Quality of FreeBSD

[Karl Denninger]

On Fri, Jul 22, 2005 at 12:45:03AM +0100, Robert Watson wrote:
> 
> On Thu, 21 Jul 2005, Karl Denninger wrote:
> 
> >ATA-NG (Soren's new code) is not (from what I understand) in the 5.x 
> >codebase.  One bone of contention is that apparently it IS in -HEAD, but 
> >there are no plans to MFC it to 5.x.
> 
> Then you misunderstand.  Soren has asked to MFC it, and we've asked him 
> to wait until it's had more testing exposure, precisely because it is a 
> sensitive code base, and we don't want to see further regression.
> 
> Robert N M Watson

I don't think I misunderstand at all Robert.

"We" (some group) has asked him not to MFC it.  

Ergo, IT IS NOT THERE NOW, and there are no plans (at present) to MFC it.

That's exactly what I said.

However, it obviously wasn't that big of a deal (to the -committers) to 
commit the ORIGINAL changes which broke the implementation going from 4.x 
to 5.something (early 5.x "early adopter" RELEASEs were ok).

What I don't understand Robert is why Soren's code is "too sensitive" to 
commit, but the explosive reduction in stability that the changes made 
between 4.x and 5.3 caused weren't enough to back THAT out until it could 
be fixed.

Its not like these problems didn't show up almost immediately when the
affected releases hit the street.  They did.  Six months later, the
problems are still there, and I see nothing in the commit logs to 
suggest that the underlying issues have been addressed.

Papering over the failures so that retries work properly (when they were
broken before) isn't a fix.  A fix would be identifying the root cause of 
the DMA_TIMEOUT errors and addressing them so that they no longer occur.

I realize that this is likely a timing issue in the code, and therefore 
is difficult to debug.

That does not, however, change the fact that this issue has been open for
more than six months without resolution, and that one potential resolution
to the problem (Soren's ATA-NG code) either (1) doesn't fix it, (2) hasn't 
been tested to see if it does, or (3) DOES fix it, but for whatever
internal reasons has not been MFC'd.

If (1), then not only should Soren's code NOT be MFC'd, but 6.x should
absolutely be held until it IS identified and resolved.

If (2), then how about trying to find out of if that solves the problem?

If (3), I think there are a few of us (myself included) that would like an
explanation.

If Soren BELIEVES (2) is the case, I'll test against -BETA1, IF I can have
confirmation that -BETA1 has the ATA-NG code in it.  

Its trivially easy for me to reproduce this problem on my sandbox machine.

--
-- 
Karl Denninger (karl at denninger.net) Internet Consultant & Kids Rights Activist
http://www.denninger.net	My home on the net - links to everything I do!
http://scubaforum.org		Your UNCENSORED place to talk about DIVING!
http://homecuda.com		Emerald Coast: Buy / sell homes, cars, boats!
http://genesis3.blogspot.com	Musings Of A Sentient Mind