectors and shrink capacity of disks so that the existing raid
> levels can access the disks according to the ata-raid layout.


> *) Get the configuration sector from disk. Analyse the configuration and
> setup disks and md-devices.

Yep and update the status of the raid if we have a device fail.

> *) All raid pers. must be able to handle I/O that requests sectors from
> more than only one disk.


> *) Partitioned raid devices must be handled somehow.


> >
> > I do not have the desire to do personality tables, but I can.

> > With about 96% of all linux boxes in the world dependent on some form of
> > ATA/ATAPI, Linus and Alan are very sensitive to even the sligthest change.
> I would avoid to change anything there, but do a raid personality...
> Well, we are back at our your solution vs. my raid personality
> discussion again!

No we are back to your "raid personality", and a functional API to the
ATA driver.

This is probably the first and last time I will openly agree for someone
to tell me were to go, and do it ;-).

You tell me what you want the driver to do, and I will make it happen.
It will be legal and technically correct. Does that sound like a good


Andre Hedrick
Linux ATA Development
ASL Kernel Development
ASL, Inc. Toll free: 1-877-ASL-3535
1757 Houret Court Fax: 1-408-941-2071
Milpitas, CA 95035 Web:

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:27    [W:0.058 / U:0.180 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site