BTX Loader crashes -- Help wanted

[Michael Powell]

Ronald F. Guilmette wrote:
[snip]
>>Try 8RC3 and see if any difference. I believe some work in this area may
>>have occurred.
> 
> I just tried it.  Alas, same result.

I follow the -CURRENT and -STABLE mail lists as well as this one. Though 
this particular problem does not pertain to me, I seem to recall some 
traffic off and on about this subject occasionally. So it is known. The 
place to get the developers to look is wrt to 8.0, so if the problem is well 
documented they may be more inclined to look into it. If the bootonly, 
install CD, or LiveFS CD for 8-RC3 can be used to reproduce the problem 
concentrate here.
 
>>I don't believe you are the first to experience this.
 
> Well, I'm just about to file a new PR on this, but I'll refrain if
> someone else has alreadyt done so.  Do you have an eisting PR number
> on this?

There are quite a number and a few of which are very similar in that they 
directly reference boot loader crashes. You could review the ones that seem 
to match the closest to your situation directly (e.g. hardware and crash-
dump wise), particularly the BTX crash associated with booting from SATA 
CD/DVD. Eye them towards using as a template to get started. The more 
exacting and succinct the PR the more likely to stimulate interest. You can 
reference the handful that match the closest by number in your own PR.

Be very exact to localize the trouble to specifically SATA CD/DVD hardware. 
If the box will boot and install fine from a PATA CD/DVD drive to a SATA 
hard drive be sure to include this. You have a VIA VT8237A controller on 
that board so that aspect should work. This will serve to isolate and 
confine the problem to be examined to a very specific issue. This increases 
the chances someone may look into it. 

 
[snip]
> 
> Sheeeesh! I literally _just_ bought this new SATA DVD drive, and I went
> with SATA because I believed that (a) the world is slowly but surely
> switching everything over to SATA and (b) SATA has been around long enough
> now that FreeBSD related bugs should have all been shaken out by now.
> 
> Please excuse my snarkiness, but... I guess I was wrong about the latter.
> 
>>If 8 does the same thing file a PR in order to bring the attention of the
>>developers. There may be one, or more, already on the subject.
> 
> Well, I did a search on the PR database for "BTX" and I'm looking at all
> those PRs... some of them going back to 2004, which doesn't exactly
> inspire confidence about a possible timely fix... and I don't see anything
> in the subjects that quite matched up to what I'm talking about.

If you just need the box to work immediately use a PATA CD/DVD. If you have 
the time to deal with it, attract the attention of developer(s), and have 
the time to work with them it serves the interests of the larger community. 
Others have had and will have your problem and getting it fixed for you will 
just mean many others will not go through what you are currently 
experiencing.
 
> And ah... while we are on the subject...
> 
> If I do file a PR on this, then at long last I'll need to know the answers
> to the two questions that have been in the back of my mind for ages,
> regarding PRs...
> 
>    1)  What do the various severity codes mean?
>    2)  What do the various proirity codes mean?

http://www.freebsd.org/doc/en_US.ISO8859-1/articles/problem-reports/

http://www.freebsd.org/doc/en_US.ISO8859-1/articles/pr-guidelines/

Might be good reading. I should read them again myself. It's been a while.

> 
> I've never filed a PR with severity "critical" or with priority "high"
> because I've always figured that this may be a good way to get the
> developers to view _all_ one's future (and past) PRs with a suspicious/
> jaundiced eye...  you know... the-boy-who-cried-wolf syndrome.
> 
> I don't want to be labeled as a nut case or an incessant complainer, but
> for _this_ issue I'm thinking that severity==critical and/or
> priority==high
> may be appropriate.  I mean jeezzz Louise!  If one can't even install from
> the distribution CDs/DVDs on perfectly good hardware...  (And it's not
> like the whole SATA interface standard is exactly ``new'' or anything
> anymore.)
> 
> So?  Any advice?  Should I stick my neck out and label this PR either
> severity==critical or priority==high ?
> 

There is always going to be a certain subjectivity present here. What is 
life and death important to one person may not be to another. I believe a 
commonly accepted dividing line can be found when you consider the usage of 
the system(s) in question. If you are a sysadmin or consultant who is being 
paid money to maintain mission critical servers then it warrants a higher 
level of concern than a single user at home with a desktop PC.

If you are racking a hundred Dell 2950's and have a problem it is critical. 
If you are a single PC user at home with a desktop, not so much. I do 
understand your concern wrt to "the boy who cried wolf" and it does matter. 
You stand more of a chance to get something fixed if the PR complaint is 
something that gives the developers a well defined starting point into the 
problem. Remember this is a volunteer effort so people need to pick and 
choose what they can spend their limited amounts of time on, so try and view 
it from a "if I was a developer what would I find interesting enough to work 
on?" perspective.

I'm certainly no expert on what you should label this. I do think that 
severity==critical/priority==high may be overreaching. This level would be 
more appropriate for the guy in the data center racking a 100 servers. So 
maybe consider severity==serious/priority==medium? This would, however, just 
be my own subjectivity speaking and may not be for the best. The guys who 
maintain the bug reports have the power to change these levels as they see 
fit.

But if you can spend the time to do a quality PR it has the potential to 
benefit the entire community. 

-Mike