Subject: Re: non-512-byte-sector devices vs. UBC
To: None <,>
From: Ross Harvey <>
List: tech-kern
Date: 06/07/1999 16:34:05
I don't think this decision can be made without asking "exactly what does
< 512-byte-dev support get us"? That is, just the fact that something out
there exists isn't good enough. Change is the enemy of reliability, so it
has to be something important to us, and something worth breaking our system
for. The hard design decisions are what Not To Support.

Also, the fact that raw audio CD blocks are slightly over 2k is very
uninteresting. There are no filesystems to deal with, and these blocks are
not addressable in the way the previous poster seemed to think.

Guess what? One of the reason the data sector (exactly 2k) is less than
the audio sector (2352) is exactly because of bits that identify the sector
number on data disks so the drive can read it!

There are subchannels within the audio data that allow approximate random
access within a track, but the only way to get to a precise point is to
backup, restart, compare the bit streams, and try to resync. Most drives
have only an approximate audio seek capability. This is not something that
the kernel can really deal with.  There are a *few* drives that will give you
the subchannel for a sector and that understand audio CD data at the sector
level. Sort of. And not very many...

So, I'm not sure what to do, but I know that the decision has to include
the benefit analysis, not just the cost analysis.

And I know it can't have anything to do with the quoted "audio CD" sectors.