Intel ICH5 SATA 150 disk controller support? (Louis LeBlanc)

[Ron Dzierwa]

Lou,

if people would stop slapping Soren on the back, and telling him what a
wonderful job he's doing, and, instead, require him to make his driver
work, i think we would all be better off.

Its not uncommon for him to drop support for something that seemed to
work previously just because he can't get his "crappy" (to use Soren's
own term) driver working!

It seems that if he can't seem to figure out a particular piece of hardware,
it becomes the hardware's problem, and he then refers to it as a "crappy
chipset", or a "broken bios"  (regardless of how many other drivers/os's
seem to work just fine with the same hardware.).  Meanwhile, the rest
of the FreeBSD community seem to think he's doing a spectacular job
and tell you that you don't have to use FreeBSD!!

Don't even try to suggest a fix, or offer help debugging a problem.  It seems
that either his arrogance is far beyond accepting help, or he's genuinely
afraid that the house of cards he calls a driver really will fall apart.

I recommend that, if you ANY sata controller working with ANY release of
FreeBSD, you stick with it.  Even the ones that he claims work don't really
perform that all that well under load.  After several weeks of struggling with
the sata controller on my tyan s2875s, i went out and bought a promise pci
sata controller.  It doesn't crash, but is doesn't run all that fast either.  I
wanted to stack the drives in a raid configuration to take advantage of their
speed.  Each drive is capable of 50-60 meg/second continuous.  If i put
two together, i get 80-90 when i add a third, i get 75!!!  All the while,
the cpu seems to spend an awful lot of time in the kernel (top tells me that
system time is about 30%!!)!!!  Oddly enough,  "crappy" old XP doesn't
seem to have a problem saturating the pci bus - with 3 drives I get 120.
The fourth drive doesn't seem to add much more, which is what i would
expect with pci/32. Additionally, the onboard SiI chip works fine with
Windoze or Linux. The excuse here is that the people at Silicon Image
don't provide enough information about their chips for Soren to create a
good driver. Ok, maybe not, but i would have expected the VIA chip on
the promise controller to work, since it's part of the supported hardware.
What i didn't expect was that FreeBSD would perform so badly, but i
guess i should have based on previous experience with the ata driver.

So much of FreeBSD is so far ahead of anything it seems such a shame
that so critical of a part be so poorly done.


sorrry, just had to vent i suppose...

ron.