Re: Bug#2063: scsi driver sequence unreasonable
Hi Jeff,
> IMHO, the fewer changes between Debian's kernels and the upstream
> ones, the better...
You are right. I introduce almost no changes at all, except as I feel
necessary or very beneficial. Even then, I restrict changes to common
and available patches. Linus has a focus and a set of goals in
releasing 1.3 patches, which I fully support (most of the time, anyway
:-). We need to support a more ``bread & butter'' user community and
this is where I may introduce few things. If they break anything,
please let me know and I will fix them or back off the change.
> I don't see why we should move any of them. They're arranged in the
> order they are in for two reasons: 1) to prevent probing problems (i.e.
> BusLogic must come before Adaptec 154x), and 2) to allow the more
> popular and/or problematic cards to come first.
I think many of us understand why the SCSi drivers are arranged as they
are. The logic is correct but the listing order is not always so. The
BusLogic case is a good example. In moving the eata_dma, for example, I
followed a slightly differet logic (?): Since it is a unique driver, it
really does not matter where in hosts.c it is defined. The reason to
move it was that most users, when having a smart/expensive controller
want to have it used for booting and for the root file system.
I think the real solution lies elsewhere; I am developing a
configuration tool which will allow us to choose the order of the
devices without editing hosts.c. It may take some time to surface and
you may beat me to it, but that's OK.
> It doesn't really matter if a 152X gets detected before a high-power
> whiz-bang SCSI-matic 2010 PCI adapter, because you can still put root
> on any SCSI controller you like.
You are right. My own machine had root on /dev/sde2 for the longest
time. I moved it as many users, during first installation do not
realize it. If it is very opjectionable, I could be convinced to move
it back, but then i will have to ship an untested kernel. I hate to do
that!
> If there is a _really_ good reason to change the probe order, we should
> discuss it on the kernel and scsi channels, and get it changed in the
> upstream source.
You are right. I am collecting opinions on what the order should be and
once my office move is over, will start the action as you suggest it.
> Adaptec provides free programming information on the 154X and 174X series,
> as well as the AIC-6X60 chips (152X), so be nice... I'm not thrilled with
> their policy on current products, but they're still excellent products. :)
I think they have not made up their mind as to what they want to do when
they grow up. An OEM chip vendor or a PC card vendor. They created so
much noise and confision with so many incompatible cards and products.
I have a 2940W in ``as new'' condition. Any offers? They are excellent
cards :-)
Happy new year
P.S. Lets not turn it into a flaming war of my card is better than your
card. Besides, thereis a company (in Colorado? Forgot their name)
which makes a FCAL card. Does about 85 MB/Sec and can take up to 256
drives. Now try & beat that!
Simon
P.S. Please ignore the below address and flame shimon@teleport.com.
He receives and answers mail :-)
Simon Shapiro Bullet Technologies, Inc.
shimon@Bullet-Tech.com 13130 SW Haystack St.
(503) 524-6631 Beaverton, OR 97005
Reply to: